He sometimes talks about periods as long as seventy years, and often talks about periods of thirty years as well as twenty years.
Peter Bernstein's introduction to the book says
So, to Bernstein, "long run" means "twenty years or more."we must keep in mind that Professor Siegel did not lightly choose Stocks for the Long Run as the title of his book. The operative number is 20. Volatility of returns is high in periods of less than 20 years.
But does Siegel himself ever commit to a number of years—or a rough number of years—or a range of number of years that constitute what he means by "long run?" On page 28, he says:
This to me suggests that he means "the long run" to be thirty years, not twenty.Never in the past 150 years has the buyer of a newly-issued 30-year government bond who held it to maturity achieved greater gains than an investor who held a diversified portfolio of common stocks over the same period.
I'm thinking of the possibility of a vicious circle. "Stocks always do well in the long run" could be an unfalsiable tautology if, given any period over which stocks didn't do well, then by definition that wasn't a long enough period to count as "the long run."