Elsevier

Food Policy

Volume 26, Issue 4, August 2001, Pages 437-452
Food Policy

Determinants of income diversification amongst rural households in Southern Mali

https://doi.org/10.1016/S0306-9192(01)00013-6Get rights and content

Abstract

Household-level panel data from a representative sample of rural households in Southern Mali describing the different sources of household income are used to examine the determinants of income diversification. A conditional fixed effects logit model is employed in the analysis to control for household-specific effects. We find evidence that poorer households have fewer opportunities in non-cropping activities such as livestock rearing and non-farm work, and hence less diversified incomes. This appears to reflect their relative lack of capital, which makes it difficult for them to diversify away from subsistence agriculture. The results also indicate that households in remote areas are less likely to participate in the non-cropping sector than their counterparts closer to local markets, while households with educated heads are more likely to participate in the non-farm sector than those with illiterate heads. The significance of entry-constraints in explaining portfolio diversification suggests that the role of government in making assets — as well as improved infrastructure — available to poorer households is still essential in promoting income diversification.

Introduction

Rural households in Sub-Saharan African countries usually have to cope with both poverty and income variability. The strategies households in this area often use to smooth income and consumption have been a topic for research in recent times (see, e.g., Reardon et al., 1992, Fafchamps, 1992, Udry, 1995, Carter, 1997). Some authors have argued that the lack of complete insurance and credit markets compels households to devote substantial resources to stabilizing the stream of income in order to protect themselves from the dire consequences of substantial income fluctuations (Bardhan and Udry, 1999).

Diversification of income sources has been put forward as one of the strategies households employ to minimize household income variability and to ensure a minimum level of income (Alderman and Paxson, 1992). In this paper, income diversification refers to the allocation of production assets among different income-generating activities, both on- and off-farm.1 Reardon et al. (1992) argue that in the failure or near absence of consumption and crop insurance markets, and given the shortcomings of an ineffectual “social net”, households must turn to income diversification. As a risk strategy, income diversification is usually taken to imply a trade-off between a higher total income involving greater probability of income failure, and a lower total income involving smaller probability of income failure. In other words, risk-averse households are willing to accept lower income for greater security (Ellis, 2000).

Empirical studies from some countries have supported the hypothesis that income diversification is linked to lowering risk. For example, findings from a study conducted by Murdoch (1990), using a sample of Indian farm households indicate that poorer households do choose less risky production strategies than do other households. But recent work in Sub-Saharan Africa appears to suggest that the poorest households are less diversified, despite the fact that they should be more risk-averse (Dercon and Krishnan, 1996, Barrett et al., 2000). Hence, Dercon and Krishnan (1996) point out that risk aversion, combined with credit and insurance market imperfections, is not the only reason for pursuing a diversified portfolio of activities. Barrett et al. (2000) argue that if households faced identical prices and constraints, then observed differences in diversification patterns could occur purely from measurement error or differences in preferences, such as with respect to risk. However, spatial variation in transaction costs and gross market prices lead to cross-sectional heterogeneity in incentives due to differential market access2.

Households willing to diversify might not have the capacity to do so because of the above-mentioned binding entry-constraints. For example, Reardon (1997) found in a review of 27 case studies in Africa that the share of non-farm income in total income in these surveys varies from 15 to 93%. He also reported that non-farm earnings were regressively distributed, with those earning the greatest farm income also enjoying a higher absolute level of and share of total income from non-farm sources than did the poor. A number of empirical studies in this area have therefore emphasized the significance of factors other than household's behavior towards risk (Dercon and Krishnan, 1996).

Given the key role income diversification can play in stabilizing incomes and alleviating rural poverty, governments in developing countries have become increasingly interested in promoting increased output diversification (Petit and Barghouti, 1992). Ellis (2000) argues that household-level diversification has implications for rural poverty reduction policies since it means that conventional approaches aimed at increasing employment, incomes and productivity in single occupations, like farming, may be missing their targets. In the African context, some of the studies mentioned above have made comprehensive efforts to identify the constraints to income diversification (Reardon et al., 1992, Dercon and Krishnan, 1996), with the objective of identifying measures that will help alleviate those constraints. The findings from these studies appear mixed, warranting further empirical investigation to shed more light on the forces driving income diversification in the region. The purpose of this study is to contribute to the empirical literature by examining the income portfolios of households in Southern Mali and analyzing the determinants of income diversification in that area. As in Dercon and Krishnan (1996), the study emphasizes the significance of factors other than household's behavior towards risk.

Most of the studies mentioned earlier have used data for cross-sectional samples and as such have not been able to control for household-specific effects — possibly unobservable — which may be correlated with other included variables in the specification. In the presence of such correlations, ordinary least squares (OLS) and generalized least squares (GLS) yield biased and inconsistent estimates of the parameters. For example, in their analysis of income diversification in Burkina Faso, Reardon et al. (1992) used OLS method that did not account for unobservable household-specific effects. Given the limitations of these studies, we utilize a conditional fixed effects logit model to control for such individual effects in our analysis of the determinants of income diversification.

