The a priori method in economics – In defence of Ludwig von Mises (essay)

Ludwig von MIses

By Detlev Schlichter

I gave a speech on this topic at the Libertarian Alliance in March. A link to the video recording of that speech is here. The following essay covers similar ground but is not identical with the speech. I develop the argument differently here. My hope is that this text is a better articulation of my views and hopefully a good basis for further debate.

The topic of this essay is broadly the method of economics: What phenomena does economics deal with? How do economists develop and test economic theories? What type of statements are the “laws of economics”, and what are they useful for?

Such far-reaching and deep issues cannot be covered adequately in an essay, so the following must remain a brief outline. The position I take on these questions is that of Ludwig von Mises, whose views remain controversial to say the least, even among many who are otherwise sympathetic to his positions (as, for example are David Ramsey Steele, David McDonagh, and J. C. Lester, all of whom I rate highly as writers and libertarian thinkers. See Steele From Marx to Mises and here and here, and Lester Escape from Leviathan).

Brief sketch of Mises’ position

Mises argued that economics was an a priori science. Economics is fundamentally different from the natural sciences in terms of subject matter, method, and the type of statements it can make. The natural sciences are (mainly, at least) empirical sciences (a posteriori as opposed to a priori). Economics is not an empirical science. It does not discover the regularities that constitute its laws through observation (for example, the collection and interpretation of statistics) or laboratory experiments (difficult in economics) but through careful logical deduction from certain starting propositions (axioms), although some observation may be involved in establishing these starting propositions. Because it is not an empirical science, it does not make empirical predictions and its key theses are not testable by empirical means. (If this sounds strange to you, don’t worry. Most economists today disagree with Mises and practice, or believe they practice, a different kind of economics, though in my view, they are quite mistaken. A lot of what is presented as “economics” to the public today does not quite deserve the label and is often intellectually weak. All this will hopefully become clearer soon.)

So is it useful? Yes, very much so, in fact, it is indispensable. Economic theory is antecedent to experience (a priori Latin: “from the earlier”). It provides a tool for understanding experience, for making experience intelligible in the first place, in this case the experience of economic phenomena, which are by definition complex phenomena. Without the laws of economics, we could not speak intelligently about observed market phenomena, make sense of what happens around us in terms of economic behavior, and in any reasonable way collect, organize and interpret economic data. Economics provides the essential abstract mental tools (which are necessarily independent of time and space, and thus generally valid) for approaching a specific situation in real life and dealing with a specific real-life problem of economics.

The laws of economics provide a searchlight, a type of X-ray, that illuminates the bony structure underlying all economic phenomena, the patterns and regularities that are at work in all human action as it relates to economic goods. (Mises famously defined economics as a subset of a wider science of human action in general, which he called praxeology, but for the purpose of this essay I stick to the narrower field of economics.)

“Falsifiability” through experience

As economics so defined is general, abstract and prior to experience, it cannot be falsified by experience, which means it is not “testable” or “refutable” in the way that most natural sciences are, and which has become in fact the generally accepted definition of what makes an inquiry a scientific inquiry in the first place, namely for the scientist to come up with testable hypotheses that can be falsified through experience.

It is this aspect that most riles many economists (the few that care about the epistemology of their science), other social scientists, and many epistemologists and philosophers. If the economist produces stuff that cannot be falsified empirically, so this criticism goes, then he may either produce tautologies (he simply re-arranges his starting axioms) or arbitrary nonsense, maybe both, and he can go on repeating it because the alleged non-falsifiability of it immunizes it against refutation.

In the following I will try and defend Mises. I believe that his position on the method of economics is correct. This topic is also extremely relevant, in my view, for any discussion of economic phenomena and ultimately for any policy discussions. We need to understand what economics can do and not do, and how it can go about its business reasonably and intelligently.

