The Limits of Economics

Philosophy asks the important questions, but as soon as an answer is in sight a new discipline splits away and sets out on its own pursuit. This possibly explains why now—as two thousand years ago—philosophers are still preoccupied with unsolvable problems: those that can be solved are always snatched out of their hands.

Locke, Hume, Adam Smith, and even Karl Marx and John Stuart Mill were still philosophers interested in economic problems among other things. By the turn of the century economics established itself as a separate academic discipline. With the adoption of the mathematical approach it became as little understandable to the layman as chemistry or physics and thus began to be considered the most advanced of the social sciences.

More recently, however, this position of pre-eminence has been undermined by creeping doubts. When translated from mathematical language into English, the answers given by modern economics seem to miss or to beg the questions that were originally raised, and economists find themselves again in the company of philosophers with more important problems on their hands than they know how to solve.

Not that the profession is tormented by internal strife. A reader of leading economic journals would hardly notice that something is amiss. “Monetarists” contend with the “neo-Keynesian fiscalists” in the same tone that bimetalists used a hundred years ago in their controversies with the partisans of an unadulterated gold standard. So far as the great majority of professional economists is concerned, their discipline is now in as satisfactory a shape as it ever was.

Most of the fundamental questioning comes from outside the profession and is echoed by a few inside. Today, as in the past, more often than not, a critic of the prevailing style of economic thought is also a critic of the dominant social and political institutions. This turns the ensuing debate into something more (or less, depending on where one stands) than a purely academic controversy. If not kept carefully apart, value judgments become confused with factual assertions and both are all too often drowned out by arguments originating in and addressed to political emotions.

Robert Heilbroner’s writing is singularly free of such confusion. His skeptical appraisal of contemporary academic economics derives from a discriminating understanding of the capabilities and limitations of modern economic theory. His criticism of capitalist and socialist systems as they exist now in both hemispheres is developed within a broad historical perspective free from the myopic distortion so characteristic of much of the new radical thinking.

Whether he calls for new “political economics” or examines the transition “from capitalism to socialism” or explores “alternatives for the future,” Heilbroner’s argument carries us toward the fundamental dilemma of wishing versus understanding, acting versus explaining, of practical pursuit of human values as against theoretical analysis of unbreakable and endless chains of causes and effects. Any social scientist endeavoring to clarify for himself the significance of his own intellectual position is bound to face these questions. A political activist cannot avoid confronting them when he challenges the social scientist to provide useful knowledge.

The potential conflict between analytical explanation derived from factual knowledge on the one hand and purposeful social action based on deliberate conscious choice on the other is well exemplified by the peculiar role played in modern economics by the so-called “maximum principle.” The pursuit of maximum profits by the capitalist, highest possible wages by laborers, and largest possible rents by the landowners was recognized by the great classical economists, and Karl Marx must certainly be counted among them, as the dominant driving force of the industrial system that grew before their eyes. Together with the Malthusian theory of population and the Darwinian theory of natural selection—all of which shared the principle of the survival of the fittest—classical economics marked the successful extension of a deterministic, as contrasted to a pragmatic teleological, explanation not only of biological but of social phenomena as well.

According to classical economics and the neoclassical version of it that prevails today, the functioning of the economic system as a whole is explained by analyzing a blind mechanical interaction of households, businesses, trade unions, governmental organization, and so on. But the behavior of each one of these units of decision is interpreted as representing a predictable result of purposeful rational choice. Here is where the “maximum principle” (or the minimum principle, as the case may be) comes in.

If there are several different methods of generating electric power, which of them should actually be used if one’s aim is to minimize the cost per kilowatt? If one assumes that there is an ideal ordering of human needs, how should government expenditures be apportioned among different agencies in view of competing claims of public health, education, housing, and, say, national defense so as to provide the maximum satisfaction of public needs according to these ideals?

The formal setting of the problem and its solution are in each case essentially the same. They always involve the relationship between means and ends. The decision maker can influence the magnitude of the variable that he would like to “maximize” or “minimize” only indirectly by manipulating other “variables” upon which that magnitude in its turn depends. The Edison Electric Company can affect the cost of power by building atomic instead of conventional thermal plants or by shifting from coal to oil or to natural gas. The government can affect the satisfaction of competing public needs by financing more hospitals and fewer schools or more of both at the expense of military outlays.

To make a purposeful, rational choice the decision maker must first of all have a well-defined aim, he must know what it is that he wants to maximize or minimize; he must have what economists call an “objective function.” Second, he has to know the “limiting condition”: the causal, structural relationship between the ultimate objective and the factors that he can in fact control. The power company must have detailed knowledge of the technical characteristics of various types of plants among which it has to choose as well as the costs of different fuels. The government must be able to assess the efficiency of various schools and hospitals that it may want to finance.

