Former Treasury Secretary Larry Summers described the Efficient Markets Hypothesis, first propounded in 1961, as “the most remarkable error in the history of economic theory” in the Wall Street Journal after the 1987 stock market crash.
Yet Nobel Prizes were given for the Hypothesis (Modigliani, 1985, Miller, 1990) or for work relying entirely on the Hypothesis for its theoretical foundation (Scholes and Merton, 1997).
Even more significant, Wall Street traders have continued to carve substantial fortunes out of trading strategies using the Hypothesis, even though every now and then spectacular crashes of such strategies are recorded. What’s going on here?
The efficient market hypothesis, in its original “strong” form, states that securities markets react immediately to all information, whether public or private. This implies that it is impossible for superior analysis to deliver superior market returns, a formulation that immediately made the hypothesis attractive to underpaid professors contemplating Wall Street analysts’ salaries.
However, on inspection, the “strong” form proved to be undoubtedly false, so two further forms were invented, the “semi-strong” form, which says that the market reacts immediately only to public information, and the “weak” form, which says that you can’t tell anything about a stock’s future movements from examining its behavior in the past.
From the efficient market hypothesis, and its corollary that markets moved in a random walk, Fisher Black and Myron T. Scholes devised in 1973 the Black-Scholes options valuation model. This model was derived by examining how an option on a security might be hedged by an opposite trade of a fractional amount of the security itself.
It therefore relied not only on the hypothesis, but also on volatility (the average amount by which the market moves, day to day), remaining constant, and on security prices themselves moving continuously, with no jumps.
The Efficient Market Hypothesis, and the Black-Scholes model, then combined with the advent of cheap computing capacity to cause an explosion of new derivatives markets, in which securities were derived, based on price movements in other securities.
Interest rate and currency swaps, futures, options and stripped bonds all originated or were greatly developed in an orgy of financial innovation in the early 1980s.
Wall Street’s profitability became increasingly driven by the trading desks, particularly those who took positions for the institution itself rather than for clients, and a range of quantitative analysts were hired, who could produce mathematical models of great plausibility, and thereby convince top management in both the Wall Street houses and their clients that particular investments were safe and potentially profitable.
These theories received their first real test in the 1987 meltdown. Portfolio Insurance had been invented by Wall Street quantitative analysts to allow investors to insure their stock holdings by triggering automatic sell programs, selling an amount calculated by reference to the precise stock price movement, if stock prices declined.
In the years to 1987, this strategy proved quite popular, with many billions of dollars in portfolio insured funds by the time the crash came.
In practice, of course, portfolio insurance proved no protection at all. Prices dropped discontinuously; consequently portfolio insurers were unable to sell quickly enough. In turn, the frantic, often computer-driven sell orders of the portfolio insurers unloaded waves of stock onto an already weakened market, exacerbating the decline. In the aftermath, market-closing triggers were introduced so that portfolio insurance could not even theoretically be foolproof, and the strategy fell out of fashion.
In the second disaster-epic episode, in 1994, a number of clients, such as Orange County, Calif., which had bought highly leveraged “exotic” interest rate derivatives products from Wall Street, discovered that the hedge they thought they had didn’t work, so they were exposed to rising interest rates.
Again, their mathematical models told them they were taking little risk, but in reality, their risk was much greater than had been predicted.
In the third, and so far final episode, in 1998, Long Term Capital Management, staffed by what were apparently Wall Street’s finest brains, plus the two winners of the 1997 economics Nobel, discovered that the “convergence” that the Efficient Market Hypothesis had predicted between different markets didn’t happen. Instead, the markets diverged, leaving LTCM with a loss that effectively extinguished its stockholders’ capital.
In these three unrelated disasters, therefore, EMH-related mathematical models, constructed by the best and the brightest on Wall Street, had dramatically failed to assess risk correctly, and had left their owners exposed to huge unexpected risks.
There would seem to be two basic reasons why EMH-related models don’t work. First, they rest on an oversimplifications of the market process, in particular perfect information and continuous pricing, that in practice are not true, and whose untruth makes reality diverge from the models’ predictions.
