for Economic Research
The legacy of civil war: The case of
J12, J22, O12
Mozambique, poverty trap,
shocks, violent conict
How do households cope with post-war economic reconstruction? Are groups of house-
holds trapped in poverty? These questions are discussed using the case of Mozambique,
a nation in Southern Africa. Shortly after gaining independence in 1975, Mozambique
delved into a civil war that lasted until 1992. Our analysis explores poverty dynamics
among rural households a decade after the civil war ended. Findings indicate that
both income and asset endowments are very low and the prevalence of poverty is
high. However, there is no evidence that households are trapped in poverty. Instead,
the rural farm-based economy as such provides very few prospects for improved liveli-
hoods. This appears to be one of the legacies of the civil war, which destroyed much
of the public infrastructure and the physical asset endowments of households and
increased the welfare gap between urban and rural areas.
In Mozambique, the period of violent conflict began with armed struggle for in-
dependence in the mid-1960s. When Mozambique gained its independence from
Portugal in 1975, it was among the poorest nations at that time. The independent
government, led by the former liberation movement FRELIMO (Frente de Libertação
de Moçambique, Front for the Liberation of Mozambique), adopted a socialist
path to development. Neighboring South Africa and Rhodesia, both ruled by white
minority regimes at that time, feared that FRELIMO would support black libera-
tion movements in their own countries. They conducted military attacks against
Mozambique and, from the end of the 1970s onwards, supported the formation of a
rebel organization in Mozambique, RENAMO (Resistência Nacional Moçambicana,
Mozambique National Resistance).
In the course of the civil war, RENAMO became increasingly independent from
its former supporters. RENAMO spread terror as a means of discouraging people
from supporting the FRELIMO government.1 The principal targets of RENAMO
attacks were public institutions, the well-off, as well as better-educated individuals,
often using brute force. This caused major population displacement away from rural
areas, where most actions of warfare took place. About 1 million individuals died
and an estimated 5 million people were displaced.2
1 Vines, A.: RENAMO: From Terrorism to Democracy in Mozambique? York, 1996. Wilson, K.B.: Cults of Violence and
Counter-Violence in Mozambique. Journal of Southern African Studies, 18(3), 1992, 527-582.
2 Hanlon, J.: Peace without prot: How the IMF blocks rebuilding in Mozambique. Oxford, 1996. Synge, R.: Mozam-
February 9, 2011
Lena Giesbert and Kati Schindler
Since neither FRELIMO nor RENAMO had a
reasonable chance of winning the war by military
means, both parties began peace negotiations in
1989. A General Peace Accord was signed in Rome
in 1992. During elections, FRELIMO was confirmed
in power and in 2011 is still the governing party in
After the transition to peace, a large number of dis-
placed people returned to the countryside, although
not necessarily to their original communities.
they arrived in devastated areas. At the end of the
war, 60 percent of primary schools were closed or
had been destroyed, 40 percent of immobile physi-
cal capital, such as farm irrigation systems, shops,
and administrative buildings, had been destroyed
or were non-operational4 and about half of all roads
were impassable.5 Moreover, both war parties had
planted landmines around public buildings and
along roads and footpaths, causing casualties for
years after the war concluded.6 Many rural house-
holds also lost their productive asset base, including
land tenure. People fleeing their homes often left
their property behind, while many of those staying
on in rural areas had their food reserves, assets,
and livestock looted by soldiers. For instance, the
number of cows fell from 1.3 million in 1980 to
250,000 in the post-war period.7
Is economic reconstruction
Given the massive destruction of assets, our re-
search8 investigates how households cope with re-
construction. What are the poverty dynamics among
rural households more than a decade after the civil
war ended? Are there groups of households still
trapped in poverty?
At first sight, there is contradicting evidence: On
the one hand, macro-economic indicators improved
steadily in the post-war period. Strict structural ad-
justment programs were implemented in the late
bique: UN Peacekeeping in Action 1992-1994. Washington, DC, 1997.
3 Wilson, K.B.: Internally Displaced, Refugees and Returnees from and in
Mozambique. Studies on Emergencies and Disaster Relief, 1, 1994.
4 Brück, T.: Mozambique: The Economic Effects of the War. In: Stewart,
F., E.V.K. FitzGerald (Eds): War and Underdevelopment. Oxford, 2001,
5 Arndt, C., Jones, S., Tarp, F.: Aid and Development: The Mozambican
Case. University of Copenhagen, Department of Economics Discussion
Papers, 06-13, 2006.
