What Reversion to Which Mean?

Posted by on 01/20/10 in Markets, Risk

Literature is all well and good, but sometimes an investor simply needs to sit with numbers and divine the future without the burden of philosophy. While there are plenty of market-based tea leaves to ponder, we thought we’d keep it simple this week by taking a look at the widely referenced Standard & Poor 500 stock index, one measure of the largest companies headquartered in the United States based on their total market capitalization (number of shares of stock * the value of each share).

Much is made about the concept of “Reversion to the Mean”—the simple notion that securities (or in the case of the S&P 500, a class of securities) will revert to the average of their previous performance. It’s an interesting notion that we think worthy of serious examination on multiple fronts, including these two:

  1. It presumes that securities and asset classes are governed by laws as one might find in physics or chemistry rather than patterns that may or may not be replicable in the future.
  2. It plays neatly into the human desire for order and predictability, and yet in so doing, may mask the reality of disorder and unpredictability in the financial markets.

Let’s examine the S&P 500 as it stood at the end of 1979, when the 20 year mean (or “average”) return of the S&P 500 had been 6.77% per year. Some would have argued that it would have been reasonable to expect equities to earn something similar going forward from there. However, by 1999, the S&P had actually earned an annualized return of 17.85% per year for 20 years, a 9% per year premium over the previous 20 years!

Next, let’s revisit year-end 1999. We know that the 20 year annualized return had been 17.85% per year, and the annualized return from 1956 to 1999 (45 years) was a similarly impressive 16.02%. So, standing at the end of 1999, what should one consider the appropriate mean of the S&P 500?  Is it the figure from 1960-1979? 1980-1999? Or perhaps 1956-1999? And more importantly, does the mean really matter? Is it really a helpful predictive tool?

In a word, “No.” During the five years following 1999 (2000-2004), the S&P 500 declined by 2.30% per year. For the next ten years (2000-2009), it lost 0.95% per year, or lost roughly 9.13% cumulative over the decade. So what? Exactly! We’re not certain that what the S&P 500 (or most other markets) returned previously matters as much as some would claim. After examining the data in 5, 10, and 20-year segments, whether unadjusted or adjusted for inflation, we would be hard pressed to even define a reasonably precise “mean,” let alone argue that investors can expect reversion to any such average. Alas, at the end of the day, investors are still left asking common sense questions like,

What are reasonable earnings assumptions going forward?
What appear to be the relative values of various asset styles or classes to each other?
What risks have I taken and how can I mitigate them?
How do I plan for unforeseen events?

We suggest removing “Reversion to the Mean” from one’s investment vocabulary unless the mean can be specifically qualified and quantified. It is less significant that the S&P 500 averaged 11.92% per year from 1956-2009 because of the extreme variability during those years. During that 54 year span almost every 10-year period that exceeded 11.92% annualized return took place between year-end 1984 and year-end 2001—hardly macro-economic conditions we could expect to see again anytime soon (e.g. falling interest rates, falling taxes, undervalued equities, etc.). The mean for all of the other 10 year periods was 7.07% (year end 1965-1983, and 2001 to 2009). That’s a big difference.

So what will the S&P 500 do this year? This decade? We don’t know. And neither does anyone else (which is extremely comforting to those who really believe in the sanctity and viability of markets). We realize (and hope) that we’ve gone and made a muddle of things again, and trust that you will forgive us for reverting to such (mean) behavior.