Evaluation of Ethiopia's Food Security Program: Documenting Progress in the Implementation of the Productive Safety Nets Programme and the Household Asset Building Program

Full text

EDRI
ETHIOPIA STRATEGY SUPPORT PROGRAM II
(ESSP II)
EVALUATION OF ETHIOPIA’S FOOD SECURITY PROGRAM:
DOCUMENTING PROGRESS IN THE IMPLEMENTATION OF THE
PRODUCTIVE SAFETY NETS PROGRAMME AND THE HOUSEHOLD
ASSET BUILDING PROGRAMME
ESSP II – EDRI REPORT
Guush Berhane, John Hoddinott, Neha Kumar, Alemayehu Seyoum
Taffesse, Michael Tedla Diressie, Yisehac Yohannes,
Rachel Sabates-Wheeler, Mulugeta Handino, Jeremy Lind,
Mulugeta Tefera, and Feyera Sima
Ethiopia Strategy Support Program II (ESSP II)
International Food Policy Research Institute
Institute of Development Studies, University of Sussex
Dadimos
May 2013

ii
THE ETHIOPIA STRATEGY SUPPORT PROGRAM II (ESSP II)
ABOUT ESSP II
The Ethiopia Strategy Support Program II is an initiative to strengthen evidence-based policymaking
in Ethiopia in the areas of rural and agricultural development. Facilitated by the International Food
Policy Research Institute (IFPRI), ESSP II works closely with the government of Ethiopia, the
Ethiopian Development Research Institute (EDRI), and other development partners to provide
information relevant for the design and implementation of Ethiopia’s agricultural and rural
development strategies. For more information, see http://essp.ifpri.info, http://www.ifpri.org/book-
757/ourwork/program/ethiopia-strategy-support-program or http://www.edri-eth.org.
The Ethiopia Strategy Support Program II is funded by a consortium of donors comprising the United
States Agency for International Development (USAID), the UK Department for International
Development (DFID), and the Canadian International Development Agency (CIDA).
This Ethiopia Strategy Support Program II (ESSP II) report contains preliminary material and research
results from IFPRI and/or its partners in Ethiopia. It has not undergone a formal peer review. It is
circulated in order to stimulate discussion and critical comment. The opinions are those of the authors
and do not necessarily reflect those of their home institutions or supporting organizations.
IFPRI – ETHIOPIA STRATEGY
SUPPORT PROGRAM
http://essp.ifpri.info
Contact:
IFPRI c/o ILRI
P.O. Box 5689
Addis Ababa, Ethiopia
Tel: +251 11 6 17 25 55
Fax: +251 11 6 46 23 18
Email: mahlet.mekuria@cgiar.org
ETHIOPIAN DEVELOPMENT
RESEARCH INSTITUTE
http://www.edri-eth.org
Contact:
P.O. Box 2479,
Addis Ababa, Ethiopia
Tel: +251 11 5 50 60 66;
+251 11 5 52 53 15
Fax: +251 11 5 50 55 88
Email: info@edri-eth.org
INTERNATIONAL FOOD POLICY RESEARCH
INSTITUTE
http://www.ifpri.org
Contact:
2033 K Street, NW
Washington, DC
20006-1002 USA
Tel: +1 202 862 5600
E-mail: ifpri@cgiar.org
About the Author(s)
Guush Berhane: Ethiopia Strategy Support Program, International Food Policy Research Institute
John Hoddinott: Poverty, Health, and Nutrition Division, International Food Policy Research Institute
Neha Kumar: Poverty, Health, and Nutrition Division, International Food Policy Research Institute
Alemayehu Seyoum Taffesse: Ethiopia Strategy Support Program, International Food Policy Research Institute
Michael Tedla Diressie: HarvestPlus, International Food Policy Research Institute
Yisehac Yohannes: Poverty, Health, and Nutrition Division, International Food Policy Research Institute
Rachel Sabates-Wheeler: Institute of Development Studies, University of Sussex
Mulugeta Handino: Institute of Development Studies, University of Sussex
Jeremy Lind: Institute of Development Studies, University of Sussex
Mulugeta Tefera: Dadimos
Feyera Sima: Dadimos

iii
Evaluation of Ethiopia’s Food Security Program:
Documenting Progress in the Implementation of the
Productive Safety Nets Programme and the Household
Asset Building Programme
Guush Berhane, John Hoddinott, Neha Kumar,
Alemayehu Seyoum Taffesse, Michael Tedla Diressie,
Yisehac Yohannes, Rachel Sabates-Wheeler, Mulugeta Handino,
Jeremy Lind, Mulugeta Tefera, and Feyera Sima
Ethiopia Strategy Support Program II (ESSP II)
International Food Policy Research Institute
Institute of Development Studies, University of Sussex
Dadimos
Copyright © 2013 International Food Policy Research Institute. All rights reserved. Sections of this material may be reproduced for
personal and not-for-profit use without the express written permission of but with acknowledgment to IFPRI. To reproduce the
material contained herein for profit or commercial use requires express written permission. To obtain permission, contact the
Communications Division at ifpri-copyright@cgiar.org.

iv
Contents
Acronyms ................................................................................................................................................. xi
Acknowledgments .................................................................................................................................. xiii
Executive Summary ................................................................................................................................. 1
1. Introduction ........................................................................................................................................ 9
1.1. Introduction ................................................................................................................................. 9
1.2. Objectives and Structure of the Report ..................................................................................... 10
2. Data Sources and Methods ............................................................................................................. 14
2.1. Introduction ............................................................................................................................... 14
2.2. Quantitative Methods ................................................................................................................ 14
2.3. Qualitative Methods .................................................................................................................. 24
3. Food Security, Assets, and Coping Strategies ................................................................................ 28
3.1. Introduction ............................................................................................................................... 28
3.2. Context ...................................................................................................................................... 28
3.3. Incidence of Shocks .................................................................................................................. 30
3.4. Food Security ............................................................................................................................ 35
3.5. Asset Levels .............................................................................................................................. 47
3.6. Subjective Well-being ................................................................................................................ 53
3.7. Summary ................................................................................................................................... 57
4. Woreda Perspectives on Implementation ........................................................................................ 58
4.1. Introduction ............................................................................................................................... 58
4.2. Woreda Resources for the PSNP ............................................................................................. 59
4.3. Woreda Differences in the Timeliness of Cash Payments ........................................................ 64
4.4. Woreda Differences in the Timeliness of Food Transfers ......................................................... 71
4.5. Woreda Resources and Early Warning and Response Capacity ............................................. 72
4.6. Contingency Funds ................................................................................................................... 75
4.7. Summary ................................................................................................................................... 77
5. Kebele and Household Perspectives on Implementation ................................................................ 78
5.1. Introduction ............................................................................................................................... 78
5.2. The Kebele and Community Food Security Task Forces ......................................................... 78
5.3. Households’ Interactions with the Community FSTF ................................................................ 81
5.4. Households’ Experiences with the Payment Process ............................................................... 83
5.5. Client Cards .............................................................................................................................. 89
5.6. Summary ................................................................................................................................... 90
6. Targeting .......................................................................................................................................... 92
6.1. Introduction ............................................................................................................................... 92
6.2. Targeting Principles in the PSNP .............................................................................................. 92
6.3. Regional Perspectives .............................................................................................................. 94
6.4. Woreda Perspectives ................................................................................................................ 95
6.5. Kebele Perspectives ................................................................................................................. 98
6.6. Household Perspectives ......................................................................................................... 101
6.7. Selection into Public Works .................................................................................................... 106
6.8. Selection into Direct Support .................................................................................................. 109
6.9. Switching from Public Works to Direct Support and vice versa .............................................. 113
6.10. Summary ................................................................................................................................. 114

v
7. The Implementation of Public Works Projects ............................................................................... 115
7.1. Introduction ............................................................................................................................. 115
7.2. Coordinating the Implementation of Public Works .................................................................. 115
7.3. Planning Public Works ............................................................................................................ 116
7.4. Linking Public Works and Local Livelihoods ........................................................................... 118
7.5. Implementation Problems ....................................................................................................... 121
7.6. Gender Considerations ........................................................................................................... 123
7.7. Summary ................................................................................................................................. 126
8. Payments for Public Works and Direct Support ............................................................................ 128
8.1. Introduction ............................................................................................................................. 128
8.2. Public Works Employment ...................................................................................................... 128
8.3. Payments for Public Works: Levels ........................................................................................ 131
8.4. Payments Levels Relative to Entitlements .............................................................................. 136
8.5. Payments for Direct Support ................................................................................................... 152
8.6. Preferences for Cash and Food .............................................................................................. 158
8.7. Summary ................................................................................................................................. 164
9. Appeals and Complaints ................................................................................................................ 166
9.1. Introduction ............................................................................................................................. 166
9.2. Sources of Appeals and Complaints ....................................................................................... 166
9.3. Who People Appeal to ............................................................................................................ 168
9.4. How Appeals Are Presented and Resolved ............................................................................ 168
9.5. Woreda Food Security Task Force and Kebele Food Security Task Force Views on the
Visit of the Rapid Response Team ........................................................................................ 174
9.6. Summary ................................................................................................................................. 176
10. The Other Food Security Program and its Transition to the Household Asset Building
Programme .................................................................................................................................... 177
10.1. Introduction ............................................................................................................................. 177
10.2. The Other Food Security Program .......................................................................................... 177
10.3. Women’s Access to Credit and Household Packages ............................................................ 180
10.4. The Household Assets Building Programme .......................................................................... 181
10.5. Rollout of the HABP ................................................................................................................ 182
10.6. Knowledge about HABP and its Linkages with the CCI, OFSP, and the PSNP ..................... 185
10.7. The Role of Development Agents in the HABP ...................................................................... 187
10.8. Household Knowledge and Access to the HABP ................................................................... 190
10.9. Access to Credit ...................................................................................................................... 192
10.10. Conclusions ........................................................................................................................... 195
11. Graduation ..................................................................................................................................... 196
11.1. Introduction ............................................................................................................................. 196
11.2. Prevalence of Graduation ....................................................................................................... 197
11.3. Understanding of Graduation .................................................................................................. 198
11.4. Graduation Benchmarks ......................................................................................................... 201
11.5. Is there Premature Graduation? ............................................................................................. 203
11.6. Incentives for Graduation ........................................................................................................ 206
11.7. Disincentives for Graduation ................................................................................................... 207
11.8. Support Needed for Graduation from PSNP ........................................................................... 209
11.9. Support Post-PSNP Graduation ............................................................................................. 210
11.10. Summary ............................................................................................................................... 211
References ........................................................................................................................................... 213

vi
Tables
Table E.S.1. Evaluation objectives covered in this report ...................................................................................... 1
Table 1.1. Evaluation objectives covered in this report ....................................................................................... 11
Table 2.1. Informants participating in the quantitative woreda survey (percentage) ............................................ 16
Table 2.2. Number of woredas, kebeles, and households sampled, by region ................................................... 18
Table 2.3. Design of the 2010 household questionnaire ..................................................................................... 20
Table 2.4. Sample numbers by round ................................................................................................................. 22
Table 2.5. Attrition by region ............................................................................................................................... 22
Table 2.6. Probit showing marginal effects of the factors affecting the probability of attrition .............................. 23
Table 2.7. Location of qualitative studies ............................................................................................................ 24
Table 2.8. Key informant interviews .................................................................................................................... 25
Table 4.1. Evaluation objectives covered in chapter 4 ........................................................................................ 58
Table 4.2. Characteristics of key woreda staff ..................................................................................................... 59
Table 4.3. Quality of access to main electricity (percent of woredas) .................................................................. 61
Table 4.4. Quality of access to landline telephones (percent of woredas) ........................................................... 62
Table 4.5. Quality of cell phone coverage (percent of woredas).......................................................................... 62
Table 4.6. Quality of cell phone and landline access (percent of woredas) ......................................................... 62
Table 4.7. Access to government-owned vehicles (percent of woredas) ............................................................. 62
Table 4.8. Access to working computers and printers (percent of woredas) ....................................................... 63
Table 4.9. Access to manuals (percent of woredas)............................................................................................ 63
Table 4.10. Aggregate PSNP resources (percent of woredas) ............................................................................ 64
Table 4.11. Regional variations in number of days taken to deliver cash payments ........................................... 65
Table 4.12. Woreda perceptions of problems affecting payments ....................................................................... 68
Table 4.13. Associations between woreda characteristics and time spent on PSNP payment activities ............. 70
Table 4.14. Regional variations in number of days taken to deliver food transfers ............................................. 72
Table 5.1. Evaluation objectives covered in chapter 5 ........................................................................................ 78
Table 5.2. Composition of the Kebele Food Security Task Force (KFSTF), by region ........................................ 80
Table 5.3. Kebele Food Security Task Force (KFSTF) record keeping, by region .............................................. 81
Table 5.4. Household contact with the Community Food Security Task Force (CFSTF), by region, year,
and beneficiary status ...................................................................................................................... 82
Table 5.5. Household knowledge of the role of the Community Food Security Task Force (CFSTF), by
region and beneficiary status ........................................................................................................... 83
Table 5.6a. Beneficiaries received all information needed to understand how the program works (percent),
by region .......................................................................................................................................... 84
Table 5.6b. Beneficiaries received all information needed to understand how the program works (percent),
by sex and schooling of household head ......................................................................................... 84
Table 5.7. Beneficiaries’ perceptions regarding being paid in full (percent) ........................................................ 84
Table 5.8. Beneficiaries’ perceptions regarding the timeliness of payments (percent) ........................................ 85
Table 5.9. Number of days’ notice that beneficiaries had that payments would be made (percent) .................... 85
Table 5.10. Beneficiaries’ perception that they were treated courteously (percent)a .......................................... 85
Table 5.11. Beneficiaries’ experiences with travel to the payment site ................................................................ 88
Table 5.12. Percentage of public works participants who reported holding a client card, by region .................... 89
Table 6.1. Evaluation objectives covered in chapter 6 ........................................................................................ 92
Table 6.2. Kebele perceptions of adequacy of resources for public works participants ..................................... 100
Table 6.3. Kebele perceptions of adequacy of financial resources to meet direct support needs ..................... 100
Table 6.4. Advice given to kebeles for addressing resource shortfalls .............................................................. 101
Table 6.5. Percentage of households reporting that the following criteria were used to select public works
participants in their locality, by region and year ............................................................................. 102
Table 6.6. Percentage of households reporting that the following criteria were used to select direct support
participants in their locality, by region and year ............................................................................. 103
Table 6.7. Consistency of household participation in the public works component of the PSNP, by region ...... 106
Table 6.8. Correlates of receipt of public works (PW), by year .......................................................................... 108

vii
Table 6.9. Correlates of receipt of public works (PW), by year and region ........................................................ 109
Table 6.10. Consistency of household participation in the direct support component of the PSNP, by
region ............................................................................................................................................. 111
Table 6.11. Correlates of receipt of direct support (DS), by year ...................................................................... 112
Table 6.12. Correlates of receipt of direct support (DS), by region and year ..................................................... 113
Table 7.1. Evaluation objectives covered in chapter 7 ...................................................................................... 115
Table 7.2. Percentage of households participating in selection of public works, by region and year ................ 117
Table 7.3. Percentage of households participating in selection of public works, by region and sex of head,
2010 ............................................................................................................................................... 117
Table 7.4. Household perceptions of whether they benefitted from public works .............................................. 120
Table 7.5. Household perceptions of whether they benefitted from public works, by sex of head, 2010 ........... 120
Table 7.6. Kebele level perceptions of the comparative importance of income transfers to the poor and the
construction of public works as priorities for the PSNP .................................................................. 123
Table 8.1. Evaluation objectives covered in chapter 8 ...................................................................................... 128
Table 8.2. Distribution of employment by sex and region, Tir–Ginbot, 2010 ..................................................... 130
Table 8.3. Mean payments (Birr) to households with public works participants, by month and region, 2009 .... 133
Table 8.4a. Mean cash payments (Birr) to households with public works participants, by month and region,
2009 ............................................................................................................................................... 134
Table 8.4b. Mean food payments (Birr) to households with public works participants, by month and region,
2009 ............................................................................................................................................... 134
Table 8.5. Mean payments (Birr) to households with public works participants, by month and region, 2010 .... 134
Table 8.6a. Mean cash payments (Birr) to households with public works participants, by month and region,
2010 ............................................................................................................................................... 135
Table 8.6b. Mean food payments (Birr) to households with public works participants, by month and region,
2010 ............................................................................................................................................... 135
Table 8.7. Number of food payments to households with public works participants, by month and region,
2010 ............................................................................................................................................... 135
Table 8.8. Entitlements and number of payments (public works), by woreda, 2009 .......................................... 140
Table 8.9. Entitlements, days worked in 2009 and 2010, number of payments, 2010, and household size,
by woreda ...................................................................................................................................... 144
Table 8.10. Spearman correlation coefficients between entitlements and other measures of PSNP
implementation (public works) ....................................................................................................... 146
Table 8.11a. Number of months worked and number of payments, 2009, Tigray ............................................. 147
Table 8.11b. Number of months worked and number of payments, 2009, Amhara .......................................... 148
Table 8.11c. Number of months worked and number of payments, 2009, Amhara-HVFB ................................ 148
Table 8.11d. Number of months worked and number of payments, 2009, Oromiya.......................................... 149
Table 8.11e. Number of months worked and number of payments, 2009, SNNPR .......................................... 149
Table 8.12. Distribution of implied wages (Birr), by region and assumptions of labor supplied ......................... 150
Table 8.13. Distribution of implied wages (Birr), by region and woreda under conservative assumptions of
labor supplied ................................................................................................................................ 151
Table 8.14. Mean in-kind and cash direct support payments 2009 and 2010, by region ................................... 154
Table 8.15. Frequency of direct support payments (percent), Tir–Ginbot 2009, by region ................................ 154
Table 8.16. Frequency of direct support payments (percent), Tir–Ginbot 2010, by region ................................ 154
Table 8.17. Median value (Birr) of direct support transfers, by month and region, 2009 ................................... 155
Table 8.18. Direct support payment data for Tir–Ginbot, by region, 2010 ......................................................... 156
Table 8.19. Median direct support payments, by household size and region, 2009 .......................................... 157
Table 8.20. Use of cash transfer mentioned, by focus groups, in order of priority usage .................................. 162
Table 8.21. Locality median price per kg, by crop, June 2010 .......................................................................... 163
Table 9.1. Sources of appeals and complaints, by region ................................................................................. 166
Table 9.2. Who appeals to whom? .................................................................................................................... 168
Table 9.3. Composition of the Kebele Appeals Committee ............................................................................... 169
Table 9.4. The composition of the Kebele Appeals Committee, by gender in study areas ................................ 171
Table 9.5a. Number of households perceiving selection process was unfair, lodging complaints, and
receiving a response, by sex of household head and region ......................................................... 173

viii
Table 9.5b. Percentage of households perceiving selection process was unfair, lodging complaints, and
receiving a response, by sex of household head and region ......................................................... 174
Table 10.1. Evaluation objectives covered in chapter 10 .................................................................................. 177
Table 10.2. Rollout status of HABP, by region and woreda ............................................................................... 183
Table 10.3. Development agent staffing at the kebele level .............................................................................. 189
Table 10.4. Household knowledge of HABP and access to development agents (DA) ..................................... 190
Table 10.5. Household knowledge of HABP and access to development agents (DA), by sex of head and
beneficiary status ........................................................................................................................... 190
Table 10.6. Information provided by development agents ................................................................................. 191
Table 10.7. Number of RUSACCOs and their operational status, by woreda ................................................... 192
Table 10.8. Sources of credit for agriculture, by region ..................................................................................... 193
Table 10.9. Source of loans for agricultural purposes ....................................................................................... 194
Table 10.10. Sources of credit, by sex of head and household beneficiary status ............................................ 194
Table 11.1. Evaluation objectives covered in chapter 11 .................................................................................. 197
Table 11.2. Elements included in benchmarks at kebele level (interviews with development agents and
Food Security Task Force) ............................................................................................................ 202
Table 11.3. Value of livestock holdings (Birr), Sene EC 2000 (June 2008), by beneficiary status between
2008 and 2010, and region ............................................................................................................ 205
Figures
Figure 2.1. Woredas surveyed in the quantitative surveys .................................................................................... 15
Figure 3.1a. Price changes, Tigray, 2006–08 and 2008–10 .................................................................................. 28
Figure 3.1b. Price changes, Amhara, 2006–08 and 2008–10 ............................................................................... 29
Figure 3.1c. Price changes, Oromiya, 2006–08 and 2008–10 .............................................................................. 29
Figure 3.1d. Price changes, SNNPR, 2006–08 and 2008–10 ............................................................................... 30
Figure 3.2. Incidence of shocks ............................................................................................................................ 30
Figure 3.3a. Incidence of shocks, Tigray ............................................................................................................... 31
Figure 3.3b. Incidence of shocks, Amhara ............................................................................................................ 31
Figure 3.3c. Incidence of shocks, Amhara-HVFB.................................................................................................. 32
Figure 3.3d. Incidence of shocks, Oromiya ........................................................................................................... 32
Figure 3.3e. Incidence of shocks, SNNPR ............................................................................................................ 32
Figure 3.4. Incidence of shocks, by public works (PW) beneficiary status and year ............................................. 34
Figure 3.5. Consequences of experiencing a drought, all households .................................................................. 35
Figure 3.6. Primary source of food, by month and year ........................................................................................ 37
Figure 3.7. Average number of months households run out of home-grown food. by drought conditions ............. 38
Figure 3.8. Average number of months households run out of home-grown food, by drought conditions and
PSNP beneficiary status .................................................................................................................. 38
Figure 3.9. Percent of households that suffered any food shortages in the rainy season, by region and year ...... 39
Figure 3.10. Percentage of households unable to satisfy food needs of the household ........................................ 40
Figure 3.11. Most severe food gap months, in 2010, 2008, and 2006 .................................................................. 41
Figure 3.12. Most severe food shortage month for PSNP and non-PSNP households ......................................... 42
Figure 3.13a. Number of meals per day adults eat (no food shortage) ................................................................. 43
Figure 3.13b. Number of meals per day children eat (no food shortage) .............................................................. 43
Figure 3.13c. Number of meals per day adults eat (during food shortage) ........................................................... 43
Figure 3.14a. Average number of meals per day—Adults ..................................................................................... 45
Figure 3.14b. Average number of meals per day—Children ................................................................................. 45
Figure 3.15. Percentage of households reporting different coping strategies, by year and region ........................ 46
Figure 3.16. Percentage of households reporting different coping strategies, by PSNP beneficiary status
and year ........................................................................................................................................... 47
Figure 3.17. Value of production assets owned (Birr) ........................................................................................... 48
Figure 3.18. Value of production assets owned (Birr), by beneficiary status ......................................................... 48
Figure 3.19. Value of livestock owned (Birr) .......................................................................................................... 49

ix
Figure 3.20. Value of livestock owned (Birr), by beneficiary status ....................................................................... 50
Figure 3.21. Tropical Livestock Units owned ......................................................................................................... 51
Figure 3.22. Tropical Livestock Units owned, by beneficiary status ...................................................................... 51
Figure 3.23. Distress sale of assets for satisfying food needs (proportion of households) .................................... 52
Figure 3.24. Distress sale of assets (proportion of households), by beneficiary status ......................................... 53
Figure 3.25. Compared to other households, you describe your household as… ................................................. 53
Figure 3.26a. Proportion of households able to borrow 100 Birr ........................................................................... 54
Figure 3.26b. Average share spent on . . . if given 100 Birr .................................................................................. 54
Figure 3.27a. Households’ food consumption (percent of households)................................................................. 55
Figure 3.27b. Household’s housing needs (percent of households) ..................................................................... 55
Figure 3.27c. Household’s income (percent of households) ................................................................................. 56
Figure 3.28. Overall economic situation of household compared with one year ago... (percent of
households) ..................................................................................................................................... 56
Figure 4.1. Type of training received: Head, Food Security Office ........................................................................ 60
Figure 4.2. Type of training received: PSNP accountant ....................................................................................... 60
Figure 4.3. Type of training received: PSNP cashiers ........................................................................................... 61
Figure 4.4. Cumulative mean number of days to deliver payments, by activity ..................................................... 65
Figure 4.5. Variations in number of days to deliver payments: 10th, 50th, and 90th percentiles .............................. 66
Figure 4.6. Timeliness of payments, by tertiles ..................................................................................................... 67
Figure 5.1. Travel times to payment sites ............................................................................................................. 88
Figure 5.2. Expenses associated with the receipt of client cards, by region (Birr) ................................................ 90
Figure 6.1. Percentage of households receiving direct support, by region and year ........................................... 110
Figure 7.1. Percentage of households reporting that public works were undertaken on their land, by wealth
category and region, 2010 ............................................................................................................. 121
Figure 8.1. Number of days worked per beneficiary household, Tir–Ginbot, by survey year .............................. 129
Figure 8.2. Mean days worked per beneficiary household, 2007 and 2009, by region and six-month period ..... 130
Figure 8.3. Age distribution of individuals employed on public works projects, by sex, 2010 .............................. 131
Figure 8.4. Mean payments to households with public works participants between Tir and Ginbot, by region
and year ......................................................................................................................................... 132
Figure 8.5. Mean payments (Birr) to households with public works participants, 2009, by region and six-
month period .................................................................................................................................. 133
Figure 8.6a. Comparison of normalized total payment (public works) to full family targeting entitlement, by
household size, Tigray, 2009 ......................................................................................................... 137
Figure 8.6b. Comparison of normalized total payment (public works) to full family targeting entitlement, by
household size, Amhara, 2009 ...................................................................................................... 138
Figure 8.6c. Comparison of normalized total payment (public works) to full family targeting entitlement, by
household size, Amhara-HVFB, 2009............................................................................................ 138
Figure 8.6d. Comparison of normalized total payment (public works) to full family targeting entitlement, by
household size, Oromiya, 2009 ..................................................................................................... 139
Figure 8.6e. Comparison of normalized total payment (public works) to full family targeting entitlement, by
household size, SNNPR, 2009 ...................................................................................................... 139
Figure 8.7. Normalized public works payments as a percentage of entitlement under full family targeting, by
household size and region, 2009 ................................................................................................... 140
Figure 8.8. Scattergram, woreda mean percentage entitlement received, by woreda mean number of
payments, public works, 2009 ....................................................................................................... 143
Figure 8.9. Mean payments to direct support beneficiary households, by region and year ................................. 153
Figure 8.10. Distribution of direct support payments across regions, between Tir and Ginbot 2010 ................... 156
Figure 8.11. Distribution of direct support payments as a percent of entitlement under full family targeting,
2009 ............................................................................................................................................... 157
Figure 8.12a. Preferences for food and cash, 2006–2010, Tigray ...................................................................... 158
Figure 8.12b. Preferences for food and cash, 2006–2010, Amhara .................................................................... 159
Figure 8.12c. Preferences for food and cash, 2006–2010, Amhara-HVFB ......................................................... 159
Figure 8.12d. Preferences for food and cash, 2006–2010, Oromiya ................................................................... 160
Figure 8.12e. Preferences for food and cash, 2006–2010, SNNPR .................................................................... 160
Figure 8.13. Locality preferences for food and cash, by maize prices ................................................................. 164

x
Figure 10.1. Percentage of development agent offices with 0, 1, 2, and 3 or more agents ................................. 188
Figure 11.1. The graduation process .................................................................................................................. 197
Figure 11.2. Percentage of households that has graduated from PSNP based on kebele reports, 2010 ............ 198
Boxes
Box 11.1. Responses to the question: “Why are people graduating too early?“ .................................................. 204
Annex
Annex A. Tag System for Direct Quotes ............................................................................................................. 214
Annex B. Characteristics of Woredas Surveyed Using Qualitative Methods ....................................................... 215
Annex Table B.1. Livelihood zones, strategies, and risks in the sample woredas............................................... 216
Annex Table B.2. Woreda population, household size, and number of PSNP and HABP/OFSP beneficiary
households ........................................................................................................................... 222

xi
Acronyms
ACSI Amhara Credit and Savings Institute
ARD Agriculture and Rural Development
BOARD (Regional) Bureau of Agriculture and Regional Development
BOFED Bureau Office of Finance and Economic Development
CBPWD Community-based participatory watershed development
CCI Complementary community investment
CFI Chronically food insecure
CFSTF Community Food Security Task Force
CIS Corrugated iron sheets
CPI Consumer Price Index
CSA Central Statistical Agency (Ethiopia)
DA(s) Development Agent(s)
DPO Disaster Prevention Office
DPPA Disaster Preparedness and Prevention Agency
DPPO Disaster Prevention and Preparedness Office
DS Direct support
EAs Enumeration areas
EW Extension Worker
EWR Early warning and response
FFT Full family targeting
FG Focus Group
FGD Focus Group Discussions
FSCD Food Security Coordination Directorate
FSP Food Security Program
FSS Food Security Survey
FSTF Food Security Task Force
GOE Government of Ethiopia
HABP Household Asset Building Programme
HVFB High Value Food Basket
IDS Institute of Development Studies
IFPRI International Food Policy Research Institute
KAC Kebele Appeals Committee
KFSTF Kebele Food Security Task Force
KII Key Informant Interview
MFIs Microfinance institutions
MOARD Ministry of Agriculture and Rural Development
NGO Nongovernmental organization
NRM Natural Resource Management
OFSP Other Food Security Program
OVC Orphans and Vulnerable Children
PA Peasant Association
PIM Project Implement Manual
PSNP Productive Safety Nets Programme
PW Public Works
RPWFU Regional public works focal units
RRT Rapid response team

xii
RUSACCO Rural savings and credit cooperatives
SME Small and Medium Enterprises
SNNPR Southern Nations, Nationalities, and People’s Region
SWC Soil and water conservation
TC Technical Committee
TLU Tropical Livestock Unit
TOR Terms of reference
TOT Training of trainers
USAID United States Agency for International Development
WARDO Woreda Agricultural and Rural Development Office
WFSD Woreda Food Security Desk
WFSO Woreda Food Security Office
WFSTF Woreda Food Security Task Force(s)
WOARD Woreda Office of Agriculture and Rural Development
WOFED Woreda Office of Finance and Economic Development
WPWFU(s) Woreda Public Works Focal Unit(s)

xiii
Acknowledgments
This work has been funded under World Bank Award 100025484/2010 Impact Evaluation of
the Ethiopia Food Security Program with additional funding from the US Agency for
International Development (USAID). The material presented here would not have been
available without the superb work undertaken by the Ethiopian Central Statistical Agency
(CSA) in implementing the household and community surveys on which this report is based.
We especially thank Weizro Samia Zekaria, Director General of the CSA, for her support.
The staff at Dadimos played a critical role in ensuring that the qualitative fieldwork was
undertaken at a high standard and it is a pleasure to recognize their contribution here. We
are grateful for the administrative support provided by the International Food Policy
Research Institute’s (IFPRI) Ethiopian Strategy Support Program and in particular the
contributions provided by Tigist Malmo and Mahlet Mekuria as well as assistance from
Caroline Guiriec and Jay Willis. At the Institute of Development Studies (IDS), Saul Butters
and Linnet Taylor provided some support for the analysis of the qualitative information. We
have benefitted from conversations with many people on aspects of the Food Security
Program. In the preparation of this report, discussions with Ato Berhanu Wolde-Micheal,
Sarah Coll-Black, Matt Hobson, and Wout Soer and comments received from participants in
a workshop held in Addis Ababa in February and June, 2011 and in Addis Ababa, Awassa,
Bahir Dar, and Mekelle in August 2011, have been invaluable. We, the authors of this report,
are solely responsible for its contents.

1
Executive Summary
This report documents progress in the implementation of the Productive Safety Nets
Programme (PSNP) and the Household Assets Building Programme (HABP) and assesses
trends in perceptions of the effectiveness and transparency of the PSNP and HABP among
different groups of clients. It also describes how living standards are evolving in PSNP and
non-PSNP beneficiary households. The report addresses the following evaluation objectives
found in the Food Security Program (FSP) Log Frame and the Terms of Reference for this
study.
Table E.S.1. Evaluation objectives covered in this report
Evaluation
objective
Issue
Chapter in
this report
Link to Log Frames and TOR
Document progress in the implementation of the PSNP
Are public works payments timely and
predictable? Do clients receive complete
entitlement?
4, 5, 6, 8
PSNP Log frame Output 1.1
PSNP Log frame Output 1.2
TOR, para 39
Are direct support payments timely and
predictable? Do clients receive complete
entitlement?
4, 5, 6, 8
PSNP Log frame Output 1.1
PSNP Log frame Output 1.2
TOR, para 35
Pregnant female participants are moved
from public works to direct support
6
PSNP Log frame Output 1.4
Transfers are received at a place no more
than three hours from home
5
PSNP Log frame Output 1.5
Households participate in PSNP for at
least three consecutive years
6
PSNP Log frame Output 1.6
Why are there woreda level differences
in timeliness of transfers?
4
TOR, para 36
How are contingency budgets used,
including as a response to the kebele
Appeals system?
4
TOR, para 41
Can gender dimensions of access be
better captured?
6, 7, 8
TOR, para 42
Document progress in the implementation of the Household
Asset Building Programme (HABP)
Can gender dimensions of access be
better captured?
10
HABP Log frame Output 1.2
TOR, para 42
HAB clients have access to financial
services
10
HABP Log frame Output 2.1
HAB clients have access to desired input
10
HABP Log frame Output 3.2
Assess trends in perceptions of the effectiveness and
transparency of the PSNP and HABP among different groups of
clients
Beneficiaries understand how the
program works
5, 6, 7, 9
PSNP Log frame Output 4.4
Beneficiaries and non-beneficiaries
report that targeting and graduation
processes are fair
6, 11
PSNP Log frame Output 4.5
Source: Authors’ compilation.

2
Several recurring themes emerge in this report:
A number of positive aspects of program implementation identified in previous reports
have remained in place or have improved further. Targeting remains good, the
appropriate administrative structures exist, and beneficiary participation in the selection
of public works has stayed the same in some regions and improved in others.
A number of important capacity strengthening initiatives are evident. More than 90
percent of surveyed woredas use the payroll and attendance sheet system (PASS).
Considerable effort has been made to increase the number of development agents at the
kebele level. There is widespread acknowledgment that this has led to an improvement
in support provided by development agents.
Apart from these positive aspects, there is considerable evidence of regional differences
in the implementation of the PSNP. Southern Nations, Nationalities, and Peoples Region
(SNNPR), Tigray, and Amhara-HVFB outperform Amhara and Oromiya across a range
of activities, most notably those related to the provision of work, payment levels, and the
processing of payments. Self-reported payments data indicate that Full Family Targeting
has not been implemented in Amhara. These data also show that direct support
payments in Oromiya are considerably below what is envisaged in the PIM.
Payment predictability has been a major concern. The evidence provided here shows a
mixed picture. On a year-to-year basis, predictability has improved. For example, access
to public works has become more consistent over time. Between 50 and 72 percent of
current public works participants received payments for public works in 2008, 2009, and
2010. However, many beneficiaries indicate that there is considerable uncertainty
regarding payment dates.
Gender dimensions also present a mixed picture. In nearly all (98 percent) surveyed
kebeles, there is at least one woman and at least one development agent on the Kebele
Food Security Task Force (KFSTF) and women are fairly represented (at least one
woman per kebele) in the Kebele Appeals Committee (KAC). Male- and female-headed
households reported similar perceptions of benefits of public works activities. But both
women and men report that women experience significant difficulties in balancing
required participation in public works with other household responsibilities. Relative to
male-headed households, female-headed households were less likely to lodge a
complaint if they perceived that selection processes were unfair, had less contact with
development agents, and were less likely to use the credit facilities established under the
HABP.
Understanding of the objectives and operations of the PSNP is good at the regional level
but becomes progressively poorer at the woreda, kebele, and household level. In
particular, we highlight the poor understanding of the concept of graduation. Below
woreda level, the understanding of the concept becomes very loose, at times completely
uninformed, and at times completely incorrect. Most disconcertingly, many graduates
interviewed had no understanding of why they had been graduated.

3
Chapter 3 describes price trends, incidence of shocks, and summary statistics of key
outcomes of interest.
In 2010, food prices continue to rise, but the rate of increase is much slower than 2008.
SNNPR experienced a decline in food prices.
Drought is the most important shock that affects a large fraction of households every
year and causes income and consumption shortfalls.
The incidence of shocks differs across regions, but does not vary much across
beneficiary status.
When looking at households’ primary source of food, we find that a large fraction
constitutes food from own production. However, this fraction has steadily declined. The
fraction sourced from PSNP has increased steadily over the period.
The food gap as measured by the number of months that the household is unable to
satisfy its food needs fell from 3.6 months to 2.3 months among all households.
FSP beneficiary households, on average, hold lower levels of assets compared to non-
beneficiaries. This is reasonable, since the FSP beneficiaries were selected to be the
more vulnerable and poor.
Over the period 2004–2010, asset levels have increased. Although beneficiary
households have not experienced accumulation of assets at a fast pace, they have
shown a steady increase.
There has been a decline in distress sale of assets, irrespective of beneficiary status.
On subjective measures of well-being, also households have fared better in 2010 as
compared to 2008. In particular, when asked about how they felt their overall economic
condition was as compared with last year, in 2010, 70 percent of households feel that
they are either the same or a little better-off as compared to 41 percent in 2008.
Chapter 4 assesses whether concerns over capacity to implement the PSNP have been
addressed and whether this has contributed to more timely transfers to beneficiaries. There
are several noteworthy findings.
Woreda level offices are increasingly well-resourced with trained staff. Nearly all now use
the PASS system and nearly all have functioning computers. Most, but not all, have
manuals that can be referred to. Training has occurred, although this could be more
widespread.
On average, it takes 38.9 days (as measured by the mean) or 32 days (as measured by
the median) from the time a woreda receives its first attendance sheet to the last
payment.
There is considerable variation in these times across woredas. The best performing
woredas manage to complete all activities associated with making payments in 21 days.
By contrast, the worst performers take nearly two months.
Factors that explain woreda level differences in the timeliness of payments include lack of
training on the PASS, the absence of frontloaded transfers, and a lack of transport.

4
Chapter 5 assesses implementation from the perspective of kebeles and households. It finds
that:
Kebele FSTFs exist in all surveyed localities. Broadly speaking, these match with what
was set out in the Project Implementation Manual (PIM). There appears to be a slight
improvement in their composition in Oromiya, where, in 2006, the KFSTFs deviated most
from the guidelines set out in the PIM. In nearly all surveyed kebeles, there is at least
one woman and at least one development agent on the KFSTF. Participant lists and
minutes of meetings are well-kept, but there appears to be a drop in the number of
KFSTFs that keep progress reports.
In both 2008 and 2010, PSNP beneficiaries were asked directly if they felt that they had
received all information needed to understand how the program works. There have been
noticeable improvements in self-reported understanding of the program in Amhara (from
65 to 83 percent), Amhara-HVFB (74 to 84 percent), and SNNPR (from 80 to 88
percent). There are slight declines in understanding in Tigray and a considerable
percentage of beneficiaries (45 percent in 2008 and 47 percent in 2010) in Oromiya who
do not feel that they have sufficient information about the PSNP.
In Oromiya, considerable concern was expressed about timeliness of payments. Only
14 percent of beneficiaries agree or strongly agree that they receive payments on a
timely basis and 37 percent strongly disagree. While Oromiya performs poorly by this
measure, it is worth noting that this concern is voiced by beneficiaries in all regions.
Beneficiaries are supposed to receive their transfers in a place no more than three hours
from home. This goal is largely met in Tigray, Oromiya, and SNNPR. However, a
considerable proportion of beneficiaries in Amhara and Amhara-HVFB must travel more
than three hours.
Across all surveyed beneficiaries, 43.6 percent report receiving a client card.
Chapter 6 assesses targeting performance and constancy of program participation.
The provision of an administrative quota begins at the federal level and extends all the
way down to the sub-kebele level. Regional and woreda officials broadly follow the
targeting criteria outlined in the PIM to determine administrative quotas, determining
allocations on the basis of previous relief caseloads, agroecological conditions,
malnutrition levels, the average size of landholdings, and the estimated population of
chronically food-insecure households in particular administrative areas. The upstream
budget review process, starting at the kebeles and moving upward to the woredas, the
regions, and the federal level to decide the total number of the PSNP beneficiaries, has
not been practiced.
Most exclusion errors are due to inadequate administrative quotas. Full family targeting
is practiced in most areas, even though officials acknowledge that it implies that fewer
households are targeted. Kebele officials frequently request additional funds, but
invariably these requests are denied.

5
Officials have used the contingency budget and stricter, locally-specific targeting criteria
to manage the limited administrative quota compared with the larger population of
chronically food-insecure that require support.
The labor cap and Full Family Targeting appear to be widely understood. However, there
has been a fall in the percentage of respondents who can identify specific criteria for
access to the PSNP and a rise in the percentage who perceive these are a consequence
of a quota system or a random allocation. Despite this, a majority of focus groups agreed
with the targeting process as well as the selection criteria used to determine public works
and direct support beneficiaries.
Access to public works and direct support has become more consistent over time.
Between 50 and 72 percent of current public works participants received payments for
public works in 2008, 2009, and 2010. Between 26 and 44 percent of current recipients
of direct support received direct support in 2008, 2009, and 2010.
Both the public works and direct support components of the PSNP are well targeted as
evidenced by estimates of the correlates of access to public work and direct support.
Outside of Tigray, there is no evidence of households being shifted from public works to
direct support or vice versa. However, focus group discussions with women indicated
that, when pregnant, they are switched out of public works and into direct support.
Chapter 7 explores aspects of the implementation of public works projects under the PSNP.
There is considerable variation in the extent of community involvement in the selection of
public works projects. There are clearly cases where these have been decided upon by
the woreda and consultation with kebeles and communities seems notional. But there
are other instances where development agents and community members have had a
significant say in deciding priorities and plans for public works.
Public works projects are generally perceived to assist in supporting livelihoods. Work on
roads is seen as particularly positively. Views on other activities, such as soil and water
conservation and natural resource management, are more mixed.
A number of implementation problems associated with lack of technical skills, difficulties
in supervision, and delays in receipt of funds for capital expenditures were noted. The
tension between the use of public works as a means of transferring funds to poor
households and as a means of strengthening community assets was also highlighted.
Participation in public works activities puts considerable strain on women. There appears
to be little willingness to address this.
Chapter 8 provides information on days worked and self-reports of payments for public
works employment, and direct support transfers.
In SNNPR and, to a slightly lesser extent in Tigray and Amhara-HVFB, there are clearly
efforts being made to ensure that work is available, that beneficiaries are paid for this
work, and that these payments reflect the amount of work undertaken. In 2009,

6
significant levels of transfers were made to public works beneficiaries in these regions
and to ensure that payments bear some relationship to entitlements.
In Tigray and SNNPR, there are clearly serious attempts to provide direct support
transfers, although the regions differ in how they implement this. In Tigray, there are
more beneficiaries, but with low levels of transfers; in SNNPR, the converse appears to
be the case.
By contrast, this dimension of program implementation lags in Amhara and Oromiya.
Total public works payment levels are lower as are the proportion of beneficiaries
receiving their entitlement.
Self-reported payments data indicate that Full Family Targeting has not been
implemented in Amhara.
These self-reports also show that direct support payments in Oromiya are considerably
below what is envisaged in the PIM.
There exists a very strong preference for food payment, given uncertainty about prices,
lack of food availability on the market, deliberate price hiking by traders when cash
payments are being disbursed, and nonequivalent value of food basket and current cash
payment. Beneficiaries were more likely to prefer cash when grain prices were below 3.5
Birr/kg.
Chapter 9 presents findings on the appeals and complaints process in the PSNP. It
describes the reasons for appeals, who people appeal to, how these are presented, whether
the appeals process is easy and transparent, and how long the appeals and complaint cases
take to be resolved.
Exclusion from the PSNP is the principal cause of appeals. Partial family targeting,
inclusion errors, deduction of payment, and delay of transfers are other reasons for
appeals and complaints.
Appeals are made to the KFSTF, kebele cabinet, the development agents, KAC, village
leaders, and others. Although Kebele Appeal Committees are widespread, they are not
perceived as being effective.
Women are fairly represented (at least one woman per kebele) in the KAC.
It appears that, relative to concerns raised about the selection process, relatively few
households appeal. There is a sense that the chances of appealing successfully are
limited and this may discourage use of appeal mechanisms.
Appeals are largely made verbally. It appears that decisions are made about these
somewhere between two weeks and two months after the complaint has been lodged.
The lack of written documentation surrounding appeals means that this figure represents
a “best guess.”

7
Chapter 10 presents an overview of the Household Assets Building Programme (HABP), and
explains the current rollout progress, challenges encountered thus far, its linkage with other
food security programs, and its role in facilitating the graduation process.
The roadmap for the HABP implementation at the regional levels is almost complete and
all necessary technical committees are established. The general comprehension about
the objectives and purpose of the program among the regional actors is very strong.
Significant variations were observed in the degree of knowledge among the actors, such
as the Woreda Food Security Task Forces and development agents at the woreda and
kebele levels. In particular, considerable confusion exists in terms of differentiating the
two programs (Other Food Security Program [OFSP] and HABP) among these actors.
Knowledge about HABP among the PSNP beneficiaries at the community level is nearly
nonexistent.
The links between the HABP and other food security programs is clear at the regional
level. However, considerable differences are observed between regional and woreda
actors in terms of understanding the relationship between the PSNP, OFSP, HABP, and
Complementary Community Investment (CCI).
Considerable effort has been made to increase the number of development agents at the
kebele level. There is widespread acknowledgement that this has led to an improvement
in support provided by development agents.
Many households report contact with development agents and, in particular, note that
they have received advice about new crops and how crops can be grown.
Rollout of the HABP below the regional level is uneven. In particular, development
agents and households who were surveyed have limited awareness of the HABP. In the
case of development agents, there is uncertainty as to how the HABP differs from the
OFSP.
Advice and assistance remains concentrated on crop production. There is limited
capacity to assist nonagricultural enterprises.
Access to new forms of credit, such as rural savings and credit cooperatives
(RUSSACOs), has been limited. Relatively few households report borrowing money to
purchase inputs or buy livestock.
Female-headed households and households receiving direct support rarely access
formal sources of credit.
Chapter 11 examines how graduation is understood and how it is applied in practice. Issues
surrounding premature graduation, incentives and disincentives for graduation, and support
needed at graduation and postgraduation are also discussed.
There appears to have been relatively little graduation to date.
There is a solid understanding at regional levels of the concepts and mechanisms of
graduation; however, only one mention of two-tiered graduation. Understanding at
woreda level is also fairly consistent. Below woreda level, the understanding of the
concept becomes very loose, at times completely uninformed, at times completely

8
incorrect. Perhaps most disconcerting is the finding that the graduates to whom we
spoke appeared to be the least informed of why they were graduated.
The knowledge of the process by which people graduate was very varied at the
community levels, with some thinking that it was time-dependent, others believing it to be
political, quota-driven, and others knowing the specific benchmark value for graduation.
The experience of graduates themselves indicated that the process was in most cases
not transparent or well-explained.
Officials at most levels are well informed of the actual graduation benchmarks. However,
when the official regional benchmarks are translated into specific criteria for targeting
potential graduate households at the kebele and community levels, a plethora of criteria
result.
Given FSP graduation targets, there is an imperative that graduates are increasingly
identified and taken off the program. This imperative is felt and discussed at all levels,
from regions right down to the communities. Sometimes respondents, particularly at the
regional level, discuss these targets in relation to quotas and required future rates of
graduation. However, at lower levels, these “quotas” are often seen as administratively
imposed and at odds with the time required for sustainable graduation.

9
1. Introduction
1.1. Introduction
The introduction to the document describing the government of Ethiopia’s Food Security
Programme 2010–2014 (GFDRE 2009a) notes that persistent food insecurity remains a
major problem in many parts of Ethiopia. To address this, the last ten years has seen a shift
away from ad hoc responses, such as those that characterized the major drought in 2002, to
a planned, systematic approach. This was embodied in the government of Ethiopia’s Food
Security Programme launched in 2005. The government of Ethiopia has noted that this
program had a number of significant achievements, inter alia:
More than seven million people have received PSNP transfers enabling them to
meet consumption needs, reducing the risks they faced, and providing them with
alternative options to selling productive assets. In addition, between 692,002
households (around 3.5 million people) received credit financed by the
government’s Federal Food Security Budget Line between 2005 and 2007 …/…
There is also significant evidence that the programme is having an impact. The
PSNP is smoothing consumption and protecting assets and a growing number of
PSNP clients are having growing access to household building efforts. Where
the two programmes are combined, particularly in areas where programmes
were well implemented (indicated by a high level of transfers), household asset
holdings have increased and crop production appears to have improved.
Despite these achievements, considerable food insecurity remains across much of Ethiopia
and graduation from the program—a major policy goal—has been limited. Consequently, in
2009 the government of Ethiopia relaunched the Food Security Programme with enhanced
efforts being made to improve a key component, the Productive Safety Nets Programme
(PSNP) and a replacement of the Other Food Security Programme (OFSP) with an
enhanced set of activities to strengthen the capacity of households to generate income and
increase asset holdings. The replacement to the OFSP, called the Household Asset Building
Programme (HABP), includes a demand driven extension and support component and
improvements in access to financial services.
The document describing the Food Security Programme (FSP) 2010–2014 notes that a
strength of the program has been the willingness of all stakeholders to ongoing learning
processes, taking information from monitoring reports and evaluation studies to strengthen
implementation. In this spirit, the new phase of the Food Security Programme pays particular
attention to ensure that this learning continues through a mixture of monitoring and
evaluation work.
This report has its origins in the intention of the government of Ethiopia (GFDRE 2009a, 77)
to carry out a biannual household survey to assess outcomes and impacts of all components
of the FSP in chronically food-insecure woredas. This biannual survey was first carried out in
2006 and again in 2008. In 2006, the sample consisted of approximately 3,700 households
located in 66 food-insecure woredas served by the PSNP. In 2008, these households were
resurveyed and an additional 1,300 households—located in woredas in Amhara that
implement the PSNP with support from NGOs—that received a High Value Food Basket

10
(HVFB)
1
were also included. A strength of the quantitative data used in this report is that it is
longitudinal—that is, it tracks the same communities and households, allowing us to see how
the program evolves and how household well-being changes over time.
An important fact is that, in 2010, this strength is complemented by the inclusion of a suite of
qualitative data collection techniques conducted in ten woredas where the quantitative
survey was fielded. There are considerable benefits to combining quantitative and qualitative
research methods. Doing so maximizes the advantages of the relative strengths of each
method, and provides a richer pool of data and greater analytic power than would have been
available with either of these methods alone. The main motivations for use of qualitative
research is the recognition that (1) processes of program implementation and how these are
received at the local level involve complex interactions, and may be different than anticipated
by program planners; understanding these interactions and their effects on outcomes
requires the use of qualitative research methods that develop trust and rapport between
researchers and respondents, allow for open-ended responses, and triangulation through
multiple respondents and participant observation; and is the recognition that (2) qualitative
methodologies can provide insights into the reasons and causes of certain impacts that will
not necessarily be uncovered by quantitative approaches alone; and that (3) views, opinions,
and interpretations of the program held by beneficiaries are important, credible, and worth
listening to.
1.2. Objectives and Structure of the Report
This is the first of three reports that will be produced, using data collected in 2010. The
objective of this report is to document progress in the implementation of the PSNP and the
HABP and assesses trends in perceptions of the effectiveness and transparency of the
PSNP and HABP among different groups of clients.
2
A second objective is to describe how
living standards are evolving in PSNP and non-PSNP beneficiary households.
The report addresses the following evaluation objectives found in the FSP Log Frame and
the Terms of Reference for this study.
1
This refers to the ‘full food basket’ provided to areas supported by USAID. This full food basket is based on the Sphere
Standarsd of 3 kgs of cereals, plus pulses and oils.
2
A second report will use double difference matching methods that measure the impact of the PSNP on the well-being of the
chronically food-insecure population and the complementary roles played by the PSNP and HABP in achieving positive and
negative outcomes for the food insecure and the reasons behind this. The third report will document livelihoods and the
implementation of the PSNP and HABP in Afar, Somali, and pastoral localities in Oromiya.

11
Table 1.1. Evaluation objectives covered in this report
Evaluation
objective
Issue
Chapter in
this report
Link to Log Frames and TOR
Document progress in the implementation of the PSNP
Are public works payments timely and
predictable? Do clients receive complete
entitlement?
4, 5, 6, 8
PSNP Log frame Output 1.1
PSNP Log frame Output 1.2
TOR, para 39
Are direct support payments timely and
predictable? Do clients receive complete
entitlement?
4, 5, 6, 8
PSNP Log frame Output 1.1
PSNP Log frame Output 1.2
TOR, para 35
Pregnant female participants are moved
from public works to direct support
6
PSNP Log frame Output 1.4
Transfers are received at a place no more
than three hours from home
5
PSNP Log frame Output 1.5
Households participate in PSNP for at
least three consecutive years
6
PSNP Log frame Output 1.6
Why are there woreda level differences
in timeliness of transfers?
4
TOR, para 36
How are contingency budgets used,
including as a response to the kebele
Appeals system?
4
TOR, para 41
Can gender dimensions of access be
better captured?
6, 7, 8
TOR, para 42
Document progress in the implementation of the Household
Asset Building Programme (HABP)
Can gender dimensions of access be
better captured?
10
HABP Log frame Output 1.2
TOR, para 42
HAB clients have access to financial
services
10
HABP Log frame Output 2.1
HAB clients have access to desired input
10
HABP Log frame Output 3.2
Assess trends in perceptions of the effectiveness and
transparency of the PSNP and HABP among different groups of
clients
Beneficiaries understand how the
program works
5, 6, 7, 9
PSNP Log frame Output 4.4
Beneficiaries and non-beneficiaries
report that targeting and graduation
processes are fair
6, 11
PSNP Log frame Output 4.5
Source: Authors’ compilation.
Below we summarize the topics covered in each chapter.
Chapter 2: Data Sources and Methods. This chapter describes the data sources and
methods that underpin this report.
Chapter 3: Food Security, Assets, and Coping Strategies. This chapter presents
statistics that describe the context in which the Food Security Programme has operated and
trends in outcomes of interest. In particular, we examine the price changes of main staple
food crops, livestock, and labor over the period 2006–2010. We use information from the
quantitative surveys to examine the extent of shocks experienced by households. It also
examines changes in asset levels, food security, and coping strategies; and subjective
measures of well-being.

12
Chapter 4: Woreda Perspectives on Implementation. This chapter examines whether
concerns regarding woreda level capacity to implement the PSNP have been addressed and
whether this has contributed to more timely transfers to beneficiaries. It also considers the
role played by early warning systems and the use of contingency funds.
Chapter 5: Kebele and Household Perspectives on Implementation. This chapter
focuses on implementation, but here the perspective is largely that of the kebele and the
household. It considers the following topics: Are local administrative structures such as
Kebele and Community Food Security Task Forces in existence and functioning as
envisaged in the Program Implementation Manual (PIM). Are households aware of these
Task Forces and do they understand their role in the implementation of the PSNP? How do
beneficiaries perceive their experiences with the payment process? Have client cards been
distributed?
Chapter 6: Targeting. This chapter covers the following topics: How is targeting supposed
to work in the PSNP? How is this understood and implemented at the regional level? How is
this understood and implemented at the woreda and kebele levels? (As part of this
discussion, it examines whether pregnant women are moved from public works to direct
support.) How is targeting understood at the household level? Who actually participates in
the PSNP? In public works and direct support? How consistent is this with the PIM? Do
households actually participate in the PSNP for three consecutive years? Does this
contribute to predictability in transfers?
Chapter 7: The Implementation of Public Works Projects. This chapter discusses a
number of aspects associated with the implementation of public works projects under the
PSNP. It explores whether there have been administrative improvements in the coordination
of public works activities planning and implementation. It assesses the extent to which those
activities that are chosen have incorporated feedback from communities and whether these
activities are linked to livelihoods. It summarizes perceptions regarding implementation
problems and the extent to which gender considerations have been incorporated into the
implementation of public works activities.
Chapter 8: Payments for Public Works and Direct Support. This chapter discusses a
number of aspects associated with payments for public works employment and direct
support. We provide information describing public works participation by gender, age, and
region. Based on self-reported payments data, we review the levels of transfers, the
frequency with which they are made, and the extent to which these reflect beneficiaries’
entitlements as outlined in the PIM. We examine transfer levels given to direct support
beneficiaries. We also present findings on household preferences for modality of payment
(food and cash) and whether these preferences are season or location dependent.
Chapter 9: Appeals and Complaints. This chapter presents the qualitative findings of the
appeals and complaints management of the PSNP in the highland areas of Ethiopia. It
describes the reasons for appeals, who people appeal to, how these are presented, whether
the appeals process is easy and transparent, and how long the appeals and complaint cases
take to be resolved. It also includes a discussion of the views and perception of the Woreda
and Kebele Food Security Taskforce on visits by the Rapid Response Team (RRT).

13
Chapter 10: The Household Asset Building Programme. This chapter presents an
overview of the Household Assets Building Programme (HABP), and explains the current
rollout progress, challenges encountered thus far, its linkage with other food security
programs, and its role in facilitating the graduation process, as well as the role of
development agents in implementing the HABP.
Chapter 11: Graduation. This chapter examines how graduation is understood and how it is
applied in practice, issues surrounding premature graduation, incentives and disincentives
for graduation, and support needed at graduation and postgraduation.

14
2. Data Sources and Methods
2.1. Introduction
There are three distinguishing features of the data sources and methods used in this report.
First, nearly all results are based on primary data collection undertaken between July and
August, 2010. Second, mixed methods—data collection techniques using both qualitative
and quantitative methods—have been employed. Doing so provides a richer pool of data
and greater analytic power than would have been available with either of these methods
used alone. Third, we adopt a “cascading” approach whereby data are collected at all levels:
federal, regional, woreda, kebele, household, and individual. Fourth, much of the quantitative
data at the household level is longitudinal—the same households have now been surveyed
three times, in 2006, 2008, and 2010.
The two qualitative methods used most frequently were focus groups and key informant
interviews. Key informant interviews were carried out at federal, regional, woreda, and
kebele levels. Focus groups were held at the kebele level. Quantitative surveys were
undertaken at the woreda level, in kebeles, and with households. We begin with a
description of the quantitative survey instruments and their implementation. We then
describe the qualitative fieldwork.
2.2. Quantitative Methods
The following quantitative survey instruments were fielded: a woreda level quantitative
capacity survey; a quantitative community survey; a community price questionnaire; and a
household survey instrument. We describe these in turn. Figure 2.1 shows the locations of
the quantitative survey.