In cases where the “objective function” is intricate and the structural relations describing the “limiting conditions” numerous and complex, not only the solution but even the formulation of a maximum problem requires much logical and mathematical finesse. Consider, for example, a political program that calls for “maximum national income” with “minimum inequality” in its distribution, or the theoretical proposition that corporations seek maximum profits with minimum risk. On close examination, these two statements turn out to make no real sense. In both instances, after a certain level of attainment, the strains of pushing against various structural limitations will be felt and the two different objectives will be found competing with each other: greater equality in the distribution of national income will be attainable only at the cost of some reduction in its total level; the reduction in the riskiness of a business venture will require some sacrifice in the expected level of profits.

Rational choice presupposes the existence of a well-defined single goal. In the first example the level of income and some measure of the equality (or inequality) of its distribution would have to be viewed not as two separate objectives of economic policies, but rather as two different factors contributing each in its own well-specified way to the attainment of the unique goal of maximizing something that, for lack of a better word, might be called public welfare. Only after we visualize a single “objective function” describing the policy maker’s assessment of the specific dependence of public welfare on these two and perhaps also many other factors, can his actions be legitimately interpreted as rational maximizing choices.

For a similar reason the profit motive can be used to explain entrepreneurial decisions only if the quantity to be maximized is specified as depending not only on the expected profits but also on some index of the incurred risk. The objective function turns out again to be not something that one can observe directly but rather an abstract entity whose existence can be ascertained only through introspection or indirect inference.

However complicated the formal solution of the related mathematical problems may be, experience has shown that the difficulty of using the maximum principle to explain the behavior of actual decision makers in the real world is incomparably greater. And this does not refer to the mechanics of large-scale data handling that have been fully mastered by modern computers, but rather to the difficulties of knowing, for example, what exactly a typical American corporation tries to maximize: What are the specific limiting conditions that its management takes into account in deciding on a particular course of action? Does it seek a certain rate of profit, for example, or a certain rate of growth, and what risks will it accept to gain either or both?

In reviewing the answers of Adolf Berle and John Kenneth Galbraith to these questions, Heilbroner finds the first to be “soft, uncritical and complaisant,” the second “essentially ambiguous.” Berle, after having first exposed and explained the immense power wielded by the few men controlling our largest corporations, indeed ended up (as quoted by Heilbroner) by calling them the “conscience earners of twentieth-century America.” Could he have prophetically been referring to Mr. Geneen of ITT? As for Galbraith, let us hope that his position will be made clearer in his future writings.

The specific data that should be fitted into the elaborate theoretical framework prepared to receive them are missing. Heilbroner opens his critical attack on contemporary neoclassical economics with the observation that formal solutions of more and more intricate “maximum” problems have contributed very little to a concrete factual explanation of the operation and development of our own or of any other economic system.

Maximizing…constitutes the bedrock on which conventional economic theory rests [p. 169)… [but] once we enter the world of oligopoly [a few large sellers], maximization of returns becomes an aim that can be translated into practice through the most contradictory activities—a fact that again makes the word “maximizing” disconcertingly empty of precise meaning. [P. 170]

As Paul Samuelson explained in his brilliant Nobel Prize address in 1971, not only the actions of consumers and producers but even the behavior of Newton’s falling apple can be interpreted as if they were solutions of appropriately formulated maximizing or minimizing problems. This clearly demonstrates, of course, that the maximum principle is in its formal generality as devoid of any empirical content as, say, the principle of multiplication or division. Its power as a tool of factual analysis depends entirely on an accurate specification of the function that is maximized (or minimized) in a particular operational context and on the full knowledge of the existing “limiting conditions.” The maximizing principle cannot serve as a substitute for factual knowledge.

Most of Heilbroner’s strictures against academic economists derive from their failure to understand the “objective functions” of the principal players on the contemporary social-economic scene and to assess correctly the technical and human factors determining the outcome of their actions. As we have seen above, he finds essentially the same faults in their academic critics too. One would expect that to set things right Heilbroner would urge his professional colleagues to roll up their sleeves and turn to systematic empirical work, however unpleasant, time consuming, and unrewarding (in academic advancement) it might be.