Second, they rest on an underlying assumption that prices move in a random manner, with constant volatility, and that price movements are independent of each other. However, in practice this is also untrue, price movements are highly correlated, both by the correlation between events themselves and by “market psychology” which when extreme affects all prices simultaneously.
Work I did — and published in “Derivatives Quarterly” three months before the LTCM debacle — demonstrates that, depending as they do on factors that are unknown but not random, financial market prices behave closer to a fuzzy logic than a probabilistic model, and in particular do not obey Bayes’ Theorem, that the probability of two events occurring is the product of their probabilities.
Of course, given this reality, a probabilistic model which assumes constant volatility is at times laughably inaccurate in predicting actual outcomes, usually in the direction of greatly underrating the risk of simultaneous adverse movement in several variables.
To a dispassionate observer, two questions must arise — why do they give Nobels for this stuff, and why is anybody still using these models?
The answers lie, I think, in the incentive structures for academics and Wall Street traders, who include substantial elements of “moral hazard.”
Academics, and academic institutions, naturally want to be relevant; this is particularly true of economists. The financial markets are of course a major part of the economic landscape, and furthermore, because of public trading of stocks and bonds, have especially high visibility. Hence, since these markets have existed, academic economists have been asked their view of values therein; no doubt Adam Smith was from time to time asked his view of the East India Company or the South Sea Company as investments.
How satisfying, therefore, by making a few of the plausible simplifying assumptions that economists make every day in other areas, to be able to construct a theory that asserts that, not only can academic economists not be expected to predict the stock market, but nor can anybody else.
If the theories are then picked up by Wall Street, as they have been, and huge amounts of money are devoted to implementing them, then such luminaries as Nobel Prize committees naturally assume that an important contribution has been made to the economics of financial markets. Thus the four Nobel prizes.
The motivations of Wall Street are, as would be expected, both more complex and more crass. For the quantitative analysts, building models for Wall Street offers mathematicians who are in practice probably not those likely to make substantial original discoveries the chance to earn serious money, at the cost only of quelling any doubts they might have about the assumptions underlying their models.
For many of them, this does not appear to have been too difficult.
As for the traders, there is a guilty secret about trading: Pure, risk free trading, with no insider information is just not that profitable. Thus traders, however good their reputations, are always looking for ways to juice up their returns. One way to do this, if their institutions let them, is to engage in a bet called a “Martingale” (nearly all successful traders appear to be exorbitantly addicted to games of chance, played for money), called after a strategy for winning (normally) at roulette, whereby a trader has a high chance of modest positive returns, and a small but finite chance of catastrophic loss.
Given poor enough controls, or sufficiently obfuscatory computer models, a trader can thus engineer his existence so he has say a 90 percent chance of a good bonus, and a 10 percent chance of bankrupting his employer. For the trader, this is fine; most years, he gets a good bonus, and if something goes wrong, he will either have moved on already, or will get fired, in which case he will have to live off past bonuses. For the institution, it is bad news; given enough traders, one of them is bound to involve you in catastrophic loss.
Badly run institutions thus have their Nick Leesons, traders who achieve a Martingale through simple fraud; well-run institutions have their proprietary investment groups. In the latter case, since positions are generally biased to work well when markets rise, in a period such as the 1990s, good year may follow upon good year with apparent regularity, thus enriching traders beyond their wildest dreams. In such cases, however, a good solid bear market, such as what we appear likely to experience now, generally causes the wheels to drop off the models. Expect some excitements ahead, therefore.
We may be drawing to the close of the era when Wall Street traders bought Ferrari Testarossas, built 30,000-square-foot houses, and lorded it over the rest of us. For many, the demise of their ill-gotten gains will be no loss. In five years, their vulgar lifestyle may be remembered like that of the late King Farouk, as symbolic of a past era of corrupt excess that is thankfully no more.
-0-
(The Bear’s Lair is a weekly column that is intended to appear each Monday, an appropriately gloomy day of the week. Its rationale is that the proportion of “sell” recommendations put out by Wall Street houses remains far below that of “buy” recommendations. Accordingly, investors have an excess of positive information and very little negative information. The column thus takes the ursine view of life and the market, in the hope that it may be usefully different from what investors see elsewhere.)
This article originally appeared on United Press International.