Correlation is not causation

The Accelerator and Say's Law

by William H. Peterson

Economists, like women, are not immune to the dictates of fashion. One such dictate in vogue among post-Keynesians is the accelerator, which enjoyed similar popularity in the early 1920s. At least a partial reason for the renewed popularity of the accelerator is that it forms an integral part of the General Theory.[1]

The acceleration doctrine holds that a temporary increase in consumer demand sets in motion an accelerated "derived demand" for capital goods. This action, according to adherents of the doctrine, explains at least part of the causation of the business cycle. As evidence supporting this theory, accelerationists point to boom-and-bust, feast-and-famine conditions prevalent in capital-goods industries.

A typical illustration of the acceleration principle follows. Assume a "normal" annual demand for a certain consumer good at 500,000 units. Production is accomplished through 1,000 durable units of capital goods; capacity of each capital unit: 500 consumer units per year; life of each unit: 10 years. Then assume a 10 percent increase in consumer demand. Thus:

Annual Consumer	Demand	Capital Goods	Annual Capital-Goods Demand ("derived")
"normal year"	500,000	1000	100 (replacements)
next yr. + 10%	550,000	1100	200 (replacements plus new)
3rd yr.-new "nor."	550,000	1100	100 (replacements)

Conclusion: 10 percent increase in consumer demand led to 100 percent increase in capital demand in same year but to 50 percent decrease in capital demand in following year.

The argument against the acceleration doctrine simply shows so many unreal assumptions and a vital non sequitur as to nullify any validity in the doctrine whatsoever. An analysis of these objections follows.

1. Rigid Specialization in Capital-Goods Industries

Accelerationists pose their doctrine on the basis of a given capital-goods industry supplying equipment for a given consumer-goods industry and no other. Thus a decrease in consumer demand or even a falling-off in its rate of growth immediately cuts off part of the capital-goods market, and the "famine" phase of the capital-goods industry begins.

Yet where is the capital-goods industry so rigidly specialized as to preclude its serving other markets, with or without some conversion of its facilities? Are we to presume that businessmen under the pressure of overhead and profit maximization will twiddle their thumbs waiting for their consumer demand to "reaccelerate"? It is clear that accelerationists deny or ignore convertibility of facilities and substitutability of markets.

Within many capital-goods industries, trends of diversification and complementarity are evident. Examples: A machine tool manufacturer that has undertaken lines of construction and textile equipment; a basic chemical producer that has engaged in the manufacture of home clotheswasher and dishwasher detergents. These trends break down the "industry" classifications, on which the accelerator is based.

2. No Unutilized Capacity in the Consumer-Goods Industry

Holders of the acceleration doctrine assume the consumer-goods industry is operating at the extensive margin of production and no intensive possibilities for greater production exist.

But very few consumer-goods industries, typically, operate at constant peak capacity. To do so is generally to operate beyond the point of optimum efficiency as well as beyond the point of maximum profit. The usual case then, other than during wartime, is that an industry operates with some unutilized capacity, some "slack." Normally this unutilized capacity is to be found among the marginal and submarginal producers, and it is these producers which could and probably would absorb any increase in consumer demand — without, of course, the purchase of new equipment.

Yet even the successful and efficient producer would likely consider other means of absorbing higher consumer demand before committing himself to more equipment and greater overhead. For example, he could expand the existing labor force, resort to overtime, add one or two additional shifts, subcontract work in overloaded departments, and so on. That such alternatives are feasible without more equipment is evidenced by the experience of even the most efficient firms in the utilization of their capital equipment. Examples: A West Coast airplane manufacturer found his gear-cutting equipment in use only 16 percent of the time; a New York newspaper plant utilized its presses only 11 percent of the time. The concept of 100 percent utilization of all capital equipment is not tenable.

