Our final model of expected returns is based on the characteristics of the
industry portfolios.
Characteristics Model
This is a four factor model. We use the dividend yield, price-earnings ratio, size
and prior year excess return (lagged one month) for the industry portfolios.
Chan, Karceksi and Lakonishok (1998) show that each of the characteristics
have strong predictive power in U.K. stock returns. A notable exception to the
list is the book-to-market ratio. We do not include the industry book-to-market
rations as they were not available on Datastream. However, Dimson, Nagel and
Quigley (2001) find that the book-to-market and dividend yield effects are very
highly correlated and dividend yield is a good proxy of the value effect in U.K.
stock returns.
Information Variables and Benchmark
We use the FTA index as the benchmark portfolio for the Jensen and FS
performance measures. To estimate the conditional versions of the linear asset
pricing models, we require to specify the information set of investors. We use
instruments that studies have found to be helpful in predicting U.K. stock
returns (Solnik, 1993; Fletcher, 1997).
10
The instruments include the lagged
dividend yield on the FTA index; lagged risk-free return; lagged excess return
on the FTA index; January dummy which equals one in the month of January
and zero otherwise.
EMPIRICAL RESULTS
We initially examine the performance of the asset allocation strategies that use
linear asset pricing models using the unconditional versions of the models.
Tables 3 and 4 report the performance results for the cases of no investment
restrictions (Table 3) and where restrictions (Table 4) are imposed. Panel A of
each table contains summary statistics of the performance of the five asset
allocation strategies and the FTA index. Panel B reports the estimated
performance measures and corresponding t-statistics of the five strategies.
Panel A of Table 3 shows that the strategy which uses the characteristics
model to forecast expected excess returns has the second highest Sharpe
performance across the five strategies when no investment restrictions are
imposed. However the Sharpe performance of the strategy which uses the
characteristics model underperforms that of the market index. This strategy is
characterised by high mean excess returns and standard deviations. This differs
262 JONATHAN FLETCHER

Get Advances in Investment Analysis and Portfolio Management, Volume 9 now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.