Rainfall shocks and intimate partner violence in sub-Saharan Africa

Global climate change makes extreme precipitation events likely to become more frequent and intense in large parts of Africa. We study the effect of rainfall shocks on intimate partner violence in sub-Saharan Africa. The analysis shows the presence of spatial autocorrelation in rainfall shocks, which compromises the exogeneity of rainfall shocks in many applications. We correct for the autocorrelation using spatial polynomials. In particular, we use three different estimation strategies. We first use the complete cross-sectional sample to analyze whether recent droughts are correlated with respondents' experience with intimate partner violence during the last year. We then use the nine countries with repeated surveys to construct a repeated cross-section analysis at the grid level. Finally, we use event history analysis on a time series constructed from the information provided by the abused women about when the violence first took place. We find no robust evidence that droughts increase intimate partner violence. Potential explanations are that the rainfall shocks do not affect spouses' power, or that the slow onset of the droughts allows for a calmer response to the crisis. We contribute to the wider literature on climate and conflict as many of the mechanisms, economic and psychological, that link climate to violence apply to both intimate partner violence and organized violence.