Review of Quantitative Finance and Accounting, 12:3 (1999): 283–301
© 1999 Kluwer Academic Publishers, Boston. Manufactured in The Netherlands.
Predicting UK Takeover Targets: Some
Methodological Issues and an Empirical Study
School of Management and Finance, University of Nottingham, Nottingham, UK, NG7 2RD, School of
Business, Rutgers University, Camden, NJ 08102
Abstract. This paper examines the methodological issues of using accounting ratios to predict takeover targets.
Certain improvements are suggested, notably the use of industry-relative ratios and the deterimination of an ex
ante cut off point which maximizes returns. The empirical study suggests that industry-relative ratios are not an
improvement upon industry-speciﬁc models and that, although the models perform better than chance, they do
not perform sufﬁciently well as to earn excess returns. Unfortunately, the results are not sufﬁciently sensitive as
to be signiﬁcantly affected by the choice of cut off point.
Key words: Mergers, takeovers, acquisitions, prediction
It is not surprising that a popular pursuit of accounting researchers has been to use
accounting data to identify and predict takeover targets and thereby earn excess stock
market returns. Studies include Simkowitz and Monroe (1971), Stevens (1973), Harris et
al (1982), Wansley and Lane (1983) and Dietrich and Sorensen (1984) in the USA,
Castagna and Matolcsy (1976) in Australia, Belkaoui (1978) and Rege (1984) in Canada
and Tzoannos and Samuels (1972) in the UK all of which used historical cost accounting
data. More recent studies using the same methodology are Barnes (1990), in the UK who
used historical cost data, and in the USA Palepu (1986), who used historical cost ac-
counting data, Bartley and Boardman (1986, 1990), who used inﬂation-adjusted account-
ing data, and Walter (1994) who used both.
With the exception of Palepu (1986) and, to a lesser extent, Walter (1994), all of these
papers boasted high predictive ability rates ranging from 60% to 90%. Palepu (1986) on
the other hand, drew attention to the methodological issues and problems involved and
showed how the earlier studies were methodologically ﬂawed and as a result overstated
their predictive accuracy. This was because the samples contained an equal number of
targets and non-targets, whereas the statistical methodology assumed random samples and
the occurrence of a takeover was relatively rare.
Of the 4,000 plus public limited com-
panies in the UK, about half of which are quoted on the International Stock Exchange in
London, between 50 and 100 are bid for each year.
These issues were well-known
elsewhere in similar applications in accounting and ﬁnance, notably in corporate failure
prediction. (For reviews of these see Zmijewski, 1984, and Altman, 1984). The fact that
Palepu’s study had a much lower predictive power further supported his argument that the
@ats-ss11/data11/kluwer/journals/requ/v12n3art4 COMPOSED: 03/12/99 12:22 am. PG.POS. 1 SESSION: 10