15
Figure 2.1. Woredas surveyed in the quantitative surveys

16
The Woreda Quantitative Capacity Survey
A number of prior evaluations of the PSNP have raised concerns regarding the timeliness
and levels of payments received by beneficiaries. These studies, however, have not been
able to pinpoint why there are delays in payments. A novel component of this study is the
introduction of a woreda level quantitative capacity survey. The objective of this instrument
was to provide data that clarify how the flow of funds from regions to beneficiaries works in
practice. It included questions on
Staffing and resources (number of cashiers and accountants dedicated to the
PSNP; use of the automated payroll and attendance sheet system (PASS);
availability and training on computers; transport constraints, and so on);
Interface with the banking system (information exchange on availability of funds;
liquidity constraints at local bank branches);
Interactions between WOFED (Woreda Office of Finance and Economic
Development) staff and the Food Security desk;
Planning and management of cash flow;
Flow of funds—“following the money” from the region to the woreda to the
beneficiaries.
This survey instrument was implemented in all 85 woredas where the quantitative household
survey was implemented: 13 in Tigray; 30 in Amhara; 21 in Oromia; and 21 in Southern
Nations, Nationalities, and Peoples Region (SNNPR). Interviews were completed by
experienced survey supervisors who were instructed to meet with staff associated with the
Woreda Food Security Office as well as with those knowledgeable of the payment system.
Ideally, they were supposed to speak with the Head of the Food Security Office, the WOFED
chief accountant, the PSNP accountant, and the PSNP cashiers. On average, 3.9 persons
were interviewed per woreda located in Tigray; 4.9 in Amhara; 5.4 in Oromiya; and 5.1 in
SNNPR. Table 2.1 shows who was interviewed.
Table 2.1. Informants participating in the quantitative woreda survey (percentage)
Tigray
Amhara
Oromiya
SNNPR
All
Head, Woreda Food Security Office
76.9
86.7
85.7
90.5
85.9
WOFED Chief Accountant
38.4
43.3
90.5
52.4
56.5
Other member, Woreda Food Security Office
84.6
86.9
100.0
88.2
89.9
PSNP Accountant
53.8
93.3
71.4
95.2
82.4
PSNP cashier1
30.7
70.0
80.9
95.2
72.9
PSNP cashier2
23.0
56.7
57.1
52.4
50.6
Other
84.6
66.7
76.2
71.4
72.9
At least one member of the WOFED Food Security Office
100.0
100.0
100.0
100.0
100.0
PSNP accountant or cashier
61.5
93.3
90.5
100.0
89.4
Source: Authors’ calculations based on the PSNP survey data.
The principal reason why key individuals were not interviewed was that they were not
available for interview. In a few cases, interviews with key individuals did not occur because
the position was vacant.

17
Quantitative Community and Price Questionnaire
In this questionnaire, the community is defined as the kebele or peasant association.
Enumerators were instructed to interview at least five people, perhaps together, who are
knowledgeable about the community (e.g., community leaders, peasant association
chairmen, elders, priests, teachers). They must include at least one member of the Kebele
Food Security Task Force and at least one woman and they are told that they may need to
meet with other members of the Kebele Food Security Task Force in order to complete some
sections of this questionnaire.
The community questionnaire covers the following topics: location and access; water and
electricity; services; education and health facilities; production and marketing; migration;
wages; prices of foodgrains in the last year; operational aspects of the PSNP, including
questions about the operations of the FSTFs, public works (PW), and direct support (DP). In
addition, a price questionnaire obtains detailed information on current food prices.
The community questionnaire builds on similar instruments fielded during the 2006 and 2008
evaluations. Two broad sets of changes were introduced in 2010. First, there was a need to
revise this instrument to include more information on operational aspects of the HABP (for
example, presence and skill-sets of development agent staff; presence and access to
financial institutions), a more nuanced understanding of beneficiary selection into direct
support, and decisions regarding allocation of labor days under the public works component.
Second, the 2006 and 2008 evaluations made relatively little use of background information
on these communities that had been collected. To ensure that the community questionnaire
was not longer than the one fielded in previous years, questions that had not proved
informative in past years were carefully reviewed and a number of them dropped.
The Quantitative Household Survey and Questionnaire
The design of the first Food Security Survey sample, fielded in 2006, was based on power
calculations conducted to determine the minimum number of sample enumeration areas and
households needed to be able to identify impacts of the Food Security Program.
3
We used
the share of chronically food-insecure (CFI) households as the outcome for the power
calculations because this is the primary targeting criterion for the program and because FSP
documents identify reducing the number of CFI households as a major goal of the program.
According to the PSNP Implementation Manual (GFDRE 2004, 4), a household is
considered CFI if it had three or more months of unmet food needs per year in each of the
past three years.
We clustered the sample at the woreda level, the administrative unit at which program
participation is assigned. Based on discussions with the Central Statistical Agency (CSA),
we assumed the sample design would include two kebeles or enumeration areas (EAs) per
woreda in Amhara, Oromiyia, and SNNPR, and three EAs per woreda in Tigray. We also
assumed 25 households would be sampled in each EA. Using 50 households per woreda as
the desired cluster size, we calculated the number of clusters needed to obtain the desired
level of statistical power, using the Optimal Design software available at
http://sitemaker.umich.edu/group-based/optimal_design_software. Treating “success” as the
3
See Gilligan et al. (2007) for a complete description of the sample and 2006 survey.

18
absence of chronic food insecurity, we assumed initially that 30 percent of the sample was
not chronically food insecure. We assumed that the sample size should be large enough to
identify an effect size equivalent to a 10-percentage-point increase in non-CFI; that is,
raising the proportion of households that were not food insecure to 40 percent. Seeking
statistical power of 80 percent and a significance level of 0.05, we found that 62 sample
clusters would be required. To account for additional sampling of kebele subclusters within
the EA and unbalanced samples of beneficiaries and nonbeneficiaries, it was decided to be
conservative and include 68 woredas as sample clusters.
Woredas were randomly sampled proportional to size (PPS) from a list of 153 chronically
food-insecure woredas (excluding the sample surveyed for USAID), stratified by region.
Within each woreda, sample kebeles serving as EAs were randomly selected from a list of
kebeles with active Productive Safety Net Programs. Within each EA, 15 beneficiary and 10
non-beneficiary households were sampled from separate lists for each group, yielding a
sample of 25 households per EA. This procedure yielded the following sample:
Table 2.2. Number of woredas, kebeles, and households sampled, by region
Region
Number of
woredas
Number of
kebeles/woreda
Number of households per
kebele
Number of households
per region
Amhara
18
2
25
900
Oromiya
19
2
25
950
SNNPR
19
2
25
950
Tigray
12
3
25
900
Total
68
3,700
Source: Authors’ calculations based on the PSNP survey data.
In 2005, a survey was conducted to study the impact of the PSNP in areas where USAID
supported the provision of the PSNP through a High Value Food Basket (HVFB). Most of the
sample for this survey covered woredas in Amhara, where USAID had its highest
concentration of PSNP-related activities, although woredas in other regions were also
included in the sample. It was decided to add part of this sample to the data collection for the
second round of the Food Security Survey fielded in 2008. This would make it possible to
compare beneficiaries in these “HVFB woredas” (as we refer to them) with beneficiaries
served solely by the government of Ethiopia. Because there may be differences in intensity
of contact between beneficiaries in these woredas and beneficiaries served by other
executing agencies, this comparison improves understanding of whether more intensive
interactions generate larger impacts.
In order to avoid having a very small sample of woredas in some regions and to control
survey costs and logistical difficulties, we decided only to revisit the sample of households
from the 2005 survey of the woredas in Amhara. There is a large sample of households
there that would permit a careful comparison with other households from Amhara in the
Food Security Survey (FSS) sample from 2006. Also, although the survey instrument used in
the 2005 survey of NGO woredas differed from that used in the 2006 Food Security Survey,
there is enough similar information to assist the analysis across the two samples. Moreover,
the questionnaire for the 2008 FSS was designed with both baseline survey instruments in
mind, and the same questionnaire was implemented in both samples in 2008, facilitating
comparison of outcomes in this year.

19
Before finalizing fieldwork arrangements for the 2008 survey round, we investigated the
statistical properties of the sample of HVFB woredas and other woredas in Amhara from the
2006 FSS to be sure that the sample would be large enough to be able to identify
differences in impact between PSNP beneficiaries in these two groups. We sought a sample
size that gives an 80-percent chance (the power of the test) of identifying a reasonable
expected difference in outcomes between HVFB and government woredas as statistically
significant at the 5 percent level. In both samples, sampling involved selection of beneficiary
woredas and then selection of EAs, which are selected kebeles or subregions of kebeles.
For these power calculations, we assumed clustering at the level of the EA because intra-
cluster correlations are typically higher at the EA level than those that are at the woreda
level.
We considered the power of the sample if the survey was fielded in all 11 HVFB woredas in
Amhara and assuming we surveyed three EAs within each woreda. In each EA, we assumed
we would survey 17 beneficiary households, yielding a sample of 561 households (11
woredas x 3 EAs per woreda x 17 households per EA). We determined that this would
provide sufficient statistical power to detect the following differences in outcomes between
woredas with NGO- versus government-run programs:
A 30-percent difference in per capita consumption expenditure.
A 50-percent difference in the size of the food gap.
A 25-percent difference in per capita caloric availability.
A 35-percent difference in the value of livestock holdings.
In fact, the Amhara HVFB sample should have somewhat greater statistical power than
described above, because the 2008 survey included four EAs rather than three in each of
the 11 woredas being surveyed and each of these EAs included 28 households, as
summarized in Table 2.3. A few EAs had one or two more households than the average of
28, yielding a total sample size of 1,237 households.
Preparations for the 2010 survey were guided by the Terms of Reference (TOR) for this
study. These made explicit reference to the design of the quantitative household
questionnaire. Paragraph 22 of the TOR stated:
The format, structure, and questions should, as much as possible, be identical to
those used in the 2006 and 2008 surveys. This will maximize comparability
across survey rounds. There is no need to repeat questions on individual or
household characteristics that do not change over time. All survey instruments
should be carefully reviewed, and modules added as required, so as to ensure
that the issues described under the Methodology section below are adequately
captured.
Table 2.3 describes the structure of the 2010 quantitative household questionnaire and how
it had been changed.

20
Table 2.3. Design of the 2010 household questionnaire
Section
Module
Number
Heading
Comments
1: Basic household
characteristics
1A
Household demographics, current
household members
No change
1B
Characteristics of the household
No change
1C
Former household members
No change
1D
Children’s education and labor
No change
2. Land and crop
production
1
Land characteristics and tenure
No change
2
Input use and crop production
Edited to ensure consistency with HABP
log frame
3
Disposition of production
No change
4
Use of household labor in crop
production
No change
3. Assets
1
Production, durables
No change
2
Housing
No change
3
Livestock ownership
No change
4
Income from livestock
No change
5
Distress sales
No change
4. Nonagricultural
income and
credit
1
Wage employment
Edited to ensure consistency with HABP
log frame
2
Own business activities
Edited to ensure consistency with HABP
log frame
3
Transfers
No change
4
Credit
Coding edited to ensure consistency with
HABP
5. Access to the
PSNP and HABP
1
Access to the PSNP—public works
Introductory questions revised
2
Access to the PSNP—direct support
Editing changes
3
Access to the HABP
MAJOR revision (previously, this section
focused on the OFSP)
4
Perceptions of benefits of assets
created by the PSNP
Review further
5
Perceptions of operations of the
PSNP
Review further (e.g., add questions on
knowledge of when payments are
coming) questions added on client cards
6. Consumption
1
Expenditure on durables and
services
No change
2
Expenditure on consumables
No change
3
Food consumption
No change
4
Food availability, access, and coping
strategies
No change
7. Health, shocks
and perceptions
1
Long term shocks
Shortened, with questions on health
status and illness (had been separate
sections)
2
Recent shocks to crops and livestock
No change
3
Poverty perceptions
No change
8. Anthropometry
1
Height, weight of children 6m to 7y
No change
2
Access to water and sanitation, child
feeding, women’s perceptions
Some questions from Module 7, Section 3
of 2008 questionnaire now included here
Source: Authors’ compilation.
All three quantitative survey rounds have been fielded at approximately the same time. The
first round was fielded in July and early August, 2006. The second round was fielded

21
between late May and early July, 2008 and the third round in June and July, 2010.
Consequently, seasonality considerations are unlikely to confound comparisons made
across rounds.
Discussions surrounding the design of the 2010 quantitative survey instruments began with
a series of dialogues involving the Food Security Coordination Directorate (FSCD), members
of the Donor Working Group, and the CSA. Following the conclusion to these, a CSA
technical team, formed specifically to coordinate the survey, began work with our team.
Tasks completed included:
1. Translation of survey instruments into Amharic. The process involved an intensive
question-by-question discussion and some modifications. The fidelity of the
translation to the original was thereby assured.
2. The preparation of the Enumerator’s Manuals (one each for the community
questionnaire and the household questionnaire) went hand-in-hand with the
translation of the questionnaire. The draft manuals were modified and finalized
through the simultaneous discussion.
3. Training of trainers followed the completion of both the survey instruments and the
manuals. It took place at CSA’s headquarters in Addis Ababa during the second half
of May, 2010.
4. Training of enumerators was subsequently conducted by trainers from the zonal CSA
offices with the help of the technical team from the CSA head office.
5. The administration of the survey followed. Enumeration and data collection took
place during June and July 2010. A team of IFPRI-supported supervisors (including
members of the IFPRI survey team) participated in the fieldwork, providing in-situ
technical support.
The 2006 survey generated data on 3,688 households in 148 EAs within 68 woredas. In
2008, the CSA enumerator assigned to each EA was provided with the list of households
interviewed during 2006 and used these along with (in some cases) maps and assistance
from local officials and residents in locating households for reinterview. Attrition was low.
Only 137 households (or 3.7 percent of the baseline sample) were not reinterviewed during
the 2008 survey. About a third of these households are from two EAs within Oromiya, where
the survey could not take place. The 2008 resurvey covered 3,551 households in 146 EAs
within 68 woredas.
As discussed above, in 2008, CSA also resurveyed households that where covered by a
USAID-organized baseline survey in 2005. That baseline sampled 41 woredas in which the
implementation of the PSNP was facilitated by eight NGOs that, in turn, were supported by
USAID/Ethiopia. From among these, the 2008 survey collected data from the Amhara
subsample alone. Again, attrition was rather small (69 households or 5.6 percent of the
original subsample). Consequently, data were collected from 1,167 households within 44
kebeles located in 11 woredas. In total, the 2008 survey generated data on 4,718
households.
Data on interviews conducted in 2010 are reported in Table 2.4. There were 3,366
households interviewed who form the 2006–2008–2010 panel. Across all three rounds,

22
3,140 households appear in all rounds, yielding an attrition rate of 14.8 percent or, over five
years, just under 3 percent. In the HVFB woredas, of 1,297 households sampled in 2005,
1,137 were interviewed in 2008 and 1,146 households were interviewed in 2010. The
effective sample of households for analysis in the impact report will be those households for
which we have baseline household characteristics. We have this information for 3,038
households across all three rounds.
Table 2.4. Sample numbers by round
2005
2006
2008
2010
All three
rounds
Number of households in the 2006–2008–2010 panel
-
3,688
3,288
3,366
3,140
Attrition rate
-
-
10.8%
8.7%
14.8%
Number of households from HVFB woredas
1,297
-
1,137
1,146
0
Number of FSS households that we have full range of
baseline characteristics for (overlap with later rounds)
3,475
3,190
3,193
3,038
Attrition rate
8.2%
8.11%
12.57%
Source: Authors’ calculations based on the PSNP survey data.
Table 2.5 presents the attrition rate by region. There is some regional variation where
households in Tigray and SNNPR are less likely to leave the sample across the three rounds
compared to Amhara and Oromiya.
Table 2.5. Attrition by region
2006
2008
Attrition rate
between
2006–08
2010
Attrition rate
between
2006–10
Panel household
(across all three
rounds)
Attrition rate
across all
three rounds
(percent)
(percent)
(percent)
Whole sample
3,475
3,190
8.2
3,193
8.1
3,038
12.6
Tigray
843
807
4.3
776
7.9
770
8.7
Amhara
806
703
12.8
742
7.9
665
17.5
Oromiya
921
813
11.7
828
10.1
770
16.4
SNNPR
905
867
4.2
847
6.4
833
8.0
Source: Authors’ calculations based on the PSNP survey data.
Lastly, we investigate whether potential differences in attrition rates can be attributed to
differences in baseline characteristics. In particular, we would like to know whether
households that are beneficiaries under the FSP are more likely to stay in the sample than
their neighbors. We examine the correlation of the probability of attrition with household
characteristics and region dummies. This is presented in Table 2.6 and shows that being a
beneficiary is not highly correlated with the probability of attrition. (This was also true when
we analyzed attrition between 2006 and 2008.) The only coefficients that are significant are
those that relate to the age of the household head and household composition, and the
magnitude of these is small.

23
Table 2.6. Probit showing marginal effects of the factors affecting the probability of
attrition
Factor
Effect
PSNP beneficiary in 2006
-0.008
(0.016)
Age of household head
-0.001**
(0.001)
Gender of household head (1 = Male, 0 = Female)
-0.021
(0.018)
Highest grade obtained by household head
0.000
(0.001)
Number of male household members aged 0-6 years
-0.019**
(0.009)
Number of male household members aged 7-15 years
-0.008
(0.007)
Number of male household members aged 16-60 years
-0.028***
(0.010)
Number of male household members aged > 60 years
-0.004
(0.028)
Number of female household members aged 0-6 years
-0.009
(0.007)
Number of female household members aged 7-15 years
-0.009
(0.007)
Number of female household members aged 16-60 years
-0.009
(0.010)
Number of female household members aged > 60 years
-0.029
(0.020)
Value of production assets owned in 2006
-0.000
(0.000)
Value of livestock owned in 2006
0.000
(0.000)
Land area owned in 2006
0.007
(0.008)
Amhara Region Dummy
0.085
(0.071)
Oromiya Region Dummy
0.084
(0.052)
SNNPR Region Dummy
-0.005
(0.028)
Observations
3,153
Source: Authors’ calculations based on the PSNP survey data.
Note: Standard errors in parentheses clustered at woreda level. * significant at 10 percent, ** significant at 5 percent, ***
significant at 1 percent.

24
2.3. Qualitative Methods
Site Selection
Ten woredas were selected for in-depth qualitative work. These were allocated across the
four regions as follows: Amhara (3), Tigray (2), Oromiya (2), SNNPR (3). These were chosen
based on the following criteria:
1. Representation of livelihood zones
2. Comparability of similar livelihoods across regions
3. Coincidence with quantitative household survey work
4. Extent of graduation (indicated by administrative data)
5. At least one woreda will be a locality where the FSP is supported by USAID.
Table 2.7 lists the woredas included in the qualitative studies. Annex B provides a
description of these.
Table 2.7. Location of qualitative studies
Region
Zone
Woreda
Kebele
Implementation arrangement
Amhara
South Gondar
Ebinate
Serawudi
Government
South Wollo
Sayinte
Shengodefer
Government
Waghimra
Sekota
Wal
NGO
Oromiya
East Hararge
Gursum
Barite
Government
East Arsi
Zeway Dugda
Dimitu Raretti
Government
SNNPR
Gomogoffa
Demba Goffa
Gurade
Government
Sidama
Shebedino
Hula
Government
Kembata
Tembaro
Soyame
Government
Tigray
Central Tigray
Aferom
Adizata
NGO
Eastern Tigray
Saesi Tseda
Sendeda
Government
Source: Authors’ calculations based on the PSNP survey data.
Once sample woredas were identified, the field research teams were responsible to select
study kebeles through discussions with woreda officials. The teams selected kebeles with a
high number of PSNP graduate households, since an important objective of the evaluation
was to better understand the situation of PSNP graduates. In places where graduation was
not exercised, kebeles with large PSNP beneficiaries were taken into account. Moreover,
since the time of this study coincided with the main rainy season, accessibility of kebeles
was considered as a secondary criterion for selecting study kebeles.
Key informant interviews were held at the federal level and at all four regional levels. These
were designed to deepen contextual understanding of implementation and outcomes (such
as the use of contingency funds, policies regarding the allocation of PSNP funds to the direct
support component, allocation of funds between the HABP and CCI) as well as learn from
institutional perspectives across different levels of administrative input and responsibility. At
the woreda, kebele, and community level, a “cascading” interview approach was used
involving interviews with officials and knowledgeable observers to build up a detailed
understanding of issues around implementation of PSNP and HABP and food security
outcomes. Key informant interviews were also used to triangulate findings from the focus

25
group discussions as well as to add nuance and texture to these. A structured interview
format was used to obtain information on a range of issues, including targeting procedures,
appeals and grievances processes, and gender representation at local level.
Table 2.8 describes the identity of individuals participating in key informant interviews.
Table 2.8. Key informant interviews
Level
Number of
interviews
Who was interviewed
Regional
2-3
One interview with the chair or a member of the Food Security Task Force. Two
Interviews with a member (or more) of the PW committee, or HABP committee or
EW committee.
If committees had not yet been formed, two interviews with other relevant people
(e.g., the PW focal person/or the OFSP person, or a direct support person).
Woreda
3
One interview with the chair or a member of the Woreda Food Security Task Force.
Two Interviews with a member (or more) of the PW committee, or HABP committee
or EW committee.
If the committees had not yet been formed, two interviews with other relevant
people (e.g., the PW focal person/or the OFSP person, or a direct support person).
Kebele
2-3
One interview with the Kebele Food Security Task Force (this could have multiple
people in the group). One or two interviews with development agents.
Source: Authors’ compilation.
Focus group discussions were held at the kebele level in all regions. Four types of focus
groups were interviewed.
1. Chronically dependent households: The main purpose of this group is to investigate
issues around direct support. Topics covered will include access constraints to the
PSNP, targeting experience, dependency, and graduation trajectories; shifts between
public works and direct support; labor constraints, labor caps, and full household
targeting; dilution and reciprocity. These groups usually consisted of six individuals
from “permanent” direct support households (such as those containing elderly,
disabled, or chronically sick persons), two participants from non-beneficiary
households (i.e., were chronically poor and should be receiving direct support), and
two participants from “temporary” direct support households.
2. Transitioning households: The main purpose of this group is to understand better the
linkages between PSNP and OFSP/HABP programs. We purposely selected a focus
group composition that may allow us to understand the interactions between PSNP
and other livelihood programs. Our aim was to interrogate the conditions by which the
activities related to the “livelihood activities” are able to strengthen livelihoods and
pathways of transition off the safety net. These groups typically included three PSNP
public works households, two PSNP households who received HABP/OFSP support,
three PSNP graduates, and two participants from non-eligible households (i.e., those
who are “better-off” according to locally-specific criteria).
3. Women’s group: In this group, we are interested in understanding a range of issues
related to (1) the gendered use of and access to PSNP cash and in-kind transfers; (2)
the gendered labor requirements of the public works and whether there are labor
substitution effects from the public works labor requirements; (3) perceived effects on

26
child outcomes in relation to education and feeding in relation to payment type; (4)
reasons for preferences concerning payment type; (5) whether access opportunities
and constraints to PSNP participation at all levels are gendered; and (6) the impact of
program participation on women’s welfare in terms of access to households resources
and gender relations inside the household and in the community. Women’s groups
consisted of two participants from female-headed households who receive temporary
direct support, two participants from female-headed households who do public works,
two participants from female-headed households who are HABP/OFSP clients, and
four participants from male-headed households who are PSNP/HABP/OFSP clients.
4. Men’s group: The discussion in this group will mirror the one held in the women’s
group. Men’s groups usually consisted of three men who were recipients of direct
support, three who were active in public works projects, and three who received
HABP/OSFP support.
Discussion in focus groups covered the following topics, although not all topics were covered
in all groups:
Gender/intrahousehold dynamics: What is the gender division of labor on public
works? Who collects the payment for public works and direct support? Who decides the
use of cash and food? What are preferences for food or cash between the sexes? What
is the experience of female-headed households? In polygamous households, are there
tensions between wives? Are there instances where not all wives from the same
polygamous household are included?
Institutional access constraints: What difficulties do households face when applying
for/collecting their PSNP payments (administration, information, physical, cultural,
discrimination constraints)? Does this differ between the type of payment received (food
or cash)? Does this differ between who applies for/collects the payments? In regard to
credit access/OSFP/HABP, are people aware of the procedures for applications and the
institutions responsible? What are the perceived and actual difficulties in accessing these
services?
Appeals and grievances: What is the composition of the Appeals Committee? Who
decides who is on the committee? Are complaints handled in a satisfactory manner? Do
people know how to appeal? If they choose not to appeal, why not? Are people’s voices
heard through existing institutional arrangements? Are there differences between men
and women concerning their voice being heard in appeals processes?
Targeting and graduation procedures: What is the composition of the Kebele Food
Security Task Force? What is the level of women’s participation (e.g., are women
chairs?)? Are women’s voices heard in any meaningful way? Is targeting done in a
transparent and fair way? Are exclusion errors perceived to be widespread? If such
exclusion errors exist, who is left out and why? How prevalent are exclusion errors in
targeting? Are households being graduated too early? What are the reasons for this
happening? Have households become overly dependent on PSNP support?
Household Asset Building Program/Other Food Security Programs: How do people
access credit and the processes through which this happens? Is HABP/OFSP demand-

27
driven? Are PSNP graduate households seeking credit through the HABP or other
existing sources? How are the kebele and woreda level FSTFs involved in the selection
process of HABP beneficiaries? How will business plans be evaluated? What is the
technical capacity of development agents to provide training and ongoing support for
HABP beneficiaries? How gender sensitive is the HABP? Given the direct support
beneficiaries of the PSNP are mainly women-headed, elderly, and disabled households,
how will the HABP be fine-tuned to target these groups?
Direct support: Are there instances of chronically food-insecure households that do not
receive direct support? What is the understanding of direct support guidelines on the
ground? Are households eligible for direct support being covered under public works?
How widespread is this?
Public works: What are household experiences of meeting their obligations to
contribute labor to public works as a condition for receiving PSNP support? Who in the
household contributes labor? What is the gender division of labor on public works? Who
carries the extra work burdens? What impact do work requirements have on primary
school enrolment? Are parents taking their children out of school to meet their labor
obligations? What are perceptions of the effectiveness of public works? In pastoralist
contexts, what is the appropriateness of public works for making stronger pastoralist
livelihoods?
Transfer dilution: What is the importance of informal social support networks alongside
PSNP transfers? What proportion of cash/food transfers do clients share, and with
whom? What is the significance of these transfers within relations of reciprocity and
interdependencies?
In addition to notes taking, these interviews were tape recorded and subsequently
transcribed and coded. In the analysis presented in subsequent chapters, qualitative quotes
are used when they are representative of broader trends.

28
3. Food Security, Assets, and Coping Strategies
3.1. Introduction
In this chapter we present some statistics that describe the context in which the FSP
operated and trends in outcomes of interest. In particular, we examine the price changes of
main staple food crops, livestock, and labor over the period 2006–2010. We also use
information from the survey to examine the extent of shocks experienced by households.
Finally, we examine changes in asset levels, food security, and coping strategies; and
subjective measures of well-being.
3.2. Context
Like the rest of the world, Ethiopia also experienced a dramatic rise in food prices in 2007–
08. This had implications for purchasing power of wages as well as food security among
households that are net buyers of food. Figures 3.1a–3.1d show the change in main staple
food crops prices, livestock prices, and wages between 2006–08 and 2008–10 by region.
As one would expect, there were significant increases in food prices in 2006–08 but not so
much in 2008–10. A point to note is that, even though the price increase in the last two years
is not as severe as that seen during the food price crisis, there is still quite a significant
upward trend (outside of SNNPR). Livestock prices also rose significantly during 2006–08 in
all four regions, with the highest increase seen in oxen prices. In the subsequent period,
there was not much change in the oxen and cow prices in Tigray and Amhara. However,
livestock prices continue to rise in Oromiya and SNNPR.
The labor market response to the sharp increase in food prices in 2006–08 is quite strong in
Tigray and Amhara—where the wages almost doubled over the period. This is not true for
Oromiya and SNNPR, even though these two regions were hit worse by the crisis. Livestock
prices also rose at a slower pace in these two regions compared to Tigray and Amhara. In
light of the analysis, it seems households in Oromiya and SNNPR have been more
vulnerable than those in Tigray and Amhara. We next examine regional differences in
incidence of various shocks experienced at the household level.
Figure 3.1a. Price changes, Tigray, 2006–08 and 2008–10
Source: Authors’ calculations based on the PSNP survey data.

29
Figure 3.1b. Price changes, Amhara, 2006–08 and 2008–10
Source: Authors’ calculations based on the PSNP survey data.
Figure 3.1c. Price changes, Oromiya, 2006–08 and 2008–10
Source: Authors’ calculations based on the PSNP survey data.

30
Figure 3.1d. Price changes, SNNPR, 2006–08 and 2008–10
Source: Authors’ calculations based on the PSNP survey data.
3.3. Incidence of Shocks
One of the goals of the FSP is to protect households from shocks. In Figure 3.2 we assess
the prevalence of shocks across the three survey rounds. In general, incidence levels of
various shocks in 2010 are almost as high, if not higher, as previous years. In particular, we
observe that a much larger fraction of households experience weather-related shocks, such
as drought, flood, and erosion. We also observe a 9 percent increase in the fraction of
households that did not have access to input markets. There is also a sharp increase in the
fraction of households that experienced illness (up from 7 percent in 2008 to 17 percent in
2010).
Figure 3.2. Incidence of shocks
Source: Authors’ calculations based on the PSNP survey data.

31
Figures 3.3a–3.3e present the incidence of these shocks by regions and show that there are
a lot of regional variations in the occurrence of various shocks. In particular, even though
drought remains the most important shock in the Amhara region, its extent is less than in the
other three regions. A much larger fraction of households in Tigray and Oromiya
experienced a drought shock in 2010 compared to 2008, whereas this fraction declined
slightly in Amhara and SNNPR. The Amhara-HVFB sample is different from the Amhara FSP
sample, where a large fraction of households (88 percent) experienced a drought shock in
2010. Households in SNNPR are more likely to experience flood compared to other regions
and the fraction of households affected more than doubled since 2008, as did the fraction of
households that experienced losses due to erosion. Lack of access to input markets and
increases in input prices are more prevalent in SNNPR.
Figure 3.3a. Incidence of shocks, Tigray
Source: Authors’ calculations based on the PSNP survey data.
Figure 3.3b. Incidence of shocks, Amhara
Source: Authors’ calculations based on the PSNP survey data.

32
Figure 3.3c. Incidence of shocks, Amhara-HVFB
Source: Authors’ calculations based on the PSNP survey data.
Figure 3.3d. Incidence of shocks, Oromiya
Source: Authors’ calculations based on the PSNP survey data.
Figure 3.3e. Incidence of shocks, SNNPR
Source: Authors’ calculations based on the PSNP survey data.
In terms of idiosyncratic shocks like death and illness, Tigray, even though the fraction of
households that were affected in any year is less than 10 percent, has experienced a decline
in the incidence of such shocks. All other regions show an increase in the fraction of
households that reported experiencing these shocks in 2010 compared to 2008. This
increase is the largest in SNNPR, where the fraction of households that experienced an

33
illness in the household went up from 10 percent to more than 30 percent. This sharp
increase is always driving the average results for the entire sample.
Next we examine whether experiencing various shocks differs by beneficiary status under
the FSP. We compare households that receive payments under the public works component
of the FSP in each year with households that were not part of the FSP at all. Figure 3.4 gives
this information. A quick look at these graphs shows that in each year, the average
experience of public works beneficiaries and non-beneficiaries is similar. The non-
beneficiary households are slightly less likely to report experiencing a drought shock. This
may seem a bit inconsistent with the fact that a drought is a covariate shock. But when we
combine it with the capability to cope with drought, this makes more sense, because non-
beneficiary households, which by definition are better off compared to public works
beneficiaries, are better placed to protect themselves.

34
Figure 3.4. Incidence of shocks, by public works (PW) beneficiary status and year
Source: Authors’ calculations based on the PSNP survey data.
We asked households in the survey about consequences of shocks experienced. Figure 3.5
gives the response of households for having experienced a drought. A large fraction of

35
households have to incur a loss in income or consumption as a result of a drought shock.
This shows that a drought not only has negative implications for household income, it also
adversely affects household consumption. About 10 percent of households affected by a
drought experienced a loss of productive assets. The pattern of responses has remained
more or less constant over the period between 2006 and 2010.
Figure 3.5. Consequences of experiencing a drought, all households
Source: Authors’ calculations based on the PSNP survey data.
3.4. Food Security
An important objective of the FSP is to enable households to smooth their consumption by
bridging periodical food gaps and eventually reduce food insecurity. In this section, we
assess the food security situation of households by examining the food availability, food
gaps, and coping strategies of households observed over the last three rounds of the survey.
We begin by analyzing the main sources of food available to households. This is essential to
understand the patterns and trends of the sources of food consumption as well as
understand how households have fared in their food security situation over the last few
years.
Food Availability
To assess the above, households were asked to indicate which sources contributed primarily
to their food consumption over the 12 months before each survey. Figure 3.6 summarizes
primary income sources by month. The figure shows that households report consistent
monthly patterns of the composition of their primary food sources over the three years
observed. Particularly, for households that participate in the food security program, the
primary food source comprises own production, purchases, and the food security program
(public works and direct support) itself. Note that the patterns of the compositions of these
primary food sources clearly follow the annual food production-food availability cycle of a
typical Ethiopian rural household. A striking observation from this figure is that the proportion
of primary food sources contributed from own production declines, while the portion
contributed by the PSNP increases over the three survey years. Clearly, comparing over the

36
three years, there is a decline in the proportion of households who reported own production
as their primary sources, which would mean an increased need for food purchases. While
these patterns suggest the food security program has become increasingly important for the
PSNP areas over the three years considered, it would be important to investigate these
relationships further using more rigorous methods.
Figure 3.6 raises several empirical questions. What happened to these households over the
years? What does this picture look like for the average household and for how many months
in a year do households fail to fulfill their food requirements from their own production. Can
shocks partly explain the declining trend in the contribution of own production to primary food
sources? Figure 3.7 helps to answer some of these questions by giving the average number
of months households reported to have run out of home-grown food over the year covered
by each survey and whether or not households suffered drought shocks in that particular
year.
Figure 3.7 shows that, on average, households had difficulty satisfying their food needs from
all production for 3.6 months in a year in 2006. The number fell to 2.2 in 2008 and remained
stable. For households that did not experience a drought, the average number of months
households fail to fulfill their food requirements from home-grown sources has declined from
a little more than 2.5 months in 2006 to about 1.5 months in 2010. The average has declined
even more dramatically for households that faced the shock (from about 4 months in 2006 to
about 2.5 months in 2010), but still remains higher than for those that did not face a drought.
This picture remains fairly similar even when disaggregated by PSNP status of households
(Figure 3.8).

37
Figure 3.6. Primary source of food, by month and year
Source: Authors’ calculations based on the PSNP survey data.

38
Figure 3.7. Average number of months households run out of home-grown food. by
drought conditions
Source: Authors’ calculations based on the PSNP survey data.
Figure 3.8. Average number of months households run out of home-grown food, by
drought conditions and PSNP beneficiary status
Source: Authors’ calculations based on the PSNP survey data.
This result seems rather contradicting to the previous result reported from Figure 3.6 that the
percentage of households that reported their own production contribution consistently
declined from 2006 to 2010. This may not necessarily contradict to the result here, however,
because the average number of food shortage months from home-grown food could be
declining while, at the same time, the number of people reporting declines in own production
are increasing.
Household Food Gaps
It is now common knowledge that households face severe food shortages around the rainy
and planting season in Ethiopia. To assess this, households were asked if they suffered any
food shortage in the rainy season, which is summarized in Figure 3.9. It can be observed
from the figure that the rainy season food gap situations vary across regions and over the

39
three years considered. Generally, Oromiya and SNNPR are among the regions with highest
proportion (a three-year average of above 25 percent) of households reporting food gaps in
the rainy season. However, comparing 2006 and 2010, the proportion of households facing
food gaps seem to be decreasing for Amhara and SNNPR, while it increases for Tigray and
Oromiya.
Figure 3.9. Percent of households that suffered any food shortages in the rainy
season, by region and year
Source: Authors’ calculations based on the PSNP survey data.
To get a sense of the bigger picture, we assess the number of months households face
difficulty fulfilling their household food requirements throughout the year.
4
Figure 3.10
presents the number of months in a year households were unable to satisfy their food needs.
The proportion of households that reported to have problems of satisfying their food needs
for a longer period of the year has increased over the years. There is a dramatic increase in
the percentage of households that reported having food gaps for the most part of the year in
the later years (2008 and 2010). There is a marginal improvement in 2010 over 2008 in
terms of the percentages of households facing above 10 months of food gaps. For example,
close to 15 percent of households reported they had food shortages for the whole year in
2008, while only a little more than 10 percent of households indicated this in 2010.
4
A household is considered as food insecure in a given month if the household was unable to satisfy its food needs for at least
five days in the month.

40
Figure 3.10. Percentage of households unable to satisfy food needs of the household
Source: Authors’ calculations based on the PSNP survey data.
To have a sense of which part of the year households are most food-insecure, households
were asked in which specific month over a period of one year the food shortage was most
acute. The distribution of the most severe food shortage month in 2010 by region is given in
Figure 3.11. Clearly, for the majority of households, the most food-insecure months fall
between February and September. Differences across regions, mainly due to seasonal
differences, are also visible here. While the northern regions (Amhara and Tigray) had their
severe food gaps in the rainy season and just before harvest (June, July, and September),
this situation occurs in SNNPR and Oromiya much earlier than the north, around March and
August, respectively. In particular, households from Amhara and Tigray suffered the most
acute food gap in September, while households from SNNPR and Oromiya suffered in May
and July, respectively. This situation did not improve much over the years. In fact, for Tigray
and Amhara regions, the severe food-gap month has moved from August (in 2006 and 2008)
to September (in 2010). Besides, the proportion of households reporting the occurrence of
the severe month has slightly increased from a little less than 30 percent in 2006 to a little
more than 40 percent in 2010.

41
Figure 3.11. Most severe food gap months, in 2010, 2008, and 2006
Source: Authors’ calculations based on the PSNP survey data.
A sensible question is whether or not the food gap situation varies by households’
participation status in the food security program (both public works and direct support). We
break down the distribution of the most acute food-gap month into PSNP and non-PSNP
household groups. This is summarized in Figure 3.12. Once again, we observe that the
months with the most acute food gap span between February and September, regardless of
PSNP status. Perhaps the most important observation from these graphs is that the severity
of the food insecurity has declined for the PSNP households between March and June in
2010 compared to the same period of 2006 and 2008. This may be attributed to the food
availability through the food security program, which, for the most part, corresponds to these
same months.
However, although relatively better off in the dry (PSNP) season, PSNP households seem to
suffer more in the end of the rainy season in 2010, particularly in September as compared
with other years. About 25 percent of the PSNP (compared to less than 15 percent of the
non-PSNP) households reported that September is the most severe food-insecure month.
Note that we have observed earlier that September is peculiarly the most severe food-
insecure month for households in Amhara and Tigray regions.

42
Figure 3.12. Most severe food shortage month for PSNP and non-PSNP households
Source: Authors’ calculations based on the PSNP survey data.
Note: PW = public works; DS = direct support.
Number of Meals
Households often adopt negative coping strategies (e.g., reduce consumption) when falling
short of meeting their daily food requirements from own or external sources. Assessing the
ability of the household to consume the ideal number of meals per day household members
eat in a normal day would thus provide a self-assessed picture of household food security.
For this purpose, households were asked to report the number of meals adult and child
members of their household eat per day on a normal and worst food-insecure day. The
results are summarized in the following graphs.
Figures 3.13a–3.13c provide the number of meals per day adult and children members eat
during a normal and worst food gap month. The number of meals adults and children eat in a
normal month is given in Figure 3.13a and 3.13b for comparison purposes. First we note that
the majority (about 75 percent) of households in all the three years reported adults eat three

43
times a day during months with no food shortages. Similarly, with the exception of 2010,
where about 40 percent of households report children eat about 5 times a day, the majority
(about 38–45 percent) of households reported children eat 3–4 meals a day in months with
no food shortages.
Figure 3.13a. Number of meals per day adults eat (no food shortage)
Source: Authors’ calculations based on the PSNP survey data.
Figure 3.13b. Number of meals per day children eat (no food shortage)
Source: Authors’ calculations based on the PSNP survey data.
Figure 3.13c. Number of meals per day adults eat (during food shortage)
Source: Authors’ calculations based on the PSNP survey data.

44
With this reference in mind, we look at the number of meals adults eat per day in food
shortage months and a considerable number of households reported adults eat two times a
day. This proportion slightly declines from 2006 to 2010. In contrast, the proportion of
households that reported one meal a day increases steadily over the same period, with
PSNP households reporting a slightly higher proportion in each year compared to non-PSNP
households. The proportion of households that reported adults eating three meals a day is
very small and seems to decline from 2006 to 2010 across both groups of households. Thus,
in contrast to the no food shortage situation, households in a food shortage situation eat
fewer meals, and this seems to deteriorate over the years among the PSNP households.
In Figure 3.14a–3.14b, we present the average number of meals household adults and
children eat in normal and food shortage months by region. This is given by region to assess
if averages vary across regions. As reported above, the average number of meals per day in
a food shortage month is 2 for adults and 3 for children. In a good month, the average
number of meals for adults is 3 and for children it ranges between 3 and 4. Region wise, all
regions have improved between 2006 and 2010 for both adults’ and children’s number of
meals except for Tigray and Amhara, which saw a slight decline in meals consumed by
adults and children in months with food shortages.

45
Figure 3.14a. Average number of meals per day—Adults
Figure 3.14b. Average number of meals per day—Children
Source: Authors’ calculations based on the PSNP survey data.

46
Coping Strategies
When facing acute food shortages, households follow a number of coping strategies that
sometimes are negative and unsustainable to the household. One such example is
consuming their productive inputs, such as seed stocks, consuming other holdings, or
consuming unusual foodstuffs. In the Ethiopian historical context, the latter had happened in
situations where food shortage was most acute and survival was at stake. As such, reports
of households eating unusual foodstuffs have been considered as an alarm to the food
security situation of households in a given area. Nevertheless, it has to be noted that eating
wild animals is sometimes part of hunting and may not necessarily indicate an acute food
security situation. Bearing these in mind, we assess if there was any likelihood of
households consuming such unusual foods.
Figure 3.15 gives the percentages of households that reported having consumed unusual
food, wild animals/plants, and seed stocks. Clearly, some households in all regions have
reported they had consumed three of these items, but in different proportions across regions
and years. It appears that a considerable number of households from Oromiya and SNNPR
in all years, and households from Tigray in 2006 reported they consumed unusual foodstuffs.
This should be taken carefully though, because what is unusual food in the other parts of the
country could be a normal food in the southern part of the country (SNNPR).
A case in point is that root crops are commonly consumed in SNNPR, which could be
considered as unusual food in some other regions. What is somehow worrying is, however,
a considerable proportion of households have reported to have eaten their seed stocks,
which appears to be slightly higher in 2008 in Amhara and SNNP than in the rest of the
years.
Figure 3.15. Percentage of households reporting different coping strategies, by year
and region
Source: Authors’ calculations based on the PSNP survey data.
Does this situation remain the same in both the PSNP and non-PSNP areas? Figure 3.16
helps to answer this question. The proportion of the PSNP households consuming seed

47
stocks increases from 2006 to 2008 but marginally declines between 2006 and 2010, while
the proportion of non-PSNP households consuming seed stocks increases between 2006
and 2010. The consumption of unusual food by all types of households declines between
2006 and 2010 with slight increases in 2010 from that of 2008. The consumption pattern of
wild animals/plants remains fairly irregular over the years with a slight decrease for the non-
PSNP households in 2008 and PSNP households in 2010.
Figure 3.16. Percentage of households reporting different coping strategies, by PSNP
beneficiary status and year
Source: Authors’ calculations based on the PSNP survey data.
3.5. Asset Levels
Households experienced widespread drought and other weather related shocks, large
increases in input prices, as well as difficulties in accessing input markets in 2010. These
conditions have prevailed for most of the period since 2006 and have worsened in many
cases. Such an environment is far from conducive to maintain asset levels, let alone growth
in their levels. We now turn to some trends in various types of assets held by households
and the likelihood that they incurred distress sale of assets for satisfying food needs.
Figure 3.17 gives the average real value of production assets (hoes, sickles, ploughs, water
pumps etc., but not livestock) held by households for the whole sample and by region.
5
This
shows that, on average, the value of production assets has fluctuated around 150 Birr
between 2004 and 2010, with a small increase observed from 2004 to 2006, followed by a
small decrease observed in 2008, and then, finally, a small increase in 2010. The regional
graphs show a very different picture and emphasize the regional differences. On the whole,
households in Tigray hold a much higher value of production assets and have experienced a
steady decline over the six-year period for which we have data. The average value of
production assets held in Tigray in 2004 was close to 272 Birr, which fell to 199 Birr in
2010—a decline of 27 percent. Oromiya, on the other hand, experienced a steady increase
5
We deflate nominal values with the regional CPIs published by the Central Statistical Agency, Ethiopia.

48
in the average value of production assets held over this period. Even though the absolute
value of assets held in Oromiya is much lower than that in Tigray, the asset growth over this
period was close to 32 percent. Value of production assets among households in Amhara
and SNNPR fluctuated from positive to negative and then again positive in the periods
2004–2006, 2006–2008, and 2008–2010, respectively.
Figure 3.17. Value of production assets owned (Birr)
Source: Authors’ calculations based on the PSNP survey data.
We examine the average real value of production assets by beneficiary status in Figure 3.18.
This figure shows that, on average, public works beneficiaries hold a lower value of
production assets compared to non-beneficiaries. However, public works beneficiaries have
experienced a slow and steady increase in the value of these assets, whereas the non-
beneficiaries’ asset growth has fluctuated.
Figure 3.18. Value of production assets owned (Birr), by beneficiary status
Source: Authors’ calculations based on the PSNP survey data.
Note: PW = public works beneficiaries.

49
Livestock is a very important asset in the rural Ethiopian context; therefore we now examine
movements in livestock held by households in our sample. In each survey round, we asked
households to give detailed livestock ownership for the survey year as well as the previous
two years. This allows us to construct annual values of livestock held from 2004–2010.
Figure 3.19 gives the average real value of livestock held by households over the period
2004–2010. Value of livestock held remained more or less constant around 3,500 Birr
between 2004 and 2006, it fell to 3,000 Birr in 2008, and then again went up to 3,500 Birr in
2009–2010. These findings can be rationalized on the basis of findings described in
subsection 3.2. We know that the period between 2006 and 2008 saw sharp increases in
food prices that were not accompanied with similar increases in livestock prices or wages.
As a result, households may have been forced to sell off assets (mostly livestock) at a faster
pace in 2008 to overcome food shortages.
The graphs for the four regions (Figure 3.19) give a much different picture. First, the average
levels of value of livestock held differs across the regions—they are much higher in Oromiya
and much lower in SNNPR than in the other two regions. Second, trends in value of livestock
held vary by region. In Tigray (except for 2004–05), there is a downward trend over the
period. The trend in Amhara mimics somewhat a business cycle—where it increases from
2004 to 2005, then decreases between 2005 and 2006, reaches an all-time low in 2008, and
then starts to increase from 2009 onward. Oromiya experiences a decline in value of
livestock held from 2004–2008, then a rise in 2009 followed by a slight decline in 2010.
Households cite pests and diseases that affect livestock as a major shock in this region over
the past six years (Figure 3.2), which improved significantly in 2010. Oromiya was also the
hardest hit in terms of the food price crisis that led to a significant increase in food prices but
a small increase in livestock prices. These two factors may explain the decline in livestock
value up until 2008. The food price situation has improved significantly in that region and
livestock prices continue to rise in 2010—thereby enabling households to accumulate
livestock.
Figure 3.19. Value of livestock owned (Birr)
Source: Authors’ calculations based on the PSNP survey data.

50
Figure 3.20 presents the real value of livestock owned by beneficiary status. This shows
that, on average, public works beneficiaries own a lower value of livestock compared to non-
beneficiaries. Another interesting feature of this graph is the extent to which livestock value
fell from 2006 to 2008 in the wake of the food price crisis. Both groups experienced large
losses, although the absolute and relative magnitudes of these losses are much lower for the
public works beneficiaries. Livestock value fell by 9.6 percent among public works
beneficiaries, whereas the corresponding number for non-beneficiaries was 19.6 percent.
That said, non-beneficiary households seem more resilient to the shock compared to public
works beneficiaries, as they were able to get back to pre-crisis levels by 2009. Although
public works beneficiaries have yet to go back to pre-crisis levels in terms of livestock value,
their asset levels have been consistently raising since 2008, which is not true for the non-
beneficiaries. We will be able to say more about the impact of the FSP on livestock holdings
in the impact report.
Figure 3.20. Value of livestock owned (Birr), by beneficiary status
Source: Authors’ calculations based on the PSNP survey data.
Note: PW = public works beneficiaries.
The value of livestock held gives a good measure of value of livestock but does not throw
much light on the number of livestock held. To get around this issue, we present the total
tropical livestock units (TLU) held by households over the period.
6
These are presented in
Figure 3.21. We observe a steady increase in TLU held by households in the sample from
2004–2007, then a slight decline in 2008, and then a rise in 2010. There are regional
variations as indicated by the regional graphs. Figure 3.22 gives the average TLU held by
beneficiary status. Not surprisingly, given the targeting of the PSNP, non-beneficiaries have
higher livestock holdings than public works beneficiaries.
6
TLU equals one for cattle, horses, and mules, 0.15 for sheep and goats, 0.005 for poultry, 0.65 for donkeys, and 1.45 for
camels (Ramakrishna and Demeke 2002).

51
Figure 3.21. Tropical Livestock Units owned
Source: Authors’ calculations based on the PSNP survey data.
Figure 3.22. Tropical Livestock Units owned, by beneficiary status
Source: Authors’ calculations based on the PSNP survey data.
Note: PW = public works beneficiaries.
Distress Asset Sales
A major cause for households to draw down assets is to satisfy food needs when faced by a
shock. One of the objectives of the FSP is to promote asset accumulation among its
beneficiaries so that they can survive shocks, such as drought, that affect their income and
agricultural output. Because these households are also the ones that are most vulnerable to
shocks, asset accumulation would happen at a much slower pace as compared to non-
beneficiaries that start off with a much higher base of stock of assets. The FSP, by providing
a means of income during the lean season, aims to protect households from drawing down
their assets in order to fulfill food needs. Figure 3.23 shows the average fraction of
households that incurred distress sale of assets over the period 2006–2010. On the whole,
this fraction has declined steadily from about 48 percent in 2006 to 43 percent in 2008 and
30 percent in 2010. The regional graphs show that there is some variation across the four
regions. In Tigray, even though a bit steeper than average, the pattern is the same as the
sample as a whole. Oromiya started out with a largest fraction of 61 percent households

52
incurring distress sale of assets in 2006, which fell to 47 percent in 2008 and remained
stable thereafter. Amhara and SNNPR experienced an increase in distress asset sales in
2008 compared to 2006 and then a sharp decline in 2010 compared to 2008.
Figure 3.23. Distress sale of assets for satisfying food needs (proportion of
households)
Source: Authors’ calculations based on the PSNP survey data.
Is there a difference between distress asset sales across public works beneficiaries and non-
beneficiaries? Figure 3.24 throws some light on this question. On average, the pattern of
distress asset sales has been similar across rounds. Public works beneficiaries started with
an average fraction of 51 percent of households incurring distress sale of assets, which has
gone down steadily and was observed to be 34 percent in 2010. The corresponding figures
for non-beneficiary households were 44 percent in 2006 and 28 percent in 2010. This shows
that the decline in the fraction of households that incurred distress asset sales is not too
different for beneficiaries and non-beneficiaries. However, in combining this finding with the
finding on asset levels among the two groups, we can say that even though beneficiaries
have much lower levels of asset holdings compared to non-beneficiaries, they have been
able to fulfill their food needs without having to sell off their assets.

53
Figure 3.24. Distress sale of assets (proportion of households), by beneficiary status
Source: Authors’ calculations based on the PSNP survey data.
Note: PW = public works beneficiaries.
3.6. Subjective Well-being
Next, we turn to data on households’ perception of poverty and well-being. In the survey, we
asked households to rank themselves as compared to other households in the village.
Figure 3.25 gives responses for the 2008 and 2010 surveys. A larger fraction of the
beneficiary households describe themselves as amongst the poorest in the village in 2010
(39 percent) as compared to 2008 (33 percent). At the same time, the fraction of
beneficiaries that described themselves as the poorest has gone down.
Figure 3.25. Compared to other households, you describe your household as…
Source: Authors’ calculations based on the PSNP survey data.

54
An important means of coping with shocks is by borrowing. In the survey, we ask whether
the household would be able to borrow easily 100 Birr within a week. Figure 3.26a shows
that the fraction of households that can easily borrow 100 Birr has gone up slightly from 54
percent to 57 percent between 2008 and 2010. However, the fraction of beneficiary
households that are able to borrow remained constant at 52 percent over this period. On
average, beneficiaries are less likely to be able to borrow 100 Birr as compared to non-
beneficiaries.
Figure 3.26a. Proportion of households able to borrow 100 Birr
Source: Authors’ calculations based on the PSNP survey data.
In the survey, we also asked households if they were given 100 Birr, how they would spend
it? Figure 3.26b shows that, on average, 50 Birr would be spent on food, with FSP
beneficiaries spending slightly more on food than non-beneficiaries. Shares on education
and other nonfood expenses are almost identical across households, whereas non-
beneficiaries are slightly more likely to save from the 100 Birr.
Figure 3.26b. Average share spent on . . . if given 100 Birr
Source: Authors’ calculations based on the PSNP survey data.

55
When asked about the household’s food consumption over the year, overall, in 2010
households are more likely to respond that they had just enough as compared with 2008
(Figure 3.27a). Although all household groups have improved on this account, the fraction of
households that reported the positive outcome is larger among the non-beneficiaries. In
terms of housing needs, conditions have remained more or less static over the two rounds
(Figure 3.27b). On average, as shown in Figure 3.27c, 32 percent of households feel that
their household incomes were just adequate in 2010 as compared to 26 percent of
households that felt the same in 2008. The gain is higher among non-beneficiary households
(29 percent to 36 percent) as compared to beneficiaries (23 percent to 28 percent).
Figure 3.27a. Households’ food consumption (percent of households)
Source: Authors’ calculations based on the PSNP survey data.
Figure 3.27b. Household’s housing needs (percent of households)
Source: Authors’ calculations based on the PSNP survey data.

56
Figure 3.27c. Household’s income (percent of households)
Source: Authors’ calculations based on the PSNP survey data.
Lastly, households were asked about their perception of the overall economic situation of
their household as compared to one year ago. The results are shown in Figure 3.28 and are
very encouraging. Responses from 2008 possibly reveal the impact of the food price crisis,
where a large fraction of households irrespective of beneficiary status said that they were
much worse off now (i.e., in 2008). The situation is much better in 2010, where the fraction of
households that feel that they are either the same or a little better off is 70 percent as
compared to 41 percent in 2008. Here, the PSNP beneficiaries feel a little better about their
economic situation as compared to the non-beneficiaries. In other words, 72 percent of the
beneficiaries feel that they are either the same or a little better off now, as compared to 68
percent among non-beneficiaries.
Figure 3.28. Overall economic situation of household compared with one year ago...
(percent of households)
Source: Authors’ calculations based on the PSNP survey data.

57
3.7. Summary
In this chapter we examined the price trends, incidence of shocks, and summary statistics of
key outcomes of interest. There are several important findings.
There were sharp increases in food prices in 2008, with some regional variation. This
sharp increase in food prices was not accompanied with a comparable increase in
livestock prices or wages. Oromiya and SNNPR were worse hit by this crisis as
compared to the two northern regions in our sample.
In 2010, food prices continue to rise, but the rate of increase is much slower. SNNPR
experienced a decline in food prices.
Drought is the most important shock that affects a large fraction of households every
year and causes income and consumption shortfalls.
The incidence of shocks differs across regions, but does not vary much across
beneficiary status.
When looking at households’ primary source of food, we find that a large fraction
constitutes food from own production. However, this fraction has steadily declined. The
fraction sourced from PSNP has increased steadily over the period.
The food gap as measured by the number of months that the household is unable to
satisfy its food needs from own production has reduced from 3.6 months to 2.3 months
among all households.
FSP beneficiary households, on average, hold lower levels of assets compared to non-
beneficiaries. This is reasonable, since the FSP beneficiaries were selected to be the
more vulnerable and poor.
Over the period 2004–2010, asset levels have increased. Although beneficiary
households have not experienced accumulation of assets at a fast pace, they have
shown a steady increase.
There has been a decline in distress sale of assets, irrespective of beneficiary status.
On subjective measures of well-being, also households have fared better in 2010 as
compared to 2008. In particular, when asked about how they felt their overall economic
condition was as compared with last year, in 2010, 70 percent of households feel that
they are either the same or a little better off, as compared to 41 percent in 2008.

58
4. Woreda Perspectives on Implementation
4.1. Introduction
The PSNP PIM states that “The woreda is the key level of government that determines
needs, and undertakes planning and implementation of safety net activities”. A partial list of
woreda responsibilities includes:
The Woreda Office of Agriculture and Rural Development (WOARD) manages
both the PSNP and the HABP.
The Woreda Food Security Task Force (WFSTF) reviews kebele annual PSNP
and HABP plans and budgets, ensures that contingency plans for PSNP risk
financing are in place, and provides assistance to kebeles.
The Woreda Food Security Desk (WFSD) coordinates safety net and household
asset building activities.
The Woreda Office of Finance and Economic Development (WOFED) ensures
that the budgets for the safety net and household asset building programs are
received in a timely manner at the woreda level and subsequent transfers to
beneficiaries are undertaken on a timely basis (World Bank 2009, 25).
Under the FSP, there has been a deliberate strategy to increase implementation capacity at
the woreda level and significant resources have been dedicated to implement this strategy.
This chapter assesses the success of this strategy in terms of improvements in measurable
dimensions of capacity (for example, training and physical infrastructure) and whether these
have resulted in more timely transfers to beneficiaries. It also considers the role played by
early warning systems and the use of contingency funds. It draws on the following data
sources: the woreda quantitative survey, and key informant and focus group interviews at
the woreda and regional level. As such, it provides a woreda level perspective on
implementation, complemented in some places with information taken from regional sources.
In doing so it addresses the following evaluation objectives described in the inception report
(Table 4.1).
Table 4.1. Evaluation objectives covered in chapter 4
Evaluation
objective
Issue
Data
sources
Links to Log Frames and TOR
Document progress in the implementation of the PSNP
Are public works payments timely and
predictable? Do clients receive complete
entitlement?
3
PSNP Log frame Output 1.1
PSNP Log frame Output 1.2
TOR, para 39
Are direct support payments timely and
predictable? Do clients receive complete
entitlement?
3
PSNP Log frame Output 1.1
PSNP Log frame Output 1.2
TOR, para 35
Why are there woreda level differences in
timeliness of transfers?
3
TOR, para 36
How are contingency budgets used, including as
a response to the Kebele Appeals system?
3
TOR, para 41
Source: Authors’ compilation.

