It is well-known that Austrians disagree strongly with other schools of economic thought, such as the Keynesians, the Monetarists, the Public Choicers, Historicists, Institutionalists, and Marxists.1 Disagreement is most conspicuous, of course, when it comes to economic policy and economic policy proposals. At times there also exists an alliance between Austrians and, in particular, Chicagoites and Public Choicers. Ludwig von Mises, Murray N. Rothbard, Milton Friedman, and James Buchanan, to cite a few names, are often united in their efforts to defend the free market economy against its "liberal" and socialist detractors.
Nonetheless, as important as such occasional agreements may be for tactical or strategic reasons, they can only be superficial, for they cover up some truly fundamental differences between the Austrian school, as represented by Mises and Rothbard, and all the rest. The ultimate difference from which all disagreements at the levels of economic theory and economic policy stem—disagreements, for instance, as regards the merit of the gold standard vs. fiat money, free-banking vs. central banking, the welfare implications of markets vs. state-action, capitalism vs. socialism, the theory of interest and the business cycle, etc.—concerns the answer to the very first question that any economist must raise: What is the subject matter of economics, and what kind of propositions are economic theorems?
Mises’s answer is that economics is the science of human action. In itself, this may not sound very controversial. But then Mises says of the science of economics:
Its statements and propositions are not derived from experience. They are, like those of logic and mathematics, a priori. They are not subject to verification and falsification on the ground of experience and facts. They are both logically and temporally antecedent to any comprehension of historical facts. They are a necessary requirement of any intellectual grasp of historical events.2
In order to emphasize the status of economics as a pure science, a science that has more in common with a discipline like applied logic than, for instance, with the empirical natural sciences, Mises proposes the term “praxeology” (the logic of action) for the branch of knowledge exemplified by economics.3
It is this assessment of economics as an a priori science, a science whose propositions can be given a rigorous logical justification, which distinguishes Austrians, or more precisely Misesians, from all other current economic schools. All the others conceive of economics as an empirical science, as a science like physics, which develops hypotheses that require continual empirical testing. And they all regard as dogmatic and unscientific Mises’s view that economic theorems—like the law of marginal utility, or the law of returns, or the time-preference theory of interest and the Austrian business cycle theory—can be given definite proof, such that it can be shown to be plainly contradictory to deny their validity.
The view of Mark Blaug, highly representative of mainstream methodological thought, illustrates this almost universal opposition to Austrianism. Blaug says of Mises, “His writings on the foundations of economic science are so cranky and idiosyncratic that one can only wonder that they have been taken seriously by anyone.”4
Blaug does not provide one argument to substantiate his outrage. His chapter on Austrianism simply ends with that statement. Could it be that Blaug’s and others’ rejection of Mises’s apriorism may have more to do with the fact that the demanding standards of argumentative rigor, which an apriorist methodology implies, prove too much for them?5
What led Mises to his characterization of economics as an a priori science? From the present day perspective it might be surprising to hear that Mises did not see his conception as out of line with the mainstream view prevailing in the early twentieth century. Mises did not wish to prescribe what economists should be doing as opposed to what they actually were doing. Rather, he saw his achievement as a philosopher of economics in systematizing, and in making explicit what economics really was, and how it had implicitly been conceived by almost everyone calling himself an economist.
And this is indeed the case. In giving a systematic explanation of what was formerly only implicit and unspoken knowledge, Mises did introduce some conceptual and terminological distinctions that had previously been unclear and unfamiliar, at least to the English-speaking world. But his position on the status of economics was essentially in full agreement with the then-orthodox view on the matter. They did not employ the term “a priori,” but such mainstream economists as Jean Baptiste Say, Nassau Senior, and John E. Cairnes, for instance, described economics quite similarly.
