A structural empirical model of firm growth, learning, and survival

In this paper we develop an empirical model of entrepreneurs' business continuation decisions, and we estimate its parameters using a new panel of monthly alcohol tax returns from bars in the state of Texas. In our data, entrepreneurial failure is frequent and predictable. In the first year of life, 20% of our sample's bars exit, and these tend to be smaller than average. In the model, an entrepreneur bases her business continuation decision on potentially noisy signals of her bar's future profits. The presence of noise implies that she should make her decision based on both current and past realizations of the signal. We observe for each bar its sales, which we assume, equals a noisy version of the entrepreneur's signal. That is, the entrepreneur's information about her bar is private. ; The entrepreneur's private information makes the estimation of our model challenging, because we cannot observe the inputs into her decision process. Nevertheless, we are able to recover from our observations the parameters characterizing the entrepreneur's learning process and the noise contaminating publicly available sales observations. The key to our analysis is to note that our ability to forecast the entrepreneur's decisions reveals the amount of noise contaminating publicly available sales observations. We infer that public and private information differ little if we can forecast entrepreneurs' business continuation decisions well. With this information, we can then determine whether the usefulness of past sales observations for forecasting future sales arises only from the noise contaminating public observations or if the observations imply the presence of additional noise contaminating entrepreneurs' observations. ; We estimate our model using observations from the first twelve months of life for approximately 300 Texas bars. We find that entrepreneurs observe the persistent component of profit without error. In this sense, their information is substantially superior to the public's.