59
4.2. Woreda Resources for the PSNP
We begin by describing the characteristics of key woreda staff involved in the
implementation of the PSNP (Table 4.2).
Table 4.2. Characteristics of key woreda staff
Head, Woreda Food
Security Office
PSNP
accountant
PSNP
cashiers
Sex (percent)
Male
95
76
65
Female
5
24
35
Age
Mean years
32
27
27
Schooling (percent)
University degree
75
40
5
High school plus some post high school education
25
52
48
High school
0
8
20
Less than completed high school
0
0
27
Mean years worked in this:
Position
2.5
3.4
3.3
Woreda
7.0
3.9
3.9
Occupation
3.2
3.1
3.2
Job status (percent)
Full-time
94
25
18
Part-time
6
75
82
Received training specific to the PSNP
Percent
93
95
50
Source: Woreda quantitative questionnaire 2010.
Levels of formal education are high; these officials are predominantly male and young. They
are relatively new to their positions. The Head of the Woreda Food Security Office (WFSO)
works full-time while the PSNP accountant and cashiers generally work part-time on the
PSNP. Nearly all heads of the WFSO and accountants working on the PSNP have received
training—see Figures 4.1 and 4.2 for specifics. This was also apparent during the key
informant interviews held at the woreda level. For example, knowledge and comprehension
of the three types of the PSNP budget lines (the Transfer, Capital and Administrative, and
the Contingency) among the Woreda Food Security Task Forces (WFSTF) was high across
all regions. In eight out of ten woredas in this study, the key actors of the program
(WFSTF/OFSP coordinators) are familiar with the different budget lines of the PSNP.
Training of cashiers is less common and focused on financial management (see Figure
4.3).
7
7
During the workshops held in August 2011, regional representatives expressed some surprise about the low level of PASS
training. They intimated that training is ongoing and that, given the timing of the survey, that staff turnover may have
temporarily reduced the number of cashiers trained on PASS.

60
Figure 4.1. Type of training received: Head, Food Security Office
Source: Authors’ calculations based on the PSNP survey data.
Figure 4.2. Type of training received: PSNP accountant
Source: Authors’ calculations based on the PSNP survey data.

61
Figure 4.3. Type of training received: PSNP cashiers
Source: Authors’ calculations based on the PSNP survey data.
The quantitative woreda questionnaire contained a series of questions about infrastructure
access and quality. Slightly less than 50 percent (40/85) of woredas had generators. There
were 15 woredas that did not have generators and electricity supply was okay, poor, erratic,
or non-existent (Table 4.3).
Table 4.3. Quality of access to main electricity (percent of woredas)
Tigray
Amhara
Amhara-HVFB
Oromiya
SNNPR
All
Excellent
0.0
0.0
30.0
0.0
0.0
3.5
Good
61.5
55.0
40.0
52.4
61.9
55.3
Okay
30.8
25.0
20.0
14.3
19.1
21.2
Poor
7.7
5.0
0.0
4.8
9.5
5.9
Erratic
0.0
0.0
0.0
0.0
0.0
0.0
No electricity
0.0
15.0
10.0
28.6
9.5
14.1
Source: Woreda quantitative questionnaire 2010.
Notes: Excellent = available all the time; Good = available most of the time; Okay = available about half of the time; Poor =
available less than half of the time; Erratic = rarely available.
There are considerable differences in phone access across these four regions. Tigray, and
to a lesser extent SNNPR, have excellent or good communications by landline, cell phones,
or both (Table 4.4 and Table 4.5). Phone access in Amhara is mixed with about half the
woredas surveyed having excellent or good access, but a quarter having both landline and
cell phone access that is “okay” or worse (Table 4.6). Phone access is poorest in Oromiya.
Internet access is largely absent. Seventy-two percent of the woredas (61/85) have no
internet access at all and of the remainder, only two woredas described access as either
excellent or good.

62
Table 4.4. Quality of access to landline telephones (percent of woredas)
Tigray
Amhara
Amhara-HVFB
Oromiya
SNNPR
Total
Excellent
76.9
50.0
30.0
19.1
10.0
34.5
Good
15.4
25.0
20.0
42.9
65.0
36.9
Okay
0.0
5.0
10.0
4.8
15.0
7.1
Poor
7.7
15.0
10.0
14.3
5.0
10.7
Erratic
0.0
0.0
0.0
0.0
0.0
0.0
No landline
0.0
5.0
30.0
19.0
5.0
10.7
Source: Woreda quantitative questionnaire 2010.
Notes: Excellent = available all the time; Good = available most of the time; Okay = available about half of the time; Poor =
available less than half of the time; Erratic = rarely available.
Table 4.5. Quality of cell phone coverage (percent of woredas)
Tigray
Amhara
Amhara–HVFB
Oromiya
SNNPR
Total
Excellent
15.4
20.0
10.0
10.0
4.8
11.9
Good
46.2
45.0
50.0
35.0
61.9
47.6
Okay
23.1
10.0
20.0
10.0
19.1
15.5
Poor
7.7
15.0
10.0
20.0
4.8
11.9
Erratic
7.7
5.0
0.0
5.0
0.0
3.6
No cell phone coverage
0.0
5.0
10.0
20.0
9.5
9.5
Source: Woreda quantitative questionnaire 2010.
Notes: Excellent = available all the time; Good = available most of the time; Okay = available about half the time; Poor =
available less than half the time; Erratic = rarely available.
Table 4.6. Quality of cell phone and landline access (percent of woredas)
Tigray
Amhara
Amhara-
HVFB
Oromiya
SNNPR
Total
Both are excellent or good
61.5
60.0
40.0
38.1
47.6
49.4
Only landline is excellent or good
30.8
15.0
10.0
23.8
23.9
21.2
Only cell coverage is excellent or good
0.0
5.0
20.0
4.8
19.0
9.4
Neither are excellent or good
7.7
20.0
30.0
33.3
9.5
20.0
Source: Woreda quantitative questionnaire 2010.
Access to government-owned vehicles is “rare or never available” in more than half the
localities (Table 4.7). In principle, woreda officials can rent private vehicles when
government vehicles are not available. However, in less than 30 percent of the woredas
where access to government-owned vehicles was “sometimes available” or “rarely or never
available” was a vehicle rented.
Table 4.7. Access to government-owned vehicles (percent of woredas)
Tigray
Amhara
Amhara-
HVFB
Oromiya
SNNPR
Total
Always available or available on short notice
15.4
30.0
10.0
14.2
28.6
21.7
Sometimes available
46.2
30.0
20.0
23.8
23.8
28.2
Rarely or never available, not known
38.4
40.0
70.0
61.9
47.6
50.6
Source: Woreda quantitative questionnaire 2010.

63
On average, there are 1.84 computers available for the exclusive use of the PSNP. More
than 90 percent were in working order at the time of the survey (Table 4.8).
Table 4.8. Access to working computers and printers (percent of woredas)
At least one working
Tigray
Amhara
Amhara-HVFB
Oromiya
SNNPR
Total
Computer
Printer
Yes
Yes
84.6
95.0
60.0
61.9
90.5
80.0
Yes
No
15.4
5.0
20.0
33.3
9.5
16.5
No
Yes
0.0
0.0
0.0
0.0
0.0
0.0
No
No
0.0
0.0
20.0
4.7
0.0
3.5
Source: Woreda quantitative questionnaire 2010.
Respondents were asked about their access to PSNP-related manuals. Generally, access is
good but not perfect (Table 4.9). Out of the 85 surveyed woredas, 75 reported that they had
the PIM, 75 had the targeting manual, 79 had the graduation manual, 72 had the financial
management manual, and 67 had the manual covering watershed management. Access to
manuals was most problematic in Oromiya.
Table 4.9. Access to manuals (percent of woredas)
PIM
Targeting
Graduation
Financial
Management
Tigray
Amhara
Amhara-
HVFB
Oromiya
SNNPR
Total
Yes
Yes
Yes
Yes
61.5
90.0
80.0
42.9
81.0
70.6
Yes
Yes
Yes/ No
Yes/No
7.7
5.0
10.0
19.1
9.5
10.6
Yes
No
Yes/ No
Yes/No
7.7
0.0
0.0
23.8
0.0
7.1
No
Yes/No
Yes/ No
Yes/No
23.1
5.0
10.0
14.3
9.5
11.8
Source: Woreda quantitative questionnaire 2010.
One way of considering these resource data in aggregate is by assessing whether woredas
have access to manuals, computers and printers, and vehicles—items that the FSP has
some control over. Table 4.10 provides these summary data. Several features are apparent.
First, relatively few woredas have all the resources needed to provide transfers on a timely
basis. Across all surveyed woredas, only 17.7 percent have all manuals, at least one
working computer and printer, and reasonable access to transport. Second, transport
emerges as the single largest gap in capacity in many woredas in Tigray, Amhara, and
SNNPR. Third, access to resources is—by a considerable margin—poorest in Oromiya
where nearly a quarter of surveyed woredas lacks working computers and/or printers, and
has poor access to transport.

64
Table 4.10. Aggregate PSNP resources (percent of woredas)
Transport
always
available or
available on
short notice
At least one
working
computer and
printer
Woreda has
all manuals
Tigray
Amhara
Amhara-
HVFB
Oromiya
SNNPR
Total
Yes
Yes
Yes
7.7
30.0
10.0
9.5
23.8
17.7
Yes
Yes
No
7.7
0.0
0.0
4.8
0.0
2.4
Yes
No
Yes
0.0
0.0
0.0
0.0
4.8
1.2
Yes
No
No
0.0
0.0
0.0
0.0
0.0
0.0
No
Yes
Yes
46.2
55.0
40.0
19.1
52.4
42.4
No
Yes
No
23.1
10.0
10.0
28.6
14.3
17.7
No
No
Yes
7.7
5.0
30.0
14.3
0.0
9.4
No
No
No
7.7
3.3
10.0
23.8
4.8
9.4
Source: Woreda quantitative questionnaire 2010.
4.3. Woreda Differences in the Timeliness of Cash Payments
How do woreda level resources affect the timeliness of payments to beneficiaries? To
answer this, we begin by describing how the cash payments system works and how long it
takes for payments to be made. With this information in hand, we look at the relationship
between the timeliness of cash payments and woreda level resources available to support
the PSNP.
The Cash Payment System and the Timeliness of Payments
The system for making cash payments consists of the following steps. Kebeles are
responsible for forwarding attendance sheets and lists of individuals eligible for direct
support to the WFSO. The WFSO enters these data using a payroll software system called
PASS. Nearly all woredas surveyed, 90 percent, use PASS to enter these data, 6 percent
use PASS only for public works participants, and 4 percent do not use PASS. When this is
complete, this information is given to the WOFED. The regional Bureau Office of Finance
and Economic Development (BOFED) notify WOFED when funds have been transferred to
the branch of the Commercial Bank of Ethiopia from which the woreda can withdraw funds.
Once funds are withdrawn, arrangements are made to pay beneficiaries.
The woreda quantitative survey asked staff from WFSO and WOFED to construct a time line
listing the different dates on which these activities took place for the last month on which a
cash payment was made; 71 out of the 85 woredas surveyed were able to provide this
information.
8
Figure 4.4 summarizes this information showing the cumulative mean number
of days taken to deliver payments. This shows that, on average, data entry into PASS begins
3.1 days after the receipt of the first attendance sheet. Data entry is completed 14.6 days
later and so the payment information is given to WOFED 17.7 days after receipt of the first
attendance sheet. On average, 5.7 days later, staff goes to the Ethiopia Commercial Bank to
withdraw funds for payment, and 3.9 days after that makes the first payment trip. It takes
11.6 days to make all payments with the result that, on average, 38.9 days elapse between
the receipt of the first attendance sheet and the last day on which payments are made.
8
In eight cases, the woreda had not made a cash transfer since January 2010 and in six cases, information was incomplete.

65
There are a number of woredas where the process of entering data into the PASS continues
after payments are made; this is described in Figure 4.4 as “to end of processing.”
Figure 4.4. Cumulative mean number of days to deliver payments, by activity
Source: Authors’ calculations based on the PSNP survey data.
Table 4.11 shows how the time taken to complete these activities varies by region showing
the time taken by the fastest woreda, the woreda at the 25th percentile of the regional
distribution of processing time, the median, the 75th percentile, and the slowest woreda. The
rows are ordered from the fastest region—as measured by the median—to the slowest.
Tigray reports the fastest time between receiving the first attendance sheet and the end of
processing; the median time being 23.5 days. SNNPR and Amhara-HVFB are close behind
with median times of 31 and 33 days, respectively. Median processing times in Amhara are
slower at 39 days and the median in Oromiya, 50 days, lags far behind the other regions.
Table 4.11 also shows that there is considerable variation within regions. If we look at the
75th percentile—which tells us that three-quarters of woredas processed payments at or
faster than this number—we see that Tigray still outperforms the other regions but that there
is now a smaller difference between Amhara and SNNPR. Generally, Oromiya processes
payments more slowly than other regions, but while it has the woreda with the slowest
processing time, it also has one of the fastest. This suggests that while a regional
disaggregation is informative, it would be helpful to focus on additional factors that are
associated with differences in processing cash payments.
Table 4.11. Regional variations in number of days taken to deliver cash payments
Sample
size
Fastest woreda
25th percentile
Median
75th percentile
Slowest woreda
Tigray
12
13
18.5
23.5
34
123
SNNP
21
14
23
31
44
78
Amhara-HVFB
6
6
30
33
34
42
Amhara
16
17
30.5
39
49
60
Oromiya
16
6
32.5
50
104.6
125
Source: Authors’ calculations based on the PSNP survey data.

66
To do so, we begin with Figure 4.5. The triangles in Figure 4.5 represent the median
cumulative amount of time taken; the lower bubble shows the cumulative time by the fastest
(10th percentile) woreda and the top bubble shows the cumulative time by the slowest (90th
percentile) woreda. It shows that the woredas that process and pay fastest do so with
genuinely impressive speed, taking only 13 days from the start of data entry to the final
payment. The median woreda completes all these activities in 32 days, while the slowest
woredas took 90 days.
Figure 4.5. Variations in number of days to deliver payments: 10th, 50th, and 90th
percentiles
Source: Authors’ calculations based on the PSNP survey data.
Figure 4.6 divides this sample of woredas into three groups based on the speed at which
they completed all activities and shows the median cumulative time taken from the receipt of
the first attendance sheet to final payment for each group.

67
Figure 4.6. Timeliness of payments, by tertiles
Source: Authors’ calculations based on the PSNP survey data.
Virtually all woredas start data entry immediately on receipt of the first attendance sheet. It is
worth noting that, generally, woredas do not receive all attendance sheets at the same time.
On average, 12 days elapsed between receipt of the first and final attendance sheets, but
this hides considerable variation. In 75 percent of the surveyed woredas, it took less than 15
days to receive all attendance sheets. In the remaining woredas, the time between receiving
the first and the last attendance sheets was between 16 and 120 days. When asked for their
perceptions of what caused delays, 8 out of 13 woredas in Tigray and 17 out of 30 woredas
in Amhara indicated that delays in receiving attendance sheets from kebeles was a problem.
In both Tigray and Amhara, a number of respondents indicated that kebeles would not send
attendance sheets until a specified amount of work had been completed. Few woredas wait
until receipt of the last attendance sheet before beginning data entry, but in nearly all cases,
this resulted in longer (than the median) processing times.
Comparing the fastest and middle group of woredas, the amount of time taken from the start
of data entry to the first payment trip is approximately the same. The median time taken by
the fastest group to complete all payments is 8 days compared to 16 days for the middle
group, and this difference largely accounts for the overall difference between the fastest and
middle groups. The slowest woredas take a long time to submit payment information to
WOFED, have to wait 17 days before making the first trip to the bank to withdraw funds, and
do not make their first payment before an additional 10 days after that.
BOFED notifies WOFEDs when funds have been transferred to the branch of the
Commercial Bank of Ethiopia from which the woreda can withdraw funds. About 72 percent
of woredas call BOFED to see if funds have been sent to these branches. In nearly all
cases, BOFED sends a letter verifying that funds have been deposited. On average, 16 days
elapsed between attendance sheets being given to WOFED and the final trip to the bank to
withdraw funds needed for payment. But variation in elapsed time across woredas is large. It
took five days or less to obtain money for payment in half of the woredas surveyed but it took

68
more than 15 days in 30 percent of woredas surveyed. Approximately one-third of all
woredas reported that they were not able to withdraw the funds they needed to pay all
beneficiaries.
Resources and Payment Delays
As part of the woreda quantitative survey, respondents were asked to provide their views on
what problems affected the timely payment of beneficiaries. These are described in Table
4.12.
Table 4.12. Woreda perceptions of problems affecting payments
Sample
size
Delay in
receiving
attendance
sheets
Delay in
getting
funds
from
BOFED
Transport
PASS
Donor
delays
Lack of
cashiers or
accountants
and staff
turnover
Lack of
other
resources
Too many
beneficiaries
to pay
Tigray
13
8
2
5
1
2
4
1
3
Amhara
19
8
9
5
6
0
4
3
1
Amhara-HVFB
11
4
6
5
2
0
5
0
0
Oromiya
21
1
19
16
4
0
2
0
1
SNNP
21
7
17
9
4
0
4
3
0
Source: Woreda quantitative questionnaire 2010.
These issues were also discussed during the qualitative key informant interviews held in 10
woredas. Many of these discussions highlighted significant delays in the release and transfer
of budgets from regions to woredas. Many, but not all, noted that the system is working
better now than it did in the past. In Tembaro woreda of SNNPR, the following was noted:
“Earlier, the budget delay was a common practice. This year there is no delay in receiving
the budget” [SN_T/W-KI-1]. By contrast, in Ebinat woreda in Amahara region, a respondent
stated: “Most of the time the funds were released on time, but they are delayed this year”
[AM_E/W-KI-1]. A similar response was generated in Shebedino woreda of the SNNPR: “In
earlier times, the money comes on time, but this year it is delayed” [SN_S/KI-1].
The respondent from Gursum woreda of Oromiya region noted that the delay resulted from
the failure of strictly following the PSNP calendar as it is indicated in the PIM as well as the
time-intensive reporting processes from woredas to regions
9
:
The PSNP calendar year is still confusing for us. According to the PSNP PIM,
PSNP uses the local (Ethiopian) budget year, while the actual budget transfer
is according to the Gregorian calendar year. Delay of transfer (capital,
administration, and 5 percent contingency) is mainly due to delay in reporting
from the woreda. The region is also not timely facilitating the transfer for us
[ORO_G/KI-1].
Somewhat surprisingly in light of Table 4.12, of the 10 woredas covered in the qualitative
study, only two (Afherom and Sasei Tsemba in Tigray) reported that the front-loaded
9
The PSNP PIM indicates that 80 percent of the capital and the administrative budgets will be front-loaded prior to the first
quarter of the season in order to facilitate the implementation of public work activities. However, the capital and the
administrative budgets were not front-loaded to the regions and woredas on a timely manner. “The region receives the capital
budget once in a year and the other budgets are split into three rounds” [SN_R/KI-1].

69
sources were insufficient to effectively implement the programs. Two others (Zeway Dugda
in Oromiya and Shebedino in SNNPR) declined to share their views and experiences
whether the resources were sufficient or not. In the remaining cases, the sufficiency of front-
loaded (i.e., pre-transferred) resources is treated in association with the timely release/delay
of resources. This could be attributed to the misinterpretation or misunderstanding of “front-
loading” of resources with the timely transfer. The following various quotes indicate this
aspect when asked, “Has the up-fronting of resources been sufficient for you to effectively
implement the PSNP in your area?” Woreda Food Security Task Forces replied as:
The capital budget has come here in March. It was also delayed last year; as
a result we weren’t able to use the capital budget [SN_T/W-FG-1].
Last year, the capital and the transfer budgets were released early in
December for January–March. We, however, did a procurement of materials
on accrual/liability in October since we are not able to work on irrigation and
natural resources activities if we do not accomplish procurement in October.
We then paid back when the budget arrived in December to WOFED
[AM_S/W-FG-2].
The Woreda Food Security Task Force in Gursum woreda in Oromiya region indicated that
they did not receive front-loaded resources. The following quote reflects this aspect.
We didn’t receive any front-loading. The region sends us the first 40 percent
PSNP resource transfer usually in March after 2–3 months implementation of
PSNP. This is inadequate to procure all required items for the public works
and other programs. As a result, we are forced to postpone procurement of
capital items for public works [ORO_G/W-FG-1].
Are these concerns consistent with what we observe across all woredas where we have
quantitative data on payment delays? To examine this, we ran regressions in which the
dependent variable is the logarithm of the number of days taken by a woreda to complete a
particular task (Table 4.13). We use logarithms for two reasons: they minimize the influence
of outliers; and the estimated coefficients can be interpreted as percentage changes
resulting from changing the explanatory variables. Because the dependent variable is (log)
days spent undertaking a task, a variable associated with this activity occurring more quickly
will have a negative coefficient and a variable associated with this activity occurring more
slowly will have a positive coefficient. These characteristics used as regressors are either
Yes/No variables or they are expressed in logarithms. Where the characteristic is a Yes/No
variable (for example, “Bank not local and more than one trip was required to obtain funds”),
the reported coefficients can, to a first approximation, be interpreted as percentage changes.
For example, the number 0.611 in column (2) means that, controlling for other
characteristics, the time taken from the submission to WOFED to the first payment trip was
61.1 percent higher in woredas where PSNP staff could not use government vehicles.

70
Table 4.13. Associations between woreda characteristics and time spent on PSNP
payment activities
(1)
(2)
(3)
(Log) Number of days between
Receipt of first
attendance sheet
and WOFED
submission
Submission
to WOFED
and first
payment trip
First payment
trip and last
payment trip
Both landline and cell phone service is excellent or good
-0.617*
-0.180
0.194
(-1.663)
(-0.423)
(0.373)
Electricity supply is excellent or good
-0.562*
(-1.737)
PSNP funds were front-loaded
-1.700***
-1.220**
(-3.909)
(-2.140)
Bank not local and more than one trip was required to
obtain funds
0.101
-0.268
(0.179)
(-0.473)
PSNP staff could not use government vehicles
0.611*
0.730**
(1.710)
(1.994)
Number of cashiers working exclusively for PSNP
-0.447*
0.066
0.108
(-1.875)
(0.195)
(0.296)
PSNP accountant trained in PASS
-0.957***
(-2.916)
Number of years PSNP accountant has worked in woreda
(log)
-0.016
-0.042
(-0.0537)
(-0.119)
Travel time to furthermost kebele (log)
0.380
-0.071
(1.343)
(-0.157)
Constant
2.574
2.075***
2.925
(1.530)
(2.983)
(1.135)
Observations
50
46
46
Adjusted R-squared
0.207
0.182
0.062
Source: Authors’ calculations based on the PSNP survey data.
Notes: t-statistics in parentheses. Standard errors robust to heteroscedasticity. * significant at 10 percent, ** significant at 5
percent, *** significant at 1 percent.
Table 4.13 presents a number of interesting results.
10
Good access to communication
infrastructure is associated with faster processing of the attendance sheets. Better electricity
supply helps too, as does having more staff. The most striking result in column (1), however,
is the coefficient on whether the PSNP accountant was trained. Recall from Figure 4.2 that
about 70 percent of PSNP accountants were trained on the PASS. Where this has occurred,
processing of attendance sheets goes much faster. We explored whether other
characteristics (not reported) of woredas affected the time spent on this task. Access to
working computers and printers did not affect this, but this may be because virtually all
woredas now have these. Access to manuals did not seem to matter nor did the
characteristics (age, sex, education) of the staff working on the PSNP.
11
What accounts for the length of time it takes from the submission of information to WOFED
to the first trip to pay beneficiaries? As described above, the key constraint is whether funds
have been transferred from the regional BOFED to the local branch of the Commercial Bank
10
Two caveats should be born in mind. Table 4.12 shows associations and accordingly caution should be used when
interpreting the results. Second, the sample sizes are relatively small.
11
We considered whether woredas that had more experience with cash transfers processed payments more quickly. Using
data collected in the woreda quantitative survey, we calculated the number of times in 2010 that the woreda had made cash
payments prior to the payment cycle described here. Neither descriptive statistics nor regression analysis showed a
correlation between prior experience with cash and processing times.

71
of Ethiopia. Where these funds have been front-loaded (i.e., pre-transferred), the amount of
time taken to complete this step is dramatically lower (column [2]). The mean time to
complete these activities in woredas where funds have not been front-loaded is about 20
days, while in woredas where front-loading occurs, it is only two days. Table 4.13 indicates
that this difference persists even when we control for other woreda characteristics.
Lack of access to vehicles appears to be the most important factor when it comes to
assessing what influences the amount of time it takes to make all payments. Column (3)
shows that where PSNP staff did not have access to government vehicles, and hired private
vehicles or walked to payment sites, it took approximately 73 percent longer to complete all
transfers. Front-loading of funds was also correlated with faster payments. In these results,
and in others not reported, other characteristics, such as the age, sex, and education of
PSNP staff, travel times, and other woreda resources (such as availability of computers) did
not have statistically significant associations with the time taken to complete these activities.
It should be noted that woreda staff expressed considerable concern about these delays. In
both focus group and key informant interviews, respondents indicated that these delays led
to distress sales of assets and borrowing money at high interest rates. In addition, a delay
resulted in the procurement of poor construction materials for the public works and this, in
turn, led to low quality of public works. In addition, delays in transfer reduce the amount of
time available to complete public works and created difficulties in managing these activities.
These problems were compounded when disbursement of the capital budget was delayed.
The budget delay causes inappropriate procurement of materials that lack
proper quality. Besides, it is difficult to properly administer the work
[AM_S/KI-1].
The budget is supposed to be released in January, but is usually delayed for
over a quarter of the year. Even when it is released, by the time we are able
to purchase materials with the capital budget, we are not in the intended fiscal
year anymore due to the delay; rather, the work is postponed for the
subsequent year [TIG_S/KI-1].
If the capital budget is delayed, we will not have time to buy industry products
or construct infrastructures since there is a limited time before the end of the
budget year. We will not be able to buy tools on time. It prevents us from
putting plans into action. The time that could be used for implementation will
be shortened [SN_D/KI-1].
4.4. Woreda Differences in the Timeliness of Food Transfers
We had intended to provide an analysis of the timeliness of food transfers similar to that
provided above for cash payments. However, it proved difficult to obtain information from
enough woredas to do so. We suspect that the principal reason for this is that responsibility
for food distribution is split across a number of actors in many woredas and it is also difficult
to obtain a full picture about how this works.
12
For example, we have data from 49 woredas
12
The regional discussion in Tigray was helpful on this issue. They suggested that this difficulty may have arisen in part
because we did not include a sufficiently broad set of knowledgeable respondents. They indicated that in some woredas,

72
on who was responsible for arranging transport of food. This was handled by WFSO in 28
woredas, by an NGO in 15 woredas, by WOFED in one woreda, and by others (not
specified) in five woredas. Consequently, here we provide an abbreviated description of the
system through which food transfers are made.
Just as with cash transfers, initially, attendance sheets are sent to the WFSO. Apart from
Amhara-HVFB, this information is entered on PASS. Where the woreda, either through the
WFSO or WOFED, is responsible for arranging the distribution of food transfers, the
attendance data are given to WOFED. This does not occur in the Amhara-HVFB woredas.
There are dedicated storage facilities for PSNP food transfers in just over half the woredas
that answered this question. In woredas where distribution is handled by the government,
there are delays in making payments as a result of difficulties in arranging transport—this
was reported by 19 out of 29 woredas. Transport is less of an issue in woredas where
distribution is handled by NGOs with only 4 out of 15 reporting that problems in accessing
transport contributed to delays in making payments. Twelve out of 45 woredas had only one
food distribution point, 13 had two, 11 had three, and 9 had five or more.
Table 4.14 lists the median number of days taken to deliver food transfers—from the receipt
of the first payment to the final delivery—for the relatively small number of woredas for which
this outcome could be constructed. As with cash payments, Oromiya lags behind other
regions.
Table 4.14. Regional variations in number of days taken to deliver food transfers
Sample size
Median
Amhara-HVFB
3
11
Tigray
6
17
Amhara
3
23
SNNP
11
30
Oromiya
3
50
Source: Woreda quantitative questionnaire 2010.
4.5. Woreda Resources and Early Warning and Response Capacity
The PSNP PIM clearly indicates that the PSNP will incorporate the information from the
findings of the Early Warning System while preparing a contingency plan to effectively
respond to the imminent risks in a given area. At the regional level, information is collected
regularly on weekly, monthly, bi-annual, and annual basis on market prices of cereals and
livestock, crop performance (pre- and postharvest crop assessment), school drop-outs,
malnutrition rates, rainfall patterns, animal health and diseases, and availability of animal
feed and water. The qualitative study carried out in 10 woredas showed that 8 have
established the system for collecting early warning information. The other two have not
started but state that they receive information from kebeles in an informal way. However, the
early warning and transfer experts in Afherom woreda of Tigray region stressed that there is
a weak linkage between the woreda and kebele early warning information flow and
management:
managers responsible for food stocks and food distribution storekeepers might be able to assist in constructing these time
lines.

73
The way the system is organized is good. There is a better flow of information
from the woreda to the region. But the information flow, particularly from the
kebele level, is not regularly/consistently responding unless there are
emergencies. They are not using the communication channel effectively
[TIG_A/W-KI-2].
Most information is collected by development agents using a format developed at the
woreda level. Community involvement is limited in the information collection process. This
was indicated in the response of the informants from Ebinat woreda, in Amhara region:
While collecting information about the distribution of rain, we identify which
kebele has fully received rains, we send an agriculture supervisor, one
supervisor for three to four kebeles, with a format. The supervisor fills the
format accordingly and then it is sent to the zones. They [zone bureau of
DPPA and FSC] need it in number [AM_E/W-KI-2].
Once the information is collected, it is consolidated by the Woreda food security and
Disaster Preparedness and Prevention Agency (DPPA)
13
and sent to the zonal offices of the
DPPA and FSC on a monthly basis. Regional FSC-DPPA offices, in turn, produce a
consolidated early warning information report on a monthly basis and pass it on to the
Federal Early Warning and Response Directorate (EWRD) in Disaster Risk Management
and Food Security (DRMFS).
There are regional differences in terms of the utilization of the early warning information. In
Oromiya, it was indicated that although the early warning team collects timely and context-
based information, they are not involved in developing a woreda contingency plan. Early
warning experts have indicated their dissatisfaction that the regions are not using the
information collected accordingly to release and send the resources for the woredas when
required. “The region was not using our regular reports to justify and allocate resource from
the 15 percent contingency resource between 2005 and 2008” [ORO_G/W-KI-2].
In the other regions, the information is used to prepare a contingency plan. The PSNP PIM
indicates that the purpose of a contingency plan is to respond in due course to an
unexpected deteriorating situation during the PSNP implementation. However, in practice,
the contingency planning is not implemented as stated in the PIM across the regions
covered in this study. Interviews with the regional and woreda level early warning and
transfer experts indicated that the contingency plan had not been prepared as it should be
and the existing planning approach is mainly based on post-incidence of the shocks. When
asked, “How does the region/woreda prepare contingency plans to monitor risks and
respond to shocks accordingly?,” they responded as follows:
We do prepare our contingency plan by building on scenarios based on the
previous year’s situation of each locality. The plan is meant to minimize future
shocks [TIG_R/KI-EW].
13
Since the last two years, the government’s Early Warning information management and structure have been revised, hence
the names of some of the agencies have changed. Currently at the woreda level the Early Warning Response and Food
Security process owner reports to the Zone and it goes up to the Regional level Early Warning and Response Core process
(REWCP). The REWCP consolidates and reports to the Federal Early Warning and Response Directorate in DRMSS.
However, it should be noted that during this study, the revised structure has not been fully functioning in most of the woredas
assessed.

74
So far, woreda contingency plans are more response oriented. The plan often
focuses on addressing the observed risk [ORO_R-KI-2].
We didn’t have contingency plans in the last five years. We started to
exercise contingency planning in 2010. We used to develop response
measures once shocks are observed. The woreda early warning technical
team will carry out an assessment on the observed shocks and associated
risks, estimate population affected, propose measures, and report to the
woreda FS and DPPA to take response measures [ORO_G/W-KI-2].
So far we did not plan for a contingency budget; it is the food security work
process that plans the contingency. They simply ask us relevant information
for the plan input. They do not involve us much in the planning process
[AM_E/W-KI-2].
One of the topics discussed during these key informant interviews was early warning and
response (EWR) capacity. Strikingly, many of the capacity concerns raised in those
discussions echo those that emerge from these quantitative data on capacity to make
transfers. While the regional bureaus were generally adequately staffed (although concerns
were raised about skill levels and training), staff and capacity limitations were widely
reported at the woreda level and this adversely affected the collection and reporting of early
warning data in an accurate and timely manner, the preparation of contingency plans, and
the ability of woreda officials to respond to shocks.
Poor communication infrastructure and lack of transport were frequently cited as significant
constraints to EWR activities. The regional EWR Case Team in Amhara described the
condition as follows: “…there is no transportation and logistics to immediately deliver
assistances to the required locations. Even it is difficult for yourself to get public transport”
[AM/ R-FG-3]. These concerns were echoed at the woreda level. A key informant on EWR
from Saesi Tsedamba woreda in Tigray stated: “Vehicles and motorbikes are nearly
nonexistent” [TIG-S/ W-KI-3]. The Ebinat woreda Early Warning and Response Case Team
stated:
Challenges for the team are lack of resources: computers with printers,
motorbikes. We cannot visit places stricken by shocks immediately within 24
hours. We simply sit down in the office due to lack of transportation. Usually
we go to such places with others [NGOs and woreda sector offices] waiting
for their field visit plans [AM-E/W -KI-3].
These resource constraints are especially problematic in localities where road access is poor
and topography rugged. A key informant in Sekota woreda explained this bitterly:
There is a place called Dagbji, where 18,000 PSNP beneficiaries are living
scattered in nine kebeles. This area is very far from the woreda capital and it
is not easily accessible. There is no access road leading there. I can take you
there to see it if you want. There is no opportunity to access them... [Am-So/
W-FG-1].

75
Respondents were asked whether there were any significant delays in responding shocks
from the federal level. Apart from SNNPR (where it was not possible to interview the EWR
team), in the remaining regions officials said that there was a delay in responding to shocks.
Across the ten focus groups that we conducted on this topic, eight indicated that the
response was not timely, often arriving after communities have deployed negative coping
strategies already. Respondents explained that the resources arrive, on average, two-to-
three months after the shocks struck. However, the early warning and transfer experts in
Shebedino and Tembaro woredas indicated that the responses were timely. In part, this
might relate to the fact that these two woredas are located in a relatively easily accessible
distance from the regional capital, Awassa, and particularly, Shebedino woreda is just about
half an hour’s drive from Awassa. This could be factored that the regional early warning
team could easily arrive to the sites whenever the shock is reported.
It is worth noting that staff working on EWR issues at the woreda level voiced concerns
about access to office support. The EWR Case Team of Ebinat woreda perceived that this
was a consequence of the relatively low priority attached to EWR activities by the woreda
administration:
The woreda administration does not give emphasis to the team considering it
is not developmental work.... We are asked information on daily basis. But we
have no facilities to fulfill these requests. Telephone, computers, and
motorbikes are the most important materials needed to be fulfilled for this
team [AM-E/W-KI-3].
4.6. Contingency Funds
The revised PSNP PIM (June 2010) indicates that the PSNP contingency fund will be used
for the following purposes both at the regional and woreda levels: (1) to provide support to
households that qualify for inclusion in the PSNP but have not been entered in the regular
program, either because their need was recognized after the annual retargeting exercise, or
because the regular program budget is insufficient to include them; (2) to respond to
transitory needs among existing PSNP clients; and (3) to respond to transitory needs
among non-PSNP households based on early warning data collected by the woreda.
In other words, the contingency fund is used to respond to both chronic and transitory cases
of food insecurity among PSNP and non-PSNP households. The responses from both
regional and woreda level key PSNP implementers confirmed that this basic principle and
the purpose of the contingency fund are well understood. When asked, “What is the
contingency fund?,” the majority of the informants associated its purpose with the
occurrence and reduction of risks within the PSNP and the non-PSNP woredas. The
informants from Oromiya region said: “The contingency fund complements other food
security programs (PSNP, HABP) through taking timely actions to reduce the impacts of
shocks” [OR_R/KI-1]. Interviews with the SNNPR food security task force revealed that “it is
a fund used when shocks occur and affect non-PSNP beneficiaries in a given woreda and if
there is acute need” [SN_R/KI-1].
Respondents in interviews conducted at the regional level were asked: “How have the
contingency funds been used in this region?”. In Oromiya, the following response was given:

76
“The contingency fund is utilized in the form of 5 percent and 15 percent in the region”
[OR_R/KI-1]. This response indicates that the principle of the allocation of the contingency
fund, rather than the actual practice and use of the fund, has been applied in the region.
Only in SNNPR we were told that the contingency fund had been used to extend the months
of transfers for the PSNP beneficiaries and to solve problems related to inclusion and
exclusion errors.
Similar responses were given when these issues were discussed at the woreda level. In
SNNPR, the contingency fund was used to support both the PSNP and non-PSNP
beneficiaries. Support to PSNP beneficiaries includes extending months of transfers,
increasing new caseloads by applying full family targeting, and covering needs during
shocks and emergencies. The same support is also provided for the non-PSNP woredas in
case of emergencies. For instance, in Tembaro woreda of the SNNPR, “the contingency
fund is used to extend the payment for PSNP beneficiaries and to provide support to non-
PSNP households” [SN_T/W-KI-2].
Almost all of the woredas covered in this study utilized the 5 percent contingency fund
allocated to them during this year for emergencies. Interviews with early warning and
transfer experts generated the following responses to the question, “How did this woreda
use the PSNP 5 percent contingency fund this year?”:
To compensate for the price increase, to increase the number of months for
the PSNP beneficiaries, and to increase a new caseload like households who
are chronically food insecure and the non-PSNP beneficiaries [ORO_G/W-KI-
2].
To increase the new caseload through providing cash transfers up to 150
households who had already graduated from PSNP and were affected by
drought and to provide two months additional transfer for the PSNP public
works graduates [ORO_Z/W-KI-2].
To increase the new caseload to those PSNP households whose family
members were not fully targeted and to accomplish activities that are already
underway [TIG_A/W-KI-2].
Support kebeles that do not get the emergency support [AM_E/W-KI-2].
Used to compensate the impact of shocks for both the PSNP and non-PSNP
households [AM_SO/W-KI-2].
Apart from two woredas (Ebinet and Dembagofa), the remaining woredas requested a
contingency budget in addition to the 5 percent already allocated to address drought-related
shocks. Four woredas indicated that they had made a request for the additional contingency
fund since the beginning of the program implementation. However, Zeway Dugda woreda
indicated that it has requested only this year and Tembaro woreda had presented the
request in each of the last three consecutive years. Woredas that received additional
contingency fund used them in different ways. The key informant interviews with the
respective woreda early warning experts produced the following responses:

77
The fund was used to support 3,463 drought-affected people who are not
benefiting from the PSNP through involving them in public works and the
payment was made in cash [ORO_Z/W-KI-2].
We used to use the fund for extending the support periods of some PSNP
beneficiary households, such as HIV-affected families, elderly, OVC (Orphans
and Vulnerable Children), etc., which are not able to feed themselves for the
rest of the six months. In addition, if we have some more funds, then we
extend the support period of all PSNP beneficiaries for one to three months
[TIG_S/W-KI-2].
Yes, we got a response from the region and used the resources to assist the
PSNP beneficiaries whose family members were not targeted [AM_SA/W-KI-
2].
We used the 5 percent for PSNP beneficiaries whereas the 10 percent was
used for non-PSNP beneficiaries [SN_S/W-KI-2].
Key informant interviews with woreda early warning and transfer experts revealed that a
number of challenges were faced in using and allocating the contingency funds. These
included delays in transfer of funds, limited funds relative to needs, lack of technical support,
limited staff capacity to prepare contingency plans, and lack of time to implement planned
activities.
4.7. Summary
This chapter has assessed whether concerns about capacity have been addressed and
whether this has contributed to more timely transfers to beneficiaries. There are several
noteworthy findings:
Woreda level offices are increasingly well-resourced with trained staff. Nearly all now
use the PASS system and nearly all have functioning computers. Most, but not all, have
manuals that can be referred to. Training has occurred, although this could be more
widespread.
On average, it takes 38.9 days (as measured by the mean) or 32 days (as measured by
the median) from the time a woreda receives its first attendance sheet to the last
payment.
There is considerable variation in these times across woredas. The best performing
woredas manage to complete all activities associated with making payments in 21 days.
By contrast, the worst performers take nearly two months.
Both the qualitative and quantitative data point to several factors that explain woreda
level differences in the timeliness of payments. These include lack of training on the
PASS, the absence of front-loaded transfers, and lack of transport.
These factors are perceived to adversely affect beneficiaries as well as the smooth
operation of the PSNP.
The PSNP contingency fund principles and purposes are well understood at both
regional and woreda levels.

78
5. Kebele and Household Perspectives on Implementation
5.1. Introduction
The sustainability of any program—the PSNP included—depends in part on whether it is
successfully implemented. As with chapter 4, this chapter also focuses on implementation,
but here the perspective is largely that of the kebele and the household. It considers the
following topics. Are local administrative structures such as Kebele and Community Food
Security Task Forces in existence and functioning as envisaged in the Program
Implementation Manual? Are households aware of these Task Forces and do they
understand their role in the implementation of the PSNP? How do beneficiaries perceive
their experiences with the payment process? Have client cards been distributed? This
chapter draws largely on the quantitative community and household surveys. It addresses
the following evaluation objective described in the inception report (Table 5.1).
Table 5.1. Evaluation objectives covered in chapter 5
Evaluation
objective
Issue
Data sources
Links to Log Frames and
TOR
Document progress in the implementation of the PSNP
Are public works payments timely and
predictable? Do clients receive complete
entitlement?
Quant_HH
Quant_Comm
Focus
PSNP Log frame Output 1.1
PSNP Log frame Output 1.2
TOR, para 39
Are direct support payments timely and
predictable? Do clients receive complete
entitlement?
Quant_HH
Quant_Comm
Focus
PSNP Log frame Output 1.1
PSNP Log frame Output 1.2
TOR, para 35
Transfers are received at a place no more
than three hours from home
Quant_HH
PSNP Log frame Output 1.5
Assess trends in perceptions of the effectiveness and transparency of the PSNP and HABP among different
groups of clients
Beneficiaries understand how the
program works
Quant_HH
Focus
PSNP Log frame Output 4.4
Source: Authors’ compilation.
5.2. The Kebele and Community Food Security Task Forces
All kebeles should have a Food Security Task Force (KFSTF). Tasks undertaken by the
KFSTF include:
Community mobilization activities so as to identify and prioritize community needs;
Plan prioritized activities with community members;
Targeting beneficiaries and participants for public works and direct support with
input from the community;
Prepare a Kebele Safety Net Plan in consultation with woreda sectoral offices;
Maintaining minutes of KFSTF meetings on Safety Net issues, lists of participants,
and progress reports;
Establishing and training Community Food Security Task Forces; and
Participating in the monitoring and evaluation of safety net activities (GFDRE
2010).

79
At the local level, Community Food Security Task Force’s (CFSTF) are supposed to be
established. Their tasks include:
Identify potential participants for the PSNP;
Undertake needs assessments so as to identify those households who can
participate in public works and those without sufficient labor or other support who
will need direct support;
Ensure that the proposed list of participants is commented on and endorsed by
the general meeting of the village residents;
Finalize the list of participants and submit it to the KFSTF;
Prepare a pipeline of projects; and
Monitor periodically public works projects to ensure that they are undertaken as
prioritized.
Tables 5.2 to 5.5 document the existence and activities of the KFSTF and the CFSTFs,
drawing on information collected as part of the community questionnaires in 2006, 2008, and
2010.
Kebele FSTFs exist in all surveyed localities. Table 5.2 shows the composition of these
KFSTFs. Broadly speaking, on an ongoing basis, these match with what was set out in the
Project Implementation Manual (PIM). There appears to be a slight improvement in their
composition in Oromiya, where, in 2006, the KFSTFs deviated most from the guidelines set
out in the PIM. In nearly all (98 percent) surveyed kebeles, there is at least one woman and
at least one development agent on the KFSTF.
Table 5.3 describes aspects of record keeping undertaken by these task forces. In addition
to asking about whether certain types of records were kept, in 2006, 2008, and 2010,
enumerators asked to see these records in order to verify that they did, in fact, exist. (They
could not determine whether the records were accurate, only that they existed.) Participant
lists and minutes of meetings are well-kept but there appears to be a drop in the number of
KFSTFs that keep progress reports. Somewhat surprisingly, the percentage of cases where
these records were actually seen by enumerators drops substantially in all regions.

80
Table 5.2. Composition of the Kebele Food Security Task Force (KFSTF), by region
Percent of KFSTF that contain:
Region
Survey
year
Chairperson
of kebele council
A member of
the kebele council
An elected
representative
from elders
An elected
representative from
youth
Tigray
2006
83
80
75
86
2008
97
100
94
97
2010
74
83
81
89
Amhara
2006
97
97
76
88
2008
75
75
78
78
2010
94
83
80
86
Amhara -
HVFB
2006
-
-
-
-
2008
100
91
95
98
2010
90
92
87
87
Oromiya
2006
84
89
84
70
2008
75
84
81
81
2010
96
96
96
88
SNNPR
2006
94
89
81
80
2008
97
95
92
82
2010
100
91
94
83
Number of elected representatives from
women’s groups
Number of development agents
0
1
2
> 2
0
1
2
> 2
Tigray
2006
3
31
19
47
8
44
14
34
2008
0
55
19
25
0
50
11
39
2010
5
37
17
40
0
54
31
15
Amhara
2006
6
39
6
49
0
58
12
30
2008
11
14
28
47
8
47
6
39
2010
0
37
40
23
0
51
29
20
Amhara -
HVFB
2006
-
-
-
-
-
-
-
-
2008
0
27
9
64
0
41
14
45
2010
0
22
42
36
0
22
8
70
Oromiya
2006
6
44
39
11
6
64
25
5
2008
11
51
30
8
11
65
25
5
2010
0
13
48
39
0
39
31
30
SNNPR
2006
0
50
31
19
0
64
17
19
2008
8
44
23
26
3
62
15
21
2010
0
37
34
29
3
46
14
37
Source: Community questionnaire, 2006, 2008, and 2010.

81
Table 5.3. Kebele Food Security Task Force (KFSTF) record keeping, by region
Does the KFSTF keep:
Region
Survey
year
Records or
minutes of
meetings?
Records or minutes of
meetings and at least
one was seen?
Progress reports of
activities?
Progress reports of
activities and at least
one was seen?
Tigray
2006
92
92
92
92
2008
92
86
97
94
2010
83
33
69
35
Amhara
2006
79
76
82
78
2008
86
78
67
64
2010
86
42
64
57
Amhara -
HVFB
2006
-
-
-
-
2008
89
86
86
84
2010
95
77
76
66
Oromiya
2006
78
61
73
68
2008
68
59
73
62
2010
79
53
79
63
SNNPR
2006
83
72
86
80
2008
85
77
85
77
2010
77
48
54
53
Keep a list of:
When was the list of participants last updated?
Participants?
Participants and at
least one was
seen?
Not known
Before
January
January–
March
April–July
Tigray
2006
97
94
3
17
22
58
2008
97
97
3
14
25
58
2010
89
52
8
58
19
14
Amhara
2006
100
92
8
14
17
61
2008
92
86
6
22
29
43
2010
100
50
15
33
33
18
Amhara -
HVFB
2006
-
-
-
-
-
-
2008
98
98
5
22
30
48
2010
100
72
0
55
35
10
Oromiya
2006
89
83
18
4
13
65
2008
89
84
13
17
16
54
2010
96
70
18
32
36
14
SNNPR
2006
97
97
11
5
18
66
2008
97
97
5
8
18
69
2010
100
60
6
56
36
3
Source: Community questionnaire, 2006, 2008, and 2010.
5.3. Households’ Interactions with the Community FSTF
Table 5.4 shifts attention to the Community Food Security Task Force. Given its role as the
link between the FSP and the community, it is useful to begin by seeing whether
respondents were aware of the existence and function of the CFSTF. The sample is
disaggregated into three groups: households that had participated in the public works
component of the PSNP, households that had received any transfers under the direct
support component of the PSNP, and households that had received both work under PSNP
and direct support payments. Table 5.4 shows a consistent, but also somewhat surprising,
pattern across all regions. Between 2006 and 2008, knowledge of the CFSTF has increased
among the program beneficiaries, particularly in Tigray and Amhara. Table 5.4 also shows

82
that the percentage of program beneficiaries that are aware of the CFSTF and had contact
with the CFSTF increased significantly in all regions and among both kinds of beneficiaries
from 2006 to 2008. However, this pattern reverses between 2008 and 2010 with awareness
of and contact with CFSTFs at best remaining static and in some cases declining. One,
admittedly speculative reason for this could be that the frequency with which beneficiary lists
are being updated has declined (see Table 5.3), resulting in less contact between CFSTFs
and beneficiaries. In general, female-headed households were 5 to 13 percentage points
less likely, depending on the region, to have contact with the CFSTFs.
Table 5.4. Household contact with the Community Food Security Task Force (CFSTF),
by region, year, and beneficiary status
Tigray
Amhara
Amhara-HVFB
PSNP beneficiary status, received:
2006
2008
2010
2006
2008
2010
2006
2008
2010
Aware that the CFSTF exists (percent)
Public Works
76
92
86
70
89
86
-
-
96
Direct Support
62
83
64
45
79
57
-
-
78
Public Works and Direct Support
64
98
88
33
-
80
-
-
93
Aware that the CFSTF exists and had contact with the CFSTF (percent)
Public Works
44
75
81
49
83
81
-
-
78
Direct Support
17
57
54
21
69
59
-
-
55
Public Works and Direct Support
34
73
71
17
-
77
-
-
76
Oromiya
SNNPR
PSNP beneficiary status, received:
2006
2008
2010
2006
2008
2010
Aware that CFSTF exists (percent)
Public Works
79
86
78
88
91
87
Direct Support
54
66
50
84
76
77
Public Works and Direct Support
73
-
80
93
-
85
Aware that the CFSTF exists and had contact with
the CFSTF (percent)
Public Works
53
80
69
67
83
86
Direct Support
24
62
61
47
69
68
Public Works and Direct Support
46
-
89
80
-
83
Source: Household questionnaire: 2006, 2008, 2010.
Note: Cells are left blank if there were fewer than 10 responses.
Households were also asked to identify the purpose of the CFSTF and Table 5.5
summarizes their responses. Gilligan et al. (2009, Table 3.4a) showed that between 2006
and 2008 there was a significant increase in the proportion of PSNP beneficiaries who could
identify specific CFSTF functions. Table 5.5 shows the extent of this awareness using data
from the 2010 household survey, disaggregated by beneficiary status and region. When
compared to 2008, there is a decline in households’ ability to identify CFSTF roles, a finding
consistent with the reduced levels of contact seen in Table 5.4. Also note that respondents in
Oromiya had the greatest difficulty in identifying the role of the CFSTF and Oromiya is the
region where contact with CFSTFs is lowest. Female-headed households, and households
headed by individuals with no formal schooling, were less likely to be able to identify these
roles.

83
Table 5.5. Household knowledge of the role of the Community Food Security Task
Force (CFSTF), by region and beneficiary status
Tigray
Amhara
Amhara-HVFB
Oromiya
SNNPR
Percentage of households who indicated that one purpose of the CFSTF was to
mobilize the community for public works
PSNP beneficiary
71.7
74.4
82.9
59.8
79.3
Non-beneficiary
68.2
46.9
68.4
41.6
63.0
All
70.7
59.5
74.3
49.6
70.6
Percentage of households who indicated that one purpose of the CFSTF was to
identify individuals to participate in public works
PSNP beneficiary
62.0
54.8
70.9
48.5
65.2
Non-beneficiary
57.3
33.8
64.6
35.7
51.4
All
60.7
43.4
67.2
41.3
57.8
Percentage of households who indicated that one purpose of the CFSTF was to
identify direct support beneficiaries
PSNP beneficiary
62.3
51.8
67.9
40.7
62.1
Non-beneficiary
56.9
30.5
58.8
29.5
48.6
All
60.8
40.3
62.5
34.4
54.9
Percentage of households who indicated that one purpose of the CFSTF was to
monitor public works activities
PSNP beneficiary
64.5
76.9
72.2
58.5
71.8
Non-beneficiary
63.6
43.7
63.6
38.6
57.7
All
64.2
58.9
67.1
47.3
64.2
Source: Authors’ calculations based on the PSNP survey data.
5.4. Households’ Experiences with the Payment Process
Output 4.4 in the PSNP Log Frame states that beneficiaries should be able to understand
how the program works. In both 2008 and 2010, PSNP beneficiaries were asked directly if
they felt that they had received all information needed to understand how the program
works. In 2008, this was asked as a Yes/No question, while in 2010, respondents could say
if they strongly agreed, agreed, disagreed, or strongly disagreed. In 2008, between
55 percent (Oromiya) and 80 percent (SNNPR) of respondents answered “yes.” If we
assume that “strongly agree” and “agree” are equivalent to “yes,” then Table 5.6 shows that
there have been noticeable improvements in understanding in Amhara (from 65 to 83
percent), Amhara-HVFB (74 to 84 percent), and SNNPR (from 80 to 88 percent). There are
slight declines in understanding in Tigray and a considerable percentage of beneficiaries in
Oromiya (45 percent in 2008 and 47 percent in 2010) do not feel that they have sufficient
information. The Oromiya results are consistent with the findings reported in Table 5.4,
showing that contact with the Community FSTFs had declined in that region between 2008
and 2010. Also consistent with lowered rates of contact is the fact that, as shown in Table
5.6b, female-headed households and households where the head had no formal schooling
were slightly less likely to strongly agree that they had sufficient information to understand
how the program works.

84
Table 5.6a. Beneficiaries received all information needed to understand how the
program works (percent), by region
2008
2010
(Percent saying yes)
Strongly agree
Agree
Disagree
Strongly disagree
Did not state
Tigray
74
22
47
14
7
9
Amhara
65
23
60
12
3
1
Amhara-HVFB
74
38
46
11
3
2
Oromiya
55
18
35
22
15
10
SNNPR
80
47
41
8
3
2
Source: Household questionnaire 2008, 2010.
Table 5.6b. Beneficiaries received all information needed to understand how the
program works (percent), by sex and schooling of household head
2010
Strongly agree
Agree
Disagree
Strongly disagree
Did not state
Sex of head
Male
30.4
45.4
12.4
7.1
4.8
Female
26.9
46.4
15.9
4.9
5.9
Education of head
No schooling
29.3
44.8
14.1
6.5
5.4
Any formal schooling
35.9
44.5
10.1
6.3
3.2
Source: Household questionnaire 2010.
Both public works and direct support beneficiaries were asked about their perceptions
regarding payments and the payment process. This included whether they thought they had
been paid in full and on time. In 2010, they were asked how many days’ notice they were
given that they would receive a payment. In both 2008 and 2010, beneficiaries were asked
how they had been treated when receiving their payment. Results are reported in Tables
5.7–5.10. As with the results reported in Table 5.6a, in 2008, these were asked as Yes/No
questions, while in 2010, respondents could say if they strongly agreed, agreed, disagreed,
or strongly disagreed.
Table 5.7. Beneficiaries’ perceptions regarding being paid in full (percent)
2008
2010
(% saying yes)
Strongly agree
Agree
Disagree
Strongly disagree
Did not state
Tigray
72
19
45
16
12
8
Amhara
74
17
59
16
8
1
Amhara-HVFB
90
42
48
7
3
0
Oromiya
58
14
30
30
17
9
SNNPR
90
43
39
10
6
1
Source: Household questionnaire 2008, 2010.

85
Table 5.8. Beneficiaries’ perceptions regarding the timeliness of payments (percent)
2008
2010
(% saying yes)
Strongly agree
Agree
Disagree
Strongly disagree
Did not state
Tigray
17
8
27
34
22
8
Amhara
40
9
35
38
16
1
Amhara-HVFB
52
32
36
24
8
1
Oromiya
15
5
9
41
37
9
SNNPR
53
27
27
28
16
1
Source: Household questionnaire 2008, 2010.
Table 5.9. Number of days’ notice that beneficiaries had that payments would be made
(percent)
Tigray
Amhara
Amhara-HVFB
Oromiya
SNNPR
1 day
68
83
72
76
72
2-3 days
21
14
24
16
25
4-5 days
2
1
2
4
2
More than 5 days
8
2
2
4
1
Source: Household questionnaire 2010.
Table 5.10. Beneficiaries’ perception that they were treated courteously (percent)a
2008
2010
(% saying yes)
Strongly agree
Agree
Disagree
Strongly disagree
Did not state
Tigray
79
19
51
13
7
10
Amhara
92
24
65
6
4
1
Amhara-HVFB
92
43
47
5
3
2
Oromiya
77
23
37
17
13
10
SNNPR
95
43
42
8
4
2
Source: Household questionnaire 2008, 2010.
Note: a/ There was no difference in the proportions of male- and female-headed households who reported being treated
courteously.
In Amhara-HVFB, respondents are most likely to perceive, in both 2008 and 2010, that they
have received their payments in full. About 82 percent of respondents in SNNPR have this
perception in 2010, a decline from the 90 percent recorded in 2008. Just over half of PSNP
beneficiaries in Oromiya think that they are receiving their full payments in 2010, a figure
unchanged since 2008. (We return to the issue of payments in full in chapter 8.) In Oromiya,
considerable concern was also expressed about timeliness of payments. Only 14 percent of
beneficiaries agree or strongly agree that they receive payments on a timely basis and 37
percent strongly disagree. While Oromiya is the worst performing region by this measure,
beneficiaries clearly perceive that there is considerable scope for improvement in the
timeliness of payments.
14
Related to this issue, Table 5.9 indicates that, as a general rule,
14
Regional level discussions indicated that there may have been some issues specifically relating to the 2010 payment cycle
which affected the timeliness of payments. In Oromiya, it was indicated that payments were delayed to those woredas that
had been tardy in the distribution of client cards. Tigrayan authorities noted that lack of grain availability led to payment delays
in some localities.