Say writes: “A treatise on political economy will . . . be confined to the enunciation of a few general principles, not requiring even the support of proofs or illustrations; because these will be but the expression of what every one will know, arranged in a form convenient for comprehending them, as well as in their whole scope as in their relation to each other.” And “political economy . . . whenever the principles which constitute its basis are the rigorous deductions of undeniable general facts, rests upon an immovable foundation.“6
According to Nassau Senior, economic “premises consist of a few general propositions, the result of observations, or consciousness, and scarcely requiring proof, or even formal statement, which almost every man, as soon as he hears them, admits as familiar to his thoughts, or at least as included in his previous knowledge; and his inferences are nearly as general, and, if he has reasoned correctly, as certain as his premises.” And economists should be “aware that the Science depends more on reasoning than on observation, and that its principal difficulty consists not in the ascertainment of its facts, but in the use of its terms.”7
And John E. Cairnes remarks that while “mankind has no direct knowledge of ultimate physical principles” . . . “the economist starts with a knowledge of ultimate causes.” . . . “The economist may thus be considered at the outset of his researches as already in possession of those ultimate principles governing the phenomena which form the subject of his study, the discovery of which in the case of physical investigation constitutes for the inquirer his most arduous task.” “Conjecture [in economics] would manifestly be out of place, inasmuch as we possess in our consciousness and in the testimony of our senses . . . direct and easy proof of that which we desire to know. In Political Economy, accordingly, hypothesis is never used as a help toward the discovery of ultimate causes and laws.”8
The views of Mises’s predecessors, Menger, Böhm-Bawerk, and Wieser, are the same: They, too, describe economics as a discipline whose propositions can—in contrast to those of the natural sciences—be given some ultimate justification. Again, however, they do so without using the terminology employed by Mises.9
And finally, Mises’s epistemological characterization of economics was also considered quite orthodox—and certainly not idiosyncratic, as Blaug would have it—after having been explicitly formulated by Mises. Lionel Robbins’s book The Nature and Significance of Economic Science, which first appeared in 1932, is nothing but a somewhat watered-down version of Mises’s description of economics as praxeology. Yet it was respected by the economics profession as the guiding methodological star for almost twenty years.
In fact, Robbins, in his Preface, explicitly singles out Mises as the most important source of his own methodological position. And Mises and Richard von Strigl—whose position is essentially indistinguishable from Mises’s10—are cited approvingly in the text more often than anyone else.11
Yet, illuminating as all this may be for an assessment of the present-day situation, it is only history. What then is the rationale of the classical economists for regarding their science as different than the natural sciences? And what is behind Mises’s explicit reconstruction of this difference as one between an a priori science and an aposteriori science? It was the recognition that the process of validation—the process of discovering whether some proposition is true or not—is different in one field of inquiry than in the other.
Let us first look briefly at the natural sciences. How do we know what the consequences will be if we subject some nature-given material to specified tests, let’s say, if we mix it with another kind of material? Obviously we do not know before we actually try it and observe what happens. We can make a prediction, of course, but our prediction is only a hypothetical one, and observations are required to find out if we are right or wrong.
Moreover, even if we have observed some definite outcome, let’s say that mixing the two materials leads to an explosion, can we then be sure that such an outcome will invariably occur whenever we mix such materials? Again, the answer is no. Our predictions will still, and permanently, be hypothetical. It is possible that an explosion will only result if certain other conditions—A, B, and C—are fulfilled. We can only find out whether or not this is the case and what these other conditions are by engaging in a never-ending trial and error process. This enables us to improve our knowledge progressively about the range of application for our original hypothetical prediction.
Now let us turn to some typical economic propositions. Consider the validation process of a proposition such as the following: Whenever two people A and B engage in a voluntary exchange, they must both expect to profit from it. And they must have reverse preference orders for the goods and services exchanged so that A values what he receives from B more highly than what he gives to him, and B must evaluate the same things the other way around.
Or consider this: Whenever an exchange is not voluntary but coerced, one party profits at the expense of the other.