6 Ascherio, A. et al.: Deaths and injuries caused by land mines in Mozam-
bique, The Lancet, 346, 1995, 721-24.
7 Brück, T.: Mozambique: The Economic Effects of the War. In: Stewart,
F., FitzGerald, E.V.K. (Eds): War and Underdevelopment. Oxford, 2001,
8 Giesbert, L., Schindler, K.: Assets, shocks, and poverty traps in rural Mo-
zambique. DIW Discussion Paper, 1073, 2010.
1980s as an attempt to achieve macroeconomic
stability.9 With large inflows of donor assistance,
investments in infrastructure, education, and the
health sector were made. For instance, between 1992
and 2004, the number of students enrolled in primary
and secondary schools increased by 174 and 537
On the other hand, these improvements do not appear
to have trickled down to rural areas, where 63 per-
cent of the population lived in 2008. Between 1995
and 2001, median income among rural households
rose by 30 percent.
Yet, this increase is largely
driven by favorable market prices for agricultural
produce in 2001. In contrast, productivity of the
smallholder farm sector stagnated at a low level.12
In order to investigate the well-being of rural house-
holds in more depth, we analyze two waves of the
Trabalho de Inquérito Agrícola (TIA) household sur-
vey. The surveys were collected by the Mozambican
Ministry of Agriculture in 2002 and 2005. Each
survey is representative of smallholder farm house-
holds in rural Mozambique. The survey question-
naire records, among other things, household de-
mography, income, assets, farming techniques, and
household exposure to shocks. Our analysis builds
on 4,058 rural households that were interviewed
in both years. These longitudinal data allow us to
observe the same households over time.
Asset ownership and income are
Table 1 presents a poverty profile, showing how
poverty varies across groups of households in 2002
and 2005. The incidence of poverty decreased
slightly over the time period of interest: In 2002,
80 percent of all rural households were poor, while
in 2005 only 76 percent were poor.13 Households
9 Hanlon, J.: Peace without prot: How the IMF blocks rebuilding in
Mozambique. Oxford, 1996.
10 Arndt, C., Jones, S., Tarp, F.: Aid and Development: The Mozambican
Case. University of Copenhagen, Department of Economics Discussion
Papers, 06-13, 2006.
11 Boughton, D., et al.: Changes in Rural Household Income Patterns
in Mozambique, 1996-2002, and Implications for Agriculture’s Contribu-
tion to Poverty Reduction. Republic of Mozambique, Ministry of Agricul-
ture and Rural Development, Directorate of Economics Research Paper
Series, 61E, 2006.
12 Datt, G., et al.: Determinants of Poverty in Mozambique:1996-97.
FCND Discussion Paper, 78, 2000.
13 The incidence of poverty found here is higher compared to studies
using household consumption data (which is not collected in the TIA
surveys). This is due to the tendency of households to underreport in-
come to a larger degree than consumption expenditures, greater annual
uctuations in income than consumption, and the fact that consump-
tion expenditures are typically valued with retail prices, while we value
income from farm production with (lower) producer prices. Hence, the
focus in the following is less on absolute levels of poverty, but rather on
changes in poverty over time and the differences in poverty across groups
The legacy of civil war
11DIW Berlin Weekly Report No. 2/2011
cultivating a large area of land had a much higher
income per person and a lower incidence of poverty
in both years compared to households cultivating
little land. This finding underlines the importance
of land as the key input for farming in Mozambique.
Moreover, there seems to be an increasing impor-
tance of knowledge on markets and prices as well
as education. Owning a radio lowers the incidence
of poverty by 12 and 19 percentage points in 2002
and 2005, respectively. Similarly, if the head of
household has some education, the incidence of
poverty is lower by around 10 percentage points
in both years.
These results suggest a strong relationship between
asset ownership and wellbeing. However, the cau-
sality of this relation is not clear: Asset ownership
may raise income; or better-off households may
invest their wealth in assets.
There is no evidence for a poverty
As a further step of analysis, we analyze households’
asset accumulation over time and test whether a
poverty trap is present in post-war Mozambique.14
14 See, for instance, Carter, M.R. and Barrett, C.B. (2006) The economics
A poverty trap implies that one threshold of as-
set ownership exists, dividing households above
and below this threshold into two groups. On the
one hand, a household above this asset threshold is
predicted to accumulate assets over time as more
profitable activities and investments become acces-
sible. Eventually, the household would move out
of poverty. On the other hand, a household below
this threshold is too poor to accumulate assets. The
household remains trapped at low welfare levels.