86
beneficiaries have no idea when their payments are coming. Between 68 and 83 percent
receive only one day’s notice that they will be paid.
15
Timeliness of payment elicited considerable comments during the qualitative fieldwork.
Twenty-four out of 30 focus group discussions (FGD) came to consensus that payments
were made irregularly and often they did not know from one month to the next when to
expect payment.
16
This causes problems for beneficiaries in terms of planning expenditure,
loan defaults, and lack of food.
We got paid 3 months in cash and 3 months in food. There is a delay in time
of the payments. For instance, PSNP starts in January every year but
payment delays up to March [TIG_A/FG-1].
The payment schedule is monthly, but there is delay in pay times, even up to
3–4 months [TIG_A/FG-3].
We are paid every 2–3 months. The payment time is not predictable and
timely information not given. We receive when transfer comes [ORO_G/FG-
4].
At the start [of the PSNP] it was regular. Now it is coming every three months;
however, it does not come timely. As a result we do not buy assets but rather
directly purchase food. In short, we are paid three times [ORO_Z/FG-3].
No clue [when we will receive next payment]. We are not informed when the
next payment will be [ORO_Z/FG-3].
We do not know when the payment is coming. We only know the arrival when
it is announced. We prefer to be paid monthly. If the payment could have
been paid without delay each month, this would have been the basis for our
growth [SN_D/FG-3].
We get our payments accidentally, in an unpredictable way [SN_S/FG-1].
We do not know when payments are to be made. We are told suddenly that
we are going to get paid. If it were given monthly, we would have been able to
do what we planned [SN_S/FG-1].
The few that mentioned that there was some predictability to payments said that payments
were usually made every 2–3 months. This is strikingly consistent with the quantitative data
reported above. The majority of people who responded to the question about preferred
payment arrangement indicated that a monthly payment was best. In about five focus group
discussions, respondents said they would like bimonthly payments. Overall, the timing of
payment caused less concern than the lack of predictability of payment.
When asked, “Do you know when you will receive your next payment?,” the majority of
responses indicated that there was a lot of uncertainty when the next payment will arrive.
Twenty-four out of 26 focus group discussions that answered this question said that they did
15
Participants in the regional workshop in Tigray provided a caveat to this finding. They noted that in Tigray, beneficiaries are
given an approximate date about a week before payments are made; this date is subsequently confirmed. Where this
approach is implemented, our data will understate the amount of notice that beneficiaries receive.
16
The section on timeliness and frequency of payments was excluded from the “transitioning” focus groups due to the length of
the survey instrument for this group.

87
not know the date. Only two groups indicated that they expected the next payment to be
made on a specific date. There was much more confidence in the amount of the payment
(food or cash) to be received: “The current level of payments are good; the problem is the
delay” [SN_D/FG-1]; “We do not know the time, but regarding how much, we know the usual
amount” [SN_S/FG-1]. In fact, most respondents indicated that they knew the payment
amount in advance.
In the (frequent) event that payments do not arrive near an expected date, many
respondents said that they resorted to borrowing and taking loans as a coping strategy. The
sale of assets to finance the time between expected payment and actual receipt was
mentioned in seven focus group discussions.
The delay in payment makes our lives difficult, particularly the households
with no reserves, as they go out to borrow from others [TIG_A/FG-1].
Some families who can afford from other sources may not be affected by the
delay as such. But in other households, we know that some sell assets such
as sheep, goats, or go out for credit and loans to sustain the periods of delay
[TIG_A/FG-3].
If we do not get payment on time, we will be forced to take credit to consume
food and we will be psychologically forced to be in tension [TIG_S/FG-4].
We are paid once in three months. But this leads to grain price increases due
to high money injection into the limited local market. We do not know the time
of the next payment, but we know how much we will receive [ORO_Z/FG-4].
When they are paid, beneficiaries are—for the most part—treated courteously by program
staff. Table 5.10 shows that only in Oromiya a relatively large percentage of respondents
disagree (17 percent) or strongly disagree (13 percent) with the statement, “I was treated
courteously by the staff.” However, the results of Tables 5.8 and 5.9 should be viewed with
concern, given that timely and predictable payments are listed as an Output (1.1) in the
PSNP Log Frame. Table 5.7 suggests that there may be issues associated with receipt of
complete entitlements too. This issue appears as Output 1.2 in the PSNP Log Frame.
Table 5.11 provides descriptive statistics on beneficiaries experiences when travelling to the
payment site. Outside of SNNPR, beneficiaries typically walk
17
12 to 16 kilometers to the
place where they will be paid. Given that beneficiaries have to cover this same distance
when they return home, and given that being paid requires long periods of standing in line, it
is not surprising that many beneficiaries must sleep at the payment site. The fact that 84
percent of beneficiaries incur no costs when travelling to receive their payment, and that the
average cost in all regions is usually less than one day’s wage, this must be put in context.
These costs are low because (outside of SNNPR) beneficiaries are walking 25–32
kilometers and are sleeping in the open. Fortunately, reports of harassment or robbery
during the travel to or from the payment site are quite low. This is true for both male- and
female-headed households.
17
Ninety-three percent of beneficiaries walk to the payment site.

88
Table 5.11. Beneficiaries’ experiences with travel to the payment site
Region
Survey
year
Distance to
payment
site
Average cost
incurred for
travel/overnight
stay at payment
site
Beneficiaries
had to stay
overnight at
the payment
site
Beneficiaries
were subject to
harassment
during the travel
to/ from the
payment site
Beneficiaries
were robbed of
food or cash
during the
travel to/from
the payment
site
(kilometers)
(Birr)
(percent)
(percent)
(percent)
Tigray
2008
11.4
2.8
39.5
3.5
3.7
2010
12.4
7.0
39.6
1.6
1.5
Amhara
2008
12.2
2.5
31.4
1.2
0.4
2010
14.2
12.4
26.9
2.6
2.0
Amhara –
HVFB
2008
18.0
8.1
66.8
3.4
3.6
2010
16.6
8.5
52.5
2.7
2.9
Oromiya
2008
13.9
3.6
27.8
2.6
1.5
2010
11.6
15.6
27.6
2.8
1.3
SNNPR
2008
5.5
0.5
6.3
6.7
4.5
2010
8.4
6.7
12.3
4.1
1.6
Source: Household questionnaire 2008, 2010.
Beneficiaries are supposed to receive their transfers in a place no more than three hours
from home (PSNP Log Frame, Output 1.5). Figure 5.1 shows that this goal is largely met in
Tigray, Oromiya, and SNNPR. However, a considerable proportion of beneficiaries in
Amhara and Amhara-HVFB must travel more than three hours.
Figure 5.1. Travel times to payment sites
Source: Authors’ calculations based on the PSNP survey data.

89
5.5. Client Cards
An important innovation in the newest phase of the PSNP has been the provision of client
cards. These cards serve several functions. They are a means of allowing beneficiaries to
keep track of payments they receive while also providing a brief summary of important
features of the program. Client cards are supposed to be free.
Table 5.12 shows the proportion of households that reported that they had participated in
PSNP public works and had received a client card. As part of the process of verifying this
information, enumerators asked respondents to show the card. Approximately 75 percent of
the households who reported holding a client card were able to do so.
Table 5.12. Percentage of public works participants who reported holding a client
card, by region
Tigray
Amhara
Amhara - HVFB
Oromiya
SNNPR
Total
Beneficiary has client card
21.8
38.6
26.4
59.6
78.0
43.6
Source: Authors’ calculations based on the PSNP survey data.
There is clearly significant regional variation in the distribution of these cards, with
distribution significantly more advanced in SNNPR and Oromiya, and the other regions
lagging. Households were also asked how much they were supposed to pay for the cards
and how much they had actually paid. Nearly all beneficiaries—95 percent—indicated that
they thought the cards should be free. Across the sample, however, only 79 percent
indicated that they had received these cards for free—as they were supposed to. But this
statistic masks considerable regional variation. Figure 5.2 shows that more than 90 percent
of households in Tigray and Oromiya reported that they had not paid anything to receive
their cards. Small payments (between 1 and 5 Birr) were reported by about one-fourth of
public works participants in Amhara and SNNPR, and a few households paid between 10
and 15 Birr. However, regional level discussions indicated that there may have been some
confusion surrounding this question. In some localities, if a beneficiary was not present when
photographs for these cards were taken, they were required to provide a photograph at their
own expense. It is possible that some of these reported expenses may have been for the
photographs and not the cards themselves.

90
Figure 5.2. Expenses associated with the receipt of client cards, by region (Birr)
Source: Authors’ calculations based on the PSNP survey data.
5.6. Summary
This chapter has assessed the implementation of the PSNP based on information provided
at the kebele and household level. It has assessed whether local administrative structures
have been established and whether these are functioning properly. It describes beneficiaries’
perceptions of the payment process and examines the extent to which payment cards have
been made available. Key findings are the following:
Kebele FSTFs exist in all surveyed localities. Broadly speaking, these match with what
was set out in the Project Implementation Manual. There appears to be a slight
improvement in their composition in Oromiya, where, in 2006, the KFSTFs deviated most
from the guidelines set out in the PIM. In nearly all (98 percent) surveyed kebeles, there
is at least one woman and at least one development agent on the KFSTF. Participant
lists and minutes of meetings are well-kept, but there appears to be a drop in the number
of KFSTFs that keep progress reports.
In both 2008 and 2010, PSNP beneficiaries were asked directly if they felt that they had
received all information needed to understand how the program works. There have been
noticeable improvements in self-reported understanding of the program in Amhara (from
65 to 83 percent), Amhara-HVFB (74 to 84 percent), and SNNPR (from 80 to 88
percent). There are slight declines in understanding in Tigray and a considerable
percentage of beneficiaries in Oromiya (45 in 2008 and 47 in 2010) who do not feel that
they have sufficient information about the PSNP.
In Oromiya, considerable concern was expressed about timeliness of payments. Only 14
percent of beneficiaries agree or strongly agree that they receive payments on a timely
basis and 37 percent strongly disagree. While Oromiya performs poorly by this measure,
it is worth noting that this concern is voiced by beneficiaries in all regions.

91
Beneficiaries are supposed to receive their transfers in a place no more than three hours
from home. This goal is largely met in Tigray, Oromiya, and SNNPR. However, a
considerable proportion of beneficiaries in Amhara and Amhara-HVFB must travel more
than three hours.
Across all surveyed beneficiaries, 43.6 percent report having received a client card.

92
6. Targeting
6.1. Introduction
The PSNP is a targeted program where the targeting methods used embody a mixed set of
approaches that include both administrative and community components. This chapter
covers the following topics. How is targeting supposed to work in the PSNP? How is this
understood and implemented at the regional level? How is this understood and implemented
at the woreda and kebele levels. (As part of this discussion, it examines whether pregnant
women are moved from public works to direct support.) How is targeting understood at the
household level? Who actually participates in the PSNP? In public works (PW) and direct
support (DS)? How consistent is this with the Project Implement Manual (PIM)? Do
households actually participate in the PSNP for three consecutive years? Does this
contribute to predictability in transfers? In so doing, it covers a number of evaluation
objectives as Table 6.1 notes.
Table 6.1. Evaluation objectives covered in chapter 6
Evaluation
objective
Issue
Link to Log Frames and TOR
Document progress in the implementation of the PSNP
Are public works payments timely and predictable?
Do clients receive complete entitlement?
PSNP Log frame Output 1.1
PSNP Log frame Output 1.2
TOR, para 39
Are direct support payments timely and predictable?
Do clients receive complete entitlement?
PSNP Log frame Output 1.1
PSNP Log frame Output 1.2
TOR, para 35
Pregnant female participants are moved from PW to
DS
PSNP Log frame Output 1.4
Households participate in PSNP for at least three
consecutive years
PSNP Log frame Output 1.6
Can gender dimensions of access be better captured?
TOR, para 42
Assess trends in perceptions of the effectiveness and transparency of the PSNP and HABP among different
groups of clients
Beneficiaries understand how the program works
PSNP Log frame Output 4.4
Beneficiaries and non-beneficiaries report that
targeting and graduation processes are fair
PSNP Log frame Output 4.5
Source: Auhtors’ compilation
6.2. Targeting Principles in the PSNP
Targeting under the PSNP combines both categorical and individual selection using both
administrative and community mechanisms. Administrative mechanisms include the
provision of a specified number of clients that can be included within a specific administrative
area (woreda, kebele, etc.); guidance on targeting criteria to be used at the community level;
and oversight to ensure transparency and accuracy. Oversight ensures upward
accountability, through regional oversight of woredas and woreda oversight of kebeles.
Accountability relations work downward as well through community targeting, which includes

93
the identification of clients by community Food Security Task Forces (FSTFs) and
verification of the client list in a public meeting, leaving open the possibility for appeals and
complaints. The PIM specifies that households who are targeted should fall into the following
categories:
be community members;
have faced continuous food shortages (three months of food gap or more per
year) in the last three years;
be acutely food-insecure due to a shock resulting in the severe loss of assets;
and,
lack adequate family support and other means of social protection and support
(GFDRE 2010, 24).
The PIM specifies supplementary criteria to assist communities in targeting:
Status of household assets: landholding, quality of land, food stock, labor
availability;
Income from agricultural and nonagricultural activities; and,
Specific vulnerability, such as female-headed households, households with
chronically-ill members, and elderly-headed households looking after orphans.
Targeting rules in the PIM distinguish between households having adequate household
labor, which should be registered for public workfare projects as a requirement to receive
their transfers, and households with labor constraints that are unable to contribute to public
works. The latter receive “direct support” and are not required to work.
According to the PIM (GFDRE 2010), public works participants include men and women over
the age of 16 who are able-bodied. Pregnant women after four months and lactating mothers
10 months after the birth of their child are exempt from public works. They should be shifted
to direct support if there is no able-bodied labor in their household to compensate for their
labor contribution. Where there is household labor available, the work requirement of
another able-bodied household member is increased up to the labor cap of 15 days.
The PIM specifies a number of administrative criteria to determine individuals who should
receive unconditional transfers, including those who are under the age of 16, the elderly,
disabled, and people who are temporarily unable to work, such as those who are sick,
pregnant women after four months and lactating mothers up to 10 months. Government and
donor officials have worried that direct support beneficiaries have been asked to contribute
to public works, although this clearly runs against the PIM. However, in all but one woreda,
officials indicate that direct support beneficiaries do not contribute to public works. In
Gursum, Oromiya state, the Public Works Focal Point explained that it is common for direct
support beneficiaries to provide childcare at public works sites and prepare coffee for
workers, but that they do not provide physical labor.
The revised PIM (June 2010) introduced a new targeting rule that all members of eligible
PSNP households should be listed as clients of the program. This rule is known as “full
family targeting” (FFT). The reasoning behind introducing the new rule was to help client
households to graduate by providing a transfer for every household member and prevent

94
dilution of transfers. Along with FFT, the targeting rules for the PSNP incorporate a “labor
cap.” This covers households with labor but who have some members who are unable to
work. The able-bodied household member(s) are required to work additional days on behalf
of those who cannot work in order to help earn their transfer. According to the labor cap rule,
the able-bodied household member cannot work more than 20 days in a month. These
households are targeted for public works assistance, even though some of the household
members may be unable to work.
Lastly, it should be noted that, in principle, the budget for the PSNP should flow from
targeting assessments. The PSNP PIM indicates that the available budget is divided
between regions according to the estimated number of beneficiaries and the woredas (the
community as well) have the opportunity to review the budgets through the upward stream
budget preparation process.
6.3. Regional Perspectives
Regional officials explained that they follow the official guidance and instructions on
targeting. In all regions, a mix of administration and community targeting approaches are
used. They affirmed the importance of community involvement in determining household
clients. An official on the regional FSTF in Oromiya commented:
We have learned that it is important to include traditional leaders and elders in
determining the targeting criteria because doing so ensures that these match
local opinions of who needs support. Involving community leaders in targeting
decisions also helps to enhance the transparency of the process [ORO-R-FG-1].
However, they made clear—albeit indirectly—that following the targeting guidelines would
not guarantee that all those identified as being eligible would actually receive assistance. In
four separate key informant interviews, regional officials were asked how they arrived at the
total number of the PSNP beneficiaries in their respective regions. They indicated that the
PSNP woredas were already identified and the quota was already set by the federal
government. Quotations taken from interviews in Oromiya and Tigray illustrate this:
In principle we are mandated to fix the quota for the woredas and zones.
However, in practice, the federal government is determining the quota for the
woredas and zones. The federal government does this based on the available
resources. We only channel what is allocated by the federal government to
woredas and zones [ORO_R/KI-1].
As of 2006, each woreda has a long list of food-insecure cases which were
under the food aid programs for many years [on average for 10 years these
people were under food aid]. These are the beneficiaries for PSNP and hence
planning is based on this data [TIG_R/KI-1].
SNNPR officials, however, indicated that they had adopted a slightly different approach to
define the total number of beneficiaries. The excerpt from the discussion with the regional
food security revealed this:

95
The [PSNP] woredas were already identified by the federal office but we had
conducted assessments and included some woredas. As a result, the number
of beneficiaries increased from 760,000 to 900,000 between 2005 and 2007
[SN_R/KI-1].
Consequently, regional officials attributed most exclusion errors not to poor targeting but
rather to inadequate budgets. An official of the regional FSTF in Oromiya stated:
A significant number of chronically food-insecure households were left outside
PSNP due to the limited quota [from the federal allocations to regions]
[ORO_R-FG-1].
Regional officials also noted a tension that exists between wider coverage of the program
that would allow inclusion of more households and the principle of full family targeting. They
also recognized that full family targeting was a means of ensuring that the PSNP was more
likely to be successful:
Consumption smoothening and the protection of a household’s assets is
undermined when all family members are not targeted in an attempt to reach
more households [ORO_R-FG-1].
[Without full family targeting] there will be dilution. You cannot achieve what
you want to. If there are households that receive payment for five members
while they have seven members, they will share the payment. As a result, the
food gap for these households is extended and they may be pushed to sell
their assets to make up for this. If they do not have land or other assets, they
won’t have physical strength. Their children won’t go to school [SN_R-FG-1].
6.4. Woreda Perspectives
Woreda officials indicated that primary responsibility for targeting lies with communities.
Woreda officials confirmed that they provide orientation and training to kebele FSTFs on how
to carry out community level targeting, including familiarizing kebele officials with the
targeting criteria that are in the PIM. They also review and approve beneficiary lists
submitted by kebele FSTFs.
Key informant interviews and focus group discussions held at the woreda level probed
attitudes and knowledge of specific targeting practices. In some localities, it was noted that
in initial targeting exercises carried out in 2005, there were cases where “model farmers”
were prioritized because they were thought to perform better and more likely to achieve
graduation. In Ebinet woreda, Amhara state, an official on the Woreda Food Security Task
Force explained that better-off farmers were targeted until 2006—when new targeting
guidelines were issued—whereas the poor were targeted for resettlement. It is unclear how
widespread this problem was. Still, the little evidence uncovered in our interviews suggests
that inclusion errors were not a serious problem. In Gursum woreda, Oromiya state, officials
noted that 400 better-off households were initially targeted before retargeting was carried
out. An official on the woreda FSTF in Sekota, Amhara state, estimated that fewer than 5
percent of all beneficiaries registered in the initial targeting exercise were better-off. After

96
new targeting guidelines were issued, the “poorest of the poor” were emphasized in
retargeting carried out during 2007.
All woreda focus groups reported that the PSNP targets “the poor.” While some groups
specified more precise criteria of who might be targeted, such as the landless and disabled,
most groups offered no more precise distinctions of who might be considered to be “poor.”
Although better-off community members are, in general, not regarded as being entitled to
PSNP support, there was recognition that households regarded as better-off were included
on beneficiary lists in some areas. In Demba Gofa woreda, SNNPR state, when asked if
there were any exclusion or inclusion errors, an official on the FSTF responded:
Since kebele leaders facilitate targeting, they do not include some poorer
individuals, so that their own relatives can be covered [S_D/W-FG-1].
When asked whether better-off individuals had been included, another official on the same
woreda FSTF stressed:
There were some areas in which the kebele chairman was included although
he was not poor. Everyone thinks that they deserve it. They [community
members] do not know about the program [targeting criteria] [S_D/W-FG-1].
Public works officials in six of ten woredas stated that the disabled, chronically ill, and elderly
were exempt from public works requirements. Woreda level officials in five areas stated that
children were exempt. However, there was some confusion evident about the age under
which children should not work. Officials in three woredas stated that children under 18 are
exempt, whereas a public works official in one woreda said children under 15 are exempt.
We encountered one anecdote about children working on public works sites. A public works
official in Shebedino woreda, SNNPR state, revealed that some children do work in spite of
entreaties by woreda officials that they should not work:
There are occasions where kebele officials allow the elderly to participate in
public works when we have told them not to do so. We have told them not to
allow children to work, as well. However, when the father is not able to work,
public works households will send their children to work out of fear that they
will be punished if they do not show up and fulfill their household’s labor
contribution. Such types of problems do exist [SN_S/W-FG-PW].
A public works official in Sayint woreda, Amhara state, stated that although children under
15 are exempt, orphaned children are required to work. No explanation was given and the
assessment came over with no other evidence to substantiate this claim.
Pregnant women and lactating mothers are categories exempt from undertaking public
works according to all woreda level public works officials that were interviewed. However,
there is some variation in the timing of when pregnant women are permitted to stop
attending public works as well as when new mothers are expected to return. In three
woredas, public works officials stated that pregnant women over three months are exempt.
In three other woredas, officials stated that women over six months are exempt. In one
woreda, a public works official stated that women are exempt once they visibly show signs of
being pregnant.

97
As with the discussions held at regional level, woreda respondents were asked about
quotas. In nine out of ten woredas, officials of the woreda FSTFs acknowledged that they
receive a quota from the regional level. In turn, woreda FSFTs determine a quota for each
kebele. A range of criteria is used to determine quotas including population size,
agroecological conditions (rainfall levels and farming potential), average size of
landholdings, levels of malnutrition, and the estimated size of the chronically food-insecure
population. There is a tendency at the woreda level to target all kebeles, although officials in
some woredas suggested they exclude a limited number of kebeles that are regarded as
higher potential farming areas with smaller chronically food-insecure populations. A FSTF
official in Shebedino woreda explained that they excluded kebeles that were comparatively
better-off as a way of working with a smaller quota:
There are kebeles that are more prone to shocks. There are a total of 19
PSNP kebeles (out of 32 rural kebeles and 2 urban kebeles). Our attention is
focused on dwellers in lowland kebeles (kola) rather than highland areas
(weynadega), since these areas are lagging behind [SN_S/ W-FG-1].
In light of these quotas, officials on woreda FSTFs were asked how they covered the needs
of chronically food-insecure households who met the PSNP eligibility criteria but were not
targeted. Responses varied. In Oromiya state, officials in Gursum and Ziway Dugda woredas
explained that they used the contingency fund to cover households that were not covered. In
Gursum, officials were able to provide transfers for three months to chronically food-insecure
households that were excluded from PSNP targeting. The contingency budget (allocated by
the region) was also used to cover excluded households that were eligible in Sayint woreda,
Amhara state, and Saesi Tseda Emba woreda in Tigray. Another approach was to base
targeting decisions on local criteria, which tended to be more specific than the criteria in the
PIM. For example, in Dimtu Rareti kebele in Ziway Dugda woreda, a development agent
noted that a household is excluded if they own a pair of oxen, regardless of other indicators
of their poverty and food security. In Saesi Tseda Emba woreda, a FSTF official explained:
We prioritize households according to their level of poverty and income as the
main tool to differentiate between eligible and non-eligible households. But in
our effort to reach the poorest of the poor, we cannot help it if some
households who actually deserve to be included in PSNP are left out. This is
just because of the limited quota. Eligible households are left out due to
limited resources; it is a tradeoff [TIG_S/W-FG-1].
Some interviews at the woreda level explored whether the concept of full family targeting
was ignored in order to increase the number of beneficiary households. Given the newness
of this, officials were asked to describe how they understand full family targeting and if the
rule was being followed. FSTFs in all woredas were knowledgeable of FSTF. All kebele
FSTFs that were asked affirmed they had heard of full family targeting. A FSTF official in
Gursum woreda, Oromiya state, explained:
Full family targeting requires that we include all the members of targeted
households, although this means that a limited number of households are
covered by the program. It leads to the exclusion of eligible households.
Targeting a greater number of households means that more households are

98
supported. But it leads to partial family targeting and this compromises the
consumption smoothening objective of the PSNP [ORO_G/W-FG-1].
A FSTF official in Saesi Tseda Emba woreda, Tigray state, noted:
We know about full family targeting and have practiced it since last year. In
the previous round of the PSNP, the tendency was to reach more households
than reach all the members within a household. By trying to reach many
households, we stretch the program and help as many people as possible to
survive hard times. But this holds back the main purpose of program. It delays
households from graduating. Some might not graduate. On the other hand,
full family targeting will quicken graduation, even though many poorer
households that are eligible for support are left out [TIG_S/W-FG-1].
Although many officials support the principle of full family targeting, there are tensions. In
one woreda (Gursum, in Oromiya state), FSTF officials acknowledged that they target fewer
members of client households, even though this runs against the principle of full family
targeting. A FSTF official in Ahferom woreda, Tigray state, recounted:
In some places (kebeles), even the selection committees swear to each other
not to disclose the confidentiality of breaking rules of full family targeting,
because they prefer to reach as many households as possible in the name of
sharing what they have with the community [TIG_A/W-FG-1].
Officials in Gursum woreda worried that full family targeting would promote dependency by
encouraging larger families. One FSTF official stressed:
We have not pushed kebele officials to adhere to full family targeting because
we want to discourage people from increasing the size of their families.
Children between the ages of 1 and 5 were not included into the program as a
result [ORO_G/ W-FG-1].
Lastly, public works officials in nine of ten woredas confirmed that they enforce a labor cap
(there was one unclear response by officials in Tembaro woreda, SNNPR state). However,
the labor cap in all areas is 20 days/month/laborer, higher than the 15 days specified in the
PIM. In Tigray, officials explained that there is a different labor cap for men (20 days) and
women (15 days).
6.5. Kebele Perspectives
Key informant and focus group discussions indicated that the provision of a quota extends all
the way down to the lowest administrative level, the kebele. FSTF officials and development
agents in all kebeles visited confirmed that the kebele quota is divided among sub-kebeles,
following a similar process of assessing population size, livelihoods, and wealth in different
areas. Community FSTFs prepare a list of beneficiaries based on the quota they receive and
development agents train community level officials on targeting procedures. All focus groups
confirmed that community members participate in targeting by verifying and agreeing on the
list of clients. In nearly all kebeles, members of the community FSTF, which are elected by

99
the community, determined the initial list of beneficiaries. A public meeting is called and the
list is read and subsequently adjusted according to the feedback and views of community
members. There were few indications that the process at this level was exclusive, with the
exception of the women’s group in Wal kebele of Sekota woreda, Amhara state, who insisted
that “low class” women stayed away from the community targeting meeting. The beneficiary
list is submitted to the kebele FSTF for review and approval. Any questions and concerns
are referred back to the community FSTF, although few instances of this happening in
practice were mentioned by kebele officials. Kebele FSTF officials gather beneficiary lists
from all sub-kebeles and submit these to the woreda FSTF.
Respondents in four focus groups indicated that kinship, including affinity to community and
kebele leaders, influenced targeting outcomes. When asked if better-off community
members were targeted, a development agent in Dimtu Rareti kebele in Ziway Dugda
woreda, Oromiya state, noted:
Aba gare (the community headman) has the upper hand in the targeting
process and this has at times undermined the transparency of targeting
because of favoritism and the inclusion of individuals that are not eligible
[ORO_Z/K-KI-2].
A development agent in Furra kebele of Shebedino woreda, SNNPR state, maintained that
kebele leaders included their wives on beneficiary lists. When questioned if the better-off had
been targeted, an official on the kebele FSTF in Soyame kebele of Tembaro woreda,
SNNPR state, remarked:
[You ask] why did you include those who should not have been targeted?
Well, when you work with the community you cannot be 100 percent without
sin! [S_T/K-KI-1].
Thus, although it is widely recognized at the community level that the poorest should be
targeted, there are reported instances of favoritism and nepotism by elites at the community
level. Further, in some areas, there were exclusion errors because of absenteeism at
targeting meetings and because community members were not fully informed about the
wealth status of their neighbors. When asked if better-off community members had been
included on beneficiary lists, a development agent in Soyame kebele explained:
Yes, they have. There were problems with targeting and some were excluded
who should have been listed. Targeting was done without a proper wealth
ranking exercise being carried out. Wealth ranking was done later (after the
initial targeting), but many were not covered because of the community vote.
The community voted on those to be included but they did not have reliable
data on which to base their decisions. Others were excluded because they
could not make it to the targeting meeting for reasons such as being ill
[S_T/K-KI-2].

100
In Adizata kebele of Ahferom woreda, Tigray state, a development agent similarly indicated
that exclusion errors occurred due to problems with wealth indicators used by communities
to select beneficiaries:
Only simple proxy indicators of wealth were considered and this meant that
some who would be eligible were excluded. Also, in-migrants were excluded,
regardless of their food gap [TIG_A/K-KI-2].
To what extent do these exclusion errors reflect quotas that are too small to meet the needs
of all those households who meet criteria for inclusion? In six of ten kebeles visited as part of
the qualitative study, officials stated explicitly that an insufficient quota contributed to
exclusion errors. For the public works component of the PSNP, this claim matches closely
with information found in the community quantitative survey. This component of the
quantitative fieldwork included a set of questions on the adequacy of resources given to
kebeles to meet the needs of both public works and direct support. Results are reported in
Tables 6.2 and 6.3.
Table 6.2. Kebele perceptions of adequacy of resources for public works participants
Tigray
Amhara
Amhara-HVFB
Oromiya
SNNPR
Sufficient
33.3
25.6
35.0
7.1
38.9
(12)
(10)
(14)
(2)
(14)
Insufficient
66.7
66.7
65.0
78.6
58.3
(24)
(26)
(26)
(22)
(21)
Did not state
0.0
7.7
0.0
14.3
2.8
(0)
(3)
(0)
(4)
(1)
Source: Authors’ calculations based on the PSNP survey data.
Note: Figures given are percentages, with the number of kebeles reporting appearing in parentheses below.
Table 6.3. Kebele perceptions of adequacy of financial resources to meet direct
support needs
Tigray
Amhara
Amhara-HVFB
Oromiya
SNNPR
Sufficient
5.6
15.4
15.0
7.1
11.1
(2)
(6)
(6)
(2)
(4)
Insufficient
94.4
76.9
85.0
78.6
86.1
(34)
(30)
(34)
(22)
(31)
Did not state
0.0
7.7
0.0
14.3
2.8
(0)
(3)
(0)
(4)
(1)
Source: Authors’ calculations based on the PSNP survey data.
Notes: Figures given are percentages, with the number of kebeles reporting appearing in parentheses below.
Across all regions, a minimum of 58 percent of kebeles perceived that resources for public
works were insufficient. For direct support, the percentage was much higher with 76.9 to
94.4 percent of kebeles across all regions reporting that the financial resources needed to
meet requirements for households eligible were insufficient.
One response to insufficient resources is to ask for more funds. Out of 119 kebeles that
reported that they had insufficient resources to employ all those eligible for public works, 79
stated that they asked the woreda for additional resources. In the remaining 30 kebeles, 24

101
did not, while 6 did not provide a response to this question. Out of 151 kebeles that reported
that they had insufficient resources to employ all those eligible for public works, 94 stated
that they asked the woreda for additional resources, 25 did not, and 32 did not say. Yet such
appeals were rarely successful—fewer than 5 percent of kebeles reported that they received
additional funds following these requests.
The vast majority of kebeles also asked for advice as to how they should allocate their
resources, given that these were insufficient to meet all needs. Responses by woredas (as
recounted by kebeles) are given in Table 6.4.
Lastly, the issue of the labor cap was discussed during the qualitative fieldwork. The cap is
followed by most kebele FSTF officials that were interviewed. Officials in one kebele, Adizata
kebele in Ahferom woreda (Tigray state), stated that they do not follow a labor cap:
The guideline says that public works participants can work a maximum of 15
days per month per person but we do not follow this guideline. We always
order everybody to work as per their family size. We do not have a labor cap.
Rather, it is the work that determines the limit to which households are
expected to contribute their labor [TIG_A/K-FG-1].
Table 6.4. Advice given to kebeles for addressing resource shortfalls
Tigray
Amhara
Amhara-HVFB
Oromiya
SNNPR
Advice given to kebeles for coping with insufficient
resources to employ all those eligible for public works
Restrict the number of households that could
receive employment
66.7
64.3
64.7
45.5
60.0
Reduce the number of days of employment
allocated to beneficiaries so that more
households could receive employment
6.7
25.0
23.5
31.8
16.7
Rotate households access to employment across
different years
16.7
3.6
11.8
18.2
10.0
No advice received
10.0
7.1
0.0
4.6
13.3
Advice given to kebeles for coping with insufficient
resources to meet demands for direct support
Restrict the number of households that could
receive direct support
51.8
61.1
54.6
35.0
48.2
Reduce the payments made to direct support
beneficiaries so that more households could
receive payments
22.2
16.7
12.1
10.0
11.1
Allow some households to receive public works
employment and direct support payments
18.5
11.1
18.2
35.0
14.8
Have these households do public works
employment instead of receiving direct support
3.7
5.6
15.2
15.0
25.9
No advice received
3.7
5.6
0.0
5.0
0.0
Source: Authors’ calculations based on the PSNP survey data.
6.6. Household Perspectives
In all three survey rounds, the same households have been asked to describe the criteria
used to select public works respondents in their locality. Results are tabulated in Table 6.5.

102
Across all regions, Table 6.5 presents a consistent, yet somewhat unsettling pattern. As
previously discussed in Gilligan et al. (2009), between 2006 and 2008 respondents
increasingly described program criteria in terms of poverty and were able to do so in
increasingly specific ways. However, between 2008 and 2010, this trend reverses with fewer
respondents describing poverty-related criteria. Instead, access to the program is
increasingly described as random or is determined by a quota system. On a more positive
note, there are fewer reports that access to public works is determined by being a friend or
family member of project staff or the village leadership. Also, fewer households identify the
public works component of the PSNP as being a response to drought.
Table 6.5. Percentage of households reporting that the following criteria were used to
select public works participants in their locality, by region and year
Tigray
Amhara
Amhara-HVFB
2006
2008
2010
2006
2008
2010
2008
2010
Poverty
People who are seen to be poor
42.4
62.1
24.1
55.7
75.9
22.7
86.2
25.8
People with small or no landholding
32.7
44.5
24.9
33.4
55.1
23.1
81.4
27.0
People with few or no cattle/oxen
9.6
22.1
13.3
10.6
30.2
11.2
51.2
16.2
Connections
Religious or ethnic groups
0.3
1.3
0.2
0.4
0.6
0.0
4.1
0.0
Family friends of project staff or
village leadership
3.2
3.0
1.5
5.0
2.8
1.5
8.7
2.9
Other
Randomly
2.3
1.5
19.0
1.8
3.0
17.8
6.1
20.4
Quota for each kebele
11.7
17.5
26.5
11.3
7.3
20.7
24.4
21.5
People badly affected by drought
14.1
19.1
6.5
14.7
22.0
4.9
35.2
8.7
Oromiya
SNNPR
2006
2008
2010
2006
2008
2010
Poverty
People who are seen to be poor
45.3
67.3
21.4
68.0
72.4
26.5
People with small or no landholding
42.0
40.6
20.3
44.4
54.5
29.6
People with few or no cattle/oxen
20.4
15.4
8.7
18.0
25.3
10.6
Connections
Religious or ethnic groups
0.5
0.3
0.0
1.2
2.6
1.5
Family friends of project staff or
village leadership
8.4
14.3
3.3
4.5
2.1
1.9
Other
Randomly
2.7
2.5
9.3
1.2
8.0
18.0
Quota for each kebele
10.6
5.4
18.7
2.5
10.8
22.4
People badly affected by drought
28.2
13.7
9.6
16.6
26.3
8.2
Source: Authors’ calculations based on the PSNP survey data.
Perceptions of the criteria applied to allocate direct support are given in Table 6.6. These
results mirror those found for identification of access to public works—namely a reduction in
the percentage of households that could provide specific criteria and an increase in those
who perceive that access is either random or governed by a quota.
In the focus group discussions, there is an overall understanding that the “poor” or the
“poorest of the poor” are targeted under the PSNP. Consistent with Table 5.5, there was a
tendency among the groups that were interviewed to view the “poor” as an undifferentiated

103
category. There was little understanding of the program criteria for determining which
population groups should be targeted. This could be interpreted as ignorance of the targeting
criteria. A different explanation is that respondents are reluctant to over-specify who should
benefit, because needs are so extensive and undercoverage is a problem. This, in turn,
relates to the preference of most communities for broader targeting, including more
households rather than larger transfers for a smaller segment of the population. Instead,
focus groups offered their own targeting criteria of who should benefit from the program.
Three of four focus groups in Shebedino woreda in SNNPR stated that the “landless” are
deserving of PSNP support. Elsewhere, landlessness was not mentioned consistently.
Disability or the inability to work was the next most common criterion, mentioned by focus
groups in all regions. Surprisingly, female-headed households do not appear to be a
significant category that is targeted for PSNP support, although, in practice, many female-
headed households were covered in targeting exercises carried out to date. Only two groups
(both in Shebedino woreda) offered this as a criterion.
Table 6.6. Percentage of households reporting that the following criteria were used to
select direct support participants in their locality, by region and year
Tigray
Amhara
Amhara-HVFB
2006
2008
2010
2006
2008
2010
2008
2010
Demographic
Old people
70.8
84.0
23.4
64.3
73.5
23.0
92.4
24.5
Disabled
55.9
75.1
23.4
54.6
72.6
19.1
88.9
23.9
Connections
Religious or ethnic groups
0.1
0.0
0.0
0.0
0.0
0.1
0.0
0.1
Family friends of project staff or
village leadership
1.5
1.0
0.9
1.9
1.6
1.4
4.8
1.8
Other
Randomly
0.3
4.2
19.5
1.7
0.9
15.8
3.6
18.7
Quota for each kebele
4.9
7.3
23.7
1.2
0.5
22.1
16.8
22.8
Oromiya
SNNPR
2006
2008
2010
2006
2008
2010
Demographics
Old people
65.7
69.5
21.9
65.5
67.7
25.2
Disabled
59.5
59.1
20.8
69.4
74.0
21.1
Connections
Religious or ethnic groups
0.2
0.5
0.2
0.3
2.3
0.8
Family friends of project staff or
village leadership
3.0
8.5
3.1
1.1
0.9
1.4
Other
Randomly
2.0
1.2
14.6
0.4
5.7
18.5
Quota for each kebele
3.1
1.4
15.2
0.3
6.6
19.4
Source: Authors’ calculations based on the PSNP survey data.
Although nearly all groups do not recall specific targeting criteria for the PSNP, nearly all
expressed a basic understanding of the criteria for targeting direct support. In general, this is
understood as those who are unable to work, whether because of disability, sickness, or old
age. The poor who can work are regarded as those who should participate in public works.
Only 3 out of 40 focus groups expressed ignorance of the criteria for receiving direct support.
Further, there was broad agreement with the rules for targeting direct support. The main

104
point of disagreement rather concerned exclusion errors, given the breadth of the population
that requires support and the limited coverage of the program.
Yet, in spite of problems with targeting, 24 of 40 focus groups that were interviewed
maintained that targeting had been carried out correctly. Ten groups suggested there were
problems with targeting, although in some groups members disagreed with at least some
claiming that targeting had been done correctly. A further six groups gave no response.
Overall, ten groups referred to a limited quota size as having contributed to the exclusion of
some chronically food-insecure households.
Furthermore, there is a general understanding that the community should also be involved in
agreeing on the list of direct support beneficiaries, even though none of the focus groups
provided any detailed description of the precise selection process. The process for selecting
those put forward to the community for agreement is slightly more varied. Some focus
groups reported that it is the community who identifies the initial list of proposed direct
support beneficiaries. In Tembaro and Shebedino woredas of SNNPR state, all focus groups
explained that the community identifies the neediest beneficiaries at a community meeting.
However, in Demba Gofa woreda, the third area of SNNPR covered by the assessment, and
in woredas in all other regions, focus groups reported a slightly different process in which a
kebele leadership committee first proposed a list of beneficiaries to the community, which
then debated the list at a public meeting. Only one focus group explicitly named the
community FSTF as having an acknowledged role in helping select direct support
beneficiaries, although it can be assumed that what many groups referred to as “kebele
leadership committees” were, in actual fact, community FSTFs. One group, a women’s
group in Ziway Dugda woreda of Oromiya state, expressed explicit dissatisfaction with the
selection process, noting that it was male-dominated and that women did not take part in
decisionmaking. A women’s group in Berite woreda, also in Oromiya, also complained:
“They [kebele officials] registered the better-off people. I complained and the
village (Ganda) leader said, ‘they have the right to cancel your name.’
Nepotism is high. They accused me of insulting the kebele ‘Arada’ leader.
The Arada people work for their kin… They do replace names of people they
want to benefit. The better-off people are registered” [ORO_G/FG-3].
Community focus group discussions touched on the issue of the labor cap. These revealed
that there is an understanding of the principle of the labor cap, although there are differences
in awareness of the precise cap. In Tigray, all but one focus group correctly recalled the
labor cap of 20 days for men and 15 days for women that is followed in the region. In
Oromiya, four focus groups stated that the labor cap is 12 days per month; a men’s focus
group in Beritie kebele (Gursum woreda) explained that the labor cap is 2 days/week
between January and March in the dry season and 1 day/week during the rains between
April and June; and a women’s group in Beritie stated that women are required to work 3–4
days/month. In Amhara state, six of nine groups stated that each individual is required to
work 5 days/month; one women’s group explained that the cap is 2 days/week increasing to
the entire week when there is a lot of work; one group did not know; and one group did not
respond. Various views were expressed by community focus groups in SNNPR state
revealing a lack of understanding. Six groups in total responded out of a possible of nine
groups. Of these, two groups stated that the labor cap was 16 days/month, although

105
members of one of these groups maintained that some households work up to 40
days/month. The responses of the other four groups that replied were not clear.
At the community level, there is widespread knowledge of the full family targeting rule.
However, views are mixed on whether the principle is being adhered to. In three of the four
regions, some focus group participants maintained that not all members of targeted
households were registered. According to a men’s focus group in Tembaro woreda, SNNPR
state, “all household members were not registered. They said there is a quota shortage. So
the full family size was not taken into consideration, either for large or small families”
[SN_D/FG-4]. Participants of a women’s group in Gursum woreda, Oromiya state, attributed
partial family targeting to a limited quota size as well. Participants of a men’s focus group in
Gursum explained that the full family targeting rule is followed for small- and medium-size
households but that some members of larger households are not targeted. According to a
men’s group in Ziway Dugda woreda, Oromiya state, 10 percent of household members, on
average, are not targeted. However, in all regions, when focus group members were asked
to report their household sizes and the number of members that were registered, a majority
indicated that all members were registered. In most cases in which households reported that
some members were not registered, further probing determined that these were newborns,
which are not covered under PSNP targeting rules if they were born after a targeting
exercise was carried out.
Views were also solicited from community focus groups on how polygamous households are
registered, who collects payment for registered polygamous households, and who makes
decisions in polygamous households on the use of transfers. Not all groups responded.
However, respondents in groups in three regions explained that a husband in a polygamous
household registers with his first (‘official” or “legal”) wife. Second and lower ranking wives—
if they meet the selection criteria—are registered as female heads of households. Yet,
respondents acknowledged many instances in which second and lower ranking wives were
not registered, usually because of a “quota,” meaning that a polygamous household could
only register one wife. There were varied responses to who collects the payment—
sometimes it is the first wife who is registered, for others it is the husband, and for still others
it is the husband and wife together. Unsurprisingly, because the elder wife is typically
registered with the husband in polygamous households, this does create tension around how
the transfer is shared. The quotes below indicate the tensions within polygamous
households around sharing the transfer:
In most cases, even if there is discussion on the use of PSNP payment
between a husband and wife, it is the man that has the final say. We say we
are equal but our tradition and culture favors the man. The bad side of our
culture has not yet been fully removed. We [women] say we have equal rights
with men, but he is the one who decides. He knows what is good and what is
bad [AM_SA/FG-3].
There are families where there is fair sharing but also there are situations in
which the sharing is unfair, in which case the family ends up in dispute
[ORO_Z/FG-3].

106
If the women are in harmony, the payment is shared. If not, there is no
sharing. However, the husband can receive the share of one person payment
(his share) and give it to the other wife [SN_T/FG-3].
The registered wife collects the money and is the one that decides its use. In
a household where there is harmony, a decision on how to use the transfer is
made, based on consultation involving the husband and the two wives.
However, such harmony is very rare [SN_T/FG-3].
If the wife receives, she brings the payment and gives to the husband. The
husband shares to both households, if he has two wives. If one wife wants to
use alone, it is not accepted and there will not be peace [SN_S/FG-4].
Buying clothing, food, sheep, etc., is decided by consultation of the wife and
husband. In polygamous households, the husband consults with both wives.
He calls both wives and then consults. The sharing depends on size of
children in the respective households. There has to be peace [SN_S/FG-4].
6.7. Selection into Public Works
Table 6.7 looks at the extent to which access to public works remains constant over time.
The top panel looks at constancy of access between 2008 and 2010, while the bottom panel
looks at 2006 to 2008. In the top panel, a relatively large fraction of households who ever
received public works between 2008 and 2010 (between 13 and 28 percent, depending on
the region), only do so in 2008. A similar percentage received these transfers in 2009 and
2010 and a much larger percentage, between 50 and 72 percent, received payments for
public works in all three years. This is suggestive of a move toward stabilizing the lists of
public works beneficiaries so as to increase the number of households who receive
payments for three consecutive years. (It also explains why households may have had less
contact with the community FSTFs.) This contrasts with results for the years 2006–2008
shown in the bottom panel. There is evidence in those data of considerably more movement
in and out of public works from one year to the next. The greater constancy observed in
2009 and 2010 is consistent with the argument that access to public works has become
more predictable.
Table 6.7. Consistency of household participation in the public works component of
the PSNP, by region
Region
Number of
households
ever receiving
public works
Only in
2008
Only in
2009
Only in
2010
In 2008
and 2009
In 2008
and 2010
In 2009
and 2010
In 2008
and 2009
and 2010
Tigray
538
14.5
0.4
0.2
0.2
0.9
23.2
60.6
Amhara
292
18.5
0.0
0.0
0.0
1.4
10.9
69.2
Amhara-HVFB
399
17.3
0.0
0.0
1.5
1.8
12.3
67.2
Oromiya
326
28.2
0.0
0.3
0.9
1.5
19.0
50.0
SNNPR
389
13.6
0.0
0.0
0.5
0.3
14.1
71.5
Source: Authors’ calculations based on the PSNP survey data.
Note: Sample restricted to households receiving public works payments in the first five months of at least one year, 2008, 2009,
or 2010.

107
Table 6.7. continued
Region
Number of
households
ever receiving
public works
Only in
2006
Only in
2007
Only in
2008
In 2006
and 2007
In 2006
and 2008
In 2007
and 2008
In 2006
and 2007
and 2008
Tigray
522
14.8
2.3
5.2
2.1
9.8
25.5
40.4
Amhara
429
17.9
5.1
2.8
7.0
5.1
27.3
34.7
Amhara-HVFB
-
-
-
-
-
-
-
Oromiya
457
24.3
3.7
2.4
5.9
4.8
13.4
45.5
SNNPR
441
16.8
0.5
0.0
2.3
0.5
11.6
68.5
Source: Authors’ calculations based on the PSNP survey data.
Note: Sample restricted to households receiving direct support payments in the first five months of at least one year, 2006,
2007, or 2008.
Has the move to greater constancy in participation across years affected the targeting of
public works employment? Because there are a number of factors that affect selection,
cross-tabulating individual characteristics against selection will not necessarily give a clear
sense of changes that may have emerged over time. Nor will tables give a sense as to
whether certain characteristics have, in terms of targeting, become more or less important
over time or whether they matter more or less across different regions.
For these reasons, we address this issue using regression analysis. The dependent variable
equals one if the household was employed in PSNP public works, zero otherwise. Probit
regressions are appropriate when the outcome is dichotomous and so we use them here.
We assess whether the selection into public works is affected by three broad sets of
household categories: household demographic categories, wealth, and “connections.” We
take into account where households reside so as to control for regional-specific factors that
affect selection. The coefficients that are estimated by the probit have been transformed into
marginal effects to make them easier to interpret. For example, the coefficient -0.001 found
in column (1) associated for the age of head means that every additional year of age of the
household head reduces the likelihood that the household was paid for public works
employment by 0.1 percentage points (0.001 x 100). Dummy variables, such as whether the
household head is female, are interpreted as “switching the variable” from 0 to 1. The
coefficient for female head in column (1) means that a female-headed household is 5.5
percentage points less likely to participate in the public works component of the PSNP than
an otherwise identical male-headed household.
18
The most striking result in Table 6.8 is the constancy of the parameters across years.
Access to public works declines with the age of the household head. Households with high
dependency ratios, or with individuals over 65, are less likely to be employed in public works;
conversely, increasing the number of adults aged 15–59 increases the likelihood of
employment. Consistent with a number of comments made in focus groups, ownership of
oxen reduces the likelihood of receiving payment for PSNP public works. Finally, while our
crude indicators of household connections (based on parents’ status within the kebele) do
not show any consistent correlations, being born in the kebele increases the likelihood of
public works participation by 5.3 percentage points. This, too, is consistent with remarks
18
For the full sample results, the calculation of standard errors takes into account the clustered nature of the sample.
Calculating heteroscedasticity-robust standard errors does not appreciably change these findings. For the region-specific
results, there are too few clusters to control for clustering and so we report heteroscedasticity-robust standard errors.

108
made during focus groups that in-migrants are excluded from participation in the PSNP. The
negative and statistically significant coefficients for the three regional variables mean that
after taking into account other household characteristics, relative to a household in Tigray, a
household in these regions is less likely to be included in the public works component of the
PSNP.
Table 6.8. Correlates of receipt of public works (PW), by year
(1)
(2)
(3)
Access to PW
Access to PW
Access to PW
Variables
2006
2008
2010
Age of head
-0.001
-0.004***
-0.004***
(-0.752)
(-3.477)
(-3.489)
Female head
-0.055*
-0.069***
-0.040
(-1.871)
(-2.591)
(-1.418)
Males, 65+
-0.083*
-0.060
-0.034
(-1.863)
(-1.318)
(-0.731)
Females, 65+
-0.048
-0.080**
-0.087**
(-1.170)
(-2.217)
(-2.037)
Dependency ratio
-0.014***
-0.006
-0.012
(-2.853)
(-0.865)
(-1.588)
Males, 15–59
0.022*
0.034*
0.017
(1.835)
(1.914)
(1.093)
Females, 15–59
0.045***
0.029
0.046**
(2.862)
(1.416)
(2.258)
Grades completed, head
-0.000
-0.003
-0.003
(-0.144)
(-1.598)
(-1.375)
Oxen owned
-0.072***
-0.067***
-0.060***
(-5.373)
(-5.093)
(-4.151)
Head born in this kebele
0.046
0.053**
0.053*
(1.565)
(2.079)
(1.826)
Father of head an important person in kebele
-0.011
-0.041*
-0.035
(-0.479)
(-1.714)
(-1.286)
Mother of head an important person in kebele
0.035
0.063*
0.031
(1.025)
(1.882)
(0.940)
If parent of head holds official position in kebele
0.010
-0.033
0.064
(0.320)
(-0.944)
(1.098)
Amhara
-0.101
-0.155**
-0.233***
(-1.362)
(-2.270)
(-3.782)
Oromiya
-0.014
-0.192***
-0.315***
(-0.249)
(-2.861)
(-5.153)
SNNPR
-0.066
-0.192***
-0.246***
(-1.096)
(-2.995)
(-4.193)
Observations
3,398
3,398
3,101
Source: Authors’ calculations based on the PSNP survey data.
Notes: Coefficients are expressed in terms of their marginal effects. Robust (clustered) Z-statistics in parentheses. * significant
at 10 percent, ** significant at 5 percent, *** significant at 1 percent.
One potentially confounding factor in the interpretation of Table 6.8 is the following. Over
time, suppose in one region, a particular characteristic became more associated with public
works participation, while in another region, the association between this characteristic and
participation weakened. Since these would tend to cancel each other out, our aggregated
results would mask these changes. Accordingly, as a robust check, we estimate the
correlates of receipt of public works by year and region. These are presented in Table 6.9.

109
Looking across these coefficients, we see no meaningful evidence of this cancelling out
hypothesis.
Table 6.9. Correlates of receipt of public works (PW), by year and region
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
Access to
PW
Access to
PW
Access to
PW
Access to
PW
Access to
PW
Access to
PW
Access to
PW
Access to
PW
Tigray
Tigray
Amhara
Amhara
Oromiya
Oromiya
SNNPR
SNNPR
Variables
2006
2010
2006
2010
2006
2010
2006
2010
Age of head
-0.005**
-0.007***
-0.001
-0.004
0.001
-0.003
0.000
-0.005**
(-2.382)
(-3.348)
(-0.392)
(-1.633)
(0.476)
(-1.434)
(0.111)
(-2.187)
Female head
0.029
-0.073
-0.070*
0.048
-0.119*
-0.150***
-0.103*
-0.053
(0.442)
(-1.083)
(-1.843)
(1.276)
(-1.852)
(-3.021)
(-1.758)
(-1.156)
Males, 65+
-0.019
-0.103
0.014
0.041
-0.118
-0.097
-0.281***
-0.076
(-0.202)
(-1.443)
(0.179)
(0.466)
(-1.516)
(-1.071)
(-3.144)
(-0.629)
Females, 65+
0.039
-0.091
0.011
-0.104
-0.057
0.082
-0.184***
-0.163**
(0.422)
(-1.056)
(0.179)
(-1.512)
(-0.704)
(1.427)
(-2.607)
(-2.050)
Dependency ratio
-0.008
0.010
-0.018***
-0.027***
-0.006
0.001
-0.006
0.008
(-0.829)
(1.093)
(-2.743)
(-3.108)
(-0.605)
(0.0680)
(-0.513)
(0.462)
Males, 15–59
0.059**
0.063**
0.026
0.053*
0.014
0.030
0.013
0.014
(2.278)
(2.058)
(1.005)
(1.675)
(0.614)
(1.392)
(0.599)
(0.509)
Females, 15–59
0.101***
0.139***
0.074***
0.066**
0.046
0.057*
0.014
-0.001
(2.624)
(3.747)
(2.601)
(2.287)
(1.135)
(1.690)
(0.718)
(-0.0397)
Grades completed, head
0.003
0.001
0.004
-0.004
-0.009**
-0.009***
-0.007
-0.006
(0.773)
(0.312)
(1.219)
(-1.045)
(-2.243)
(-2.653)
(-1.379)
(-1.101)
Oxen owned
-0.064***
-0.065**
-0.117***
-0.128***
-0.016
-0.025
-0.102***
-0.099***
(-2.792)
(-2.085)
(-5.608)
(-7.330)
(-0.554)
(-1.036)
(-3.468)
(-2.927)
Head born in this kebele
0.039
-0.021
0.005
0.088
0.084
0.059
0.020
-0.001
(0.775)
(-0.385)
(0.0893)
(1.314)
(1.249)
(0.977)
(0.345)
(-0.0106)
Father of head an important
person in kebele
0.006
-0.024
-0.032
0.024
-0.008
-0.002
0.014
0.029
(0.171)
(-0.445)
(-0.611)
(0.619)
(-0.218)
(-0.0300)
(0.211)
(0.419)
Mother of head an
important person in kebele
0.073
0.057
0.179***
0.004
0.013
0.009
-0.085
-0.044
(1.095)
(0.666)
(4.066)
(0.0719)
(0.198)
(0.146)
(-1.206)
(-0.663)
If parent of head holds
official position in kebele
0.049
-0.049
-0.090
-0.011
-0.018
-0.061
0.063
0.034
(0.857)
(-0.588)
(-1.498)
(-0.153)
(-0.307)
(-0.661)
(1.238)
(0.939)
Observations
844
849
831
849
824
845
899
911
Source: Authors’ calculations based on the PSNP survey data.
Notes: Coefficients are expressed in terms of their marginal effects. Robust Z-statistics in parentheses. * significant at 10
percent, ** significant at 5 percent, *** significant at 1 percent.
6.8. Selection into Direct Support
We now examine the evolution of access to direct support over time. We begin with Figure
6.1. This shows the percentage of households in our sample that received direct support in
2006, 2007, 2008, 2009, and 2010. In Tigray, the percentage of households receiving direct
support jumped markedly between 2005 and 2006 from 13 to 20 percent. Since 2007, this
figure has hovered between 20 and 22 percent. Access is much higher in Tigray than in any
other region. In 2010, it ranged from 8 to 10 percent in the other regions, a percentage that
has changed little over time.

110
Figure 6.1. Percentage of households receiving direct support, by region and year
Source: Authors’ calculations based on the PSNP survey data.
Table 6.10 looks at the extent to which access to direct support remains constant over time.
The top panel looks at constancy of access between 2008 and 2010, while the bottom panel
looks at 2006 to 2008. In the top panel, a relatively large fraction of households who ever
received direct support between 2008 and 2010 only do so in 2008 (between 25 and 39
percent, depending on the region). A similar percentage received these transfers in 2009
and 2010 and somewhere between 26 and 45 percent received direct support in all three
years. This is suggestive of a process in which the list of direct support beneficiaries was
revised in 2009—with the result that some, but not all, households were dropped and others
added—but not revised in 2010 (which is why the percentage of households getting direct
support in both those years is also relatively high). This contrasts with results for the years
2006–2008 shown in the bottom panel. There is evidence in those data of considerably more
movement in and out of direct support from one year to the next. The greater constancy
observed in 2009 and 2010 is consistent with a claim that along this dimension access to
direct support has become more predictable.
In an earlier analysis of the targeting of direct support, Gilligan et al. (2009) found that
households receiving direct support payments had heads that are considerably older, had
fewer members, and had fewer able-bodied workers. They were also poorer as measured by
livestock and landholdings. To assess whether targeting of direct support has improved,
remained the same, or worsened, we estimate a probit regression where the dependent
variable equals one if the household received direct support, zero otherwise. Results are
reported in Table 6.11.

