You are here:
There is a part of your brain, in the left hemisphere, that neuroscientists have dubbed "the interpreter." The apparent role of the interpreter is to assign a cause to every effect it sees. Generally, it associates good results with lots of skill and poor results with a lack of skill. The interpreter is very effective at sorting cause and effect most of the time, and this ability is what allows us to understand the world around us. But the interpreter stumbles when confronted with randomness. The interpreter construes positive results that come from favorable luck as something good. Bad outcomes are to be avoided. This is our natural mode and presents a problem to us as investors. Perhaps the most dismal numbers in investing relate to the difference between three investment returns: those of the market, those of active investment managers, and those of investors. For example, the annual total shareholder returns were 9.3 percent for the S&P 500 Index over the past 20 years. The annual return for the average actively managed mutual fund was 1.0–1.5 percentage points less, reflecting expense ratios and transaction costs. This makes sense because the returns for passive and active funds are the same before costs, on average, but are lower for active funds after costs.
The "Dumb Money Effect"
But the average return that investors earned was another 1–2 percentage points less than that of the average actively managed fund. This means that the investor return was roughly 60 – 80 percent that of the market. At first glance, it does not make sense that investors who own actively managed funds could earn returns lower than the funds themselves. The root of the problem is bad timing. Spurred on by the interpreter, investors tend to extrapolate recent results. This pattern of investor behavior is so consistent that academics have a name for it: the "dumb money effect". When markets are down investors are fearful and withdraw their cash. When markets are up they are greedy and add more cash.
Getting Excited But Losing Money
The key to understanding the dumb money effect is the distinction between time-weighted and asset-weighted returns. Let's use a mutual fund as an example. The time-weighted return measures the performance of the fund over time based on net asset value. The asset-weighted return incorporates not only the performance, but also the money going in and out of the fund. Here's a simple illustration. Let us say an investor buys 100 shares of a fund that starts a year with a net asset value of 10 US dollars, representing an outlay of 1,000 US dollars. In the next year, the fund's net asset value rises to 20 US dollars, doubling the investor's money. Excited, the investor buys an additional 100 shares, spending another 2,000 US dollars. In the second year, the net asset value of the fund declines to 10 US dollars, back where it started. How did the fund and our investor fare over the two years? The time-weighted return for the fund is zero, of course, as the fund ended at the same price as it started. But the asset weighted return for the investor is –27 percent , calculated as the internal rate of return based on the timing and magnitude of the investor's cash flows. The return would have been zero had our investor used a simple buy-and-hold strategy, and there would have been no nominal gain or loss. But in the scenario we outlined, our investor lost 1,000 US dollars of the 3,000 US dollars total investment because of the purchase after the fund rose. This basic example reflects the experience of one investor over two years, but we can apply the same methodology to many investors over multiple years. Because of investor behavior, returns for major indices substantially overstate the returns that investors actually earn. The adjacent figure shows the difference between the buy-and-hold return and the asset-weighted return for 19 countries around the world. On average, investors earn 1.5 percentage points less per year than a buy-and-hold strategy as a result of the dumb money effect. So our minds encourage us to act at extremes and buy when the market is up and sell when the market is down. The question is: How do we sidestep this behavioral bias of buying high and selling low?
We Should Learn From the Past
Daniel Kahneman is a psychologist who is renowned for his work in judgment and decision-making and who won the Nobel Prize in economics in 2002. Shortly after he won the Nobel Prize, he was asked to name the favorite paper he had written. He replied with "On the Psychology of Prediction," which he wrote with Amos Tversky in 1973. The paper is rich with insight, but, for our purpose, the main lesson is how to make a thoughtful prediction. They argued that three types of information are relevant. First is the base rate, or the outcome of an appropriate reference class. In the case of the stock market, for instance, this would reflect the historical record of returns that the Global Investment Returns Yearbook series documents. The Yearbook provides a remarkably robust database from which to consider long-term returns. Second is the specific information about the case that you are examining. For markets, that would represent some sense of valuation and what that valuation implies about future returns. The final element is how to weight the base rate and the specific information at hand in order to create a sensible prediction. In some cases, most of the weight should be accorded to the base rate. In other instances, the specific information should carry the most weight. Kahneman and Tversky suggested that we tend to underweight the base rate in many of our predictions.
