The wise investor digs deeply into market data

In last week’s column, I made the point that the statistical term “average” is widely used as an estimator of future conditions. If the average stock market return over the past 50 years was, say, 11 percent, then we are told to expect that same 11 percent going forward.

However, I noted that, if the average annual daytime temperature at Lake Tahoe is 60 degrees, should I expect it to be 60 degrees when I go skiing in January?

Certainly not. I need to sort the database first and look only at past January temperatures. I find that a better estimator of conditions is 38 degrees.

When forming an estimator of future market conditions, we cannot merely look at the entire past string of figures. We need to look at subsets of those figures that most reflect the conditions we face right now.

Using a data set of stock market prices, earnings and dividends going back to 1900, I can calculate that the average annual return to a buy-and-hold investor was 6.53 percent, including dividends. From 1950 till now, the return was 8.73 percent.

Many analysts claim that we should only consider data from 1950 forward. Should I, then, use 8.73 percent as my estimator for future returns?

In keeping with my weather theme, I would want to look at periods in the past where stock prices started out at valuation levels similar to today’s. When I read or hear bullish stock market forecasts, they usually involve reference to the price/earnings ratio.

Commentators will note that the price/earnings ratio on the market is about 16.7. The average ratio since 1950 has been 15.6, so by this measure, the market is within a reasonable historic range.

But, I don’t care about the full period of averages. What I want to know is this: What have been the returns on stocks in a five- or 10-year period following a time when the price/earnings ratio is 16.7 or higher?

I want to know what I should expect going forward from today. To get at that estimator, I want to look at past periods that were like today. If I had bought a portfolio of stocks at a time in the past when the price/earnings ratio was 16.7 or higher, my average annual return was only 5.9 percent.

If I make a chart of the data, I can see that the second half of the 1990s was an anomaly. Despite already high prices, returns continued to be strong. That, of course, led to the bubble and the bust.

I am inclined to toss out that period since I really don’t want to go through that again. I liken this to ignoring the fact that, in one January many years ago, the Tahoe weather was 60 degrees and sunny. I’m going to toss out that data and pack my winter clothes anyway.

If I throw out the returns for the period of irrational exuberance from 1996 to 2000, I find that the average five-year return to stock holders following a time when the price/earnings ratio was 16.7 or higher was only 3.3 percent.

How do you like the market now?

In order to achieve satisfying stock returns for a five-year period - which I would define as 8 percent or better - the price/earnings ratio has, on average, been 13.3 at the time the stock investment was made. At the earnings levels today, that would put the S&P 500 at a price more than 300 points below its current level.

It is simply not enough to link the average historic price/earnings ratio to the average historic rate of return and conclude that we should expect average results going forward. The wise investor will dig more deeply into the data and find guidance.