We show that the data strongly support the model’s predictions about bidding behavior. Petrobras’ bids should have been highly correlated with profitability, while its competitors’ should not. Thus, notification of hydrocarbons should be positively correlated with Petrobras’s valuation of tracts and orthogonal to its competitors’. Indeed, Petrobras notified that it found hydrocarbons in 95 out of the 403 tracts that it won, which is almost 50 percent higher than the same ratio for its competitors, equal to 63 out of 400. Moreover, the ratio between Petrobras’ average winning bid for tracts where hydrocarbons were eventually detected and the average of all its winning bids is larger than the same ratio for its competitors.
We also test a prediction that explores the fact that we observe many tracts being offered repeatedly. We start from the observation that if Petrobras held superior information that contained any knowledge its competitors might have, then any information that the latter might have revealed through their bids in previous auctions should not affect how Petrobras bid in future auctions for the same tract. On the other hand, Petrobras’ bid in an auction should have affected its competitors’ bids in future auctions for the same tract because the latter’s bid contained some of its private information about the tract’s value that should have caused its competitors to adjust their own estimates of it. Because these repeated tracts had been offered in a past round, received no bids and were offered again later, this reasoning predicts that Petrobras should have valued tracts that had been offered before relatively more than its competitors. Bidders’ participation decisions support our prediction about how bids relate to whether a tract had been offered before. For both onshore and offshore tracts, the ratio between auctions for tracts that were never offered that received bids from Petrobras and from its competitors is lower than for tracts that had been offered before, which indicates that Petrobras valued tracts that had been offered before as opposed to tracts that had never been offered relatively more than its competitors.