Portfolio Management

Owen Lamont, Phd, Portfolio Manager, DKR Capital
Originally published in the February 2007 issue

What is behavioural finance?

Behavioural finance is the study of how asset prices can be wrong, in the sense that the actual price of an asset can be above or below the fundamental value. Behavioural finance stands in opposition to the efficient market hypothesis, which states that asset prices are always equal to their fundamental value. Behavioural finance has emerged in the past 20 years as a competing branch of academic research.

Behavioural finance consists of two key components. First, there has to be something that causes prices to be wrong in the first place. For example, perhaps some investors are irrationally gloomy about IBM, and thus value it at less than its true value. Second, there has to be something that prevents other, more rational investors from correcting the mistakes made by these irrational gloomy investors. This second component is called 'limits to arbitrage'.

The first component, the original causeof the mispricing, could be either a specific feature of investor psychology, or it could be some sort of institutional effect. Much of behavioural finance has focused on investor psychology. Briefly, a large body of evidence shows that investors make systematic errors in evaluating information. They may over-react to dramatic news and under-react to mundane news. So, for example, if Apple has an exciting new Ipod product, investors become over-enthusiastic about Apple and it becomes overpriced. Meanwhile, investors may undervalue IBM despite its solid but boring fundamentals. There are many other ways in which some investors seem to behave irrationally. Both emotion and simple cognitive errors play a part.

There are several limits to arbitrage that prevent rational arbitrageurs from quickly pushing prices back to fundamentals. These include various trading costs as well as risk. The risks associated with correcting mispricing include both sentiment risk and fundamental risk. The classic example of sentiment risk is the tech stock bubble of 19992000. NASDAQ stocks were overpriced in early 1999, but got even more overpriced by early 2000. A rational arbitrageur attempting to correct this mispricing faced huge sentiment risk.

Evidence supporting behavioural finance

A variety of different evidence supports the idea that investors make mistakes and that these mistakes affect asset prices. One type of evidence comes from the behaviour of individuals, both in lab settings and when making actual investment decisions. For example, many studies have shown individuals make basic errors in choosing how to diversify, what types of evidence to consider, and how much to trade. A second type of evidence comes from asset prices themselves. According to the efficient market hypothesis, it should be impossible to predict returns or to earn profits from trading, yet a variety of robust patterns have been identified that allow one to construct profitable trading strategies.

A third type of evidence comes from connecting the behaviour of different classes of investors with asset prices. Under the efficient market hypothesis, the trades of any one class of traders should not predict future returns. In fact, recent research indicates that some investors are smart money, and are able to earn excess returns. Other investors are dumb money, and consistently hurt themselves through their trading. The smart money appears to be professional investors (such as short sellers) and the issuing firms themselves, while the dumb money appears to be individual retail investors. When you see, for example, retail investors frantically buying tech stocks in 1999, and tech companies frantically issuing stock, that's a good sign the dumb money is buying and the smart money is selling.

Strategies from behavioural finance

Behavioural finance produces three categories of investable strategies. First, there are under-reaction strategies. This group of strategies reflects the tendency of stock prices to under-react to specific events. For example, when bad news (say, an unexpectedly low earnings announcement) occurs for a specific stock, the stock price immediately falls, but does not fall enough. On average, it will continue to go down in the next few months. An under-reaction strategy would hold long positions in stocks that recently had good news, and short positions in stocks that recently had bad news.

The second category is value-type strategies, reflecting mispricing that persist for many months or years. The value effect is the fact that over long periods of time, value stocks (measured by price/book or some other valuation ratio) outperform growth stocks. The value effect has been studied by academics since the 1980's and has been shown to hold true in many different countries and time periods. It reflects the fact that sentiment causes some stocks to be overpriced (growth stocks) and some to be underpriced (value stocks).

One can use many signals in addition to valuation levels. Value-type strategies go long on firms that are repurchasing stock, have high quality earnings, and have high free cash flow. The strategies go short on issuing, low quality firms that seem 'speculative'. Value-type strategies are contrarian. They tend to buy stocks that investors don't like and sell short stocks that investors love.

These different measures of mispricing – issuance, valuation ratios, earnings quality – all tend to be correlated across stocks. This fact suggests that these various measures are all reflecting the same underlying phenomenon, namely mispricing. In thinking about this cluster of attributes associated with mispricing, it is useful to consider an example that captures many of these common elements. Telecom stocks in 1999 had many of the attributes correlated with overpricing. They had huge valuation ratios and were issuing large amounts of shares. Not surprisingly, telecom stocks had low returns subsequent to this episode.