3. Automaton Role for Entrepreneurs

Accelerationists share the danger common to all holistic and macro approaches to economic problems — namely, the submergence of individual and entrepreneurial decisions (human action) to a constant factor within a pat formula. Such treatment implies on the part of entrepreneurs irrationality or sheer impulsiveness. Boulding described this situation thusly:

The picture of the firm on which much of our analysis is built is crude in the extreme, and in spite of recent refinements there remains a vast gap between the elegant curves of the economist and the daily problems of a flesh-and-blood executive.[2]

Accelerationists argue that a temporary rise in consumer demand automatically calls into being additional capital goods. If this were true, it follows that entrepreneurs in capital-goods industries witlessly expand their capacity and thereby commit themselves to greater overhead without regard to future capital-goods demand.

True, entrepreneurs can and do err in gauging future demand. But the concept of automatic response to any rise in demand, on the order of the conditioned-reflex salivation of Pavlov's dogs, is not warranted. Increased capacity is less of a calculated risk in response to increased current demand than it is to anticipated future demand. This anticipation, in turn, is likely to be based on market research, price comparison, population studies, cost analysis, political stability, etc., rather than on impulse.

4. Static Technology

It is not surprising that the accelerator perhaps reached the zenith of its popularity when professional journals were replete with terms like "secular stagnation" and "technological frontier." (Nowadays the term is "automation." Apparently we have moved from the one extreme of too little technology to the opposite extreme of too much.) Such heavy-handed treatment of technology does not coincide with experience. Science and invention do not hibernate during depressions. Du Pont introduced both Nylon and Cellophane during the 1930s.

Adherents of the acceleration principle must either minimize or ignore the impact of technology on rising productivity, for, after all, a strict ratio of capital-goods to consumer-goods output must be maintained to substantiate the action of the accelerator. Technology, however, can and does obviate such ratios. Technological advances not only serve to increase the unit volume of given capital goods through superior technical design but also through the improvement of fuel, the refinement of raw materials, the use of time-and-motion studies, the rearrangement of layout and production flow, and so on.

While the growth of technology is somewhat irregular, there can be no question of its progression. Progression tends to "accelerate" the obsolescence component of depreciation and thereby crimps the acceleration model, which, ceteris paribus, ignores the unpredictable dynamics of technology.[3]

5. Arbitrary Time Periods

Accelerationists must use time as a frame of reference for their doctrine. The most frequent time period used is a year. Such a time period, however, implies an even spread of the increase (or the decrease) of consumer demand in the time period. Thus a spasmodic strengthening and weakening of demand within the time period could distort the artificial taxonomies of the accelerator.

For example, a January–December period may carry one peak demand, whereas a July–June period may yield two peak demands. An accelerationist may read the first period as having an 8 percent increase and the second as having a 10 percent increase, which, in the long run, may average out to 9 percent or some other figure.

Moreover, within a time period, the accelerationist assumes a fixed relationship between consumer goods and capital goods. Let alone the problem of technological advances, were such a fixed relationship to exist it would necessarily mean that the cycles of production for both sets of goods were perfectly synchronized. This, however, is rarely the case. Consumer goods generally have a short cycle; capital goods, a long cycle. Thus, current capital-goods production may be based on orders originating in an earlier "period." Two consecutive increases in consumer demand could conceivably be followed by a decrease, which may well mean that the latest order for capital goods would be cancelled. The flow of goods from the capital pipeline is not irrevocable.

6. Implicit Denial of Say's Law

Previous objections to the acceleration doctrine were of the "other-things-are-not-equal" variety. In short, with so many independent variables ceteris paribus would not hold.

This objection — the implicit denial of Say's law of markets — is more fundamental. If it is valid, it would strike at the heart of the acceleration principle and reduce it to a non sequitur.

According to Say's law, the source of purchasing power lies within production — i.e., supply creates its own demand — and therefore generalized overproduction or underconsumption is not possible. Barring external distortions to the economy, such as war or drought, Say's law is operative under two conditions — the flexibility of prices and the neutrality of money. Thus it is not astonishing that a major accelerationist like Keynes who shunned price flexibility and upheld inflation should attempt a refutation of Say's law and resurrect the dead body of underconsumption, rebaptized as the "consumption function" or "the propensity to consume."

If it is true, as accelerationists claim, that a rise in consumer demand will thereby create a demand for capital goods, then it must be explained what causes the rise in consumer demand in the first place. Should accelerationists concede that the rise is due to capital — or as Böhm-Bawerk put it, "the technical superiority of roundabout production" — they would then be forced to admit, logically, that they have put the cart before the horse, that the growth of capital preceded the growth of demand.

