Poverty, risk and insurance in rural Madagascar

Anne-Claire Thomas

Décembre 2011

Université catholique de Louvain

Sous la direction de B. Henry De Frahan et F. Gubert

This thesis aims at bringing some elements to untangle the causality behind the persistence of rural poverty.
In the vast literature on poverty, the small contribution of this work is:
‐ to propose a method to measure and find the determinants of poverty dynamics (Chap 1)
‐ to test one of the theory underlying persistent poverty, namely the risk poverty trap (Chap 2 et 3).
We mobilize a large and original panel database providing information on the evolution of the rural households’ standard of living in Madagascar. We are thus able to examine poverty determinants (chap 1), the impacts of agricultural shocks on growth (chap 2) and the distributional impact of informal risk management strategies (chap 3) at the microeconomic level of households. The first chapter indicates that large households with many dependants, namely children are more likely to be chronically poor. Family planning appears as an important poverty reduction tool in rural Madagascar[1]. The perpetuation of poverty is also due to poverty itself. Poverty leads households to take behaviours that durably hampered future income earnings opportunities. This last result appeals to a deeper analysis of the mechanisms trough which poverty impacts the future evolution of households’ standard of living. The following two chapters of this thesis examine one possible mechanism. It shows how the combination of a risky environment and informal risk management strategies create a poverty trap.
Besides, the first chapter of the thesis indicates that poverty is punctually linked to household’s characteristics. Education, assets owned, the nature of income generating activities, the composition of the activities portfolio and the riskiness of the environment shapes the probability to be poor. Undergoing damages on crop and demographic shocks like birth or death of household members are causes of poverty. However, in the first chapter, only idiosyncratic risks are explicitly introduced in the regression since covariate shocks are controlled by the area and year dummies. We are thus unable to evaluate the overall impacts of shocks.
The second chapter is aimed at measuring the impact of both idiosyncratic and covariate shocks on household consumption growth. We show that rainfall shocks have persistent impacts on consumption growth for poor households. A large part of rainfall shocks is uninsured for both the poor and non poor. The richer households however manage to cope with the shock and catch up within a year whereas poor’s consumption is permanently set on a lower growth path because of the consequence of the shock. This result suggests that risk management strategies of poor households are less efficient to smooth consumption that those from the rich households.
The third chapter thus examines risk management strategies by wealth level. Since the market fails to provide efficient credit and insurance services, households have to develop informal risk management strategies. We assess that these strategies are such that poor households have lower capacities to cope with shocks ex‐post than the richer households. On the one hand they have fewer buffer stocks. On the other hand they participate less to informal mutual insurance. They are thus obliged to reduce exposure to risk ex‐ante by adopting less risky and less remunerative activities. Since poor are confined in doing less remunerative activities, they cannot accumulate assets that would eventually increase their capacity to cope with shocks ex‐post. Moreover since they adopt less risky activities, they have lower incentives to participate to mutual insurance. Their benefit is limited because sharing income with richer households ex‐post them to the higher risks taken by the rich. Wealth differentiated pattern of informal risk management strategies creates poverty traps.
The second and third chapters show that welfare costs due to shocks and foregone profitable opportunities contribute to persistent poverty. The presence of risk poverty traps has important policy implications. Protection against shocks may matter as much as exogenous improvements of productivity. The poverty trap hypothesis suggests high returns to providing insurance that excludes poor people from more remunerative (and more risky) activities. The easy answer would be to recommend formal insurance as the policy response to break the risk poverty trap. Microinsurance, meaning the provision of insurance tailored to the poor who would otherwise not be able to take insurance has recently received a very large attention. However, if insurance provision is theoretically the right tools, the few empirical evidences available show that microinsurance has not lead to many successes. Uptake is indeed very low. Some of difficulties concern the difficulty of households to understand the product, to trust the provider or simply to afford insurance (Mc Cord, 2001; Radermacher, Dror, & Noble, 2006). Implementation if microinsurance should be done with careful examinations of the local conditions to design a product that fit the needs and gain the trust of potential clients. Marketing efforts have to be important since it is a much more complex product than microcredit. In some places, other insurance‐like mechanisms can do at least as good as the provision of insurance. A caveat of microinsurance literature is to forget to compare insurance with alternative mechanism such as microsaving, emergency credit or public safety nets. Moreover, if micro insurance is aimed at fighting poverty, it is needed to evaluate its distributional impact.


[1] Birth Control was a priority of Ravalomanana’s government program. The Ministry of health was renamed in 2003 the Ministry of Health and Family Planning making Madagascar one of the few place in the world where voluntarily family planning is as a key health intervention is explicitly recognized.