We develop a model of firm learning in volatile markets with noisy signals and test its predictions using historical German data. Firms’ forecasts improve with age. We exploit German Reunification as a natural experiment where firms in the East are treated with ignorance about the distribution of market states. As theoretically predicted, Eastern firms forecast worse than Western ones, but this gap gradually closes over the decade following Reunification. The slow convergence stems from differences in expectations rather than market conditions. We find evidence for the model’s predictions that improvements from learning are faster where market signals are noisier.
expectation formation learning business cycle forecasting transition dynamics