Elsevier

Journal of Econometrics

Volume 49, Issues 1–2, July–August 1991, Pages 51-104
Journal of Econometrics

A posterior odds analysis of the weekend effect

https://doi.org/10.1016/0304-4076(91)90010-BGet rights and content

Abstract

The weekend effect is the tendency for Monday stock returns to be negative. The paper reports a posterior odds evaluation of the day-of-the-week and weekend effect that largely reverses earlier findings. The interaction of large sample sizes and fixed significance level hypothesis testing is identified as the likely source of disagreements between p-values and posterior probabilities. Analysis with informative and relatively diffuse prior distributions indicates this divergence does not apparently reflect special distributional assumptions. Further analysis suggests that earnings announcement behavior and a small number of outliers may account for much of the evidence of systematically negative Monday returns in the few years where posterior odds favor the weekend effect hypothesis.

References (52)

  • James O. Berger et al.

    Testing a point hull hypothesis: The irreconcilability of p values and evidence

    Journal of the American Statistical Association

    (1987)
  • George E.P. Box et al.

    Bayesian inference in statistical analysis

    (1973)
  • Stephen J. Brown et al.

    A new approach to testing asset pricing models: The bilinear paradigm

    Journal of Finance

    (1983)
  • George Casella et al.

    Reconciling Bayesian and frequentist evidence in the one-sided testing problem

    Journal of the American Statistical Association

    (1987)
  • Richard P. Castanias

    Macroinformation and the variability of stock market prices

    Journal of Finance

    (1979)
  • Robert A. Connolly

    GARCH volatility models for bond and stock returns

    (1988)
  • Robert A. Connolly

    An examination of the robustness of the weekend effect

    Journal of Financial and Quantitative Analysis

    (1989)
  • Brad Cornell

    Money supply announcements and interest rates: Another view

    Journal of Business

    (1983)
  • Brad Cornell

    The money supply announcements puzzle: Review and interpretation

    American Economic Review

    (1983)
  • James M. Dickey

    Bayesian alternatives to the F-test and least squares estimates in the normal linear model

  • James M. Dickey

    Is the tail area useful as an approximate Bayes factor?

    Journal of the American Statistical Association

    (1977)
  • James M. Dickey

    Approximate coherence for regression model inference with a new analysis of Fisher's Broadbalk Wheatfield example

  • Mark Flannery et al.

    From T-bills to common stock: Investigating the generality of intra-week return seasonality

    Journal of Finance

    (1988)
  • Kenneth M. Gaver et al.

    Discriminating among alternative models: Bayesian and non-Bayesian methods

  • John Geweke

    Exact inference in models with autoregressive conditional heteroskedasticity

  • Michael R. Gibbons et al.

    Day of the week effects and asset returns

    Journal of Business

    (1981)
  • Cited by (48)

    • Stock market anomalies: An extreme bounds analysis

      2023, International Review of Financial Analysis
    • Significance testing in empirical finance: A critical review and assessment

      2015, Journal of Empirical Finance
      Citation Excerpt :

      This occurs largely because the classical method of hypothesis testing is conducted with a fixed level of significance, while the Bayesian methods imply the critical values as an increasing function of sample size, as in Eqs. (2) and (3). In finance, Neal (1987) and Connolly (1989, 1991) acknowledge the effect of large sample size on significance testing and propose the Bayesian methods as an alternative. Connolly (1991) finds that earlier empirical evidence for the weekend effect in stock return is reversed if the Bayesian method is used.

    • The Monday effect revisited: An alternative testing approach

      2011, Journal of Empirical Finance
      Citation Excerpt :

      All kinds of competing explanations have been presented — the timing of corporate releases after Friday's close (see Damodaran, 1989), short sellers closing positions over non-trading periods such as weekends (Chen and Singal, 2003) as well as the comparative advantage of informed traders when markets first open after a period of no trading (see Foster and Viswanathan, 1990). Meanwhile various methodological issues related, among others, to time series properties of stock returns have been addressed (see e.g. Chang et al., 1993; Connolly, 1991). This paper is concerned with the problem that empirical tests of day-of-the-week effects do not appropriately account for the so-called multiplicity effect.1

    View all citing articles on Scopus

    Comments and assistance from the following individuals are gratefully acknowledged: Nai-fu Chen, John Clarke, Charles Cuny, Phil Dybvig, Robert Eisenbeis, Robert Haugen, Chuan-Yang Hwang, Chris Lamoureux, Edward E. Leamer, Ron Masulis, Richard McEnally, Henry McMillan, Marlene Puffer, Erik Sirri, Seymour Smidt, Paul Spindt, Laura Starks, Neal Stoughton, Rex Thompson, and Arnold Zellner. I am grateful to Dale Poirier (the editor) for considerable help and assistance. An associate editor and two anonymous referees also provided very useful comments. I benefited from comments of seminar participants at Rutgers University, Southern Methodist University, University of California at Irvine, University of North Carolina at Chapel Hill, Washington University, and the 36th meeting of the NBER-NSF Seminar on Bayesian Inference in Econometrics held at the School of Business Administration, The University of Michigan, April 1988. Juli Jones and Kwang Lee provided capable research assistance. Several grants from the University of California-Irvine Academic Senate Committee on Research and a UCI Faculty Research Fellowship provided financial support during the time this research was underway. The usual caveat applies.

    View full text