Firm Experimentation in New Markets

An important feature of the data on new exporters is the evolution over time of their export size and exit rate. As the exporting age of firms increases, their average export volumes grow and the exit rate decreases. In particular, we observe cases of new exporters expanding slowly initially, and switching to high levels of exports after some time. I propose a quantitative model of active learning that can explain these patterns. In the presence of demand uncertainty and high sunk costs of entry, the firm can postpone paying this cost and learn more about its demand by paying for the testing technology. This is the first model in the trade literature to allow for multiple period and varying sample size experimentation. The duration of the learning stage, the intensity of learning, and the total entry costs are determined endogenously. The model permits us to produce rich comparative statics predictions with respect to market characteristics, firm characteristics, and features of uncertainty.