International Tests Of Factor Anomalies: Most Don’t Survive

  • For the entire world market, 74 anomalies yielded a statistically significant long-short return for equal-weighted portfolios—56 percent of the anomalies replicated in international data based on this weighting scheme. The average absolute annualized anomaly long-short return was 6.5 percent.
  • The number of significant long-short returns shrank to 40 (30 percent) for the value-weighted portfolios. The average absolute annualized anomaly long-short return also shrank compared to equal-weighted portfolios, from 6.5 percent to 5.2 percent.
  • Only a few anomalies replicated when mitigating the impact of tiny stocks, accounting for multiple testing (multiple comparisons arise when a statistical analysis involves multiple, simultaneous statistical tests, each of which has a potential to produce a “discovery” of the same data set or dependent data sets; a stated confidence level generally applies only to each test considered individually, but often it is desirable to have a confidence level for the whole family of simultaneous tests) and using prominent factor models to adjust for expected returns. Accounting for the impact of tiny stocks and multiple testing, only 19 of 132 anomalies (14 percent) yielded significant long-short returns in the ex-U.S. world cross-section.
  • A few anomalies were strong across regions. Most of these were value anomalies (such as price-to-book, price-to-cash flow, and price-to-earnings). Others included gross profitability, earnings surprise, and new shares issuance.
  • The Fama-French size factor (SMB) and the Pastor and Stambaugh liquidity factor do not survive international tests.
  • Moving from equal-weighted portfolios to value-weighted portfolios with breakpoints derived from big stocks cut the number of significant anomalies in half for most regions—important because the stocks in international markets are on average considerably smaller than those in the U.S.
  • Prominent factor models, including Stambaugh and Yuan’s mispricing (anomaly)-based model, shrank the anomaly universe. The best-performing models contained factors related to investment and profitability in addition to the long-established risk factors (beta, size, and value).
  • For the individual regions, the share of anomalies that did not survive even the single-testing framework (using only equal weighting) was also very large. It was 80 percent for Europe, 92 percent for Japan, 88 percent for the Middle East, and 81 percent for North America. For the Asia Pacific and South America, which both consist mostly of emerging markets, the shares were 65 percent and 75 percent, respectively.
View single page >> |

Disclaimer: Performance figures contained herein are hypothetical, unaudited and prepared by Alpha Architect, LLC; hypothetical results are intended for illustrative purposes only. Past ...

How did you like this article? Let us know so we can better customize your reading experience.


Leave a comment to automatically be entered into our contest to win a free Echo Show.