Narasimhan Jegadeesh

Evaluating the performance of value versus glamour stocks The impact of selection bias

We examine whether sample selection bias explains the difference in returns between ‘value’ stocks (high book-to-market ratios) and ‘glamour’ stocks (low book-to-market ratios). Selection bias on Compustat is not a severe problem: for CRSP primary domestic firms, the proportion missing from Compustat is not large and the average return is not very different from the Compustat sample. Mechanical problems with matching Cusip identifiers account for much of the discrepancy between CRSP and Compustat. The superior performance of value stocks is confirmed for the top quintile of NYSE-Amex stocks, using a sample free from selection bias.

Long-Term Performance of Seasoned Equity Offerings: Benchmark Errors and Biases in Expectations

I investigate the long-term performance of firms that issue seasoned equity relative to a variety of benchmarks. I find that these firms significantly underperform all of my benchmarks over the five years following the equity issues. Across SEOs, I find similar levels of underperformance for both small firms and large firms, and both growth firms and value firms. The paper also shows that factor-model benchmarks are misspecified. Hence inferences on SEO underperformance based on such benchmarks are misleading. I also find that SEOs underperform their benchmarks by twice as much within earnings announcement windows as they do outside these windows.

Value of Analyst Recommendations: International Evidence

We evaluate the value of analysts’ recommendations in the G7 countries. Stock prices react significantly to recommendation revisions in all countries except Italy. We find the largest price reactions around recommendation revisions and the largest post-revision price drift in the US. Neither differences in the timing of recommendation revisions relative to earnings announcements nor differences in industry coverage explain the superior performance of the US analysts’ recommendations. Tests within a subsample of ADRs indicate that the most likely explanation for the superior performance is that the US analysts are more skilled at identifying mispriced stocks than their foreign counterparts.

The Behavior of Interest Rates Implied by the Term Structure of Eurodollar Futures

This paper considers an equilibrium model of the term structure that is determined by two stochastic factors: a short-term interest rate and a target level to which the short rate is expected to revert. A Kalman filter technique that uses a time series, cross-section of Eurodollar futures prices is developed to estimate the parameters of the model. The term structures of spot LIBOR and Eurodollar futures volatility are compared to that predicted by the model. The empirical results indicate that the two-factor specification represents a significant improvement over its one-factor version. Copyright 1996 by Ohio State University Press.

Does Market Risk Really Explain the Size Effect

This paper critically evaluates the claim in recent papers that precisely estimated betas explain the cross-sectional differences in expected returns across size-based portfolios. In these studies, the correlations between firm size and betas across the test portfolios are close to one in magnitude, yielding potentially spurious inferences. This paper shows that when the test portfolios are constructed so that the correlations between firm size and beta are small, the betas explain virtually none of the cross-sectional differences in portfolio returns.

