[This article is part 2 of a series of articles and a section taken out of the longer essay “Do You Even Prax, Bro?” Over the span of the following days the corresponding sections will be published until finally the essay is published in full along with the missing  and additional sections added since first publication]


Elswehere I discussed a priori knowledge and how it pertains to Austrian Economics. I discussed how Austrians deduce from the statement “man acts”, plus the performative contradiction involved in attempting to refute it. I also addressed the objections of it not being “scientific,” plus how a priori and deductive knowledge was common prior to the empirical approach.


In the following article, we will be discussing the Austrian view on empirical observation. Typically, it is repeated more frequently as an objection to Austrian methodology compared to the deductive approach that was explained in the previous section, usually from competing schools such as the Neo-classical or Chicago School.  The  grievance is that Austrians reject empirical observation, implying Austrian Economics may be false because they do not test their theories.  Opponents to the Austrian school point to some counter argument that is predicted by a model, which they claim is correct because it is supposedly “empirical.”


Empirical evidence means as much to one economist as to any other, because each one has boundless ability to interpret historical fact into explanations that bolster their own theories. Their valuable methods of demonstrating improper application of models, or in showing that models are built upon unsound and illogical foundations, are precisely what Austrians are being admonished for by their opponents, that is, for “failing” to use empiricism. We must remember though that all economists discuss empirics, and apply theory to the real world. Austrians specifically, however, object to doing the opposite: letting empirical observations inform theory, or pretending it even can. In other words, Austrians do use empirical data to evaluate whether they applied all the relevant aspect of their theories, and whether they may be missing some aspect that needs to be further derived. What they do not do, and what cannot be done, is show that “models are built upon unsound and illogical foundations.” Empirical evidence is mere observation.  Sound economic theory is required to interpret that which you observe. Some empiricists may contend that empirical observation pans out in the real world, since it is tested. Yet, we must remember that it’s not a question of whether it “pans out” or not. It’s a question of theory being utterly untestable empirically. Economics in particular has no way of running controlled experiments on unchanging units of nature, like physics or chemistry do.


As Rothbard states:


“In the sciences of human action [which economics is the study of], on the other hand, it is impossible to test conclusions.  There is no laboratory where facts can be isolated and controlled; the “facts” of human history are complex ones, resultants of many causes. These causes can only be isolated by theory, theory that is necessarily a priori to these historical (including statistical) facts.” [1]


Because controlled experimentation is impossible in economics, any interpretation of real world data must make use of a theory of a set of assumptions, either implicit or explicit. The difference, then, between Austrian economics and “empirical” economics is that the Austrians are explicit in their use of carefully considered axioms as the basis of their analysis, while the “empiricists” are themselves blind to the ad hoc mish-mash of unexamined assumptions that inform their own interpretations.


Austrians are lectured about empirical observation or what theories like those of the Keynesians, Chicagoans, Neo-Classicals are intending to do. Austrians are aware of the status and nature of such propositions, which is exactly why they are incompatible with economics; economics is the logic of human action, human action and behavior cannot be confined or predicted through empiricist observation.  The Austrian School knows socialism won’t work just by logical deduction, where other schools actually think it must be tested first to see if it is really bad or not. The only way to understand human action is to follow the action axiom. Historians, for example, need not follow the action axiom as a theoretical formulation — they have their own methods of studying. The logic of action, honestly, is but one analytical tool to help us expand on the knowledge of human action. The logical categories of action provide us with a clear framework of what is going on, so long as the formulations are occurring in reality at that time. As an empiricist, you will never fail to find what you’re looking for in the data, never fail to validate your hypothesis. If someone finds data that appear to falsify your hypothesis, as he will every single time he wants to, you can never fail to find complicating factors and assign them whatever significance is required to protect your preferred conclusion — not because you are a sloppy thinker, but because there is no consistent way to make this assignment. By relying on formal logic, I can be (and have been) absolutely proven wrong with no recourse to anything. This is not to say man is infallible — but logic is far, far, far more amenable to the discovery of error than empirical observation in the field of economics. Of course, sometimes logic doesn’t pan out because there are more variables than you realized. In which case, one can logically point out those variables and their effects — no need for empirical analysis. Further, the logic is still just as valid when analyzing these same things, onlyceteris paribus. For example, that if demand increases, then price rises (all else being the same) is ironclad — but it doesn’t explain all the factors that might be involved. For that, you need historical analysis, but the theory is still sound.


Think of it as though each ceteris paribus condition was a “force” — gravity is bound to follow an equation, but electricity may also be acting on an object. And so forth.


You can’t observe a variable acting on the outcome and point it out. You could observe that things did not turn out as expected and then infer an unaccounted-for variable exists. But you could not observe a causal relationship. Empirical analysis can only lead one to infer such a relationship if the experiment is controlled — if it is, essentially, ceteris paribus.


There are no tests that can be run in economics. There are no experiments that can be controlled. There are only results, and the only way to explain those results is a combination of logically deduced causal relationships.


And because of this Austrians are radically different from other schools such as the Neo-classicals and the Chicago School. Even if we reach some of the same conclusions, which inevitably occus, the epistemological difference is quite fundamental. There is no reason to pretend otherwise — not even the hope of becoming one of the cool kids on the block.


