Промышленный лизинг Промышленный лизинг  Методички 

[ 1 ] 2 3 4 5 6 7 8 9 10

sample and methodology

We consider all domestic, primary stocks listed on the New York (NYSE), American (AMEX), and Nasdaq stock markets. Closed-end funds, Real Estate Investment Trusts (REITs), trusts, American Depository Receipts (ADRs), and foreign stocks are excluded from the analysis. Since we require information on earnings, the sample comprises all companies with coverage on both the Center for Research in Security Prices (CRSP) and COMPUSTAT (Active and Research) files. The data for firms in this sample are supplemented, wherever available, with data on analysts forecasts of earnings from the Lynch, Jones, and Ryan Institutional Brokers Estimate System (I/B/E/S) database.

At the beginning of every month from January 1977 to January 1993, we rank stocks on the basis of either past returns or a measure of earnings news. To be eligible, a stock need only have data available on the variable(s) used for ranking, even though we provide information on other stock attributes. The ranked stocks are then assigned to one of ten decile portfolios, where the breakpoints are based only on NYSE stocks. In our earnings momentum strategies, the breakpoints in any given month are based on all NYSE firms that have reported earnings within the prior three months. This takes into account a complete cycle of earnings announcements. All stocks are equally-weighted within a given portfolio.

The ranking variable used in our price momentum strategy is a stocks past compound return, extending back six months prior to portfolio formation. In our earnings momentum strategies, we use three different measures of earnings news. Our first is the commonly used standardized unexpected earnings (SUE) variable. Foster, Olsen, and Shevlin (1984) examine different time series models for expected earnings and how the resulting measures of unanticipated earnings are associated with future returns. They find that a seasonal random walk model performs as well as more complex models, so we use



it as our model of expected earnings. The SUE for stock i in month t is thus defined as

SUEa = e*~gig-4 (1)

where eiq is quarterly earnings per share most recently announced as of month t for stock i, eiq 4 is earnings per share four quarters ago, and ait is the standard deviation of unexpected earnings, eiq - eiq 4, over the preceding eight quarters.

Another measure of earnings surprise is the cumulative abnormal stock return around the most recent announcement date of earnings up to month t, ABR, defined as

ABR = 2 (-rmj) (2)

J = -2

where rtj is stock is return on day j (with the earnings being announced on day 0) and rmj is the return on the equally-weighted market index. We cumulate returns until one day after the announcement date to account for the possibility of delayed stock price reaction to earnings news, particularly since our sample includes Nasdaq issues that may be less frequently traded. This return-based measure is a fairly clean measure of earnings surprise, since it does not require an explicit model for earnings expectations. However, the abnormal return around the announcement captures the change over a window of only a few days in the markets views about earnings. The SUE measure incorporates the information up to the last quarters earnings and hence in principle measures earnings surprise over a longer period.

Our final measure of earnings news is given by changes in analysts forecasts of earnings. Since analyst estimates are not necessarily revised every month, many of the monthly revisions take the value of zero. To get around this, we define REV6, a six-month moving average of past changes in earnings forecasts by analysts:

REV6tt= 2fu~ifu~i~1 (3)

i=0 Pit-j-i

where fit is the consensus (mean) I/B/E/S estimate in month t of firm is earnings for the current fiscal year (FY1). The monthly revisions in estimates are scaled by the prior months stock price.1 Analyst estimates are available on

1 Scaling the revisions by the stock price penalizes stocks with high price-earnings ratios. To circumvent this possibility, we also scaled revisions by the book value per share. We also experimented with the percent change in the median I/B/E/S estimate, as well as the difference between



a monthly basis2 and dispense with the need for a model of expected earnings. However, the estimates issued by analysts may be colored by other incentives such as the desire to encourage investors to trade and hence generate brokerage commissions.3 As a result, analyst forecasts may not be a clean measure of expected earnings.

For each of our momentum strategies, we report buy-and-hold returns in the periods subsequent to portfolio formation. Returns measured over contiguous intervals may be spuriously related due to bid-ask bounce, thereby attenuating the performance of the price momentum strategy. To control for this effect, we skip the first five days after portfolio formation before we begin to measure returns under the price momentum strategy and, for the sake of comparability, under the earnings momentum strategy as well. If a stock is delisted after it is included in a portfolio but before the end of the holding period over which returns are calculated, we replace its return until the end of the period with the return on a value-weighted market index. At the end of the period we rebalance all the remaining stocks in the original portfolio to equal weights in order to calculate returns in subsequent periods. In addition to returns on the portfolios, we also report two attributes of our portfolios-the book-to-market value of equity and the ratio of cash flow (earnings plus depreciation) to price-at the time of portfolio formation. Finally, we also track our three measures of earnings surprise (SUE, ABR, and REV6) at the time of portfolio formation and thereafter.

II. Price and Earnings Momentum: Univariate Analysis

A. Price Momentum

We first examine the ability of each of the momentum strategies to predict future returns, and the characteristics of the momentum portfolios. To lay the groundwork, Table I reports correlations between the various measures we use to group stocks into portfolios. The correlations are based on monthly observations pooled across all stocks. Although the variables are positively correlated with one another, the coefficients are not large. In particular, the differ-

the number of upward and downward revisions as a proportion of the number of estimates. Our results are robust to these alternative measures of analyst revisions.

2 In the context of an implementable investment strategy, all stocks are candidates for inclusion in our price momentum or earnings momentum portfolios in a given month. The strategy based on analyst revisions automatically fulfills this requirement, since consensus estimates are available at a monthly frequency. The portfolios based on standardized unexpected earnings and abnormal announcement returns will pick up an earnings variable that may be somewhat out-of-date for those firms not announcing earnings in the month of portfolio formation. This may lead to an understatement of the returns to these two earnings momentum strategies, but in any event we are able to compare directly the results from the price momentum and from the earnings momentum strategies.

3 Several recent examples of these kinds of pressures on analysts are described by Michael Siconolfi in A rare glimpse at how Wall Street covers clients, Wall Street Journal, July 14,1995, and Incredible buys: Many companies press analysts to steer clear of negative ratings, Wall Street Journal, July 19, 1995.



[ 1 ] 2 3 4 5 6 7 8 9 10