REASON this month inaugurates a new feature, which will appear from time to time: a column of commentary and advice on various aspects of investment. Alastair MacLeary is one of the new breed of young, mathematically oriented investment analysts. His current interests center on the use of computer analysis on commodity trading. Mr. MacLeary commences his column with a look at the gullibility of some libertarians regarding silver investments.
Fundamental analysis of expected supply/demand relationships is a useful tool for predicting forwardization of prices of a commodity only when at least the following conditions hold:
1. Speculation in the commodity is closely tied to the economic interests producing and utilizing the commodity. That is, the futures market in the commodity has as its primary function provision of liquidity for the economic interests. Speculative interest in the silver futures markets, for example, enables producers of silver to ensure that their product will be sold when produced in the future at a price they can determine today and the users of silver, say for the semiconductor industry, to ensure that they will be able to obtain silver when needed.
2. Factors affecting the supply and demand of the commodity are relatively predictable. If the users of a commodity are not tied in to the defense industry and consumer demand is relatively constant or slowly increasing (as with wheat), reasonable predictions can be made on the demand for the coming year. If the commodity is some rare metal or, like copper, a metal used high in the structure of production, major perturbations, such as defense cutbacks or a recession will cause wild decreases in the demand for the commodity. Government use in activities influenced by controversial political decisions, such as minting of silver coins or the amount of the defense cutback or devaluation will cause the demand to be unpredictable.
3. Major uses of the commodity do not have a propensity to be tied to the latest technical developments. A new alloy, for example, may make silver entirely obsolete in electrical conduction. On the other hand, it may greatly increase the electrical demand for silver. A fad may greatly increase or reduce the amount of silver demanded for jewelry.
SILVER—UP AND DOWN
Unfortunately, the commodity in which libertarians speculate most is that very same commodity which least fulfills these three conditions: silver. Additionally, as I have discussed in my articles for COMMODITY JOURNAL, JOURNAL OF COMMODITY TRADING, and SILVER MARKET LETTER, the method of price prediction used by fiscal conservatives and most libertarians is of the most naive fundamental type. Depending on only one fundamental prediction for a change in the demand—a devaluation, premier political maneuver—libertarians have invested in waves since 1967.
That this prediction was false is now known. What is less known is that even were the prediction correct, the chances are only very small that the price, after initial shock (remember, after the government's announcement of the two-tier gold market, the price of gold rapidly increased for a couple weeks then decreased below previous levels), would have increased beyond the high price of over $2.50 per troy ounce of June 1968.
Currently the price is about $ 1.65 per troy ounce; after three years of inflation, the price in 1968 dollars is perhaps $1.45 per troy ounce. Even the current price represents speculative interests far exceeding any economic requirements (once on Commodity Exchange in New York, three times the United States' annual production of silver, or one-half the world's annual production of silver, changed hands in one day's trading. And the Commodity Exchange, albeit the largest, is only one of four organized silver futures markets in the United States. The others are the Chicago Board of Trade, the West Coast Commodity Exchange, and the American Board of Trade).
The price of $2.50/troy ounce more than compensated for the increase in economic value were a devaluation to occur. Probably a devaluation, after the initial shock and temporarily increasing prices, would have been followed with a price decrease in silver, only one that was smaller than the actual decline that has occurred since June 1968, the difference in the declines approximating the amount of the devaluation. An explanation of the artificially high prices for silver since 1967 is offered below.
TWO CLASSES TO CONSIDER
If one is to speculate in commodity futures, or spot commodities, he must distinguish two broad classes traded in organized commodity exchanges. The first class is those for which speculation is at least not grossly distorted from its economic function of providing liquidity and predictability for producers and users of the commodity; the second class is those in which much of the speculation serves other than economic functions—a sophisticated form of gambling, political expression (silver!), curiosity regarding the futures markets (Maine potatoes) or some fad (pork bellies were as now live beef cattle is, the fad commodity, while everyone was investing in IBM stock several years ago).
