Ch. 7: IPO Underpricing 405
to expect lagged index returns to affect lawsuits many years later. This makes lagged
index returns a plausible instrument for underpricing. Damages generally increase in
the number of shares traded at the allegedly misleading prices, so stock turnover may
be a plausible instrument for litigation risk a priori.^15
The OLS and 2SLS estimates give rise to radically different conclusions. The OLS
results suggest that underpricing decreases in the incidence of (actual) lawsuits, sug-
gesting that firms underprice less the more often they are sued. The sign of this relation
flips in the 2SLS model. Here, underpricing increases in the predicted probability of
lawsuits, consistent with the lawsuit avoidance hypothesis. Interestingly, greater under-
pricing does not appear to have much deterrence effect: the probability of being sued
does not decrease in the instrumented underpricing return, at least not at conventional
significance levels.
Lowry and Shu’s study is sensitive to econometric concerns, and using more careful
tools than prior work it finds evidence consistent with the proposition that firms use un-
derpricing as a form of insurance against future litigation. Unfortunately, their empirical
model is not able to gauge theeconomicmagnitude of this effect (because their system
cannot identify all relevant parameters). They are thus unable to say if litigation risk has
a first-order effect on underpricing.
4.2. Price stabilization
Rather than forming a symmetric distribution around some positive mean, underpricing
returns typically peak sharply at zero and rarely fall below zero. In a controversial paper,
Ruud (1993)takes these statistical regularities as her starting point to argue that IPOs
arenotdeliberately underpriced. Rather, IPOs are priced at expected market value but
offerings whose prices threaten to fall below the offer price are stabilized in after-market
trading. Such price stabilization would tend to eliminate the left tail of the distribution
of initial returns, and thus lead to the appearance of a positive average price jump. Thus
what we observe in the data may not be the unconditional expectation of true initial
returns but the mean conditional upon underwriter intervention in the aftermarket. Esti-
mating the unobserved unconditional mean of the return distribution in a Tobit model,
Ruud finds that average (logged) first-day returns are indeed close to zero.
This largely statistical view of the origins of IPO underpricing leaves little room
for economics. Why would underwriters stabilize prices in the first place? Subsequent
theoretical work on price stabilization has stressed its role in reducing underpricing.
Benveniste, Busaba, and Wilhelm (1996)formalizeSmith’s (1986)notion of price sta-
bilization as a mechanism that ‘bonds’ underwriters and investors. Because their dollar
fees increase in gross proceeds, underwriters have a natural incentive to raise the offer
(^15) Though note that empirically, stock turnover does correlate with underpricing, violating the order condi-
tion. Strictly speaking, the system estimated in Lowry and Shu relies for identification on the functional form
of the probit equation modeling litigation risk, not on the use of instrumental variables.