Exploring the Socio-economic Dimensions of Rainfall Variability on Subsistence Agriculturalists in Mozambique

The development literature has increasingly emphasized the role of climate variability in shaping the vulnerability of agrarian households. The growing interest in the intersections of economic and climate risk has pointed to the need for better understandings of how extreme weather impacts the well-being of subsistence farmers in less advanced economies. We explore these linkages in Mozambique, one of the world's poorest countries, which has high rainfall variability and a great degree of exposure to both droughts and floods. Utilizing satellite-based estimates of rainfall that we spatially analyze within a GIS, we establish a 12-year rainfall climatology and calculate monthly rainfall anomalies for 551 villages during three growing seasons from 2002 to 2005. We also approximate storm-total rainfall from tropical cyclones entering the Mozambique Channel. We find that rainfall patterns vary from season to season over the study period, and no season closely approximates the 12-year climatology. Hierarchical cluster analysis places the villages into nine regions according to wet season monthly rainfall anomalies and rainfall received from Cyclones Delfina (2002-03) and Japhet (2003). Then, using longitudinal household data from the National Agricultural Survey of Mozambique conducted in 2002 and 2005, we regress household income change on various measures of changing household dependency on subsistence agricultural production and other demographic and social factors over the study period. Our method expands on other socio-economic studies by defining regions according to shared weather patterns as opposed to more standard political administrative boundaries. The model was estimated via ordinary least squares for all nine rainfall regions. This method allows us to investigate the differential effects of climate on economic well-being while avoiding problems of endogeniety regularly encountered when estimating the effects of weather on the income of rural agriculturists. We find that, in general, an increasing household reliance on subsistence agriculture is associated with declining incomes, particularly for regions where monthly rainfall varies the most. Only in the region affected by Japhet did our measure for change in subsistence agricultural production as percentage of overall income have a positive and statistically significant relationship. This suggests that cyclone-related rainfall may benefit areas experiencing drought in certain cases. In the region where Cyclone Delfina produced more than 150 mm of rain, gender has a statistically significant negative parameter estimate, indicating that female-headed households may face more difficulties in coping with extreme weather events associated with some tropical cyclones. Taken together, our findings suggest that day to day rainfall variability may be as detrimental as extreme weather events to household well-being in rural Mozambique. Future detailed analyses of daily or monthly rainfall variability in rural agricultural regions of the developing world could identify areas of high economic vulnerability, even in the absence of extreme events.