Lit in Review: Three from Rural Ethiopia

From the American Journal of Agricultural Economics, November 2009, Vol. 91, No. 4
Ethiopia's Blue Nile Falls, right.

Tadesse and Shively, "Food Aid, Food Prices, and Producer Disincentives in Ethiopia", 942-955 ungated.
Methods: Seemingly unrelated and recursive regressions, 3 geographic markets [capitol, food-surplus, and food-deficit], 3 commodities [teff, wheat, maize], monthly: 1994-2006
Facts: "Food aid accounted for 9% of the country's cereal budget ... 1994-2006" and 16% in 2003, 1/3 of calories for 3/4 of the households in food-deficit areas.
Findings: Because of good market integration, producers in supply and deficit areas are affected equally, but disincentive effects do not apply to nontradables (such as the principle consumption food, teff), and are sensitive to a threshold of about 10%. They recommend that food aid should only be used in production shortfalls to reduce negative effects.
Areas for future research: Identifying time-and-context-specific elasticities of substitution between nontradables and tradables.

Dercon, Gilligan, Hoddinott, and Woldehanna, "The Impact of Agricultural Extension and Roads on Poverty and Consumption Growth in Fifteen Ethiopian Villages", 1007-21 via IFPRI.
Methods: Longitudinal survey, (1994x2, 95, 97, 99, 04), GMM-IV, Hhd fixed effects, poverty line is 'cost-of-needs
Facts: 85% of employment nationally (96% rural) is in agriculture. Ethiopia increased extension support over the last two decades while other countries and agencies have decreased it. Durable goods improved over time and school enrollment increased, though these are not accounted for in the regressions.
Findings: At least one extension visit reduces headcount poverty by 9.8 percentage points and increases consumption growth by 7.1%. Access to all weather roads reduces poverty 6.9 p.p. and increases consumption growth 16.3%. Poverty overall decreased from 48% to 36% and Gini coefficient dropped from 0.44 to 0.41 between 94-99. They recommend caution in interpreting the extension results.

Yesuf and Bluffstone, "Poverty, Risk Aversion, and Path Dependence in Low-Income Countries: Experimental Evidence from Ethiopia", 1022-37.
Methods: Respondents from Amhara [Amharans, right] were asked which of six coin-flip bets they wanted to accept (gains only), then received a payoff accordingly. This then becomes the dependent variable for random effects and ordered probit models. A gains and losses bet was also volunteered to the large winners from the gains-only portion (226 of 262 players).
Findings: 28% of the populace exhibit extreme relative risk aversion (15). Risk aversion increases (stat. significantly) with the payoff, when there are potential real losses, and decreases with family wealth (oxen ownership). In the poorer district, the median respondent is severely risk averse. Past failure at the games also increases risk aversion in later rounds.