Instead he argues, in effect, that academic economists should give up the attempt to arrive at a better understanding of the operation of the economic system as it is and turn to the more relevant, more urgent task of helping to improve it. “The erstwhile determinacy of economic behavior is weakened to such an extent that it is no longer capable of supporting a superstructure of traditional theory.” Quoting Adolph Lowe,¹ whose decisive influence on his own thinking he repeatedly acknowledges, Heilbroner concludes, “‘Economics as a medium of passive contemplation must be converted into political economics, namely, an implement of active interference with the course of these [economic] processes.”‘

“Erstwhile determinacy” harks back to Marx’s leap from the implacable laws of development under capitalism to the freedom of rational choice he claimed would be possible under socialism. But for the economists Heilbroner is criticizing, the timing of the leap is now advanced to the present. “The question ultimately resolves in unstable beliefs in determinism and free will. But even if we take an extreme deterministic point of view it is clear that the problem of amassing the relevant data is qualitatively different in this case…. An inner core of the decision making process will be for all intents and purposes beyond any possible information system.” In Heilbroner’s view, we cannot possibly know with precision how and why producers, consumers, workers, bankers will decide things. And what we do not and cannot know cannot restrict our freedom.

This observation is not only true in a subjective sense, it also provides a convenient starting point for understanding the relationship between economic knowledge—for example, a systematic (“deterministic”) explanation of the operations of a given economic system—and action intended to bring about some desired change in the economy.

We can understand such a “deterministic” explanation better if we visualize it as being presented in the form of a mathematical system containing as many limiting conditions (i.e., equations) as it contains unknowns, with the values of all the unknowns, being uniquely determined by the given shapes of the equations. Let us consider, for example, the new, sophisticated approach to the problem of inflation. Reduced to the simplest terms, the analytical system used consists of two equations that are supposed to describe two different relationships between the same two variables: the rate of unemployment (let it be called x) and the rate of the corresponding price rise (y).

When the rate of unemployment is quite low (say 3 percent), i.e., when the labor market is tight, wage raises as high as 12 percent per annum might be needed to attract additional workers. On the other hand, when the unemployment rate is very high (say 18 percent) those who are lucky enough to have jobs would be willing to go on working—so goes the argument—even if wages were rising by only 2 percent per annum. In extreme cases they might be ready even to accept wage cuts, but this is most unlikely.

Both situations described above—as well as all the intermediate situations—can be described concisely by the following equation:

y = 36/x

If x, i.e., the rate of unemployment appearing on the right-hand side, is set at 3 the corresponding magnitude of y—i.e., the rate of wage and price rise “required” to keep the 3 percent rate of unemployment—turns out to be 12 percent (= 36/3). The same equation shows that a very high rate of unemployment of 18 percent would be compatible with a wage and price rise of only 2 percent per annum.

This equation—its correct “shape,” i.e., the actual rates of wage and price rise, would have to be determined through factual observation—describes all combinations of unemployment rates and of wage and price inflation rates compatible with the existing structure of what might be called the supply side of the labor market.

The second equation is intended to describe all the different combinations of wage rate changes (y) and employment levels (x)—as reflected in corresponding percentage rates of unemployment—compatible with all the different factors governing the behavior of the employers. The technologically conditioned dependence of labor productivity on the total level of employment is probably the most important structural condition limiting the range of possible alternatives on this side.

Let that second equation be of the following simple shape:

y = 4x

This says that if the wages were rising at the rate of, say, 24 percent per annum the total number of jobs available would be such as to keep the rate of unemployment at 6 percent (24 = 4×6).

Remembering that it describes only one part of the complete structure, we also find that according to the same relationship unemployment could be kept down to 1 percent provided the annual wage increases were held at 4 percent. However, turning back to the first equation we find the structure of workers’ attitudes and behavior to be such that the annual wage rise could be kept down to the low rate of 1 percent only under the pressure of unemployment as high as 36 percent. In satisfying one set of structural conditions the system seems to run against the limitation imposed by the other set.

There exists one and only one combination of the two rates that would comply with both sets of structural relationships at the same time. It can be easily verified that the solution y = 12, x = 3 satisfies indeed both equations simultaneously. An unemployment rate of 3 percent combined with a wage rise of 12 percent per annum is the only one that can coexist without violating any of the given structural conditions.² Had the equations set up above for illustrative purposes described correctly the structural characteristics of an existing economy, the rates of 3 and 12 percent respectively would indeed have been observed.

An economist assigned the task of explaining the prevailing rates of unemployment and of wage and price rise could discharge this task fully only by ascertaining the shapes of both equations. In such a completely determined system there is no elbow room for action designed to make the solution different from what it actually is. Things are as they are and nothing can be done about it. But let us drop one equation, i.e., one of the limiting conditions, and the uniqueness of the solution disappears. The number of remaining equations being now smaller than the number of the variables, the solution becomes indeterminate; that is, instead of only one, many (usually infinitely many) different combinations of the values of the unknown variables will be compatible with the restrictions imposed by the remaining limitations.

