How does publication bias affect the anomalies literature?

Question

AcadiFi · Accepted Answer

Publication bias (or the file-drawer problem) is the tendency for journals to publish significant positive results and reject null or negative findings. In a field like factor investing where researchers test thousands of candidate signals, only the ~5% that hit significance by chance get published — creating an inflated impression of 'real' anomalies. Combined with data mining, this means the universe of published anomalies contains many false positives. Example: McLean and Pontiff (2016) examined 97 published anomalies and documented that post-publication returns fell by ~58% on average. Some interpretations: (1) arbitrage — once publicized, investors trade away the anomaly; (2) publication bias — the in-sample alpha was inflated by luck, and out-of-sample returns regress to zero; (3) regime shifts. Research by Hou-Xue-Zhang replicated 447 anomalies and found only ~36% survived tests for microcaps, transaction costs, and multiple-testing corrections. CFA candidates should be skeptical of any 'factor' with a short track record and limited theoretical grounding. Journals now increasingly require pre-registration and out-of-sample tests to address the problem.

How does publication bias affect the anomalies literature?

Master Level II with our CFA Course

Related Questions

Practice Questions