Post-Earnings-Announcement Drift: The Role of Revenue Surprises

The study reported here consisted of estimating earnings and sales (or revenue) surprises either with historical time-series data or with analyst forecasts. Post-earnings-announcement drift was found to be stronger when the revenue surprise was in the same direction as the earnings surprise. This result proved to be robust to various controls, including the proportions of stock held by institutional investors, arbitrage risk, and turnover (prior 60-month average trading volume). This finding is consistent with prior evidence that earnings surprises have a more persistent effect on future earnings growth when they consist of higher revenue surprises than when they consist of lower expense surprises. The study we report shows that the magnitude of post-earnings-announcement drift in security returns after the announcement of earnings depends on the magnitude of contemporaneous revenue surprises. When the two signals for a company confirm each other, drift is larger. Drift is stronger when the revenue surprise is in the same direction as the earnings surprise quite probably because revenue surprises identify companies for which earnings surprises should have a more persistent effect on future earnings growth. This result held even after we controlled for the sophistication of investors (as indicated by the proportion of shares held by institutions), arbitrage risk (based on the correlation of the company returns with S&P 500 Index returns), and turnover (prior 60–month average trading volume). These results proved to be robust across subperiods, including the 1998–2002 period, which spanned extreme market fluctuations. The results proved to be robust also for three different subsamples of data—a large subsample of companies that had historical earnings and revenue data in the Compustat quarterly database, a subsample of companies for which at least two analysts provided earnings forecasts and historical revenue data were used, and a much smaller subsample of companies for which at least one analyst forecasted revenues as well as earnings. In all three cases, a portfolio of companies with extreme earnings surprises and extreme revenue surprises in the same direction earned significantly higher abnormal returns in the quarter after the preliminary earnings announcement. These higher abnormal returns were most pronounced when we used analyst forecasts of both earnings and revenues to estimate the surprises. The difference between analyst forecasts of earnings and the actual earnings reported by I/B/E/S provided a better measure of earnings surprise than the difference between time-series estimates of earnings surprise garnered from Compustat data. Naturally, the improvement in performance when we used a trading strategy based on revenue surprises as well as earnings surprises came at a minor cost; the portfolio of companies with extreme earnings and revenue surprises in the same direction comprised a smaller number of companies than the portfolio of companies with extreme earnings surprises alone. The portfolio of extreme revenue+earnings surprises nevertheless contained a sufficient number of companies for adequate portfolio diversification. Investors and other market participants can use the evidence in this study to improve trading strategies that use earnings surprise as a signal. Revenue surprises can be useful as an additional signal to determine future earnings growth. Similarly, analysts can use revenue surprises when forecasting future earnings. The results also indicate that analysts should carefully examine the sources of earnings surprises to forecast future earnings growth. Fundamental analysis should take into account the differential effects of revenue and expense surprises on the persistence of future earnings growth. In particular, earnings growth driven by revenue growth has a stronger effect on future earnings levels than does earnings growth stemming from expense reduction. Finally, research into the reasons and implications of post-earnings-announcement drift should take into account the differential drift implications of revenue and expense surprises.

Gender and Job Performance: Evidence from Wall Street

We study the relation between gender and job performance among brokerage firm equity analysts. Womens representation in analyst positions drops from 16% in 1995 to 13% in 2005. We find women cover roughly 9 stocks on average compared to 10 for men. Womens earnings estimates tend to be less accurate. After controlling for forecast characteristics, the difference in accuracy is roughly equivalent to four years of experience. Despite reduced coverage and lower forecast accuracy, we find women are significantly more likely to be designated as All-Stars, which suggests they outperform at other aspects of the job such as client service.

An analysis of bidding in the Japanese government bond auctions

We examine the bidding patterns and auction profits in the Japanese Government Bond (JGB) auctions and empirically test the predictions of auction theory. We find that the average profit in JGB auctions is not reliably different from zero, and the degree of competition and the level of uncertainty are insignificant in determining auction profits. The winning shares of the U.S. dealers are positively related to auction profits, whereas the winning shares of their Japanese counterparts show a negative association. We also find that the share of winnings of Japanese dealers tends to be correlated with the share of winnings of their compatriot dealers but a similar relation is not found for U.S. dealers. Copyright The American Finance Association 1998.

A Non-Parametric Prepayment Model and Valuation of Mortgage-Backed Securities

A non-parametric technique called generalized additive model (GAM) estimation is particularly useful in high-dimension non-parametric estimations and in situations that involved mixed parametric and non-parametric specifications. The relationship between prepayment rates, and variables such as the age of the mortgage, the ratio of the mortgage coupon rate and prevailing interest rates, and expected and unexpected burnouts is highly non-linear, and it is difficult to capture these relations with parametric functions. Decomposition of pool burnouts into expected and unexpected components improves the model fit and has important pricing implications. The prepayment model estimated here fits the data significantly better than other models in use, and illustrates the factors that affect prepayments and the prices of mortgage-backed securities.

Market-Based Evaluation for Models to Predict Bond Ratings

Previous studies have examined different statistical models to predict corporate bond ratings. However, these papers use agency ratings as the benchmark to assess models and ignore the evidence that agency ratings may not be accurate in a timely manner. In this paper, we propose a new approach which incorporates ex-post bond returns to evaluate rating prediction models. Relative rating strength portfolios, formed by buying under-rated bonds with agency ratings lower than model ratings and selling over-rated bonds with agency ratings higher than model ratings, are employed to test the performance of different statistical models in rating predictions. Our results show that one version of multiple discriminant analysis model can generate a statistically significant abnormal return of 5% over a 5-year horizon. The ordered probit model which is believed to possess theoretical advantages in classifying bonds does not perform better. This suggests that using traditional measures to evaluate models can be misleading. The existence of a profitable trading strategy also raises the concern of market efficiency in the corporate bond market.