Ch. 6: Security Offerings 325
the various theoretical models listed inTable 12suggest a link betweenARand a set of
characteristics:
AR=f(m,C,k,q,β,σ,I,b/a,P), m=ur,sr,fc,pp, (5)where the parameters are the flotation method choice (m∈[ur,sr,fc,pp]), direct and
indirect issue costs (C), expected shareholder takeup of the issue (k), signal quality or
the informativeness of the available issue-quality certification technology (q), private
benefits of control (β), the ex ante risk that the security is overpriced (σ), growth as
given by the size of the project’s investment amount (I) and the size of the project’s NPV
relative to the value of assets in place (b/a), and market beliefs about the nature of firms’
equilibrium flotation strategies. These beliefs imply an issue market price ofP, which
in some equilibria are lower than the true, intrinsic value, resulting in an undervaluation
cost-component inC.
A caveat before proceeding with the results: it should be noted that the explanatory
power of regressions of the type in equation(5)as reported in the literature is uniformly
low, almost always less than 10%. More seriously, these cross-sectional regressions are
typically estimated using linear estimators (such as OLS).Eckbo, Maksimovic, and
Williams (1990)show that linear estimators (such as OLS and GLS) are biased and in-
consistent when the issuer self-selects the timing of the event (in this case security issue)
and derive a consistent, non-linear estimator.^38 Some studies (e.g.,Bøhren, Eckbo, and
Michalsen, 1997) report results with the nonlinear estimator, while others (e.g.,Eckbo
and Masulis, 1992) report that key inferences are unchanged when using OLS. More-
over, the potential for bias is smaller for utilities that are constrained by the regulatory
process. However, for the vast majority of studies reporting cross-sectional regressions,
the magnitude of the bias introduced by self-selection is largely unknown.
Adverse Selection and growth opportunities.InMyers and Majluf (1984), the market
prices firms correctly only on average, causing some highly undervalued firms to avoid
dilutive equity issues. Here, the information content is simply the adverse selection re-
vealed by the firm’s willingness to issue (separating equilibrium). The negative average
market reactions to SEOs soldto the marketin the U.S., such as in standby rights and
firm commitment offerings, is consistent with this generic framework. Moreover, as
pointed out byEckbo and Masulis (1992), equity issues that are purchased by current
shareholders (i.e.,notsold to the market) results in pooling and therefore do not convey
information. This prediction is also supported by the evidence on uninsured rights in
Table 16, both in the U.S. and internationally.
The adverse selection model also implies that the market reaction to equity offerings
should be more negative the greater the issue size (Krasker, 1986) and the greater the
ex ante uncertainty that the issue is overpriced. The uncertainty hypothesis is supported
by the evidence that debt offerings are met with little or no market reaction, while con-
vertible debt offerings produce a negative effect that is only about half the size of the
(^38) This issue is surveyed extensively inLi and Prabhala (2007)(Chapter 2of this volume).