Content uploaded by Furaha Rashid
Author content
All content in this area was uploaded by Furaha Rashid on May 06, 2020
Content may be subject to copyright.
International Journal of Food and Agricultural Economics
ISSN 2147-8988, E-ISSN: 2149-3766
Vol. 8, No. 1, 2020, pp. 79-95
79
COMMERCIALISATION PATHWAYS: IMPLICATIONS ON
SMALLHOLDER RICE FARMERS’ PRODUCTIVITY AND
WELFARE IN MBARALI DISTRICT, TANZANIA
Furaha Ndakije Rashid
Sokoine University of Agriculture, Department of Agricultural Economics and
Agribusiness, P.O Box 3007, Morogoro, Tanzania, Email: rashidfuraha.@gmail.com
Roselyne Alphonce
Sokoine University of Agriculture, Department of Agricultural, Food and
Resource Economics, Tanzania
Isaac Joseph Minde
Michigan State University, Department of Agricultural, Food and Resource
Economics, USA
Abstract
This study aimed at evaluating the most effective commercialisation pathway (smallholder
and inclusive) and its impacts on productivity and welfare on smallholder rice farmers in the
pathways versus rain-fed farmers in Mbarali District. Output and input commercialisation
indices (CCI and ICI) and propensity score matching were used for data analysis. The overall
output commercialisation was more than half of the produced rice (CCI=59%) but the use of
improved inputs in the study area was low (ICI = 27%). The proportion of rice sold was higher
in the inclusive pathway (80%) relative to smallholder pathway (70%) and rain-fed scheme
(41%). Total factor productivity ranged between 1.17 - 1.21 and 0.98 – 1.02 in the smallholder
and inclusive pathways respectively more than that in the rain-fed scheme. In terms of welfare,
inclusive pathway was better-off relative to the two groups. Therefore, both smallholder and
inclusive pathways should be adopted to explore the synergies.
Keywords: Commercialisation pathways, Productivity, Welfare, Smallholder, Propensity
Score Matching.
JEL Codes: D24, O12, O21, I31, Q13, Q18
1. Introduction
Transforming the agricultural sector from low productivity to high productivity
commercialisation has been a critical policy of concern in most of Sub Saharan African (SSA)
countries. Agricultural sector plays a critical role in SSA since more than 70% of the
population (about 904 million) live in rural areas, 43% live under poverty line of US $ 1.90
per day, 22% of the population are food insecure and over 75% of the poor are smallholders
whose primary source of livelihood is agriculture (IFAD, 2012; Zhou et al., 2013; World Bank,
2016). In Tanzania, 81% of the population live in rural areas where 31.3% are under poverty
line relative to 15.8% in urban areas (MoFP-PED, 2019). The World Economic Forum (2015)
estimated that, growth generated from agriculture is 2- 4 times more effective at reducing
poverty than the growth in other sectors in SSA.
Commercialisation Pathways: Implications on Smallholder Rice Farmers …
80
In Tanzania, agriculture is dominated by smallholder farmers where 91% of cultivated
farms are considered to be small scale (up to 5ha) and account for about 80% of food
production (Jayne et al., 2016; URT, 2016). They are characterized by rain fed agriculture
(95%), low use of improved inputs specifically fertilizer where farmers apply only 7 – 9kg of
nutrients /ha which is very low relative to Malawi (27 kg), South Africa (53 kg) and China
(279 kg) and far less than the 2006 Abuja declaration commitment of increasing fertilizer use
to at least 50 kg nutrients/ha (Masso et al., 2017; URT, 2016; Nkuba et al., 2016). This in turn
has led to low productivity particularly in staple foods including rice whose average
productivity is 2 tons/ha relative to global average of 4.3 tons/ha (Zhou et al., 2013). As in
other parts of SSA, Tanzania has encountered low improvement of cereal crops in terms of
productivity.
To overcome this poor performance of the sector, transformation of the agricultural sector
from smallholder subsistence to commercial oriented agriculture is inevitable. Agricultural
commercialisation is process by which subsistence/semi-subsistence oriented production is
transformed into market oriented production through increased productivity and greater
surplus that enhance the rise in output and input markets participation based on principles of
profit maximisation. (Von Braun, 1995; Pingali & Rosegrant, 1995). There are different
pathways that agricultural commercialisation can take. Different scholars (Newsham et al.,
2018; Oya, 2012; Jayne et al., 2014) have identified four pathways of commercialisation: (i)
Estate/plantation/large-scale commercial farming involving large land holding, growing a
single cash crop, involves high mechanisation and relies on hired labour. (ii) Out
grower/contract farming – farmers produce and sell output to a buyer based on pre-agreed
arrangements. (iii) Medium-scale commercial agriculture – farmers with land holding ranging
from 5 – 100 ha and (iv) Smallholder pathway- owning less than 2 ha, rural, depend on family
labour and sell surplus. In view of these pathways, during 2010, an initiative called the
Southern Agricultural Growth Corridor of Tanzania (SAGCOT) was established as a pathway
to commercialize agriculture in Tanzania and further fuel KILIMO KWANZA initiative
(SAGCOT, 2013; Herrmann, 2017). This is an inclusive smallholder and medium/ large scale
agribusiness model that aimed at increasing smallholders’ productivity, income and welfare
through adoption of modern technology (high-yielding seeds, fertilizer, machinery and good
agronomic practices) and markets by linking them with medium/large scale commercial
farmers (SAGCOT, 2013; West & Haug, 2017). Despite the efforts made by the government
of Tanzania to commercialize agriculture through the establishment of the SAGCOT, crop
productivity particularly for rice is still low on average at 2 tons/ha and 26.4% of rural farmers
are still faced by basic needs poverty (URT, 2016; MoFP-PED, 2019). Previous studies have
mostly focused on determinants, levels, processes and outcomes of agricultural
commercialisation on employment, profitability, income and nutrition (Von Braun, 1995;
Hailua et al., 2015; Okezie et al., 2012; Mitiku, 2014).
Moreover, no study took into consideration on which commercialisation pathway should
be adopted particularly in Tanzania. This study therefore aimed at addressing this gap by
providing the empirical evidence that will inform policy in Tanzania on the pathway that
should be adopted to shape smallholder commercialisation and livelihood. The study specific
objectives were:-
i. To determine the extent of smallholder rice commercialisation in the area.
ii. To evaluate the impacts of smallholder and inclusive commercialisation pathways on
productivity of smallholder rice farmers in the pathways versus rain-fed farmers.
iii. To evaluate the impacts of smallholder and inclusive commercialisation pathways on
the welfare of smallholder rice farmers in the pathways versus rain-fed farmers.
