Divergence of Opinion, Overallotment, and IPO Long-Run Performance

Abstract

We apply an ex ante measure of divergence of investor opinions – industry-level dispersion in analyst earnings forecast – to test the Miller (1977) prediction about IPO overvaluation. Our measure of divergence of opinion is the dispersion in analyst long-term earnings growth forecasts divided by the absolute value of the mean long-term growth forecast averaged at the industry level in the month prior to the IPO. We focus on the effect of divergence of opinion on long-run IPO performance, and test two hypotheses. First, greater divergence of opinion should predict lower long-run performance. Second, since the Miller effect depends on the demand of the offer relative to the supply, we further predict that the effect of belief dispersion should be stronger for offers with high investor demand, as measured by issue overallotment at the time of the offer. Using a sample of about 7,000 IPOs from 1982 to 2005, we find that IPOs in industries with high divergence of opinion have lower long-run returns in the following 3-5 years than IPOs in industries with low divergence of opinion. The effect of belief dispersion on long-run performance primarily exists among IPOs with high demand, as measured by the overallotment at the time of the offer, and this effect remains strong when returns are value-weighted. These findings support Miller’s hypothesis that in markets with restricted short-selling, valuations reflect the most optimistic investor’s appraisal in the short-run, and the initial overvaluation is corrected in the long-run. There is evidence that IPOs with high divergence of opinion also have lower first-day returns, but this association is weakened by the amount of overallotment, suggesting that the correction of initial overvaluation starts shortly after the offer, but the initial correction is dampened by investor demand at the time of the offer.