111
Table 6.10. Consistency of household participation in the direct support component of
the PSNP, by region
Region
Number of
households
ever receiving
direct support
Only in
2008
Only in
2009
Only in
2010
In 2008
and 2009
In 2008
and 2010
In 2009
and 2010
In 2008
and 2009
and 2010
Tigray
241
29.8
0.0
0.0
0.8
2.5
27.8
39.0
Amhara
130
37.2
0.0
0.8
0.0
5.4
20.2
36.4
Amhara-HVFB
184
30.9
1.6
0.5
0.0
1.6
23.9
41.3
Oromiya
99
38.8
0.0
3.0
0.0
1.0
30.6
26.5
SNNPR
89
25.8
0.0
0.0
0.0
0.0
29.2
44.9
Source: Authors’ calculations based on the PSNP survey data.
Note: Sample restricted to households receiving direct support payments in the first five months of at least one year, 2008,
2009, or 2010.
Region
Number of
households
ever receiving
direct support
Only in
2006
Only in
2007
Only in
2008
In 2006
and 2007
In 2006
and 2008
In 2007
and 2008
In 2006
and 2007
and 2008
Tigray
224
8.0
2.2
11.6
0.5
2.7
39.7
35.3
Amhara
125
4.0
5.6
8.8
0.0
0.0
52.0
29.6
Amhara-HVFB
-
-
-
-
-
-
-
-
Oromiya
181
4.4
6.6
47.5
2.8
1.7
20.4
16.6
SNNPR
124
0.8
0.0
37.1
3.2
0.0
24.2
34.7
Source: Authors’ calculations based on the PSNP survey data.
Note: Sample restricted to households receiving direct support payments in the first five months of at least one year, 2006,
2007, or 2008.
In Table 6.11, the number associated with each variable denotes its marginal effect on the
probability that a household receives direct support. For example, the number “0.069” for
female head in the column marked “Access to DS 2006” means that holding all other
characteristics constant, a household headed by a female was 6.9 percentage points more
likely to receive direct support. In 2006, demographic characteristics dominated the
likelihood that a household received direct support. Households with older heads, older men,
and fewer younger men were also more likely to receive these payments. Characteristics
related to poverty played a much smaller role in determining selection, suggesting that
infirmity (as proxied by age) and absence of labor power are the factors driving selection.
Household “connections” as proxied by family status within the kebele appear to have no
impact on receipt of direct support.
The magnitudes of these effects are meaningfully large. As with our discussion of public
works, consider two households. Household A has a female head aged 70; household B has
a male head aged 40. In all other respects, they are identical. The results reported in Table
6.11, column (1) (for 2006) indicate that the probability of receipt of direct support by
Household A is 12.9 percentage points higher than by Household B. Doing the same
exercise with column (3) (for 2010), we find that the probability that Household A receives
direct support rises by 23.5 percentage points. Assuming that advanced age is correlated, in
these settings, with reduced work capacity, these results indicate that the targeting principles
laid out in the Project Implementation Manual for direct support continue to be met.

112
Table 6.11. Correlates of receipt of direct support (DS), by year
(1)
(2)
(3)
Access to DS
Access to DS
Access to DS
Variables
2006
2008
2010
Age of head
0.002***
0.003***
0.004***
(3.934)
(7.180)
(7.716)
Female head
0.069***
0.092***
0.115***
(4.732)
(6.797)
(7.314)
Males, 65+
0.064***
0.057***
0.020
(2.808)
(2.781)
(0.944)
Females, 65+
0.074***
0.002
-0.004
(3.742)
(0.113)
(-0.247)
Dependency ratio
0.004
-0.000
0.005**
(1.528)
(-0.121)
(2.031)
Males, 15–59
-0.019***
-0.034***
-0.013
(-2.668)
(-4.376)
(-1.640)
Females, 15–59
0.003
-0.017**
-0.018***
(0.330)
(-2.099)
(-2.613)
Grades completed, head
-0.001
0.001
0.001
(-0.898)
(1.117)
(0.468)
Oxen owned
-0.029***
-0.023***
-0.012**
(-3.937)
(-3.557)
(-2.205)
Head born in this kebele
0.007
-0.024
0.007
(0.408)
(-1.574)
(0.527)
Father of head an important person in kebele
0.003
0.015
-0.006
(0.284)
(1.175)
(-0.521)
Mother of head an important person in kebele
0.006
-0.010
0.003
(0.448)
(-0.842)
(0.210)
If parent of head holds official position in kebele
-0.016
-0.014
0.003
(-0.798)
(-0.623)
(0.142)
Amhara
-0.063***
-0.071***
-0.070***
(-2.670)
(-3.003)
(-4.366)
Oromiya
0.027
-0.071***
-0.070***
(0.803)
(-3.628)
(-4.165)
SNNPR
0.003
-0.070***
-0.050***
(0.133)
(-4.438)
(-3.163)
Observations
3,669
3,264
3,120
Source: Authors’ calculations based on the PSNP survey data.
Notes: Coefficients are expressed in terms of their marginal effects. Robust (clustered) Z-statistics in parentheses. * significant
at 10 percent, ** significant at 5 percent, *** significant at 1 percent.
Lastly, in Table 6.12, we assess whether these changes are observed across all regions. To
do so, we estimate the probit separately by region and examine how the coefficients change,
or do not change, over time. Consistent with what we observed in Table 6.11, over time
being a female head has an increasingly large effect on the likelihood of being selected for
direct support. This is especially so in Tigray, where, holding other factors constant, a
female-headed household had a 10.4 percentage point higher likelihood of receiving direct
support in 2006 and a 24.1 percentage point higher likelihood of receiving direct support in
2010.

113
Table 6.12. Correlates of receipt of direct support (DS), by region and year
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
Access to
DS
Access to
DS
Access to
DS
Access to
DS
Access to
DS
Access
to DS
Access to
DS
Access to
DS
Tigray
Tigray
Amhara
Amhara
Oromiya
Oromiya
SNNPR
SNNPR
Variables
2006
2010
2006
2010
2006
2010
2006
2010
Age of head
0.003***
0.008***
0.000
0.002**
0.004***
0.003***
0.001
0.003***
(3.091)
(4.666)
(0.231)
(2.215)
(4.088)
(4.403)
(1.010)
(3.949)
Female head
0.104***
0.241***
0.034*
0.089***
0.031
0.077***
0.082**
0.044*
(5.572)
(4.340)
(1.815)
(3.595)
(0.842)
(3.804)
(2.085)
(1.789)
Males, 65+
0.128**
0.099
0.025
0.005
-0.028
0.003
0.166***
-0.017
(2.283)
(1.270)
(1.051)
(0.161)
(-0.585)
(0.114)
(3.958)
(-0.531)
Females, 65+
0.038
-0.009
0.072***
0.036
0.105*
-0.029
0.077*
-0.004
(1.330)
(-0.174)
(2.764)
(1.274)
(1.730)
(-1.152)
(1.858)
(-0.0955)
Dependency ratio
-0.006
0.007
0.007**
0.015***
0.005
0.002
0.005
-0.004
(-0.947)
(1.243)
(2.122)
(3.398)
(0.846)
(0.505)
(0.727)
(-0.777)
Males, 15–59
-0.031**
-0.017
-0.006
0.001
-0.034
-0.008
-0.017
-0.028**
(-2.271)
(-0.727)
(-0.657)
(0.0426)
(-1.598)
(-0.613)
(-1.370)
(-2.376)
Females, 15–59
-0.046**
-0.061***
0.015
0.011
0.031*
-0.022
0.008
-0.011
(-2.413)
(-3.835)
(1.553)
(0.456)
(1.951)
(-1.607)
(0.790)
(-1.455)
Grades completed, head
-0.003
0.000
-0.002*
0.000
0.003
0.002
-0.001
-0.002
(-1.302)
(0.0477)
(-1.665)
(0.0676)
(1.247)
(1.220)
(-0.584)
(-0.977)
Oxen owned
-0.017
-0.015
-0.023**
-0.014
-0.032**
-0.001
-0.034**
-0.018
(-1.121)
(-0.929)
(-2.466)
(-1.451)
(-2.068)
(-0.145)
(-2.106)
(-1.468)
Head born in this kebele
-0.028
0.033
0.024*
0.037
0.074**
0.005
-0.068*
-0.029
(-0.909)
(0.909)
(1.666)
(1.598)
(2.557)
(0.242)
(-1.907)
(-1.267)
Father of head an important
person in kebele
-0.014
-0.037**
-0.009
-0.052***
-0.005
-0.015
0.039*
0.054***
(-0.516)
(-2.167)
(-0.567)
(-2.936)
(-0.243)
(-0.652)
(1.667)
(4.301)
Mother of head an important
person in kebele
0.046
0.057*
0.016
0.053*
-0.004
0.014
-0.015
-0.059***
(1.469)
(1.880)
(1.353)
(1.679)
(-0.104)
(0.545)
(-0.684)
(-4.849)
If parent of head holds
official position in kebele
0.093**
0.109
-0.048
-0.023
-0.042
0.002
(2.068)
(1.075)
(-0.874)
(-0.782)
(-1.460)
(0.0704)
Observations
894
775
845
667
937
824
944
821
Source: Authors’ calculations based on the PSNP survey data.
6.9. Switching from Public Works to Direct Support and vice versa
Outside of Tigray, there is virtually no evidence of households being shifted from public
works to direct support or vice versa. Fewer than 5 percent of public works households were
switched to direct support, and fewer than 5 percent of direct support households were
switched to public works in Amhara, Amhara-HVFB, Oromiya, and SNNPR. In Tigray,
approximately 12 percent of households were switched from public works to direct support
after 2008, while about 33 percent of direct support households were switched to public
works. Female-headed households, smaller households, and households with older heads
were more likely to be switched from public works to direct support, while larger households
were more likely to be switched from direct support to public works. However, focus group
discussions with women indicated that, when pregnant, they are switched out of public works
and into direct support.

114
6.10. Summary
The provision of an administrative quota begins at the federal level and extends all the
way down to the sub-kebele level. Regional and woreda officials broadly follow the
targeting criteria outlined in the PIM to determine administrative quotas, determining
allocations on the basis of previous relief caseloads, agroecological conditions,
malnutrition levels, the average size of landholdings, and the estimated population of
chronically food-insecure households in particular administrative areas. The upstream
budget review process—starting at the kebeles and moving upward to the woredas, the
regions, and the federal level—to decide the total number of the PSNP beneficiaries has
not been practiced.
Most exclusion errors are due to inadequate administrative quotas. Full family targeting
is practiced in most areas, even though officials acknowledge that it implies that fewer
households are targeted. Kebele officials frequently request additional funds but
invariably these requests are denied.
Officials have used the contingency budget and stricter, locally-specific targeting criteria
to manage the limited administrative quota compared with the larger population of
chronically food insecure that require support.
The targeting criteria appear to be widely understood. However, there has been a fall in
the percentage of respondents who can identify specific criteria for access to the PSNP
and there is a rise in the percentage who perceives these are a consequence of a quota
system or a random allocation. Despite this, a majority of focus groups agreed with the
targeting process as well as the selection criteria used to determine public works and
direct support beneficiaries.
Access to public works and direct support has become more consistent over time.
Between 50 and 72 percent of current public works participants received payments for
public works in 2008, 2009, and 2010. Between 26 and 45 percent of current recipient of
direct support received direct support in 2008, 2009, and 2010.
There has been little change in the application of targeting criteria as seen through
estimates of the correlates of access to public works or direct support.
Outside of Tigray, there is virtually no evidence of households being shifted from public
works to direct support or vice versa. However, focus group discussions with women
indicated that, when pregnant, they are switched out of public works and into direct
support.

115
7. The Implementation of Public Works Projects
7.1. Introduction
This chapter discusses a number of aspects associated with the implementation of public
works projects under the PSNP. It explores whether there have been administrative
improvements in the coordination of public works activities planning and implementation. It
assesses the extent to which those activities that are chosen have incorporated feedback
from communities and whether these activities are linked to livelihoods. It summarizes
perceptions regarding implementation problems and the extent to which gender
considerations have been incorporated into the implementation of public works activities.
While it includes perceptions of the usefulness of public works, it does not provide a
technical assessment of their quality.
19
In so doing, it covers a number of evaluation
objectives as Table 7.1 notes.
Table 7.1. Evaluation objectives covered in chapter 7
Evaluation
objective
Issue
Link to Log Frames and TOR
Document progress in the implementation of the PSNP
Can gender dimensions of access be better captured?
TOR, para 42
Assess trends in perceptions of the effectiveness and transparency of the PSNP and HABP among different
groups of clients
Beneficiaries understand how the program works
PSNP Log frame Output 4.4
Source: Authors’ compilation.
7.2. Coordinating the Implementation of Public Works
An objective of the current phase of the PSNP is to improve coordination and technical
support to public works activities. To this end, Regional Public Works Focal Units (RPWFU)
have been established in all regions. They are responsible for coordinating the
implementation of public works activities under the Natural Resource Management (NRM)
Work Process of the Bureau of Agriculture and Regional Development (BoARD) at the
regional level. They are also responsible for providing technical supports to woredas in the
planning and implementation of public works. Regional Public Works Technical Committees
also provide some technical support. In focus group and key informant interviews, some
concerns were expressed that there were insufficient resources to support these
coordination activities. Limitations included shortages of vehicles, limited operational budget,
insufficient number of staff compared to woredas required to be covered, and limited
experience.
There are variations in the establishment of Woreda Public Works Focal Units (WPWFU)
across regions. In Amhara and SNNPR, WPWFUs have been established and have started
to provide technical support to the implementation of public works at the grassroots level.
19
Such an assessment is beyond the Terms of Reference for this study. A number of other instruments assess the quality of
public works sub-projects, such as the Public Works Reviews and Public Works Impact Evaluation.

116
The woreda public works focal unit coordinator of Denba Goffa woreda of SNNPR explained
their situation as follows:
There is a public work focal unit at woreda level. The head of natural resource
management case team is the coordinator of the focal unit. Crop, livestock,
food security, and natural resource management experts are assigned to
monitor the whole activities of public works starting from the planning stage
[SN-D/W-FGD-WPWFU].
The existence of WPWFU with different experts in Amhara and SNNPR woredas is
perceived to provide more coordinated technical support during the implementation of public
works. A respondent from a WPWFU in Ebinate woreda explained:
There was only a focal person responsible for public works, but now there are
a group of experts and they are jointly planning, monitoring, and evaluating
the implementation of public works [AM-E/W-FGD-WPWFU].
That said, concerns were also expressed about coordination with sectoral government
offices and technical assistance they provide. The Sokota woreda Public Works Focal Unit
indicated, “There is not much coordinated technical support from sectoral offices”
[AM_SO/W-KII-WPWFU].
7.3. Planning Public Works
Interviews with actors involved at different administrative levels provided a variety of
contrasting views on how public works are planned. In most areas, officials at the kebele
level and below prioritize projects for public works. Five woreda Public Works Focal Points
that were interviewed suggested a certain degree of community involvement in helping to
plan public works and set priorities. They indicated that they provide an initial list of indicative
projects, with development agents and community FSTF officials modifying these on the
basis of local priorities and circumstances. In all three woredas visited in Amhara, the Public
Works Focal Point indicated that a woreda level Natural Resource Technical Committee
(“Watershed Committee”) determines public works plans and sends these to kebeles.
However, kebele officials in the three woredas offered a contrasting view, indicating that
development agents and community members decide priorities and plans for public works.
In all but two kebeles, members of kebele FSTFs and development agents explained that
public works planning is conducted at the community level. In Gurade kebele in Demba Gofa
woreda, SNNPR state, a development agent explained that the woreda sends a list of
indicative activities and that the community has no decisive role in changing this. In Beritie
kebele, Gursum woreda, Oromiya state, a development agent explained that public works
plans were determined by the kebele FSTF but that these did not deviate greatly from plans
sent by the woreda:
The kebele FSTF develops a detailed implementation plan based on the
indicative plan sent from the woreda. The kebele FSTF makes minor
adjustments to this. It can increase the volume of public works but cannot
reduce the number of proposed works. Also, it can substitute one activity with
a similar and locally relevant activity, such as constructing a soil bund rather

117
than a stone bund. However, the kebele FSTF does not have the mandate to
make major modifications to the plan sent from the woreda. Public works
planning is not participatory, because the main activities and the amount of
work to be done is fixed by woreda officials [ORO_G/K-KI-2].
These varying responses are suggestive of the possibility that there is just a good deal of
substantial variation in community participation in the selection of public works projects
across regions and over time. Evidence consistent with this comes from the household
surveys where respondents were asked if they had been involved in the selection of public
works that had been completed during 2010. Results are shown in Table 7.2. This shows
that, over time, there has been a gradual increase in the proportion of households indicating
some role in project selection. However, as Table 7.3 shows, significant differences exist in
the percentages of male- and female-headed households participating in decisionmaking,
with male-headed households more likely to do so in all regions.
Table 7.2. Percentage of households participating in selection of public works, by
region and year
Year
Tigray
Amhara
Amhara-HVFB
Oromiya
SNNPR
2006
8
10
11
9
2008
20
31
22
31
2010
34
30
35
28
28
Source: Household questionnaire, 2006, 2008, 2010.
Table 7.3. Percentage of households participating in selection of public works, by
region and sex of head, 2010
Sex of head
Tigray
Amhara
Amhara-HVFB
Oromiya
SNNPR
Male
40
31
38
32
29
Female
20
25
24
18
26
Source: Household questionnaire, 2010.
Two other aspects of the planning of public works are worth noting. First, there appears to
be little effort to explicitly include women in decisionmaking. In three of ten woredas, women
are represented on watershed committees. Some woreda officials justified this by stating
that there were other ways that women could contribute to decisionmaking, such as through
their representation on kebele FSTFs or by attending meetings at the community level to
discuss public works plans.
Second, one of the principles of public works specified in the PIM is that they are to be
integrated with woreda development plans. In all woredas that were visited, public works
officials confirmed that public works are planned in this way. As well as being consistent with
woreda development plans, public works are also to be implemented in accordance with the
Ministry of Agriculture and Rural Development (MoARD) guidelines on community-based
participatory watershed development (CBPWD). The overall objective of CBPWD is “to
improve the livelihood of communities and households in rural Ethiopia through
comprehensive and integrated natural resource development” (GFDRE 2010, 40). Many of
the indicative public works activities in the PIM reflect the CBPWD approach, such as the

118
establishment of area enclosures and woodlots, the construction of hillside terraces, shallow
wells and ponds, and stream diversion.
7.4. Linking Public Works and Local Livelihoods
Regional and woreda officials were asked to describe how public works had supported
people’s livelihoods. The most frequent response given was the construction of conservation
works such as hillside terraces, tree planting, and rock dams. Terraces, in particular, were
thought to have prevented further soil erosion and limited crop damage due to flooding.
Other ways in which public works supported people’s livelihoods was by the establishment of
enclosures for growing high value fodder and constructing roads to improve market access
(indicated by three out of nine officials that were asked). At the kebele level, all development
agents that responded (seven in total) said that soil and water conservation activities had
helped improving livelihoods. A development agent in Furra kebele in Shebedino woreda
(SNNPR state) summarized the impacts of public works on people’s livelihoods as the
following:
The roads that were built helped members of this community to take their
produce to the market easily. Soil and water conservation activities have
helped to improve the productivity of farm plots. Planting trees has brought
about positive ecological changes, which are observable in improved
productivity of coffee farms. Not only do the trees provide shade for the
coffee, but the wood is cut for fuelwood and as construction material
[SN_S/K-KI-2].
Woreda level public works officials in Shebedino explained:
Public works have helped to build community assets. For example, terraces
have helped to recover land that previously was thought to be useless. Trees
and grasses have been planted to produce animal feed; as well as stabilizing
land that is susceptible to erosion; these have a direct economic benefit for
the community, because there is more fodder available for cattle. The other
activity that has helped is road construction. Before, it was difficult for vehicles
to reach rural areas and to transport farm produce to bigger markets. Roads
have opened up outlying areas and vehicles can move much more easily than
before. Connections to markets have increased [SN_S/W-FG-PW].
A development agent in Adizata kebele in Ahferom woreda (Tigray state) offered a similar
assessment:
Water for irrigation is more readily available for growing vegetables and fruits.
Now you can get banana, orange, and lemon in this area, which you could not
before. These also provide a source of income. Communal grazing areas
have been established as well through public works and this has increased
access to livestock fodder and, hence, increased household incomes. The
construction of schools and clinics has also helped supporting livelihoods.
The construction of roads has made it easier to transport farm produce and
increase incomes [TIG_A/K-KI-2].

119
Officials were also asked to give examples of particular public works activities that had been
successful. Five of ten woreda public works officials who replied mentioned watershed
improvements such as water harvesting structures and terraces. However, there were some
dissenting views as to whether public works have helped to improve people’s livelihoods,
even though these were the exception to the general outlook, which was positive. Members
of the Public Works Technical Committee in Ebinet woreda in Amhara state explained:
Public works are supporting people’s livelihoods in terms of protecting assets.
But if you take a long-term perspective and having a look at an impact beyond
the life of the program, then public works are having less impact, because
they do not create assets. Public works in this woreda consisted mainly of
constructing hillside terraces and micro-basins, but an integrated approach
was not used, so these will not support livelihoods in the long run. So we
cannot say confidently that public works are supporting livelihoods in this
woreda [AM_E/W-FG-PW].
A development agent in Berite kebele in Gursum woreda (Oromiya state) recounted
numerous problems with public works:
Few public works were completed in this kebele because we have received
limited support from the woreda, such as few seedlings, although tree
planting was one of the main activities. The kebele is over 40 km from the
woreda center and the road is bad, particularly during the wet season.
Watershed activities have been localized and the quality of the work carried
out is poor. Hence, there has been no reduction in natural resource
degradation. Part of the problem is that public works sites are far away.
People complain about the distance they must travel to reach the sites. Some
of the seedlings that were planted were not cared for and dried up. Soil and
water conservation structures were also subsequently ruined [ORO_G/K-KI-
2].
As part of the household survey, respondents were asked if they perceived that they had
benefitted from these public works. As Table 7.4 shows, households have consistently
perceived that they have benefitted from road construction and maintenance. In 2010,
between 56 and 80 percent of households felt that the construction of water-harvesting
infrastructure had been helpful and, generally, this percentage had risen since 2006. By
contrast, soil and water conservation (SWC) activities and natural resource management
(NRM) are not as frequently perceived as being beneficial and the percentage of households
reporting that these were helpful had generally declined since 2008. Broadly speaking, these
views are consistent with those of regional and woreda officials. Across all activities, female-
headed households are slightly less likely to report that they benefitted from these (Table
7.5). Separately, we assessed whether perception of benefits differed by wealth status. To
do this, we disaggregated the sample into quartiles based on livestock holdings. Across
most forms of public works activities, there are few differences in perceptions. However, the
wealthiest quartile was more likely to indicate that they benefitted from soil and water
conservation activities on private land and other natural resource management activities (52
and 49 percent, respectively) than the poorest quartile (44 and 39 percent, respectively).

120
Table 7.4. Household perceptions of whether they benefitted from public works
Percentage of households indicating that their household benefited from
construction or maintenance of
Region
Survey
year
Roads
Water
harvesting
infrastructure
SWC,
communal
land
SWC,
private
land
Other
NRM
activities
Wells
Tigray
2006
53
33
38
51
21
58
2008
72
55
54
68
52
46
2010
69
58
59
44
56
45
Amhara
2006
57
29
36
67
16
48
2008
83
39
68
59
69
80
2010
78
56
50
42
40
75
Amhara – HVFB
2010
85
62
58
61
49
69
Oromiya
2006
76
63
39
28
13
62
2008
90
78
53
60
52
66
2010
79
80
49
24
30
96
SNNPR
2006
70
36
35
22
8
-
2008
84
42
54
37
53
91
2010
85
66
63
46
46
77
Source: Household questionnaire, 2006, 2008, 2010.
Notes: SWC = soil and water conservation. NRM = natural resource management. Cells are left blank if there were fewer than
10 responses.
Table 7.5. Household perceptions of whether they benefitted from public works, by
sex of head, 2010
Percentage of households indicating that their household benefited from construction or
maintenance of
Sex of head
Roads
Water
harvesting
infrastructure
SWC,
communal
land
SWC,
private land
Other NRM
activities
Wells
Male
83
65
57
47
44
64
Female
77
59
54
44
41
60
Source: Household questionnaire, 2010.
Notes: SWC = soil and water conservation. NRM = natural resource management.
One reason why wealthier households would be more likely to say that they benefitted from
soil and water conservation activities on private land is that these activities were actually
undertaken on their own land. Figure 7.1 shows that in all regions, wealthier households
were more likely to report that a public works activity was undertaken on their land than were
poorer households. By contrast, there was no difference between male- and female-headed
households (not shown).

121
Figure 7.1. Percentage of households reporting that public works were undertaken on
their land, by wealth category and region, 2010
Source: Authors’ calculations based on the PSNP survey data.
7.5. Implementation Problems
Regional and woreda public works officials recounted many implementation difficulties that
contributed to these problems. Limited capacity was the most common complaint. Capacity
issues that were mentioned included a lack of training and technical knowledge, few staff to
monitor works being carried out, limited access to vehicles for woreda officials to visit public
works sites, and a shortage of funds, making it difficult to procure equipment for technically
complex works, such as GPS for watershed planning. The Public Works Focal Point in
Tembaro woreda, SNNPR state, stated:
Roads that have been constructed have poor technical qualities. This is
because of a lack of monitoring by experts who are knowledgeable about
road construction. Sometimes we take experts from rural road construction
office with us to the construction sites. But since it is not possible to take them
to all places where roads have been constructed, there is still a problem in the
quality of the roads that have been constructed [SN_T/W-FG-PW].
Three of five woreda public works officials that responded to questions on implementation
difficulties indicated delays in transferring the capital and transfers budgets. The Public
Works Focal Point in Sayint woreda, Amhara state, explained how budget delays had
adverse consequences for public works that had been planned in 2010:
The timing of public works was extended this year because we did not receive
funds until May. Because of this, we could not pay public works participants
directly after they had completed their work. The delay led to unwanted
things. For example, it was difficult for us to process bids and employ

122
contractors or to procure materials, given that the fiscal year ends in June,
just one month after we received the funds. By the time we received the
budget, the rains had set in. The timing was bad for the public works we had
planned, such as installing water pumps. The quality of some public works
was compromised; costs increased. Some public works that were planned
had to be cancelled and carried over to the next fiscal year [AM_SA/W-FG-
PW].
Another problem referred to by an official in the Oromiya Regional Public Works Office is the
tendency of development agents to work with more successful farmers rather than the
poorest:
The development agents working at the community level lack the required
capacity to mobilize the community for public works. This is because they
tend to work with the model farmers that adopt technologies and show
progress rather than the poorest of the poor who need diverse support for
which the development agent may not have the knowledge, attitude, or skills
to provide [ORO_R/R-KI-PW].
The same official also raised the possibility that a dependency attitude had set in among
clients of the program, thereby limiting its achievements:
The most important factor affecting the performance of public works is the
dilemma of whether the PSNP is an entitlement or a tool for development.
The development concern of PSNP has been overridden by the entitlement
concern of PSNP and this has significantly impacted on the performance of
public works. Donors more often than not are pushing for the PSNP to be
treated as an entitlement. They say people have to receive the transfer and
that public works should not be considered as a condition for being paid. In
most cases the community knows this and as a result people are not
motivated to be involved in public works [ORO_R/R-KI-PW].
This tension between the twin objectives of the public works component of the PSNP
emerges from other sources. As part of the quantitative survey conducted at the kebele
level, respondents were asked, “Which is more important for public works programs,
transferring income to the poorest households or building productive community assets?”
and “Is hiring skilled labor to assist with public works projects sometimes justified, even if it
means that less people can be hired or that other lower skilled people must work fewer
days?” Table 7.6 shows responses by regions.

123
Table 7.6. Kebele level perceptions of the comparative importance of income transfers
to the poor and the construction of public works as priorities for the PSNP
Tigray
Amhara
Amhara-HYFB
Oromiya
SNNPR
All
Which is more important for public works programs, transferring
income to the poorest households or building productive community
assets?
Only transferring income to the poor
is important
2.9
8.8
10.8
0.0
6.3
6.3
Transferring income to the poor is
more important than building
productive assets
41.2
38.2
27.0
17.4
12.5
28.1
Transferring income to the poor and
building productive assets are
equally important
17.7
20.6
29.7
60.9
46.9
33.1
Building productive assets is more
important than transferring income
to the poor
38.2
26.5
29.7
21.7
25.0
28.8
Only building productive assets is
important
0.0
5.9
2.7
0.0
9.4
3.8
Is hiring skilled labor to assist with public works projects sometimes
justified even if it means that less people can be hired or that other
lower skilled people must work fewer days?
Agree
85.7
94.3
94.7
86.4
93.3
91.3
Disagree
14.3
5.7
5.3
13.6
6.7
8.7
Source: Authors’ calculations based on the PSNP survey data.
7.6. Gender Considerations
The revised PIM (June 2010) incorporates principles to ensure that implementation of public
works is gender sensitive. For example, priority is to be given to projects that will reduce
women’s work burden. Work is to be organized in a way that considers women’s domestic
responsibilities. Therefore, women may be asked to work fewer hours each day and less
days overall than men, as well as be permitted to arrive late and leave early to look after
their responsibilities for cooking and caring for children at home. Also, women are also to be
assigned tasks that are less physically demanding or alternatively to participate in nutritional
classes or other activities that are thought to be particularly beneficial to them and their
children.
Men’s and women’s focus groups were asked to provide examples of “men’s work” and
“women’s work” on public works to determine if women were assigned tasks that were less
physically demanding. Ten out of 20 groups responded. Groups in three kebeles reported
different tasks for women and men. For example, members of a women’s group in Adizata
kebele in Afherom woreda (Tigray state) stated that men will do the digging, such as for a
pond, while women remove the soil. The same example was given by a men’s group in
Furra kebele in Shebedino woreda (SNNPR state). Members of the men’s group in Adizata
explained that men will break stone and women will transport the stone to the public works
site. Another example provided by members of a men’s group in Beritie kebele in Gursum
woreda (Oromiya state) is that men do digging and excavation, while women prepare and
serve tea and perform other lighter tasks. Community focus groups were also asked if men
and women do the same types of activities on public works. Seven women’s focus groups

124
(of seven who responded) stated that women and men do the same activities; six men’s
groups (of seven who responded) said that both sexes do the same activities. A respondent
in a women’s group in Wal in Amhara complained, “We go for work together with the men
and we come back home together. It is our foremen that are pressurizing us to do hard labor
like men, not the government. We shoulder and transport stones from one place to another
and dig the ground just like men” [AM_SO/FG-3]. A respondent in a men’s group in Wal in
Amhara state explained, “As far as public works are concerned, women are given the same
type of work as men. She is required to work like men. We feel that this puts a lot of
pressure on women. The work is even very hard for us, let alone for women. In light of this,
they are oppressed” [AM_SO/FG-4].
Thus, although women in some areas do lighter tasks, in general, women and men do the
same work on public works projects.
Woreda officials were asked whether public works interfere with women’s other
responsibilities around the home. Public works officials in six of ten woredas responded. Of
these, four acknowledged that public works do interfere with women’s domestic work;
however, officials in two of these woredas said the evidence was “anecdotal” and that no
formal assessment had taken place. Still, a member of the Woreda FSTF in Sokota in
Amhara state exhorted, “It goes without saying [that public works interfere with women’s
other domestic work]. Public works are continuous. This situation needs attention. It has
created an additional workload for women” [AM_SO/W-FG-1]. Kebele FSTF members were
not asked directly whether public works interfere with women’s domestic work, but
responses to other questions concerning who participates in public works gave some
indication that public works in some areas are overburdening women, as the following
quotes illustrate:
Yes, public works interfere with women’s domestic and childcare
responsibilities. But we did not take this issue into account or develop any
supportive actions. You know, if a housewife works with her husband until
6 pm in the evening at a public works site, then she also will need to finish her
domestic activities till 9 pm. This is really a burden which needs further
attention [TIG_A/K-KI-1].
Yes, public works affect women.... Women do not get any rest. They are
engaged again in domestic work when they return home from doing public
works. A member of the KFSTF added that women are mentioning that the
public works is becoming harder and harder for them [AM_SO/K-KI-1].
Women are usually engaged in lighter activities than men. But the number of
hours they are working is the same as men. A woman leaves behind many
other household activities to work on public works [SN_S/K-KI-1].
Nine of ten women’s focus groups that were asked report that public works interfere with
women’s domestic tasks. By comparison, four men’s groups (out of eight that responded)
stated that public works interfered with their other work activities. However, of these, three
stated that public works interfere with their agricultural work in the harvesting period in
December and January. Yet, as woreda and kebele officials alike state that public works are

125
timed around the lean agricultural activity period, we can assume that the problem is
minimized.
The following statements from women’s groups give a sense of the pressures existing for
women who must fulfill their labor contribution to public works alongside their work at home:
Yes, we face problems. When we face serious problems we go to the
foreman to get permission to be absent, such as in the event of a funeral or to
care for someone sick at home. We are permitted up to two days off if we
need to travel far for a funeral. If it is more than two days then there will be
deduction in our salary. There is no one who will help us for free. Those that
help you on public works want to be paid for what they have worked. So if I
miss work at a public works site, then I must expect to pay 10 Birr to whoever
steps in. There is no one from home who can help. There are those who say
that if you have a problem, then give me my 10 Birr, otherwise there are no
relatives who give support for free [SN_S/FG-3].
We feel that women’s time share in public works should have been half that of
men to help us accomplish our additional responsibilities at home. We’d even
suggest that public works should include projects that help female heads of
households to repair their homes and complete other difficult tasks
[TIG_S/FG-3].
The burden of balancing public works and domestic chores fall the hardest on
us—female heads of households [ORO_Z/FG-3].
Those not having a helping hand in the family are faced with the problem.
Especially, those women who do not have an elder girl at home to help out
face the greatest challenge of contributing to public works and working at
home [ORO_G/FG-3].
Yes, it interferes with our domestic responsibilities. But since it is
development, we decrease the time we spend on domestic tasks and
concentrate on the public works [AM_E/FG-3].
There are women who have children that feed on the breast. In such cases,
when there are no elder children at home to take care of the younger ones,
we are forced to leave them behind with neighbors. If there is a milking cow,
we face a shortage of time to fetch it grass. There is also no time to prepare
food for school children [SN_T/FG-3].
Public works interfere with our domestic work. The workload in the house
awaits her when she is back home. If she does not work the public works, she
will not get payment, and if she does not accomplish her domestic work at
home, the family cannot survive, therefore she is affected by the public works
[AM_SO/FG-3].
There is no one in the village or household to assist us. Both the public works
and the domestic work are managed by us women alone. The children are in
the school. The cattle are kept in the shade while we are away for public
works. There is nobody to take care of them [AM_SA/FG-3].

126
Woreda and kebele officials alike were asked what measures had been taken to help women
cope with the additional burden of contributing to public works, such as being permitted to
arrive late and leave early and being provided with childcare at public works sites. Public
works officials in eight of ten woredas report that no steps have been taken to make it easier
for women to participate in public works. The exceptions were Ziway Dugda (Oromiya) and
Shebedino (SNNPR) woredas, where officials affirmed that women can arrive late and leave
early from public works sites. Members of four KFSTFs (out of eight that responded) report
that there are no special measures in place to help women. KFSTF members in
Shengodefer and Serawudi kebeles (Ebinet and Sayint woredas, respectively, in Amhara
state) and in Beritie kebele (Gursum woreda, Oromiya state) stated that measures exist to
help women contribute to public works. In both Shengodefer and Serawudi, other household
members are expected to cover the contribution of women who arrive late or leave early. In
Beritie, direct support recipients provide childcare at some public works sites, according to
woreda public works officials.
An interesting insight was provided by members of the FSTF in Saesi Tsemba woreda in
Tigray state:
As per the guideline [in the PIM], women are allowed to start work late and
leave early to support them in doing their household chores. However, this
has raised a very serious issue. Although women are not penalized for
coming late and leaving early, other community members, and men, in
particular, have complained that this is not proper. They have questioned the
practice, saying ‘why not reduce the work norm for the women, as is the case
in the non-PSNP community free labor?’ [TIG_S/W-FG-1].
The women’s affairs member of the same FSTF added:
The principle of allowing women to arrive late or leave early is not widely
known for fear that women might claim this as a right and, hence, deter them
from participating in the community’s work. Even some complain that women
time the birth of their children to coincide with public works so that they can
receive free support [TIG_S/W-FG-1].
7.7. Summary
This chapter has explored aspects of the implementation of public works projects under the
PSNP. There are several noteworthy results.
There appears to be considerable variation in the extent of community involvement in the
selection of public works projects. There are clearly cases where these have been
decided upon by the woreda and consultation with kebeles and communities seems
notional. But there are other instances where development agents and community
members have had a significant say in deciding priorities and plans for public works.
Public works projects are generally perceived to assist in supporting livelihoods. Work on
roads is seen as particularly positively. Views on other activities, such as water
conservation and natural resource management, are more mixed.

127
A number of implementation problems associated with lack of technical skills, difficulties
in supervision, and delays in receipt of funds were noted. The tension between the use
of public works as a means of transferring funds to poor households and as a means of
strengthening community assets was also highlighted.
Participation in public works activities puts considerable strain on women. There appears
to be little willingness to address this.

128
8. Payments for Public Works and Direct Support
8.1. Introduction
This chapter discusses a number of aspects associated with payments for public works
employment and direct support. We provide information describing public works participation
by gender, age, and region. We then review the levels of transfers, the frequency with which
they are made, and the extent to which these reflect beneficiaries’ entitlements as outlined in
the Project Implement Manual (PIM). We examine transfer levels given to direct support
beneficiaries. We also present findings on household preferences for modality of payment
(food and cash) and whether these preferences are season or location dependent. Table 8.1
notes how this work links to the evaluation objectives set out in the PSNP Log Frame.
Table 8.1. Evaluation objectives covered in chapter 8
Evaluation
objective
Issue
Links to Log Frames and TOR
Document progress in the implementation of the PSNP
Are public works payments timely and
predictable? Do clients receive complete
entitlement?
PSNP Log frame Output 1.1
PSNP Log frame Output 1.2
TOR, para 39
Are direct support payments timely and
predictable? Do clients receive complete
entitlement?
PSNP Log frame Output 1.1
PSNP Log frame Output 1.2
TOR, para 35
Source: Authors’ compilation.
8.2. Public Works Employment
Figure 8.1 displays the mean number of days worked, per household, by region, and year,
conditional on having worked at least one day in that year. It is important to note that,
because of the timing of the 2006, 2008, and 2010 surveys, these data pertain to the first
five months of each year, the months of Tir, Yekatit, Megabit, Miaza, and Ginbot in the
Ethiopian calendar (approximately January 9–June 9). Public works are also planned for the
month that follows Ginbot, but because this is not captured here, Figure 8.1 understates the
amount of employment that individuals receive under the PSNP. In Tigray and SNNPR,
beneficiary households averaged 25 days of work per month. In Amhara and Amhara-HVFB,
they averaged about 20 days per month. While the level of employment has increased in
Oromiya relative to its performance in 2006, the number of days worked remains
considerably lower than that found elsewhere.

129
Figure 8.1. Number of days worked per beneficiary household, Tir–Ginbot, by survey
year
Source: Authors’ calculations based on the PSNP survey data.
In the 2008 and 2010 surveys, we also asked about employment over the previous year
(2007 and 2009, respectively). This allows us to assess whether, for example, employment
levels in Oromiya are low or, instead, whether employment occurs slightly later in the
calendar year. Figure 8.2 reports these data for two subperiods, Tir–Sene (roughly mid-
January to mid-July) and Hamile–Tahese (mid-July to mid-January). Until 2008, he program
design called for employment to be concentrated in the period Tir to Sene so what we should
see is little employment in the subsequent period, Hamile to Tahisas; from 2009 onwards,
public works can be carried out at any time during the year. Across the full sample, the
average household employed in public works received 129 days of employment in 2009.
Three-quarters of this work occurs in the first six months of the year. There are, however,
some differences by region. Tigray, Amhara, Amhara-HVFB, and SNNPR largely conform to
this pattern. For example, the average public works household in SNNPR received 134 days
employment in 2009 of which 103 days occurred between Tir and Sene. By contrast,
employment levels are lower in Oromiya. There, beneficiary households received, on
average, 104 days employment and a larger fraction of this, 40 percent, occurred in the
second half of the year.

130
Figure 8.2. Mean days worked per beneficiary household, 2007 and 2009, by region
and six-month period
Source: Authors’ calculations based on the PSNP survey data.
Table 8.2 examines the distribution of employment by sex between Tir and Ginbot, 2010. In
Tigray, employment is fairly evenly split between men and women. Around 40 percent of
public works in Amhara and Amhara-HVFB is undertaken by women. Women’s participation
in public works is lowest in SNNPR and Oromiya.
Table 8.2. Distribution of employment by sex and region, Tir–Ginbot, 2010
Region
Tigray
Amhara
Amhara-HVFB
Oromiya
SNNPR
Men
Number who worked
309
285
277
262
323
Mean days worked
69.7
51.0
59.0
41.4
66.0
Women
Number who worked
293
221
233
122
204
Mean days worked
63.9
43.8
49.1
36.9
55.0
All
Percent workers female
48.4
42.2
45.4
31.7
38.6
Percent labor days by women
46.4
38.6
40.9
29.2
34.4
Source: Authors’ calculations based on the PSNP survey data.
Lastly, we examine the age distribution of public works participants. At various times, there
have been concerns expressed in terms of either children or the elderly undertaking these
activities. Figure 8.3 shows that the reported number of children (individuals aged 15 or less)
as a percentage of the total number of participants is small (not shown, but available on
request). In fact, the quantitative data show that there are only 10 children in the more than

131
2,500 individuals who undertook public works employment in the first five months of 2010.
20
Individuals aged 20–29, 30–39, and 40–49 make up nearly 70 percent of the workforce
engaged in public works and only 1.2 percent is 69 or older.
Figure 8.3. Age distribution of individuals employed on public works projects, by sex,
2010
Source: Authors’ calculations based on the PSNP survey data.
Notes: Figures are in percentages.
8.3. Payments for Public Works: Levels
The 2006, 2008, and 2010 surveys collected information on how much, and when,
beneficiaries were paid for their work in both cash and kind.
21
The community questionnaire
collected data on prices of the goods provided in kind and these data are the main source
used to value the in-kind transfers. Note that all figures are nominal. They do not account for
inflation, which at times has been substantial.
Figure 8.4 shows the mean payments received by beneficiary households in the first five
months of 2006, 2007, 2008, 2009, and 2010 for the full sample and also by region. In three
locations, Tigray, Amhara-HVFB, and SNNPR, there are impressive increases in the amount
20
We were concerned that, if parents understate the age of their children, these data would understate the extent of child labor.
If this were true, we might expect to see age heaping around the age cutoff, 15 years. However, examination of the age
distribution of the sample shows only slightly higher percentage of individuals aged 15 than those aged 14 or 16.
21
It is important to note that these are self-reported data. It would be immensely helpful to validate these self-reports against
administrative payroll data. This was discussed during the regional workshops in August 2011 and with federal FSCD staff.
However, as of mid-October 2011 (when the final versions of this report was prepared), these administrative data have not
been made available to us.

132
of money paid to respondents over this period, particularly between 2006 and 2009. For
example, this rises from 251 to 865 Birr per household in Tigray and from 281 to 985 Birr per
household in SNNPR. The value of transfers in Amhara-HVFB in the first five months of
2009 exceeded 1,000 Birr; it was the first time this has occurred in the six years the PSNP
has operated. In these localities, however, payments fell in 2010. While payments in
Oromiya increased between 2008 and 2009, they remained below Tigray and SNNPR. In
Amhara, there has been little increase in mean payments over the first five months between
2007 and 2009, and the mean level of payments in 2010, 332 Birr, was the lowest level
recorded since 2006.
Figure 8.4. Mean payments to households with public works participants between Tir
and Ginbot, by region and year
Source: Authors’ calculations based on the PSNP survey data.
One possible explanation is that, just as employment is delayed in some regions, so, too, is
payment. For the year 2009, we can explore this possibility by examining payments made
month-by-month. This is reported in Table 8.3 and summarized, in six monthly intervals, in
Figure 8.5. Table 8.3 shows that payments, for at least some beneficiaries, continue after
Sene, the last month when employment is supposed to occur. That some payments are
made in Hamile is not especially noteworthy; it is likely that in some woredas, the distances
involved mean that work undertaken in one month is paid in the following month. But Table
8.3 shows that while payments do continue after Hamile, they slow to a trickle everywhere
by Meskerem.

133
Table 8.3. Mean payments (Birr) to households with public works participants, by
month and region, 2009
Region
Tir
Yekatit
Megabit
Miaza
Ginbot
Sene
Hamile
Nehase
Meskerem
Tikmit
Hidar
Tahisis
Total
Tigray
113
138
197
180
212
204
109
104
26
14
14
14
1,329
Amhara
46
65
125
102
97
132
109
53
47
6
12
9
807
Amhara-HVFB
92
138
263
209
253
161
199
60
6
11
16
9
1,419
Oromiya
122
72
106
122
141
102
89
99
56
25
21
16
974
SNNPR
185
175
222
193
199
192
97
78
55
33
37
44
1,512
Source: Authors’ calculations based on the PSNP survey data.
Figure 8.5. Mean payments (Birr) to households with public works participants, 2009,
by region and six-month period
Source: Authors’ calculations based on the PSNP survey data.
Tables 8.4 and 8.5 disaggregate these monthly payments by food and cash for 2009 and
2010. Table 8.4 indicates that the predominant form of transfer in SNNPR was cash, while
food transfers comprised a much larger fraction of total transfers in Tigray and Amhara-
HVFB. Cash accounted for just over half the value of all payments in Amhara, while food
accounted for approximately 60 percent of transfers in Oromiya. Food transfers in the first
three months of 2009 in Amhara are so small—53 Birr per household to be essentially
nonexistent. The largest food transfer, in value terms, in Amhara occurs in Meskerem,
several months after transfers are supposed to be completed.

134
Table 8.4a. Mean cash payments (Birr) to households with public works participants,
by month and region, 2009
Region
Tir
Yekatit
Megabit
Miaza
Ginbot
Sene
Hamile
Nehase
Meskerem
Tikmit
Hidar
Tahisis
Total
Tigray
40
41
64
20
38
26
22
27
11
15
37
32
303
Amhara
51
60
99
70
60
61
25
10
25
19
41
20
433
Amhara-HVFB
1
10
24
53
36
40
28
9
6
2
34
12
197
Oromiya
43
39
46
57
54
62
28
36
40
24
31
34
369
SNNPR
162
151
192
150
153
129
74
41
80
65
101
105
1,155
Source: Authors’ calculations based on the PSNP survey data.
Table 8.4b. Mean food payments (Birr) to households with public works participants,
by month and region, 2009
Region
Tir
Yekatit
Megabit
Miaza
Ginbot
Sene
Hamile
Nehase
Meskerem
Tikmit
Hidar
Tahisis
Total
Tigray
81
105
166
125
198
173
86
85
59
37
31
16
1,026
Amhara
1
12
40
38
41
75
88
45
135
3
2
11
368
Amhara-HVFB
104
145
276
186
224
126
176
53
24
55
51
32
1,220
Oromiya
91
40
70
77
120
64
82
85
95
36
20
4
604
SNNPR
25
26
33
46
52
70
28
40
76
29
3
20
358
Source: Authors’ calculations based on the PSNP survey data.
Table 8.5 shows payments by month for 2010. Table 8.6 shows payments disaggregated by
cash and food. There is a sharp distinction between payments in Tigray, Amhara-HVFB, and
SNNPR, and Amhara and Oromiya. SNNPR managed to deliver, on average, relatively large
and (apart from Ginbot, which coincided with the 2010 election) equal amounts of cash on a
monthly basis. Tigray and Amhara-HVFB were able to transfer, on a consistent basis,
meaningful amounts of food. By contrast, food transfers in Amhara and Oromiya were low in
2010 and cash transfers were considerably smaller than in SNNPR.
Table 8.5. Mean payments (Birr) to households with public works participants, by
month and region, 2010
Region
Tir
Yekatit
Megabit
Miaza
Ginbot
Total
Tigray
98
131
116
142
127
614
Amhara
48
66
64
65
89
332
Amhara-HVFB
76
133
190
151
223
772
Oromiya
40
62
64
46
80
292
SNNPR
161
161
200
118
86
725
Source: Authors’ calculations based on the PSNP survey data.

135
Table 8.6a. Mean cash payments (Birr) to households with public works participants,
by month and region, 2010
Region
Tir
Yekatit
Megabit
Miaza
Ginbot
Total
Tigray
28
10
19
35
39
132
Amhara
47
42
28
47
87
250
Amhara-HVFB
4
11
18
23
60
117
Oromiya
34
35
47
42
71
228
SNNPR
153
153
183
81
62
633
Source: Authors’ calculations based on the PSNP survey data.
Table 8.6b. Mean food payments (Birr) to households with public works participants,
by month and region, 2010
Region
Tir
Yekatit
Megabit
Miaza
Ginbot
Total
Tigray
70
121
97
107
88
482
Amhara
1
23
37
18
2
82
Amhara-HVFB
71
121
172
128
163
655
Oromiya
6
27
17
4
8
63
SNNPR
8
7
17
37
24
93
Source: Authors’ calculations based on the PSNP survey data.
A limitation of Table 8.6 is that, by averaging across all beneficiary households, it does not
distinguish between payment regimes where many beneficiaries get small-ish payments
regularly from one where periodically beneficiaries receive large-ish payments. Given the
especially low values for food transfers (see above), in Table 8.7 we look at the number of
PSNP public works beneficiaries receiving a food payment in the first five months of 2010.
This confirms that there were virtually no food transfers occurring in Amhara in the first five
months of 2010.
Table 8.7. Number of food payments to households with public works participants, by
month and region, 2010
Region
Tir
Yekatit
Megabit
Miaza
Ginbot
Tigray
Number households receiving food payment
100
147
129
110
130
Number households not receiving food payment
279
232
250
269
249
Amhara
Number households receiving food payment
1
15
22
16
2
Number households not receiving food payment
252
238
231
237
251
Amhara-
HVFB
Number households receiving food payment
61
107
137
150
166
Number households not receiving food payment
288
242
212
199
183
Oromiya
Number households receiving food payment
31
27
15
4
30
Number households not receiving food payment
192
196
208
219
193
SNNPR
Number households receiving food payment
27
21
25
66
50
Number households not receiving food payment
318
324
320
279
295
Source: Authors’ calculations based on the PSNP survey data.

136
8.4. Payments Levels Relative to Entitlements
While the preceding tables give a good sense as to how much beneficiaries get paid, it does
not tell us if they are receiving their full entitlement. Assessing this is tricky for several
reasons: (1) payments may come more than one month after the work has been done. For
example, our data will not capture payments made after Ginbot 2010; (2) if there are errors
in respondents’ recollections of how much they work—for example, overstatements of the
number of days worked or understatements of payments will increase the proportion of
beneficiaries that appear to be underpaid; (3) in order to construct estimates of total
payments, we need to value payments made in-kind. This can also lead to errors in the
determination of whether beneficiaries have received their full entitlement.
Mindful of these caveats, we do the following. We focus on 2009 for which we have a full
year of payment data. We assume that every three kilograms of grains received as a
payment for public works is worth 10 Birr and construct a “normalized cash value” of these
grain payments. We add all cash payments to this “normalized cash value” to create a
“normalized total payment” in 2009. We then calculate the level of payment, in Birr, that a
beneficiary household should have received, given full family targeting. Under full family
targeting, each member is entitled to five work days per month for six months. At a wage
rate of 10 Birr per day, this yields a payment of 300 Birr per household member. This means
that a three-person household should receive 900 Birr, a four-person household should
receive 1,200 Birr, and so on. We compare these expected levels of payment against mean
notional total payments by region for households ranging in size from three to nine
persons.
22
Results are shown in Figure 8.6.
23
There are a number of striking findings. First, SNNPR appears to come closest to meeting
the transfer levels envisaged under full family targeting. Second, there are large gaps
between transfer levels under full family targeting and these notional total payments in all
other regions. These gaps vary by region, with the result that identically-sized households
receive markedly different levels of transfers, depending on where they live. Using the
methodology described in the previous paragraph, a five-person household (about average
household size in these regions) received 573 Birr if they lived in Oromiya, 785 Birr in
Amhara, 800 Birr in Amhara-HVFB, 972 Birr in Tigray, and 1,333 Birr in SNNPR—a 133-
percent difference between the lowest (Oromiya) and highest (SNNPR) paying regions.
Third, in Tigray and Amhara-HVFB, there appears to be an attempt to follow the spirit of full
family targeting as these transfer levels rise with family size. This is less apparent in
Oromiya. In Amhara, all households appear to receive about the same amount, regardless of
household size.
Figure 8.7 complements these results, showing the percentage of entitlement received by
household size and region. Within each level of size, it is ordered SNNPR, Tigray, Amhara-
HVFB, Oromiya, and Amhara so as to illustrate not only the magnitude of the shortfalls, but
also the regional differences in these. For small- to medium-sized households, SNNPR
meets or comes very close to meeting entitlements under full family targeting, while larger
22
There are insufficient observations to do this for households with fewer than three people or more than nine. Also, there were
not enough observations of nine-person public works beneficiary households in Amhara to calculate a mean notional total
payment.
23
When reviewing these findings, participants in the Tigray workshop noted that, “If people do not work, they do not get paid.”
So, for example, if a household meets only 80 percent of their work requirement, they receive only 80 percent of their
payment. This is done to prevent free riding.

137
households receive 70 to 80 percent. By contrast, nearly all small households in Oromiya
and Amhara receive less than 50 percent of entitlement, and larger households receive as
little as one-third of their entitlement.
Figure 8.6a. Comparison of normalized total payment (public works) to full family
targeting entitlement, by household size, Tigray, 2009
Source: Authors’ calculations based on the PSNP survey data.

138
Figure 8.6b. Comparison of normalized total payment (public works) to full family
targeting entitlement, by household size, Amhara, 2009
Source: Authors’ calculations based on the PSNP survey data.
Figure 8.6c. Comparison of normalized total payment (public works) to full family
targeting entitlement, by household size, Amhara-HVFB, 2009
Source: Authors’ calculations based on the PSNP survey data.

139
Figure 8.6d. Comparison of normalized total payment (public works) to full family
targeting entitlement, by household size, Oromiya, 2009
Source: Authors’ calculations based on the PSNP survey data.
Figure 8.6e. Comparison of normalized total payment (public works) to full family
targeting entitlement, by household size, SNNPR, 2009
Source: Authors’ calculations based on the PSNP survey data.

140
Figure 8.7. Normalized public works payments as a percentage of entitlement under
full family targeting, by household size and region, 2009
Source: Authors’ calculations based on the PSNP survey data.
Are these results plausible? To assess, we take our household level data and aggregate
them up to woreda level means. We present these in Table 8.8, ordering the data from the
woreda, where it appears that households received the smallest percentage of their
entitlement, to the woreda where payments, as a percentage of entitlements, were highest.
We also include the woreda mean value of normalized payments and the number of
payments—disaggregated by number of food and cash payments—for 2009.
Table 8.8. Entitlements and number of payments (public works), by woreda, 2009
2009 Woreda mean:
Region
Woreda
Percent
entitlement
received
Normalized
payments
received (Birr)
Total
number of
payments
Number of
cash
payments
Number of
food
payments
Oromiya
Zeway Dugda
13.0
254
2.80
2.80
0.00
Amhara
Angolela Tera
14.9
236
2.25
2.25
0.00
Oromiya
Gelana
18.5
313
1.71
0.00
1.71
Amhara
Ebenat
19.1
251
2.33
1.60
0.73
Oromiya
Goro
21.3
332
5.41
5.35
0.06
Oromiya
Anchar
23.1
390
1.75
0.88
0.88
Oromiya
Seweyna
25.2
510
3.38
1.54
1.83
SNNPR
Alaba
28.2
459
2.96
2.38
0.58
Oromiya
Boset
29.5
497
4.91
4.91
0.00
Oromiya
Boke
29.9
359
3.71
3.43
0.29
Amhara
Artuma fursi
32.1
491
3.00
2.25
0.75
Oromiya
Mema Angetu
34.5
588
4.75
4.25
0.00
Amhara
Kelala
35.2
386
5.00
4.56
0.44
Amhara
Ambasel
35.7
536
4.84
4.26
0.58
Amhara
West Belesa
36.3
462
3.91
2.91
1.00

141
2009 Woreda mean:
Region
Woreda
Percent
entitlement
received
Normalized
payments
received (Birr)
Total
number of
payments
Number of
cash
payments
Number of
food
payments
Tigray
Naedei Adet
37.0
613
2.67
1.29
1.38
Tigray
Atsegede Tsimbela
38.3
511
3.86
3.43
0.57
Amhara
Janamora
39.1
568
2.83
2.08
1.33
Oromiya
Fentale
42.0
797
3.06
0.71
2.18
Oromiya
Yabello
43.9
771
2.43
0.00
2.43
Oromiya
Gola Oda
44.2
950
5.31
4.38
0.94
Amhara
Beyeda
46.7
765
4.50
2.38
2.13
Oromiya
Gursum
47.5
915
5.81
3.81
1.96
Oromiya
Daro Lebu
48.5
678
2.83
2.50
0.33
Tigray
Kolla Temben
48.9
912
5.29
2.86
2.38
Amhara-HVFB
Dehana
49.3
611
3.07
0.00
3.00
SNNPR
Soro
49.4
764
4.36
4.36
0.00
Tigray
Atsbi Wenberta
50.3
759
5.39
1.09
4.28
Amhara-HVFB
Tach Gayint
52.7
725
3.24
2.45
2.40
Oromiya
Babile
53.9
1,220
3.20
1.80
1.40
Tigray
Saesi Tsaedamba
56.2
927
4.90
0.95
3.83
Amhara-HVFB
Habru
56.8
574
3.75
2.50
2.50
Amhara
Dawa Chefa
56.9
744
4.43
4.10
0.33
Amhara-HVFB
Wadla
57.3
917
3.72
1.76
2.28
Oromiya
Fedis
57.6
914
4.10
1.20
2.90
SNNPR
Amaro
59.0
939
6.00
3.50
2.25
Amhara
Were IIu
59.2
616
5.73
4.55
1.18
Tigray
Ahiferom
61.2
939
3.05
1.36
1.56
Amhara-HVFB
Kobo
61.4
812
4.06
1.87
2.19
Tigray
Tanqua Abergele
63.7
1,101
7.53
1.86
5.58
Amhara-HVFB
Gubalafto
63.8
1,070
5.00
0.86
4.09
Amhara
Legambo
65.4
967
4.39
3.87
0.52
Amhara
Sayint
65.5
1,022
3.06
0.76
2.29
Amhara-HVFB
Lay Gayint
65.8
806
3.02
1.81
1.72
Oromiya
Guba Koricha
68.0
1,297
5.40
2.87
2.60
Tigray
Enderta
69.4
1010
4.67
1.61
3.03
Amhara-HVFB
Sokota
69.6
876
3.47
0.00
3.38
Oromiya
Mieso
70.3
1,147
5.95
3.26
2.74
SNNPR
Burji
71.0
1,378
5.35
3.65
1.70
SNNPR
Badewacho
73.1
1,277
6.05
4.38
1.67
Amhara-HVFB
Habru
74.4
950
5.94
3.06
2.97
SNNPR
Damot Gale
75.3
1,154
5.65
3.65
2.00
Amhara-HVFB
Bugna
75.4
820
5.69
1.62
4.15
Tigray
Gulomehada
76.3
1,269
6.40
3.08
3.38
SNNPR
Zala
77.5
1,125
6.00
6.00
0.00
Tigray
Ofla
77.9
1,386
6.16
3.45
2.68
Amhara-HVFB
Zukela
80.4
1,406
4.77
0.00
4.77
Amhara
Kalu
81.5
1,085
3.71
3.24
1.00

142
2009 Woreda mean:
Region
Woreda
Percent
entitlement
received
Normalized
payments
received (Birr)
Total
number of
payments
Number of
cash
payments
Number of
food
payments
Amhara-HVFB
Simada
81.6
1,324
3.52
0.57
2.96
Tigray
Alamata
82.1
1,395
6.13
1.82
4.18
Amhara
Debresina
82.4
1,021
4.50
2.80
1.50
Amhara
Gidan
82.6
1,034
6.31
3.00
3.31
SNNPR
Aleta Wendo
87.8
1,458
5.05
4.82
0.23
Tigray
Wukro
88.0
1,575
5.69
0.00
5.63
Amhara
Tena
91.1
1,233
6.15
5.54
0.62
Amhara
Habru
91.3
1,059
5.45
2.70
3.40
SNNPR
Omo Sheleko
92.3
1,218
5.54
2.69
2.85
SNNPR
Boloso sore
95.6
1,458
7.00
4.65
2.26
SNNPR
Shebedino
97.1
1,957
5.54
5.54
0.00
SNNPR
Kemba
97.2
1,629
7.61
4.94
2.67
SNNPR
Bonkie
98.6
1,940
6.00
6.00
0.00
SNNPR
Gofa zuria
100.8
1,552
6.33
4.22
2.06
SNNPR
Kindo Koisha
114.2
1,611
7.94
5.35
2.59
Oromiya
Jarso
125.8
2,222
8.59
4.59
3.71
SNNPR
Damote Weyede
137.6
1,875
10.50
10.50
0.00
SNNPR
Konso
151.4
2,291
10.59
8.07
2.46
Source: Authors’ calculations based on the PSNP survey data.
Table 8.8 shows that there is considerable variation in the percentage of entitlements
received, from a low of 13 percent (Zeway Dugda woreda in Oromiya) to a high of 151
percent (Konso woreda in SNNPR). While the poorest performing woredas tend to be in
Oromiya and Amhara, and the best performing woredas in SNNPR, this is not universally
true. Alaba woreda, in SNNPR, performs poorly by this measure, while the third best
performing woreda is Jarso in Oromiya. This variation suggests that it is unlikely that these
results are driven by systematic measurement errors across regions. In Figure 8.8, we
present a scattergram of the data on percentage of entitlements received (the vertical axis)
and the number of payments received in 2009 (the horizontal axis). This figure is consistent
with what emerges from a causal “eye-balling” of the data in Table 8.7, namely that in
woredas where payments were more frequent, beneficiaries were more likely to receive a
greater share of their entitlement.