The next sections are arranged as follows. Section two gives an analysis of literature
relevant to this study, section three comprises the methodology that was used in this study.
F. N. Rashid, R. Alphonce and I. J. Minde
81
Section four gives the results and discussion while section five consists of conclusion and
policy recommendations.
2. Commercialisation Pathways
Different regions in the World have adopted different models towards agricultural
commercialisation. Some including the south East Asian region transformed the agrarian
sector through smallholder commercialisation during the period of green revolution in 1960s
(Asfaw et al., 2012) while other countries including Brazil have succeeded through investing
in large scale commercial farming. Other commercialisation pathways that have been widely
adopted particularly in SSA include out grower/contract farming and currently through the rise
of medium scale commercial farming (Leavy & Poulton, 2007).
Unlike South East Asia where smallholders stand as vital route in the transformation
process brought by their efficiency in the use of resources including family labour,
intensification and production mediated to the market (Rigg et al., 2016), efforts that have
been devoted to transform smallholder agriculture in SSA including the CAADP and in
particular Kilimo Kwanza “Agriculture First” initiative in Tanzania have not led to expected
impacts and the region still realizes production below its potential (Byerlee & Deininger,
2013). This can be explained by agronomic factors (low use of inputs, climate change, low
tech-know how) and policy issues that are not mediated towards the market (market controls)
and lack of clear focus to smallholder farmers in sub Saharan Africa (Van Donge et al., 2012).
It is argued that large scale agricultural investments-plantations model (LSAIs) should then
be the priority for poverty alleviation and economic growth (Otsuka, 2013; Oya, 2012; Henley,
2012; Bellemare, 2012), since they are the source of technology transfer, employment and
produce towards the markets (Kleemann, 2015). However, it is also argued that, plantations
give low wage employment to few people while leaving the majority, affect local food
production through dispossession of land from smallholders as well as diverting labour from
smallholders. For example, Adewumi (2013) argued that, there was an efficiency level of 90%
on the frontier by farmers within inclusive investments as compared to less than 50% by
farmers outside the investment after adoption of new technology in Nigeria. Contrary to these
positives of the model, the model is also argued to lead to social differentiation by including
only top tiers of smallholders endowed with resources, leads also to food insecurity due to over
reliance on cash crops rather than food crops (Wang et al., 2014). The current available
literature is still inconclusive on which model is suitable in bringing about the expected
outcomes.
2.1 Rice production in Tanzania
In Tanzania, rice is the second most important food grain after maize and is a priority crop
in the second Agricultural sector development Program (ASDP II), and in the Southern
Agricultural Growth Corridor of Tanzania (SAGCOT), with annual consumption per capita of
25.4 kg (RCT, 2015). The rice sector is 90% dominated by smallholder farmers who produce
both local varieties and it employs about 1.5 – 2 million people in the country (RCT, 2015).
About 25% of total rice produced in Tanzania comes from Mbeya (Mbarali and Kyela
districts) and Morogoro regions. Despite the potential for rice production in the country due to
availability of irrigable land (29.4 million hectares), the sector is still faced with various
challenges including over dependence on rainfall, inadequate financing and low productivity
which is averaged at 2 ton/ha and the noted increase in production has been attributed to area
expansion (RCT, 2015; URT, 2016) as shown in Figure 1.
Commercialisation Pathways: Implications on Smallholder Rice Farmers …
82
Source: FAOSTAT, 2017
Figure 1. Rice production, harvested area and productivity trend in Tanzania (2008 -
2016)
3. Methodology
3.1 Study Area
The study was carried out in Madibira scheme – a smallholder commercialisation pathway
in Madibira ward, Kapunga scheme – an inclusive smallholder-Medium/large scale pathway
and the rain-fed smallholder farmers located in Itamboleo ward of Mbarali district. The district
lies within the Usangu basin which is potential for rice production and is characterized by
extensive irrigated rice schemes consisting of both smallholder and medium/large scale
farmers. The economy of Mbarali district depends on agriculture sector which employs over
80% of the inhabitants in the district.
3.2 Research Design, sampling and data collection
The study used quasi-experimental design, utilizing cross sectional survey data collected
from rice farmers in Mbarali District in May/June 2018. Since participating in the smallholder
and medium/large scale commercialisation is voluntary, random allocation to treatment or
control group is not possible (White & Sabarwal, 2014). A comparison group that is similar as
possible to the treatment group in terms of baseline characteristics was established. The
comparison group captures what would have been the outcome if the program had not been
implemented (Caliendo & Kopeinig 2005). Thus, three groups consisting rain-fed rice farmers
was used as a baseline (treatment 1), rice farmers participating in the smallholder pathway
(treatment 2) and smallholder rice farmers in Kapunga inclusive pathway (treatment 3) were
evaluated using both ANOVA and Econometric analysis. The study used two stage probability
sampling. At first, a random selection of wards producing rice were selected. In this stage, a
list of rice farmers participating in both smallholder and inclusive commercialisation pathways
were established from the Madibira and Kapunga Agricultural and Marketing Cooperatives
respectively. A list of rain-fed farmers found in Itamboleo ward was also established. Then,
0,0
0,5
1,0
1,5
2,0
2,5
3,0
0
500
1000
1500
2000
2500
3000
3500
2008 2009 2010 2011 2012 2013 2014 2015 2016
Productivity(ton/ha)
Area
harvested(ha),production(ton)
Year
"000"
Area harvested (ha) Production(tons) Yield (Ton/ha)
F. N. Rashid, R. Alphonce and I. J. Minde
83
proportionate probability sampling was established based on strata that were identified. A
total sample of 256 smallholder farmers were interviewed in this study of which 90 were
farmers participating in the smallholder pathway, 110 were rain-fed farmers and 56 were
farmers in the Kapunga inclusive commercialisation pathway.