Don't Rely On Your Luck When Playing Roulette
Here is one way to think about the problem of how to weight the information. If you are dealing with an activity where luck is the main factor in determining outcomes, you should place almost all of the weight on the base rate. For example, think of the spin of a roulette wheel or the roll of a die. The best estimate is some measure of the average, with an appropriate variance. If, by contrast, you are dealing with an activity where luck plays almost no role, you should place almost all of the weight on the specific information. For example, if you line up five people off the street against a world class sprinter, you know the sprinter is going to win.
We can quantify the role of luck through the correlation coefficient, which statisticians denote by the letter r. The correlation coefficient measures the degree of linear relationship between variables in a pair of distributions. When the correlation coefficient is zero, what happens next is unrelated to what happened before. Results are random. When the correlation is 1.0, what happened before tells you what will happen next. The correlation coefficient takes a value from –1.0 (perfect negative correlation) to 1.0 (perfect positive correlation). The main point, for our purpose, is that r gives you an indication of how to weight the base rate and specific information. If r is close to zero, rely on the base rate. If r is 1.0, the specific information is all you need. The correlation coefficient gives you an indication of the rate of reversion to the mean.
Making Predictions on Solid Ground
Let us consider a couple of examples to make this concrete. Take a look at the chart. On the left is the correlation between the heights of fathers and sons, which is 0.50. Part of a son's height is hereditary and part is environmental. Say a father is 76 inches tall and the average height of all men is 70 inches. To predict the son's height, you would equally weight the father's height of 76 inches (specific information) and the average height of 70 inches (base rate) for a prediction of 73 inches. Naturally, this prediction does not hold for any particular son, but it is the best prediction for a population of fathers of that height. Now examine the picture on the right of the chart, which shows the correlation between 2011 and 2012 cash flow return on investment (CFROI®) for more than 1,000 consumer staples companies around the world. Here, r approaches 0.90, which tells us that what happened last year is a very good indicator of what will happen this year. Assume a company has a CFROI of 13.5 percent and the average for the sector is 9.2 percent. The expected CFROI for the subsequent year is about 13 percent, as most of the weighting in the forecast goes to the specific information. There is some reversion to the mean, but overall the results are very persistent from year to year.
Don't Rely on Last Year's Performance
Now we turn our attention back to markets. This chart shows the correlation coefficient for year-to-year total shareholder returns for the S&P 500 from 1928 to 2013 as well as the MSCI World Index from 1970 to 2013. In both cases, the r is very close to zero. In practical terms, this means that the best prediction of next year's return is something consistent with the base rate. For the S&P 500 from 1928 to 2013, for instance, the base rate is a nominal arithmetic return of 11.3 percent with a standard deviation of about 20 percent. In 2013, most developed markets realized total shareholder returns above historical averages, led by Japan's gain of more than 50 percent and the greater than 30 percent rise in the Unites States. The MSCI World Index gained 27.4 percent. Emerging markets fared poorer, with the MSCI Emerging Market Index down 2 percent. And what about 2014? Andrew Garthwaite, Global Equity Strategist at Credit Suisse, forecasts total shareholder returns in the range of 9 percent for the United States equity market and 13 percent for global equities for 2014. The basis for this short-term forecast is that Credit Suisse's strategy team continues to believe that equity valuations remain attractive relative to bonds and that flows into equities have more to go. Naturally, a long-term forecast should appeal to the accumulation of data in the Yearbook. Since 1900, the return for US equities has exceeded that of ex-US equities by 1.9 percentage points per annum. The lesson should be clear. Since year-to-year results for the stock market are very difficult to predict, investors should not be lured by last year's good results any more than they should be repelled by poor outcomes. It is better to focus on long-term averages and avoid being too swayed by recent outcomes. Avoiding the dumb money effect boils down to maintaining consistent exposure.