Are There Sectoral Anomalies Too? The Pitfalls of Unreported Multiple Hypothesis Testing and a Simple Solution

  • Michael Greenstone
  • Paul Oyer
Article

DOI: 10.1023/A:1008313703909

Cite this article as:
Greenstone, M. & Oyer, P. Review of Quantitative Finance and Accounting (2000) 15: 37. doi:10.1023/A:1008313703909

Abstract

The recent emphasis on sector-specific investment strategies has led to the emergence of industry-specific calendar anomalies, notably the technology sector “summer swoon”. A standard t-test implies that these price movements provide arbitrage opportunities. However, this test fails to account for the many tests that may have preceded the swoon’s discovery. We propose the use of the Bonferroni correction to account for this unreported testing. Its application reverses the conclusions about the summer swoon and finds no evidence of calendar-based price patterns in any other sector. We also use the Bonferroni correction to revisit previously documented, market-wide, anomalies. Conclusions about the most widely cited anomalies (e.g., the January effect) are unchanged, but evidence for some other “anomalies” is substantially weakened. Our results emphasize that in evaluating a proposed anomaly, sectoral in nature or otherwise, it is crucial to account for the hypotheses that were likely to have been tested but not reported.

calendar anomaliestechnology stocksdata miningBonferroni correction

Copyright information

© Kluwer Academic Publishers 2000

Authors and Affiliations

  • Michael Greenstone
    • 1
  • Paul Oyer
    • 2
  1. 1.Robert Wood Johnson Foundation ScholarUniversity of California-BerkeleyBerkeley
  2. 2.J.L. Kellogg Graduate School of ManagementNorthwestern UniversityEvanston