3.3 Analytical Framework
Objective 1: To Determine the Extent of Smallholder Rice Commercialisation in the Area
Following Von Braun (1995), this objective was addressed by the use of both crop (rice)
output and input commercialisation indices (CCI and ICI) as described below;-
≥; 0 ≤ CCI ≤ 100 (1)
Where = Crop (rice) commercialisation index of house hold growing rice, =
Value of rice sold in monetary terms and is the monetary value of total quantity of total
rice produced. Similarly, from input side, input commercialisation index (ICI) is given by;
≥; 0 ≤ ICI ≤ 100 (2)
is the gross value of crop inputs acquired from the market and is the gross value of
total rice produced. With reference to the work by FAO (1989), households whose
≥50% are commercial oriented, 25%≤ <50% are in transition and those with < 25%
are subsistence oriented.
Objective 2: To Evaluate the Impacts of Smallholder and Inclusive Commercialisation
Pathways on Productivity of Smallholder Rice Farmers in the pathways versus Rain-fed
Farmers in the Study Area.
The decision to participate in either of the commercialisation pathway is modelled using
random utility framework (Becerril & Abdulai, 2010).
With
(3)
Where is a latent binary variable for participation, = 1 if a household participated in
either of the pathway and if the household did not participate. The conventional
approach commonly used to measure the impacts of an intervention in this case, participation
in either of the commercialisation pathway on smallholder rice farmers’ productivity and
welfare would be through the use of an Ordinary least square (OLS) comprising of a dummy
variable given by;-
(4)
Where is the average outcome variable of household i, is a vector of household socio-
economic characteristics and is a dummy variable taking the value of 1 for participants and
0 for rain-fed farmers. However, the use of OLS in impact evaluation would yield biased
estimates since the model assumes that participation in an intervention is exogenously
Commercialisation Pathways: Implications on Smallholder Rice Farmers …
84
determined while it is potentially endogenous (Herrmann & Grote, 2015). Assignment into
treatment is not always random, but maybe due to purposive placement into the program or
self –selection. To solve this problem, a counterfactual group was established through the use
of propensity score matching (Caliendo & Kopeinig 2005). Propensity score (P) is the
conditional probability of being assigned to a particular treatment given a vector of observed
covariates Xi to facilitate causal inference (Dehejia & Wahba, 2002). It is given by;-
P (X) = P ( =1|Xij) (5)
Where Z (0, 1) is an indicator for exposure to treatment and is a matrix of covariates
influencing the outcome variable, in this case productivity. Following Rosenbaum and Rubin
(1983), binary logit models were used for both smallholder scheme and Inclusive Scheme
using rain-fed rice farmers as control group to estimate the propensity scores P(X). The logit
model (Gujarati, 2004) can be described as:-
(6)
Where P is the propensity score, = is a matrix of observed values influencing
participation and productivity based on economic theory and literature review, j is the response
category and is the matrix of unobserved random effects. The model can further be specified
as; = α+β1X1+β2X2+β3X3+β4X4+ …+β7X7+δ1D1+δ2D2 + …+ δ5D5+ ε (7)
Where X1 is age of household head, X2 is education level of household head, X3 is the
household size, X4 is the farm size, X5 is the distance to the market place, X6 is off-farm
income, X7 is the net income from rice, and D1, D2, D3, …. D5 are dummies for sex, access to
improved seed, access to extension services, access to market information, and
producer/marketing organisational membership respectively. The response probabilities can
be obtained by equation 8 given by,
(8)
Equation 8 is intrinsically linear because the logit model is linear in. It shows that the
probability of participation in either of the commercialisation pathway P lies between 1 and 0
and they vary non-linearly with. The partial effects for continuous variables to account for
the causal – effect can be calculated using quotient rule as;
(9)
The partial effects for discrete variables was calculated as the difference of mean
probabilities estimated for the respective discrete variable. Then, the covariates in each block
were matched using the nearest neighbour matching since it is a most straight forward
estimator among other estimators. Furthermore, kernel matching estimator was used. The
average treatment effect on the treated (ATT) which is the average difference in outcome
between the matched control and the treated group was then estimated using equation 10, 11
and 12 (Hailua et al., 2015; Rosenbaum and Rubin, 1983). Let Y1 be the productivity when
the household is subject to treatment (Z=1) and Y0 be the same variable when the household
did not receive treatment (Z=0), then the observed productivity outcome can be given by;
F. N. Rashid, R. Alphonce and I. J. Minde
85
Y= Z + (1-Z) (10)
ATT = E ( − | = 1) = E (11)
ATT =E (12)
Where, ATT is the average difference in productivity between smallholder rice farmers
receiving treatment relative to rain-fed, P(X) are the propensity scores, and is an indicator
for treatment which equals 1 if individual received treatment and 0 otherwise. From equation
10, we can only observe the outcome variable of participants E (Y1 | = 1), but we cannot
observe the outcome of participants if they had not participated E (Y0 | = 1).
Objective 3: To Evaluate the Impacts of Smallholder and Inclusive Commercialisation
Pathways on Welfare of Smallholder Rice Farmers in the Pathways versus Rain-fed Farmers
in the Study Area.
Propensity score matching was also used as in objective two. Welfare was measured by the
use of food consumption score (FCS), access to health insurance, value of durable assets
owned like farm implements and income unlike previous studies (Amare et al., 2012; Asfaw
et al., 2012) that have used single measure of welfare. The household food consumption score
was measured by the frequency of food group consumption for the last 7 days before the survey
which reflects on food security and nutritional adequacy. Other measures included household
access to health insurance fund and household total annual income from all sources.
4. Results and Discussion
4.1 Descriptive Results
Table 1 and 2 present the descriptive statistics of socio-economic characteristics of sample
respondents of the three treatments: rain-fed/reference farmers, smallholder commercialisation
pathway and the inclusive (small-medium/large scale) pathway farmers. From the descriptive
statistics in Table 1, the average age of the sample respondents (household heads) was about
44.2 years. The difference in age between the three comparable groups was insignificant
implying that the household heads age was almost the same although respondents in the rain-
fed group had higher average age relative to the respondents in the two groups. The observed
average age implies that most farmers were still in their productive age in the country (15 – 64
years).
The sampled respondents had an average family size of about 5.8. Households in the two
pathways had large family size relative to rain-fed farmers. About 86% of sampled respondents
were males (Table 2). Similarly, about 81.7% of the respondents had formal education. The
literacy rate was higher in the smallholder pathway (91%) followed by respondents in the
inclusive pathway (82.1%) and 73.6% in the rain-fed scheme. The difference in the literacy
rate among the three groups was significant in all the education levels. Education helps farmers
to make informed decisions and respond to market dynamics through the acquired skills and
exposure (Ochieng et al., 2015).
