342 B.E. Eckbo et al.
effective holding period for stockiisTi, whereTiin the analysis below is either five
years or the time until delisting or the occurrence of a new SEO, whichever comes first.
Kothari and Warner (1997), Barber and Lyon (1997)andLyon, Barber, and Tsai (1999)
provide simulation-based analyses of the statistical properties of test statistics based on
long-run return metrics such as BHR. InChapter 1of this volume,Kothari and Warner
(2007)survey the main statistical conclusions from this analysis.^51
The matched-firm technique equates theexpectedreturn to issuing firms with the
realizedreturn to a non-issuing firm, usually matched on firm characteristics such as
industry, size and book-to-market ratio. The abnormal or unexpected return BHAR is
then
BHARIssuer≡BHRIssuer−BHRMatched firm. (7)Table 17shows average five-year buy-and-hold returns following security offerings
by U.S. firms that took place over the period 1980 through 2000, classified by the type
of issuer.^52 As inEckbo, Masulis, and Norli (2000)andEckbo and Norli (2005),the
matched firms are selected from all CRSP-listed companies at the end of the year prior
to the issue-year and that are not in our sample of issuers for a period of five years prior
to the offer date. We first select the subset of firms that have equity market values within
30% of the equity market value of the issuer. This subset is then ranked according to
book-to-market ratios. The size and book-to-market matched firm is the firm with the
book-to-market ratio, measured at the end of the year prior to the issue year, that is
closest to the issuer’s ratio. Matched firms are included for the full five-year holding
period or until they are delisted, whichever occurs sooner. If a match delists, a new
match is drawn from theoriginallist of candidates described above.
(^51) An alternative toBHR is to estimate the average monthly return to a strategy of investing in the stocks
of issuers and hold these for up toTperiods. TheT-period return would then be formed as thecumulative
average(portfolio) return, or
CMR≡
∏T
t=τ
[
1 +
1
ωt
∑Nt
i= 1
Rit
]
− 1.
As noted byKothari and Warner (2007), depending on the return generating process, the statistical properties
ofBHR andCMR can be very different. Notice also that whileCMR represents the return on a feasible
investment strategy,BHR does not. You obtainCMR by investing one dollar in the first security issue at the
beginning of the sample period, and then successively rebalancing this initial investment to include subsequent
issues as they appear (andNincreases), all with aT-period holding period. In contrast,BHR is formed in
event time—and thus presumes prior knowledge of the magnitude ofN. Thus, estimates ofCMR are better
suited than estimates ofBHR to address the question of whether investors have an incentive to take advantage
of a potential market mispricing of security issues. Most of the empirical studies using the matched firm
technique report results based onBHR, which we follow here. In the subsequent section, however, we discuss
portfolio benchmark returns based on asset pricing models, which uses the return conceptCMR on a monthly
basis, i.e., without theT-period cumulation.
(^52) Utilities are firms with CRSP SIC codes in the interval [4910, 4939].