Or the law of marginal utility: Whenever the supply of a good increases by one additional unit, provided each unit is regarded as of equal serviceability by a person, the value attached to this unit must decrease. For this additional unit can only be employed as a means for the attainment of a goal that is considered less valuable than the least valued goal satisfied by a unit of such good if the supply were one unit shorter.
Or take the Ricardian law of association: Of two producers, if A is more productive in the production of two types of goods than is B, they can still engage in a mutually beneficial division of labor. This is because overall physical productivity is higher if A specializes in producing one good which he can produce most efficiently, rather than both A and B producing both goods separately and autonomously.
Or as another example: Whenever minimum wage laws are enforced that require wages to be higher than existing market wages, involuntary unemployment will result.
Or as a final example: Whenever the quantity of money is increased while the demand for money to be held as cash reserve on hand is unchanged, the purchasing power of money will fall.
Considering such propositions, is the validation process involved in establishing them as true or false of the same type as that involved in establishing a proposition in the natural sciences? Are these propositions hypothetical in the same sense as a proposition regarding the effects of mixing two types of natural materials? Do we have to test these economic propositions continuously against observations? And does it require a never-ending trial and error process in order to find out the range of application for these propositions and to gradually improve our knowledge, such as we have seen to be the case in the natural sciences?
It seems quite evident—except to most economists for the last forty years—that the answer to these questions is a clear and unambiguous No. That A and B must expect to profit and have reverse preference orders follows from our understanding of what an exchange is. And the same is the case concerning the consequences of a coerced exchange. It is inconceivable that things could ever be different: It was so a million years ago and it will be so a million years hence. And the range of application for these propositions too is clear once and for all: They are true whenever something is a voluntary exchange or a coerced exchange, and that is all there is to it.
There is no difference with respect to the other examples given. That the marginal utility of additional units of supply of homogeneous goods must fall follows from the incontestable statement that every acting person always prefers what satisfies him more over what satisfies him less. It is simply absurd to think that continuous testing would be required to establish such a proposition.
The Ricardian law of association, along with a once-and-for-all delineation of its range of application, also logically follows from the very existence of the situation described. If A and B differ as described and accordingly there exists a technological substitution ratio for the goods produced (one such rate for A and one for B), then if they engage in a division of labor as characterized by the law, the physical output produced must be greater than it otherwise would be. Any other conclusion is logically flawed.
The same is true regarding the consequences of minimum wage laws or an increase in the quantity of money. An increase in unemployment and a decrease in the purchasing power of money are consequences which are logically implied in the very description of the initial condition as stated in the propositions at hand. As a matter of fact, it is absurd to regard these predictions as hypothetical and to think that their validity could not be established independently of observations, i.e., other than by actually trying out minimum wage laws or printing more money and observing what happens.
To use an analogy, it is as if one wanted to establish the theorem of Pythagoras by actually measuring sides and angles of triangles. Just as anyone would have to comment on such an endeavor, mustn’t we say that to think economic propositions would have to be empirically tested is a sign of outright intellectual confusion?
But Mises by no means merely notices this rather obvious difference between economics and the empirical sciences. He makes us understand the nature of this difference and explains how and why a unique discipline like economics, which teaches something about reality without requiring observations, can possibly exist. It is this achievement of Mises’s which can hardly be overrated.
In order to better understand his explanation, we must make an excursion into the field of philosophy, or more precisely into the field of the philosophy of knowledge or epistemology. In particular, we must examine the epistemology of Immanuel Kant as developed most completely in his Critique of Pure Reason. Mises’s idea of praxeology is clearly influenced by Kant. This is not to say that Mises is a plain and simple Kantian. As a matter of fact, as I will point out later, Mises carries the Kantian epistemology beyond the point at which Kant himself left off. Mises improves the Kantian philosophy in a way that to this very day has been completely ignored and unappreciated by orthodox Kantian philosophers. Nonetheless, Mises takes from Kant his central conceptual and terminological distinctions as well as some fundamental Kantian insights into the nature of human knowledge. Thus we must turn to Kant.
