Ch. 7: IPO Underpricing 399
However, these 38 investment bank IPOs appear to have been underpriced by roughly
as much as other IPOs, which Muscarella and Vetsuypens interpret as contradicting the
agency models.
There are only so many investment banks taking themselves public, soMuscarella
and Vetsuypens’ (1989)approach does not lend itself straightforwardly to large-sample
testing. But over the course of the 1990s, investment banks emerged as an important pre-
IPO shareholder group in many IPO companies (Ljungqvist and Wilhelm, 2003). Often,
they acquired stakes in these companies indirectly, via their venture capital operations.
By the year 2000, investment banks were pre-IPO shareholders in 44% of companies
going public. These equity stakes should reduce their incentives to underprice the stock
to the issuer’s detriment, and the size of this effect should be proportional to the size of
their equity stake.
The evidence reported inLjungqvist and Wilhelm (2003)supports both these pre-
dictions. The greater the investment bank’s equity holding, the lower are first-day
underpricing returns. This finding contrasts with the earlier result ofMuscarella and
Vetsuypens (1989)that investment banks underwriting their own IPOs suffered as much
underpricing as other issuers. However, the negative relation between investment bank
equity holdings and underpricing does not appear to depend on whether the investment
bank acted as lead underwriter. Focusing on venture-backed IPOs only,Li and Masulis
(2003)also find that initial returns decrease in the size of investment banks’ pre-IPO eq-
uity holdings, though in their case, the effect is more pronounced for lead underwriters
than for other syndicate members.
How widespread is the self-dealing behavior alleged in recent regulatory investiga-
tions into IPO practices? In general, this is hard to address empirically. For instance,
banks do not typically publish the kind of allocation data necessary to examine ‘spin-
ning’. Notwithstanding Congressional disclosure of IPO allocations to executives at
WorldCom and the class action suit over spinning against eBay, Inc., the relevant data
are unlikely to become available in a systematic fashion.
The link between allocations and trading commissions is potentially more readily ob-
servable. In an innovative paper,Reuter (2004)combines data on the recipients of the
brokerage commissions paid by U.S. mutual funds with data on the mutual funds’ eq-
uity holdings. The fund holdings data are used to approximate IPO allocations, on the
assumption that funds do not trade their IPO allocations in any systematic way (that is,
in a way that is correlated with the variables of interest). Reuter finds a positive relation
between the commissions mutual funds paid to lead managers and the size of reported
holdings in the managers’ IPOs. One interpretation is that fund managers ‘buy’ un-
derpriced IPO allocations with their trading commissions. Another is that underwriters
allocate IPOs to clients they have strong relationships with, which includes executing
much of the clients’ trades.
Reuter’s (2004)point estimates suggest that investment banks received 85 cents in
trading commissions per dollar of underpricing gain allocated to mutual funds in 1996–
- Assuming trading commissions were used to ‘buy’ underpriced IPO allocations,
banks appear to have been very good at capturing the lion’s share of the rent over that