Decades of Negative Returns: A Long-Term Look at the Dow

Oct.11.08 | About: SPDR Dow (DIA)

With the severe market decline this past week, the 10-year percentage change for the Dow Jones Industrial Average (($DJI)) has fallen to single digits, and it appears likely that we could head into negative territory in the weeks to come.

A long-term historical look at the Dow since 1910 finds that ten-year periods of negative returns are not so unusual. They occurred in 1915, during the Great Depression years, and then during the severe recessionary period in 1974. Both the 1930s episode and that beginning in 1974 lasted a number of years: once 10-year performance went negative, it stayed that way for quite a while.

These periods of 10-year underperformance are not common -- they account for only 4469 of the more than 26,000 days in my sample. Interestingly, only the occasions falling during 1937 showed losses over the following ten years: 4314 occasions showed subsequent 10-year rises, 155 showed losses. The median size of the 10-year gain following 10-year weakness was actually a bit smaller (61%) than the median size of 10-year rises for the remainder of the occasions (85%). In short, staying out of stocks following 10-year periods of underperformance did not benefit investors, but on average they did not see unusual outperformance either.

What this little exercise suggests is that periods of long-term market underperformance can persist over time -- they do not necessarily lead to durable rebounds simply because of oversold conditions--but that fleeing markets during these periods of long-term underperformance has not been a winning strategy over the long run. Even after momentum lows were registered in 1932 and 1974, markets engaged in long periods of basing before new secular bull markets kicked in. So far, it is not at all clear that we've seen those momentum lows. I'll need to see those -- and some evidence of meaningful basing -- before aggressively pursuing the next secular bull market of triple-digit 10-year returns.