Chapter 4

“What do you want to run for this year, Joe?”
“Treasurer.”
“But you cannot even add!”
“I didn’t say I wanted to run for assistant treasurer.”

MODERN CAPITAL THEORY encompasses efficient-market theories and efficient-portfolio theories.

The efficient-market theory is based on the assumption that most of the markets for common stocks can be regarded as having a sufficiently high volume of trade and sufficient flexibility so that no individual trader has a perceptible effect on price and special information will be discounted by the market almost immediately. It indicates that the small outside trader can neither influence nor beat the market by himself.

The basic idea behind an efficient portfolio is as follows. Suppose that an individual knows how he reacts to risk and that he must choose a portfolio from a collection of different financial instruments, such as bonds, stocks or cash. Furthermore, suppose that he must evaluate each instrument accurately in terms of risk and expected return, and that he is required to evaluate how the risk on one instrument is related to another. (He must know, for example, how changes in price of the stocks of two companies in the same industry are related; that, say, tobacco stocks or chemical stocks generally show a tendency to move as a group.) Given all of this information and the presumption that it is accurate, an efficient portfolio is one that contains a mixture of holdings such that it provides the largest expected return for a given level of risk.

It is our view that there is nothing in this book to indicate that the theories noted above are anything but valid and useful when viewed narrowly from the perspective of outside investors who strive primarily for total return. That is, efficient portfolios and efficient markets seem to describe well the environment faced by the stock trader who does not happen to be someone who also obtains something off the top in the form of fees, commissions or other trading advantages.

Total return at any given moment is a maximum valuation as measured by the sum of the present market price of securities held, and by the cash income derived from interest and dividends during the period securities are held. But many investors cannot be characterized as the kind of total-return, outside investors or traders to whom modern capital theories are applicable. In any event, this book is not directed toward nonactivists who attempt to beat the market continuously.

For all others engaged in the purchase, sale and holding of equity securities, modern capital theory as embodied in efficient-market and efficient-portfolio hypotheses is irrelevant.²¹ It is to them that this book is directed. Specifically, technical theories lack relevance to those outside investors who are primarily interested in income, to dollar averagers, to special-situation investors who ignore timing considerations and to all activists.²²

Efficient-portfolio theory assumes implicitly that at any given moment, the price of any stock is in equilibrium, balancing the perceived prospects of reward against the perceived risk of loss. As this book points out, however, even though stock prices may appear to be equilibrium prices to total-return, outside investors, there is no reason why that should be true for anyone else. Indeed, considering the premiums paid to acquire control of companies, it is rather obvious that the equilibrium price that an acquirer is willing to pay for control of a business is different from the equilibrium price that outside investors pay to purchase shares of stock in the open market. These two equilibrium prices are, at most, only loosely related to each other. And, of course, the two prices should not be expected to be close, because the variables the control buyer considers to be important in making his investment decisions are usually different from the variables most outside investors believe are crucial to their determinations of value.

We believe that there are many efficient markets or close-to-efficient markets, such as markets for money, many commodities and high-grade corporate bonds. From the perspective of the total-return, outside investor—that is, the securities trader seeking a maximum short-run return—even markets for common stocks may be efficient. However, such efficiencies seem less than appropriate considerations for common-stock investors who have goals other than short-run profit maximization. For one thing, specific items of information mean absolutely different things to different common-stock investors. For example, a large supply of stock overhanging the market would logically be a reason for encouraging the total-return, outside investor to sell. Equally logically, such information would be a reason for encouraging a potential control buyer to start a long-term stock accumulation program.

For another thing, efficient-market theories assume that outside investors and their advisers are highly capable and are able on the average to interpret information properly. As far as we can tell, there is no empirical evidence to support such a view. Indeed, we suspect that many stock researchers and security analysts are relatively incompetent in either analyzing or understanding businesses. Incompetence of analysts, incidentally, is just as logical an explanation of why equity markets seem efficient as the view that information is competently absorbed and immediately reflected in market prices. Unquestionably, there are many, many extremely competent practitioners within the financial community. Relatively few of them, however, seem to make a career out of servicing the perceived needs of total-return, outside investors. Rather, the best and finest brains on Wall Street seem to gravitate to fields such as arbitrage, corporate finance, private placements, mergers and acquisitions, and becoming control principals in companies.

THE COMPUTER AND MATHEMATICAL ANALYSIS

To date, the contributions of computers and mathematical analysis to security analysis, corporate finance and portfolio analysis appear to be limited to technical rather than fundamental approaches. These methods simply are not useful much of the time in most investment situations. This is so because most of the time in the complex world of finance, the sneaky little nonquantifiable variable or the ugly little fact lurking in some financial legal document that has been left out of the model happens to be a key factor in an analysis.

