5. % Change in FY1 Estimates over 3 Months

Description of Factor:

This factor calculates the percentage growth of projected 1-Year EPS from that stated 3 months prior. The estimates are based on IBES consensus forecasts. The calculation is (the current consensus 1-Year projected EPS divided by the consensus 1-Year EPS 3 months ago) minus 1. The rationale behind this variable is the expectation that stocks with upward revised forecasts exhibit some earnings momentum and earnings surprise characteristics. We surmise that stocks which contain a high value for this variable will outperform stocks with low or negative values for this variable.

Analysis

Equal Weighted: The equal weighted graph below shows a significant spread in annualized average returns between quintile 1 and quintile 5, but does not exhibit the pure step distribution which we would like to see. In particular, quintile 3 outperforms quintile 2. Quintile 1 has the highest annualized average return of 27.61% with a standard deviation of 17.07%, while quintile 2 exhibits the lowest annualized average return of 9.23% and the second highest standard deviation of 16.91%. The average excess return of quintile 1 is 7.31% and (8.45)% for quintile 5. The beta for quintile 1 is 1.09 with an R² of 69%, while beta for quintile 5 is 1.02 with an R² of 62%.

The returns for a long-short strategy based on this factor are positive for all years sampled except 2001 where the strategy would have returned (14.41%). Although this result is certainly a favorable one, an investment philosophy based solely on this strategy would have provided much weaker results in recent years. Over the past five years, quintile 1 has experienced the highest returns twice and quintile 5 has experienced the worst results twice. Furthermore, there were no occurrences of quintile 1 performing the best and quintile 5 performing the worst in the same year in any of the past five years. The monthly turnovers for all quintiles is high (96.57% for quintile 3) with quintile 1 turning over 58.24%, and quintile 5 turning over 52.22%. Although a long-short investment strategy utilizing quintiles 1 and 5 provides a reasonable spread over the sixteen years sampled, the significant transaction costs due to the high turnover could negate any excess returns achieved through this strategy.

Value Weighted: Utilizing the value weighted portfolio, the returns exhibit a step distribution but the spreads are much smaller than the equal weighted portfolio. Quintile 1 again has the highest average annual return at 21.23% and standard deviation of 15.67%. Likewise, quintile 5 also repeats the lowest average annual return of 11.68% and second highest standard deviation of 14.91%. The average excess return of quintile 1 is 1.75% and (6.55)% for quintile 5. The beta for quintile 1 is 1.03, while beta for quintile 5 is 0.92.

A long-short strategy based on value weights would produce a negative return seven of the sixteen sampled years. Clearly this strategy would not suffice. Quintile 1 attains the highest return as often as the lowest return: four years. Quintile 5 predicts the lower extreme seven of the sixteen years, but still attains the highest return three times.

Conclusion:

Based on the discussion above, it is clear that the equal weighted portfolio is better than the value weighted portfolio. But due to the high turnover and expected significant transaction costs, this strategy will likely be unsuccessful. Furthermore, the recent poor results of the equal weighted strategy hints at a potential structural change among the EPS estimates of analysts. Due to the increased scrutiny of analyst reports in recent years, some analysts may be reluctant to be too bullish on a company’s projected results, clouding the true earnings momentum of firms. Overall, we believe this factor has significant potential. However, when used alone, it does not adequately provide enough insight into quintile performance. We consider using this variable on an equal weighted basis in a bivariate or multivariate analysis.