Momentum strategies exploit a tendency for a stock's prior returns and prior news
about its earnings to predict future returns. The authors confirm momentum for
subsequent six-month and one-year periods. Prior returns and prior earnings
contribute to predicted future returns after controlling for the other. The
authors find that the results are not related to company size or book-to-market
ratios. Sluggish response of analysts' earnings forecasts to past news also
indicates that the market responds slowly to new information.
The finance literature demonstrates that past stock returns help to predict future
returns. Although this literature explains stock price reversals, it does not explain
stock price momentum. Potential sources of price momentum are underreaction to
earnings-related information, market overreaction resulting from feedback strategies,
and earnings momentum. The authors evaluate evidence of stock price momentum and its
causes.
The data set includes all primary stocks listed on the New York and American stock
exchanges and on the Nasdaq market from January 1977 to January 1993. Earnings and price
information are from CRSP, Compustat, and I/B/E/S. The authors test price and momentum
strategies by comparing performance of a group of companies for the six months prior to
portfolio formation with subsequent performance. The methods used include individual
tests of each momentum strategy, multivariate tests of conditional performance (holding
other strategies constant), and cross-sectional regressions. Separate tests consider
only large companies and adjust for company size and book-to-market ratios.
The authors compare returns on an equally weighted portfolio during the six months prior
to portfolio formation with returns over the subsequent six-month, one-, two-, and
three-year periods. Persistence of returns during the later periods provides evidence of
momentum. Price momentum tests compare the stock's past six-month compound return with
unexpected earnings, cumulative abnormal return around the most recent earnings
announcement date, and revisions of analysts' forecasts of earnings.
Price momentum is evident in that portfolios with high returns (winners) in the prior six
months are also winners in the following six months and the year after portfolio
formation. This momentum relates positively to portfolios' book-to-market and
cash-flow-to-price ratios. Earnings performance, abnormal announcement returns, and
revisions in analysts' forecasts help to explain price momentum. The results suggest
that stock price momentum partially reflects slow adjustment to information about
earnings.
Superior performance of a strategy of investing based on standardized unexpected
earnings, at least in the short term, confirms the existence of earnings momentum.
Earnings data also indicate gradual adjustment of stock prices to earnings surprises.
Stocks with large favorable earnings announcements subsequently tend to outperform those
with unfavorable announcements, but the returns tend to be more short lived. Analysts
tend to adjust their earnings forecasts with a lag, but these changes in forecasts still
influence subsequent prices.
Two-way tests show that returns and earnings news over the six months prior to portfolio
formation explain returns in the subsequent periods. Each variable provides incremental
predictive power over the other. Thus, each momentum strategy reflects market
underreaction to differing information. Again, the impact of earnings surprise is not as
long lasting as that of prior return. The authors suggest that this phenomenon occurs
because short-term earnings uncertainty is resolved more rapidly than the broader
sources of uncertainty that asset prices reflect. Regression analysis confirms that past
returns and each earnings measure are statistically significant in explaining subsequent
returns. Similar results follow when only large firms are used in the regressions.
Adjusting for size and book-to-market ratios does not alter the results.
The authors consider whether their data support the hypothesis that positive-feedback
trading causes market overreaction, which leads to subsequent reversals. Neither the
one-way nor the two-way classification indicates that stock prices subsequently correct.
A reversal does occur, however, in cases in which high past returns are not supported by
subsequent favorable earnings news. The authors conclude that a stock's prior return and
each measure of recent earnings surprise help to predict the stock's future price. Stock
prices display a delayed reaction to information on past returns and earnings news. A
general absence of evidence of subsequent reversals in returns suggests that
positive-feedback trading does not account for the success of momentum strategies.