410 A. Ljungqvist
hand, assume that the IPO is designed topreventa transfer of control in spite of the
partial transfer of ownership. Who is right? The empirical evidence is more nearly con-
sistent with the staged-sale notion.Pagano, Panetta, and Zingales (1998)document that
most Italian IPOs are followed by private sales of controlling blocks to large outside
investors. Indeed, control turnover is twice as common in newly listed firms as in the
universe of unlisted companies. In the U.S., control turnover in the first five years is
29 percent in IPO firms with at least five years of trading history prior to flotation and
13 percent for younger companies (Mikkelson, Partch, and Shah, 1997). Similarly, offi-
cers and directors in U.S. IPOs on average own 66 percent of equity before the IPO and
44 percent immediately afterwards, which is reduced to 29 percent over the subsequent
five years, and to 18 percent ten years later (Mikkelson, Partch, and Shah, 1997).
Underpricing-induced ownership dispersion is not the only way to protect private
benefits of control. An obvious alternative is to put in place takeover defenses or simply
to issue non-voting stock.Field and Karpoff (2002)show that a majority of U.S. firms
deploy at least one takeover defense just before going public, especially when private
benefits of control appear large and internal monitoring mechanisms look weak—that
is, when managers’ compensation packages are unusually generous, their own equity
stakes are small, and non-directors play a smaller role in corporate governance. Inter-
estingly, however, these firms are still underpriced—though we do not know whether
they arelessunderpriced than firms that choose to entrench their managers via the
Brennan–Franks mechanism—so the protection of private benefits is unlikely to be the
only explanation of underpricing, at least in the U.S.
Issuing non-voting shares would guarantee that managers could retain control of the
company and all attendant private benefits. Whether it dominates the Brennan–Franks
underpricing mechanism is an empirical matter. Non-voting shares tend to trade at lower
multiples than voting shares. This voting discount could be smaller or larger than the
money left on the table via underpricing.Smart and Zutter (2003)find that U.S. com-
panies that issue non-voting stock in their IPOs are less underpriced and have higher
institutional ownership after the IPO. This is consistent with the notion that non-voting
stock can substitute for the Brennan–Franks mechanism. At the same time, Smart and
Zutter find that non-voting IPO shares trade at lower multiples, though they do not in-
vestigate how these compare with the monetary benefit of reduced underpricing.
Arugaslan, Cook, and Kieschnick (2004)take issue withSmart and Zutter’s (2003)
study on econometric grounds, pointing out that the main reason why IPOs involving
non-voting stock are less underpriced than voting-stock IPOs is that they are larger. Size
in turn is an important determinant of institutional investors’ stock selection, and may
thus be driving the higher post-IPO institutional ownership Smart and Zutter observe
among non-voting-stock IPOs.
Underpricing and the resulting excess demand will shield managers from outside
monitoring only to the extent that outside investors do not assemble large blocks once
trading has begun.Brennan and Franks (1997)suggest that such open-market purchases
may not be profitable. If the market anticipates the gains that would accrue if man-
agement were monitored by a sufficiently large outside shareholder, prices will rise in