Kant, in the course of his critique of classical empiricism, in particular that of David Hume, developed the idea that all our propositions can be classified in a two-fold way: On the one hand they are either analytic or synthetic, and on the other they are either a priori or a posteriori. The meaning of these distinctions is, in short, the following. Propositions are analytic whenever the means of formal logic are sufficient in order to find out whether they are true or not; otherwise propositions are synthetic ones. And propositions are a posteriori whenever observations are necessary in order to establish their truth or at least confirm them. If observations are not necessary, then propositions are a priori.
The characteristic mark of Kantian philosophy is the claim that true a priori synthetic propositions exist—and it is because Mises subscribes to this claim that he can be called a Kantian. Synthetic a priori propositions are those whose truth-value can be definitely established, even though in order to do so the means of formal logic are not sufficient (while, of course, necessary) and observations are unnecessary.
According to Kant, mathematics and geometry provide examples of true a priori synthetic propositions. Yet he also thinks that a proposition such as the general principle of causality—i.e., the statement that there are time-invariantly operating causes, and every event is embedded into a network of such causes—is a true synthetic a priori proposition.
I cannot go into great detail here to explain how Kant justifies this view.12 A few remarks will have to suffice. First, how is the truth of such propositions derived, if formal logic is not sufficient and observations are unnecessary? Kant’s answer is that the truth follows from self-evident material axioms.
What makes these axioms self-evident? Kant answers, it is not because they are evident in a psychological sense, in which case we would be immediately aware of them. On the contrary, Kant insists, it is usually much more painstaking to discover such axioms than it is to discover some empirical truth such as that the leaves of trees are green. They are self-evident because one cannot deny their truth without self-contradiction; that is, in attempting to deny them one would actually, implicitly, admit their truth.
How do we find such axioms? Kant answers, by reflecting upon ourselves, by understanding ourselves as knowing subjects. And this fact—that the truth of a priori synthetic propositions derives ultimately from inner, reflectively produced experience—also explains why such propositions can possibly have the status of being understood as necessarily true. Observational experience can only reveal things as they happen to be; there is nothing in it that indicates why things must be the way they are. Contrary to this, however, writes Kant, our reason can understand such things as being necessarily the way they are, “which it has itself produced according to its own design.”13
In all this Mises follows Kant. Yet, as I said earlier, Mises adds one more extremely important insight that Kant had only vaguely glimpsed. It has been a common quarrel with Kantianism that this philosophy seemed to imply some sort of idealism. For if, as Kant sees it, true synthetic a priori propositions are propositions about how our mind works and must of necessity work, how can it be explained that such mental categories fit reality? How can it be explained, for instance, that reality conforms to the principle of causality if this principle has to be understood as one to which the operation of our mind must conform? Don't we have to make the absurd idealistic assumption that this is possible only because reality was actually created by the mind? So that I am not misunderstood, I do not think that such a charge against Kantianism is justified.14 And yet, through parts of his formulations Kant has no doubt given this charge some plausibility.
Consider, for example, this programmatic statement of his: “So far it has been assumed that our knowledge had to conform to observational reality”; instead it should be assumed “that observational reality conform to our knowledge.”15
Mises provides the solution to this challenge. It is true, as Kant says, that true synthetic a priori propositions are grounded in self-evident axioms and that these axioms have to be understood by reflection upon ourselves rather than being in any meaningful sense “observable.” Yet we have to go one step further. We must recognize that such necessary truths are not simply categories of our mind, but that our mind is one of acting persons. Our mental categories have to be understood as ultimately grounded in categories of action. And as soon as this is recognized, all idealistic suggestions immediately disappear. Instead, an epistemology claiming the existence of true synthetic a priori propositions becomes a realistic epistemology. Since it is understood as ultimately grounded in categories of action, the gulf between the mental and the real, outside, physical world is bridged. As categories of action, they must be mental things as much as they are characteristics of reality. For it is through actions that the mind and reality make contact.
