Uncertainty, Imperfect Information, and Expectation Formation over the Firm's Life Cycle

Using a long-panel data set of Japanese firms that contains firm-level sales forecasts, we provide evidence on firm-level uncertainty and imperfect information over their life cycles. We find that firms make non-negligible and positively auto-correlated forecast errors. However, they make more precise forecasts and less auto-correlated forecast errors when they become more experienced. We then build a model of heterogeneous firms with endogenous entry and exit where firms gradually learn about their demand by using a noisy signal. We present our novel approach to cleanly isolate the learning mechanism from other mechanisms by using expectations data over time. We combine the model with our data to perform a non-parametric decomposition of forecast errors and find that learning accounts for between 20% to 40% of the overall decline in forecast errors over the life cycle. Our model shows that, within the context of our cross-regional data, productivity gains from removing information frictions ranges from 3% to 12%. We find a prominent role of firm entry and exit in generating high productivity gains.