Monday, August 11, 2008

Too Many Safety Margins

I’m a big fan of safety margins. When I drive over a bridge, I’m glad it has been designed for several times the weight of the cars on it. Investing strategies should be designed so that you’ll be okay financially even if your returns are lower than you hope. But, we can sometimes make the mistake of layering too many safety margins and lose track of likely outcomes.

In his book, The Intelligent Portfolio, Christopher L. Jones does some computer simulations to show that even though the S&P 500 returned an average compound rate of 6% above inflation for the past 40 years, there was a 1 out of 20 chance that the return could have been as low as 1.2% above inflation.

This analysis is based on assumptions about the distributions of stock market returns. What happens if we take a look at actual returns over the past century?

According to a chart produced by Crestmont Research, in the 58 rolling 40-year periods since 1910, the S&P 500 compound average return has ranged from inflation plus 3% to inflation plus 8%. This is the average compound rate taking into account transaction costs such as bid-ask spreads, commissions, and other fees (but not taxes).

So, even taking into account transaction costs, there has never been a 40-year period in the past century with returns as low as inflation plus 1.2%. Either we have been lucky, or Jones’ model is wrong.

It’s amazing what many people choose to worry about: very poor stock market returns for more than a generation, terrorist attacks, and meteors. If you want something real to worry about, then think about cancer and heart disease. I’m much more likely to die in the next 40 years than I am to see stocks lose out to inflation.


  1. Excellent post.

    The lowest nominal annualized return for the TSX over rolling 30 year periods since 1952 was 8.6%. I don't know what inflation was in Canada during that time, but if we can use the US numbers, inflation averaged 4.3% during that time period. So, assuming the Canadian inflation was not too far off of that, that represents a real return in the neighbourhood of 4.3%.

    The best 30 year period for the TSX on a nominal basis was 1970 t0 2000 with a nominal return of 12.7%. I DO have the Canadian inflation numbers for this time period - on an annualized basis inflation averaged 5.2%. Therefore the real return was 7.5% during this time.

    If there was a time period as long as 40 years where stocks lose to inflation, that would mean capitalism has failed and we would have much bigger problems to worry about than how much we had saved.

  2. Preet: Thanks for the Canadian numbers. Even with transaction costs the story in Canada is similar to the US -- stocks win over a long time. You're absolutely right about having bigger problems than savings if capitalism fails.

  3. Yes, bonds could outperform stocks over the long run but the probability is quite low. If stocks did under perform, an investor simply has to work and save for a couple more years.

    I agree with you that it is pointless to worry about it because what's the alternative? If most of a portfolio is in bonds, lower returns than stocks are the most likely outcome. Isn't it better to pick the alternative that provides the highest likelihood of success?

  4. CC: You make a good point about adapting to the return that you get. There is a lot of uncertainty in the returns that you will get over 30 or 40 years. Fortunately, you can adapt your number of years working and spending levels.

    The main problem with the tech bubble bursting was that people made plans based on peak prices and a continued rise. The S&P 500 returns from 1995 to the market bottom after the bubble burst were a healthy 5% per year above inflation.

  5. That's a pretty solid posts Michael. If you have a nice diversified mix of income producing assets ( stocks, bonds, real estate) who cares about long-term returns averages.

  6. Dividend Growth Investor: You're right that short-term volatility won't affect income much for a dividend investor, but you do have to pay attention to long-term returns. If you consistently underperform the market, in a couple of decades you will be generating income from a smaller base than you could have had. For that reason, I'm not a fan of bonds for the long term. Dividend paying stocks usually have a total return (dividend + capital gains) that at least matches the returns on non-dividend paying stocks, and so I'm a fan of dividend-paying stocks.