Climate Change and Seasonal Variability in Rice Production: Evidence from Sri Lanka

This paper examines the impact of seasonal climate on rice yield in Sri Lanka and uses district-level panel data covering the period from 1979 to 2024. Yala and Maha seasons are considered separately, and a fixed-effects Spatial Durbin model with a queen contiguity matrix is used to capture both local and spatial spillover effects. In the Yala season, spatial spillovers are predominant: increases in harvested area boost local yields, but expansion into neighbouring districts reduces yields. Significant positive spillovers stem from minimum temperature and rainfall in adjacent districts, while the interaction of rainfall extent shows negative spillovers. Although most individual weather variables have insignificant direct effects, their total effects are positive and significant, reflecting the overall influence of climatic conditions across the spatial system. In the Maha season, rice yield is mainly influenced by local factors: harvested area and rainfall have notable positive direct effects, with minimal spillover effects from neighbouring districts. Night-day humidity differences negatively impact local yields. Cross-seasonal spatial models show that rainfall in the earlier Maha season positively affects Yala yields, and humidity differences in the Yala season negatively affect the next Maha season rice production. Our estimates indicate that climatic variations in one season can affect another season, underscoring the need for seasonal, spatial interdependencies considerations for policy-making and implementing climate adaptation strategies.