Included in the first group are commodities such as wheat, cotton, cocoa, sugar, live hogs, oats, corn, soybeans, soybean oil, soybean meal, coffee, and copper; best representing the second group are silver, Maine potatoes, pork bellies, live beef cattle, frozen orange juice, and lumber.
The most rigorous attempts to predict forwardization of prices of a commodity of the first class, based on rigorous mathematical interpretations of international commodity agreements between governments, black market leaks, supply of stock in storage, expected demand, and expected production, has been performed at the University of California at Berkeley in the cocoa market and by the Food Research Institute Studies at Stanford.
Based substantially on the theory of storage developed by Brennon and others associated with the Chicago Board of Trade, it was a regression equation giving the future price as a function of several dependent variables. The constants in the equation were determined by the least-squares method. The least-squares method finds those constants for which the sum of squares of the differences between the analytic predictions and the historical prices actually occurring is least. Yet the correlation between the actual and analytic prices was only 0.66 (for the cocoa market) which if applied to speculation would prove disastrous. The cocoa market was used at UC-Berkeley as it is a far more stable commodity than most. Markets such as silver would have analytic predictions so far diverging from the actually occurring prices that the equations would be ludicrous as anything but an exercise in a problem set for a statistics class.
When speculators enter the silver market, they make implicit regression analysis, i.e., they predict a future price based on primitive, one might say intuitive (libertarian conventional wisdom!), expected supply/demand ratios. We have seen that even rigorous, mathematical regression analysis does not work, let alone the intuitive, conventional wisdom variety.
Why? Because the supply/demand ratio is inadequate to ascertain the actions of speculators who are using the market to express something besides their expected supply/demand ratios. A number of economic factors enter the market as parameters via the speculators, factors which determine not only the number of marginal speculators, but the size of the marginal speculation. These conditions include, among many others, monetary conditions and monetary policy, the degree and intensity of relative inflation in other nations, actual or anticipated government intervention in the marketplace, wars, labor unrest, changing technology, foreign competition, substitute goods, and the kinds of people who are helped or hurt by government redistribution of resources.
The whim factor was highly relevant to the debacle in the recent silver market: It became popular among fiscal conservatives and libertarians who were displeased with government monetary insanity, especially during the silver certificate phase-out, to enter the silver market as a means of expressing their dissatisfaction and, in many cases, fright. The rightist financial advisors leaped at the opportunity to make substantial monetary profits for themselves by influencing the actions of libertarians and conservatives who wanted a behavioral correlate to their political outrage. They sold books by the tens of thousands and issued advisory letters by the hundreds of thousands, all suggesting that monetary profits were to be had simultaneously with expressing political outrage by investing in the silver market. By and large, since the 1968 high price, over $500,000,000 has been lost by silver speculators. Undoubtedly a large portion has been lost by fiscal conservatives and libertarians. Only the advisory services and the would-be financial messiahs have profited in the silver market in recent years.
The silver market thus provided an emotional outlet for a large number of people; for them, the market had a noneconomic as well as an economic function. Speculators were willing to pay almost any price for the psychological satisfaction of expressing their hostility to the government. Consequently, the price, from the economic point of view, was overdiscounted (especially in view of the fact that the Fed was drying up the money supply and industrial production was entering a sustained slump). In order to accommodate the influx of buy orders of those entering the market for psychological reasons, the price of silver increased far beyond what any purely economic analysis would have permitted. Quite literally, a fad developed; it was tribally de rigueur to be in the silver market if you opposed government fiscal and monetary policy. Buying waves developed. Those who bought on the first wave and saw their investment increased in value pointed out to subsequent waves of buyers the "correctness" of speculating in silver.
Eventually, however, the market had absorbed that sub-set of individuals who were buying at any price for psychological and political reasons—and suddenly there were no more people to expand the balloon further. The last wave of buyers found out quickly that they had bought at the top. In the meantime, the predicted and expected "imminent devaluation" failed to materialize. The fad quickly abated and those who had rushed into silver speculation found themselves heavily invested in an overdiscounted market. The market reacted as markets do by reverting to a more economic function, that is, the price fell as those in for political reasons exited. Those who bought at a very high price for political reasons paid for their emotional satisfaction by substantial economic loss.