143
Figure 8.8. Scattergram, woreda mean percentage entitlement received, by woreda
mean number of payments, public works, 2009
010 20 30 40 50 60 70 80 90 100 110
Mean percent entitlement rec'd
1 2 3 4 5 6 7
Mean number of payments
Source: Authors’ calculations based on the PSNP survey data.
Table 8.9 presents data on some other dimensions of implementation performance at the
woreda level: days worked in 2009 and 2010, number of payments made in 2010, and
household size. To simplify the analysis of these relative to entitlements, we calculate
Spearman correlation coefficients. A Spearman correlation coefficient is a correlation
measure based on ranks. It ranges in value from -1 to +1. A positive coefficient means that
the best performing woredas by one measure (here, entitlements) are also the best
performing woredas by some other measure. A negative coefficient means that the best
performing woredas by one measure (here, entitlements) are the worst performing woredas
by some other measure. These are presented in Table 8.10 for all woredas and also by
region.

144
Table 8.9. Entitlements, days worked in 2009 and 2010, number of payments, 2010,
and household size, by woreda
Region
Woreda
Percent
entitlement
received,
2009
Days
worked,
January-July
2009
Days
worked,
January-June
2010
Total number of
payments,
January-June
2010
Household
size, 2010
Oromiya
Zeway Dugda
13.0
115
86
1.07
6.40
Amhara
Angolela Tera
14.9
78
49
1.00
5.56
Oromiya
Gelana
18.5
41
31
1.47
6.35
Amhara
Ebenat
19.1
56
64
0.93
5.27
Oromiya
Goro
21.3
48
34
3.18
5.94
Oromiya
Anchar
23.1
64
63
1.00
6.06
Oromiya
Seweyna
25.2
90
72
0.71
7.04
SNNPR
Alaba
28.2
70
81
2.65
5.65
Oromiya
Boset
29.5
49
43
2.86
5.59
Oromiya
Boke
29.9
61
37
1.00
4.43
Amhara
Artuma fursi
32.1
87
71
1.18
5.43
Oromiya
Mema Angetu
34.5
23
0
0.00
6.75
Amhara
Kelala
35.2
96
71
1.44
4.78
Amhara
Ambasel
35.7
110
92
3.79
5.16
Amhara
West Belesa
36.3
111
100
1.04
4.65
Tigray
Naedei Adet
37.0
122
86
0.81
6.00
Tigray
Atsegede Tsimbela
38.3
71
26
1.00
4.86
Amhara
Janamora
39.1
73
60
1.42
4.58
Oromiya
Fentale
42.0
42
21
0.53
6.35
Oromiya
Yabello
43.9
82
65
1.00
5.79
Oromiya
Gola Oda
44.2
15
36
1.19
7.25
Amhara
Beyeda
46.7
91
75
0.00
5.44
Oromiya
Gursum
47.5
21
18
0.38
6.58
Oromiya
Daro Lebu
48.5
130
83
0.17
6.33
Tigray
Kolla Temben
48.9
123
43
0.95
6.52
Amhara-HVFB
Dehana
49.3
116
93
1.93
4.80
SNNPR
Soro
49.4
56
49
2.86
6.45
Tigray
Atsbi Wenberta
50.3
113
101
2.00
5.43
Amhara-HVFB
Tach Gayint
52.7
117
101
2.27
5.02
Oromiya
Babile
53.9
50
40
1.60
6.60
Tigray
Saesi Tsaedamba
56.2
114
75
1.59
5.56
Amhara
Dawa chefa
56.9
114
76
1.81
4.86
Amhara-HVFB
Wadla
57.3
98
80
2.04
5.16
Oromiya
Fedis
57.6
73
61
1.00
6.00
SNNPR
Amaro
59.0
90
66
2.75
5.25
Amhara
Were IIu
59.2
108
66
2.45
4.00
Tigray
Ahiferom
61.2
142
86
1.13
5.74
Amhara-HVFB
Kobo
61.4
96
74
2.29
4.74
Tigray
Tanqua Abergele
63.7
100
76
3.19
6.09
Amhara-HVFB
Gubalafto
63.8
121
97
3.40
5.60
Amhara
Legambo
65.4
127
107
0.78
4.74

145
Region
Woreda
Percent
entitlement
received,
2009
Days
worked,
January-July
2009
Days
worked,
January-June
2010
Total number of
payments,
January-June
2010
Household
size, 2010
Amhara
Sayint
65.5
80
65
1.65
5.24
Amhara-HVFB
Lay Gayint
65.8
115
78
0.96
4.79
Oromiya
Guba Koricha
68.0
133
76
1.00
6.53
Tigray
Enderta
69.4
130
103
2.73
5.42
Amhara-HVFB
Sokota
69.6
55
44
2.22
4.38
Oromiya
Mieso
70.3
82
66
1.84
5.74
SNNPR
Burji
71.0
164
137
2.48
6.78
SNNPR
Badewacho
73.1
138
118
3.33
6.14
Amhara-HVFB
Habru
74.4
142
117
1.40
4.40
SNNPR
Damot Gale
75.3
98
78
3.39
5.39
Amhara-HVFB
Bugna
75.4
85
69
4.08
3.85
Tigray
Gulomehada
76.3
127
105
1.42
5.63
SNNPR
Zala
77.5
113
90
3.38
5.06
Tigray
Ofla
77.9
102
45
1.29
5.92
Amhara-HVFB
Zukela
80.4
123
101
3.68
6.09
Amhara
Kalu
81.5
143
119
1.24
4.88
Amhara-HVFB
Simada
81.6
89
54
1.04
5.43
Tigray
Alamata
82.1
110
81
2.62
5.62
Amhara
Debresina
82.4
123
101
1.40
4.40
Amhara
Gidan
82.6
103
86
3.31
4.56
SNNPR
Aleta Wendo
87.8
125
103
2.68
5.95
Tigray
Wukro
88.0
142
119
3.35
5.98
Amhara
Tena
91.1
92
79
4.38
4.62
Amhara
Habru
91.3
73
47
1.85
4.05
SNNPR
Omo Sheleko
92.3
132
110
2.85
5.62
SNNPR
Boloso sore
95.6
106
90
3.55
5.19
SNNPR
Shebedino
97.1
97
82
3.67
6.92
SNNPR
Kemba
97.2
118
100
4.22
5.22
SNNPR
Bonkie
98.6
158
141
4.56
6.72
SNNPR
Gofa zuria
100.8
87
65
2.72
5.50
SNNPR
Kindo Koisha
114.2
77
63
3.59
5.53
Oromiya
Jarso
125.8
56
32
1.12
6.29
SNNPR
Damote Weyede
137.6
103
91
4.60
5.10
SNNPR
Konso
151.4
126
100
3.02
5.29
Source: Authors’ calculations based on the PSNP survey data.
Table 8.10 confirms the evidence shown in Figure 8.8—woredas that made more payments
were more likely to provide beneficiaries with levels of transfers that reflected their
entitlements. In SNNPR, where most transfers took the form of cash, the mean number of
cash payments made in 2009 was more strongly correlated with the mean percentage of
entitlements received than with the mean number of food payments. In the other regions, the
ability to make food payments was much more strongly correlated with meeting entitlements.

146
We wondered if the ability of woredas to provide work was correlated with payments. Tables
8.9 and 8.10 suggest that some correlations exist but these are relatively smaller than those
found for payment frequency and, at the regional level, less precisely measured. We note,
however, that woredas that provided larger transfers as a percentage of entitlements in 2009
tended to make more payments in the first five months of 2010 in Tigray, Amhara, and
SNNPR. This suggests that, at least over 2009 and 2010, woredas that performed well in
terms of payments in one year tended to do so in the other year and vice versa. We also
wondered if some of these woreda differences were a result of differences in mean
household sizes across woredas. Table 8.9 suggests that there is some variation in size by
location; however, Table 8.10 suggests that these variations are not, generally, correlated
with household size. The one exception is Amhara. Figure 8.6b showed that in Amhara,
there appears to have been no differentiation of payments by household size and so it is not
surprising that within Amhara, woredas with larger households do less well in terms of
providing payments that reflect entitlements.
Table 8.10. Spearman correlation coefficients between entitlements and other
measures of PSNP implementation (public works)
All
Tigray
Amhara
Amhara-
HVFB
Oromiya
SNNPR
Number of payments, 2009
0.68***
0.65**
0.62***
0.43
0.55**
0.78***
Number of cash payments, 2009
0.29**
0.01
0.27
-0.27
0.12
0.72***
Number of food payments, 2009
0.40***
0.62**
0.58**
0.44
0.70***
0.22
Days worked, Jan–Jul 2009
0.41***
0.20
0.30
-0.15
0.16
0.16
Days worked, Jan–Jun 2010
0.42***
0.46
0.29
-0.31
0.04
0.13
Number of payments, Jan–Jun 2010
0.55***
0.71**
0.48**
0.11
-0.09
0.58**
Household size
-0.13
0.04
-0.65**
0.14
0.01
-0.29
Source: Authors’ calculations based on the PSNP survey data.
Notes: * Significant at the 10 percent level; ** Significant at the 5 percent level; *** Significant at the 1 percent level.
If payments are not being made on the basis of entitlements, how are they being made?
Tables 8.11 and 8.12 provide some insights. Table 8.11 cross-tabulates the number of
months worked by the number of payments received in 2009. Cells to the right and above
the diagonal line represent cases where the number of payments exceeded the number of
months worked. Cells with the diagonal line are cases where the number of payments
equaled the number of months worked and cells below and to the left of the diagonal line are
cases where frequency of payment was less than frequency of work. To interpret Table 8.11,
it is helpful to work through several examples. In 2010, in SNNPR (Table 8.11e), 347
households reported that they did public works under the PSNP for which they were paid.
About 45 percent reported that they worked for six months (the row total for number of
months worked is 157 and 157/347 = 0.45). A small number of households worked fewer
than four months
24
. 88 out of those 157 households were paid on six different occasions (88
is the number in the cell where number of months worked and number of payments both
equal 6).
24
The number of households reporting that they worked 12 months, 64, seems high. It is possible that at least some of these
cases arise where respondents claim, “I worked every month.”

147
The usefulness of Table 8.11 is the following. A simple explanation for low payments as a
fraction of entitlements and low frequency of payments being made in some regions is that,
for whatever reason, the number of months households reported working was lower in some
regions than in others. But Table 8.11 indicates that this is not a plausible explanation. In
Amhara, only 11 out of 301 households report that they worked in four, or fewer, months in
2009. But roughly half of PSNP public works beneficiaries in Amhara report that they were
paid on fewer than five occasions. While we observe a similar pattern in Amhara-HVFB, a
comparison of Figures 8.6b and 8.6c tells us that transfer levels in Amhara-HVFB are higher
than in Amhara. (Note that Figure 8.6 is based on normalized transfer levels, so the
difference between Amhara-HVFB and Amhara is not because of differences in grain prices
or because of the additional pulses and lentils provided in Amhara-HVFB.) Put in another
way, while the distribution of months worked and the frequency of payments look similar in
Amhara and Amhara-HVFB, in Amhara-HVFB the payments that are made are larger. This
is also consistent with the finding in Table 8.10 that food payments frequency is more weakly
correlated with percent entitlements received in Amhara-HVFB than in Amhara.
In Oromiya, there is a slightly higher number of households reporting working fewer than six
months. While the number of households reporting working 12 months is high, the more
salient number is the sum of households receiving six or more payments, only 72 out of 265
public works beneficiaries.
Table 8.11a. Number of months worked and number of payments, 2009, Tigray
Number of payments
Number
of
months
worked
1
2
3
4
5
6
7
8
9
10
11
12
Row
total
1
0
0
0
0
0
0
0
0
0
0
0
0
0
2
1
1
0
0
0
1
0
0
0
0
0
0
3
3
1
0
4
1
0
3
0
0
0
0
0
0
9
4
2
3
1
1
0
0
0
0
0
0
0
0
7
5
0
2
1
1
2
6
2
0
0
0
0
0
14
6
1
3
14
4
13
50
5
3
0
0
0
0
93
7
0
0
6
2
10
17
12
5
1
0
0
0
53
8
0
2
6
6
5
9
7
29
0
0
0
0
64
9
0
2
6
1
3
10
3
5
0
0
0
0
30
10
0
0
13
6
3
7
8
1
0
1
0
0
39
11
0
0
3
2
0
0
0
3
0
0
0
0
8
12
2
2
21
17
7
21
17
10
0
0
0
4
101
Column
total
7
15
75
41
43
124
54
56
1
1
0
0
421
Source: Authors’ calculations based on the PSNP survey data.

148
Table 8.11b. Number of months worked and number of payments, 2009, Amhara
Number of payments
Number
of
months
worked
1
2
3
4
5
6
7
8
9
10
11
12
Row
total
1
0
0
0
0
0
0
0
0
0
0
0
0
0
2
0
0
1
0
0
0
0
0
0
0
0
0
1
3
3
1
1
0
0
0
0
0
0
0
0
0
5
4
1
2
1
1
0
0
0
0
0
0
0
0
5
5
1
3
7
10
3
4
0
1
0
0
0
0
29
6
1
10
32
20
12
43
1
3
1
0
0
0
123
7
3
14
14
4
3
21
9
0
0
0
0
0
68
8
1
14
15
5
2
6
1
0
0
0
0
0
44
9
0
2
0
2
0
0
0
0
0
0
0
0
4
10
0
0
0
0
0
1
0
0
0
0
0
0
1
11
0
0
0
0
0
1
0
0
0
0
0
0
1
12
0
3
4
1
0
9
1
1
0
1
0
0
20
Column
total
10
49
75
43
20
85
12
5
1
1
0
0
301
Source: Authors’ calculations based on the PSNP survey data.
Table 8.11c. Number of months worked and number of payments, 2009, Amhara-HVFB
Number of payments
Number
of
months
worked
1
2
3
4
5
6
7
8
9
10
11
12
Row
total
1
0
0
1
0
0
0
0
0
0
0
0
0
1
2
0
0
1
0
0
0
0
0
0
0
0
0
1
3
0
0
1
1
0
2
0
1
0
0
0
0
5
4
0
0
4
2
0
1
0
0
0
0
0
0
7
5
0
0
14
3
3
3
0
0
0
0
0
0
23
6
0
6
54
46
2
26
1
6
1
1
0
0
143
7
0
5
21
16
8
4
0
4
0
1
0
0
59
8
0
3
21
12
7
0
1
1
1
0
0
0
46
9
1
1
7
3
7
0
0
0
0
1
0
0
20
10
0
1
3
2
1
0
0
0
0
0
0
0
7
11
1
0
3
0
0
0
0
0
0
0
0
0
4
12
1
6
5
4
2
1
1
7
1
0
0
0
28
Column
total
3
22
135
89
30
37
3
19
3
3
0
0
344
Source: Authors’ calculations based on the PSNP survey data.

149
Table 8.11d. Number of months worked and number of payments, 2009, Oromiya
Number of payments
Number
of
months
worked
1
2
3
4
5
6
7
8
9
10
11
12
Row
total
1
1
0
0
1
0
0
0
0
0
0
0
0
2
2
0
1
0
1
0
0
1
0
0
1
0
0
4
3
0
2
2
1
7
2
2
0
0
0
0
0
16
4
2
1
1
5
4
1
0
1
0
0
0
0
15
5
2
1
1
2
8
4
1
0
0
0
0
0
19
6
0
2
3
1
16
13
0
1
0
0
0
0
36
7
0
2
0
2
2
1
0
2
0
0
1
0
10
8
0
7
3
6
2
2
3
6
0
0
0
0
29
9
0
6
1
2
1
0
0
1
1
0
0
0
12
10
2
0
0
4
1
0
2
0
0
0
0
0
9
11
1
0
3
0
1
0
0
0
0
0
0
0
5
12
9
29
24
13
7
11
3
3
1
1
1
6
108
Column
total
17
51
38
38
49
34
12
14
2
2
2
6
265
Source: Authors’ calculations based on the PSNP survey data.
Table 8.11e. Number of months worked and number of payments, 2009, SNNPR
Number of payments
Number
of
months
worked
1
2
3
4
5
6
7
8
9
10
11
12
Row
total
1
5
0
0
0
0
0
0
0
0
0
0
0
5
2
0
0
0
0
0
0
0
0
0
0
0
0
0
3
0
0
3
1
0
0
0
0
0
0
0
0
4
4
0
0
1
0
0
0
0
0
0
0
0
0
1
5
0
0
3
1
1
12
0
1
0
0
0
0
18
6
0
0
18
17
20
88
3
10
0
1
0
0
157
7
0
1
4
4
2
17
4
3
1
0
0
0
36
8
0
0
0
1
1
5
1
12
3
0
0
0
23
9
0
0
0
0
6
1
0
4
5
0
0
0
16
10
0
0
0
0
1
0
0
0
0
1
0
1
3
11
0
0
0
0
1
16
1
2
0
0
0
0
20
12
0
1
0
0
4
7
1
13
1
2
1
34
64
Column
total
5
2
29
24
36
146
10
45
10
4
1
35
347
Source: Authors’ calculations based on the PSNP survey data.
Finally, we calculate an “implied wage” for 2009. That is, we take the total normalized
payments for 2009 (so as to strip out the impact of variations in grain prices and the
provision of legumes and oils in Amhara-HVFB) and divide it by the total number of days
worked in 2009. This quotient is referred to as “Implied Wage (1)” in Table 8.12. To correct
for possible overreporting of days worked, or cases where respondents confuse unpaid
community labor with PSNP public works, we construct a second measure where—

150
arbitrarily—we restrict the number of days a household could report working to a maximum
of 180. This quotient is referred to as “Implied Wage (2)” in Table 8.12
Table 8.12. Distribution of implied wages (Birr), by region and assumptions of labor
supplied
Distribution of implied wages
Region
10th
percentile
25th
percentile
Median
75th
percentile
90th
percentile
Tigray
Implied Wage (1)
2.54
4.02
6.67
10.00
11.67
Implied Wage (2)
3.33
4.81
8.33
10.00
13.33
Amhara
Implied Wage (1)
2.13
3.25
5.44
10.00
12.50
Implied Wage (2)
2.22
3.37
5.93
10.00
12.50
Amhara-HVFB
Implied Wage (1)
2.96
4.29
6.38
10.00
17.50
Implied Wage (2)
3.46
4.67
6.86
10.30
17.50
Oromiya
Implied Wage (1)
1.74
3.73
6.94
12.00
25.42
Implied Wage (2)
2.22
4.10
7.39
12.11
26.67
SNNPR
Implied Wage (1)
4.00
6.02
10.00
15.28
22.22
Implied Wage (2)
4.33
7.69
10.98
16.67
23.25
Source: Authors’ calculations based on the PSNP survey data.
The combined effect of full family targeting (FFT) and the labor cap means that the implied
wage can exceed 10 Birr. Given this, the implied wages greater than 10 Birr/day, seen at the
75th and 90th percentiles of the distribution of implied wages, suggests that the labor cap is,
in fact, being implemented to some extent. Even after adjusting for possible overreporting of
days worked, the median implied wage in Amhara is only 5.93. This suggests that, at least
for 2009, that not only were beneficiary households not receiving their full entitlement of
payments, they were not even receiving the equivalent of 10 Birr per day. But as with the
analysis presented above, it is again instructive to disaggregate to the woreda level. Being
careful to note that in some woredas, sample sizes get very small, Table 8.13 shows the
distribution of implied wages—with the arbitrary correction for overreporting of days worked.
Table 8.13 shows that just as there is variation across regions, there are also variations
within regions. Even in Amhara, there are woredas such as Habru and Tena that are paying
most PSNP public works participants an implied wage close to the 10 Birr per day that
households should receive. But there are many woredas in Amhara, and elsewhere, that did
not even meet this payment level.
Finally, in chapter 4, we showed that 82 percent of public works beneficiaries in SNNPR
“strongly agreed” or “agreed” that they received their payments in full. This percentage is
consistent with the good performance of SNNPR highlighted in all tables in this section.
Similarly, only 44 percent of beneficiaries in Oromiya agreed with this statement and this is
consistent with the relatively low percentages reported in Figure 8.7. But, puzzlingly, 89
percent of beneficiaries in Amhara “strongly agree” or “agree” with this statement, even
though payments are well below entitlements.

151
Table 8.13. Distribution of implied wages (Birr), by region and woreda under
conservative assumptions of labor supplied
Region and woreda
10th
percentile
25th
percentile
Median
75th
percentile
90th
percentile
Sample
size
Tigray
Atsegede Tsimbel
2.12
2.63
7.51
12.46
26.47
7
Ahiferom
2.78
4.00
6.67
7.50
9.56
39
Naedei Adet
1.67
2.71
4.07
4.56
7.78
21
Kolla Temben
3.06
4.91
8.83
10.00
10.00
21
Tanqua Abergele
3.33
8.15
10.00
11.43
13.33
43
Gulomehada
5.00
7.32
10.00
10.00
12.00
48
Saesi Tsaedamba
3.59
4.48
6.02
8.33
10.56
41
Wukro
5.56
8.33
10.00
13.33
17.76
49
Atsbi Wenberta
3.33
3.61
4.57
8.68
11.54
44
Enderta
2.64
4.17
6.29
8.23
11.48
32
Alamata
3.06
7.78
10.00
13.33
18.75
39
Ofla
5.58
7.67
10.00
11.17
22.22
38
Total
3.33
4.81
8.33
10.00
13.33
422
Amhara
Beyeda
5.42
5.42
6.89
8.57
8.57
32
Janamora
2.30
2.92
5.52
15.07
20.00
10
West Belesa
2.50
2.96
3.54
5.00
6.62
23
Ebenat
3.33
3.37
3.89
4.67
7.86
15
Gidan
6.86
8.27
10.11
11.39
13.61
16
Habru
9.00
10.00
11.81
14.17
20.83
20
Tena
7.29
8.00
9.11
10.00
20.56
10
Ambasel
2.13
3.13
4.21
6.25
6.80
18
Kalu
4.33
5.75
9.63
10.30
11.00
16
Legambo
2.29
3.06
8.33
10.00
16.00
23
Sayint
4.76
7.69
10.53
16.67
37.04
15
Debresina
1.11
5.00
7.73
13.06
38.85
9
Kelala
0.83
2.10
3.00
5.47
7.50
8
Were IIu
1.85
2.08
5.10
9.00
10.00
11
Angolela Tera
1.30
1.39
3.00
5.65
7.00
14
Dawa chefa
2.71
3.61
3.61
7.50
8.83
21
Artuma fursi
1.43
2.38
4.26
9.74
19.97
27
Total
2.38
3.61
6.25
10.00
12.86
292
Amhara-HVFB
Simada
3.41
4.00
8.25
10.91
19.56
15
Tach Gayint
3.33
4.31
5.57
7.92
9.96
62
Lay Gayint
2.92
3.89
5.69
8.52
12.00
46
Habru
4.18
4.23
4.95
8.16
10.69
4
Wadla
2.24
4.06
7.75
14.66
20.49
24
Bugna
7.04
7.22
7.78
8.33
12.52
26
Habru
4.17
5.08
6.81
10.76
15.00
35
Gubalafto
3.70
4.63
6.15
8.94
13.33
34
Kobo
4.22
5.42
5.96
9.48
13.23
30
Sokota
5.00
6.22
20.00
28.44
41.67
30
Zukela
8.33
10.00
11.83
15.00
25.47
22
Dehana
3.50
3.96
4.67
4.67
5.00
15
Total
3.50
4.67
6.91
10.33
17.50
343

152
Region and woreda
10th
percentile
25th
percentile
Median
75th
percentile
90th
percentile
Sample
size
Oromiya
Fentale
4.31
10.00
15.63
20.41
29.03
17
Boset
6.58
7.81
8.62
9.26
9.80
21
Zeway Dugda
0.56
0.74
1.48
3.36
5.56
12
Mieso
5.00
6.94
7.89
13.89
25.00
19
Guba Koricha
4.67
5.83
8.44
12.00
17.71
15
Anchar
1.63
2.22
2.94
4.12
5.19
14
Daro Lebu
1.92
2.50
4.44
7.50
7.96
5
Boke
1.96
1.99
5.56
9.46
10.54
7
Jarso
14.56
22.08
25.00
29.69
36.92
17
Gursum
7.30
10.78
24.06
55.00
72.41
20
Babile
1.94
5.66
18.85
38.67
49.01
4
Fedis
2.47
3.06
5.64
8.00
8.33
19
Gola Oda
3.70
4.17
16.30
65.97
100.00
14
mema Angetu
4.07
4.07
12.22
50.00
50.00
3
Goro
4.17
5.00
10.00
10.00
12.00
16
Seweyna
2.22
2.78
4.82
7.92
10.42
24
Yabello
4.17
4.44
6.32
8.33
10.53
14
Gelana
3.47
3.47
3.47
4.86
5.30
17
Total
2.42
4.27
7.80
12.58
26.67
258
SNNPR
Badewacho
4.05
5.00
7.69
10.00
11.90
21
Soro
1.79
2.08
16.67
25.00
25.00
21
Omo Sheleko
4.86
5.00
6.94
7.78
10.00
13
Shebedino
11.11
14.79
18.75
20.64
30.56
24
Aleta Wendo
10.00
10.00
10.00
13.33
16.67
22
Boloso sore
6.59
8.70
9.70
13.33
15.24
31
Damot Gale
5.83
6.08
13.35
16.67
20.83
23
Damote Weyede
4.44
7.50
11.04
13.33
18.05
10
Kindo Koisha
11.43
13.33
16.79
20.00
23.33
17
Kemba
5.43
9.26
10.00
13.64
18.52
18
Bonkie
10.00
10.00
12.00
13.60
23.33
18
Gofa zuria
8.58
12.86
15.97
23.30
40.83
18
Zala
4.17
7.50
10.00
11.67
20.00
16
Amaro
5.75
7.43
9.76
13.54
16.67
4
Burji
5.95
6.94
11.11
17.01
20.83
23
Konso
7.33
9.67
12.59
16.67
26.67
41
Alaba
2.41
2.67
3.85
6.41
9.62
26
Total
4.44
7.73
11.04
16.67
23.25
346
Source: Authors’ calculations based on the PSNP survey data.
8.5. Payments for Direct Support
We now examine direct support payments, starting with the trend in the mean level of these
payments between 2006 and 2010. We define a direct support beneficiary in 2006 as
someone who received at least one direct support payment in the first five months (Tir to
Ginbot) of 2006, a 2007 beneficiary as someone who received at least one direct support
payment in the first five months (Tir to Ginbot) of 2007, and so on. The amounts shown in
Figure 8.9 are the sum of cash payments received plus the value of food transfers. They are

153
nominal amounts, that is, they do not take inflation into account, which was substantial over
this period.
Figure 8.9 shows that mean direct support payments are, in all years, highest in SNNPR,
followed by Amhara-HVFB and Tigray. In these localities, there are increases in nominal
payments between 2006 (2007 for Amhara-HVFB) and 2009, followed by a reduction in
2010. In Amhara, mean payments have fallen in every year since 2007. While payments
rose between 2006 and 2009 in Oromiya, there is little difference in mean payments in 2010
and 2007.
Figure 8.9. Mean payments to direct support beneficiary households, by region and
year
Source: Authors’ calculations based on the PSNP survey data.
Table 8.14 shows the extent to which payments were received in cash and in-kind during the
first five months of 2009 and 2010. As with payments for public works, cash payments
predominate in SNNPR, while in-kind transfers are more important in Tigray and Amhara-
HVFB. There were virtually no food transfers made to direct support beneficiaries in Amhara
in 2010.
In addition to the level of payments, we are also interested in their frequency—how often do
beneficiaries receive their payments. Tables 8.15 and 8.16 report the number of times a
direct support household received payments in the first five months of 2009 and 2010. There
are distinct regional patterns. In SNNPR, nearly 80 percent of beneficiaries received at least
four payments in the first five months of 2009, as did 43 percent of beneficiaries in Tigray. By

154
contrast, less than 16 percent did so in Oromiya. A similar regional pattern is observed in
2010 but with lowered frequency of payments found everywhere.
Table 8.14. Mean in-kind and cash direct support payments 2009 and 2010, by region
Tir–Ginbot 2009
Tir–Ginbot 2010
Value of in-kind payments
Cash
Value of in-kind payments
Cash
Tigray
388
66
256
55
Amhara
73
159
20
133
Amhara-HVFB
490
54
303
68
Tigray
233
139
74
211
SNNPR
139
483
51
469
Source: Authors’ calculations based on the PSNP survey data.
Table 8.15. Frequency of direct support payments (percent), Tir–Ginbot 2009, by
region
Region
Number of payments
Tigray
Amhara
Amhara-HVFB
Oromiya
SNNPR
1
22.4
33.3
12.2
31.9
3.3
2
21.9
25.3
33.3
24.6
5.5
3
10.4
13.1
26.0
27.5
13.2
4
13.1
7.1
2.4
8.7
16.5
5
32.2
21.2
26.0
7.2
61.5
Source: Authors’ calculations based on the PSNP survey data.
Table 8.16. Frequency of direct support payments (percent), Tir–Ginbot 2010, by
region
Region
Number of payments
Tigray
Amhara
Amhara-HVFB
Oromiya
SNNPR
1
32.3
34.5
21.9
60.3
4.6
2
23.7
39.1
32.1
20.6
15.9
3
27.3
16.1
18.3
14.3
28.4
4
9.1
5.7
8.0
3.2
22.7
5
7.6
4.6
19.7
1.6
28.4
Source: Authors’ calculations based on the PSNP survey data.
It is possible that Figure 8.9 as well as Tables 8.15 and 8.16 understate the amount of
transfers being made because they only cover the first five months of 2009 and 2010. For
this reason, we report, in Table 8.17, the median level of transfers per month for all 12
months of 2009 along with the number of households receiving direct support.
Table 8.17 indicates that direct support payments do continue into Sene, and to a lesser
extent into Hamile and Nehase. It also shows that the number of beneficiaries falls in the
second half of the year. For example, there are 392 households receiving direct support in
Megabit but only 114 in Meskerem. Table 8.17 shows that the regions that did not provide
significant levels of payments in the first five months of 2009 did not compensate with either
higher payments or more extensive coverage in the second half of the year.

155
Table 8.17. Median value (Birr) of direct support transfers, by month and region, 2009
Region
Month
Tigray
Amhara
Amhara-HVFB
Oromiya
SNNPR
Tir
91
(78)
50
(30)
113
(47)
60
(35)
150
(74)
Yekatit
95
(98)
50
(42)
106
(70)
100
(13)
150
(73)
Megabit
100
(130)
55
(66)
112
(74)
73
(38)
150
(84)
Miaza
75
(148)
97
(61)
83
(88)
100
(34)
150
(78)
Ginbot
95
(115)
75
(56)
119
(86)
67
(42)
150
(80)
Sene
98
(134)
75
(77)
68
(88)
86
(40)
150
(74)
Hamile
75
(94)
112
(48)
54
(51)
95
(28)
150
(44)
Nehase
100
(86)
117
(25)
109
(35)
90
(36)
150
(38)
Meskerem
36
(54)
59
(17)
169
(6)
90
(15)
100
(22)
Tikmit
50
(28)
43
(10)
0
(0)
103
(8)
100
(14)
Hidar
50
(19)
54
(11)
50
(13)
50
(13)
100
(18)
Tahisis
52
(15)
50
(9)
121
(10)
60
(9)
100
(14)
Source: Authors’ calculations based on the PSNP survey data.
Note: Number of beneficiaries in parentheses.
Table 8.17 also indicates that in 2009, the number of direct support beneficiaries is highest
in Tigray. We explore this further in Table 8.18, which looks at payment size, the number of
payments, and the number of households receiving direct support between Tir and Ginbot
2010. (Note that we present both the mean and median so as to see whether outliers in the
data affect the pattern we observe.) What is striking here is that while median payments are
nearly two-times higher in SNNPR compared to Tigray, Tigray has slightly more than twice
the number of beneficiaries. Both payment levels and the number of beneficiaries are lowest
in Amhara and Oromiya.
In light of this information, Figure 8.10 is instructive. It shows the share of all payments made
to direct support beneficiaries in Tigray, Amhara, Oromiya, and SNNPR between Tir and
Ginbot 2010. The samples for these four regions are approximately similar
25
and so if each
region was allocating the same level of direct support payments, the pie chart would show
each region accounting for a quarter of the overall budget. Instead, we see that
approximately 44 percent of total direct support payments made in this sample are disbursed
in Tigray and 33 percent in SNNPR. Amhara and Oromiya collectively account for only 23
percent of direct support spending.
25
In fact, the sample sizes in Amhara and Oromiya are slightly larger. We exclude Amhara-HVFB from this figure because it is
drawn from a much larger sample.

156
Table 8.18. Direct support payment data for Tir–Ginbot, by region, 2010
Region
Mean (Birr)
Median (Birr)
Number of beneficiaries
Tigray
Payment size
124.9
82.2
198
Number of payments
2.4
2.0
198
Amhara
Payment size
79.9
50.0
87
Number of payments
2.1
2.0
87
Amhara-HVFB
Payment size
130.6
75.7
137
Number of payments
2.7
2.0
137
Oromiya
Payment size
199.6
69.8
63
Number of payments
1.7
1.0
63
SNNPR
Payment size
152.5
150.0
88
Number of payments
3.5
4.0
88
Source: Authors’ calculations based on the PSNP survey data.
Figure 8.10. Distribution of direct support payments across regions, between Tir and
Ginbot 2010
Source: Authors’ calculations based on the PSNP survey data.
Lastly, we assessed how direct support payments varied by household size. Interpreting
these data is a little tricky as a result of the small numbers of beneficiaries in Amhara and
Oromiya—an atypically high or low level of payment reported by one household can skew
mean values. We partially remedy this in Table 8.19 by reporting transfer levels in terms of
median values. These show that in SNNPR direct support payments rise with household
size. In all other regions, direct support payments are effectively the same irrespective of
how many people reside in the household.

157
Table 8.19. Median direct support payments, by household size and region, 2009
Household size
Region
1
2
3
4
5
6
Tigray
405
410
602
392
617
585
Amhara
388
212
336
525
408
338
Amhara-HVFB
277
540
592
543
400
168
Oromiya
112
765
319
463
379
330
SNNPR
483
755
900
1,178
1,200
1,472
Source: Authors’ calculations based on the PSNP survey data.
We complement Table 8.19 by assessing the extent to which direct support payments are
consistent with Full Family Targeting in 2009. This is shown in Figure 8.11 which shows the
distribution of payments in terms of the percentage of entitlement that is actually received.
Only in SNNPR we do see that more than 50 percent of direct support beneficiaries receive
their full entitlement. Tigray has the highest number of beneficiaries but payment levels (as a
percentage of entitlements under FFT) are lower than they are in SNNPR. Amhara and
Oromiya come out especially badly—not only are their relatively few direct support
beneficiaries, direct support households typically get only a small fraction of their entitlement.
Figure 8.11. Distribution of direct support payments as a percent of entitlement under
full family targeting, 2009
Source: Authors’ calculations based on the PSNP survey data.
Taken collectively, these data present a mixed picture on the level of direct support
transfers. On the positive side, there has been a marked increase in the nominal values of
these transfers after 2006 in Tigray and SNNPR. These regions have adopted somewhat

158
different approaches. In Tigray, there are more beneficiaries but with lower transfer levels,
while in SNNPR, there are fewer beneficiaries receiving higher levels of transfers. By
contrast, there are fewer recipients of direct support in Amhara and Oromiya and these
recipients receive smaller transfers.
8.6. Preferences for Cash and Food
An initial goal of the PSNP was “to shift the financing of the program from food aid to cash”
(GFDRE 2004, 1). This was more than a signal of the government’s intent to phase out
nonemergency food aid; it also recognized the developmental potential of cash transfers. It
was hoped that through the provision of cash transfers, the PSNP would enable
smallholders to increase consumption and investment levels while stimulating the
development of rural markets.
In the 2006, 2008, and 2010 quantitative surveys, respondents were asked if they would
prefer their payments entirely in cash, 75 percent cash and 25 percent food, an even split
between cash and food, 25 percent cash and 75 percent food, or entirely in food. These
responses for both public works and direct support recipients are summarized by region in
Figures 8.12a–8.12e.
Figure 8.12a. Preferences for food and cash, 2006–2010, Tigray
Source: Authors’ calculations based on the PSNP survey data.

159
Figure 8.12b. Preferences for food and cash, 2006–2010, Amhara
Source: Authors’ calculations based on the PSNP survey data.
Figure 8.12c. Preferences for food and cash, 2006–2010, Amhara-HVFB
Source: Authors’ calculations based on the PSNP survey data.

160
Figure 8.12d. Preferences for food and cash, 2006–2010, Oromiya
Source: Authors’ calculations based on the PSNP survey data.
Figure 8.12e. Preferences for food and cash, 2006–2010, SNNPR
Source: Authors’ calculations based on the PSNP survey data.

161
Across all regions, there are several common features. First, beneficiary preferences shift
toward food and away from cash between 2006 and 2008, a consequence of rapidly rising
food prices in the first six months of 2008. Second, in all regions apart from Tigray, this is
reversed between 2008 and 2010, although this increased preference for cash is more
pronounced among public works beneficiaries. Third, in all locations and in all years, direct
support beneficiaries are more likely to express preferences for food than public works
beneficiaries. Outside of SNNPR, less than 25 percent of direct support recipients would
prefer that more than half of their payments would be made in cash. Fourth, regional
differences have persisted over time with PSNP beneficiaries in Tigray always having the
strongest preference for food and SNNPR beneficiaries the strongest preference for cash.
Preferences for cash and food were explored in 40 focus group discussions.
26
When asked
about their preference for payment type, 17 of 38 groups (2 did not respond) reached
consensus that they would prefer to be paid only in food. A further 15 said they wanted to
receive a mix of food and cash and only 5 groups indicated that they would like cash only.
In focus groups that indicated a preference for food, most justified this due to the lack of
availability of food in the local markets and poor growing conditions. Furthermore, many said
that prices increased significantly when they receive cash.
We want food. The reason we said food is that although we may take the
cash, the great part of it is used to buy food. Moreover, once the cash goes
into the pocket of men, they do not get happy to give out for grain buying
[SNNPR_S/FG3].
If the payment is in cash, we could not be able to buy food from the market
easily, because there is no sufficient supply of food items like grains in the
marketplace [AM_S/FG1].
If you want to buy food, it is very expensive. When money is given, the
traders talk to each other and raise the price [AM_S/FG4].
We prefer food payment. Our community knows which payment is important
and we can get money by working in different areas but cannot get easily
grains in our area [TIG_SA/FG1].
When we are paid in cash the market price of food items increases, but when
paid in kind the market prices are fair [TIG_SA/FG2].
Most of the respondents who revealed a preference for a mixed transfer explained that they
use the food for their household’s own consumption and the money for other needs, such as
education and clothes—in other words, reasons related to the differential uses of cash and
food. A direct support group in SNNPR agreed that when
We are paid in food; we do not face a problem of having to eat with other
families and people in the community. The cash is needed to cover expenses
of fertilizer and improved seeds, tax, to buy and create assets, and to cover
expenses of clothing [SNNPR_T/FG1].
26
Across these focus groups, 22 received their payments as a mix of food and cash, 14 received cash only, and 4 received
only food.

162
The Transitioning group in the same kebele indicated that
We prefer both...if we say we can buy grain using the money, the grain prices
get expensive. Thus there are times that we prefer grain. The reason we say
cash is to cover school expenses, tax, to buy fertilizer and others we need
money. Thus, we prefer to continue with a mixture—as it is now
[SNNPR_T/FG2].
Of the few households with a preference for cash only, this was attributed to the flexibility of
cash to enable them to use it for a variety of things (Table 8.20), as well as investing in
growing their own food.
We prefer cash payment. All of us prefer cash. This is because of two
reasons. First, if we bring the grain home, it is not enough for a month and
second, if it is sold it does not fetch as much as the value of cash payment
[SNNPR_D/FG3].
Cash, because it can be used to pay back the lent money. Nowadays, food
such as maize can be bought for 300ETB/100kg and wheat for
600ETB/100kg. So the cash can be flexibly used [ORO_Z/FG1].
Table 8.20. Use of cash transfer mentioned, by focus groups, in order of priority
usage
Use
Total
First
Second
Third
Food
27
19
6
2
Schooling
16
2
6
8
Livestock
14
3
5
6
Seeds/fertilizer
13
6
4
3
Clothes
8
0
3
5
Other
9
2
5
2
Source: FSP qualitative evaluation, 2010.
Focus group participants were asked the hypothetical question, “If you could receive 15 kg of
grain a month or 50 Birr, what would you chose?” Of the 37 focus groups that provided
responses, 19 agreed that they would prefer 15 kg of grain, 5 said they would prefer 50 Birr,
and 3 said that they would prefer to switch payment across seasons—specifically, post-
March there is a preference for grain and pre-March a preference for cash. When probed,
the reasons for the strong preference for food-only payment can be clustered into three
categories: (1) the unequal current value of food and cash (“We told you that 15 kg is more
than 50 Birr and hence we prefer it” [TIG_SA/FG1].); (2) general market/price uncertainty
and therefore a preference for a certain value of food (“Market uncertainty makes it difficult
to answer. Today’s price is not the same as yesterdays” [SNNPR_S/FG2].); and (3)
deliberate price hiking at the time of payment (“We prefer food transfers, even if the cash
payment is increased, because the moment merchants know the payment is in cash, the
price increases. Thus we are forced to purchase food items before the payment days. The
merchants exchange information with the mobile telephone and play with the food prices”
[TIG_SA/FG3]. “We still prefer to be paid in food, because the cash does not guarantee that

163
we get food items in the market. The merchants also increase prices/play a kind of game
when they know that we are paid in cash” [TIG_SA/FG2].).
Respondents preferring food-only were asked whether they would still prefer food if the cash
amount was increased seasonally. All of these respondents still preferred food. The main
reasons given for this were the lack of food available on the market and increasing food
costs. Many respondents were not convinced that an increase would ever be sufficient
enough to cover the continual rise in prices. “No, if the cash increases, we [still] cannot buy
the amount that we need” [AM_SA/FG4]. “We prefer food payments, because we use it for
household consumption; if it is money, the money cannot buy enough food from the market”
[AM_S/FG3]. One women’s focus group discussed the possible misuse of cash in the hands
of men. This led them to agree that, “Still we prefer food. When the pay is cash, men usually
go to collect it, but also misuse the cash before reaching home” [ORO_G/FG3].
The dominant theme within these focus group discussions is the differential value of the
cash and food payments. To explore this further, we exploit data collected in the market
price survey that was fielded at the same time as the quantitative household survey. Table
8.21 shows locality (EA) median prices for the six most frequently consumed grains in rural
Ethiopia as of June 2010.
Table 8.21. Locality median price per kg, by crop, June 2010
Food
Median price per kg
Number of observations
White teff
7.60
100
Black/mixed teff
7.12
23
Barley
4.61
128
Wheat
6.11
35
Maize
3.88
137
Sorghum
4.30
112
Source: Authors’ calculations based on the PSNP survey data.
The 10 Birr cash payment should be sufficient to purchase 3 kg of maize. This is based on
an assumed price of 3.33 Birr/kg. As Table 8.16 shows, only maize prices are close to this
assumed price. Mindful of this, we calculate the mean preference for the mix of food and
cash payments at the locality (EA) level and categorize these as “Mostly cash,” “Even split,”
and “Mostly food.”
27
We cross-tabulate these with mean EA maize prices grouped into the
following tertiles: 1.33–3.35 Birr/kg; 3.35–4.48 Birr/kg; and 4.48–16.00 Birr/kg. Results are
show in Figure 8.12.
Figure 8.13 shows that when maize prices are below 3.35 Birr/kg, a majority of beneficiaries
would prefer mostly cash payments or an even split of food and cash. However, once prices
rise above 3.35 Birr/kg, fewer than 15 percent want mostly cash payments and more than
half want mostly food payments. These results are consistent with what we observe in the
qualitative data.
27
To construct these categories, we calculated the mean EA values based on individual responses. These equaled one where
all beneficiaries indicate a preference for cash, three where, on average, there is a preference for an equal split between food
and cash, and five where all beneficiaries indicate a preference for food. ”Mostly cash” were EAs where the mean value
ranged from 1.0 to 2.50. “Even split” EAs were those where the mean value ranged from 2.51 to 3.50 and “Mostly food” EAs
were those where the mean value ranged from 3.51 to 5.00.

164
Figure 8.13. Locality preferences for food and cash, by maize prices
Source: Authors’ calculations based on the PSNP survey data.
Respondents were asked whether there are certain times of the year that they would prefer
to have payments in food or cash. The general consensus on this was that people require
food during the rainy season and cash during the dry season. These vary across the survey
sites, possibly due to different varying regional climatic conditions. Only in one region,
Amhara, did all groups indicate that they preferred food all year round.
R1. When our crops are not ready to harvest—that is when we prefer food.
R2. During December and January or before the month of May, they prefer
cash, because the cash is used to cover expenses to send children to school
(non-beneficiary made this comments) [AM_E/FG-1].
R1: We prefer cash from January to March, because it helps us buy farm
inputs for the following planting season. Also, we have food from our own
harvest during these months.
R2: We prefer food transfer during planting season (from April-June),
because the food improves farm labor performance [ORO_G/FG-4].
8.7. Summary
These data on days worked, payments for public works employment, and direct support
transfers are strongly suggestive of a “two-tiered” PSNP.
In SNNPR and, to a slightly lesser extent, in Tigray and Amhara-HVFB, there are clearly
efforts being made to ensure work is available, that beneficiaries are paid for this work,
and that these payments reflect the amount of work undertaken. In 2009, significant

165
levels of transfers were made to public works beneficiaries in these regions and to
ensure that payments bear some relationship to entitlements.
In Tigray and SNNPR, there are clearly serious attempts to provide direct support
transfers, though the regions differ in how they implement this. In Tigray, there are more
beneficiaries, but with low-ish levels of transfers; in SNNPR, the converse appears to be
the case.
By contrast, this dimension of program implementation lags in Amhara and Oromiya.
Total public works payment levels are lower as are the proportion of beneficiaries
receiving their entitlement.
Self-reported payments data indicate that Full Family Targeting has not been
implemented in Amhara.
These self-reports show that direct support payments in Oromiya are considerably below
what is envisaged in the PIM.
There exists a very strong preference for food payment, given uncertainly about prices,
lack of food availability on the market, deliberate price hiking by traders, and non-
equivalent value of food basket and current cash payment. Beneficiaries were more
likely to prefer cash when grain prices were below 3.5 Birr/kg.
Nearly all those who responded to a preferred payment arrangement were monthly. But
it seems the main problem is predictability of payment. Many respondents indicated that
they would prefer a seasonally-specific payment modality: food during the rainy season
and cash during the dry season. The months seem to vary, possibly due to different
varying regional climatic conditions.

166
9. Appeals and Complaints
9.1. Introduction
This chapter presents the qualitative findings of the appeals and complaints management of
the PSNP in the highland areas of Ethiopia. It describes the reasons for appeals, who people
appeal to, how these are presented, whether the appeals process is easy and transparent,
and how long the appeals and complaint cases take to be resolved. It also includes a
discussion of the views and perception of the Woreda and Kebele Food Security Task Force
on visits by the Rapid Response Team (RRT).
9.2. Sources of Appeals and Complaints
Individuals, households, and groups have a right to complain if they feel that they are left out
of the program and/or any wrong decisions during the targeting process have been made.
Table 9.1 presents data drawn from the qualitative fieldwork on responses to questions
about the causes of appeals and complaints across regions.
Table 9.1. Sources of appeals and complaints, by region
Regions
Causes of appeals and complaints
SNNPR
Oromiya
Amhara
Tigray
Total
Exclusion
15
7
13
10
45
Inclusion
0
0
3
0
3
Partial family targeting
4
5
1
0
10
Delay of payment
0
2
1
0
3
Deduction of payment
1
2
0
0
3
Form of payment
0
0
1
0
1
Source: Authors’ calculations based on the PSNP survey data.
Fifty focus group discussions were conducted with the woreda and kebele level task forces
and four community groups comprised of those who are benefiting and not benefiting from
the PSNP. From 50 focus group discussions, exclusion emerged as the first cause for
appeals and complaints. The response pattern was similar across the regions and all the
respondents (the four community groups) strongly stressed that they felt they were excluded,
despite having met the targeting criteria of the PSNP. In general, the inclusion error as a
cause for appeal did not appear as strong as exclusion. This could be attributed to the fact
that after the first year of the PSNP, lessons were learned and the community involvement in
the targeting process stronger and, hence, this contributed to a lower number of appeal and
complaint cases related to inclusion errors. During the focus group discussions with the
Woreda Food Security Task Force, this was echoed as follows:
After 2007, there are no significant complaints. Most complaints are about
why I am not included, what can you do about it, rather than why he/she is
included while I am not [SN_S/W-FG-1].
Among people who are not targeted in the program, there is a general perception that they
were excluded from a government support package that should benefit all. People felt that
the PSNP is a free government support that all poor people are entitled to benefit from. This

167
notion was eloquently captured in one focus group discussion with the KFSTF when they
explained people’s common perception of this issue as ”the government is the father to
everyone and not an uncle for some”[ SN_S/KI-1]. This indicates there is a significant
sentiment about exclusion from the program. This understanding could be attributed to the
previously implemented emergency response and people’s misunderstanding of the
difference between the PSNP and an emergency response.
The one exception to this was conversations held with community focus groups in Amhara.
When asked whether “anyone in the community has complained or appealed about the
safety net targeting?,” the “inclusion error” during the first phase of PSNP targeting was cited
as a major cause for appeals and complaints. This was due to the fact that the Amhara
region exercised a different targeting approach during the first year of the PSNP. At that
time, Amhara targeted relatively better-off households who were the most likely to graduate
from the program. (This was subsequently changed.)
The second cause for appeal is partial family targeting. A significant number of respondents
mentioned that one or more of their household members were not included in the program.
Given the PSNP resource limitations during the first years of the PSNP, almost all regions
exercised a partial family targeting approach rather than a full family targeting (FFT).
However, it is widely noticed that, currently, most of the woredas under this study has
adopted or are planning to adopt a FFT. This was reflected in response of the Woreda Food
Security Task Forces in different woredas:
In 1999 EC (2007) we did retargeting, because there were complaints that
there were both inclusion and exclusion errors and the targeting exercise did
not take into account all family members of the selected households. It was
not full family targeting. Then we undertook a fundamental retargeting by
involving communities. At present, we have the list of beneficiaries of
1999 EC (2007) retargeting results [AM_E/W-FG-1].
On 2007 retargeting was done uniformly in all kebeles. Earlier, if you have 10
household members, only two or three will be taken; this was corrected during
retargeting. The number of beneficiaries has also increased from around
12,000 to 14,808 [SN_S/W-FG-1].
From the outset, our woreda used to practice a full family targeting (FFT) in
the previous years; later on changed toward reaching as many households as
possible. But this year, the FFT has become mandatory for all woredas and,
hence, we are implementing it in all kebeles [TIG_A/W-FG-1].
The third cause of appeals that were mentioned, particularly from Oromiya and Amhara,
related to the delay of transfer and payments. Despite some improvements in this regard,
respondents expressed their frustration with the delay of transfer of payments.
In addition, deduction of payment has also emerged as equally important to delay of
transfers in SNNPR and Oromiya regions. The focus group discussions with men, women,
and KFSTF groups’ respondents revealed this. One respondent said, ”…40 Birr are
deducted from my payment. I do not know why it is being deducted. Where can I go to
appeal? If we have grievances in other administrative issues, we talk to the kebele chairman.

168
But there is no one to talk to about the PSNP” [SN_S/FG-3/4]. The focus group discussion
with a women group in Amhara region indicated that different forms of payment (in cash
payment instead of food) are a concern and a cause for appeals.
9.3. Who People Appeal to
Ways of appeals and complaints against targeting decisions are illustrated in Table 9.2.
When excluded from the PSNP, people made their appeals and complaints to various actors
who are directly and indirectly involved with the PSNP in their locality. Out of the focus group
discussions with men and women in four regions, 10 communities reflected that people
made their appeals and complaints to the development agents (DAs), kebele cabinet, kebele
administrator, Kebele Food Security Task Force (KFSTF), Kebele Appeals Committee
(KAC), traditional/village leaders, Kebele Women’s Affairs Office, woreda administration,
Agricultural and Rural Development Office (ARDO), and Disaster Prevention Office (DPO).
Table 9.2. Who appeals to whom?
Appeals made to
Appeals made
by (in gender)
DA
Kebele
cabinet
Kebele
administrator
KFSTF
KAC
Traditional
(ganda)
leader
Woreda
administration/
ARDO/DPO
Kebele
women’s
affair
Men’s group
0
5
3
2
2
0
1
0
Women’s group
4
4
4
2
2
1
0
1
Total
4
9
7
4
2
1
1
1
Source: Authors’ calculations based on the PSNP survey data.
Notes: development agent (DA), Kebele Food Security Task Force (KFSTF), Kebele Appeals Committee (KAC), Agricultural
and Rural Development Office (ARDO), Disaster Prevention Office (DPO).
Table 9.2 disaggregates the responses in gender differences. Men appear to prefer
appealing to the kebele cabinet, kebele administrator, and KFSTF and KAC, respectively,
whereas women prefer appealing to the development agent, kebele cabinet, kebele
administrator, KFSTF, and KAC. In addition, some women indicated that they felt more
comfortable in appealing to the Kebele Women’s Affairs Office. Interestingly, none of the
women mentioned that they would go up to the woreda level to present their appeal and
complaint. The focus group discussion with women in Gursum, Oromiya region, confirmed
this:
The appeal is usually to the development agent. We do not have the
conviction that if we go to the woreda, we get justice. We do not feel that the
woreda officials listen to us either. We believe that they listen to the peasant
association and kebele leader rather than us. Moreover, we do not even know
which office we should go to [ORO_G/W-FG-3].
9.4. How Appeals Are Presented and Resolved
Communities present their appeals and complaints against the decision of the PSNP
targeting to the Kebele Cabinet, administrator, development agent, KFSTF, KAC, Kebele
Women’s Affairs Office, and village leaders. Focus group discussions with all categories of
beneficiaries across all woredas indicated that 95 percent of the appeals and complaints are

169
undertaken verbally. In rare occasions, mostly for higher administrative levels, appellants are
asked to present their appeal in writing. One respondent from the direct support group
pointed out the following about appealing in writing:
They excluded me from the program after two years. I requested to the then
kebele chairman who was also the chairman of KFSTF verbally, but his
response was negative. Then I presented my complaint to the woreda
agriculture office in writing and they have responded. I got a timely response
from them. I was not aware of why I was excluded [SN_D/FG-1].
The Kebele Appeals Committee (KAC) is responsible to listen to the appeal and complaint
cases and provide timely responses. Table 9.3 illustrates the composition of KAC in the
study areas.
Table 9.3. Composition of the Kebele Appeals Committee
Composition of Kebele Appeals Committee
Region
Kebele
Kebele chairman
Kebele manager
Two elders/
religious leaders
Women’s
representative
School teacher
Health extension
worker
Kebele spokes-
person
Kebele militia
commander
Development
agent
Food security
secretary
Kebele secretary
Number of
members
SNNPR
Gurade


5
Furra



3
Soyame
0
Oromiya
Barite
0
Dimtu Raretti





6
Amhara
Serawudi



3
Shengo Defar






5
Wal

3
Tigray
Adizata



5
Sendeda



7
Source: Authors’ calculations based on the PSNP survey data.
KACs had been established in 8 out of 10 kebeles covered in this study. In principle, the
overall appeals and complaints process should be managed by an independent team of
KACs that is different from the one involved in the targeting process. The composition of
KAC members varies from religious leaders, women, and school teachers, to health
extension workers, kebele administrators, and kebele managers. In Amhara region, Sayint
woreda, however, the composition is slightly different, consisting of the Kebele chairman, the
Kebele manager, a teacher, the Kebele cabinet’s spokesperson, and the Kebele
militia/security commander.