In the above example, if one of the two equations is dropped what remains is a system containing two unknowns, but only a single equation. By inserting on one side of this equation an arbitrarily chosen value of one of the two unknowns we can compute the corresponding value of the other. Instead of a single, unique solution we have a whole range of possible combinations. It is as if one part were removed from a rigid metal structure and as a result of this the structure had acquired flexibility, that is, capability to form configurations different from that in which it was originally locked. If some of these different feasible configurations can serve a given purpose, whatever it may be, better than other configurations, action can be taken to bring about a state that will serve that purpose best. The flexibility, the freedom of action thus acquired, is still, however, limited; it is circumscribed by the properties of the remaining structure, properties determined by the configuration of its remaining parts.

The advance in our understanding of socio-economic reality can be visualized as a progressive closing, a gradual tightening of the explanatory scheme. The introduction of new, previously unknown relationships between the variables involved reduces step by step the range of indeterminacy that characterizes all empirical knowledge—or the lack of it—at any stage of this cumulative process. A fully validated closed system can be conceived only as a practically unattainable ideal.

Hundreds or even thousands of variables, depending on the amount of details required, are used in descriptions of the complex modern economy. To explain its actual state we would have to ascertain the specific shapes of hundreds and thousands of structural relationships. The practical challenge of such a task increases with its size.

Large amounts of data have been collected and systematically organized in the United States, Japan, the Soviet Union, and many other countries on the structure of all productive sectors that transform natural resources, labor, and all kinds of other inputs into final goods. But our knowledge of the “human equations” that govern the behavior of those who supply the labor and consume the final output, not to speak of those who manage the so-called public sectors, is still quite rudimentary.

This should not, however, be taken as an indication that such equations do not exist but rather that their structure has not yet been ascertained. Our ignorance of the actual sources of human behavior is not surprising: no corn can be grown, no cars built without someone knowing how to do it and also knowing that he knows it. It is possible that human actions and reactions have a structure as rigid or even more rigid than the technological processes involved in growing corn or making automobiles; still the actors themselves need not be aware of the processes that impel them to act or even of their existence.

This brings up the question whether it is possible to take purposeful action that is rationally designed to bring about a fundamental socio-economic change. If based on past experience, the answer is no, or hardly. While Keynesian fiscal policies have proved to be capable of providing strong defenses against economic cave-ins of the type experienced in the early Thirties, they seem to aggravate the problem of inflation. As a matter of fact, the original Keynesian model is being recalled now for modification. That modification, in fact, consists in accepting the two simple equations I described earlier. It demonstrates, as we have seen, that full employment cannot be maintained without inflation.

Consider the disappointing performance of the Russian and other socialist economies, the limited success of developmental planning in backward areas, the obvious futility of diverse attempts to induce the leadership of large American corporations to transcend the profit motive. Such efforts at “active regulation of behavior through the use of reason” fall at least as short of their avowed goals as conventional economic theory falls short of a satisfactory realization of its aim to explain the operation of the economic system as it actually is, not as one assumes or wishes it to be.

A failure of a policy to attain its ends can usually be attributed to insufficient knowledge of conditions and relationships on which the outcome of that policy actually depends. From this, critics are apt to conclude that if he only had acquired such knowledge, the policy maker should certainly be capable of designing a course of action that would ensure the attainment of his ends. But this is not necessarily true. Better, more complete knowledge might simply make it plain that certain ends cannot be attained.

Imagine that the Council of Economic Advisers had found out the shape of the second of the two equations presented above and decided on the policy (consistent with the structural condition described by it) of keeping the annual wage rise down to 4 percent and the rate of unemployment to 1 percent. But imagine also that before leaping into action it did some more factual research and discovered the existence of the additional structural restriction described by the first equation. This, of course, would have dashed all hope of reducing the unemployment rate below 4 percent.

Is such knowledge useful? The answer should be yes. Even partial knowledge is useful, since it can protect us from at least some, if not all, mistakes. Before the existence of the first structural relationship was discovered, the policy makers could have already seen that the restrictions inherent in the second would have frustrated any effort to attain a 1 percent rate of unemployment combined with a wage rise of, say, 3 percent.

Heilbroner is concerned to point out that such technical, natural, and ecological limitations on the freedom of human choice must be accepted.

