There is no disagreement of the profile of the ideal stock -- ceteris paribus, every investor would like to purchase those that have low valuation, high projected earnings growth, positive recent performance, high profitability, etc. Given the thousands of active managers and investment analysts, it is not surprising that such an "ideal" stock does not exist. For example, screening the Russell 1000 universe for just three favorable characteristics results in only 37 securities.
This result is not surprising. On average, investors recognize that companies with higher expected earnings growth should have higher valuations, and bid their prices up. The same is true for those companies that have had strong relative recent performance or have high book values. This rational behavior makes it virtually impossible to find companies with low valuations and unambiguously positive growth prospects.
In such an environment, how does the aspiring active manager attempt to beat the market?
In their search for undervalued securities in the presence of uncertainty about its future prospects, investment managers are therefore forced to make complex trade-offs in selecting their stocks. Value managers focus on companies with low valuations, but attempt to identify those whose low earnings growth is temporary. Growth managers attempt to identify those companies that they believe will experience higher growth than is reflected in their stock prices. Others focus on multiple themes such as "growth at a reasonable price."
Making trade-offs among different characteristics is exceedingly difficult when there are multiple characteristics to consider.
Human mind's limitations
There is growing evidence pointing to the limited ability of the human mind to make optimal trade-offs when faced with tradeoffs in more than two or three dimensions. Professors Daniel Kahneman and Amos Tversky, who pioneered empirical research in this area, note that humans "when faced with a complex problem .*.*. use computational shortcuts and editing operations."
They attribute this type of behavior to two human shortcomings: the tendency for emotional biases to affect the ability of individuals to make rational decisions; and the inability of individuals to fully understand the trade-offs inherent in any decision.
Or equivalently, in the jargon of psychologists, they experience "cognitive difficulties." Investors often fixate on one variable as opposed to the whole range of variables. Analysts frequently get hung up on the most recent piece of information they receive or make distorted assessments of the long-run consequences of the information -- witness the often precipitous drop in a stock's price when earnings are under expectations by a penny.
These cognitive failures manifest themselves in inefficiencies in the equity market. Recent studies have demonstrated numerous such anomalies, including:
*Stocks that are "cheap" by some valuation measure -- the ratio of price-to-sales, price-to-book, price-to-earnings or price-to-cash flow -- tend to outperform stocks that are expensive by the same measure. Investors appear to forget that competitive forces limit the ability of firms to earn abnormal profits in the long run.
*Stocks that have positive performances as measured over the prior six to 12 months tend to outperform their peers over the next 12 months -- nobody wants to miss out on owning a "winner." This behavior leads to short-run momentum.
*Poor performance over the prior three- to five-year period is associated with out performance over the next period. Investors appear to apply too high a discount to stocks that have had negative returns in the past.
*Stocks that have unexpectedly high earnings tend to decline in value when a component of these earnings is reversed over the following year and vice versa. In other words, analysts tend to expect the high level of earnings growth to continue too far into the future.
Good news? Maybe not
These findings should be good news for active managers who -- if they can overcome their cognitive biases -- will be able to consistently outperform the market. In reality, however, active managers find it hard to beat the market as measured by the Standard & Poor's 500 Stock Index. Both individual investors and institutions have responded to this underperformance by indexing a greater portion of their equity assets.
Does this suggest that the market does in fact reflect rational prices?
On the contrary, Peter Bernstein, in his recent book "Against the Gods," ascribes this systematic inability of the average investment manager to beat the S&P 500 as evidence that market prices do not reflect rational valuations.
The S&P 500 contains a broad cross section of stocks, not just those that investment managers consider attractive. Managers arguably underperform the index because they systematically overestimate the value of the stocks they purchase. They also likely underestimate the value of those that they sell. The S&P 500 index manager does not fall into this trap because he or she purchases stocks regardless of their level of over or under valuation -- thus paying, on average, a rational price.
Overcoming cognitive biases
One way to attempt to overcome these biases is to use a process of sequentially screening stocks into portfolios based on one or more variables that are correlated with known valuation errors.
The most common approach consists of focusing on a single characteristic -- such as price to book -- that is systematically related to valuation errors.
The irony in using such an approach to exploit market mispricings that arise as a result of behavioral biases is that their users are falling into the same trap that caused the inefficiency in the first place -- the tendency for investors to take shortcuts in dealing with the complexity of equity valuation. More variables can be considered in the screening process but this often results in a portfolio containing an impracticably small number of securities.
Another shortcoming of the screening process is its failure to account for the fact that the value ascribed to characteristics should be different both in terms of their magnitude and variability. The impact of different characteristics varies in magnitude over the business cycle.