Kant had hinted at this solution. He thought mathematics, for instance, had to be grounded in our knowledge of the meaning of repetition, of repetitive operations. And he also realized, if only somewhat vaguely, that the principle of causality is implied in our understanding of what it is and means to act.16
Yet it is Mises who brings this insight to the foreground: Causality, he realizes, is a category of action. To act means to interfere at some earlier point in time in order to produce some later result, and thus every actor must presuppose the existence of constantly operating causes. Causality is a prerequisite of acting, as Mises puts it.
But Mises is not, as is Kant, interested in epistemology as such. With his recognition of action as the bridge between the mind and the outside reality, he has found a solution to the Kantian problem of how true synthetic a priori propositions can be possible. And he has offered some extremely valuable insights regarding the ultimate foundation of other central epistemological propositions besides the principle of causality, such as the law of contradiction as the cornerstone of logic. And he has thereby opened a path for future philosophical research that, to my knowledge, has hardly been traveled. Yet Mises’s subject matter is economics, and so I will have to lay to rest the problem of explaining in more detail the causality principle as an a priori true proposition.17
Mises not only recognizes that epistemology indirectly rests on our reflective knowledge of action and can thereby claim to state something a priori true about reality but that economics does so too and does so in a much more direct way. Economic propositions flow directly from our reflectively gained knowledge of action; and the status of these propositions as a priori true statements about something real is derived from our understanding of what Mises terms “the axiom of action.”
This axiom, the proposition that humans act, fulfills the requirements precisely for a true synthetic a priori proposition. It cannot be denied that this proposition is true, since the denial would have to be categorized as an action—and so the truth of the statement literally cannot be undone. And the axiom is also not derived from observation—there are only bodily movements to be observed but no such things as actions—but stems instead from reflective understanding.
Moreover, as something that has to be understood rather than observed, it is still knowledge about reality. This is because the conceptual distinctions involved in this understanding are nothing less than the categories employed in the mind’s interaction with the physical world by means of its own physical body. And the axiom of action in all its implications is certainly not self-evident in a psychological sense, although once made explicit it can be understood as an undeniably true proposition about something real and existent.18
Certainly, it is not psychologically evident nor is it observable that with every action an actor pursues a goal; and that whatever the goal may be, the fact that it is pursued by an actor reveals that he places a relatively higher value on it than on any other goal of action he could conceive of at the start of his action.
It is neither evident nor observable that in order to achieve his most highly valued goal an action must interfere or decide not to interfere (which, of course, is also an interference) at an earlier point in time to produce some later result; nor that such interferences invariably imply the employment of some scarce means (at least those of the actor’s body, its standing room and the time absorbed by the interference).
It is neither self-evident nor can it be observed that these means must also have value for an actor—a value derived from that of the goal—because the actor must regard their employment as necessary in order to effectively achieve the goal; and that actions can only be performed sequentially, always involving the making of a choice, i.e., taking up that one course of action which at some given point in time promises the most highly valued result to the actor and excluding at the same time the pursuit of other, less highly valued goals.
It is not automatically clear or observable that as a consequence of having to choose and give preference to one goal over another—of not being able to realize all goals simultaneously—each and every action implies the incurrence of costs. For example, forsaking the value attached to the most highly valued alternative goal that cannot be realized or whose realization must be deferred because the means necessary to effect it are bound up in the production of another, even more highly valued goal.
And lastly, it is not plainly evident or observable that at its starting point every goal of action must be considered worth more to the actor than its cost and capable of yielding a profit, i.e., a result whose value is ranked higher than that of the foregone opportunities. And yet, every action is also invariably threatened by the possibility of a loss if an actor finds, in retrospect, that the result actually achieved—contrary to previous expectations—has a lower value than the relinquished alternative would have had.
