62 K. Li and N.R. Prabhala
7.3. Takeovers:Eckbo, Maksimovic and Williams (1990)
Eckbo, Maksimovic and Williams (1990)—henceforth EMW—propose variants of the
“truncated regression” specification, rather than the Heckman selection model used in
Acharya (1988)model to analyze announcement effects. Their empirical application is
to takeovers, the subject of Chapter 15 (Betton, Eckbo and Thorburn, 2007).
EMW develop two models for announcement effects. In both models, managers an-
nounce eventEif the stock market gain,yi=xiγ+ηiis positive. As before,ηiis
private information, normally distributed with mean zero and varianceω^2 andxide-
notes publicly known variables. In model 1, eventEcompletely surprises the capital
markets. In this case, the bidder’s announcement effect is
F(xi)=E(yi|yi=xiγ+ηi> 0 )=xiγ+ω (39)φ(xiγ/ω)
Φ(xiγ/ω).
In model 2, the market learns about the impending takeover on a prior rumor date. The
probability that the takeover will be announced is the probability that the takeover gain
is positive, i.e.,Pr(xiγ+ηi> 0 )=Φ(xiγ/ω). If the takeover occurs, the gain isF(xi),
while the absence of the takeover is assumed to lead to zero gain. Thus, the expected
stock return on the rumor date isF(xi)Φ(xiγ/ω). On the actual merger announcement
date, the takeover probability rises to 1 and the announcement effect is
G(xi)= (40)[
xiγ+ωφ(xiγ/ω)
Φ(xiγ/ω)]
[
1 −Φ(xiγ/ω)]
.
The EMW expression in equation(40)is different from the Acharya model because
EMW assume that private information has value only conditional on the takeoverE,but
has no value if there is no takeover. Thus, EMW model the real gains specific to mergers
rather than non-specific information modeled by Acharya. In the actual empirical appli-
cation. EMW find that bidder gains decrease with the size of the bidder relative to the
target, the concentration of firms in the industry, and the number of previous takeovers
in the industry. As a model diagnostic, they show that OLS estimates differ from those
of the non-linear model(40), which is supported by theVuong (1989)test statistics.
EMW also report thatω^2 is significant, indicating that bidders’ private information is
valued by capital markets.
The EMW framework has been widely applied in other event-studies with cross-
sectional regressions.Eckbo (1990)examines the valuation effects of greenmail pro-
hibitions and finds that the precommittment not to pay greenmail is value enhancing.
Maksimovic and Unal (1993)estimate the after-market price performance of public of-
fers in thrift conversions recognizing that management’s choice of issue size reflects the
value of growth opportunities and conflicts of interest between managers and investors.
Servaes (1994)relates takeover announcement effects to excess capital expenditure.
Hubbard and Palia (1995)find an increasing and then decreasing relation between
merger announcement effects and managerial ownership levels.Bohren, Eckbo and