Learning by investing

We test for learning among heterogeneous individuals who are engaged in bidding and who possess limited information about the auctioned assets. We use an extensive dataset taken from an auction through which state‐owned enterprises were privatized in the Czech Republic in 1993–94. To test for learning, we develop new measures of individual performance to accommodate the varying prices of assets available in six successive stages of bidding. We present evidence that learning took place among heterogeneous agents during a large‐scale multi‐stage auction that is unique in terms of its size, incentives and variation. The auction’s design, with multiple market periods, allowed agents to learn based on accumulated experience. As individuals had to pay a fee to participate, and as the potential gain was in the magnitude of several months’ salary, we conclude that large incentives were driving our results.