than the two-year-ahead BR, a result consistent with Stone, Guerard, Gul-
tekin, and Adams (2002). The CTEF variable produces statistically signifi-
cant total active returns and asset selection. The CTEF variable loading on
the risk index is statistically significant at the 10 percent level because of its
loading on the earnings yield and nonlinear size indexes, as was the case
with its breadth components. There are no statistically significant sector ex-
posures in the CTEF variable. The CTEF model offers statistically significant
asset selection in a multiple-factor model framework.
We test the Frank Russell large market capitalization (the Russell 1000),
middle market capitalization (mid-cap), small capitalization (the Russell
2000), and small and middle market capitalization (the Russell 2500) uni-
verses. We test the equally weighted composite model, CTEF, of I/B/E/S earn-
ings forecasts, revisions, and breadth, described in the previous section.
The portfolio optimization algorithm seeks to maximize the ranking of
the CTEF variable while minimizing risk. The underlying CTEF variable is
statistically significant, having a monthly information coefficient of 0.049
over the 491,119 observations. The CTEF variable is used as the portfolio
tilt variable in the ITG optimization system using the BARRA risk model,
and statistically significant total excess returns are found in the Frank Rus-
sell universes (see Table 8.6). We create 100-stock portfolios monthly dur-
ing the 1990–2001 period. A lambda tilt value of one is initially used in
producing efficient portfolios. Active returns rise as the average stock size
diminishes, a result consistent with the inefficient markets literature sum-
marized in Dimson (1988) and Ziemba (1992). The highest total active re-
turns are found in the Russell 2000 stocks, the smallest stocks in the largest
3000 stocks in the Frank Russell universes, each year (see Table 8.6).
The CTEF tilt variable does not produce statistically significant sector
exposures, as reported in Table 8.7, as we previously noted in the Russell
3000 universe. The factor exposures of the CTEF variable in the Russell
224 THE USE OF FINANCIAL INFORMATION IN THE RISK AND RETURN OF EQUITY
TABLE 8.6 Risk and Return of Mean-Variance Efficient Portfolios, 1990–2001
Total Asset Risk
Universe Active T-Value Selection T-Value Index T-Value Sectors T-Value
RMC 1.98 1.37 0.99 0.86 0.97 1.45 –0.88 –0.97
R1000 2.47 2.52 1.85 2.12 0.82 2.13 –0.11 –0.23
R2500 7.76 4.37 6.48 3.96 1.61 2.85 –0.33 –0.62
R2000 9.68 5.83 8.81 5.57 0.90 2.36 –0.02 –0.07
RMC—Frank Russell mid-cap universe.
R1000—Frank Russell largest 1,000 stock universe.
R2000—Frank Russell small-cap universe.
R2500—Frank Russell small and mid-cap universe.