The average cultivated farm size was 2 ha, where farmers in the smallholder and inclusive
schemes cultivated 2.2 ha and 2.5 ha respectively relative to 1.6ha for rain-fed farmers. The
difference was significant implying that smallholder farmers in either of the commercialisation
pathway cultivated larger parcels of land compared to rain-fed farmers partially due to
Commercialisation Pathways: Implications on Smallholder Rice Farmers …
86
mechanisation. An increase in farm size helps farmers to produce surpluses thereby stimulating
higher levels of commercialisation (Martey et al., 2012).
Table 1.Household, Farm and Land Characteristics
Variable
Rain-fed
group
(n=110)
Smallholder
Pathway
(n=90)
Inclusive
Pathway
(n=56)
Total
sample
(N=256)
F
Prob>F
Age of the household head
44.6
43.6
44.1
44.2
0.20
0.82
Family size
5.56
5.92
5.93
5.8
1.15
0.32
Land and farm characteristics
Farm size (ha)
1.60
2.50
2.20
2.00
5.70
0.00***
Distance to input market(km)
3.50
2.70
2.60
3.00
11.33
0.00***
Distance to output market
2.40
2.30
1.50
2.10
13.24
0.00***
Land productivity (t/ha)
1.85
4.31
4.37
3.27
125.6
0.00***
Total factor productivity(tfp)
2.20
2.49
2.03
2.26
1.89
0.15
Output commercialisation Index
0.411
0.69
0.76
0.586
119.1
0.00***
Input commercialisation Index
0.283
0.28
0.23
0.270
2.83
0.06*
Household welfare indicator
Food consumption score
52.4
66.90
63.50
59.90
64.88
0.00***
Value of assets(“0000”Tsh)
264.4
854.50
809.60
591.10
4.57
0.01**
Annual income(“0000”Tsh)
225.30
961.03
881.60
626.22
29.70
0.00***
Notes: *= Significant at 10%; ** = Significant at 5%; ***= Significant at 1%.
Table 2. Social, Institutional, Access and Welfare Variables
Variable
Rain-fed
group
(n=110)
Smallhold
er Pathway
(n=90)
Inclusive
Pathway
(n=56)
Total
sample
(N=256)
χ2
Prob>
χ2
Sex (1=male, 0=female)
92(83.6)
74(82.2)
53(94.6)
219(85.5)
4.9
0.087*
Education level of the household head
No formal education
29(26.4)
8(8.9)
10(17.9)
47(18.4)
10.1
0.01***
Primary education
43(39.1)
67(74.4)
23(41.1)
133(52)
28.2
0.00***
Secondary education
26(23.6)
13(14.4)
19(33.9)
58(22.7)
7.6
0.02**
Tertiary education
12(10.9)
2(2.2)
4(7.1)
18(7)
5.7
0.06*
Access/institutional variables
seed (1=improved, 0=local)
28(25.5)
37(41.1)
22(39.3)
87(34.0)
6.3
0.04**
Access to irrigation(1=yes)
4(3.6)
90(100)
56(100)
150(58.6)
240
0.00***
Access to extension(1=yes)
60 (55)
65(73)
42(75)
167(65.7)
9.8
0.01***
Applied fertilizer(1=yes,0=no)
87(79.1)
85(94.4)
56(100)
228(89.1)
20.8
0.00***
Access to credit(1=yes,0=no)
12(10.9)
52(57.8)
22(39.3)
86(33.6)
49.8
0.00***
Access to market
information(1=yes)
70(63.6)
67(74.4)
35(62.5)
172(67.2)
3.3
0.19
Cooperative member(1=yes)
1(0.9)
65(72.2)
49(87.5)
115(44.9)
154
0.00***
Access to health
insurance(1=yes)
43(39.1)
45(50)
34(60.7)
122(47.7)
7.3
0.03*
Notes: *= Significant at 10%; ** = Significant at 5%; ***= Significant at 1%.; Figures in
parentheses are percentages.
Total factor productivity expressed as returns to factors of production was 2.26 denoting
increasing returns to scale. The average land productivity was 3.3 ton/ha where land
productivity among farmers participating in the smallholder and inclusive schemes averaged
F. N. Rashid, R. Alphonce and I. J. Minde
87
at 4.3 ton/ha and 4.4 ton/ha respectively while for rain-fed farmers averaged at 1.86 ton/ha.
The difference in distance to the nearest input market was found to be statistically significant
implying that differences existed between the three groups. On the output market, the average
distance to the nearest market was two kilometre from the farmer’s residence. As distance to
the nearest input or output market increases, market participation decreases (Hailua et al.,
2015). About 66% of rice farmers had access to extension services, Furthermore, of the
sampled farmers (Table 2), only 34% had access to improved seeds.
For irrigation facilities, about 59% of the sampled respondents had access to irrigation
facilities. Similarly, 89.1% of the respondents applied fertilizer in their rice fields. About 33%
of the respondents had access to agricultural credits. Credit could be used to purchase inputs
that may lead to an increase in productivity and thereby generate surplus production (Martey
et al., 2012). About 67% of farmers in the study area had access to market information and it
was not statistically different between the three groups. Furthermore, 45% of sampled farmers
were cooperative members. Cooperatives/associations act as social networks in which farmers
can have access to information related to production and marketing as well as social capital
formation (Martey et al., 2012; Camara, 2017).
Based on the food consumption score (FCS), the sampled households were food secure
following World Food Program classification (WFP, 2008). However, the level of frequency
of consumption of the food groups was higher (66.9 and 63.5) among farmers in the two
pathways compared to those in the rain-fed group whose FCS was approximately 52.4. On the
pattern of consumption, grains constituted the largest share of all food groups consumed in a
week compared to other food groups in all the three comparison groups. On average, a typical
household consumed food grains for six days a week. The food grains was supplemented by
vegetables, oil/fats, pulses, root tubers and fruits which had also higher frequency of
consumption in all the three groups (Figure. 2).
