Field Exchange Nov 2002: Dynamics of livelihood diversification in post-famine Ethiopia

Summary of published paper1
Income diversification has been shown to be positively associated not only with wealth accumulation but also with an increased ability to withstand exogenous shocks. The commitment to diversification as an explicit objective within livelihood development strategies assumes that poorest households in risky environments can, and indeed want, to avail themselves of opportunities presented to diversify. As a result, the promotion of off-farm employment as policy has gained widespread support across a spectrum of development agencies from the World Bank to INGOs, especially in countries facing repeated income and consumption shocks. However, according to a recently published study, little is known about how shocks affect diversification over time as households reconsider past choices and adapt to new conditions.

This study sought to identify relevant associations between income diversification, household perceptions of livelihood risks and changes in consumption outcomes over time in post-famine Ethiopia. Four key questions were explored:

i. to what extent did households emerging from the famine period with relatively higher income and calorie consumption levels also have a more diversified income base

ii. was higher income diversification in 1989 associated with a greater increase in diversification by 1994, using a household's ranking according to income or consumption

iii. which households increased their share of income from non-cropping activities most during the inter-survey years - those already strongly committed to diversification or those previously less diversified

iv. did household heads perceive a lack of non-farm income activities to be an important risk factor in famine vulnerability, and did such perceptions play an identifiable role in determining the nature of their income portfolio.

Survey context and approach

In 1989/90 over 500 households were interviewed in 7 communes located in different parts of the country. Communities were purposively sampled according to a range of famine experiences, forms of public intervention, agro-ecologies and ethnic groups. Almost 300 of the original households were revisited in 1994 as part of an expanded survey. The five years straddled by the survey saw improvements in farm output that allowed for fairly rapid post-famine recovery. Part of the increase was due to improved weather and part due to political and economic reforms, e.g. market liberalisation, currency devaluation, land reforms and redistribution, and a period of greater donor support. Households in highland sites increased cereal and non-cereal yields by 40 and 400 per cent respectively. The effects were less dramatic in the lowlands where cereal yields fell and non-cereal yields only improved by 20%.

Cropping provided the main source of earnings for all income groups in both periods. Crop income as a share of total income was 68% among poorest households in the first survey, falling to 58% (statistically significant) in the second. For wealthier households crop income represented only 44% of total income, falling to 41% (not significant) in the second survey.

Mean income rose in real terms from $14 per capita to $39 among poorest households and from $233 to $519 among upper income households during the same period. The share of food in total household expenditure among poorest households was close to 90% in the first period, falling to 73% in the second. Among upper income households, the share fell from 70% to 54 per cent.

Findings and conclusions

Households surviving the famine with higher than average income and food consumption levels also had a more diversified income base and more valuable assets in hand (especially livestock). Analysis of the determinants of diversification at that point in time (1989) indicates that greater income diversification (out of cropping) was positively associated with per capita income level, higher dependency ratios, location in the highlands, and ownership of non-farm assets. No significant association was found between education (literacy) and changes in well-being over time, arguably because there were only minimal opportunities for formal salaried employment in the inter-survey period.

While many households perceived older age of household head and small farm size to be predictors of household risk, such an association was not confirmed empirically. However, most households did believe that earning income outside of cropping was a key to reducing risk. Yet, there was an inverse association observed between female-headed households and diversification despite the fact that female headed households were significantly more likely to believe that off-farm income protects a household against famine.

Nevertheless, the post-famine recovery period was a time of dynamism and change as highlighted by the movement between income and consumption deciles. This suggests that opportunities for change are neither uniformly spread nor captured only by the already wealthy. In fact the households that most increased their flow of non-crop income between the two surveys were the poorest and least diversified in the initial period. By contrast, the already well diversified (in 1989) maintained their level of diversification but still increased their income flow and asset wealth.

The authors summarise the findings by suggesting that complex trade-offs exist between perceptions of risk and diversification practices - trade-offs that vary considerably by household type and location. Furthermore, such differences in household choices and opportunities need to be better understood if development agents are to deliver effectively the benefits of off-farm employment policies to the poor (particularly women farmers) in risky environments.


1 Block S and Webb P (2001): The dynamics of livelihood diversification in post-famine Ethiopia. Food Policy, 26, pp 333-350