Importance and examples

It is important to stress from the start that Mises did not suggest that this method should be adopted, or that this was a method that should distinguish the Austrian School economists from other economists; that this was one available method next to others, and that the a priori one was just better than any alternatives. He claimed – correctly, in my view- that this was the method of economics. All the key tenets of economics developed over 300 years of systematic economic investigation, from Cantillon to Hume to Adam Smith to Ricardo to Carl Menger, were of such an a priori nature. Observation might have led these economists to develop their theories. Observation might have provided an initial spark; provided ideas or aroused interest. But economic theory proper is always logically derived from human conduct; it tells us something about the logic of action. To be an economist in the sense of being an economic theoretician is to think in terms of a priori concepts. To apply the laws of economics to specific economic problems in a given situation is to apply again a priori concepts. Some examples may help illustrate this:

The law of diminishing returns is an a priori law. It is essential for any analysis of real life economic situations. If economics were an empirical science and if its laws were subject to empirical testing, must we not fear that tomorrow we might encounter a maverick economic good to which the law of diminishing returns did not apply? This is impossible. An economic good that was not subject to the laws of diminishing returns would not be an economic good, as there would be no reason to economize on it. Being an economic good means being subject to the laws of diminishing returns.

Interest is the phenomenon that we value the same or similar goods differently if they are available at different points in time, and specifically, that we value future goods lower than present goods. If economics were an empirical science, would we not have to fear that tomorrow we encounter a group of people to whom that law did not apply? No, this is impossible (at least as long as these people have not yet discovered the secret of eternal life). Interest follows directly from time preference, and time preference is an integral part of what constitutes an economic good. “To want something means to want it, all else being equal, sooner rather than later” (George Reisman). Or, to put it differently, to be indifferent as to whether you enjoy a good today, in five years time, or in twenty years time, is equivalent to not caring about it, which means, the good in question is not an economic good to you in the first place. “Economic good” means you care about it, which means you experience time preference in relation to this good, which means the concept of interest applies to it, which means all a priori laws of interest apply to it.

Ricardo’s law of comparative cost has led to the law of association, which shows that it is advantageous (always marginally wealth-enhancing) for individuals and groups of individuals (nations) to engage in cooperation via free trade, even if one or more members of this trade network do not possess any comparative advantage, meaning their marginal productivity is lower in every relevant area than the marginal productivity of other members. Even then, everybody benefits from the inclusion of this member/these members in the network. This law is ultimately derived from the law of marginal utility, of which the law of diminishing returns is a subset. If economics were an empirical science, would we not have to fear that tomorrow we might encounter a group of people to whom the law of association did not apply? That is, again, impossible. Wherever there is trade in economic goods, to which the laws of marginal utility apply (otherwise they wouldn’t be economic goods), the law of association applies. The law of association is a priori. It is not subject to empirical testing but is logically derived from the key tenets of economic action and it is therefore a tool for making something as complex as trade and cooperation of people with different skills intelligible in the first place. (It is also the most powerful argument against any type of restriction to free trade.)

Let’s take an example from the field of money. David Ramsey Steele who is very knowledgeable about Mises and generally sympathetic to his work but rejects Mises’ apriori-ism, correctly emphasizes in his book and also here that it is often difficult to ascertain precisely which financial assets fulfil the function of money at a specific time and place. Over time, what is used as money has changed, and in a modern economy with a highly developed financial system the distinction between money and non-money can be blurred. Seemingly safe assets for which very liquid markets exist have occasionally assumed the role of quasi-monies. However, these are not problems of economic theory but of application of theory to specific situations. Once we ascertain what is used as money in a specific economy at a specific time, the laws of money as specified my monetary theory necessarily apply to this form of money.

Money is a medium of exchange, a facilitator of trade that is so widely used by the public that money prices also function as a reasonable basis for economic calculation and the monetary asset itself frequently as a store of value. Because money is the medium of exchange, demand for money (cash balances) is demand for readily exercisable purchasing power, and any changes in money demand can therefore in principle be met by changes in the purchasing power of the monetary unit and therefore without any additional production of money. At different purchasing powers the same quantity of money can satisfy different degrees of money demand, something that non-money goods cannot do. If economics were an empirical science, would we not have to fear that tomorrow we might encounter a form of money that was a maverick and that would falsify this rule, or to which this rule would not apply? No, this is nonsensical. By determining that something is money we also imply that changes in demand for it can be met by changes in its price. This is a priori.