Figure 2. Weekly Food Consumption Pattern Among the Three Groups of Farmers
The frequency of consumption of the highly consumed food groups was higher in the
inclusive pathway relative to the smallholder pathway and rain-fed farmers. However, the
widely consumed food groups are those rich in carbohydrates and starch while foods rich in
protein including meat, fish, milk and their products were consumed less frequently. This has
health implications due to over reliance on monotonous starchy staples. For example, a typical
household consumed milk once a week for the rain-fed farmers and three times for households
in the commercialisation pathways. Furthermore, from the food consumption score estimates,
4.3% of respondents were food insecure (Borderline food consumption category; FCS= 28.5 -
0,0
1,0
2,0
3,0
4,0
5,0
6,0
7,0
Grains
Root tubers
Pulses
Oilnuts
Milk
Meat
Fish
Vegetable
Fruits
Oil/fats
Beverages
Food groups
Mean number of times
consumed
Rain-fed farmers Smallholder pathway Inclusive pathway
Commercialisation Pathways: Implications on Smallholder Rice Farmers …
88
42) while 95.7% were food secure households (Acceptable food consumption category;
FCS>42) as shown in Table 3.
Table 3. Status of farmers’ commercialisation and food consumption categories
Variable
Rain-fed/Ref-
group (%)
Smallholder
pathway (%)
Inclusive
pathway (%)
Total (%)
χ2
Output commercialisation index
0 - 0.249
15(13.6)
1(1.1)
0(0)
16(6.3)
134.54***
0.25 - 0.499
71(64.5)
9(10)
1(1.8)
81(31.6)
0.5 - 1
24(21.8)
80(88.9)
55(98.2)
159(62.1)
Input commercialisation Index
0 - 0.249
52(47.3)
45(50)
36(64.3)
133(52)
6.50
0.25 - 0.499
49(44.5)
36(40)
19(33.9)
104(40.6)
0.5 - 1
9(8.2)
9(10)
1(1.8)
19(7.4)
Food consumption category(FCC)
Borderline
9(8.2)
2(2.2)
0(0)
11(4.3)
7.49**
Acceptable
101(91.8)
88(97.8)
56(100)
245(95.7)
Note: FCC (0-28) =Poor consumption, FCC (28.5-42) = Borderline, FCC>42 = Acceptable
(WFP, 2008); percentages in parentheses; **P >0.05, ***P>0.01.
Most of the food insecure households were found among the rain-fed farmers (8.2%) and
the rest were in the smallholder pathway (2.2%) while there were no food insecure households
among farmers in the inclusive pathway. On average, a smallholder rice farmer had a gross
annual income of about 6.26 million Tanzania shillings. Farmers in the smallholder and
inclusive pathways had higher income than rain-fed farmers by 7.36 and 6.56 million Tanzania
shillings respectively.
Similarly, smallholder farmers in the smallholder pathway had more valued assets relative
to the rest of the two groups though rain-fed farmers had the least valued assets of the three
groups. On access to health insurance, about 47.7% of the sampled households had access to
health insurance (either the National Health Insurance Fund or Community Health Fund
insurer). Smallholders in the inclusive scheme had more access to health insurance (60.7%)
compared to 50% and 39.1% of farmers in the smallholder scheme and rain-fed farmers
respectively.
4.2 Rice Output and Input Commercialisation
From Table 1 and based on FAO (1989), rice in the study area is a commercial crop (CCI
≥ 50%) with commercialisation index of 59%. This implies that, on average, 59% of the total
rice produced by smallholder farmers in the study area was sold. The crop was more
commercialized among farmers in the inclusive smallholder-medium/large scale pathway
where 80% of the rice produced was sold compared to 70% for smallholder pathway and 41%
for rain-fed farmers. On the input side, results showed that, the extent of use of improved inputs
was low in all the three groups since only 27% of the inputs used by farmers were purchased
from the market while the rest of the inputs used were either low productive local inputs or
retained inputs from previous year. This signifies less use of improved inputs purchased from
the market, leading to low productivity caused by the use of low-productive retained seeds.
On the status of smallholder farmers’ level of commercialisation, 62.1% of the respondents
were commercial oriented, 31.6% were still in transition (likely to commercialize) while only
6.30% were subsistence farmers. Of the three comparison groups, 98.2% of smallholder
F. N. Rashid, R. Alphonce and I. J. Minde
89
farmers in the inclusive scheme, 88.9% of farmers in the smallholder scheme and 21% of rain-
fed farmers sold more than 50% of total rice produced. This implies that, smallholders in the
inclusive pathway were more commercialized than the smallholder pathway and rain-fed
farmers. On the input side, only 7.4% of farmers used improved purchased inputs while 40.6%
used both local/retained and purchased inputs and 52% used entirely local low productive
inputs.
Based on farm size, 72.7% of the households cultivated less than 2 ha while only 3.1%
cultivated more than 5 ha. Results show also a positive relationship between farm size and the
degree of commercialisation. This implies that households with larger farms tend to produce
and sell more relative to farmers with small farms due to economies of scale. These results are
consistent with previous studies (Martey et al., 2012). Figure 3 show the relationship between
commercialisation and farm size.
Figure 3. Degree of Agricultural Commercialisation by Farm Size
From the descriptive statistics, the three comparable groups are similar but they differ in
some few characteristics. However, evaluating the impacts of commercialisation based on
comparison of simple means from ANOVA and chi-square tests would yield biased estimates
due to unobservable characteristics. In order to address the self –selection bias, PSM was used
following previous studies (Hailua et al., 2015; Amare et al., 2012; Herrmann & Grote, 2015)
as shown in section 4.3.
4.3 Econometric Results
4.3.1 Determinants of Agricultural Commercialisation
Binary logit models (Mitiku et al., 2014; Herrmann & Grote, 2015) were used to estimate
the propensity of participation in the pathways.
The estimated logit models were both significant at 1%. Since the coefficients estimated in
the logit model show only the direction (positive or negative) of the effects of the hypothesized
covariates on smallholders’ probability of participation in the commercialisation pathways,
average marginal effects were further estimated to infer the extent of the effect of covariates
on the treatment variable. From Table 4, the propensity scores indicates that most of the
72,7
24,2
3,1
0,53
0,72
0,82
0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
0
10
20
30
40
50
60
70
80
< 2 ha 2 - 5 ha > 5 ha
Proportion of rice sold
% Farmers by farm size
Farm size in hectares
% Farmers by farm size Proportion of rice sold
Commercialisation Pathways: Implications on Smallholder Rice Farmers …
90
hypothesized variables that affect farmers participation in either of the treatment including age,
family size, education level of the household head, sex, farm size, access to extension services
and access to market information were insignificant. This implies that the comparable groups
were similar in these covariates.