The third category is technical strategies, which are a variety of different strategies based on variables such as volume, events and short-term price trends. An example of a technical strategy is the 'earnings announcement premium'.

The earnings announcement premium

It is a striking fact that, as first noted in a 1968 accounting journal, all stocks tend to rise around their scheduled earnings announcement date. That is, some stocks rise on earnings news, some stocks fall, but on average the winners systematically outweigh the losers.

The accompanying graph (Returns to Anticipated Announcement Strategy, Jan 1973 – Nov 2005) shows the twelve month moving average returns from the following simple strategy. Buy every stock on the first day of the calendar month in which you expect an earnings announcement, and sell it at month end. For example, if you expect IBM to announce quarterly earnings on 15 July, buy it 1 July and hold until 31 July. We will hedge this long portfolio by shorting all stocks not expected to announce this month. Thus for every stock we will go long four times a year and short the other 8 months. The graph shows that on an annual basis, the announcing stocks outperform the non-announcing stocks by six percent per year (these calculations do not include transactions costs, which would of course lower the portfolio returns).

What explains this effect? One explanation from behavioural finance is the attention-grabbing effect. When IBM announces its quarterly earnings, this news is noticed by unsophisticated investors. These investors tend to buy IBM when it is in the news, and this wave of buying pushes up stock prices. Studies have shown that individual retail investors tend to buy on any news, whether it is good or bad.

Why do these strategies keep working?

One might expect that these various strategies would stop working as more and more investors become aware of them. Yet this doesn't appear to have happened. For example, the graph seems to show that the earnings premium is fairly constant over time, even though several academic studies have addressed it since 1968. The value effect has worked amazingly well in the past five years, just as well as it has in the previous 70 years. Why hasn't this smart money eliminated these patterns?

One answer to this question is that these strategies keep working on average precisely because they only work on average. Not every value stock goes up. One can only detect and exploit these patterns by looking at very diversified portfolios consisting of hundreds of stocks. In addition to working only in large groups of stocks, you also need to look at long time periods consisting of many years. For example, many observers wrongly concluded that value investing had 'stopped working' as of 19992000, only to see value stage a huge comeback in the next few years. As long as these strategies are hard to detect and difficult to evaluate, they can continue to generate profits on average.

Another answer to this question is that it is hard to judge which out of the thousands of possible investment strategies will work going forward. There are many voices claiming to have a way to beat the market, so the item in short supply is not information, ideas, or even IQ, but rather wisdom. These judgment calls require knowledge of the underlying behavioural theory, experience in evaluating empirical evidence, statistical sophistication and formal training in using financial modeling to construct portfolios. These attributes are scarce commodities, but one place they can be found is in individuals with academic finance backgrounds. Access to cutting edge research is particularly valuable right now, since financial economics is in an exciting state of transition. Scientific understanding of financial markets is changing rapidly as behavioural finance researchers make new discoveries.

Why systematic strategies?

Having identified various patterns that have been uncovered by researchers in behavioural finance, the next step is to design a system to exploit these patterns. The proper way to do this is with a systematic approach. As suggested by the limits of arbitrage discussion, one major issue is risk. What empirical studies have shown is that, for example, on average, stocks with good earnings news beat stocks with bad earnings news. But for any individual stock, there is a lot of random noise. Thus to properly exploit the systematic pattern, one needs to construct a well diversified portfolio that minimizes idiosyncratic risk.

In addition to diversifying across stocks, it is also useful to diversify across strategies. That is, a portfolio that combines both value-type strategies and under-reaction strategies is far superior to a portfolio with only one of those categories. The reason is that value-type strategies tend to have returns that are negatively correlated with returns from under-reaction strategies. After all, value strategies tend to buy firms with falling prices while under-reaction strategies tend to buy firms with rising prices. Thus putting these two types of strategies into a combined portfolio results in a substantial decrease in risk.

A properly constructed systematic approach can produce returns that are consistent over time. A key element to engineering consistency is a sufficient historical period with which to test the various substrategies. Generally, the academic studies mentioned here use long historical sample periods to test for patterns. These studies typically go back to 1963, and in some cases back to 1926. Using these long periods to simulate various strategies allows one to be confident that the resulting strategies are robust to different market conditions. It also allows one to estimate how the different substrategies move together over time.

Exploiting for profit

Behavioural finance says that prices can be wrong, and it is possible to exploit this mispricing to earn high returns. By using patterns uncovered by rigorous scientific studies, combining a wide variety of strategies that exploit various behavioural effects and superimposing risk management tools, one can construct portfolios that deliver consistent profits over time.