Are these not simple tautologies? Are we not re-stating what is already entailed in the original concepts? – In a way yes. Mises was quite ready to accept this. We may label them tautologies but it does not make these deductions trivial, useless or arbitrary. In fact, more elaborate theories can and should be built on these basic theories. These theories are the keys to unlocking the logical structure inherent in all economic activity. As we are all economic actors we experience and we use these concepts daily, mostly without much further reflection. But the economist illuminates the underlying regularities and laws of economic action and makes intelligible the patterns that are necessarily at work.

Mises was not proposing a new method of economics. He was clarifying what it meant to do economics. As economics is a fairly young science (Mises said the youngest of all the sciences), it does not have a long history of epistemological analysis. Many of those who thought carefully about how economics works as a science came indeed to conclusions that are similar to Mises’: Nassau William Senior, John Stuart Mill, John Elliott Cairnes (maybe closest to Mises), Frank Knight, and Friedrich von Wieser. But Mises argued this position most consistently and convincingly.

What about modern mainstream economists?

Most modern economists appear to be doing something completely different. They are evidently using copious amounts of statistical data and applying mathematical procedures to it. How does this relate to what we just said?

Statistical data always describes historical events. It shows us what happened in a specific place at a specific time. It is impossible to approach statistical data, and to organize and interpret it without any theory whatsoever. Theory-free statistical analysis is impossible. The combination of (correct) a priori theory and historical data may allow us to understand what happened in the past, including the very recent past. Following Steele (as referenced before) we may, for example, try and understand to what degree liquid AAA-rated floaters functioned as near-monies in the run-up to the 2007 financial crisis. Or we may analyze how consumer good prices and producer goods prices behaved in the United States from 1929 to 1933. Or we may try to quantify the extent to which QE has probably lowered the borrowing costs of the state over the past 5 years.

What we cannot do is two things: We can neither verify nor falsify the a priori laws of economics, such as the ones listed above. Even more important (and potentially disappointing to those who derive their expectations as to what science is all about from the natural sciences) we cannot derive the laws of economics from mere observation, and that includes even the most extensive collection of data and the most elaborate and sophisticated analysis of it. The economists who claim to do this are either confused or simply play to the gallery (Piketty?) and are frequently not really proper economists, although some of them may even win Nobel Prizes. This may sound harsh but I believe it is true. The reasons for why we must fail to achieve these two things (test/verify/falsify economic laws and discover economic laws through statistics) are fundamental and I will give them below. Of course, if economics were a natural science, if it were an empirical science, these two things would not only be possible, they would be essential to its modus operandi as a science. Crucially, economics is not an empirical science in the sense that the natural sciences are.

This is in fact the reason why no amount of data mining and statistical analysis will ever settle disputes in the field of economics. Keynesian economists will forever quote historical data from around the Great Depression as evidence of their crisis theories and policy recommendations, just as those who subscribe to monetary explanations of the business cycle (as we “Austrians” do) will forever cite the same or similar data in support of their theories. It is a common complaint that anything can be proven with statistics, and in the field of economic debate this seems to be true to a large degree. (I subscribe to the “Austrian” explanation of economic crises not because it fits the data better but because it fits the principles of economics, the laws of economics that allow us to analyse the cycle in the first place. A detailed analysis of Keynesian theories leads to conflicts and mismatches with some key economic principles. This makes this theory much less convincing.)

Paul Krugman, praxeologist?

Any serious discussion of economic matters must ultimately drill down to first principles, to the a priori level. This means that every economist is ultimately forced to use Mises’ method, even someone like the belligerent archtypical Keynesian Paul Krugman, who would not want to be associated with any tenets of the Austrian School. But if we discussed an economic issue with Paul Krugman long and hard enough, and assuming for a moment the honest intention on both sides to get to the heart of the matter, we would ultimately have to arrive at the level of fundamental theory.