Cooperative membership, off-farm income and type of rice seed had positive significant
effect on the probability of smallholders to participate in either of the commercialisation
pathways while distance to the nearest market, negatively affected the probability of
participation in the commercialisation pathways and the rain-fed scheme.
Table 4. Propensity Score Estimation of Covariates Affecting Treatment in Each of the
Commercialisation Pathway with Rain-Fed Farmers as a Baseline Category
Variable
Inclusive pathway
Smallholder pathway
Coefficient
Marginal effect
Coefficient
Marginal effect
Age of household head
0.0212
(0.0490)
0.0006
(0.0013)
-0.0238
(0.0272)
-0.0023
(0.0026)
Education of household head
1.384
(0.874)
0.0366
(0.0223)
-0.507
(0.333)
-0.0484
(0.0313)
Household size
0.141
(0.321)
0.0037
(0.0084)
0.144
(0.170)
0.0138
(0.0162)
Farm size (ha)
0.522
(0.708)
0.0138
(0.0188)
0.201
(0.194)
0.0192
(0.0184)
Distance to market(km)
-2.895**
(1.116)
-0.0765
(0.0270)
0.363*
(0.176)
0.0347
(0.0162)
Off-farm income(Tsh)
3.91E-7*
(1.75E-7)
1.03E-8
(4.30E-9)
-2.32E-7
(2.41E-7)
-2.22E-8
(2.30E-8)
Sex(1=male, 0=female)
5.310
(3.363)
0.1403
(0.0878)
-0.301
(0.741)
-0.0287
(0.0708)
Seed(1=Improved, 0=local)
-2.133
(1.783)
-0.0564
(0.0472)
1.177*
(0.558)
0.1124
(0.0517)
Extension service (1=yes)
2.992
(1.929)
0.0791
(0.0511)
-0.168
(0.518)
-0.0161
(0.0494)
Market information(1=yes)
-3.275
(1.711)
-0.0865
(0.0433)
0.0991
(0.567)
0.0095
(0.0542)
Cooper member(1=yes,0=no)
13.48**
(4.878)
0.3561
(0.1166)
6.163***
(1.133)
0.5890
(0.0892)
Constant
-9.623
(5.615)
-1.485
(1.606)
N
166
200
Pseudo R2
0.8597
0.5456
LR chi2(11)
182.45
150.19
Prob > chi2
0.0000
0.0000
Loglikelihood
-14.89
-62.53
Standard errors in parentheses; * p<0.05, ** p<0.01, *** p<0.001
The influence of cooperative membership on participation in commercialisation was
positive and significant. Cooperatives act as social networks in which farmers can have access
to information related to production and marketing as well as social capital formation (Martey
et al., 2012; Camara, 2017). This further helps farmers in reducing transaction costs. Being a
member of a cooperative increases the probability of participation in the smallholder and
inclusive commercialisation pathway by 42.8% and 80.2% respectively relative to rain-fed
F. N. Rashid, R. Alphonce and I. J. Minde
91
farmers
1
. Off - farm income also positively and significantly influenced participation in either
of the treatment. An increase in the household’s off-farm income by one Tanzania shilling,
leads to an increase in the probability of participation in the inclusive commercialisation
pathway by 1.03 x 10-6 % ceteris paribus. This is similar to the findings by Hailua et al. (2015)
who found that having off-farm income positively influenced participation in commercial
agriculture. The plausible reason could be, farmers tend to use off-farm income earned to
invest in rice production aiming at increasing production volume and sales.
Furthermore, consistent with other studies (Ochieng et al., 2016; Hailua et al., 2015), an
increase in distance to the nearest output market negatively and significantly affected
participation in commercialisation. At the margin around the mean values, an increase in
distance by one kilometre leads to a decrease in the probability of participation in the inclusive
pathway by about 7.7%. This could be caused by higher transaction costs involved in accessing
the output market as well as market information dynamics. Improved seed positively and
significantly affected participation in the smallholder commercialisation pathway. For every
one kg increase in the use of improved rice seed, the probability of participating in rice
commercialisation increased by 11.9%. This could be brought as a result of high yielding type
of improved rice seeds including SARO 5 a mostly used seed variety in the smallholder
pathway relative to low yield local seeds due to its expected returns.
4.3.2 Impacts of Commercialisation Pathways on Productivity and Welfare
This section summarizes the results of the propensity score matching (Nearest Neighbour
matching and Kernel Matching algorithms) estimated to evaluate the impacts of the two
commercialisation pathways on productivity and welfare (Table 5) using rain-fed farmers as a
reference group.
Table 5. Impacts of Commercialisation Pathways on Productivity and Welfare
Treatment
Inclusive pathway
Smallholder pathway
Variable
ATT
ATT
NNM
KM
NNM
KM
Total factor productivity
1.02 *
(0.42)
0.98*
(0.39)
1.21**
(0.73)
1.17***
(0.35)
Income(“000000” TShs)
5.48***
(1.51)
5.42***
(1.11)
7.68***
(1.55)
7.65***
(1.43)
Food consumption score (FCS)
14.04***
(3.31)
13.59**
(1.66)
12.21***
(9.21)
12.90***
(2.73)
Assets(“000000”TShs)
6.65***
(1.24)
6.61***
(1.96)
-0.60
(3.21)
-0.66
(4.53)
Access to health insurance
0.36***
(0.55)
0.37***
(0.06)
-0.456***
(0.30)
-0.41***
(0.07)
Notes: * p<5%, ** p<10%, *** p<0.1%; figures in parentheses are standard errors,
NNM=Nearest Neighbour Matching, KM= Kernel Matching, ATT= Average Treatment
effect on the Treated.
Smallholder and inclusive commercialisation pathways significantly and positively
impacted on productivity and welfare. Productivity was measured in terms of returns to factors
of production expressed as the ratio of gross value of output to the sum of values of factors of
production employed. Similarly, welfare was measured by several indicators including annual
household income, food consumption score (FCS) and access to health insurance. Smallholder
Commercialisation Pathways: Implications on Smallholder Rice Farmers …
92
and inclusive commercialisation pathways were compared with the rain-fed farmers used as a
reference group.