Does this mean that the mathematical model builders who study financial problems and who grind out large computer runs are useless both in practice and in the creation of a body of knowledge that may eventually be useful? No, it does not. What it does mean is that they, like cost accountants, lawyers, bookkeepers, and other technicians and consultants, have a limited applied value to actual investing. They can help to work out details and to structure the banks of information they are given. They can tell you what the real yield will be from a debenture with an 8 percent coupon, selling at 84¾, prepayable at the issuer’s option at 105 after five years and maturing in 1990. But there is nothing to indicate that in general there is anything in mathematical finance that makes it a very good tool for picking the key controlling variables and factors in an equity-investment situation that involves more than a straightforward problem in actuarial science or a comparison of relative costs of money.

This will become clearer as we look at the three major topics that fall under the broad category of mathematical systems applied to finance. They are (1) systems for playing the market, (2) arbitrage and (3) the design and balancing of portfolios.

ON SYSTEMS FOR PLAYING THE MARKET

I’ve a system that’s devilishly clever
that I have learned from a croupier friend,
and should go on winning forever,
but I do seem to lose in the end.

CANDIDE

The dreams of the roulette player, the horseplayer and the technical-market analyst are all variants of the same belief: that just by studying the previous spin of the wheel, the form sheet or the action of the market, a magic mathematical formula will enable the market player to use a “scientific system” to beat the game.

Granger and Morgenstern ²³ and several others have shown that for the individual devotee who confines himself to following the movements of the market, tomorrow’s prices are linked to today’s by a random walk. In other words, there is no system that can remove the uncertainty of period-to-period fluctuations.

In particular, Granger and Morgenstern cast considerable doubt on a series of old wives’ tales. Are there significant seasonal variations in stock prices? They find no evidence. Do certain stock-price movements lead others? The answer is no! Can stock prices be predicted from the “technical analysis” of price charts? The answer is no.

There are two groups of individuals who limit their interests to prices and trading volume, without concern for detailed study of the firms behind the stocks and the economy behind the firms. These are (1) chartists, or technical-market analysts, and (2) random-walk theorists.

The chartists believe that by studying the charts alone they can divine market psychology and the dynamics of price movements. A true chartist, who drinks in his numbers “straight,” may even avoid reading company reports because they would dilute his thinking.

The best-known chartist theory is the Dow theory, and a devotee of this and the many other chartist techniques soon learns his own language and looks for “triangles,” “heads and shoulders,” “significant reversals” and so forth.

The first question we might ask about the chartist approach is, Is it necessarily nonsense or illogical or irrational? The answer is no. One could agree that the movements of the market reflect aggregate behavior. Perhaps in the future, some behavioral scientist will be able to find a chartist method that works. Up until now, however, the track record of the chartists taken as a whole seems poor.

Opposed to the chartists are the random-walk theorists. They are frequently completely misunderstood by most of those who have heard of them. All that the random-walk theorists claim is the reverse of what the chartists claim. The random-walk proponents argue that at any instant, price changes follow no predictable pattern. In the language of the statisticians, they act like a set of random numbers.

It is important to understand that the random-walk theorists are playing the same game as the chartists. They are allowing themselves as information only prices and volumes of trade. They are not concerned with prediction based upon any type of basic economic analysis or inside information.

The evidence supports the random-walk theorists. Using the relatively advanced statistical techniques of spectral analysis, Granger and Morgenstern²⁴ as well as others such as Mandelbrot²⁵ have obtained clear negative results. Given the trading information alone, there is no evidence that short-term price changes can be predicted. We agree basically with the conclusions of the random-walk theorists. Using only trading information, we think there is no predictability of short-term price movements in the markets.

The results of the random-walk theorists should tell the intelligent investor something that some of us strongly suspected—that is, that outside investors could spend their time more wisely than by trying fancy calculations or even simple rules on the numbers that do not have the story. The random-walk results imply that the various systems—filter rules, formula-timing theory and so on—will not work.²⁶

The general idea behind a filter rule is as follows: using a 2 percent filter, if the daily closing price of a security goes up at least 2 percent, buy it and hold it until the price moves down at least 2 percent from a high. At that time, sell and go short until another significant reversal, at which point you cover and go long again.

The formula-timing plans provide simple investment rules. Graham and Dodd attribute the original plans to the administrators of the Yale University and Vassar College endowments. There was a flurry of interest in these plans in the late 1940’s, as can be seen from the writings of Ketchum,²⁷ Weston²⁸ and others. The simplest plan is dollar averaging, which has the investor purchase equal dollar amounts of securities at equal intervals of time.

