Research (working papers)

Abstract

I examine whether recency bias—forming expectations from recent experiences— induced by recent rainfall shocks reduces fertilizer use among maize farmers in Nigeria. Economic theory predicts that a profit-maximizing farmer should not be influenced by recent weather shocks except through liquidity constraints. Using nationally representative panel data merged with historical rainfall data, I show that, conditional on past profit (proxied by livestock accumulation and prior season's maize price), recent rainfall shocks significantly influence current fertilizer use even though shocks are serially uncorrelated and transitory. Importantly, I find that this effect is negatively asymmetric: prior negative shocks depress fertilizer use more than prior positive shocks raise it, thereby lowering average fertilizer use. Back-of-the-envelope calculations imply that weather-induced recency bias is associated with about 5\% reduction in the already low fertilizer use (56.2 kg/acre) in Nigeria. Improving access to accurate and timely weather forecasts could help households make more efficient input use decisions.