4.3.2.1 Impacts on Smallholder Rice Farmers’ Productivity and Welfare
Total factor productivity index ranged between 0.983 and 1.02 for the inclusive and
between 1.17 and 1.21 for the smallholder commercialisation pathways respectively more than
that of rain-fed farmers. From these results, farmers in the inclusive pathway experienced
almost a constant return to factors of production (returns ≈ 1) while farmers in the smallholder
commercialisation pathway experienced increasing returns (returns >1). This implies that,
farmers in the smallholder commercialisation pathway had more returns to factors of
production relative to the other two groups of farmers.
4.3.2.2 Impacts on Smallholder Rice Farmers’ Welfare
Using the income estimates from Table 5, the results shows that smallholders in the
inclusive small-medium/large commercialisation pathway earned an annual income ranging
between 5.42 and 5.48 million Tanzania shillings more than rain-fed farmers. Similarly,
smallholders in the smallholder pathway earned an annual income ranging between 7.65 and
7.68 million Tanzania shillings more than rain-fed farmers. These results imply that both
commercialisation pathways led to a significant increase in the household income but the
impact was higher in the smallholder pathway relative to inclusive pathway. This is consistent
with Hailua et al. (2015) study in Ethiopia who found that farmers participating in
commercialisation projects had higher incomes than rain-fed farmers.
Similarly, farmers in either of the commercialisation pathways were more food secure than
the rain-fed farmers. The weighted frequency of food consumption among farmers in an
inclusive pathway ranged between 13.59 and 14.04 times more while it was between 12.2 and
12.9 times more in the smallholder pathway relative to rain-fed farmers respectively. This
implies that, participating farmers had more economic ability of food access. The results are
consistent with that found by Ochieng et al. (2015) in the Great Lakes Region (Rwanda and
DRC) who found that, banana and legumes commercial oriented farmers were more food
secure than rain-fed farmers. This could be explained by the fact that, commercial oriented
households could easily purchase more food varieties to supplement their own production.
Furthermore, farmers in the inclusive pathway had more valued assets relative to non-
participating farmers. On average, the value of assets owned by farmers in the inclusive
pathway ranged between 6.61 and 6.65 million Tanzania Shillings more than the value of
assets owned by rain-fed farmers. Similarly, about 36% of smallholders in the inclusive
pathway had access to health insurance more than that of rain-fed farmers. Generally, the
results do not confirm the postulated hypothesis that commercialisation has no significant
impacts on productivity and welfare of smallholder rice farmers in Mbarali district.
5. Conclusion and Policy Recommendations
This study aimed at evaluating the most effective commercialisation pathway (smallholder
or inclusive) and its impacts on productivity and welfare on smallholder rice farmers in the
pathways versus rain-fed farmers in Mbarali District. The findings indicated that both
smallholder and inclusive commercialisation pathways positively impacted on the extent of
smallholder rice commercialisation, productivity and welfare. The overall level of rice
commercialisation was more than half of what was produced (CCI=59). However, the extent
of rice commercialisation was higher in the inclusive smallholder –medium/large scale
pathway where 80% of rice produced was sold relative to 70% for smallholder farmers in the
F. N. Rashid, R. Alphonce and I. J. Minde
93
smallholder commercialisation pathway and 41% for rain-fed farmers. Similarly, smallholder
production system is characterized by low improved input use with input commercialisation
index of 27%. This is caused by high transaction costs involved in purchasing the inputs due
to poor roads, unavailability and untimely delivery of inputs in the study area. Since each
pathway brought some impacts relative to the other, investing in both commercialisation
pathways is crucial to explore the synergies. Therefore, programs intending to increase
smallholder agricultural productivity should put more focus on productivity enhancing inputs,
establishment of producer and marketing cooperatives. There should be also an investment in
irrigation facilities which provide incentives for farmers to increase investment in farm
production since the risk associated with farm failure is reduced.
Acknowledgements
The authors would like to thank the ASPIRES Tanzania project for funding this study. We
thank also Prof. Milu Muyanga of Michigan State University and Prof. Ntengua Mdoe of
Sokoine Univeristy of Agriculture for their ideas that helped in accomplishing this study.
References
Adewumi, M. O., Jimoh, A. & Omotesho, O. A. (2013). Implications of the Presence of Large
Scale Commercial Farmers on Small Scale Farming in Nigeria. The Case of Zimbabwean
Farmers in Kwara State. Knowledge Horizons. Economics 5(4): 67 - 73.
Amare, M., Asfaw, S. & Shiferaw, B. (2012). Welfare impacts of maize – pigeon pea
intensification in Tanzania. Agricultural Economics 43: 27–43.
Asfaw, S., Shiferaw, B., Simtowe, F. & Lipper, L. (2012). Impact of modern agricultural
technologies on smallholder welfare : Evidence from Tanzania and Ethiopia. Food Policy
37(3): 283–295.
Becerril, J. & Abdulai, A. (2010). The impact of improved maize varieties on poverty in
Mexico: a propensity score-matching approach. World Development 38(7): 1024-1035.
Bellemare, M. F. (2012). As you sow, so shall you reap: The welfare impacts of contract
farming? World Development 40(7): 1418-1434.
Busso, M., DiNardo, J. & McCrary, J. (2014). New evidence on the finite sample properties
of propensity score reweighting and matching estimators. Review of Economics and
Statistics 96(5): 885-897.
Byerlee, D. & Deininger, K. (2013). Growing resource scarcity and global farmland
investment. Annu. Rev. Resour. Econ. 5(1): 13-34.
Caliendo, M. & Kopeinig, S. (2005). Some practical guidance for the implementation of
propensity score matching. Journal of Economic Surveys 22(1): 31-72.
Camara, A. (2017). An Analysis of Welfare Effect of Market Participation of Smallholder
Farm Households in Guinea. Economic Research Guardian 7(1): 2 - 23.
Dehejia, R. H. & Wahba, S. (2002). Propensity score-matching methods for nonexperimental
causal studies. The Review of Economics and Statistics 84(1): 151-161.
FAO (1989). Horticultural Marketing: A resource and Training manual for Extension
Officers. Agricultural Services Bulletin no. 76, FAO, Rome. 118pp.
FAOSTAT (2017). Food and Agriculture data by country. Online available at:
www.fao.org/faostat/en/#home. Accessed on 12/08/2017.
Gujarati, D. (2004). Basic Econometrics. (4th ed). The McGraw−Hill, New York. 1024pp.