170
In Tigray region, however, the KAC members are independent and impartial and were
selected from the community members. Further, they have a separate food security
secretary at the kebele levels. When asked, “Who are the members of the KAC?,” the
response was
Those people who are impartial, honest, and with better level of know-how
than the community and elected by each sub-kebele and they should not be
members of the kebele administration [TIG_S/K-FG-1].
Prior to the introduction of the KAC, the appeals and complaints procedures were managed
by the Kebele Cabinet, traditional leaders, the Woreda ARDO/Disaster Prevention and
Preparedness Office (DPPO) or council. The majority of respondents pointed out that the
appeal process is transparent and easy to use. The focus group discussion with a men’s
group in Tigray region, Ahferom woreda, strongly supported this view:
Yes, the process of appeals and complaints is transparent and easy for those
who have cases. You can present your case at any time in any form that
could be written or verbally. If you need to present your appeal in written form,
they can provide you a format to be completed there. They invite the person
who complains and the other party and they listen to each party’s idea
attentively and collect the necessary evidence from the concerned sources
and finally decide based on the data. If one of the parties does not agree on
the decision, the case will be sent back to the public for approval [TIG_A/FG-4].
However, there is a stark regional variation on the appeals and complaints case
management. The discussion resulting from men’s groups in Oromiya and SNNPR indicated
an opposite result about the transparency and easiness of the appeal process. Respondents
expressed dissatisfaction with the appeal and complaint process and with the composition of
the members of the KAC in general.
The appeal process is not transparent. People working on the appeal case
management are those who have created the problem during the targeting
process. The same people involved in targeting are also involved in the
appeal case management. That is why the process is not transparent
[ORO/SN_G/S-FG-4].
The appeal process is difficult. This is because the committees are running to
fulfill their private daily activities and do not give time to this. It would be
difficult to say to them, “sit and listen to our complaints” [OR_D/K-FG-3].
There is an appeal committee in our kebele. But they usually endorse the
proposals of the CFSTF and, hence, return complainers back [TIG_A/K-FG-1].
The gender dimension of the appeals and complaints process at the kebele level is
investigated by looking at the following aspects: the composition of the KAC and the gender
of the appellants. Table 9.4 illustrates the gender composition of the KAC.

171
Table 9.4. The composition of the Kebele Appeals Committee, by gender in study
areas
Region
Kebele
Number of the KAC
members
Number of men in the
KAC
Number of women in
the KAC
SNNPR
Gurade
5
4
1
Furra
3
2
1
Soyame
0
0
0
Oromiya
Barite
0
0
0
Dimtu Raretti
6
5
1
Amhara
Serawudi
3
2
1
Shengo Defar
5
4
1
Wal
3
2
1
Tigray
Adizata
5
3
2
Sendeda
7
6
1
Total
37
28
9
Source: Authors’ calculations based on the PSNP survey data.
The above result indicated that while women are not equally represented on the KAC as
men, they are evenly represented (at least one woman per kebele) across the study areas.
No stark difference is observed in the number of women members in the KAC across the
regions except Adizata kebele in Tigray, where two out of five members are women. Further
attempt was made to capture the gender dimension of the appeals process by looking at the
gender of appellants who have made appeals against the decision of the PSNP targeting
and other program aspects. The focus group discussion with all KFSTFs across the regions
could not yield a good sense about the gender of the appellants. The KFSTFs were not able
to provide the research team with the records of the appeal cases during the study.
However, the focus group discussion with the women’s group in all 10 kebeles
overwhelmingly indicated that they know women in their community who have complained or
appealed about the safety net targeting.
An attempt was made to understand why there were so few appeal cases in the survey,
despite such a large number of the PSNP beneficiaries nationwide. Acquiring clear and
consistent information on this proved very difficult. We propose that the low incidence of
appeals could be attributed to (1) the historical lack of success in appeal cases; (2) the
passive role the KAC members, given that the role is likely to be underfunded due to all the
other PNSP responsibilities; (3) the non-independence of the KAC; or/and (4) lack of
confidence in the KAC by the community. Further, it was observed that the appeal cases are
presented not only to the KAC, but also to various administrative officials, village leaders,
and the development agents at the kebele and woreda levels. Often these appeals are made
verbally, hence it is difficult to trace how many cases were presented and resolved.
The focus group discussion with the women’s group in Gurade kebele of SNNPR also
complained that the appeal and complaint process was not transparent. “The focus group
discussion results with communities indicated that the type of responses from the KAC is
known in advance, often similar, and negative.” The focus group discussion with a women’s
group in Tigray region revealed their frustrations as follows: “…when we go to KAC with
complaints, they tell us that it is because of the shortage in beneficiary quota allocated to the
kebele and sub-kebele. Otherwise, you could have been included among the beneficiaries”

172
[TIG_ A/FG-3]. Discussion results with a men’s group in Oromiya region also indicated that
despite persistent appeal and complaints, the response was generally negative. “I have five
family members. Only three of my family members were targeted. I appealed to the Kebele
Targeting Committee to consider the remaining two family members. They said no to my
appeal” [ORO-Z/FG-4].
It proved difficult to get a good sense as to how long it took for appeals to be heard. The
response to the question, “How long did it take from complaint to resolution?,” tended to vary
from “two weeks to two months.” But it is important to note that because most appeals
appear to have been handled verbally, there is relatively little formal documentation of these
appeals.
It is also worth noting the perception that there were limited gains to be had by appealing as
the following quotations reveal:
There are many people who appealed. They appeal to the kebele
administration. There is nothing that they got after they have appealed. They
are only moaning. The response they receive is that the included households
are not any better off than they themselves, and that there is no regulation
that allows to exclude them in order to include the appellant [SN_D/K-FG-3].
Two of my family members were not registered for PSNP. I have brought this
case to the Kebele Targeting Committee. But the kebele administrator
refused my case [ORO_Z/K-FG-4].
They usually get a response immediately because the answer is a quota
[SN_D/FG-4].
There are also appeals related to targeting. This could be such that I am not
better off than some beneficiaries, but you excluded me. You favor based on
kinship or money. We complain to the chairman. The outcome is that nothing
happens. They may say you will be included during the next targeting
[SN_S/K-FG-3].
The following comment from a focus group with direct support beneficiaries in Tigray region
also suggests a lack of confidence in the KAC in resolving the appeal cases.
It is not hard to complain, but as far as we know these people are not
important in solving the problem, so we prefer keeping quiet rather than
shouting endlessly [TIG_S/FG-1].
Responses to previous appeals may have also discouraged use of the appeals
mechanism.
28
Focus group discussions with the KFSTF in Amhara and Tigray regions
highlighted how the targeting errors and grievances were resolved:
The appeal committee was following an approach to ask the appellant to
identify someone who is better off but included in the safety net. If he/she
identified a household who was incorrectly included, then the complaint or
28
In fact, the quantitative household survey showed that only 185 households (out of more than 4,000) had made an appeal
based on their exclusion from the PSNP.

173
appeal is supported. If he/she was not able to identity someone who is in the
program incorrectly, his/her complaint will not be pursued [TIG-A/K-FG-1].
Such an approach clearly discourages an appellant to present his/her case in front of the
appeals committee. Even if they could identify the better-off who is included in the program,
this would create tension and possible confrontation within the community.
Although it lacks the depth of the information gleaned from the qualitative studies, there is
some information on appeals in the household quantitative survey. Specifically, respondents
were asked three questions: Did you ever feel the selection made [of beneficiaries] was
unfair? If you thought the process was unfair, did you or anyone in your household ever
lodge a complaint/appeal/grievance about the selection? And if you did lodge a complaint,
did someone respond to your complaint? Table 9.5a gives the numbers of responses to
these questions, while Table 9.5b expresses these as percentages.
Table 9.5a. Number of households perceiving selection process was unfair, lodging
complaints, and receiving a response, by sex of household head and region
Region
Sex of
household head
Sample size
Perceive that
selection was
unfair
Lodged a
complaint
Perceived selection
was unfair and
lodged a complaint
Received a
response
Tigray
Male-headed
584
165
119
76
38
Female-headed
252
54
27
19
6
All
836
219
146
95
44
Amhara
Male-headed
620
191
107
89
44
Female-headed
226
66
27
22
7
All
846
257
134
111
51
Amhara-HVFB
Male-headed
886
285
166
102
43
Female-headed
261
67
18
10
2
All
1,147
352
184
112
45
Oromiya
Male-headed
640
225
74
63
15
Female-headed
222
67
20
18
3
All
862
292
94
81
18
SNNPR
Male-headed
714
223
97
78
24
Female-headed
203
56
18
13
4
All
917
279
115
91
28
Source: Authors’ calculations based on the PSNP survey data.

174
Table 9.5b. Percentage of households perceiving selection process was unfair,
lodging complaints, and receiving a response, by sex of household head and region
Percent of sample that:
Region
Headship
status
Perceived that
selection was
unfair
Lodged a
complaint
Perceived
selection was
unfair and
lodged a
complaint
Perceived
selection was
unfair, lodged
a complaint,
and received a
response
Received a
response given
that they
perceived
selection was
unfair and had
lodged a
complaint
Tigray
Male-headed
28.3
20.4
13.0
6.5
50.0
Female-headed
21.4
10.7
7.5
2.4
31.6
All
26.2
17.5
11.4
5.3
46.3
Amhara
Male-headed
30.8
17.3
14.4
7.1
49.4
Female-headed
29.2
11.9
9.7
3.1
31.8
All
30.4
15.8
13.1
6.0
45.9
Amhara-HVFB
Male-headed
32.2
18.7
11.5
4.9
42.2
Female-headed
25.7
6.9
3.8
0.8
20.0
All
30.7
16.0
9.8
3.9
40.2
Oromiya
Male-headed
35.2
11.6
9.8
2.3
23.8
Female-headed
30.2
9.0
8.1
1.4
16.7
All
33.9
10.9
9.4
2.1
22.2
SNNPR
Male-headed
31.2
13.6
10.9
3.4
30.8
Female-headed
27.6
8.9
6.4
2.0
30.8
All
30.4
12.5
9.9
3.1
30.8
Source: Authors’ calculations based on the PSNP survey data.
Some care is needed in interpreting Table 9.5. A household could perceive that the selection
process was unfair either because they were excluded, or because they felt another
household should have been included, or because they felt that another household should
have been excluded.
29
Households could lodge a complaint even though they thought the
selection process was fair.
30
Mindful of this, even when households perceive that the
selection process was unfair, they are relatively unlikely to lodge a complaint. There is little
regional variation. Everywhere, male-headed households are more likely to submit a
complaint than female-headed households. Between 22 percent (Oromiya) and 46 percent
(Tigray) of households that lodge a complaint when they perceive the selection process has
been unfair report that they received a response; the questionnaire did not include a
question as to whether their complaint was successfully resolved. These results are broadly
consistent with results from the qualitative work.
9.5. Woreda Food Security Task Force and Kebele Food Security Task Force Views
on the Visit of the Rapid Response Team
Rapid Response Teams (RRTs) are supposed to undertake field visits, identify
implementation problems and bottlenecks, report these to relevant policy and administrative
bodies, provide guidance and technical support, and track previously identified problems
29
The household questionnaire did not distinguish between these possibilities. This could be added to the 2012 survey.
30
In the 2012 survey, asking the nature of the complaint would be useful as would knowing the outcome of the complaint.

175
(GFDRE 2010; Sharp, Brown, and Teshome 2006, 43). As per this mandate, the RRT
should visit most PSNP woredas on regular basis. The Oromiya regional FSTF indicated
reasons why it did not do so:
The regional rapid response team is expected to have a monthly field visit
schedule. The intension is to cover all PSNP woredas every year. However,
we were unable to realize this, mainly due to logistic constraints. We have
limited vehicles to conduct regular visits. Also, we have a series of other
duties which prevented us from conducting regular visits. The number of visits
we conducted so far ranges between 8–10 times in a year [ORO_R-KI-1].
The focus group discussion results show that the majority the WFSF across all regions found
that the RRT team visit was helpful in providing timely support and corrective measures.
However, it seems there is a significant disconnect between the Woreda Food Security Task
Forces and Kebele Food Security Task Forces. The focus group discussions with the KFSTF
indicated that the RRTs from the federal, regional, and woreda levels paid infrequent and
irregular visits to all kebeles over the past years, spending few days on the ground.
However, a focus group discussion result suggested that the overall support received from
the RRT is not well received among the KFSTF across all the regions. Compared with the
WFSTF, overall, there is limited knowledge and comprehension among the KFSTF on the
role and objectives of the RRT visit and the scope of responsibilities attached to their visit on
the kebele level. It seems the KFSTF regarded the RRT team visit as a panacea for all
PSNP-related problems in their kebele.
They did not look for us. They did not talk with us [SN_S/K-FG-1].
The visit of the RRT was of no use to the KFSTF [SN_D/K-FG-1].
In some regions (Tigray and Amhara), the role and responsibilities of the RRT are seen as
trouble shooting and on-site technical backstopping rather than correcting issues that are
affecting the overall PSNP implementation. The focus group discussions in Seiseamba
Tigray indicated that, “They are not helpful in settling administrative related issues”
[TIG_S/K-FG-1]. “This team assesses and technically backstops the undergoing PSNP
activities, especially public works” [TIG_S/K-FG-1]. The focus group discussions with the
WFSTF in Shebedino woreda indicated that the visit of the RRT helped them to resolve
misunderstandings related to the graduation process.
It is very helpful. Sometimes when the kebele didn’t believe what we say, we
would call the RRT. Especially around graduation, since there is a resistance,
they will come to persuade them. They play a great role in clarifying that it is
not something we created [SN_S/FG-W-FG-1].

176
9.6. Summary
This chapter has summarized information collected on the appeals process. There are
several salient findings.
Exclusion from the PSNP is the principal cause of appeals. Partial targeting, inclusion
errors, deduction of payment, and delay of transfers are other reasons for appeals and
complaints.
Appeals are made to the KFSTF, kebele cabinet, the development agents, Kebele
Appeal Committees (KAC), village leaders, and others. Although KACs are widespread,
they are not perceived as being effective.
Women are fairly represented (at least one woman per kebele) in the KAC.
It appears that, relative to concerns raised about the selection process, relatively few
households appeal. There is a sense that the chances of appealing successfully are
limited and this may discourage use of appeal mechanisms.
Appeals are largely made verbally. It appears that decisions are made about these
somewhere between two weeks and two months after the complaint has been lodged.
The lack of written documentation surrounding appeals means that this figure represents
a “best guess.”
Views on the usefulness of RRT support differ. Several Woreda Food Security Task
Forces believed that RRT visits were helpful in assisting with bottlenecks associated with
PSNP implementation and in providing back-up support. However, Kebele Food Security
Task Forces believed that the RRT visits are not as helpful as they should be.

177
10. The Other Food Security Program and its Transition to the
Household Asset Building Programme
10.1. Introduction
This chapter presents an overview of the Other Food Security Program (OFSP), challenges
faced in implementation, women’s access to credit/household packages, and the
subsequent transition of the OFSP to the Household Asset Building Programme (HABP) in
the revised National Food Security Program of the Government of Ethiopia. It further
explains the current rollout progress of HABP including the provision of credit, challenges
encountered thus far, its linkage with other food security programs, and its role in facilitating
the graduation process, as well as the role of development agents in implementing the
HABP. It draws on a variety of data sources, including key informant interviews (KII) at the
regional, woreda, and kebele level, focus group discussions in woredas and kebeles, and
also the quantitative surveys fielded at the kebele and household level. Table 10.1 gives an
overview of the evaluation objectives covered in this chapter.
Table 10.1. Evaluation objectives covered in chapter 10
Evaluation
objective
Issue
Links to Log Frames and TOR
Document progress in the implementation of the HABP
Can gender dimensions of access be better
captured?
HABP Log frame Output 1.2
TOR, para 42
HAB clients have access to financial services
HABP Log frame Output 2.1
HAB clients have access to desired input
HABP Log frame Output 3.2
Source: Authors’ compilation.
10.2. The Other Food Security Program
In the initial phase of the Food Security Program (FSP), the PSNP was complemented by
the “Other Food Security Program” (OFSP). The OFSP encompassed a suite of activities
designed to support agricultural production and food security, and to facilitate asset
accumulation. This included access to credit; assistance in obtaining livestock, small
livestock, or bees, tools, seeds; and assistance with irrigation or water-harvesting schemes,
soil conservation, and improvements in pasture land. In some cases, beneficiaries were
provided with subsidized credit to purchase “packages,” combinations of agricultural inputs
sometimes based on a business plan developed with support from the extension service. In
the first evaluation of the FSP, Gilligan et al. (2007) noted that outside Tigray, access to the
OFSP was low. While this improved between 2006 and 2008, access to the OFSP remained
limited (Gilligan et al. 2009) and few households had consistent access to OFSP resources.
This limited coverage reflected a number of other challenges associated with the
implementation of the OFSP. One challenge was that the agricultural extension system was
under resourced and there were too few development agents with sufficient skills to play
their role effectively (World Bank 2010).
Focus group discussions and key informant interviews revealed considerable regional
variations in the targeting of the OFSP, partly because, at least as perceived by

178
respondents, there were no clear guidelines on OFSP implementation, particularly who
should be targeted. Some regions, concerned about the number of prospective graduate
rates they would achieve by the end of the program, targeted the better-off households who
were most likely to graduate. For instance, during the first phase of the PSNP, the Amhara
region gave a priority to households who could easily achieve the graduation target, rather
than the ultra-poor. Similar situations were noticed in Tigray region as well.
Lack of clear guideline and procedure for targeting has hampered the
implementation of OFSP [ORO-Z/W-KII-OFSP].
OFSP had been implemented in an uncoordinated way by NGOs and
government [Am/ R-FG-2].
There was an absence of coordinating body for household packages in the
woreda. Further, government provided household packages are not
coordinated with NGO efforts in the area. Different actors intervening in the
household asset building activities rarely work in coordinated manner to
maximize impact [ORO-Z/W-KII-OFSP].
We did not have information on who has to benefit from the federal resources
(OFSP) in the first three years of the PSNP period. We were not informed on
the targeting process too. As a result, we provided the federal resources to
PSNP and non-PSNP beneficiaries. This has negatively impacted our
intention for graduation by reducing the number of PSNP households
benefited from OFSP [ORO-G/W-FGD-1].
At the initial period, there were trade-offs, even tensions between pro-poor
and pro-progress/growth opinions. Among the challenges was the question,
“Is it development or addressing the households’ food gap, which should
come first?” [TIG_R/FG-1].
In Oromiya, targeting was based on instructions received from the Woreda Food Security
Office and the DPPO, whereas in SNNPR, households were targeted in a lottery system. In
addition, household’s motivation and determination is seen as a positive step toward the
selection criteria for the OFSP. This was well illustrated during the key informant interviews
with the development agents when asked, “How were the OSFP clients selected?”
The OFSP is based on the assumption of a daily income of 1 US dollar.
Based on this assumption, any member of the community who is willing and
able to take credit and is willing to implement the credit for the intended
purposes [as in the business plan] can be targeted to receive the
OFSP/household package assistance [TIG_S/W-KI-2].
At first, when they collected the socioeconomic data, each PSNP household
is surveyed and those who have more assets are selected and a business
plan is prepared [AM_S/K-KI-1].
Provision of credit was also problematic. In 2008, only 72 percent of loans that had fallen
due had been repaid. A slow start in establishing a revolving funds system meant that only
36 percent of loans that had been repaid were refinanced (World Bank 2010). In fact, there

179
was confusion as to whether beneficiaries were receiving a grant, a loan, or participating in a
revolving funds scheme. This created confusion for both implementers and beneficiaries.
Many recipients considered it a free grant, took it, and later were informed to repay. The
focus group discussions with a men’s group and the OFSP team illustrated these aspects
mentioned above.
For example, a lady who did not even have a pot and cannot earn money also
took the loan thinking that it was for free [AM_SO/K-FG-4].
Initially the development agents told us that it was free, and then when we
were ready to collect the money after finalizing the loan procedures, they told
us to collect the money from the cooperatives. The cooperatives told us that it
is a loan, and to be repaid with interest [AM_SO/W-FG-4].
It is expected that the OFSP resources operate on revolving fund basis. But
no one knew whether it was a grant or a revolving fund. We are asked now to
re-collect it and make it seed money for implementing the HABP. The
challenge is that the extended funds were impossible to trace [ORO_Z/W-KI-2].
In addition to the Ethiopian government’s OFSP support, various other actors are involved in
funding and implementing household package programs in different regions. For instance, in
Oromiya region, Gursum woreda, and SNNPR, in 13 woredas, the World Bank’s Food
Security Project (also supported by CIDA and the Italian government) provided similar
support as the federal OFSP through a revolving funding mechanism. In Tigray and Amhara
region, USAID funded various nongovernmental organizations (NGOs) to implement similar
household package programs. The involvement of various actors in the implementation of
the OFSP created considerable confusion at the beginning of the program, especially
concerning the intended target population. This was well illustrated during the focus group
discussion with the Woreda Food Security Task Force in Gursum woreda, Oromiya region:
At the beginning, there was confusion between the World Bank (WB) credit
and the federal credit schemes. Unlike the WB credit, the federal fund does
not have clear guidelines as to how to go about it, its purpose and focus. We
did not have information on who has to benefit from the federal resources in
the first three years of the PSNP period. We were not given any information
on targeting process for the federal government credit. As a result, we
provided the federal resource to PSNP and non-PSNP beneficiaries. This has
negatively impacted on the achievement of the graduation objectives [ORO-
G/W-KI-1].
Finally, there were persistent concerns about the appropriateness of the OFSP packages
and the lack of coordination across stakeholders.
The other problem is that the menus did not fully address the interest of the
beneficiaries; the menu is narrow and the community asks for activities
outside it, and we influenced the beneficiaries to work within the given menu
[AM_SO/W-FG-2].

180
10.3. Women’s Access to Credit and Household Packages
The focus group discussions with women groups in all ten kebeles in the study areas
overwhelmingly indicated that women can, in theory, access credit. However, many women
indicated that the nature of the credit/OFSP targeting and implementation made it difficult for
them to access. As indicated above, some regions gave priority of credit/packages to those
who have better collateral or who have relatively better assets in their livelihood. In addition,
the high interest rate was a significant barrier for women not being able to take the credit.
The following has been captured during focus group discussions with the women’s group
across the regions when asked, “Can women get credit?”
We can take credit from microfinances (MFIs). However, because the interest
rate is too high, we do not dare to take loans from MFI. The interest is around
30 percent [SN-D/FG-3].
Yes, women can get credit. But we are not sure, since many of us did not ask
for credit [SN-S/FG-3].
Yes, women can get credit. We go to the development agent declaring our
interest to get credit. The development agent asks us what assets we have in
the house. We tell him what assets we have, such as sheep, goats, house,
etc... [AM-SA/FG-3].
Women can get loan. For example, I got registered for the loan and they
asked me what resources I have, like trees, vegetables, etc… They might
think that I may not be able to repay the loan, so they did not give me the
loan. But there are some women who have got this loan [AM-SO/FG-3].
The packages include poultry, goat and sheep reproduction, water pump
motor, modern bee hive, fattening, credit services. For instance, the financial
institutions purchase and give us dairy cows in kind, not the cash. But when
they give us the loans, they strongly advise us not to use the money for
consumption [TIG-S/FG-3].
Beneficiaries across all groups in all the regions indicated that the amount of credit/package
they received through the OFSP was generally insufficient to support the purpose for what
they took the package. Despite the significant increase of the size of the credit/package
since the early years of the PSNP, they expressed their dissatisfaction regarding the
amount. From focus group discussions held with nine groups of transitioning households that
received household packages over the last five years, seven said that the amount of
credit/loan or package received through OFSP was insufficient to implement the overall
business plan intended.

181
Further, the beneficiaries provided reasons why the amount of credit/package was not
sufficient to implement the business ideas they had:
Not sufficient. The amount was only 1,500 Birr. Most of us planned for animal
fattening. Fattening is a lucrative business in our area. However, the credit
given could not help us to buy a well-grown bull. It only buys small bulls,
which required at least three years feeding. We were forced to buy small bulls
and sold them before they had fully grown up to repay the debt. We sold them
early because we were required to repay the debt within two years according
to the agreement. The limited credit volume coupled with the short repayment
period has reduced the impact of fattening, which has very high potential in
the woreda [ORO-G/FG-F-2].
The amount of credit we received was not enough. The 4,000 Birr do not
exceed an ox. It would have been good if it has been raised to 8,000 Birr
[SN-T/FG-2].
It is not sufficient. Nowadays, goods are very expensive; I am involved in
trade activities and have a small shop. The credit I received could not even
buy 10 packs of beer. I am working for others and for myself; it is still not
sufficient to take a better share of the market [AM-S/FG-2].
10.4. The Household Assets Building Programme
In light of the problems mentioned above, the Ethiopian government, in collaboration with
donors and development partners, extensively redesigned the OFSP, naming the new
program as the Household Assets Building Programme (HABP). The HABP is one of the
four components of the Ethiopian government’s National Food Security Program. As such, it
contributes to the achievement of the FSP’s expected outcome of an improved food security
status of male and female members of food-insecure households in chronically food-
insecure (CFI) woredas. The specific targeted outcome of the HABP is diversified income
sources and increased productive assets for food-insecure households in CFI woredas
(GFDRE 2009b).
The HABP differs from the OFSP in three ways. Along with the injection of new resources,
there is an emphasis on increased contact and coordination with the extension services as
well as other actors, such as the Small and Medium Enterprise Development Agency,
programs for women and youth, and off-farm technical officers. Each kebele should have
three development agents, one crop science development agent, one animal husbandry
development agent, and one natural resources management development agent. They are
supposed to disseminate “technology packages” and provide on-farm technical advice.
These are demand-led with clients involved in the identification of new opportunities as well
as the development of tailored business plans that can, where appropriate, include off-farm
activities. Second, credit services have been de-linked from the extension service. Instead,
credit will be provided through microfinance institutions (MFIs) and Rural Savings and Credit
Cooperatives (RUSACCO) (GFDRE 2009b).
A third significant change has been the clarification of access to the HABP. The government
of Ethiopia (GFDRE 2009b) states that, “The clients of the Household Asset Building

182
component are food-insecure households in chronically food-insecure woredas” (GFDRE
2009b, 9). Initially, priority is to be given to expand the coverage of the HABP component as
rapidly as possible to ensure graduation at scale. For this reason, PSNP clients are to be
prioritized for support under HABP. There is increased emphasis on meeting the specific
needs of women and female-headed households and an attempt to reach rural youth.
Finally, direct support clients were not to be excluded from the HABP (GFDRE 2009b).
Key informant interviews with the main food security actors at the regional and woreda levels
indicated that to facilitate the rollout and implementation of the HABP, new joint technical
committees were planned or had been established. In Tigray and Amhara, the committees
are already established. According to the HABP policy document,
The Regional Agricultural Extension Directorate is in charge of the HABP,
accountable to BOARD Head, and responsible for leading the coordination
with other agencies through the HABP Joint Technical Committee (TC). In
this coordination role, it is responsible for: (i) developing and consolidating
annual implementation plans and budgets for the HABP in the region in line
with proposals from the woredas and line bureaus; (ii) ensuring close
collaboration with and coordination of implementing agencies both within and
outside the Agriculture and Rural Development (ARD) sector; (iii) ensuring
that NGO plans with household asset building activities contribute to the
overall objective of the FSP; (iv) ensuring appropriate integration of the HABP
plans and activities with plans and activities of other FSP programs in the
region (including, notably, ensuring synergy between the HABP household
level activity and the development of enabling community infrastructure
through the PSNP public works and complementary community investment
(CCI), and with other development interventions (GFDRE 2009b).
The Ethiopian government’s policy document describing its food security program also
clearly specified that complementary community investment is an important complementary
food security component.
The CCI component aims to facilitate an adequate level of infrastructure by
populations in chronically food-insecure woredas in order that other
development interventions, which might focus on households, can achieve
their potential impacts. Therefore, CCI is not only complementary to the
HABP component, it is inseparable from the household investment
interventions. Activities should have a direct relevance to food security
(GFDRE 2009b, 40).
10.5. Rollout of the HABP
The HABP was supposed to be rolled out in the highland regions from January 2010
onwards. Key informant interviews with the relevant technical committees and administrative
bodies at the regional level indicated a positive response about the current state of the
HABP rollout process. Overall, across all the regions covered by this study, it is found that
the HABP launching awareness workshops have been undertaken and technical committees
established. The overall ownership, technical aspects, and the facilitation role of HABP have

183
been given to the regional Agricultural Extension Unit, while the CCI is under the regional
Natural Resources Unit, as stated in the policy document. The key informant interviews also
indicated that at the regional level, overall, there is clear comprehension about the objectives
of the program and the institutional arrangement. An excerpt from the regional food security
team revealed the following about the objectives of HABP and institutional arrangement:
As of 2010, the HABP is under the agricultural extension program…. What
differentiates HABP from the OFSP is that it is going to be implemented in an
improved way and with a new approach. But the thought is similar. Some
work has been done to organize cooperatives in each kebele as initial work
for the HABP, since there is difference in the credit system. Previously credit
was disbursed through the governmental structure, but now it is going to be
disbursed by microfinance institutions. Then, the agricultural extension
department will provide technical support, especially in business plan
development and providing different package menus that could be
appropriate for a specific area [SN_R/FG-2].
HABP has more potential to help farmers exert their efforts and benefit from
the different alternatives in the package, while OFSP had limited opportunities
and alternatives [TIG_R/KI-2].
However, considerable regional variations were observed in the HABP rollout progress. As
of August 2010, HABP regional committees had been established to lead the program rollout
and implementation process in all regions. In Amhara, Oromiya, and Tigray, HABP
orientation activities were taking place at woreda and kebele levels. However, in SNNPR,
HABP orientations were limited to regional level offices. Key informants in SNNPR explained
that the delays had occurred as a result of delays in receiving program implementation
manuals (PIM) and Training of Trainers (TOT). All regions were finalizing the establishment
of technical committees for HABP at the woreda level and providing orientation and training
to these (see Table 10.2). However, in these key informant interviews, it was reported that
credit disbursements had not yet begun.
Table 10.2. Rollout status of HABP, by region and woreda
Region
Woreda
HABP
Committee
established
Training
provided on
HABP
HABP
manual
distributed
Training
cascaded for
DAs/kebeles
HABP credit
provided for
households
Amhara
Ebinate
√
√
√
√
X
Sayinte
√
√
X
X
X
Sekota
√
√
√
√
X
Oromiya
Gursum
√
√
X
X
X
Zeway D
√
X
X
X
X
SNNP
Demba G
X
X
X
X
X
Shebedino
X
X
X
X
X
Tembaro
X
X
X
X
X
Tigray
Aferom
√
√
X
X
X
Saesi Tsedamba
√
√
X
√
X
Source: Interviews with regional and woreda level officials.
Note: DA = development agent.

184
For instance, the Amhara region has progressed well ahead of the three remaining regions
in the rollout of the program and cascading to the lower levels. The technical committee of
the HABP has already conducted three meetings since the establishment, the sensitization
trainings are already conducted, and a draft Amharic version of the HABP manual has been
prepared and distributed in the region. By contrast, in the Oromiya, SNNP, and Tigray
regions, the progress is relatively slow. The following has been captured during key
informant interviews with the regional Food Security Task Force (FSTF) when asked, “As of
now, how much progress has been made in rolling out HABP in this region?”
Not much. Even, the HABP committee is not functioning as anticipated
[ORO_R-KI-1].
We do not have the HABP PIM, and the training of trainers (TOT) has not
been provided yet [SN_R-KI-1].
Last December (2009), there was a big consultation workshop held at which
all concerned parties were trained and a regional level technical steering
committee has been established, composed of the various offices of
extension (The Chair), cooperative, food security, youth and sports affairs,
micro and small enterprises, and women’s affairs. We believe that these
parties have been well aware of the HABP, but actual implementation phase
will be in 2003 EC (2010/2011). After that, the technical committee will be
established and trainings [at different levels] are planned to be rolled out in
early August (2010) [TIG_R/FG-2].
The overall progress in general and the claim made about the preparation of a draft HABP
manual in the Amharic language by the Amhara region food security office is somehow
surprising as the preparation of the HABP PIM at the federal level had not been completed
and disseminated to the regions and lower administrative levels as of August 2010. Asked,
“When HABP will be rolled out?,” two key informants (a rural development advisor for the
donor country and the donor group coordinator) at the federal level indicated the following
about the progress of the HABP PIM:
The HABP has been rolled out in most of the cropping areas (Amhara, Tigray,
Oromiya, and SNNPR) since the beginning of 2010. Significant preparatory
works have already been done and we are in the process of developing the
program implementation manual (PIM) [FED_AD/KI-1].
The key informant interviews from the WFSTF indicated that the overall rollout process has
not been uniformly cascaded to the woredas and kebeles. The Amhara region also
progressed ahead in this aspect in comparison to the other three regions—woreda level
HABP committees had been established and relevant training is provided for all woredas in
the qualitative study by regional experts. It was indicated that the training was cascaded to
the development agents at the kebele level. However, there is a dissonance between the
WFSTF claim over the provision of HABP training to the development agents at the kebele
level and the development agents’ indications about the HABP in general and training in
particular. The development agent in Sayint woreda revealed the following:

185
I have no information; I heard from you about it for the first time
[AM_SA/K-KI-1].
The focus group discussion with WFSTF in Sayint woreda also indicated a mixed result
about the overall status of HABP.
It is [HABP] rolled out since March 2010. Expert is employed for the program.
However, no guideline, no training provided so far on the program
[AM_SA/W-FG-1].
In SNNPR, Oromiya, and Tigray, the rollout of the HABP to the woreda level is slower than
in Amhara. In SNNPR, in Demba Gofa and Tembaro woredas, there was no information
about the HABP and no training has been provided to the WFSTF. In Shebedino woreda, the
HABP awareness training is only provided to the experts from the woreda finance office but
not for the Woreda Agricultural and Rural Development Office (WARDO). In Oromiya region,
in Gursum woreda, training was provided for the WFSTF, while no information and training
was provided in Ziway Dugda. Similar patterns were noticed in Tigray region. In Afheworm
woreda, training was given to a staff member, but the information was not shared among the
rest of the WFSTF, whereas in Saesi Tsaeda Emba, some training was provided, but there
remains a lack of adequate knowledge and understanding about the program.
When asked, “What are the main challenges in rolling out HABP?,” key informant interviews
with the regional food security task force and OFSP/HABP committee in all regions indicated
that the absence of the HABP PIM, the formalized training manual, and the limited capacity
and awareness gap about the program were the main challenges they faced in the HABP
rollout process. In addition, respondents indicated that limited financial capital and limited
capacity of RUSSACOs to handle the fund management are other main challenges the
rollout of the program has faced. Last but not least, limited numbers of RUSSACOs,
absence of concise instructions and clarity from the federal level, and lack of adequate
training of trainers are other obstacles encountered by the regions during the rollout process.
Key informant interviews with the rural development advisor for the donor country at the
federal level also revealed that the limited number of RUSSACOs at the country level is a
main challenge with significant impact on the overall implementation of the HABP. The
excerpt from the interview is as follows:
The plan for the HABP covers eight regions (countrywide). So we need to
establish 5,000 RUSACCOs. Some RUSACCOs were created under an IFAD
program. Establishing a RUSACCO is not simple. It requires mobilizing the
community, providing training, then implementation support. Under HABP, the
intention is to establish RUSACCOs across the country within one or two
years [FED_A/KI-D].
10.6. Knowledge about HABP and its Linkages with the CCI, OFSP, and the PSNP
Overall, among key stakeholders at the regional level, there is significant knowledge and
understanding about the objectives and the purpose of the HABP and its linkages with other
ongoing food security programs. This was illustrated during the key informant interviews,
when asked to explain the differences between OFSP, HABP, and PSNP.

186
Earlier, credit was dispersed and collected by the government agriculture
extension unit. This has created a problem in collecting the credit. However,
organizing RUSACCOs in each kebele where this program will be
implemented is now mandatory, as they will disperse and collect the credit.
The government’s role will then be to facilitate and give technical support. The
credit will flow through RUSACCOs. The ‘Cooperatives renaissance
movement’ document has been developed. There were discussions and
plans to establish at least one RUSACCO in areas where there is no
RUSACCO yet, and to strengthen RUSACCOs where they are already in
place and available [SN_R-FG-2].
HABP is a new program; previously it was OFSP or household package,
hence it is the extension of the OFSP. We had provided household packages
for more than 160,000 PSNP beneficiaries. What differs between HABP from
OFSP is that it will be implemented in a strong way and with a new approach.
But the idea is similar. Some work has been done to organize cooperatives in
each kebele as an initial work for HABP, since there is a difference in the
credit system. Previously, credit was disbursed through the governmental
structure, but now it is going to be disbursed by local finance institutions
[SN_R-FG-2].
The HABP has more potential to help farmers exert their efforts and benefit
from the different alternatives in the package, while the OFSP has limited
opportunities and alternatives [TIG-R-KI-1].
Further, the respondents clearly outlined the linkage between different components of food
security programs and how this linkage could facilitate the overall graduation process that
the government set as an ultimate objective of the program. Excerpt from the regional food
security actors strongly signaled this:
PSNP alone cannot bring about graduation. If we strengthen the HABP and
provide credit services through preparing business plans with the intended
beneficiaries, we will register better results in graduation [AM_R/KI-1].
Graduation can’t be realized with PSNP only. Other household asset building
interventions should be rolled out to enhance graduation [ORO_R/KI-1].
Overall linkages between the complementary community investment (CCI), HABP, and
PSNP do not appear well-coordinated across regions. For instance, in Oromiya region, CCI
has already been started in lowland areas with potential for large-scale irrigation agriculture.
CCI has already commenced in Oromiya. It intends to graduate communities.
The region has been allocating more funds for this than the HABP
[ORO_R/KI-1].
Whereas, in the remaining regions, there is not much progress in terms of rollout and efforts
to link CCI with PSNP and HABP. In Tigray region, CCI has not been planned and started
yet.

187
CCI are infrastructures developed to facilitate graduation as part of any food
security program. So far, in this region practically this program has not yet
been implemented; we are in the process of preparation for the near future
[TIG_R/KI-1].
A similar trend was noticed in SNNPR region.
CCI is not yet implemented. CCI is a huge investment. If activities require 20
percent of capital budget, then it will be considered as CCI, for example,
bridge constructions. We heard that the CCI document is getting prepared,
but we haven’t received it yet [SN_R/KI-2].
However, significant variations were observed in the degree of knowledge among the actors,
such as the WFSTF and WOFSP focal persons at the woreda levels. Particularly,
considerable confusion exists in terms of differentiating the two programs (OFSP and HABP)
among these actors. Focus group discussions with KFSTFs in all 10 kebeles covered in this
study indicated that they have not heard about HABP. This could be attributed to the lack of
harmonized training and awareness creating campaigns, which are yet to be carried out
across most of the woredas. The following has been captured during key informant
interviews with the WFSTFs and WOFSP focal persons across the regions when asked,
“How is HABP different from previous Household Credit/Livelihoods Packages?”
There is basically no difference between the two programs; the real difference
is the focus. The HABP has received a lot of attention this time [AM-E/W-KI-1].
I do not know this program [HABP]. There is the letter I heard about it
[AM-SO/W-KI-1].
HABP has a business plan for three years, assuming that the beneficiaries
will graduate after that. However, the OFSP business plan is prepared on an
annual basis and consists of different activities—SME [(small and medium
enterprises)] and income-generating activities—and it does not forecast
graduation period [AM-SO/W-KI-1].
As to me, HABP is better targeted; independent on and off-farm committee is
established and better focused than OFSP [AM-SO/W-KI-1].
There could be a possibility in which non-PSNP beneficiaries could get credit
through HABP. The cooperatives will become stronger and this benefits the
community. They will make saving a tradition and as the result of this, their
living condition will be improved [SN-T/W-KI-1].
10.7. The Role of Development Agents in the HABP
Development agents have a key role to play in the overall food security and livelihood
program of the Ethiopian government, including the implementation of the HABP. This
includes, among others, the following tasks: (1) be member of both the KFSTF and CFSTF;
(2) support the CFSTF in prioritizing community needs and preparing annual HABP plans;
(3) assist households in the preparation and implementation of business plans, ensuring that

188
the business plans are the outcome of household decisions, not the supply (package) driven
approach of the past (GFDRE 2009b, 65).
In order to undertake this wide range of activities, considerable efforts have been made to
increase the number of development agents working at the kebele level.
Figure 10.1 and Table 10.3 provide some information on the availability of development
agents at the kebele level. Apart from SNNPR, nearly all kebeles have a development
agent’s office. In SNNPR, the median distance to the closest office, when one was not
situated in the kebele, was 4.5 km. Apart from Oromiya, most have a crop specialist. Fewer
have livestock specialists and development agents with specialist knowledge of off-farm
income are rare. There are a relatively large number of development agents whose specialty
is described as “Other”; it is possible that many of these individuals have training in Natural
Resource Management. Respondents perceive that there have been improvements in
development agent support since the 2008 survey. This may reflect the fact that many of
these development agents are relatively recent appointments. Many of these individuals
have been in place for a relatively short period of time, with 48 percent having been working
in their current location for one year or less, and 67 percent having been in place for no more
than two years. Approximately 16 percent of development agents are women. In Tigray,
Amhara, and Oromiya, approximately one-third of kebeles have a female extension agent.
This percentage is much lower, 18 percent, in Oromiya.
Figure 10.1. Percentage of development agent offices with 0, 1, 2, and 3 or more
agents
Source: Authors’ calculations based on the PSNP survey data.

189
Table 10.3. Development agent staffing at the kebele level
Kebeles with a development agent specialist in:
Region
DA office
in kebele
Three or
more DAs
Crops
Livestock
Off-farm
income
Other
Improvement in
DA support in
last two years
Tigray
91.7
75.7
83.3
72.2
8.3
69.4
66.7
Amhara
100.0
55.3
74.3
79.5
23.1
61.5
82.1
Amhara-HVFB
100.0
82.5
85.0
47.5
45.0
90.0
67.5
Oromiya
85.7
34.8
60.7
50.0
10.7
25.0
64.3
SNNPR
72.2
88.5
94.4
52.8
33.3
80.6
52.8
Source: Kebele quantitative questionnaire.
Note: All figures are percentages. DA = development agent.
However, linking development agents to HABP implementation has been limited. The key
informant interviews with the development agents across all regions under this study
indicated that the overall knowledge regarding their role in HABP is nearly nonexistent. Of a
total of 10 kebeles covered in the qualitative study, only 3 development agents in Oromiya,
Amhara, and Tigray regions had heard about the HABP and were (somewhat) able to
explain their role in its implementation. Among the development agents who had heard
about HABP, the overall knowledge and understanding about the program was shallow. For
instance, the development agent at Sokota woreda of Amhara region has associated HABP
with only the public works beneficiaries:
The HABP is for PSNP public works beneficiaries only. I wish this credit would
have been extended to non-PSNP households as well [AM_SO/K-KI-1].
Quotations from the development agents in Oromiya and Tigray regions below indicate
similar levels of understanding:
I do not understand the difference between OFSP and HABP. I feel both
focuses on household asset creation. I think PSNP beneficiaries and drought
affected communities should benefit from the HABP [ORO-Z/K-KI-1].
We were given training on the HABP; for three development agents and six
sub-kebele representatives. But we cannot describe its difference from the
OFSP [TIG_S/K-KI-1].
It is unclear whether they are equipped with the necessary knowledge to carry out the HABP
within their existing capacity. This issue was well illustrated when they responded to the
question, “Do you know what the role of development agents is in implementing the HABP?”:
Yes, my role is to provide technical support to HABP beneficiaries in the
future. But we expect details of roles from the woreda offices in the future
ORO_Z/K-KI-1].
My role is to assist households to create assets [AMR_E/K-KI-1].

190
10.8. Household Knowledge and Access to the HABP
Given this information on the roll-out of the HABP as seen at the regional, woreda, and
kebele level, it is interesting to consider whether households know about the HABP and
have gained access to its products. Tables 10.3 and 10.4 give some insights into this. Less
than half the households living in Oromiya and SNNPR had heard of the HABP. This is
consistent with the findings of the qualitative study. Somewhat surprising in light of the
qualitative data described above, however, knowledge of the HABP appears widespread in
Tigray, but much less so in Amhara.
In all regions, at least 40 percent of households had had some sort of contact with a
development agent, an increase relative to the number of households that had received
OFSP packages. Male-headed households were more likely to have had contact with
development agents than female-headed households. Direct support beneficiaries were less
likely to have contact than non-beneficiaries or households receiving public works payments.
This contact occurred primarily in meetings convened by development agents, although
between 20 and 30 percent of these contacts (depending on the region) occurred on an
individual basis.
Table 10.4. Household knowledge of HABP and access to development agents (DA)
How contact with DA came about
Region
Have heard
of HABP
Have had
contact with
DA
DA convened a
meeting that I
attended
Had individual
meeting with DA at
my home or their
office
Other
Tigray
72.5
50.3
54.5
23.1
22.4
Amhara
37.8
57.8
61.9
16.8
21.5
Amhara-HVFB
59.2
61.8
64.5
19.9
15.6
Oromiya
48.7
44.1
50.4
29.8
19.8
SNNPR
47.6
55.7
47.4
28.1
24.5
Source: Household quantitative questionnaire.
Note: All figures are percentages. DA = development agent.
Table 10.5. Household knowledge of HABP and access to development agents (DA),
by sex of head and beneficiary status
How contact with DA came about
Have heard
of HABP
Have had
contact with
DA
DA convened a
meeting that I
attended
Had individual
meeting with DA at
my home or their
office
Other
Male-headed
51.3
58.7
56.3
24.0
19.2
Female-headed
38.3
39.7
59.4
18.1
21.9
Receives PW
57.8
66.1
56.7
22.3
20.2
Receives DS
28.6
31.8
54.4
24.6
21.0
Non-beneficiary
46.6
51.8
58.1
21.1
20.5
Source: Household quantitative questionnaire.
Note: All figures are percentages. PW = public works; DS = direct support; DA = development agent.

191
A key component of HABP-supported activities is the development of a business plan.
Slightly more than 15 percent of surveyed households reported that they had developed
such a plan and had showed it to the development agent. Among households with a plan, 58
percent reported receiving assistance from a development agent when developing it. Kebele
officials also appear to play a role in the development of these plans, with 22 percent of
respondents stating that they had been helped by these individuals. Consulting either
development agents or kebele officials was clearly an important part of getting approval for
these plans. Approximately 90 percent of plans where development agents or kebele
officials had been consulted were approved by development agents compared to 78 percent
of plans where no prior consultations had occurred. Female-headed households and
households receiving direct support were less likely to develop business plans than other
household types. Conditional on developing a plan, however, approval rates were
comparable across all household types.
Table 10.6 describes the type of information that households received from development
agents.
Table 10.6. Information provided by development agents
Development agents:
Tigray
Amhara
Amhara-
HVFB
Oromiya
SNNPR
All
Suggested:
new crops
69.95
43.55
56.51
33.49
50.27
51.06
when to plant
71.39
49.10
67.34
38.31
55.31
56.94
Assisted:
in getting credit
70.67
50.30
61.18
30.32
43.59
51.66
in obtaining seeds
73.56
57.42
67.34
38.54
58.60
59.55
in obtaining fertilizer
76.68
55.97
63.56
39.72
58.05
59.02
Advised:
when to apply fertilizer
73.56
54.52
61.27
37.13
55.31
56.59
how to apply fertilizer
74.04
53.44
61.36
37.49
55.20
56.54
how to deal with insects
65.50
49.58
64.26
33.25
43.48
51.87
how to deal with crop disease
65.63
48.97
65.14
33.02
44.58
52.18
when to harvest
65.87
48.49
64.17
37.72
48.85
53.63
how to market crops
60.82
45.72
57.13
32.20
44.36
48.52
how to construct bunds
72.00
59.95
69.10
42.66
52.57
59.72
on irrigation and water harvesting
65.87
52.96
57.04
34.55
37.13
49.73
Helped:
with poultry production
65.14
42.94
49.74
29.96
38.88
45.45
with honey production
57.81
41.25
42.34
23.38
28.15
38.59
with non-agricultural income
43.39
33.41
32.39
21.03
21.36
30.26
obtain credit for agricultural production
40.02
34.62
31.25
18.57
20.04
28.85
Provided:
training on financial management
41.35
36.31
34.33
23.74
23.11
31.75
Source: Authors’ calculations based on the PSNP survey data.
The overwhelming impression provided by Table 10.6 is that development agents have been
active in the provision of a wide array of information. A closer examination, however,
provides a more nuanced view. First, households in Tigray are more likely to report that they
have received information on virtually every topic, compared to households located
elsewhere. The provision of this information is also relatively high in Amhara-HVFB localities
followed by other areas in Amhara and SNNPR. Provision of information and assistance
appears lowest in Oromiya. This pattern corresponds to development agent staffing levels

192
described in Table 10.3. Second, much of this assistance focuses on standard cropping
issues, such as when to plant, how to deal with plant or insect infestations, and so on.
“Nontraditional” topics such as helping establish nonagricultural income sources or the
provision of financial management trainings are less frequently reported. Third, it is less
clear whether development agents are providing generic advice or information specific to the
needs of individual households. One piece of evidence consistent with the view that generic
information is being provided comes from questions posed to households about
nonagricultural own business activities that they operated. Only 15 percent of these
enterprises received technical support or credit and only one-third of this 15 percent received
this support directly from development agents or local financial institutions.
10.9. Access to Credit
An important component of the revised HABP is improved access to credit. This is to be
provided by microfinance institutions (MFI) and Rural Saving and Credit Cooperatives
(RUSACCO). As part of the qualitative fieldwork, information was obtained about the
activities of these organizations.
Table 10.7. Number of RUSACCOs and their operational status, by woreda
Region
Woreda
Number of existing
RUSACCOs
Operational status
Amhara
Ebinate
None
-
Sayinte
No RUSACCOs
Three multipurpose cooperatives are engaged in saving
and credit services for their members.
Sekota
No RUSACCOs, but 23
basic cooperatives
Twenty-three multipurpose cooperatives established
and providing agricultural loans
Oromiya
Gursum
15
All have started saving and completed one-time
disbursement of credits. Credits are being repaid
without default.
Zeway D
12
Three are administering revolving funds from a World
Bank project.
SNNPR
Demba G
No RUSACCOs, but 22
basic cooperatives
Sometimes cooperatives are providing credit.
Shebedino
None
-
Tembaro
None
-
Tigray
Aferom
18
Thirteen are at the resource mobilization stage. Only
five have started disbursing loans.
Saesi Tsedamba
None
-
Source: Qualitative PSNP survey.
Table 10.7 shows that across these 10 woredas, there are some where RUSACCOs have
been established but these are by no means widespread.
In the Zewaydugda and Gursum woredas of Oromiya, RUSACCOs have benefitted from the
presence of a (now-ended) World Bank Food Security Project. Twelve RUSACCOs have
been established, three of which administer revolving funds from the World Bank project. In
Gursum, 15 RUSACCOs were established in 15 kebeles where the World Bank project had
been active. These have started collecting savings from and disbursing credit for their
members. While they are showing good progress in managing these funds, it appears that

193
their organization capacity may be limited. Leaders have relatively limited training on
cooperative management, only a few have a bank account, and most have no office space
and facilities to support their operations.
At the time of the survey, RUSACCOs were being established in Ahferom woreda, Tigray.
Out of 33 kebeles, 18 have RUSACCOs. Most of these cooperatives are at the stage of
resource mobilization by initiating members and only five had started disbursing loans. In
Amhara Region, no RUSACCOs had been established. However, there were some
multipurpose cooperatives in Sayint and Sekota woredas (3 in Saynit and 23 in Sekota) that
have been providing agricultural credit services. However, these are not formally linked to
either the OFSP or HABP. Similarly, there were no RUSACCOs in the sample woredas of
SNNPR. According to the Regional Food Security Case Team, only 15 woredas, where the
World Bank Food Security Project had been operational, have established RUSACCOs—
none of these woredas were in our sample. However, the region plans to establish
RUSACCOs in the future. “There is a clear goal set to establish at least one RUSACCO per
kebele in areas where there are no RUSACCOs. Strengthening and establishing
RUSACCOs is one of the future directions of the government” [SN/R-KII-OFSP].
There were three MFIs operating in these woredas: the Amhara Credit and Saving Institute
(ACSI), the Omo Micro-Finance, and the Dedebit Saving and Credit Institute operating in
Amhara, SNNPR, and Tigray regions, respectively. We did not come across any MFIs
operating in the sample woredas of Oromiya.
We triangulate these data with information on sources of agricultural credit identified in the
quantitative community questionnaire. This, too, shows that RUSACCOs, savings and loan
associations, and cooperatives have very limited operations in Oromiya and SNNPR.
Table 10.8. Sources of credit for agriculture, by region
Region
Government
RUSACCOs, savings and loan
societies, and cooperatives
NGOs
Private firms
Tigray
54.3
36.1
27.8
11.4
Amhara
66.7
29.0
18.2
0.0
Amhara-HVFB
82.5
44.7
10.5
0.0
Oromiya
32.4
2.9
20.6
0.0
SNNPR
62.2
2.9
11.8
0.0
Source: Authors’ calculations based on the PSNP survey data.
This limited rollout suggests that relatively few households will have had access to new
credit sources. To confirm this, we examined the credit data found in the household
questionnaire.
Approximately 20 percent of households reported borrowing more than 20 Birr in the last
12 months. Most (87 percent) reported only one or two loans. The vast majority of loans
were taken out for consumption purposes, with 45 percent of these loans being used for
purchasing food or other goods; 12.5 percent were used to pay for health expenses and
11.2 percent were used to repay other debts. The two most common income-generating
reasons for borrowing were to buy livestock (12.3 percent of all loans) or buying inputs such
as seeds, fertilizer, or pesticides (6.3 percent of all loans). Only 33 out of the 2,420 loans

194
recorded in the quantitative survey were to support the establishment of a new
nonagricultural business and most of these loans came from relatives, friends, or neighbors.
Table 10.9 shows the source of loans taken for agricultural purposes. Strikingly, households
borrowing to buy inputs rely heavily on informal sources, such as relatives, friends, and
neighbors. RUSACCOs and village savings and lending associations are a relatively larger
source of loans for livestock, but fewer than 7 percent of all surveyed households were
borrowing money for this purpose.
Table 10.9. Source of loans for agricultural purposes
Source of loan
Reason for loan
Buy inputs
Buy livestock
(percent)
Moneylender/arata
5.33
1.34
Relative
26.67
7.38
Friend/neighbor
21.33
3.02
Equub
2.67
0.34
Iddir
2.00
0.00
Cooperative, including rural savings and credit cooperative (RUSACCO)
10.67
30.87
Other local organization including village savings and lending association
8.00
19.46
Bank
0.67
3.69
Government/ministry/kebele
8.67
19.80
Microcredit institution or program /NGO
2.00
10.40
Other
12.00
3.69
Number of loans
150
298
Source: Authors’ calculations based on the PSNP survey data.
In Table 10.10, we explore whether access to formal and informal credit sources differs by
sex of head and household beneficiary status. Formal sources include RUSACCOs, other
cooperatives, village savings and loan societies, banks, government, MFIs, and NGOs.
Informal sources are moneylenders, relatives, friends, neighbors, equub, and iddir.
Consistent with the qualitative data noted earlier, women can borrow money from these
formal sources but are less likely to do so compared to male-headed households. Very few
households receiving direct support borrow money from formal sources.
Table 10.10. Sources of credit, by sex of head and household beneficiary status
Did not borrow
Borrowed from formal sources
Borrowed from informal sources
Male-headed
52.66
11.17
36.17
Female-headed
61.54
6.50
31.97
Non-beneficiary
55.72
9.97
34.31
Receives PW
49.04
12.06
38.90
Receives DS
70.22
3.43
26.35
Source: Authors’ calculations based on the PSNP survey data.
Note: PW = public works; DS = direct support.

195
10.10. Conclusions
This chapter has assessed the rollout of the HABP, an important component of the
revamped Food Security Program. Evidence from both qualitative and quantitative data
sources point to a number of areas where progress has been made.
The roadmap for the HABP implementation at the regional levels is almost complete and
all necessary technical committees are established. The general comprehension about
the objectives and purpose of the program among the regional actors is very strong.
Significant variations were observed in the degree of knowledge among the actors, such
as the Woreda Food Security Task Forces and development agents at the woreda and
kebele levels. In particular, considerable confusion exits in terms of differentiating the
two programs (OFSP and HABP) among these actors. Knowledge about HABP among
the PSNP beneficiaries at the community level is nearly nonexistent.
The links between the HABP and other food security programs is clear at the regional
level. However, considerable differences are observed between regional and woreda
actors in terms of understanding the relationship between the PSNP, OFSP, HABP, and
CCI.
Considerable effort has been made to increase the number of development agents at the
kebele level. There is widespread acknowledgement that this has led to an improvement
in support provided by development agents.
Many households report contact with development agents and, in particular, note that
they have received advice about new crops and how crops can be grown.
That said, a number of challenges remain:
Rollout of the HABP below the regional level is uneven. In particular, development
agents and households who were surveyed have limited awareness of the HABP. In the
case of development agents, there is uncertainty as to how the HABP differs from the
OFSP.
Advice and assistance remains concentrated on crop production. There is limited
capacity to assist nonagricultural enterprises.
Access to new forms of credit, such as RUSSACOs, has been limited. Relatively few
households report borrowing money to purchase inputs or buy livestock.
Female-headed households and households receiving direct support rarely access
formal sources of credit.

196
11. Graduation
11.1. Introduction
The notion of “graduation” has been integral to thinking about the PSNP since its inception.
“Graduation” describes a process whereby recipients of support move from a position of
depending on external assistance to a condition where they no longer need this support, and
can therefore exit the program. A “Graduation Guidance Note” (GFDRE 2007) describes
graduation from the PSNP as a transition from “chronically food insecure” to “food sufficient,”
defined as follows:
A household has graduated when, in the absence of receiving PSNP
transfers, it can meet its food needs for all 12 months and is able to withstand
modest shocks (GFDRE 2007, 1).
While the PSNP is designed to protect existing assets and ensure a minimum level of food
consumption, the HABP (and the OFSP before it) is designed to assist households in
increasing incomes generated from agricultural activities and to build up assets so that they
will be able to “graduate” off the program. The diagram below (Figure 11.1) illustrates the
theory behind the two stages of graduation. Starting at the lower left-hand quadrant of the
diagram, ultra poor and chronically food-insecure households are targeted with PSNP
transfers. Simultaneously, intensive support in the form of tailored products and financial
literacy and savings facilities are encouraged and provided so that households can stabilize
assets and, over time, move out of poverty. As households become less poor, extensions
services, complementary community investment (where appropriate) and business advice
(OFSP and HABP) are provided. As households’ economic base becomes stronger, they
reach the first threshold for graduation—the first red line, denoting graduation from the
PSNP. These households will likely need further support through extension and credit
provision, provided under the FSP. This will enable them to accumulate assets. At some
point the households will become strong enough to support themselves and will graduate off
the FSP altogether (this is the second level of graduation).

