Does food aid really have disincentive effects?
New evidence from sub-Saharan Africa
University of Kiel, Kiel, Germany
Christopher B. Barrett*
Cornell University, Ithaca, NY, USA
International Food Policy Research Institute, Washington, DC, USA
[Download complete version - 215Kb ~ 1 min (39 pages)]
[ Share with a friend
Per capita food production in sub-Saharan Africa declined precipitously between the early 1970s and the mid-1980s. While there has been modest recovery over the past fifteen or so years, per capita food production in sub-Saharan Africa remains almost 20 percent below the levels observed thirty years ago. Over the same period that food production per capita declined, food aid flows into sub-Saharan Africa increased nearly fivefold. Food aid flows then became extremely volatile, but remaining in the 2.0-4.0
million metric tons per year range for the past decade (Figure 1). This juxtaposition of falling per capita food production and rising food aid receipts raises widespread concern among donors, government policymakers and development practitioners in the NGO
community. Some observers interpret these data as a sign that food aid flows have caused a decline in African agriculture.
The logic behind this assertion is that food aid deliveries increase supply faster than they stimulate demand, depressing the food prices received by recipient country producers and traders, and thereby creating disincentives for producers to invest in
improved technologies or for marketing intermediaries to invest in storage and transport capacity. Longstanding concerns about the food price effects of food aid deliveries – which date at least from Schultz (1960) – have been corroborated empirically in several cases in sub-Saharan Africa in recent years (Donovan et al. 1999, Barrett and Maxwell 2005). These concerns are echoed at the household level where concern is raised that the receipt of food aid causes households to reduce their labor supply, discourages household investment in agricultural production and crowds out private transfers and other means of informal responses to shocks. Special concerns have been voiced about the labor market effects of ill-conceived food-for-work (FFW) projects, which may distort local labor markets by attracting workers away from vital activities during the agricultural year, especially if the wages offered under FFW are at or above prevailing market wage rates (Maxwell and Singer, 1979). Further, critics assert, food aid permits recipient country governments to continue to neglect the needs of their agricultural sectors for adequate rural infrastructure, agricultural research and extension, as well as price and trade policies that are not biased against agriculture.
Given the importance of reversing productivity declines in African agriculture and the contested assessments of the role food aid plays in agricultural development, one would expect to find a considerable body of carefully analyzed empirical evidence on the dependency and disincentive effects of food aid on farm households and the agricultural sectors of recipient economies. This is not the case. Lentz (2003) provides a careful annotated bibliography covering papers on the dependency and disincentive of food aid published in the last 25 years. Her review shows that, strikingly, while a number of authors claim that such disincentive effects exist, these claims are based only on case studies and anecdotes.1 While case studies can be highly informative, one would like some quantitative corroboration of such claims. No such evidence exists to date.
In this paper, we examine the impact of food aid at both the micro-level, using household survey data from Ethiopia, the largest recipient in Africa, and the macro-level, using national production data from across the continent. At the household level, when we examine descriptive statistics such as unconditional means, we find evidence that food aid has marked disincentive effects on a range of behaviors. These effects are large in magnitude and statistically significant. However, these comparisons do not control for other factors. To take a simple example, free food allocations are sometimes targeted towards individuals who are unable to work because they are infirm. A comparison of the unconditional means of labor supply between two groups, disabled individuals receiving free food and able-bodied individuals not receiving free food, would likely find labor supply lower amongst the former but arguably this is a consequence of physical disability that induces food aid receipt and is not a consequence of food aid receipt itself. The everimportant distinction between causality and correlation is critically important.
In our household level analysis, we find that when we control for characteristics such as age, sex and education of head, land holdings, size and location – attributes on which food aid distributions are sometimes targeted – food aid’s apparent disincentive
effects vanish. Our findings are even more marked when we examine the impact of food aid flows on per capita agricultural production across Africa. Once we introduce appropriate econometric controls, we find that contrary to received wisdom, food aid
flows may even have stimulated (rather than depressed) per capita food production in sub-Saharan Africa in the short-term following receipt of food aid shipments.
As with any empirical study, care should be taken not to over-interpret these results. There may be other mechanisms by which food aid generates disincentive effects that we do not consider here. For example, it is possible that food-for-work projects
crowd out participation in community efforts to manage common property resources. If poorly targeted food aid depresses local market prices, it may create disincentives for farmers who do not receive food aid, although our macro-level evidence suggests this effect is not large as a general phenomenon. At the national level, the impact of food aid on recipient government effort to promote agricultural production remains an open research question. Given these and other lacuna, we stress that we see the paper’s
principal contribution as methodological, namely the importance of controlling for potentially confounding effects when assessing the impact of food aid.
Section 2 lays out the conceptual issues associated with evaluating the impact of food aid. Section 3 uses household-level data from Ethiopia to examine the impact of food aid on labor supply, on-farm investment and mutual aid. Section 4 then explores the dynamic relationship between food production and food aid at the macro-level of countries’ aggregate food production per capita. The final section summarizes our findings and concludes with policy implications.2
* Corresponding author: 315 Warren Hall, Department of Applied Economics and Management, Cornell University, Ithaca, NY 14853-7801 USA, email: , tel: 1-607-255-4489, fax: 1-607-255-9984.
We note too that it is possible to construct an opposing perspective on the food aid – food productivity relationship, which holds that increased food aid has been and could prospectively be beneficial to African agriculture. This line of argument emphasizes food aid’s role in increasing poor households’ access to and consumption of food in the face of exogenous climatic or political shocks, thereby improving human nutritional status, health, labor productivity and income earning capacity relative to what would transpire in the absence of food aid. In Ethiopia, Quisumbing (2003) and Yamano et al. (2005) find that food aid plays an important role in protecting the nutritional status of pre-school children, particularly in the aftermath of
shocks. Bezuneh et al. (1988), Barrett et al. (2001) and Holden et al. (forthcoming) provide evidence that in some localities, food aid may have relieved seasonal credit constraints that limit farmers’ capacity to invest in mineral fertilizer, modern seeds and other productivity-enhancing inputs and to have sparked investment in terracing and other on-farm improvements that likewise increase yields with a lag (Holden et al. forthcoming). Von Braun et al. (1999) report on the multiplier effects of a FFW-built road in the Ethiopian lowlands, where improved market access directly attributable to that road led to the establishment of water
mills and fruit plantations and the revival of traditional cotton spinning and weaving in the three years after the road was built. Well-conceived and managed FFW projects that invest in necessary materials to complement labor inputs can ‘crowd in’ private investment, as Holden et al. (forthcoming) find in the case of private investment in soil and water conservation structures in Tigray region of Ethiopia and Bezuneh et al. (1988) find in purchased farm inputs in Baringo District, Kenya.
We are aware of the important debate over whether food aid is the best resource to use in promoting agricultural market development in recipient countries, particularly in sub-Saharan Africa. In the abstract, food aid seems an undeniably inferior resource because it is doubly tied, in the following sense. Food aid is inflexible in the form of the transfer – it comes only as food, not as cash or other useful goods or services – and it is typically inflexible in sourcing that food, relying on commodities procured in and shipped directly from the donor country. However, if the volume of international transfers depends on the form in which donors provide it, particularly if donor country governments are willing to provide food aid in part because
it supports their own farm and agribusiness constituencies, then food aid that is relatively inefficient per dollar transferred may nonetheless generate greater absolute transfers than less restricted cash appropriations. It is within this context that we place our analysis: whether food aid violates a necessary (but not sufficient) condition for its use in promoting recipient household and country development.