Indeed, if demand could arise without prior production to give it effectiveness, then we should witness the overnight industrialization of India, where such astronomical "consumer demand" exists as to induce the full flowering of the accelerator.

Say's law not only points to the fallacy of the accelerator but to its corollary, "derived demand." There is a germ of truth in "derived demand" — "primary" consumer demand does affect "secondary" capital demand. But the consecutive sequence should be reversed. The effect of consumer demand on capital is not demand for capital per se. Capital is always in demand as long as time preference exists — as long as capital yields the reward of interest. Rather, the effect of "derived demand" will be, if strong enough, merely to change the form of capital goods, no more. If not otherwise impeded, capital will always flow to the most urgent of the least satisfied demands. The point is that capital accumulation — saving and investment — must comebefore "derived demand." So-called derived demand merely shifts already existing productive resources from present applications to alternative but more rewarding applications.

Insofar as the acceleration explanation of the business cycle is concerned, accelerationists view deceleration with equal alarm to acceleration. The dilemma was stated by Samuelson:

It is easy to see that in the acceleration principle we have a powerful factor for economic instability. We have all heard of situations where people have to keep running in order to stand still. In the economic world, matters may be worse still: the system may have to be kept running at an ever faster pace just in order to stand still.[4]

To maintain such an argument, Samuelson and other accelerationists must discount the fact that a cut in consumer demand in one line releases consumer demand for other lines. Thus, the change in the composition of consumer demand releases factors engaged in certain suspended lines of capital-goods production for new lines of endeavor. That this would cause frictional unemployment of factors is not denied, but frictional unemployment is far less of a problem than generalized unemployment. The notion of ever-accelerating consumer demand to achieve stability within its related capital-goods industry thus loses sight of the interchangeability of factors.

The essence of capitalism, as in life, is change. While some industries may be in decline, others will be in ascendancy. Capital is not eternally fixed; it can be liquidated and "recirculated." Nor does capital idly wait for consumer demand to "reaccelerate." Disinvestment and reinvestment, business mortality and business birth, industry expansion and industry contraction constantly adjust the supply and form of capital to the demand for consumer goods. Samuelson overlooks the dynamics of capital in his essentially static, timeless acceleration thesis.

Say's law places production as the controlling factor over consumption. The accelerator reverses this order. Thus accelerationist Keynes sought to accelerate consumer demand by having the unemployed uselessly dig holes or build pyramids, the important thing being to put "purchasing power" in the hands of spenders. Productionless "purchasing power," according to Say's law, is a contradiction in terms; it is nothing but inflation. In short, the false premise of "derived demand" in the acceleration principle has led to other false premises.

Conclusions

Four findings spring from this article. One, the accelerator is groundless as a tool of economic analysis. Two, Say's law has yet to meet an effective refutation. Three, acceptance of the acceleration doctrine leads to false conclusions in other areas of economics. And four, accelerationists must look elsewhere for an answer to the business cycle.

While there is evidence that capital-goods industries do suffer wide extremes of business activity during the course of the business cycle, it is also true that consumer-goods industries undergo much the same cycle, even if their amplitudes are smaller. That there is correlation between the two phenomena is not denied. But correlation is not causation. This is the heart of the error in the accelerator.

Notes

[1] See, e.g., J.M. Clark, "Business Acceleration and the Law of Demand," JPE, March 1917, pp. 217–235; T.N. Carver, Principles of National Economy, 1921, pp. 436–440; J.M. Keynes, General Theory, 1935; and R.F. Harrod, Trade Cycle, 1936.

[2] K.E. Boulding, "The Theory of the Firm in the Last Ten Years," AER, December 1942, p. 801.

[3] Cf. J.R. Hicks, The Theory of Wages, 1932, pp. 112–135. Even though it is incidental to his distribution theory, Hicks formulates a theory of invention which could profit the accelerationists.

[4] P.A. Samuelson, Economics, 2nd ed., 1951, p. 391.