Testing whether such a threshold exists involves sev-
eral steps.15 First, for every household, we combine
different kinds of assets – land, fruit trees, livestock,
tools for farming, human capital – into one common
index. Second, we estimate the relationship between
a household’s asset index in 2002 and 2005. Third,
we account for other household and community
characteristics and exposure to shocks that may in-
fluence a household’s asset accumulation.
Findings reveal that there is no evidence for a pov-
erty trap. Instead, it appears that all households in
rural Mozambique converge towards the same as-
of poverty traps and persistent poverty: An asset-based approach, Jour-
nal of Development Studies, 42(2), pp. 178-199.
15 For a detailed description of methods used, see Giesbert, L., Schindler,
K.: Assets, shocks, and poverty traps in rural Mozambique. DIW Discussion
Paper, 1073, 2010.
Mean household income per adult
equivalent (in Euro
2002 2005 2002 2005 2002 2005
All households 100 10 0 76.1 91. 8 80 76
Household cultivates more than 0.5 hectare
of land per adult equivalent
yes 28 44 108.8 111. 6 71 72
no 71 55 63.2 73.9 83 79
Household owns livestock yes 77 69 78.4 96.7 79 74
no 23 31 67. 9 77.2 81 80
Household owns a bike yes 24 32 9 7.3 94.1 71 70
no 76 68 69.6 89.3 83 79
Household owns a radio yes 51 53 94.7 117 . 6 74 67
no 49 47 5.7 59.9 86 86
Household uses fertilizers or pesticides yes 15 11 107.5 12 7.3 74 68
no 85 89 70.8 86.6 81 77
Head has some education yes 60 65 87. 4 103.0 76 73
no 39 34 58.7 6 7.5 86 82
Household is member of an association yes 4 6 103.6 157. 1 76 68
no 96 93 74.8 85.7 80 77
1 At constant 2005 values.
Results are weighted with population weights and inverse probability weights. Official, regional-specific food poverty
lines are used. Household income is calculated per adult equivalent in order to make income comparable across
households of different size and composition.
Source: TIA 2002/2005 panel households; own calculations.
DIW Berlin 2011
Asset ownership is strongly correlated with well-being.
Tel. +49-30-897 89-0
Fax +49-30-897 89-200
Translated from the
All articles are protected
set level in the medium-term. Surprisingly, this asset level is at an extremely low
income level, worth 1.11 times the poverty line. This corresponds to a yearly
income per adult of about 142 Euro. Households above this asset level are better
off for stochastic reasons, such as luck and favourable weather for farming. Their
livelihood is not grounded in a sustainable asset base and households are expected
to lose assets in future until they reach the common asset equilibrium. Households
below the asset equilibrium will eventually improve their well-being and approach
the equilibrium from below. To conclude, there appears to be relative stagnation
and little differentiation across households in rural Mozambique.
Our findings contrast with most other empirical studies on Sub-Saharan Africa that
do find poverty traps. The Mozambican civil war may explain the unusual results
in two regards. First, there may have been a selection in the back-migration of
displaced people to rural areas after the civil war ended. Possibly, most returnees
to rural areas did not succeed during the war, having less economic incentives to
stay on in urban settlements. This would explain the low differentiation across rural
households, while the wealth divide between urban and rural households is large.
Second, the massive war-related destruction of assets may have amplified the
impact of unfavourable economic conditions in rural Mozambique. These include
low population density, low degree of market integration, high transport costs, high
frequency of natural disasters, and very few off-farm employment opportunities.
While there is no evidence that groups of rural households are trapped in poverty,
one may think of the rural farm-based economy as such being trapped in poverty
compared to the urban economy.
The long-lasting civil war in Mozambique disrupted economic activities in rural
areas and caused major destruction of physical property. In the early post-war period,
many displaced persons returned to rural areas and reconstructed their livelihoods.
While overall economic indicators improved, economic reconstruction fell short
of expectations at the household level.
Poverty decreased slightly between 2002 and 2005. Yet, in absolute terms, poverty
remained at a high level, even by regional standards. Surprisingly, there is no evi-
dence of a poverty trap that discriminates against some households based on their
initial welfare endowments. Rather, all rural households are expected to converge
to a common asset level associated with a very low income level. This may indicate
an overall development trap in the farm-based rural economy, being one legacy of
the civil war. This indicates the need for structural change in rural areas, including
improved access to, and integration of markets.
(First published as “Das Erbe des Bürgerkrieges in Mosambik: Landbevölkerung
verharrt in Armut”, in: Wochenbericht des DIW Berlin Nr. 4/2011.)