ABSTRACT
Many agrarian economies are associated with seasonal patterns of income and consumption expenditure. Chambers (1982) notes that many poor people in developing countries live in tropical environments characterized by wet–dry seasonality with varying levels of income, food availability, incidence of disease and supply of and demand for labour. However, studies that model the determinants of poverty tend to ignore the issue of seasonality in the estimates of poverty even though the household-level data used in the analysis are collected over a long period of time. Several studies looking at the determinants of poverty in different African countries do not account for seasonality (Grootaert, 1997; Geda et al., 2005; Datt et al., 2000; Mukherjee and Benson, 2003; Okwi et al., 2007). Geda et al. (2005) acknowledge that the data collection did not take seasonality into account, and therefore could not control for seasonality in the estimation of the determinants of poverty in Kenya. Other studies, however, explicitly discuss the role of seasonality in household welfare. Dercon and Krishnan (2000), for instance, use seasonal price indices and seasonal wages to capture seasonality and these proxies were highly significant with per capita consumption increasing in peak labour periods and decreasing in high price periods. Appleton (2002), using a nationally representative survey in Uganda, controls for the time of interview in regression models and finds it statistically significant in welfare and poverty functions. Similarly, Khandker (2009) controls for a season associated with food deprivation and finds a negative and significant impact on welfare in Bangladesh.
