Survival Bias and the Equity Premium Puzzle


  • Haitao Li,

  • Yuewu Xu

    Search for more papers by this author
    • Li is from Johnson Graduate School of Management, Cornell University and Xu is from TIAA-CREF. We thank Warren Bailey; Peter Bossaerts; Stephen Brown; Alexei Egorov; Will Goetzmann; Roger Ibbotson; Jonathan Ingersoll; Robert Jarrow; Charles M.C. Lee; Stephen Ross; Mark Schroder; Fan Yu; and seminar participants at Cornell University, the Yale School of Management, and the 11th Annual Financial Economics and Accounting Conference at the University of Michigan Business School for helpful comments. We are especially grateful to the editor Richard Green and the anonymous referee for their comments and suggestions that greatly improved the paper. We are responsible for any errors.


Previous authors have raised the concern that there could be serious survival bias in the observed U.S. equity premium. Contrary to conventional wisdom, we argue that the survival bias in the U.S. data is unlikely to be significant. To reach this conclusion, we introduce a general framework for modeling survival and derive a mathematical relationship between the ex ante survival probability and the average survival bias. This relationship reveals the fundamental difficulty facing the survival argument: High survival bias requires an ex ante probability of market failure, which seems unrealistically high given the history of world financial markets.