197
Figure 11.1. The graduation process
Source: GFDRE 2009b.
Conversations—prompted and unprompted—about graduation occurred frequently in key
informant and focus group discussions.
31
In this chapter, we summarize these, paying
particular attention to how graduation is understood and how it is applied in practice, issues
surrounding premature graduation, incentives and disincentives for graduation, and support
needed at graduation and postgraduation. In so doing, it covers a number of evaluation
objectives, as Table 11.1 notes.
Table 11.1. Evaluation objectives covered in chapter 11
Evaluation
objective
Issue
Link to Log Frames and TOR
Assess trends in perceptions of the effectiveness and transparency of the PSNP and HABP among different
groups of clients
Beneficiaries understand how the
program works
Quant_HH
Focus
PSNP Log frame Output 4.4
Beneficiaries and non-beneficiaries
report that targeting and graduation
processes are fair
Quant_HH
Focus
PSNP Log frame Output 4.5
Source: Authors’ compilation.
11.2. Prevalence of Graduation
As part of the community survey, key informants were asked to estimate the proportion of
households that had graduated from the PSNP since Meskerem 2000 EC (September 2008)
in their kebeles. Results are shown in Figure 11.2. In nearly half of these kebeles, no
31
A more complete and comprehensive picture of the processes, constraints, and experiences of graduation is provided in
“Assessment of Graduation” by Sandford et al. (2010).

198
graduation had been reported to have taken place and fewer than 5 percent reported
graduation rates above 10 percent of beneficiaries. There are regional variations in these
reports, with kebeles in Amhara and Amhara-HVFB more likely to report that graduation had
occurred. Amhara-HVFB was the only area where more than 10 percent of kebeles reported
graduation rates in excess of 10 percent.
Figure 11.2. Percentage of households that has graduated from PSNP based on
kebele reports, 2010
Source: Community questionnaire 2010.
As part of the household questionnaire, respondents were asked if they were not hired for
public works because they had been “graduated” from the PSNP. Only 84 households (71
male-headed and 13 female-headed)—3.3 percent of those not receiving PSNP benefits—
provided this response. It is possible that this understates the extent to which graduation has
occurred; for example, respondents who had graduated might describe themselves as “not
being eligible” for the program. When we look at the data we have on PSNP payments for
public works between 2008 and 2010, we find that 18 percent of households who received
these payments in 2008 were no longer receiving these in 2010. Unfortunately, using our
data in this way does not allow us to distinguish between households that have graduated
and those that are being excluded so as to make room for others.
11.3. Understanding of Graduation
Interviews with regional level officials revealed a solid grasp of the concept and processes
required for graduation—as described above. A focus group discussion with members of the
OFSP task force at the regional level in Amhara provided a concise view of graduation within
the program:
R1: The safety net is to fill the food gap of households that have food security
problems. This transfer by itself cannot enable them to create assets, become
self-dependent, and graduate from the program. It is only to cover the food
gap. Following this logic, other food security programs or different packages
were designed to build their assets. The HABP is a program that provides

199
credit services, various technologies, and inputs, and supports the creation of
assets to enable households to graduate [AM/R-FG-2_OFSP].
A key informant interview with a member of the Regional Food Security Task Force in
SNNPR likened the process to graduating from school:
R1: When you attend a school and reach at some level, you will receive a
certificate. Just like that, our vision is to protect their [beneficiary] household
assets and, along with that, they will receive a credit that will help them to
graduate. So, our final goal is graduation [SNNPR/R-KII-1_FSTF].
Regional respondents are aware of the main tenets of the graduation hypothesis
underpinning the program. They typically emphasize the importance of “other” programs that
facilitate assets creation, as well as credit provision and larger scale infrastructure, as
fundamental to graduation. The logic of how all the separate components of the FSP work
together is summed up by a respondent from a special OFSP task force:
R2: The ultimate goal is the graduate [beneficiaries] and to change their
livelihoods. To protect their [the beneficiaries] asset depletion and push them
back up when they encounter localized shocks, they are supported by risk
financing according to assessments. Mainly, the driving force of graduation is
what we currently call HABP. Therefore, if food gaps are filled, localized
shocks are resisted by risk financing, their lives are strengthened through
different mini-HABP projects, then they improve the living status of their
family, create assets, and after that they become free from CFI, become self-
dependent, and graduate from the program [AM/R-FG-2_OFSP].
The concept, therefore, is well understood at the regional level. Less well understood was
the finer distinction between graduating from the PSNP transfer and graduating from the
FSP program. It was only at the woreda level in Tigray, Afherom, that officials spoke of the
two-tier graduation concept: “there are two types of graduation: PSNP, that is, attaining 12
months of food gap and food security attainment”; “Food security attainment is difficult
through food supply alone, rather building assets is the only option for food security”
[TI_AF/W_FG_OFSP].
The understanding of graduation at the woreda level is also fairly consistent, although with
some interesting variations. This may reflect, in part, the fact that the guidance notes on
graduation were found in virtually all (92 percent) woredas where the woreda quantitative
survey was carried out. Statements of graduation characteristics included “asset rich,” “able
to secure food independently,” “able to secure health and education independently,” “we
consider different assets like ox, land, and permanent plants.” A minority of interviews
revealed that some woreda level officials believed that “graduation is quota driven,” a topic
discussed later in this chapter. At the kebele level, the understanding of graduation becomes
very loose, at times completely uninformed, and sometimes incorrect as the diversity of
responses below suggests:
Graduation is the process of creating economically capable PSNP
beneficiaries [TI_A/DA].

200
For me, one is able to graduate if he has the capacity to withstand
unfortunate conditions like drought, flood, or ice storm, independently at least
for some time [TI_A/FGD_KFSTF].
According to what we have heard, it is about the poor being able to farm, able
to construct a house, have a pair of ox. Those who do not have grain and
attain self-sufficiency through home production [SNNPR_D/FGD_KFSTF].
But the minimum [for graduation] should be those who have a pair of ox and a
milking cow, those who transformed from thatched houses to corrugated iron
sheets roofing , and who met their food gaps better than others. This is
between 10,000 and 15,000 Birr after adding everything up. This is the
practical benchmark that we apply [SNNPR_T/KII_DA].
In focus group discussions with households, some indicated that graduation was a time-
defined phenomenon—after five years of support, the household will graduate:
We heard that people graduate from PSNP by the end of the fifth year. We
also heard that some who leave safety net early by their own initiative
[TI_A/FG1].
Yes we heard about graduation from PSNP. We heard it from the agricultural
office workers at gatherings/meetings, where they explained to us that in the
five years time, if the living condition of a household changes, they graduate
from the PSNP program. For some kebele leadership and sub-
kebele/Gujile/Got representatives, a small booklet-like document is distributed
which explains about the process [TI_S/FG-2].
Others, in SNNPR, only knew about it when they witnessed households in their community
being taken off the program:
We were not told initially. We heard about graduation. When the graduates
went out, they told us that all of you will go out like them sometime in the
future [SN_S/FG-2].
We only saw those leaving out who should not have been excluded but have
not heard clearly about graduation. What we heard is that they are given
things essential to support graduation when they get graduated [SN_T/FG-2].
Others had a more informed understanding of the process.
32
A household having more than 18,000 Birr is supposed to graduate. The
directive came from the WFSTF [OR_F/FGD-1].
Individuals graduate if they can cover food gaps plus assets of 4,000 Birr per
individual. We have been told by the development agents and in meetings to
make ourselves ready for graduation [AM_S/FGD-2].
32
We also explored households understanding of graduation criteria in the household quantitative survey. Most (48 percent) of
respondents stated that households were eligible for graduation if they “had sufficient assets to ensure food security” or if it
“could feed itself all through the year” (17 percent). However, in light of these qualitative findings, it is clear that this question
was not sufficiently nuanced to elicit deep understanding of graduation criteria.

201
Disconcertingly, a number of graduates we spoke with appeared to be the least informed of
why they were graduated.
The process was not clear. We were told to graduate because we were in the
program for five years. Two elders were assigned in each village to propose
households for graduation. We were screened for graduation by these elders
and finally we left the program. We did not know the process and how we
were selected [OR_G/FGD-2].
We first understood [we would graduate] on that day of the meeting. First,
many individuals were selected saying ‘you will be graduate in the future,
however later on, after three months, they again included us and we started
to work. Later on they called us, in January, and told us we are graduated.
And we accepted this, since it is a directive of the government. However, we
are not ready to graduate [AM_E/FGD-2].
No, the process [graduation process] was not clear. They kicked me off prior
to attaining enough capacity. I was kicked off after two years in the program
and immediately I got a package with it, giving me forage seed or feed for the
ox. I was not self-sufficient [SN_DG/FGD-2].
A PSNP coordinator working at the woreda level, Sokota, Amhara, summed up the
confusion surrounding graduation in the following way:
I feel that there is a problem with the concept of graduation. People,
especially implementing bodies and individuals of the program, at all levels
understand graduation in their own way [AM_SO/KII-PSNP Coordinator].
11.4. Graduation Benchmarks
The key source of guidance for graduation is the Graduation Guidance Note (GFDRE 2007).
It identifies seven core principles for the introduction and use of benchmarks as well as 16
steps that regions, woredas, kebeles, and communities should undertake in identifying
graduates. According to the guidance notes, benchmark levels of assets for graduation are
as follows: Oromiya, 19,187 Birr per household; Tigray, 5,600 Birr per capita; Amhara, 4,200
Birr per capita; and SNNPR, 2,998 Birr per capita. Discussions with officials across the four
regions consistently revealed that there was accurate knowledge of these benchmarks.
At the kebele and community levels, knowledge of the benchmark was much more
imprecise. In the kebele quantitative survey, respondents were asked, “What criteria were
used for graduation?” Out of the 92 kebeles that provided this information, 65 gave
responses coded as “Household has sufficient assets to be food secure” and 60 gave
information consistent with the pre-coded answer, “Household is able to feed itself
throughout the year.” Other responses are a source of concern: 13 stated that the household
has enough able-bodied labor; 30 reported that the household had repaid the credit it had
borrowed; it has repaid credit; 38 said that the household “is well off”; and 10 stated that the
household had received its share of PSNP benefits (four of these responses came from
Tigray).

202
Further, the benchmark values, while easy enough to quote at the regional and even woreda
levels, need to be translated into measurable indicators at the kebele and community levels.
Key informant interviews, especially with the development agents, showed that the
determination of whether someone has reached a benchmark is often interpreted in different
ways. Instead of specifying a single value benchmark, respondents in focus groups believed
that if a certain set of criteria was met, such as having corrugated iron sheets roofing on
one’s house, or owning two oxen, or having a certain amount of land, or credit, then that
would define graduation. While these indicators do not reflect accurate knowledge of the
benchmarks, they do, in fact, reflect how people actually experience the process of
graduation.
We see from these interviews, especially with the development agents, that how to
determine whether someone has reached a benchmark is interpreted in different ways.
Table 11.2 shows the different elements included in the benchmarks at the kebele level
(development agent and Food Security Task Force interviews).
Table 11.2. Elements included in benchmarks at kebele level (interviews with
development agents and Food Security Task Force)
Kebele
CIS roofing
Land
Livestock
Technology
Income stream
Crops
Farms
SNNPR
Shebedino
√
√
√
Tembaro
√
√
√
Demba Gofa
√
√
√
√
Amhara
Ebenat
√
Sayint
√
√
√
Sokota
√
√
√
√
Oromiya
Ziway Dugda
√
√
√
√
Gursum
√
√
√
√
√
Tigray
Ahiferom
√
√
Saesi Tsaedamba
√
√
Source: Authors’ calculations based on the PSNP survey data.
Note: CIS roofing = corrugated iron sheets roofing.
A variety of views were expressed on the appropriateness of the benchmark.
For me, one can graduate if he has the capacity to withstand unfortunate
conditions like drought, flood, or ice storm independently at least for some
time, animal may not indicate even they can be abolished in one day but
someone who have saved capital, better innovativeness, model farmers
should be the criteria for graduation [TI_A/KII-KFSTF].
Yes, we already discussed that there was no formal graduation, rather
exclusion due to budget limitation; the excluded people are also poor and
incapable to lead their family safely. Some of them are suffering food
insecurity problems [TI_S/FG-4].
If we talk openly, we do not know what will happen to us tomorrow. The two
who were excluded should not have been so. Here in this kebele if one

203
constructs corrugated iron sheets roofed house, then it is considered as if he
has made improvement [SN-D/FG-3].
Those who should graduate should be those who can teach other PSNP
beneficiaries or attain a stage where they can be examples for others. This is
when they fully use agricultural extension services, from living house
construction to using livestock barns. But now as a resistance to their
exclusion they go to court sometimes, even selling their assets to cover their
expenses [SN_S/KII-DAS].
The benchmark from the region is not appropriate. Because I have an opinion
that it is too low. Ours is appropriate. There is a pressure from above to have
graduation. The minimum should be those who have a pair of ox and a
milking cow, those who transformed from thatched houses to corrugated iron
sheets, and those who met their food gaps better than others. This is between
10,000 and 15,000 Birr after adding everything [SN_S/KII-DAS].
The 18,000 Birr benchmark is not adequate to guarantee year-round access
to food for families. Also, some assets like house are basic and not
convertible to food in case of shortage [OR_Z/KII-EW].
The second thing is there are assets incorporated when we do the calculation
[or graduation benchmark]. I think these things have to be revisited. For
example, there are livestock dependent areas and crop dependent areas.
Now since we consider only productive assets, in the livestock region they
can easily reach graduation. Whereas in crop dependent areas, people can
harvest 30 or 40 or 50 quintals, but there is no common understanding to
consider this harvest...there is also variation: for example, the standard
[quality] of ox [AM/KI-OFSP].
We were told to graduate safety net households from this kebele two, three
times. We know that there is a benchmark and a calculation for graduation.
When this calculation was made, there was no household that can graduate
from our kebele [AM_S/KII-KFSTF].
11.5. Is there Premature Graduation?
As noted in Section 11.2, relatively little graduation has occurred. But given the importance
of graduation within the context of the Food Security Program, an imperative is increasingly
sensed that graduates should be identified and taken off the program. This imperative is felt
and discussed at all levels. Sometimes respondents, particularly at regional level, discuss
these targets in relation to quotas and required future rates of graduation. However, at lower
levels, these “quotas” are often seen as administratively imposed and at odds with the time
required for sustainable graduation.
These pressures provoked discussion as to whether households were graduating too early.
Box 11.1 provides a range of these responses. The perception of quota-driven graduation
was mainly seen in a negative light. In the SNNPR woredas we worked in, we found there
are no standard benchmarks applied for graduation; instead, graduation was based on the

204
local perceptions that somebody has attained self-sufficiency. Whether a household is ready
for graduation was often determined by a relative comparison of people who are subject to
community judgment in regard to the attainment of a set of criteria. Development agents and
community leaders in most cases are not comfortable with the criteria, because, according to
them, emphasis is given to quota fulfillment rather than a real change in the situation of the
household.
Box 11.1. Responses to the question: “Why are people graduating too early?“
Woreda level
The woreda had planned for 100 percent graduation of all public work beneficiary households in
the last five years. In 2009 alone, the region gave us 30 percent graduation quota. But so far,
only 9,017 people graduated from the program. This was forced graduation because most of the
graduates left the program before they were able to fill their family’s food gap [OR_G/KII-
WFSTF].
Yes, most of the graduates left the program early. This is because there is a quota that the
woreda FS and DPPO gives us to fulfill [OR_Z/WKII-EW].
Kebele level
This is because of the quota limitation given to us from the woreda to exclude 126 household
heads...of the excluded beneficiaries some of them are suffering food gaps [TI_A/KII-DAS].
Most of them are not able to be food secure; it is just that they have been excluded to fulfill the
quota given from the kebele [TI_A/KII-KFSTF].
We were told to graduate safety net households from this kebele two, three times. We know that
there is a benchmark and a calculation for graduation. When this calculation was made, there
was no household that can graduate from our kebele [AM_S/KII-KFSTF].
Community level
People are graduating too early. When they are subject to a pressure from upper bodies, they
said if graduation is a must, then somebody has constructed a house, has a calf, and should get
graduated, but in reality it was not on the basis of attainment of the required capacity.
Particularly when one has a household package, the cattle from this package attracts the
attention of people [SN-D/FG-2].
The chairman deleted my name to make a replacement for his own relatives. On that day, the
chairman included five extra people. The father and mother of the chairman are in the program. I
was excluded based on the alleged retargeting. They told me that the government said that is
enough for you and another person should be included [SN-D/FG-2 case study].
We believe that it is because that the development agents are told to get promotion and rewards
as a result of our graduation...it is only for the purpose of reporting to the kebele and to the
woreda but did not do it based on knowledge about our improvement. It is not done to help us.
When we hear on the radio about other areas, the graduates get graduated when they
themselves declare that is enough for us. But here graduates are excluded even without making
them clear about its objective [SN_S/FG-2].
Source: PSNP survey data.
Given the variety of criteria that appears to be applied when considering if a household
should graduate, it is difficult to assess quantitatively the extent of premature graduation.
This is exacerbated by the fact, as discussed above, that few respondents explicitly
indicated that they were not receiving transfers because they had graduated. However, what
we can do is assess whether those individuals who were receiving public works payments in

205
2008 but not in 2009 or 2010 had characteristics that, in light of the criteria described above,
looked like they were candidates for graduation.
Table 11.3 considers three groups of households by their PSNP status: those that do not
receive payments for public works in 2008, 2009, or 2010 (“No PW”); those that received
payments for public works in 2008 only; and those that received payments for public works
in 2008, 2009, and 2010. For each group, Table 11.3 shows values of livestock holdings as
of June 2008—the period at which the 2008 round of public works activities was starting to
draw to a close—at different points along the distribution (poorest 10th percentile, 25th
percentile, median, 75th percentile, and richest 90th percentile) for each region and group.
Table 11.3. Value of livestock holdings (Birr), Sene EC 2000 (June 2008), by
beneficiary status between 2008 and 2010, and region
Region
Status
10th
percentile
25th
percentile
Median
75th
percentile
90th
percentile
Tigray
No PW
120
1,736
4,859
9,660
14,860
PW in 2008 only
320
2,600
5,596
13,400
21,000
PW in 2008, 2009, 2010
470
2,270
4,950
8,400
13,000
Amhara
No PW
0
2,000
6,000
11,060
17,962
PW in 2008 only
0
600
3,385
5,600
9,110
PW in 2008, 2009, 2010
0
1,005
3,000
5,245
9,000
Amhara-HVFB
No PW
360
3,000
6,240
10,590
17,398
PW in 2008 only
0
1,800
4,315
6,920
9,450
PW in 2008, 2009, 2010
0
1,270
3,560
6,000
8,495
Oromiya
No PW
80
1,785
5,625
11,200
20,400
PW in 2008 only
400
2,700
7,600
13,200
23,400
PW in 2008, 2009, 2010
0
1,200
4,700
9,030
16,000
SNNPR
No PW
0
1,000
3,000
6,000
11,800
PW in 2008 only
0
1,000
1,600
2,560
6,890
PW in 2008, 2009, 2010
0
200
1,750
3,600
6,800
Source: Household survey, 2010.
Note: PW = public works.
We begin with Tigray. Looking at the row “PW in 2008 only,” we see that households in this
category had median livestock holdings of 5,596 Birr. Given the regional benchmarks
described in Section 11.4, this looks low. However, those households at the 75th or 90th
percentiles have livestock holdings that are close to the benchmark (and, of course, this
ignores consideration of other assets). Further, at all points in the distribution, those
households that were removed from the PSNP after 2008 look like households that were not
receiving payments in 2008, 2009, and 2010 (if anything, they look slightly better off). They
are clearly better off than households that continued to receive benefits after 2008. Data
from Oromiya tell a similar story. But in Amhara, households that were dropped after 2008—
either because they were “graduated” or because they were rotated off—do not look any
better off than those households that were retained in the program. While this also appears
true in SNNPR, the generally low levels of livestock holdings suggest, as does Table 11.1,
that other criteria were being used.

206
11.6. Incentives for Graduation
Sandford et al. (2010) classify and document the various incentives and disincentives that
PSNP clients face with respect to graduation. Regarding the former, they report that
graduates at the individual and community level shared five factors that motivate them to
graduate: (1) pride in graduating; (2) receipt of agricultural inputs and other prizes during
graduation; (3) availability of other more attractive livelihood options; (4) encouragement by
model farmers; and (5) woreda level incentives. In our sample, we find little to no evidence
with respect to (1), (3), and (4). We find a lot of evidence about (2); however, this concerns
the “promise” of inputs and prizes rather than the actual receipt of inputs and prizes. We also
find evidence of (5). As well, we document a further incentive related to the nonequivalence
of transfers between PSNP, contingency funds, and risk financing.
The Promise (or Expectation) of Material Incentives
Responses from the community focus groups and also at the woreda and kebele levels
indicated that there was much confusion about how households actually graduate. This was
particularly evident in respect to whether incentives had been given to potential graduates.
Focus groups recounted that they had ‘heard of’ material incentives being given to promote
graduation:
When we hear from other areas or kebeles when people get graduated, they
are awarded about two thousand Birr and farm implements. Here, the
graduates did not get such support. Here, there is not any support. If
somebody constructs a house from his personal effort and has an ox, then he
is classified as a graduate. Here, besides the advice, there is no other support
[SN_D/FG-2].
We heard that when there is graduation, money is going to be given
[AM_E/FGD-3].
Respondents from transitioning households, that is, those households who actually have the
potential to graduate or who have graduated, provided first-hand experience:
Yes, we were promised to be given ox and/or cow after graduation
[OR_G/FGD-2].
Some graduates were promised to get support through various development
activities, but this did not happen in reality. We were promised to be given
beehives, but this did not happen [OR_Z/FGD-2].
Last year 1,000 Birr was given as a credit for graduating households. This
year we were expecting graduation money, but, due to a shortage of money,
graduation did not take place. They were told to graduate and collected
1,000 Birr loan, but their expectation was 5,000–6,000 Birr [AM_E/FGD-1].
Are the perceptions held by the community about material incentives corroborated at higher
levels of administration? Interviews with Kebele and Woreda Food Security Task Forces
suggest that there is limited “encouragement” given for promoting graduation (for instance,
access to credit or extension). For instance, in an Amhara woreda, a key informant interview

207
said that “The region did a good job that encouraged graduation. PSNP graduates were
recognized as heroes of development and awarded” [AM_S/KII-WFSTF]. However, we find
little to no evidence that specific prizes, awards, or payment (in cash or kind) is given in the
event of graduation. The notion that graduation is associated with prizes was common in all
woredas in SNNPR and many woredas in other regions.
The Nonequivalence and Fungibility of Transfers
The “nonequivalence of different transfers” is an interesting issue that was raised in some
woreda and kebele level interviews. There appears to be an increasing awareness among
beneficiaries that there is some room for “maneuver” between different types of transfer and
assistance within the food security program. For instance, in Tigray, a focus group
discussion with woreda level officials revealed that a significant number of people are “self-
graduating” from the PSNP in order to then appeal, on the basis of food insecurity, to be put
on another form of transfer (usually the contingency fund or risk financing). This has multiple
benefits for the households: first, it is a timely payment (unlike the PSNP). Second, it is a
grant and does not require a public work commitment. Therefore, people are able to claim
support, receive it on time, and pursue other work opportunities. This is a possible challenge
for the coming years.
It is also possible that enforcing graduation quotas will lead to premature graduation for
many households, which will appeal to other transfer benefits within the FSP. A key
informant interview with a development agent in Oromiya said that “due to pressure from
woreda from graduation quota this led to early graduation” and “all graduated households
were re-targeted through the 5 percent contingency fund” [In OR_ZD/KII_DA]. Premature
graduation may simply transfer the food-insecure caseload to different budget lines within
the FSP program.
11.7. Disincentives for Graduation
While sustainable graduation is seen as the ultimate goal of the FSP for the majority of
households, a number of factors constrain the potential for households to graduate in a
program-linear way. These, at times, can be administrative- or procedure-specific (the
benchmark may not be set appropriately), beneficiary-specific (a lack of desire to graduate),
or exogenously determined (due to weather-related shocks). In this section, we review the
survey evidence for the constraints to sustainable graduation. We distinguish benchmark
graduation (a static benchmark threshold either related to the PSNP or the FSP, as shown
above) from sustainable graduation (the ability of the household to remain above the
benchmark in the medium to long term). We make this distinction because identifying
households according to a benchmark will attract different constraints (such as hidden
information and administrative problems) than those enabling longer term fulfillment of that
benchmark (such as weather shocks and access to markets).
Reluctance to be Self-Reliant
As with all forms of social assistance, concerns exist about whether the very provision of
them will create a reluctance to be self-reliant or to become dependent upon them by the

208
recipients, and thus undermines (particularly in this case) the prospect for graduation. It is
reasonable to expect that the transfer, in the case of the PSNP, for direct support
beneficiaries may lead to what Lentz, Barrett, and Hoddinott (2005) refer to as positive
dependency, as there is no expectation that this group will graduate. However, for the
majority of the public works beneficiaries, dependency on the program could severely
undermine the graduation and livelihood strengthening objectives.
Many respondents at the administrative levels mention “an attitude” of dependency as a
critical problem facing the success of the FSP/PSNP.
The most important problem is attitudinal, where households expect the
safety net to remain there forever. The other issue is lack of hard work among
some households [TI_S/KII-DAS].
Of course, our people do not like graduation. They always want to take. There
are some people that I do not agree with their inclusion. Many people have no
sense of tomorrow. There are people who say the government will feed us
and simply leave [SN_D/KII-DAS].
There are individuals that say, “I will continue to get support for ten years as
long as PSNP is continuing for ten years” [SN_S/KII-WFSTF].
It seems that some graduates are reluctant to accept graduation because of
dependency syndrome. Though not all, there are some graduates who can
feed their family all the time, send their children to school without PSNP
support, and buy and use improved farm inputs from own sources. For
example, there is one graduate who has managed to produce number of
livestock heads including two oxen, irrigation motor pump, and relatively good
farm, but this graduate insisted that his family still has a food gap [Comment
on OR_G/FGD-2].
Our kebele people do not want to graduate. There is no positive thinking
about graduation. There is a problem of attitude towards graduation
[AM_S/Case study].
The data from Amhara, Ebinat woreda, indicate that dependency on assistance is highly
developed and deeply rooted within the culture of this woreda. “The problem is very grave.
Every member of the society demands assistance because there has been blanket
assistance of food aid for the last 30 to 40 years in this woreda for all members of the
community” [AM_E/field report].
One graduate honestly admitted to a desire to remain dependent:
R: Yes, I can feed my family throughout the year, and have some livestock
heads. But I was not interested to lose PSNP transfer. I did not say I am
enough with the transfer money. But, they pushed me and I left the program.
[It seems dependency on aid is at the root of resistance to graduation.]
[OR_Z/FGD-2]

209
Hidden Information
Many interviews with woreda and kebele level officials suggested that these officials
believed that beneficiaries often attempt to hide true information about their assets, as they
do not want to graduate.
There is a huge challenge on what people do so as not to graduate. There
are individuals that physically hide their cattle. From this, I think that such
individuals that have been receiving support for the last five years, even if
they will continue to get support for ten years, they will never graduate
[SN_S/KII-WFSTF].
There is hiding of assets on the part of households when it comes to
graduation. This attitude has to be discouraged [AM_E/FGD-OFSP].
For the graduation purpose there is a form that development agents fill
through indirect community asset valuation, since the community hide their
assets [AM_S/KII-OFSP].
When we go for calculating individuals [ready to graduate], we cannot get as
needed. Individuals hide what they have; they take their livestock to their
relatives. They do not give you the right information [AM_S/KII-DA].
11.8. Support Needed for Graduation from PSNP
Interviews with members of the Woreda Food Security Task Forces, Kebele Food Security
Task Forces, and other special committees at the woreda level provided some suggestions
for necessary provision to support graduation. An interview with a member of the Woreda
Food Security Task Force in Saesi Tsaedamba, Tigray, summed up some points held by
other similarly placed officials in other regions:
When we look at the overall achievement of the PSNP support provided so
far, we felt it was not implemented well. Even the OFSP packages were not
utilized as per the intended plans. For instance, many took credits for
weddings and other unproductive investments. But now we have embarked
on full family targeting and are working very closely on the use of household
packages, thus we hope to achieve better implementation and get more
graduates [TI_S/KII-WFSTF].
The reference to the importance of full family targeting and effective use of packages is
critical to the national vision of pathways for graduation. On the subject of packages, a key
informant interview in SNNPR said that “There should be other packages in addition to the
household package, like the World Bank that supports in other woredas...when one option is
provided for a farmer, the other farmers will observe and learn from that person” [SN_D/KII-
OFSP]. This view resonated with a more widely held perception that in woredas and kebeles
where there are multiple players and packages (over and above the government), there
would be a higher likelihood of sustainable graduation: “Graduation is more likely in areas
where beneficiaries have access to adequate and productive farmland, credit, and irrigation.
We have already observed this in some woredas like Fentalle, Melka Ballo, Goro Gutu
woredas, etc. World Bank credit has played a key role in this regard” [O/KII-RFSTF].

210
In addition to full family targeting and more effective packages, respondents at woreda and
kebele levels frequently discussed the importance of large-scale irrigation, dams, and
access to agricultural technologies as critical activities to support large-scale graduation.
Complementary community infrastructures like expansion of irrigation facilities
should be strengthened [TI/FGD-FSTF].
There should be full family targeting; big community investments like check
dam need to be strengthened; agricultural technologies like improved seed,
fertilizer, and irrigation facilities need to be introduced widely [TI_A/FGD-
OFSP].
The main thing is awareness creation. The beneficiaries should believe that
they could improve themselves if they work hard. And we need to prepare
ways in which farmers could be able to produce two or three times per year.
New agricultural technologies should be transferred to the farmers. They
should be able to get water nearby; if the water table is near to the surface,
they should be assisted to take the water out by pump and use it to cultivate
vegetables in their garden, so that they can sell their vegetables [SN/KII-
FSTF].
If the climate is conducive and when people continue working on irrigation
and year-round production is possible, we hope that 70 percent will graduate
in the coming five years (2010–2014). If irrigation is successfully used,
graduation will be realized; 18/28 kebeles have got water. Therefore if (a) we
enhance work ethics training, (b) enhance irrigation facilities, (c) water pump
maintenance training for farmers, and (d) buy quality motors [OR_Z/FGD-
OFSP].
Views from the focus group discussions with community members on how to facilitate
graduation were sometimes of a different nature than those expressed from administrative
levels. They tended to focus on household-specific needs, rather than activities and
investments that would lead to graduation at scale. So, for instance, many said that the
provision of oxen to households or schooling for children would be facilitators for graduation.
One respondent explained the assistance that PSNP households need to graduate as, “if the
households does not have ox, then provision of farming ox, if the households does not have
milking cow, then making a provision, if living in grass thatched house make him have a
house with 42 corrugated iron sheets, if all these are provided by being told after this you
have to be self-dependent, then they can do.” Some, especially those who were in the
transitioning focus groups, identified very specific assets as necessary for graduation. “If
there is availability of irrigation with technological packages given to them, I think they will
reach the benchmarks” [TI_S/FG-1].
11.9. Support Post-PSNP Graduation
Graduates from the PSNP are supposed to continue to obtain support in the form of
extension and credit for a defined period. After this time, they are expected to graduate from

211
the FSP. Interviews with officials at regional and woreda levels were aware of this, and
claimed to provide this support:
Once households are graduated from PSNP, they are still supported to fill
their food gaps for the subsequent one year, with the amount of food
equivalent to what they have been getting when enrolled for safety net. For
the higher level graduates (who passed to full self-reliance), to prevent them
from going back to their previous status, they are supported in different ways,
including close follow-up, repeated trainings, creating market linkages,
constructing postharvest storages [TI/KII-OFSP].
There is technical support and training; the graduates will not be out of our
focus [AM_E/FGD-OFSP].
However, in keeping with the findings reported by Sandford et al. (2010), we find that none
of the PSNP graduates interviewed continued to receive credit or further support from the
FSP.
11.10. Summary
There appears to have been relatively little graduation to date.
There is a solid understanding at regional levels of the concepts and mechanisms of
graduation; however, there was only one mention of two-tiered graduation.
Understanding at woreda level is also fairly consistent. Below woreda level, the
understanding of the concept becomes very loose, at times completely uninformed, and
at times completely incorrect. Perhaps most disconcerting is the finding that the
graduates we spoke with appeared to be the least informed of why they were graduated.
The knowledge of the process by which people graduate was very varied at the
community levels, with some thinking that it was time-dependent, others believing it to be
political, quota-driven, and others knowing the specific benchmark value for graduation.
The experience of graduates themselves indicated that the process was in most cases
not transparent or well-explained.
Officials at most levels are well informed of the actual graduation benchmarks. However,
when the official regional benchmarks are translated into specific criteria for targeting
potential graduate households at the kebele and community levels, a plethora of criteria
result.
Given the intention of the government to meet and fulfill the FSP graduation targets,
there is an imperative that graduates are increasingly identified and taken off the
program. This imperative is a reality and is felt and discussed at all levels, from regions
right down to the communities. Sometimes respondents, particularly at the regional level,
discuss these targets in relation to quotas and required future rates of graduation.
However, at lower levels, these “quotas” are often seen as administratively imposed and
at odds with the time required for sustainable graduation.
The incentives for graduation include (1) the promise (or expectation) of material
incentives; (2) the nonequivalence and fungibility of different transfers; and

212
(3) regional/woreda level incentives. These incentives may increase the incidence of
benchmark graduation, but may undermine sustainable graduation. The main finding on
disincentives for graduation has to do with a reluctance to be self-reliant.
A number of constraints/enablers of sustainable graduation were identified, including (1)
full family versus partial family targeting; (2) the level of household and community
assets, particularly land; (3) price changes; and (4) natural, climatic shocks. Many
respondents believed that if sustainable graduation was to be achieved at scale, then
large investments in community level agricultural technologies and irrigation are critical.
This is in keeping with the vision of complementary community investment, however
complementary community investment is currently intended for lowland areas only.
While support is available for new PSNP graduates, we found little evidence of graduate
households receiving this support.

213
References
Gilligan, D., J. Hoddinott, A.S. Taffesse, S. Dejene, N. Tefera, and Y. Yohannes. 2007.
“Ethiopia Food Security Program: Report on 2006 Baseline Survey.” Mimeo,
International Food Policy Research Institute, Washington, DC.
Gilligan, D., J. Hoddinott, N. Kumar, A.S. Taffesse, S. Dejene, F. Gezahegn, and Y.
Yohannes. 2009. “Ethiopia Food Security Program: Report on 2008 Survey.” Mimeo,
International Food Policy Research Institute, Washington, DC.
GFDRE (Government of the Federal Democratic Republic of Ethiopia). 2004. Productive
Safety Net Programme: Programme Implementation Manual. Addis Ababa: Ministry
of Agriculture and Rural Development.
———. 2007. Graduation Guidance Note. Addis Ababa: Food Security Coordination
Directorate, Ministry of Agriculture and Rural Development.
———. 2009a. Food Security Programme, 2010–2014. Addis Ababa: Ministry of Agriculture
and Rural Development.
———. 2009b. Household Asset Building Programme, 2010–2014. Addis Ababa: Ministry of
Agriculture and Rural Development.
———. 2010. Productive Safety Net Programme: Programme Implementation Manual. Addis
Ababa: Ministry of Agriculture and Rural Development.
Lentz, E., C. Barrett, and J. Hoddinott. 2005. “Food Aid and Dependency: Implications for
Emergency Food Security Assessments.” Mimeo, International Food Policy Research
Institute, Washington, DC.
Ramakrishna, R., and A. Demeke. 2002. “An Empirical Analysis of Food Insecurity in
Ethiopia: The Case of Northern Wollo.” Africa Development XXVII (1 and 2): 127–
143.
Sandford, J. et al. 2010. Ethiopia Productive Safety Net Programme: Assessment of
Graduation Report. August 2010.
Sharp, K., T. Brown, and A. Teshome. 2006. “Targeting Ethiopia’s Productive Safety Net
Programme.” Mimeo, Overseas Development Institute, London.
World Bank. 2009. Project appraisal document on a proposed grant (APL III) in support of
the third phase of the Productive Safety Net Program. Report No. 48633-ET.
Washington, DC: World Bank.
———. 2010. Designing and Implementing a Rural Safety Net in a Low Income Setting:
Lessons Learned from Ethiopia’s Productive Safety Net Program 2005–2009. Washington,
DC: World Bank.

214
Annex
Annex A. Tag System for Direct Quotes
Region
Woreda
SNNPR [SN]
Shebedino [S]
Tembaro [T]
Demba Gofa [D]
Amhara [AM]
Ebenat [E]
Sayint [SA]
Sokota [SO]
Oromyia [ORO]
Zeway Dugda [Z]
Gursum [G]
Tigray [TIG]
Ahiferom [A]
Saesi Tsaedamba [S]
Method
Respondent
R-KI-1 or R-FG-1
Regional key informant or focus group FSP committee
R-KI-2 or R-FG-2
Regional Special committee (EW, HABP, or PW—specify)
W-KI-1 or W-FG-1
Woreda key informant or focus group FSTF
W-KI-2 or W-FG-2
Woreda special committee (EW, HABP, PW—specify)
K-KI-1 or K-FG-1
Kebele FSTF
K-KI-2 or K-KI-2
Kebele DA interview (KI or FG)
K-KI-3
Other kebele level key informant interview
FG-1
Direct Support Focus Group
FG-2
Transitioning Focus Group
FG-3
Women’s Focus Group
FG-4
Men’s Focus Group
Source: Authors’ compilation.
Notes: DA (development agent), EW (extension worker), FG (focus group), FSP (Food Security Program), FSTF (Food Security
Task Force), HABP (Household Asset Building Programme), KI (key informant), PW (public works).
Examples of how these tags are used in the report are as follows:
A quote from the transitioning focus group discussion from Gursum is referenced as
[ORO_G/FG-2].
An interview with one development agent in Tembaro is referenced as [SN_T/K-KI-1].

215
Annex B. Characteristics of Woredas Surveyed Using Qualitative Methods
B.1. Introduction
This annex provides a description of the livelihoods found in woredas surveyed using
qualitative methods. It also gives their population size and the proportion of households
enrolled in the PSNP or having access to the HABP or OFSP.
B.2. Livelihood Profiles
The livelihood zones and the major crops and livestock produced are shown in
Annex Table B.1. Most woredas are characterized by a mixed farming system in which crop
and livestock husbandry goes together. Climate-related risks such as drought, erratic rainfall,
and hailstorms are common. The severe drought crisis of 2009 affecting the Greater Horn of
Africa was felt in some of the woredas covered by this evaluation. Crop pests, livestock
diseases, and malaria were also found to be most important sources of livelihood risks.
Moreover, cash crop growing woredas, mainly coffee growing SNNPR woredas, were
identified as areas affected by market shocks due to reduced terms of trade between coffee
and food crops during 2007–2009. In Annex Table B.1 and the discussion that follows,
livelihoods are based on characterizations provided by the Livelihood Integration Unit,
Disaster Management and Food Security Sector, MOARD (July 2010 for Amhara and
Oromiya; January 2009 for Tigray and SNNPR).

216
Annex Table B.1. Livelihood zones, strategies, and risks in the sample woredas
Woreda
Livelihood zone
Major crops grown
Dominant livestock
type
Livelihood risks/shocks
Ziway Dugda
(Oromiya)
Rift Valley Maize and
Haricot bean (RVM)
Maize, haricot beans,
teff
Cattle, goats, sheep,
donkeys, and
chickens
Drought, crop pests and
diseases, flooding, and
livestock disease
Gursum
(Oromiya)
Gursum Babile
Groundnut (GBG)
Sorghum, maize,
groundnut, chat
Cattle, shoats
Shortage of moisture
North East Agro-
pastoral (NAP)
Sorghum, maize,
groundnut
Camels, cattle, and
shoats
Shortage of moisture
Sorghum and Maize
and Chat (SMC)
Sorghum, maize, chat
Cattle, shoats
Shortage of moisture
Ahferom
(Tigray)
Central Mixed Crop
(CMC)
Sorghum, teff, hanfets,
finger millet
Cattle and shoats
Erratic rainfall
Gesho and Wheat,
Highland (GWH)
Gesho, wheat, hanfets
Cattle and shoats
Erratic rainfall
Mereb Basin (MRB)
Sorghum, finger millet,
maize, teff
Cattle and shoats
Environmental degradation
Saesi Tsaeda
Emba
(Tigray)
Atsbi Womberta
Highland (AWH)
Barley, wheat, pulses
Cattle and shoats
Erratic rainfall
Eastern Plateau
Livelihood (EPL)
Barley, hanfets, wheat,
maize
Cattle and shoats
Erratic rainfall
Irob Mountain (IRM)
Barley, wheat, maize,
pulses
Cattle and shoats
Erratic rainfall
Ebenat and
Sekota
(both are in
Amhara)
North East
Woynadega Mixed
Cereal (NMC)
Barely, wheat, sorghum,
teff
Environmental degradation and
erratic rains
Tekeze Lowland
Sorghum and Goat
(TSG)
Sorghum, teff, haricot
beans
Environmental degradation and
erratic rains
Sayint
(Amhara)
Abay Beshlo River
Basin West (ABB)
Teff, sorghum, maize
Cattle, goat, and
sheep
Drought, hailstorm, and
landslide
South Wollo Meher
(SME)
Wheat, teff, barley, red
sorghum (zengada)
Cattle, goat, and
sheep
Drought, hailstorm, and
landslide
Demba Gofa
(SNNPR)
Gamo Gofa Enset and
Barley (GGE)
Enset, barley, wheat,
sweet potato
Goats and poultry
Erratic rainfall
Gamo Gofa Maize
and Root Crops
(GMR)
Maize, enset, sweet
potato, taro, teff
Goats and poultry
Erratic rainfall
Shebedino
(SNNPR)
Sidama Coffee (SCO)
Enset, coffee, maize,
sorghum
Cattle
Drought, hailstorms and frost,
crop diseases, market shock
(fluctuation in crop price,
particularly coffee)
Sidama-Gedeo
Highland Enset and
Barley (SEB)
Enset, barley, wheat,
horsebeans, peas
Cattle
Drought, hailstorms and frost,
crop diseases, market shock
(fluctuation in crop price,
particularly coffee)
Tambaro
(SNNPR)
Hadero Ginger
Livelihood Zone
(HGZ)
Maize, haricot beans,
sweet potato
Cattle and goats
Drought, livestock disease, and
malaria
Hadiya-Kembata
Cereal and Enset
(HWE)
Enset, wheat, potatoes,
baley
Cattle and goats
Drought, livestock disease, and
malaria
Source: Authors’ compilation.
More detailed descriptions are given below.

217
Ziway Dugda Woreda
Ziway Dugda woreda lies in the Rift Valley Maize and Haricot Bean (RVM) livelihood zone
(LZ). The woreda is located in central parts of Oromiya and lies in the central rift valley. The
agroecology is mostly midlands or woinadega with some lowland or kolla areas. The
topography includes hills, plains, and undulating landscapes with the rift valley escarpment.
The major types of vegetation are bush scrub and grasslands.
The main crops grown are maize, haricot beans, and teff, all of which are both for
consumption and sales. Although the woreda is a crop producing area, livestock is also very
important to livelihoods and is a major source of income. The main livestock kept are cattle,
goats, sheep, donkeys, and chickens. Milking cattle and goats provide milk as food and to
generate income. Sale of livestock (cattle and shoats), and livestock products, such as
butter, skins, and eggs, are sources of cash income. Poorer households sell their labor in
rural areas and towns. Firewood collection and selling is also quite an important income
generating activity for poor households. The road access in the woreda is good and includes
many large trading centers. The main hazards are drought (failure, delay, and/or erratic
rainfall), crop pests and disease, flooding, and livestock disease. Drought occurs about once
every five years, affecting not only crop production, but also leading to a decline in water and
pasture availability, and thus the physical deterioration of livestock.
Gursum Woreda
Gursum woreda straddles three livelihood zones, namely Gursum and Babile Groundnut
(GBG), North East Agro-pastoral (NAP), and Sorghum, and Maize and Chat (SMC). Parts of
Gursum woreda found in GBG and NAP LZs encompass areas that share a boarder with the
Somali region to the east. Agroecologically, Gursum is dominantly kola or lowlands with
some woinadega or midland areas.
In all the three livelihood zones, sorghum and maize are the major food crops grown and, in
addition to these, groundnut is produced as a cash crop in GBG and NAP livelihood zones.
Besides groundnut, chat is also produced as a cash crop in the woreda with the exception of
NAP LZ. The major livestock kept in the woreda include cattle, sheep, and goats, while
camels are also important in NAP LZ. However, the very poor households do not own
livestock. Those households who have livestock sell them whenever cash is needed to cover
some of their expenses and food needs, although a large source of annual income for the
households comes from sale of chat and groundnuts where they are grown. Poorer
households also generate income from the collection of firewood, agricultural labor, and
through participation in the Safety Net program. Market access is considered good as most
parts of the woreda, with the exception of NAP LZ, have adequate access to all-weather
roads and good access to markets. Market access in NAP LZ is moderate. In the woreda,
moisture shortage either due to inadequate rainfall (in terms of total volume) or in the
distribution of showers over the growing season is a major chronic hazard, which may result
in loss of overall crop production.

218
Ahferom Woreda
Ahferom woreda encompasses three livelihood zones, namely Central Mixed Crop (CMC),
Gesho and Wheat Highland (GWH), and Mereb Basin (MRB). MRB LZ is predominantly a
kolla lowland agroecological zone, with undulating terrain; CMC is a midland ecology
characterized by undulating hills and imposing mountains alternating with plains; and GWH
is a mountainous area in a dega agroecology zone.
In Ahferom woreda, the production system is mixed farming, with both crop and livestock
production. In CMC and MRB livelihood zones, the major crops produced are sorghum,
finger millet, maize, and teff, but with varying order of importance. In CMC LZ, hanfets (a
crop combining properties of barley and wheat) is also an important crop. Hanfets is
cultivated to mitigate the risk of drought, because it is relatively drought resistant. Teff,
hanfets, and finger millet are the most important food crop in CMC, while the staple food
crops in MRB LZ are sorghum and finger millet. The main crops cultivated in GWH LZ are
gesho (equivalent to hops) and wheat. Gesho is a perennial shrub whose leaves are used
for the preparation of local beer, and it is a valuable cash crop. Wheat is the staple food crop
in GWH LZ. In the woreda, the main livestock kept are cattle, sheep, and goats. Livestock
are important for providing draught power, income, and food. The woreda faces a chronic
food insecurity problem primarily because land available for cultivation is limited and cannot
support the dense population residing there. The sale of agricultural labor, both locally and
outmigration, is an important source of income for the very poor and poor households. Poor
households also rely heavily on food from the PSNP program to supplement what can be
grown and purchased. The condition of road access in the woreda varies from place to place
and is rated well in MRB LZ, moderate in CMC, and poor in GWH LZ. Poor road conditions
in GWH LZ are a significant impediment to marketing activities and there are no local
markets. Commodities are transported to and from the market on donkeys. Erratic rainfall is
the biggest natural hazard in CMC and GWH LZs, which occurs once every three years. In
MRB LZ, land degradation arising from deforestation and poor land management practices
poses a major threat to food security.
Saesi Tsaeda Emba Woreda
Saesi Tsaeda Emba woreda has three livelihood zones: Atsbi Wonberta Highland (AWH),
Eastern Plateau Livelihood (EPL), and Irob Mountain (IRM). Most of Saesi Tsaeda Emba
woreda is an area with heavily deforested plains, and the remaining vegetation is
predominantly scattered bush and acacia trees. The landscape in most parts of the woreda
is mainly rugged mountains and hills spanning the dega, woina dega, and kolla altitudes.
Mixed farming is the major livelihood activity in the woreda, where crop production is more
important in AWH and EPL livelihood zones while livestock rearing is the main economic
activity in IRM LZ. Agriculture is dependent on kiremt rains. In the woreda, barley and wheat
are the major crops produced, followed by maize and pulses. Hanfets (a mixture of wheat
and barley) is also one of the important crops produced in EPL LZ. In IRM and EPL LZs, wild
cacti that grow on private plots are also harvested for household consumption. The main
livestock reared in the woreda are cattle, sheep, and goats. Cattle are the more valuable
livestock as a source of draught power, food, and cash income. In much of the woreda
(AWH and EPL LZs), wage labor is an important economic activity and even during normal

219
years, there is seasonal migration to Afar and Humera. Afar provides opportunities for
participation in the salt trade. Humera is a source of agriculture labor on sesame farms
during the weeding and harvesting seasons. The Productive Safety Net Programme (PSNP)
is a source of additional income for the poor. The condition of market access in the woreda
varies from place to place, ranging from good access to no access at all. Population
pressure and erratic rainfall combined with shortage of suitable land for crop cultivation
contribute to make this woreda a food-insecure area.
Ebenat and Sekota Woredas
Ebenat and Sekota woredas have two livelihood zones in common, namely North East
Woynadega Mixed Cereal (NMC) and Tekeze Lowland Sorghum and Goat (TSG).
The main livelihood strategy of the two woredas comprising these two livelihood zones is
mixed farming with smallholder crop production, livestock rearing, and paid agricultural labor.
Agricultural performance is generally poor and this is attributable to environmental
degradation and erratic rains. They are a chronically food deficit areas. In NMC livelihood
zone, of the two woredas, barley, wheat, sorghum, and teff are the main crops cultivated for
consumption, while faba bean, lentils, and oil seeds are grown in these areas for sale. In
TSG livelihood zone of the two woredas, the main crops cultivated are sorghum, teff, and
haricot beans, and the cereals are often quick varieties adapted to the rainfall conditions,
including the lower-yielding bunyi teff. The cereals and pulses are used for household
consumption, and sesame and Niger seed are grown for cash. Generally, market access in
these two livelihood zones of the two woredas is poor; this is particularly due to the poor
road network and rugged topography.
Sekota falls into NMC and TSG livelihood zones, while Ebenat woreda lies in the Tana Zuria
(TZA) livelihood zone. TZA LZ of Ebenat woreda is situated around the biggest Ethiopian
lake, Lake Tana. Here too, mixed farming is the major economic activity. The major crops
produced in this part of Ebenat woreda are maize and finger millet. As opposed to NMC and
TSG livelihood zones of Ebenat and Sekota woredas, here the market access is rated to be
good. The good market access is attributed to the good road conditions in the area and the
proximity of the area to the major urban centers, including the regional capital, Bahir Dar.
Sayint Woreda
Sayint woreda consists of two livelihood zones, namely Abay Beshlo River Basin West
(ABB) and South Wollo Meher (SME). Agroecologically, the part of Sayint woreda with ABB
LZ is predominantly kola with few peripheral parts characterizing woinadega, while SME LZ
consists of both dega and woina dega agroecologies.
The major economic activity of the woreda in the two livelihood zones is cereals crop
production supplemented by livestock rearing. Teff, sorghum, and maize are the major crops
produced in the ABB LZ, while wheat, teff, barley, red sorghum (zengada), and pulses are
the main crops grown in SME LZ. The part of the woreda with SME LZ is a chronic food
deficit area unlike the ABB LZ of the woreda where there is no history of food aid in the past
years. The soil is of a degraded nature in many parts and loses its potential for crop
production, although there are some parts with moderate soil fertility. Cattle, goat, and sheep

220
are the prominent livestock types commonly reared. Sale of charcoal and firewood and also
migratory labor are the major sources of cash income in the woreda apart from the crop and
livestock production. Much of the woreda is inaccessible. There are few roads across this
rugged topography and the woreda lies quite far from the main highways and big towns.
Drought, hailstorm, and landslide are the major hazards of the woreda and copping
strategies are very wealth-specific.
Demaba Gofa Woreda
Demba Gofa woreda has two livelihood zones, Gamo Gofa Enset and Barley (GGE) and
Gamo-Gofa Maize and Root Crop (GMR). The area that is found in GGE LZ is mountainous
and includes wet woina dega and dega agroecological zones. The ecology in GMR is
midland (woina dega) and upper lowland, with a hilly or undulating topography. Most of the
rural population in GGE LZ is self-sufficient in food, while the part of Demba Gofa woreda
that is found in GMR LZ is highly food-insecure.
The major crops produced in the chronically food-insecure part of the woreda are maize,
enset, sweet potatoes, taro, teff, and yams. The combination of cereals and perennial/root
crops offers some insurance against at least moderate rain failure, since maize is more
susceptible than either root crops or enset to shortage of rain. Lack of grazing and fodder
affects oxen production, so that only the better off and middle wealth group households,
which own plow-oxen, are able to till the land efficiently. Most households possess goats and
poultry, but livestock numbers are modest among all households. Market accessibility is
generally poor in the woreda due to the poor state of the roads, most of which are only
suitable for dry-weather transportation and are crossed by seasonal rivers. Better-off
households use horses, mules, and donkeys for transport, but seasonal rivers often cannot
be crossed during the rainy season and it is difficult to get to markets. Apart from the state of
the roads, most areas of the woreda are distant from major urban markets and major
transport routes in the region. As a result, the prices of the goods that households sell tend
to be low and the prices of the goods that they purchase tend to be high. Erratic rainfall is a
major risk factor affecting livelihoods in the woreda. It is a periodic problem and can include
a late-start and/or an uneven distribution. A late start to the belg rains is especially
significant, resulting in an extended and more severe hunger season than usual by delaying
the green maize harvest.
Shebedino Woreda
Shebedino woreda falls in Sidama Coffee (SCO) and Sidama-Gedeo Highland Enset and
Barley (SEB) livelihood zones. The part of Shebedino under SEB LZ is relatively food
secure, with no history of food aid distributions.
The area that is found in SCO LZ is a relatively productive midland area that attracts migrant
laborers from nearby highland areas during the busy coffee picking season. The landscape
is characterized by undulating hills and, due to the high population density, most of the land
is cultivated. This is a visibly green part of SNNPR, with eucalyptus, fruit, and coffee trees
prominent throughout the area and enset growing around every household. However, there
is no natural forest and very limited communal grazing land. Coffee is the main cash crop
and enset is the main food crop, and these are supplemented by small quantities of other

221
rainfed food crops (including maize, sorghum, haricot beans, yams, taro, and sweet
potatoes) and fruits (including avocado and pineapple). Annual food crops are generally
intercropped among the coffee and enset plants. As a result, plough oxen are rarely used for
cultivation in this part of the woreda; most cultivation is done by hand. Cattle are the most
important type of livestock in this area. Grazing land is in short supply, however, so cattle are
generally raised using a “zero-grazing” system. Labor migration is relatively uncommon, but
poorer households do resort to this income-generating option in bad years. In normal years,
poor households find casual work locally, including agricultural work for better-off farmers
and daily labor in the pulping stations during the coffee harvest season. Market access is
generally good in this part of the woreda and major urban markets for crops and livestock
are nearby. Due to small landholding sizes and the large proportion of land that is dedicated
to coffee production, most households do not produce enough food crops to last throughout
the year, even in a year of good crop production. Market reliance is therefore quite high in
this area, suggesting that both cash crop and staple food prices should be closely monitored.
The main hazards affecting the woreda are shortage of rain and drought, hailstorms and
frost, crop diseases, fluctuating coffee production and prices, and increase in prices of staple
foods.
Tambaro Woreda
Tambaro woreda consists of two distinct livelihood zones, Hadero Ginger livelihood zone
(HGZ) and Hadiya-Kembata Cereal and Enset (HWE). Part of the Tambaro woreda that is
found in HWE LZ has historically been self-sufficient in terms of crop production and
households are generally food-secure and thus this section describes the livelihood situation
of HGZ LZ, which is chronically food-insecure. The area that falls in HGZ LZ consists of
rugged terrain and vast areas of unproductive land that do not support the cultivation of
crops due to poor soil. Agroecologically, this area stretches from kolla (lowland) to woina
dega (midland).
The major food crops grown are maize, haricot beans, and sweet potato. In recent years,
there has been a gradual shift of emphasis from cereal crop production to root crop
production (i.e., from maize to sweet potatoes), to the extent that sweet potatoes are now
the most important food crop. The major income earners for households in this area are
ginger and coffee. However, the continuous dry spell is affecting coffee production and
farmers are increasingly relying on ginger production for their cash income. Although ginger
is susceptible to wide price fluctuations, it is drought resistant and a significant income
earner in most years. Self-employment and labor migration are additional sources of income
for very poor and poor households. Self-employment includes timber, grass, and firewood
sales. Cattle and goats are the main livestock types reared in this area. There is a form of
arrangement for sharing cattle (and sometimes goats) whereby poor households care for the
livestock of the rich in return for the skimmed milk and a share of the offspring. As a result of
this type of agreement, all households in the area keep cattle. Despite poor feeder roads
within the area, market access is good due to the geographical location of the area, which is
proximate to major roads and market centers. The major natural hazards experienced by
households in the ginger zone are drought, livestock disease, and malaria. Drought is a
recurring problem in recent years, resulting in crop failure and loss of livestock assets.

222
B.3. Population and Participation in the PSNP and HABP/OFSP
The population of the ten woredas is estimated at 1.7 million (Annex Table B.2). Approximately 28 percent—445,328 people—are PSNP
beneficiaries. The male-to-female proportions of the PSNP beneficiary populations are roughly equivalent. About 47 percent of PSNP
beneficiary households are led by women.
Annex Table B.2. Woreda population, household size, and number of PSNP and HABP/OFSP beneficiary households
Region
Woreda
Total
population
Total
households
PSNP beneficiary
PSNP/OFSP beneficiary
Non-PSNP
population
Female
Male
Total
Female-
headed
households
Male-
headed
households
Female-
headed
households
Male-
headed
households
Amhara
Ebinate
259,053
51,688
40,763
36,855
77,618
10,107
3,626
1,748
3,168
181,435
Sayinte
152,091
36,881
27,053
28,392
55,445
7,474
13,018
139
1,446
96,646
Sekota
112,259
22,452
15,315
25,215
40,530
6,565
6,857
132
505
71,729
Oromiya
Gursum
211,705
37,338
21,653
22,639
44,292
3,314
6,336
1,432
4,447
167,413
Zeway Dugrad
136,286
17,933
5,400
14,590
19,990
375
500
318
842
116,296
SNNPR
Demba Goffa
88,425
17,685
11,715
11,144
22,859
2,228
4,696
754
2,262
65,566
Shebedino
239,881
40,796
6,738
7,048
13,786
507
1,834
317
1,463
226,095
Tembaro
115,999
23,199
12,383
11,827
24,210
2,367
2,975
91,789
Tigray
Aferom
194,488
36,524
40,783
32,516
73,299
5,657
10,825
8,362
21,272
121,189
Saesi Tsedamba
153,112
53,159
40,783
32,516
73,299
32,364
30,842
5,207
14,349
79,813
Total
1,663,299
337,655
222,586
222,742
445,328
70,958
81,509
18,409
49,754
1,217,971
Source: Woreda Office of Agriculture and Rural Development of Assessment woredas.