Possibly the best thing that can be said about the blind application of mechanical rules is that they should prevent the investor from being suckered into go-go markets or into the mass-psychology flights of fancy that sweep across the exchanges on occasion.

ON ARBITRAGE

Arbitrage is an occupation for the professional. It is a special topic, and it calls for a detailed understanding, plenty of calculation, a minimization of trading costs (a common-stock arbitrager probably should be a member of a New York Stock Exchange firm) and a meticulous attention to detail. Thorpe and Kassouf have a good professional book on the subject, which describes what arbitragers do and leaves the reader with a clear impression that a professional arbitrager of this variety earns his extra returns from the market by work.²⁹ Since these authors are what they claim to be, they can afford to publish their system because they know that few readers will be dedicated and skilled enough to follow their advice in detail. The disadvantages of the possible increase in competition is more than offset by the publicity and professional recognition resulting from publishing their book. Furthermore, much of Thorpe’s competitive edge is protected by his own computer programs. Professional arbitrage, especially risk arbitrage, is discussed briefly in Chapter 17.

PORTFOLIO BALANCING

An important development in the application of relatively mathematical methods to the stock market came about with the work on portfolio selection, originally developed by Harry Markowitz.³⁰ On Wall Street today, fat books of calculations on security-risk evaluation and the beta coefficient are churned out monthly by such firms as Merrill Lynch Pierce Fenner and Smith. “Beta” is the estimated market sensitivity of a stock, measured in terms of an expected incremental percentage return associated with a one percent change in return of the Standard and Poor’s 500 Index or a comparable index.

What are the assumptions behind the portfolio-selection work? What does it mean? For whom is it useful and what are its limitations? These are the fundamental questions we must answer in order to put this work into perspective.

Markowitz assumes that an analysis of a group of stocks has been performed. Furthermore, the portfolio manager not only can form a picture of the individual riskiness of any stock in particular, he also can calculate the correlation among the returns from given stocks. For example, industries as a whole may encounter good or bad times; thus there may be a more direct relationship between the behavior of the stock of General Motors and Ford than, say, General Motors and AT&T.

A problem which faces a mutual-fund manager or any other individual who has to run many millions of dollars is that in his selection of a portfolio he may have to consider trading expected returns for safety. If there were the choice of portfolios A and B, both of which had the same expected returns but one of which, say B, had greater uncertainty, clearly A would be superior. The whole of the Markowitz analysis is aimed at finding efficient portfolios, or portfolios such that any improvement in expected returns could be obtained only at the cost of increasing risk. Suppose we had three portfolios, A with 7 percent expected return and 10 percent risk, B with 6 percent expected return and 10 percent risk, and C with 8 percent expected return and 11 percent risk. Portfolio B would be called inefficient because A offers more with no added risk. C, however, has a bigger risk than A, but a higher expectation. Thus, the choice between A and C is a matter for the portfolio manager to decide—trading risk versus expected return.

The basic approach of the Markowitz method is to select a list of stocks, measure the historical average return from each over some selected time range, determine a measure of the variability in the returns from each stock by calculating a statistic known as the standard deviation, and then calculate the level of correlation in the stock movements—that is, the degree that various stocks appear to move together in a market. These calculations serve as the basic data from which to examine the expected returns and risk associated with a portfolio consisting of any mixture of holdings of the basic stocks.

Markowitz stresses that a portfolio analysis begins where security analyses leave off. The security analyst need not use historical data to judge the expected return and variability of a stock; he may have many other methods to do so. Furthermore, consideration must be given to taxes, assets other than stocks, trading costs and many other detailed variables that a realistic model must take into account.

We stress security analysis of the most fundamental sort, corporate finance and self-analysis. We emphasize how to evaluate individual stocks, and how the merits of these stocks fit in with individual goals and financial plans. We believe that in the case of the individual investor who is reasonably sophisticated, problems occur far more with security analysis than with portfolio selection. All of the computer runs in the world will not help the investor if he is selecting from a menu of a dozen poorly analyzed issues. If, on the other hand, by following a financial-integrity approach and seeing angles that others have not seen he has four or five potentially superior stocks, he scarcely needs portfolio analysis to calculate what to do.

If you are managing the Dreyfus Fund and the daily pressures of several hundred million are on you, our advice is to obtain computer runs which suggest changes in the composition of your portfolio. You may not like what they suggest and will probably end up by doing what you intended to do in the first place. Or you may modify both your portfolio and the one suggested by the computer. Nevertheless, as the runs are now relatively cheap and the basic idea behind the portfolio-selection method is sound, it is a useful exercise when the sums and the need for diversification are large enough.

FUNDAMENTAL SECURITY ANALYSIS AND CORPORATE FINANCE

In spite of the frequent assumptions of many economists about the perfection and efficiency of markets, it has been our observation that where control of businesses is not involved, there are fewer individuals capable of analyzing equity securities in depth than there are opportunities to analyze.