When speculating in a commodity whose futures market serves functions which dwarf the economic function of increasing the rational allocation of that commodity in the structure of production, the speculation must be regulated by predicting not the relationship of the stock of the commodity to its use but by predicting the reactions of the type of speculators that use the commodity futures for extra-economic activity.
Predicting the behavior of speculators in a futures market is best accomplished not by syllogisms beginning with the axioms of a psychology system but by statistical tests on what has actually occurred in the past trading of that commodity in the marketplace.
It is necessary to create operational definitions of one's claims to understanding the behavior of people in the marketplace. In order to make money, one must have at least the following:
1. Quantitatively defined stimuli to which the speculators react. One cannot say, "After a sharp rise in a short period of time, speculators will profit-take and a small decline will fill the gaps." Or, rather, one may say this, and many do, but they don't have any money. What is a sharp rise? Or a short period of time? We must be specific. "If the silver prices in the May contract increase at least such and such in at most so many days, then when those conditions have occurred in the data representing the trading of the May contract since 1964.…"
2. The stimuli must be detectible. If we run time series analysis on the relationship between stated unemployment rates, published money in circulation and the price of silver futures during the month the figures describe, our psychology is irrelevant, because the stimuli is not detectible until the response has occurred. We might, however, run correlations of the monetary conditions one month and the price behavior of silver futures the next month, remembering that while we may find positive statistical correlations, we should not commit the post hoc ergo propter hoc fallacy.
3. The predicted response must be operationally defined. A statement such as, "The market did x, so the speculators will get scared," is meaningfree, except to those who wish to lose money. The response must be operationally stated. Given parameter relationship x, since 1964 the average price movement was y in n days, along with a number of other statistically interesting numbers.
That the process of which I am speaking is difficult goes without saying. Most people, including libertarians, like to pretend they know market behavior that they don't know at all. They create a metaphysics of the market, which like all metaphysics is rife with meaningfree, symbolic assertions. They pretend to know the market by giving their ignorance labels: "accumulation," "distribution," "short-term reversal," "market adjustment," "unexpected pressure on the downside." The terms mean nothing. People lose money on them every day.
In silver, we can obtain at least four prices describing each day's trading. The WALL STREET JOURNAL lists the previous day's trading, and daily prices are available from all brokerages that trade commodity futures. An entry in the WALL STREET JOURNAL may appear as follows: Silver, New York, May '71: 161.90 162.30 161.40 162. 10-.60. The prices are in cents per troy ounce.
These four prices represent the actual behavior of speculators on one day. The first figure is the opening price; the second the highest price at which silver was traded that day; the third the lowest price at which silver was traded that day; the last is the closing price for that day's trading session. The fifth number represents the change in one day's closing price from the previous trading session's closing price.
It is with the accumulation of sequential daily price listings that we design our stimuli, and it is with subsequent listings that we measure our responses.
To test responses to various market stimuli requires a data bank sufficiently large to permit valid statistical predictions. I would suggest the price history of at least two delivery months as far back as 1966, when the silver market became modern—and on a large scale.
It makes sense that a stimulus likely to elicit a response of rapid price movement is one which consists of recent volatility in the price behavior of the commodity. This is experimentally verified. Thus we may test a number of levels of volatility of the prices and measure the market response when such a parameter set occurs. Two tests of price volatility immediately come to mind. The first is measuring how violent intraday trading is, and the second is measuring interday violence. A simple Fortran program can be written to generate these parameters routinely.
Mere printouts of the range and change for each day are only a trivial first step in designing the stimuli, let alone measuring a germane response and performing valid statistical tests on their reliability. We must operationally define high volatility and high price movement and, finding when these parameter sets occur, test subsequent price movements.
Without listing the specific Fortran expressions, let me note that what is required is a measurement of the difference in price levels over a specified time period, a parameter varying that specified time period, and a parameter testing the number of days we are to search for subsequent price movements.
