آیا کمک های غذایی دارای اثرات بازدارنده در تولید داخلی است؟ چشم انداز تعادل عمومی در کمک های غذایی در اتیوپی

This paper examined impacts of food aid on domestic food production employing a computable general equilibrium modelling technique and using data from Ethiopia. The simulation experiments have shown that food aid has unambiguous disincentive effects on domestic food production. The removal of food aid caused a modest increase in food prices but this stimulated food production. Employment and income generation effects of the latter outweighed the adverse effect of the former. Consequently, the removal of food aid led to improvements in aggregate household welfare. Contrary to some concerns in the food aid literature that any reduction in food aid would hurt the poor, the simulation experiments suggested that actually poor rural households and urban wage earners are the ones who benefit most in absence of food aid but entrepreneurs are more likely to encounter a marginal welfare decline. We have distinguished between in-kind food aid and cash equivalent transfers in order to isolate the disincentives that in-kind transfers would make to domestic production from those that are related to household purchasing power problem. The expansionary effect of removing food aid becomes significantly larger when it is accompanied by cash equivalent payments because the latter would provide demand side stimulus to agriculture while the removal of in-kind transfers would stimulate supply side, with the supply and demand side effects reinforcing each other. In our modelling framework, the only adverse effect would be a modest deterioration in the external current account, because the expansionary effects of food aid would cause imports to rise but exports to fall.

There has been a growing need to explain the role of food aid in alleviating poverty in less developed countries. Many researchers have critically examined associations between the level of food aid and rising level of poverty in the recipient countries. Although food aid has been a heavily researched topic, most studies have assessed only one aspect of the transfer such as: food aid’s impact on domestic agricultural production (Schultz, 1960); domestic price (Gabre-Madhin et al., 2003; Barrett, 2003; Barrett et al., 1999), factor supply (Barrett and Clay, 2003 and von Braun et al., 1999; Holden et al., 2003; Bezuneh et al., 1988) relative efficiency of in-kind food aid and cash payments (Sen, 1986, Coate, 1989, Basu, 1996 and Faminow, 1995); and whether or not food aid reaches target households (Clay et al., 1999 and Jayne et al., 2001). The impact of food aid on domestic economy in the recipient country are mostly analysed in a partial equilibrium context. However, the existence and importance of system-wide effects of food aid are widely acknowledged in the literature. In spite of some attempts at developing some analytical frameworks tracing general equilibrium effects of food aid as early as 1980s (Bhagwati, 1985), general equilibrium analysis has rarely been applied to food aid issues. This is despite the fact that there has been vastly increased application of computable general equilibrium (CGE) on a wide range of policy analysis in LDCs in recent years. Few existing CGE applications on this subject so far have generally focused more on assessment of general food aid requirements (Wobst, 2001 and Fontana et al., 2005) or specific food aid targeting (Arndt and Tarp, 2001) than disincentive of food aid on domestic food production and the feed-back effects of this on the rest of the economy. This paper would make a modest attempt to fill this gap in the literature. We concentrate on the disincentive of food aid on food production in Ethiopia and employ a computable general equilibrium modelling approach, relying on a social accounting matrix and associated database created recently for Ethiopia through a project sponsored by the World Bank. More specifically, the general equilibrium analysis focuses on differential impacts of in-kind transfers and cash equivalent payments. It is found that cash payment is by far the most effective form of aid because it simulates domestic production, factor employment and household welfare. The paper is structured as follows. The first section provides an overview of food aid to Ethiopia. The second section highlights key features of the CGE model developed for this study with a focus on the specification of food aid. Section Simulation results are discussed the third section. Concluding remarks are made in a final section.

This paper has examined whether or not food aid has disincentive effects on agricultural production in Ethiopia. It employed a computable general equilibrium modelling approach, a suitable technique to account for feed-back effects of such policy shocks and the relationship between food aid and food production in Ethiopia. We have distinguished between tied food aid (that is in-kind food aid) and cash equivalent payments. Accordingly, the impact of the removal of tied food aid (in-kind transfers) was simulated in two scenarios: no cash equivalent transfer payments to households in scenario 1 and cash equivalent transfer payments to households in scenario 2. The motivation behind this simulation set-up was to isolate the disincentive effects of additional grain supply in the agricultural commodity market from the purchasing power problem of households. The simulation experiments suggested that the impact of the policy shock on the food producing sector of agriculture, the sector that receives the exogenous shock, was positive in both scenarios. The removal of food aid generates a demand stimulus for agriculture and hence the food producing sector expands by 2.2% in scenario 1 and by 4.5% in scenario 2. It is interesting to note that the differential impacts between the two scenarios: the removal of food aid has smaller expansionary effects in scenario 1 because of limited purchasing power of households or household budget constraints. In this scenario, the food crops sector was the only commodity group that experiences a positive impact with most other agricultural and non-agricultural sectors encountering adverse effects although mostly by negligible proportions. However, in scenario 2, when households receive cash equivalent transfer payments, then not only that the food sector experience a larger expansion but also most other sectors receive positive stimulus which comes via inter-sectoral multiplier effects. The simulation results indicate that the overall impact of removing food aid is positive with significant expansion in factor employments in agriculture. Employment rises by 4% and 2% in agriculture and non-agricultural sectors, respectively. Moreover, agricultural land use expands by about 1.4%. In terms of aggregate macroeconomic effects, expenditure measure of GDP marginally rises by 0.45% and this is a net outcome of expansionary effects on household consumption and imports by (increases by 1.64% and 2.36%) but an adverse export effect, a decline by 3.76%. As we expect with most macroeconomic policy shocks, this shows a conflicting outcome from the policy shock occurring due to the removal of in-kind food aid, with its expansionary effect on the domestic economy causing imports to rise but exports to fall; because diversion of exports for domestic use would have an adverse consequences on Ethiopia’s external current account. The finding of this study supports the disincentive hypothesis originally put forward by Schulz (1960). It is useful to compare our results with studies cited earlier that have used Ethiopian data, employed descriptive or econometric techniques and have rejected the disincentive hypothesis. For instance, the key to Levinsohn and McMillans’ analysis was the assumption that food price in Ethiopia would increase considerably if food aid is abolished. They estimated this using a single commodity (wheat) demand and supply equations and predicted that food price would rise by 52.8% if food aid was removed from the system. However, we have employed a multi-commodity and multi-sectoral general equilibrium modelling framework, accounting for multiplier effects, and have shown that food prices rise by a maximum of 2.51%. The adverse effect of this increase in food price was outweighed by the positive effects to agriculture and the ripple effect of this throughout the economy. Similarly, Levinsohn and MicMillan have argued that food aid benefits poor households than rich ones but we have argued that actually the opposite is likely to be the case. Our simulation experiments have indicated that poor rural households are likely to benefit much more than any other household group. Also, “net-buyers” of food, such as wage earners, do not necessarily get hurt as a result of abolishing food aid, because the expansionary effect of removing food aid in terms of generating employment and income outweighs any adverse effect that comes from food price increases. In our simulation experiments urban entrepreneurs are the only group who encounter minor decline in welfare mainly because increase in agricultural prices would squeeze their profit margins as cost of raw materials increase. It should be noted that the aggregate nature of the model used for this study allows only a cautious policy recommendation in favour of cash transfers. As noted earlier, if cash transfer causes regionally differentiated food price increases, then welfare losses may be suffered by groups of households who are not targeted. Firmer policy recommendations will be made with future improvement of this model with its disaggregation both spatially and across different beneficiary categories.