Price clustering in initial public offerings (IPOs) is extreme, pervasive, and long-lived. In a 25 year sample of 7,805 IPOs, offering prices range form one tenth cent to $5,000 per share. However, the price frequency distribution yields a clear triple peak with three discrete values, $5, 10, and $15 per share, accounting for 21.5% of all IPO prices. Clustering pervades every one of the top 20 two-digit SIC code groupings and every year from 1970 through 1994. Firms, underwriters and investors appear to have such consistent preferences for certain nominal prices per share that the mean IPO price per share dropped by 67.7% in real terms during the 25 year period. Additionally, the median IPO price has lagged inflation; it was $10 per share in 1994 as it was in 1990, 1989, 1983, 1977, 1973, 1972, and 1971. Two-day returns are 147 basis points (13%) higher for IPOs priced at $10 per share than otherwise similar firms. These results may indicate that underwriters choose the more highly-rounded prices to signal less precise valuations and that shareholders may earn higher returns for greater uncertainty. These findings build on Harris (1991) concerning clustering in seasoned stocks.

The agency literature in corporate finance rests for the most part on the assumption that highly diffuse ownership is commonplace and managerial ownership small; the result is costly for shareholders. Evidence provided here is to the contrary. Even where ownership is dispersed, managerial ownership is often significant and the mean and ownership statistics are surprisingly large. Furthermore, there is not strong consistent relation, linear or non-linear, between ownership and valuation or profit. These relations vary greatly over time. These findings support Demsetz and Lehn (1985) and contradict the agency arguments in Jensen and Meckling (1976), Morck, Shleifer, and Vishny (1988) and McConnell and Servaes (1990).

This paper supports and extension of the negotiation hypothesis begun in Harris (1991) and calls into question the evidence of implicit collusion developed in Christie and Schultz (1994). This paper documents the distribution of prices per share for initial public offerings, mergers and acquisitions, and self-tender offers. These markets were selected because they lack the potential for bid-ask collusion. Despite the lack of bid-ask collusion in IPOs, M&As, and self-tenders, prices are often quoted in rounded amounts per share – avoiding quotes ending in add eighths, off quarters, and odd halves. Two new hypotheses are offered to explain the lack of off-eighths quotes and other previously unexplained pricing phenomena.