Every now and then, well all the #@$@#$ time, we hear that stocks are a good long term investment. Apparently, so the reasoning goes, stocks have historically provided good returns when averaged over periods of several decades(*), and therefore stocks will provide good returns in the long run in the future as well.

(*) Several decades is a long time. FYI a couple=2, a few=3-5, several>5.

This reasoning is akin to jumping out of an airplane without a parachute and arguing that things have been going splendid this far and therefore things will go splendid in the future as well.

Until of course the point when they are not.

As investors are finding out.

The thing that amazes me the most is that some investors still think a random basket of stocks is a good long term investment. As a saying in certain parts of Texas goes: Fool me once, shame on … shame on you. Fool me … you can’t get fooled again! 😀

Extrapolation as a predictive model stops working as soon as you hit a curve.

The thing we can agree on is that stocks HAVE BEEN a good long term investment in the past. What can we conclude from that? Not much!

Past performance is no indication is no indication of future returns. It’s in every prospectus I have ever read through. Every one of them.

Here are some reasons why stocks do well

  • Low population density / unexploited resources.
  • Immigration (population growth).
  • Rising energy production.
  • Shareholder profits.
  • Technological wave.
  • Workers saving for retirement.
  • Surplus on capital account from trade deficit gets invested in stocks.
  • Sheeple follow the bears (widespread pessimism)

Here are some reasons why stocks do bad

  • Declining energy production.
  • Resource constraints.
  • Corporate taxes.
  • Decaying infrastructure.
  • Employee profits.
  • Aging population
  • Retired workers selling.
  • Sheeple follow the bulls (widespread optimism).

In the past, these factors have favored stocks as an investment. However, there is absolutely nothing fixed about these factors and no guarantee that they are going to be the same in the future.

Determining whether stocks are a good long term investment requires an a priori analysis rather than an a posteriori analysis. Statistics is useless in unchartered territory.

The reason statistics is useless is because you are trying to gauge whether the 21st century will be like the 20th century. With statistics you are basing your conclusions on a sample size of one. Consider for instance, whether the US in the 21st century will be much like the US in the 20th century or whether it will be more like Japan or Italy, say? Statistics is only worthwhile when you have multiple observations to base the conclusion on.
This makes it possible to make predictions about the business cycle and the medium term boom and crash behavior that is caused by the credit cycle. Using statistics to predict the long term is folly though. Such prediction has to be done a priori.

Originally posted 2009-01-24 07:59:11.