In the 1990s, Paul Krugman was known as a free-trade Keynesian. When financier Sir James Goldsmith published his anti-free trade pamphlet “The Trap” in 1994, Krugman criticized it and Krugman correctly pointed out that Sir James failed to grasp even the basics of trade. Appropriately, Krugman referred Goldsmith to Ricardo’s work and the great economists’ essential a priori insights as to the benefits of trade, benefits that must even accrue to allegedly “inferior” (less productive) trading partners (see my earlier point on Ricardo’s theorem). Ricardo, who had then already been dead for 170 years, did not have the better data but the better theory, as Krugman rightly acknowledged.

Because of the very nature of its subject matter – purposeful human action as opposed to natural phenomena – economics can, on the one hand, make incredibly powerful generally valid statements about human action of the kind of “no country can lastingly be a loser in free trade”, or “no minimum wage law can lastingly improve the material position of those on lower income”, which are true regardless of time and space, while, on the other hand, it cannot make the type of statements, in particular specific predictions, that one is used to from the natural sciences, such as “if the price of chocolate goes up by X, demand for chocolate will go down by Y”, or “if the minimum wage is raised by a unemployment will go up by b”, or “if the central bank doubles the money supply, the prices of milk will go up by Z, and the average wage by M.” (To solve these problems, we usually use entrepreneurs and speculators, not economists.)

Those who are disappointed by the latter and do therefore not consider economics “scientific”, unfortunately close their minds to the deep insights that can be derived from proper economic theory correctly applied.

Economic science versus natural science: The fundamental difference

“’All daffodils I have seen have been yellow, so the ones I have still to see will probably also be yellow’; refinements apart, the generalizations of natural science all rest on reasoning of this type, and none of them are certain, in the sense that we can see them to be necessarily true.” (Brand Blanshard, Reason and Analysis, 1962).

Natural scientists observe that A always coincides with B and make inferences from this “coincidence”. In analyzing inanimate objects and instinct-driven non-human animals, this has been a very powerful technique. Why? – Because in the “natural world” there appear to be many regularities and reasonably stable relationships that allow us to make these inferences. Or, to put it differently, the natural world does not know valuing, purposeful behavior, or “free will”. This changes fundamentally when we introduce human action.

Humans appear to be unique in that they consciously act, that is, evaluate a situation, make choices, purposefully interfere with their surroundings, and consciously re-shape part of their environment. At the core of this process is the act of valuation, of preferring one thing to another. None of this is observable in non-human affairs. It is the unique feature of human action, and human action itself (not the consequences of it in the physical world) is the subject matter of economics. As Mises pointed out, one day we may be able to determine which chemical or physical processes cause a person to prefer A to B in a specific situation, but until we have done so there remains an unbridgeable gap between natural phenomena and the phenomena of human action, and they require fundamentally different techniques (this is called methodological dualism). When dealing with humans we have to assume an element of “free will”.

Put in the same or similar situation, two people may value and act differently, and the same person may value and act differently at different times. (This does, at this stage, not even relate to David Ramsey Steele’s points about rationality or consistency of preferences, as raised here.) Furthermore, the complexity of the world we experience as humans means that we can certainly not “step into the same river twice.” If people responded to the rise in chocolate prices one way in the past, they may still respond differently the next time. But all that observation of history (statistics) can do, is ascertain how they acted at a specific time and place. The problem is simply that people are not automatons, billiard balls, light rays, or amoebae. They do not respond simply to stimuli.

This puts the student of human action at a disadvantage to the natural scientist in one respect – namely, that he or she cannot assume the stability of observable relationships in the same way that the natural scientist can – but also provides a fundamental advantage: The scientist is himself or herself a rational human being, and as a social scientist uses human reason to analyze rational human behavior. While the natural scientist remains forever outside the very forces he observes (he or she never has access to those “prime movers” that make billiard balls behave the way they do), the economist (or praxeologist) can relate to what he or she observes in a much deeper way.