Without explicitly forecasting the value of different characteristics (profitability, earnings growth, operating margins, interest coverage) it is impossible to make rational tradeoffs in choosing between different stocks.
Multitude of factors needed
Clearly, overcoming these cognitive deficiencies requires a valuation model that takes into consideration the multitude of factors that are important in valuing a security; in short, the model should mimic the behavior of a rational investor.
One of the first systematic studies of the benefits of using such a "rational" approach, in which more than 50 characteristics are quantified and explicitly taken into account in valuing a security, was conducted by Professor Robert A. Haugen and Nardin Baker in 1996. In this study, a quantitative model is used to make trade-offs between different characteristics when evaluating the attractiveness of each stock.
The weight given to each characteristic is modified based on the observed statistical relationship between it and subsequent returns. Relationships are not assumed to follow a simple linear pattern, but rather are empirically determined. Finally, an optimization process is used to capture the inter-relationships among securities in combining them into a portfolio and to take full advantage of the power of diversification to control risk.
In an efficient market, such a strategy would not generate above-market returns. However, using realistic estimates of transaction costs, the Haugen study demonstrates it is possible to build portfolios that consistently outperform the market by average rates of 4% to 6% on an annual basis.
The average characteristics of the portfolios are unusual in that they have profiles, in terms of profitability, earnings growth, and valuation, that are not met by any single stock. This indicates that institutional investors do not accurately evaluate and subsequently arbitrage away the complex trade-offs among different equity characteristics.
Thus, the optimized portfolio is given all the characteristics sought in an equity investment but never available in any single security.
In a market dominated by rational investors it would be nearly impossible to construct a portfolio with such desirable characteristics. This result also is found to hold in major international markets including Great Britain, Germany, France and Japan.
Evaluating an REV model
How well can a rational model identify under- and overvalued securities? We evaluated the performance of such a rational equity valuation, or REV, model over a 13-year period ending in 1996. The performance of this model was compared to the commonly used method of focusing on one particular characteristic or valuation parameter.
Model performance is evaluated in terms of its information ratio, or IR, which is a measure of the ability of the model to discriminate between good and bad stocks. The IR is computed by calculating the average return from a "long-short" portfolio -- consisting of a long position in undervalued stocks and a short position in overvalued stocks -- and dividing the average return by the standard deviation of these returns.
If the model has no power, the average difference in returns between the long and short portfolios will be zero, and on a monthly basis, the difference will vary randomly around zero.
In these tests, long-short portfolios were constructed by computing the performance differential between the most undervalued and most overvalued quintiles. The sample universe consists of the largest 1,000 stocks on the NYSE, AMEX and NASDAQ.
The performance of the REV model is superior to the more simple models not only in terms of its ability to find over- and under-valued securities, but also in terms of the consistency of its performance. The information ratio of this model far exceeds that of the other models.
Can it outperform the S&P 500?
Such a strategy can be used in a number of different strategies, varying from traditional core equity to leveraged "long-short." The potential for value added will be determined by the level of "active risk" inherent in the strategy.
In the Analytic/TSA Global Asset Management enhanced equity strategy, we use such a rational equity valuation approach to construct an equity portfolio with the objective of outperforming the S&P 500 index.
The resulting portfolio tends to be well-diversified, with approximately 70 securities, and an average market capitalization similar to that of the S&P 500.
As our research indicates that inefficiencies do not persist across industry groups, the portfolio's industry exposure is constrained to mirror that of the S&P 500
The performance of such a strategy is shown at left over the time period 1991 through 1997. Prior to October 1996, the results reflect simulated returns. Subsequent to September 1996, the performance represents the actual returns of the Analytic/TSA enhanced equity composite.
Overall portfolio risk as measured by annualized standard deviation is 9.7% vs. 10.1% for the S&P 500 over the entire time period. The beta of the portfolio is 0.91. Despite this risk profile, the strategy outperformed the S&P 500 by approximately 4 percentage points over this period. Over the one-year period ending September 1997, the 12-month period in which the returns represent performance on a live account, the strategy outperformed the S&P 500 by more than 11 percentage points.
We believe that this approach represents a superior alternative to overcoming cognitive biases than simply investing in an index fund. Committing funds to such an approach, however, requires the fundamental belief that a quantitative valuation model, when applied in a disciplined fashion, is capable of overcoming the emotional irrationalities to which humans often succumb.
Roger G. Clarke is chairman and Harindra de Silva is managing director of Analytic/TSA Global Asset Management Inc., Los Angeles.