Have you checked the latest returns for the Thrift Savings Plan’s investment funds? You’ll find them summarized in the accompanying chart.

Interesting? Maybe. Useful? For what? So you can calculate the recent value of your account? You can go online at www.tsp.gov or call the TSP ThriftLine at (877) 968-3778 any time and check your account value without the need for a calculator. The effects of history on your account are always reflected in its current balance. There’s really no need for an investor to know the details of the returns that produce the result, only the result itself.

Of most interest to investors should not be what has happened in the past, but what will happen in the future. This is where historical information can be useful. The problem is that short-term results — those for a month, a quarter or even a year — are extremely unreliable predictors. The fact that the S Fund produced a 0.37 percent return during the month of December tells us nothing special about what will happen in January or any other future period.

Since TSP is a retirement plan, it is reasonable to assume that the relevant period of importance to most participants is measured in years, even decades, rather than months. Like any gambling proposition, the law of large numbers applies. That is, the more data points we consider, the closer their average value will be to the expected or true result. Flip a fair coin 10 times and you may get seven heads. Flip it 10,000 times and you’ll get a number much closer to 5,000 heads — the expected result.

If we use history to predict future investment results, longer history will produce more reliable predictions than shorter history. Similarly, predictions will be more reliable when applied to long-term future results than to those in the short term. Using the past day’s return to predict the next day’s is an unreliable method, to say the least. The same is true for monthly returns. Even annual returns prove extremely unreliable predictors. In order to maximize the reliability of our predictions, we’d like to use all available history as our basis. The problem is that the availability of accurate performance records diminishes as we go back in time. Additionally, the history of our economy is relatively short and it has changed fundamentally during that history. About the best we can do is to use the last 100 years or so to make some predictions. Once produced, these predictions will be more reliable when applied to long-term average results than short-term results in the future.

As an example, let’s consider an investment in the C Fund, which invests in the stocks of large, established domestic companies — large-cap stocks — as represented by the Standard & Poor’s 500 index. I looked at the past 80 years of data for large-cap domestic stocks and calculated the average annual return for 50 unique 30-year periods.

The lowest 30-year compound return was 8.47 percent for the period from 1929 through 1958; and the highest, 13.72 percent for 1970 through 1999. The median return was 10.82 percent for 1948 through 1977.

The results can be used to make predictions about the future performance of the C Fund. First, we need to adjust the returns for the expense incurred for the operation of the fund. For 2004, the expense ratio for the C Fund was 0.06 percent, but the expenses can vary from year to year. Based on history, 0.10 percent might be a conservative estimate for future fund expenses. This adjustment would lower the realized median compound return from large-cap stocks from 10.82 percent to 9.82 percent. This will serve as our expected future result from the C Fund — a compound average annual rate of return of 9.82 percent. In other words, over the next 30 years, the expected, or most likely, result from the C Fund is 9.82 percent per year, compounded.

The fact that there will be years when the actual return varies considerably from this expected result will not lead us to change our expectation. And, as a long-term investor, you should not be depressed by unusually bad years or excited by good ones. These variations were considered in our prediction and were instrumental in producing the expected long-term result. While there are no guarantees, as long as your time horizon remains sufficiently long — say 15 years or more — you’ll likely recover your extraordinary short-term losses and give back those excessive short-term gains before you are done.

Written by Mike Miles
For the Federal Times
Publication January 16, 2006