The paper is organized as follows. In the next section, we discuss the setting in which the data was collected and the evidence already available. A descriptive analysis of the data then follows. In Section 3, we present a simple model of activity choice. Section 4 contains our empirical results and the final section our summary and conclusions.

Section snippets

Data and geographical setting

The data used in this study come from a farm household survey that was carried out in Southern Mali3. The survey covered the three farming years of 1993/94–1995/96. This study uses the data from the last two agricultural seasons of the survey (1994/95–1995/96), for which complete information is available

Income sources in Southern Mali

Table 1 provides information concerning income sources. The table shows that crop production remains the principal source of income for households in the sample — roughly 70% on average. Cotton accounts for 44%, while food-crops accounts for just 26% of total household income per capita. On average, non-cropping income constitutes about 30% of total income. This proportion is closer to the figure reported by Reardon et al. (1992) for the Sudanian zone of Burkina Faso — 26%. Much of household's

Empirical model of income diversification

Our empirical approach focuses on the significance of factors other than household's behavior towards risk in explaining the household's resource allocation over time. As indicated earlier, decreasing absolute risk aversion requires that poorer households diversify their sources of income than their wealthier counterparts. However, poorer households facing various sources of risk but with little insurance mechanism to rely on, are likely to devote substantial resources to food self-sufficiency

Estimation

A conditional fixed effects logit model is employed to examine the likelihood of a household holding one of the three possible income portfolios outlined in the previous section. The underlying equation for the logit model isqit=βXititwhere qit is an unobservable latent variable for participation in any of the three activities mentioned, Xit is a vector of explanatory variables, β is a vector of parameters to be estimated, and ηit is the error term. The subscripts i and t index the

Empirical results

Table 3 presents estimates of the portfolio choice model for the three activities. All utilize the Chamberlain technique11. The results include the parameter estimates, the b̂ values, with their asymptotic t-statistics in the left half of each column, and estimated marginal effects at the mean, the γ̂ values, in the second half. A household γ̂i is defined as the derivative of the probability with respect to Xi,

Summary and conclusions

In this paper we have examined the income portfolios of farm households in Southern Mali, focusing on four main activities: food-crop production; cash-crop production; livestock rearing; and non-farm work. The results show that households hold very different portfolios of incomes and these in turn are related to the different levels of income and asset holdings. As Dercon and Krishnan (1996) have pointed out, while some part of the explanation for these observed differences can be attributed to

Acknowledgements

This paper was prepared while the first author was visiting fellow at the Economic Growth Center, Yale University, whose hospitality is gratefully acknowledged. The authors wish to thank Thomas Reardon, Christopher Barrett, Douglas Gollin, Christopher Udry and three anonymous journal reviewers for their detailed comments and suggestions on an earlier version of the paper. The assistance of the Malian Textile Development Company during the data collection is also appreciated.

References (23)

  • A Abdulai et al.

    Estimating labour supply of farm households under non-separability: empirical evidence from Nepal

    Agricultural Economics

    (2000)
  • M Fafchamps et al.

    Drought and savings in West Africa: are livestock a buffer stock?

    Journal of Development Economics

    (1998)
  • A Abdulai et al.

    Determinants of nonfarm earnings of farm-based husbands and wives in Northern Ghana

    American Journal of Agricultural Economics

    (1999)
  • A Abdulai et al.

    The impact of agricultural price policy on cocoa supply in Ghana

    Journal of African Economies

    (1995)
  • Alderman, H., Paxson, C.H., 1992. Do the poor insure? A synthesis of the literature on risk and consumption in...
  • P Bardhan et al.

    Development Microeconomics

    (1999)
  • Barrett, C.B., Benuneh, M., Clay, D.C., Reardon, T., 2000. Heterogeneous constraints, incentives, and income...
  • W Berckmoes et al.

    L'intensification agricole au Mali-Sud. Souhait ou réalité?

    KIT/DRSPR/IER, Les bulletins de l'Institut Royal des Tropiques

    (1990)
  • M.R Carter

    Environment, technology, and the social articulation of risk in West Africa agriculture

    Economic Development and Cultural Change

    (1997)
  • G Chamberlain

    Analysis of covariance with qualitative data

    Review of Economic Studies

    (1982)
  • De Groote, H., 1994. Women's income versus family income as a determinant for food security, an example from Southern...
  • Cited by (196)

    • Distributive impacts of non-farm income on output and farm income of cassava farmers in Southwestern Nigeria

      2023, Scientific African
      Citation Excerpt :

      Farmers with access to credit are better placed to overcome financial constraints and tend to engage in non-farm works. In support of this finding, Abdulai and Crolerees [2] reveal that the nonexistence of an effective formal credit market in developing countries is one of the factors affecting the development of non-farm activities in those countries. Owusu et al. [78] also found a positive effect of credit access on participation in non-farm works in Ghana.

    View all citing articles on Scopus
    View full text