Mean Reversion in Stock Prices: Evidence and Implications
Read the paperopens www.nber.org in a new tab
What they found
Using variance-ratio tests on U.S. stock indices since 1871 and on 17 other countries' markets, the authors found that returns are positively autocorrelated over short horizons (months) and negatively autocorrelated over long horizons (several years), meaning a substantial transitory component in prices that decays slowly. They estimate that a large fraction of the variance of monthly returns comes from this temporary component, and interpret it as evidence of slowly correcting mispricing driven by noise traders rather than changing risk premia.
What you can use
- Prices contain a large 'temporary' component: some of every big move eventually unwinds, but over years, not days.
- Short horizons show momentum-like continuation; long horizons show reversal. Your holding period determines which you face.
- Long-horizon tests have very few independent observations, so the statistical evidence is weaker than it looks.
Caveats
With roughly a century of data, there are only a couple of dozen non-overlapping five-year periods; the authors acknowledge low statistical power. Results are index-level, not stock-level.
Tags: mean-reversion, variance-ratio, long-horizon, indices
Summaries are our own reading of the paper, not the authors' words. Educational only, not advice. Discuss it in Book Club.