In the future as in the past the development of the technology of production seems bounded by the constraints of knowledge and capabilities and thus, in principle at least, open to prediction as a determinable force of the historic process;…underdeveloped countries can never hope to achieve parity with the developed countries. Given our present and prospective technology, there are simply not enough resources to permit a “Western” rate of industrial exploitation to be expanded to populations of four billion—much less eight billion persons.

If man cannot free himself from the limits imposed on the modes of his existence by the laws of nature, can he nevertheless not guide the system as a whole toward freely chosen goals through purposeful control of the human variables? Or does the system appear to be open in this respect only because the social sciences have not yet discovered the missing human equations? These might turn out to be as rigid and unyielding as the other already ascertained limiting conditions.

The subjective perception of the freedom of an individual will does not of course provide an answer to these questions. The strands of causal relationships capable of explaining human purposes and actions—if they can be traced at all—will be found strung through the very core of subjective motivations.

In his critical comments on the contemporary scene Heilbroner tends to emphasize not so much questionable values as inadequate knowledge: “The absence of a satisfactory foundation of knowledge beneath the socialist conception of human nature” and in particular its “failure to examine the nature and consequences of alternative motivations to those of material incentive” accounts, according to Heilbroner, for the disappointing performance of centrally planned socialist countries.

In an essay entitled “Innocence Abroad” he chides the managers of the Western as well as the underdeveloped countries for their “failure to recognize that the central inescapable and indispensable precondition for economic development is political and social change on an unending and tearing scale.” Here, as on several other occasions, he drops or at least waives the reservations advanced in his methodological essays and offers a concrete factual observation on the connection between economic growth and social structure. What he says is true and even important; but Marx said it more than a hundred years ago.

Marx illuminated the role of conflicting economic interests as driving forces in the social and political process. His followers, however, never made good on the vast claims staked out in that brilliant flash of insight along these inaccessible frontiers of knowledge; neither, one hastens to add, has anyone else. The structure of relationships that governs the interplay of these conflicting interests in the capitalist and the socialist society or in transition from one to another has not yet been articulated with a clarity even remotely approaching that with which economists can now describe various processes of production and consumption.

Inspired by the urge to lend a helping hand in the building of a better world, and impatient to put to use their new tool kits, many economic theorists are now engaged in “maximizing” all kinds of arbitrarily conceived social objective functions. Moreover, they derive their ideal solutions within the framework of palpably incomplete sets of limiting conditions. For example, a technical paper presented recently at a scientific meeting developed a proposal to incorporate the length of human life into the conventional measure of national income: the underlying assumption of the paper was that longer life is desirable because it permits more consumption. The author accordingly included in his social objective function the total amounts of goods consumed by citizens belonging to various income groups over their entire life spans. A peculiar implication of this particular index of social welfare is that the life of a rich person capable of maintaining year after year a high level of consumption is valued much higher than the life of a less fortunate citizen spending his days under conditions of constant deprivation.

While an economist certainly has as much right as anyone else to express his views on the ultimate objective of social policies, there is no reason why his view of human values should carry greater weight than anybody else’s. But when it comes to assessing the restraints imposed on the attainment of a given goal—whatever it may be—by the existing limiting conditions, the expert can be expected to have a definite advantage over the layman.

When the expert overlooks in his computations some of the relevant limiting conditions he naturally will err in his advice, and the error will necessarily be in the overoptimistic direction. This probably is the main reason why most five-year plans have been underfulfilled and practically all expense budgets—private as well as public—underestimated. More knowledge in this respect should bring about a lowering of targets.

Discussing the “limits of economic prediction,” Heilbroner presents what he characterizes as a “generous case” in favor of “negative predictions.” The less one knows about a social or economic system, the more flexible it might be thought to be; hence each additional bit of specific information about its structure might add to our knowledge of what cannot be accomplished with it or within it. If the advisers of three successive Presidents had known better we would not be in Vietnam now. But probably those who knew better could not have become Presidential advisers. If this is true, it was probably inevitable that the United States would intervene in Vietnam.

The antinomy of free will and determinism, the quest for the knowledge of the world as it is and the desire to change it, is as unresolved now as it was at the time of Calvin. No head-on clash is imminent between the two points of view since the gulf of ignorance separating the adherents of one from those of the other is vast, particularly in the social sciences.

Moreover, the proponents of the principle of free will do in fact seek the knowledge of structural relationships—relationships between means and ends—that would enable them to assess correctly the consequences of alternative actions they might decide to take: they are anxious to preserve just enough openness, i.e., indeterminacy in our view of the world, to secure elbow room for meaningful choice. But how much is enough? Heilbroner does not answer this question; he does not even pose it.