Cash Transfers and Food Security in Sub‐Saharan Africa

• Author: Luca Pieroni,Francesco Burchi,Margherita Scarlato,Giorgio d’Agostino
• Published date: 01 December 2018
• Date: 01 December 2018
• DOI: http://doi.org/10.1111/saje.12204

© 2018 Economic Society of Sout h Africa . doi : 10.1111/ saje .122 04

383

South African Journal of Economics Vol. 86:4 December 2018

CASH TRANSFERS AND FOOD SECURITY IN

SUB-SAHARAN AFRICA

FRANCESCO BURCHI†, GIORGIO D’AGOSTINO‡, LUCA PIE RONI§ AND MARGHE RITA SCAR LATO*‡

Abstract

This paper compares the impact of social protection strategies based on cash transfers in a selection

of lower-middle-income and low-income sub-Saharan African countries. We use the synthetic

control method to evaluate changes in the trajectories of the prevalence of undernourishment

indicators during the 1990s and 2000s for several relevant countries (Ethiopia, Kenya, Malawi and

Rwanda). The results suggest that cash transfers exert a significant effect on access to food, though

there are differences in magnitude. Robustness analysis with placebo experiments confirms the

soundness of our results, and their implications for policy-makers are discussed.

JEL Classificati on: Q1, Q18, O13

Keywords: Sub-Saharan Africa, undernourishment, cash transfers

1. INTRODUC TION

Social protection strategies and featu res of cash transfers (CTs) in sub-Sahara n African

countries (SSA) are highly diversif ied (Niño-Zarazúa et al., 2012). In particular,

lower-middle and low-income countries in the region differ signif icantly with regard

to their social protection models. In middle-income countries, CTs often take the form

of programmes that are planned for the long term, managed by government institu-

tions and domestically funded. These CTs are usually addressed to specific vulnerable

groups, e.g. the elderly and children (Garcia and Moore, 2012; Niño-Zarazúa et al.,

2012). Conversely, low-income countries have limited room for extended redistributive

policies and are unable to collect taxes. They have implemented small CTs, which are

often inconsistent collections of fragmented projects, aiming to combat food ins ecurity

and extreme poverty or provide emergency responses to natural disasters and conflicts

(del Ninno and Mills, 2015). These interventions are typically non-governmental pro-

grammes, partially or fully funded by donors, with weak national political commit-

ment and precarious long-term sustainability (Garcia and Moore, 2012; Niño-Zarazúa

et al., 2012).

*Corresponding author: Ma rgherita Scarlato, Roma Tre University, Via Silvio D’Amico 77,

00145 Roma. E-mail: margherita.sca rlato@uniroma3.it

†Department of Sustaina ble Economic and Social De velopment, German Development

Institute/Deutsches Institut für Entwicklungspolitik (DIE).

‡Department of Economics, Roma Tre University.

§Department of Political Science, Universit y of Perugia.

South African Journal

of Economics

© 2018 Economic Society of Sout h Africa .

South African Journal of Economics Vol. 86:4 December 2018384

Empirically, the marked differences in t hese programmes have been ana lysed by mea-

suring the impact of specific CTs on several socioeconomic outcomes1 or through quali-

tative discussion of social protection strategies across emerging and developing countries

(Barrientos et al., 2008; Hanlon et al., 2010; Devereux, 2013; de Haan, 2014; Giovannetti

et al., 2015). Some of the impact evaluations of these programmes have specifically

looked at their effects on food security. However, these studies rely on micro-level data

(often collected ad hoc) and therefore lack external validity. It is, in fact, difficult to

compare results across countries to identify which kind of programme works better and

under which conditions.2

The present paper tries to fill this gap by employing the synthetic control method3 –

originally proposed by Abadie and Gardeazabal (2003) and Abadie et al. (2010, 2014) –

for analyzing and comparing the effects of different CTs on food security in SSA. In

order to estimate food security, we used the FAO indicator of the prevalence of under-

nourishment (PU), which has been proposed to track the progress of Sustainable

Development Goal 2 and for which long historical country-level data are avai lable. Three

main criteria were identified to detect suitable ca se-study countries for investigating food

security issues: (1) the presence of a CT as an important component of anti-poverty strat-

egies; (2) similar timing of implementation and comparable transfer size; and (3) differ-

ences based on certain programme objectives and targeting of CTs. Based on these

criteria, we selected four countries: Malawi, Kenya, Ethiopia and Rwanda.

In line with the synthetic control method, we then selected a comparison unit for

each case study as the weighted average of all potential comparison units (or control

group) that best resembles the characteristics of the country of interest in the years prior

to the introduction of the intervention. We created a counterfactual sample (synthetic

control) for each country of interest, which provides a close approximation of the treated

country prior to the intervention. We then compared the post-intervention outcomes

of the treated country, which was exposed to the intervention, with the outcomes that

occurred in the synthetic control, which was not exposed to the intervention. A discrep-

ancy between the trajectory of these outcomes is interpreted as being produced by the

intervention (Abadie et al., 2 014).

Our results show that CTs had a significant effect on food security, but high hetero-

geneity emerges across the four selected African countries. In the short term, the largest

effect was detected in Rwanda with a reduction of 18 percentage points of the under-

nourished population, whereas the lowest effect was encountered in Ethiopia with a sig-

nificant reduction of only 2 percentage points. A lthough the substantial heterogeneity in

the implementation of these policies makes the generalisation of the specif ic mechanisms

at work difficult, it provides suitable assessments for policy-maker decisions.

1For a review, see Davis et al. (2012) and Fiszbein and Schady (2009). Also see the Transfer

Project, funded by Save the Children UK and UNICEF, and From Protection to Production

(PtoP), funded by the FAO.

2Moreover, these studies usual ly look at the impacts on the CT beneficia ries, therefore disregard-

ing possible (positive or negative) externa lities on non-beneficiaries.

3For recent contributions to the synthetic control approach, see Bi llmeier and Nannicini (2013),

Horiuchi and Mayerson (2015), El-Shagi et al. (2016) and Grier and Mayna rd (2016).