Status, patterns, and microeconomic drivers of the extent of diversity in crop production: Evidence from Afghanistan

We analyse the microeconomic determinants of the extent of crop diversification in Afghanistan by employing instrumental variable Tobit analysis using data from a cross-section survey of 8,613 farm households in 364 districts in all 34 provinces collected in 2013/14. The estimates of the Composite Entropy Index (CEI) show a relatively low level of crop diversification. Regression results indicate that landholding size, access to sufficient irrigation water, ownership of tractor, oxen, and cattle by the farm households, household size, landscape, and quality of land significantly increase the level of crop diversification. A significantly lower degree of crop diversification is found for farm households with higher off-farm income, and farmers living in communities with low access to all-season drivable roads. Since off-farm income is highly likely to be associated with the unobserved household characteristics such as household entrepreneurial skills and risk preferences which are omitted from the regression analysis, one would expect biased estimates of the relevant coefficients. We allow off-farm income to be endogenous and use Instrumental variable (IV) Tobit analysis revealing that endogeneity masks the true effect of off-farm income; the negative impact of off-farm income on the level of crop diversification is even greater when endogeneity is accounted for. This is consistent with the hypothesis that risk-aversion behaviour of famers generates an upward bias in the coefficient of off-farm income if endogeneity is not allowed for. Unobserved risk-aversion behaviour drive household's decision to diversify into both nonfarm income and crop mix.