OUTLAY BEFORE INCOME
As an indication of the ultimate complexity of such a search, my current programs testing such patterns and subsequent behavior require about 20 minutes to cover one 250 day contract price history, and this is on a CDC 6400, which is from 2 to 5 times faster than most moderate sized models of IBM 360s. Since computer time costs at least $80 an hour on a 360/40, a check of ten silver contracts would require 80 x 1/3 x 4 x 10 = $1080. And this is only the beginning. At least double this must be spent to reach the level of sophistication to titrate all the possible parameter relationships to what I have written now; and, further, my current programs are only the intermediate stage in my searches for statistical predictions.
The point of this description of my work is to indicate to libertarians that profit production in the silver market is as difficult as profit production in any other form of economic activity: Information costs are considerable and the technology is complex. There is no more reason for a libertarian to expect to make money in silver speculation by referring to the incantation, "A devaluation is bound to occur," or the alternative incantation, "As people lose faith in paper, they will prefer silver and gold," than there is for him to expect to make money in, say, the electronics protection industry by calling forth the magic powers of the incantation, "As the police become more politicized and protect less, people will buy electronic protection devices as an alternative."
Do not be seduced by the siren songs of the high priesthood of the silver cult. Of all the advisory services and commodity letters including silver in discussion, those specifically rightist or libertarian are among the worst, for they know how to appeal to the noneconomic motives of their customers and they do so. Hayden-Stone, Dunn & Hargitt, Commodity Research Bureau, and the Silver Market Letter are four reasonably good predictors of the silver market price movements. Read them and compare their operationally precise statements to the obscure and ambiguous statements of the movement-oriented advisory services. I feel the subject of silver trading is a relevant topic for discussion in a theoretical magazine for several reasons Libertarians have less money for marginal activities, including the support of REASON magazine, because they speculated in silver in large numbers and large amounts due to inadequate analysis (and incredibly easy seduction) by the medicine men of the financial world that belies their own advocacy of reason. These libertarians have confused names with objects, incantations with scientific knowledge. When next impelled to speculate in silver, pretend it is pork bellies, and ask yourself, "If the competence I have in predicting the price behavior of the silver market were the competence I have in predicting the price behavior of the pork bellies market, would I speculate?" If the answer is no, don't.
This article was written several months ago. Several additional comments are now in order. My conversations with commodity analysts for two large brokerages recently bear out the misfortune of the dear cost of confusing ideology and economic prediction. First, silver coins. Two commodity exchanges are trading silver coins, the respective face values being very high. The price of the bags has been inflated far beyond the face value and has increased remarkably since the initiation of trading. The analysts indicate that feedback from their brokers shows that the vast majority of trading is being done by politically-oriented, e.g., libertarian and fiscal conservative types, speculators who plan to accept delivery. The coins were originally put up and short sales are initiated still by the very dealers who have always advertised to the ideological speculators. Apparently almost no dealers are speculating long. They are unloading the coins at prices much, much higher than they were charging for the coins through their advertisements just a few months ago. The speculators are bidding not for trading but for delivery. In about a year the market will fade, for the coins will have been delivered at prices the dealers previous to the futures market could never have hoped for. The prices will then decline, and another debacle will have occurred, albeit on a much smaller scale than the $500,000,000 debacle in silver bar trading. By Crom, this disturbs me.
Secondly, we have the continuing story of silver futures. The prices this year have been fluctuating in a $.30 per ounce range. Here, too, the ideologues have been getting screwed or, I should say, have been screwing themselves. My analyst friends again say the politically oriented speculators are taking the losses. As soon as the market makes a jump, all the "advisory" services catering to the right-wing say, "See, the apocalypse is come, buy silver to save yourself from the imminent collapse," or a more mild variant. They buy, the price goes up a bit more; more buy, price reaches a plateau; the prices drop to the bottom of the range, the libertarians-conservatives get scared and sell. This has happened four or five times this year. The losses are not trivial. They run to millions of dollars.