Tuesday, October 13, 2009

What Follows a Lost Decade?

Commentators bemoan the lost decade for stocks. In Canada, returns were positive, but low over the last decade, but in the U.S., returns of the S&P 500 were actually negative. According to Larry Swedroe, this is the first time since the great depression that this has happened over 10 or more years.

However, if we subtract out inflation, returns look even worse. As Larry shows, there have been three periods of at least 10 years since the depression that the S&P 500 has failed to beat inflation. These periods ended in 1947, 1983, and now (at least we hope it doesn’t continue).

This is all very depressing, and can make us want to give up on stocks altogether. However, I wondered what happened after these bad times. Here are the S&P 500 total returns (including dividends) for the decade after these “lost decades”:

1948 to 1957: 14.4% above inflation
1984 to 1993: 10.7% above inflation
2009 to 2018: ?

As you can see, those first two decades were spectacular! There is no guarantee that the upcoming decade will match these impressive results, but it does give us some hope.


  1. I was reading something similar recently. The commentator was saying how there's a general consensus that the recovery from this recession will be slow.

    He said this doesn't match historical reality, since recoveries after deep recessions are usually robust. I think there's a bias among humans to extrapolate the recent past into the future. Most of the time, this serves us well, but sometimes we get it wrong.

  2. Gene: You're right that we often extrapolate from the recent past. This doesn't work well when we're watching events that are mostly random. Short term investing results are mostly random. So are dice rolls, but craps players swear that dice sometimes get "hot".

  3. Your comment about craps reminds me of a tendency I have even though I know better. If I were to flip 5 heads in a row, I'd think the next flip is more likely to be a tail.

    I've read that when people are asked to simulate a random string of digits, they tend to under-represent the occurrence of doubles. I guess we think random means the number tends to change every time.