All of these categories—values, ends, means, choice, preference, cost, profit and loss, as well as time and causality—are implied in the axiom of action. Yet, that one is able to interpret observations in such categories requires that one already knows what it means to act. No one who is not an actor could ever understand them. They are not “given,” ready to be observed, but observational experience is cast in these terms as it is construed by an actor. Nor is their reflective reconstruction a simple, psychologically self-evident intellectual task, as proved by a long line of abortive attempts along the way to the just-outlined insights into the nature of action.
It took painstaking intellectual effort to recognize explicitly what, once made explicit, everybody recognizes immediately as true and can understand as true synthetic a priori statements, i.e., propositions that can be validated independently of observations and thus cannot possibly be falsified by any observation whatsoever.
The attempt to disprove the action-axiom would itself be an action aimed at a goal, requiring means, excluding other courses of action, incurring costs, subjecting the actor to the possibility of achieving or not achieving the desired goal and so leading to a profit or a loss.
And the very possession of such knowledge then can never be disputed, and the validity of these concepts can never be falsified by any contingent experience, for disputing or falsifying anything would already have presupposed their very existence. As a matter of fact, a situation in which these categories of action would cease to have a real existence could itself never be observed, for making an observation, too, is an action.
Mises’s great insight was that economic reasoning has its foundation in just this understanding of action; and that the status of economics as a sort of applied logic derives from the status of the action-axiom as an a priori-true synthetic proposition. The laws of exchange, the law of diminishing marginal utility, the Ricardian law of association, the law of price controls, and the quantity theory of money—all the examples of economic propositions which I have mentioned—can be logically derived from this axiom. And this is why it strikes one as ridiculous to think of such propositions as being of the same epistemological type as those of the natural sciences. To think that they are, and accordingly to require testing for their validation, is like supposing that we had to engage in some fact-finding process without knowing the possible outcome in order to establish the fact that one is indeed an actor. In a word: It is absurd.
Praxeology says that all economic propositions which claim to be true must be shown to be deducible by means of formal logic from the incontestably true material knowledge regarding the meaning of action.
Specifically, all economic reasoning consists of the following:
(1) an understanding of the categories of action and the meaning of a change occurring in such things as values, preferences, knowledge, means, costs, etc;
(2) a description of a world in which the categories of action assume concrete meaning, where definite people are identified as actors with definite objects specified as their means of action, with some definite goals identified as values and definite things specified as costs. Such description could be one of a Robinson Crusoe world, or a world with more than one actor in which interpersonal relationships are possible; of a world of barter exchange or of money and exchanges that make use of money as a common medium of exchange; of a world of only land, labor, and time as factors of production, or a world with capital products; of a world with perfectly divisible or indivisible, specific or unspecific factors of production; or of a world with diverse social institutions, treating diverse actions as aggression and threatening them with physical punishment, etc; and
(3) a logical deduction of the consequences which result from the performance of some specified action within this world, or of the consequences which result for a specific actor if this situation is changed in a specified way.
Provided there is no flaw in the process of deduction, the conclusions that such reasoning yield must be valid a priori because their validity would ultimately go back to nothing but the indisputable axiom of action. If the situation and the changes introduced into it are fictional or assumptional (a Robinson Crusoe world, or a world with only indivisible or only completely specific factors of production), then the conclusions are, of course, a priori true only of such a “possible world.” If, on the other hand, the situation and changes can be identified as real, perceived and conceptualized as such by real actors, then the conclusions are a priori true propositions about the world as it really is.19
Such is the idea of economics as praxeology. And such then is the ultimate disagreement that Austrians have with their colleagues: Their pronouncements cannot be deduced from the axiom of action or even stand in clear-cut contradiction to propositions that can be deduced from the axiom of action.
And even if there is agreement on the identification of facts and the assessment of certain events as being related to each other as causes and consequences, this agreement is superficial. For such economists falsely believe their statements to be empirically well-tested propositions when they are, in fact, propositions that are true a priori.