The fashion in thought among many of those who teach the more mathematical brands of financial analysis is indicated in this quote on fundamental security analysis.

There are thousands of professional fundamental security analysts at work in the United States . . . As a result of the efforts of this army of professional fundamental analysts, the price of any publicly listed and traded security represents the best estimate available at that moment of the intrinsic value of that security. In fact, the fundamental analysts do such a good job, there is no reason for anyone who is not a full time professional to bother with fundamental analysis.³¹

As far as we can tell, the above statement could not be more wrong. Good fundamental security analysis involves perception, training, understanding and a high degree of abstraction in implicit or explicit model building—that is, in picking the right variables and causal relations. There are not too many skilled practitioners, especially among that vast army more interested in predicting stock market actions than in following fundamental approaches.

An interesting book on the simulation of trust investments was devoted to building a computer simulation of the behavior of a bank trust officer in selecting portfolios for customers.³² The book showed how this could be done with considerable success. It has an important lesson to teach us: the legal and institutional constraints on the average trust officer are so large that it does not require an extremely sophisticated program to be able to perform roughly as well as the officer.

In our opinion, it will be a long, long time before anyone builds a computer program of a first-class security analyst that can perform in anywhere near the manner he performs.

CALCULATION OR EVALUATION

There is little doubt that a number of talented analysts have prospered using complicated computer calculations, especially in areas such as options and commodity trading. There will be more complicated methods that will replace them or improve upon them. But these individuals, like other superior analysts or deal makers, earn their livings by being professionals and by being able to judge the limits as well as the power of their type of analyses. Markowitz and Thorpe may well have been able to use their skills to do better than average; but it seems as though Ben Graham and some of his noncomputer colleagues have also prospered.

Much of the work of economic theorists has been based on such assumptions as the existence of perfect capital markets, firms run by managements with the single purpose of looking after their stockholders, a world with perfect accounting and clearly understood information. In such a world, the advanced economic theorist is willing and able to make allowances for uncertainty in the form of probability distributions over various events. But in the world that we live in, statistical uncertainty is one of our least worries. Of course we want to be able to get figures and statistics. The problem that faces any serious analyst, however, is more what the figures mean, and not so much what the figures happen to be. It may not be significant to know only that Company X’s real estate holdings are carried on its books at $1.5 million or $1.3 million; but those numbers, coupled with the information that they represent 100,000 acres of California coastal land carried at the 1880 purchase price, are significant.

We believe that economic thinking is invaluable to good financial analysis. A full understanding that investment in one project implies investment forgone in another is a lesson that many individuals find hard to learn. Yet, at the same time as we endorse economic thinking, we feel that the relevance of such economic theory to public or personal economic problems is minimal, not because we are opposed to abstract thinking, but because we are opposed to poor models of economic reality, a misemphasis on the controlling variables in our economy and a blind belief that by leaving out the institutional facts of life, somehow a mathematical financial analyst is going to produce a great general abstract theory of value.

A textbook on portfolio analysis begins with the following statement:

Changes are occurring rapidly in the teaching of investments. Investigation of the legal intricacies of the various securities, the tax status of different sources of income, how a stock exchange operates, the needs of the various investing institutions, and other descriptive and institutional matters are giving way to deeper analysis. The newer courses treat problems on a more abstract and general level.³³

We believe that such a statement really means that some of the business schools in the United States have faculty who prefer the comfort of teaching mathematical techniques in the hope that it will better train students to learn about the real world later, rather than teaching them about the real world now. Perhaps they are right. We disagree. It is not that we think that abstractions are unimportant. It is just that, we think, the more that people who are engaged in investments understood about the real world, the better off they will be.

One further example should help to illustrate our point. In the perfectly liquid, friction-free, tax-free world of much of microeconomic theory, the statement that on December 18, 1974, Company X had a net worth of $475,000 has a specific meaning. In the world that we live in, virtually any individual who by any measure is worth around half a million is not in a position to calculate his wealth at any point of time with much accuracy unless the purpose for evaluation and conditions for liquidation are all specified. It will be one value for estate planning, another for income taxes and yet another in obtaining a loan.

Economic thinking is critically important in interpreting and understanding tendencies in economic systems. But tendencies should not be confused with actualities. It is likely that there is some tendency toward efficiency, even for equity markets. Among other things, rules and regulations of the Securities and Exchange Commission have resulted in increased informational efficiency of U.S. securities markets. However, when we look at the history of the New York Stock Exchange over the last sixty years, it can scarcely be described as a mad rush toward economic efficiency, even though there probably has been some tendency in that direction.