Weather shocks, economic growth and damage function for India: A varying coefficient semi-parametric approach

Weather shocks associated with global climate change engender substantial damages, in the order of multi-billion dollars annually, to the Indian economy. Using data from 33 states during 1981--2022, we explore the effect of weather shock on India's economic growth, in the presence of interplay of temperature and precipitation levels. To avoid arbitrary assumptions of parametric estimation, we estimate the economic damages resulting from weather shocks using semi-parametric varying coefficient generalised additive models (VC-GAM). We select the optimal class of VC-GAM among 29 possible classes based on four relevant criteria. From the optimal class, out of 84 possible specifications, we determine the optimal damage specification using the out-of-sample and in-sample performance. We find that the contemporaneous year-on-year weather change and lagged year-on-year precipitation change have an impact on the per capita economic growth through total factor productivity channel, whereas only contemporaneous precipitation level have an impact on the per capita economic growth through labour productivity channel. We observe that the marginal effect of a contemporaneous weather change varies with the level of lagged precipitation level, whereas high lagged precipitation level combined with a low to moderate lagged temperature level exacerbates the detrimental impact of a positive lagged precipitation change on the per capita economic growth for India. One potential mechanism through which contemporaneous and lagged weather variables could have an impact on the per capita economic growth, is based on the impact of soil moisture quality. We have demonstrated our results to be considerably robust.