According to the author, the historical stock market data that are oft
en used to forecast future returns are biased upward by virtue of the
fact that they are a sample from the most successful country (the Unit
ed States) in the most prosperous century in the history of the world.
Future returns, the author argues, will probably be lower than sugges
ted by past data. Moreover, many investors use time diversification to
reduce the risk of holding stocks over the long run. This will. work
only if stock returns are mean-reverting, which may be a statistical i
llusion. In the author's view, these two factors make long-run investm
ent in equities riskier than it appears.