Valuation, Anchoring, and Availability

William Hester, CFA
March 2004
All rights reserved and actively enforced.

The economist Paul Samuelson once said that a good question outranks an easy answer. Now that a year has passed since the cyclical low in stock prices, and the market's price-to-peak-earnings multiple is above anything seen either before or after the peak of the bubble, a few questions do come to mind.

How cheap did stocks get a year ago? How rich are they now? Do these two levels of valuation provide new upper and lower limits for future bull and bear markets?

Answers are readily available for the first two questions. The median price-to-earnings multiple for the companies in the S&P 100 index was 17 a year ago. Today it's 20, the net result of increases in both prices and earnings. The third question, though, can only be answered with the passage of time.

The boundaries of what "normal" valuations are is a vitally important issue for investors. A sure answer to it would nearly silence the debate about what returns buy-and-hold equity investors can expect. If you assume low multiples at the end of say, a 10-year holding period, it would take heroic assumptions about the growth of dividends and earnings to get a respectable return from stocks (see: Estimating the Long Term Return on Stocks).

All serious analysts assume stock returns will be lower over the next ten or fifteen years when compared to the last two decades. The debate is centered on how much lower. Those who expect stock returns to fall only slightly - delivering about an 8 percent annual return - argue that P/E multiples aren't going to fall to bargain levels again. Investors have wised up because they know that stocks are the best investment for the long run. This newfound courage in equities will turn them into buyers long before P/E multiples reach the low levels of the past.

Those less optimistic about the dominance of stocks believe that reversion to the mean is a powerful force. The recent rally in P/E multiples is only a temporary detour on the way to valuations more representative of history, they say.

If this latter group turns out to be correct, what explains these high-multiple detours? Behavioral finance provides some explanations. 1

Anchoring and Adjustment

Much of the research in the field of behavioral finance grew out of the work of two psychologists: Daniel Kahneman, who shared the Nobel Prize in Economics in 2002, and Amos Tversky (who passed away before the award was given). Their work showed that people take mental shortcuts when making decisions that involve uncertainty.

Kahneman and Tversky found that when people make a decision they start from a "reference point." This is the case even if the reference point has little to do with the decision.

For example, in one study researchers spun a roulette-type wheel labeled with the numbers 1-100. Then they asked participants the percentage of African countries that were members of the United Nations. The responses were heavily dependent on the number the wheel landed on. For example, when the number on the wheel was 65, the median guess was 45 percent of countries. When the wheel landed on 10, the median guess was 25. 2

Numerous other studies duplicated these results. Cornell MBA students were asked what year Attila the Hun was defeated. 3 Before answering the question, the students were asked to add 400 to the last three digits of their phone number. This nearly random number affected the student's answers. When the sum was between 400 and 599, the student's average guess was that Attila was defeated in AD 629. When the number was between 1200 and 1399, their average guess was AD 988. (Attila the Hun was defeated in AD 451.)

Objects in Mirror May Be Less Important Than They Appear

Kahneman and Tversky also showed that people estimate the likelihood of an event on the basis of an association that comes to mind. In other words, people estimate patterns or results by the ease in which they can recall similar items. They called this the "availability heuristic."

In one study, participants were asked if the letter K is more likely to be the first or third position of a word. The scientists assumed that if it were easier to think of words that started with K rather than those that have K in the third position, participants would overestimate the size of the first group.

In fact, twice as many participants guessed that more words began with the letter K than had this letter in the third position. This was the case even though K is found in the third position of words twice as many as in the first.

In this and other studies, the psychologists found that the mental shortcuts people use to make decisions are heavily affected by the experience of the participants. "If one's experiences are biased, one's perceptions are likely to be inaccurate," they concluded.

Stocks, Anchoring, and Availability

What are the "anchors," or reference points that investors use to interpret commonly-used metrics of stock valuation? I calculated the median ratios of price to earnings, price to book value, price to sales, and price to cash-flow for the companies in the S&P 100 index in March of each of the following years: 1990, 2000, 2003, and 2004. The month of March in each of these years provides us with market valleys (in 1990 and 2003) and peaks (in 2000 and now).

I used the median valuations within the S&P 100 because they are more conservative measures of valuation than the more popular market-value weighted measures, and are less influenced by overvaluation in the largest capitalization components. In March of 2000 the median P/E of the S&P 100 companies was 25. The market-value weighted P/E was 35.

The graphs below show a wide range between possible reference points. If you focus only on recent experience and use March 2000 as your reference point, stocks would have looked particularly attractive last year. Median price earnings multiples had declined from 25 to 17. The recent rally has brought these multiples back up to 20, still five points below the levels of March 2000. The ratios of price to book value, price to sales, and price to cash flow show the same pattern. 4

On the other hand, if you change the reference point to March 1990, the drop in the level of valuation in 2003 looks less enticing. The median P/E ratio of the S&P 100 companies was thirteen in 1990, nearly half of the March 2000 peak and 4 points below last years bottom.

Keep in mind that March 1990 is not a particularly depressed reference point. Even after a market decline in late 1990, the lowest levels of valuation that year simply matched the long-term median for P/E multiples, and were considerably above the levels reached at the depth of the market bottoms in 1982 and 1974.

The stock valuations at both the beginning and the end of an investor's holding period are paramount to the investment's return. Using the last four years as a reference point, stocks sit comfortably between the boundaries of the recent bull and bear market. Using the last fourteen years as a reference, they sit high above the lower boundary. Buy-and-hold investors, especially those that have known future expenses, need to consider both possibilities.

____________________________________________

1 For a paper outlining recent advancements in Behavioral Finance see Barberis, Nicholas and Richard Thaler, A Survey of Behavioral Finance at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=327880 . A more extensive review can be found in Shefrin, Hersch, 2000, Beyond Greed and Fear , Harvard Business School Press. See also Shleifer, Andrei, 2000, Inefficient Markets: An Introduction to Behavioral Finance , Oxford University Press. For an accessible discussion of behavioral finance and individual investors, see Belsky, Gary and Thomas Givlovich, 1999, Why Smart People Make Big Money Mistakes, Simon & Schuster.

2 Kahneman, Daniel, Paul Slovic, and Amos Tversky, 1982, Judgment Under Uncertainty: Heuristics and Biases , Cambridge University Press.

3 Belsky, Gary and Thomas Givlovich, 1999, Why Smart People Make Big Money Mistakes, Simon & Schuster.

4 I calculated the same valuation metrics for the companies in the S&P 500 index. There is no difference in the relative patterns.