It is, I hope, now becoming clearer, why the laws of economics are of the nature they are, that is, a priori, as illustrated before. They reflect the inherent logic of rational behavior and are thus essentially restatements and careful further elaborations of the starting axioms that man prefers one thing to another, that he values and then acts. Notions such as economic good (and therefore marginal utility), time preference (and therefore interest), cost, benefit, profit, loss, are all logically deducted from these axioms.

The a priori in natural sciences

When one raises the issue of the a priori in economics there is usually a lot of push-back from natural scientists and this seems to reflect the harsh treatment the a priori concept had to endure in their discipline in the 20th century at the hand of the philosophy of analysis, of logical positivism and extreme empiricism. A priori concepts had always been indispensable for making sense of things, including natural phenomena, but particularly since David Hume there has also been doubt as to the validity of these concepts and their ability to tell us anything meaningful about reality. This skepticism was taken to new extremes in the 20th century. The question was raised whether the standard tools of abstract human reasoning, such as logic, mathematics, and geometry, that is, the classic a priori disciplines, did even meaningfully correspond to anything in the real world at all. Bertrand Russell seemed to have had this in mind when he said: “I thought of mathematics with reverence and suffered when Wittgenstein led me to regard it as nothing but tautologies.” Or, as Einstein said: “As far as the theorems of mathematics refer to reality, they are not certain, and as far as they are certain, they do not refer to reality.”

A key event behind this new trend seems to have been the discovery (by Bolyai and Lobachevsky) of non-Euclidean geometry, which seemingly knocked the undisputed paragon of a priori thinking of its pedestal. Here was the former superstar of a priori reasoning: entirely man-made, a creation of abstract thinking, yet a powerful and indispensable tool for dealing with the real world: Euclidean geometry. But after a more than 2,000-year unassailable reign it now had to face a newcomer. Today it seems to be widely agreed that Euclidean geometry is still useful for building bridges on earth but that when it comes to analyzing big stuff in space, the new type is much better.

Should this shake our faith in the a priori method in economics? – No, said Mises. First of all, the idea that the tools of human reasoning are often just adequate rather than perfect did not surprise or shock Mises. “There is no such thing as perfection in human knowledge, nor for that matter in any other human achievement. Omniscience is denied to man. The most elaborate theory that seems to satisfy completely our thirst for knowledge may one day be amended or supplanted by a new theory.” (Human Action, Introduction) But importantly, the arrival of a new geometry did not mean that these a priori concepts were just arbitrary or simply the result of convention. We could not redesign a new geometry at will. In fact, to Mises, the history of geometry showed that the human mind was quite capable of developing a priori concepts that did meaningfully respond to the real world (as Euclidean geometry still does, even now that it has lost some of its luster).

And furthermore, the epistemological problem of whether or to what degree the products of abstract human reasoning correspond to the physical universe may trouble the natural scientist, but it is in fact irrelevant for the economist. As we have seen, the economist deals with purposeful human behavior, with human rationality at work (and, again, not even with the physical manifestations of human action, but with the inner logic of human action). He or she applies human reasoning to the work of human reasoning. In the “natural world” there may be no self-evident truths that allow for any necessarily valid deductions. Here, the human mind must always guess. But, to the human mind, rational human action is the ultimate self-evident truth, and here necessarily valid deductions are not only possible, such a priori inferences are the only statements we can make with any certainty.

There is, of course, much more to say about this important topic but I hope that the above provides at least a good basis for further discussion. The last word belongs to Mises:

“…the sciences of human action differ radically from the natural sciences. All authors eager to construct an epistemological system of the sciences of human action according to the pattern of the natural sciences err lamentably.

The real thing which is the subject matter of praxeology, human action, stems from the same source as human reasoning. Action and reason are congeneric and homogeneous; they may even be called two different aspects of the same thing. That reason has the power to make clear through pure ratiocination the essential features of action is a consequence of the fact that action is an offshoot of reason.” (Human Action, Chapter II).