Empirical evidence suggests that better-informed investors in bookbuilt IPOs submit more informative bids and receive better allocations than do investors with less precise information. While the traditional bookbuilding argument accounts for this evidence as better-informed investors being rewarded with more favorable allocations for providing more useful information, the present paper adopts the winner's curse argument and shows that better-informed investors get better allocations by being better able to pick underpriced issues, even though in equilibrium investors' bids fully reveal their information. The paper offers empirical implications that allow the two arguments to be separated.

The empirical evidence on initial returns in IPOs reveals average overpricing as well as underpricing, depending on the type of security offered for sale. Consistent with this evidence, the present paper develops a model in which an IPO may be overpriced in equilibrium relative to its expected (or average) aftermarket price. Overpricing disappears, however, once the offer price is compared to a ‘float-weighted’ expectation, where the weights are given by the extent to which the number of securities that are floated in the offering (at the posted price) is positively related to the demand for allocations. Empirically, the model implies that equally-weighted returns underestimate initial returns in IPOs, and hence that inferences based on equally-weighted returns may be misleading.

Extant literature shows that IPO first-day returns are correlated with market returns preceding the issue. We propose a rational explanation for this puzzling predictability by adding a public signal to Benveniste and Spindt (1989)’s information-based framework. A novel result of our model is that the compensation required by investors to truthfully reveal their information decreases with the public signal. This “incentive effect” receives strong empirical support in a sample of 6300 IPOs in 1983–2012. Controlling for the incentive effect, the positive relation between initial returns and pre-issue market returns disappears for top-tier underwriters, where the order book is held to be most informative, effectively resolving the predictability puzzle.