Food Security and Social Protection in Sub-Saharan Africa: an Evaluation of Cash Transfer Programs Giorgio d Agostino 1 Margherita Scarlato 1 Luca Pieroni 2 1 University of Rome III (Italy) 2 University of Perugia (Italy) Between Crisis and Development: Which Role for the Bio-Economy. AIEAA Conference 2013, Parma, 6-7 June 2013
Introduction: The literature on social protection is gradually shifting the focus of interest from purely redistributive actions to active welfare measures that aim to promote capabilities and incentives. Emphasis is given to non-transitory effects of social policies targeted to the root of "poverty traps", i.e. social policies able to change the expectations and behaviours of poor people (Barrett, Carter and Ikegami, 2008). An even bolder approach refers to changing marginalised individuals self - perception by increasing their participation in democratic processes and involvement in the implementation of public policies (Devereux and Sabates-Wheeler, 2007; Cook and Kabeer 2009). According with this change in the theoretical perspective, a new generation of social protection measures have been implemented throughout the developing and emerging countries.
Aims: To analyse the impact of social protection interventions, through cash transfers programs, on food security in a comparative perspective for a sample of Sub- Saharan African countries, for the period 1992-2010. To propose an non-experimental evaluation methodology to compare the trajectory of post-intervention food security of treated economies with the trajectory of a combination of similar but untreated economies. To assesses the role of cash transfer programs in achieving food security and reducing vulnerability of the poorest segments of the population.
Method: The empirical analysis is based on a dataset covering 48 sub-saharan African countries for the period 1992-2010. Treated units LSO Lesotho 2005, 2009 Upper-middle income SWZ Swaziland 2005 Upper-middle income BFA Burkina Faso 2008 Low income ETH Ethiopia 2005 Low-income GHA Ghana 2008 Low-income MLI Mali 2000-2007 Low-income MWI Malawi 2005 Low-income NER Niger 2008 Low-income RWA Rwanda 2006 Low-income SLE Sierra Leone 2005 Fragile state ZWE Zimbabwe 2004 Fragile state Donor pool countries AGO Angola Upper-middle income CMR Cameroon Upper-middle income GAB Gabon Upper-middle income GNQ Equatorial Guinea + Upper-middle income MUS Mauritius Upper-middle income SYC Seychelles + Upper-middle income COM Comoros Low-income GIN Guinea Low-income MDG Madagascar Low-income MRT Mauritania Low-income TCD Chad Low-income UGA Uganda Low-income BDI Burundi Fragile state CIV Cote d Ivoire Fragile state COG Congo Republic of Fragile state LBR Liberia Fragile state SDN Sudan + Fragile state SOM Somalia + Fragile state STP Sao Tome and Principe Fragile state TGO Togo Fragile state
Method: Abadie and Gardeazabal (2003); Abadie et al. (2010, 2012) propose to use a synthetic control method, which is a weighted combination of potential control countries to approximate the most relevant characteristics linked to food security of the treated country. As a measure of food insecurity, we use the estimates of the proportion of population in condition of chronic undernourishment (see Musset 2011 for a review of hunger indices and methods). Covariates 1) Access to improved water sources, 2) Access to improved sanitation facilities, 3) Cereal import dependency ratio, 4) Primary education females, 5) Political stability and absence of violence, 5) Agricultural population, 6) Growth rate of percapita GDP, 7) Mean of prevalence of undernourishment.
An example: (a) Ethiopia (b) Ethiopia placebo test Comparison countries for Ethiopia: Angola (0.434), Burundi (0.151), Comoros (0.380), Uganda (0.035).
Results: (c) Lesotho (d) Swaziland (e) Burkina Faso (f) Ethiopia
Results: (g) Ghana (h) Mali (i) Malawi (j) Niger
Results: (k) Rwanda (l) Sierra Leone (m) Zimbabwe
Conclusions: We find heterogeneous results which depend on whatever the policy have been implemented in upper-middle income countries or in low-income/fragile states. We do not find any significant relationship when these policies have been introduced in high-income countries such as Swaziland or Lesotho, whereas positive and significant impacts have been founded in low-income countries such as Rwanda and Ethiopia as well as in Sierra Leone and Zimbabwe, classified by World Bank as fragile states.
Conclusions: A possible explanation for these heterogenous results is linked to the target of these policies - e.g., more/less universal the application of the public scheme for reducing the economic vulnerability, and to administrative issues behind the implementation of these policies. Since these policies are commonly introduced to recover natural disasters or famines, the choice is driven by preferences of the donors.
Policy suggestions: To drive these policies from external donors to government agencies. To chose the population target more accurately. To implement long-term cash transfers programs instead of short-term policies to recover natural disasters or famines.