³ÉÈË´óƬ

Several important auction settings, including treasury auctions in Canada and the U.S., have the feature that some bidders (dealers) observe the bids of a subset of other bidders (customers). Quantifying the economic advantage that informationally advantaged bidders derive from this institutional feature requires that we empirically distinguish between private vs. interdependent values paradigms. Bidders with private values who obtain information about rivals' bids use this information to update their beliefs about the distribution of residual supply. With interdependent values, bidders also update their beliefs about the value of the good being auctioned. We use these differential updating effects to construct formal hypothesis tests of the presence of private vs. interdependent values. Using data from Canadian treasury auctions, we cannot reject the null hypothesis of private values in auctions of 3- and 12-month treasury bills. We also do not find evidence supporting the alternative hypothesis of interdependent values. We use the estimated model to quantify the value of observing customer bids to a dealer. We find that the extra information contained in customers' bids leads on average to an increase in payoff equal to 13 ? 35% of the expected surplus of dealers.