Cragg and Malkiel (1968) and Niederhoffer and Regan (1972) found that an-
alysts could not effectively forecast annual earnings relative to naive time se-
ries models, and Elton, Gruber, and Gultekin (1981) found little excess
returns associated with purchasing securities solely on the basis of predicted
earnings forecasts (EP). Niederhoffer and Regan (1972) and Elton, Gruber,
and Gultekin (1981) found that the securities achieving the highest earnings
growth produced significant excess returns. Thus, there is a significant reward
to correctly forecasting earnings, but analysts’ forecasts may be not sufficient.
Guerard, Blin, and Bender (1996a) found that analysts’ forecasts were
not sufficient in Japan to outperform the market during the 1987–1994 pe-
riod. The lack of excess returns associated with consensus forecasts should
not be the end of the analysis, because changes in the mean values of the
forecasts divided by the stock price have been shown to add value in the
United States (Hawkins, Chamberlain, and Daniel 1984; Wheeler 1995)
and Japan (Guerard, Blin, and Bender 1996a). The changes in mean fore-
casts are referred to as “earnings revisions” (EREV), and one purchases
stocks when analysts are raising their forecasts (Keon 1996). Wheeler
(1995) found substantial value to using the breadth of earnings (defined as
the number of forecasts raised less the number of forecasts lowered, the re-
sult divided by the total number of forecasts) to rank stock, where one pur-
chases stocks when a (net) increasing number of analysts are raising their
forecasts (EB). The breadth measure may well be less susceptible to the un-
due influence of a single analyst.
A composite growth variable (CTEF) is created from consensus I/B/E/S
forecasts, forecast revisions, and breadth of forecasts and is of the general
form described in Wheeler (1990) and discussed in Chapters 8 and 9. The
composite I/B/E/S variable, PRGR, greatly enhances return even after
transactions costs have been included.
In this study we test several forms of an earnings forecasting (EF)
variable:
1.EQ(FY1, FY2EP)
2.EQ(FY1, FY2 EREV)
3.EQ(FY1, FY2 EB)
4.EQ(FY1 EP, EREV, EB)
5.EQ(FY2 EP, EREV, EB)
6.CTEFThe model may be summarized in equation (10.1):
TRt= a0 + a1EPt+ a2BPt+ a3CPt+ a4SPt+ a5DYt
+ a6NCAVt+ a7EFt+ et(10.1)