Precipitation and economic growth

The ongoing process of global warming goes along with changes in precipitation patterns around the globe. As a consequence, the question whether and how precipitation affects economic development has recently gained new interest. Numerous empirical studies have considered the effect of rainfall on agricultural productivity, either in the form of case studies, single country time-series analyses or, more rarely, in panel studies. However, they almost exclusively focus on the very short-run and on countries with a comparatively high importance of the agricultural sector. Interestingly enough, they nevertheless fail to deliver a clear picture as various studies find a positive effect of precipitation while others find the opposite or no effect. To a significant extent, this is due to the enormous variety in sample countries, sample periods, data sources, empirical methodologies and time horizons. Moreover, a major source of heterogeneity is the employed rainfall indicator.

This paper delivers a systematic empirical panel analysis of the short- but especially potentially occurring long-run effects of precipitation on economic growth. Our (unbalanced) sample covers the period of 1950 to 2014 and more than 150 countries around the globe, allowing us to deliver results in a worldwide perspective but also for several subgroups of countries. In order to solve the so-called overcontrolling-problem we run our estimations in a two-way fixed effects setting with heteroscedasticity and autocorrelation (HAC) corrected standard errors. We also control for spatial dependencies which likely play an important role in the context of spatially correlated rainfall. Most important, we make use of three most advanced measures of precipitation, standardized rainfall anomalies (SPA), the standardized precipitation index (SPI) and the standardized precipitation and evapotranspiration index (SPEI), which have yet rarely been used in the related literature. Besides using the indicators in their classical linear version in our regression approach, we also allow for asymmetric effects for comparatively dry and overly wet periods. Besides numerous stability tests, we also deliver results for three alternative rainfall datasets to study whether our results depend on the source of precipitation
data.

While rainfall has little effect in highly developed countries, we find highly robust empirical evidence for negative effects of rainfall shortages in poor country samples (while overly wet periods have no significant effect). These effects persist over longer periods and thus have the potential to affect the long-run development of the affected countries significantly. Interestingly enough, the effects are not driven by Sub-Saharan countries.