Hailua, G., Manjureb, K. & Aymutc, K. M. (2015). Crop commercialisation and smallholder
farmers’ livelihood in the Tigray Region, Ethiopia. Journal of Development and
Agricultural Economics 7(9): 314 – 322.
Commercialisation Pathways: Implications on Smallholder Rice Farmers …
94
Henley, D. (2012). The agrarian roots of industrial growth: Rural development in SouthEast
Asia and SubSaharan Africa. Development Policy Review 30: 25 - 47.
Herrmann, R. T. (2017). Large-scale agricultural investments and smallholder welfare: A
comparison of wage labour and out grower channels in Tanzania. World Development 90:
294-310.
Herrmann, R. & Grote, U. (2015). Large-scale agro-industrial investments and rural poverty:
evidence from sugarcane in Malawi. Journal of African Economies 24(5): 645-676.
IFAD. (2012). Rural Poverty in Africa. Available online at: https://www.ruralpovertyportal.
org/region/home/tags/africa. Accessed on 12/6/2017.
Jayne, T. S., Chamberlin, J., Traub, L., Sitko, N., Muyanga, M., Yeboah, F. K. & Kachule, R.
(2016). Africa's changing farm size distribution patterns: the rise of mediumscale farms.
Agricultural Economics 47(1): 197-214.
Kleemann, L. & Thiele, R. (2015). Rural welfare implications of large-scale land acquisitions
in Africa: A theoretical framework. Economic Modelling 51: 269-279.
Leavy, J. & Poulton, C. (2007). Commercialisations in agriculture. Ethiopian Journal of
Economics 16(1): 1-37.
Li, M. (2013). Using the propensity score method to estimate causal effects: A review and
practical guide. Organisational Research Methods 16(2): 188-226.
Martey, E., Al-hassan, R. M. and Kuwornu, J. K. M. (2012). Commercialisation of smallholder
agriculture in Ghana. A Tobit Regression Analysis 7(14): 2131–2141.
Masso, C., Baijukya, F., Ebanyat, P., Bouaziz, S., Wendt, J., Bekunda, M. & Vanlauwe, B.
(2017). Dilemma of nitrogen management for future food security in sub-Saharan Africa–
a review. Soil Research 55(6): 425-434.
Ministry of Finance and Planning – Poverty Eradication Division (MoFP – PED) and National
bureau of Statistics (2019). Tanzania Mainland House Hold Budget Survey 2017/18, Key
Indicators Report. Dodoma, Tanzania. 42pp.
Mitiku, A. (2014). Impacts of smallholder farmer Agricultural commercialisation on rural
household’s poverty. The International Journal of Applied Economics and Finance 8(2):
51 – 61.
Nkuba, J., Ndunguru, A., Madulu, R., Lwezaura, D., Kajiru, G., Babu, A. & Ley, G. (2016).
Rice value chain analysis in Tanzania: identification of constraints, opportunities and
upgrading strategies. African Crop Science Journal 24(1): 73-87.
Ochieng, J., Knerr, B., Owuor, G. & Ouma, E. (2017). Commercialisation of Food Crops and
Farm Productivity : Evidence from Smallholders in Central Africa. Journal of Agrekon
55(4): 458-482.
Okezie, C. A., Sulaiman, J. & Nwosu, A. C. (2012). Farm–level determinants of agricultural
commercialisation. International Journal of agriculture and Forestry 2(2): 1-5.
Otsuka, K. (2013). Food insecurity, income inequality, and the changing comparative
advantage in world agriculture. Agricultural Economics 44(1): 7-18.
Oya, C. (2012). Contract farming in sub-Saharan Africa: A survey of approaches, debates and
issues. Journal of Agrarian Change 12(1): 1-33.
Pingali, P. L. & Rosegrant, M. W. (1995). Agricultural Commercialisation and Diversification:
Processes and Policies. Food Policy, 20(3): 171 – 185.
RCT (2015). RCT’s Strategic Plan 2015–2019. Online available at:
rct.co.tz/documents/RCTS-STRATEGIC-PLAN-2015-2019]. Accessed on 16/7/2018
Rigg, J., Salamanca, A. & Thompson, E. C. (2016). The puzzle of East and Southeast Asia's
persistent smallholder. Journal of Rural Studies 43: 118-133.
Rosenbaum, P. R. & Rubin, D. B. (1983). The central role of the propensity score in
observational studies for causal effects. Biometrika 70(1): 41-55.
F. N. Rashid, R. Alphonce and I. J. Minde
95
SAGCOT (2013). The SAGCOT green print. Online available at: www.sagcot.com. Accessed
on 30/7/2018.
URT (2016). National Five year Development Plan 2016/17 – 2020/21. Nurturing
Industrialisation for Economic Transformation and Human Development. Ministry of
Finance and Planning, Dar es Salaam. 316pp.
Van Donge, J. K., Henley, D., & Lewis, P. (2012). Tracking development in SouthEast Asia
and sub-Saharan Africa: The primacy of policy. Development Policy Review 30: 5 - 25.
Von Braun, J. (1995). Agricultural commercialisation: Impacts on income and nutrition and
implications for policy. Food Policy 20(3): 187 – 202.
Wang, H. H., Wang, Y. & Delgado, M. S. (2014). The transition to modern agriculture:
Contract farming in developing economies. American Journal of Agricultural Economics
96(5): 1257-1271.
West, J. & Haug, R. (2017). Polarised narratives and complex realities in Tanzania’s Southern
Agricultural Growth Corridor. Development in Practice 27(4): 418-431.
WFP (2008). Food consumption analysis: calculation and use of the food consumption United
Nations Vulnerability Analysis and Mapping Branch. Online available at: https://scholar.
google.com]. Accessed on 8/7/2018.
White, H. & Sabarwal, S. (2014). Quasi-experimental design and methods. Methodological
Briefs: Impact Evaluation 8, UNICEF Office of Research, Florence 2014(8): 1–16.
Zhou, S., Minde,I. J. & Mtigwe, B. (2013). Smallholder Agricultural Commercialisation for
Income Growth and Poverty Alleviation in Southern Africa. A Review: African Journal of
Agricultural Research 8(22): 2599 – 2608.
1
For log-lin models (ln Yi = β0 + β1Di), the semi-elasticity with respect to the dummy Di
regressor with value 1 or 0, was calculated by the formula (e β1 – 1)*100 (Gujarati, 2004: 333).