"There are more buyers of wheat in Ethiopia than sellers of wheat at all levels of income -- an important result because it means that at all levels of living standards, more households benefit from food aid (and a subsequent reduction in wheat prices) than are hurt by it."
The delivery of food aid to developing countries seems like an uncontroversial policy -- a straightforward effort that helps the poor and underscores the generosity of donor nations. Yet, economists have long debated the merits of food aid. By increasing the local supply of food, such aid may depress prices and thus undercut the income of rural farmers in the recipient nations, for example; it also may discourage local production. And, since the poor often are concentrated in rural areas, food aid in fact may disproportionately hurt the poor.
NBER researchers James Levinsohn and Margaret McMillan tackle this debate in Does Food Aid Harm the Poor? Household Evidence from Ethiopia. The impact of lower food prices on the poor, they reason, hinges on whether poor households tend to be net buyers or net sellers of food. The authors seek to answer this question by examining consumption and expenditures survey data from both urban and rural households in Ethiopia. They focus on Ethiopia because it receives more food aid than almost any other nation in the world, but also because it is widely recognized that raising the productivity and profitability of small-scale Ethiopian farmers is essential to reducing poverty in the country.
Food aid can take several forms, but some portion of all types of food aid (including emergency relief aid) is eventually sold in local markets and thus competes with domestic producers. Therefore, food aid will benefit Ethiopia's net food buyers and hurt its' net food sellers. To carry out their study, Levinsohn and McMillan merge data from two nationally representative surveys and create a data set of 8,212 urban and 8,308 rural Ethiopian households.
They find that households tend to earn income from only one or two different cereal grains, and that rural households rely more heavily on such income than urban households. Indeed, 21 percent of rural households report positive income from teff, 12 percent from wheat, 10 percent from barley, 24 percent from maize, 11 percent from sorghum, and 12 percent from coffee. Meanwhile, urban households report figures of less than 3 percent for each of these products. The authors also find that households spend a large fraction of their income on cereal grains, ranging from 26 percent to 12 percent for rural households and from 16 percent to 5 percent for urban households. Since wheat is the only cereal imported in the form of food aid, it is the 12 percent of rural households that report income from wheat who stand to gain most from price increases and lose most from price declines.
The authors classify households as either net buyers of wheat (if they buy more than they sell) or as net sellers. To determine the poverty impact of food aid, they also classify the households by expenditure per capita and assess whether the poor households are net buyers or sellers of food. Finally, they estimate the magnitude of the price changes caused by food aid and hence the welfare effects of an increase in the price of food.
Levinsohn and McMillan offer several conclusions. First, net buyers of wheat are poorer than net sellers of wheat in Ethiopia. Indeed, roughly 85 percent of the poorest households are net buyers of wheat. Second, there are more buyers of wheat in Ethiopia than sellers of wheat at all levels of income -- an important result because it means that at all levels of living standards, more households benefit from food aid (and a subsequent reduction in wheat prices) than are hurt by it. Third, the proportion of net sellers is increasing in living standards, and fourth, poorer households (in rural and in urban areas) benefit proportionately more from a drop in the price of wheat. "In light of this evidence," the authors conclude, "it appears that households at all levels of income benefit from food aid and that -- somewhat surprisingly -- the benefits go disproportionately to the poorest households."
Levinsohn and McMillan estimate that, in the absence of food aid, the price of wheat in Ethiopia would be $295 per metric ton, compared to an actual price of $193 per metric ton in 1999. On average, the authors conclude, "the loss in consumer surplus works out to roughly 37 US dollars per household per year for households that consume wheat and the gain in producer surplus works out to roughly 157 US dollars per household per year for households that sell wheat." In a nation such as Ethiopia, where the poverty line is about $132 per year, the impact is therefore substantial.
This paper uses household-level data from Ethiopia to investigate the impact of food aid on the poor. We find that food aid in Ethiopia is "pro-poor." Our results indicate that (i) net buyers of wheat are poorer than net sellers of wheat, (ii) there are more buyers of wheat than sellers of wheat at all levels of income, (iii) the proportion of net sellers is increasing in living standards and (iv) net benefit ratios are higher for poorer households indicating that poorer households benefit proportionately more from a drop in the price of wheat. In light of this evidence, it appears that households at all levels of income benefit from food aid and that - somewhat surprisingly - the benefits go disproportionately to the poorest households.