Many Austrians have addressed the issue of empirics time and again.  For instance, Walter Block, Senior Fellow at the Mises Institute and author of Defending the Undefendable, speaks of a time when his former mentor stated “Austrian economics could benefit greatly from developments in empirical economics […”] in written correspondence.  Walter Block replied:


“I don’t know whether you remember this or not, but when you were my dissertation advisor at Columbia University, my topic was rent control.  I was, presumably, trying to test the usual implications of this law: that it causes rental housing deterioration, reduced investment in residential rental units, shortages, etc.  These were the dependent variables in my regression equations.  The independent variable was presence of rent control.  My observations were cities.  I tried to control for a few other things, such as wealth, income, race, etc.  Most of the time the sign of my rent control variable was the correct one, indicating consistency between my model and the usual economic analysis of rent control (in which both Austrians and mainstreamers fully concur).


But every once in a while I got the wrong sign for this variable, indicating, if you believed my results, that rent control actually improved the housing situation.  On these occasions did you brag to your colleagues that you had this young genius, Block, on your hands, who was in the process of overturning everything we all knew about rent control?  Did you urge me to send my earth shaking results to the AER for publication?  To ask these questions is to answer them: you did no such thing. Very much to the contrary you said, “Block, go out and do this again until you get it right!” That is what you explicitly said to me. In contrast, what I heard was: “Block, you dummy, go out and do this again until you get it right!” Of course, you were far too polite and supportive to actually say this to me, but I could tell that this was what you were thinking; at least, this is the way I felt, at the time.


So, what was “testing” what? Were my equations really testing the usual supply and demand analysis of rent control, or was the usual supply and demand analysis of rent control testing my econometrics? Obviously, the latter was the case.”[2]


You just aren’t going to be able to find an empirical observation, or collection thereof, which vindicates one theory and fails to vindicate another.  There will always be more than enough relevant facts to dig up and point to as the most relevant causal factor, and there’s no way — without proper theory and therefore ability to determine causality — to resolve this quantitative dispute over which of the known facts was most important in bringing about the observed effect. You think you can prove monetary policy is the cure for recessions; I can just as easily “prove” the recession was caused by ice cream sales or volcanic activity. Empirical analysis can only demonstrate the improper or incomplete application of a model — whether that model is “empirical” or not. The fact that, for instance, unemployment may have gone up in situation A, despite an increase in the minimum wage (which serves to prohibit cheap labor) can never invalidate the analysis involved in wage controls — it can only indicate that other factors were also at work (some form of growth that would have raised wages in spite of the intervention, most likely). Even it were possible to do experimentation — or even if interpretation of events in the past weren’t guaranteed to yield a validation of almost any possible economic hypothesis and falsifications of all others — these things are as unnecessary for establishing the relationship between prices and the money supply as they are for evaluating the assertion “no two straight lines can enclose a space.” I doubt anyone has ever set out to discover whether this is true by observing the real world, and any attempt to do so would be utterly silly.


In a discussion of methodology, one may suggest that we need both theory and empirical observation. For instance,  if you only have data, or you only have theory, there is a good chance that you made a mistake at some point. If you have both data and theory, the odds of both showing the same error are very small. No,  that isn’t true at all, in economics. There’s a reason you’ve never seen an economist look at some data and decide he has to abandon a theoretical position. The data is no good for that, because there are always plenty of complicating factors. If the theoretical prediction didn’t work out, maybe there was a complicating factor you didn’t account for… ah yes, there it is. And if it does work out, even if the theory was wrong, which is entirely possible depending on the theoretical proposition, the theory is confirmed. . If the evidence doesn’t seem to line up with a deduction you need to check you deduction then if it still checks look for outside influences that you have no control over.


The Austrian point is not one against the scientific method, but against using it in economics because it is virtually impossible to run a valid experiment, that is, control for all the variablesIf the theory really is incomplete, there’s a question that hasn’t been answered yet – the data can help you identify those questions, decide in which direction to further develop theory. But if any of your existing conclusions are wrong, they’re not incomplete – they’re just wrong.


There’s almost nowhere mainstream economists haven’t fallen short, but at least when it comes to the Austrian school it’s possible for such errors to be discovered and corrected by others. What does distinguish the Austrians from everybody else? We use logic, and we use it exclusively, so that if one of us is wrong another of us can point out the error and we can easily be corrected. Everybody else uses empirical observation, and unsurprisingly, none of them ever fail to find data that corroborate their theories. This makes the Austrian school unique, because as of now it is the only school of thought where empiricism is not relied upon, and where empiricism’s problem — the inability to make economic laboratories where complicating factors can be isolated out so the subject at hand can be examined in unadulterated purity — is essentially irrelevant. When it comes to economics, I can ask almost any statement and ask that it either be validated or falsified by observation, and neither of those efforts can ever possibly fail. We can observe all day long that event B was preceded by A, but only logic can establish causality, not by observing relations that are temporal.


In proving this we turn to Hoppe:


“ […] A proposition regarding the relationship between economic events can never be validated once and for all with certainty. Instead, it is forever subject to the outcome of contingent, future experiences. Such experience might confirm the hypothesis. But this would not prove the hypothesis to be true, since the economic proposition would have used general terms (in philosophical terminology: universals) in its description of the related events, and thus would apply to an indefinite number of cases or instances, thereby always leaving room for possibly falsifying future experiences. All a confirmation would prove is that the hypothesis had not yet turned out wrong. On the other hand, the experience might falsify the hypothesis. This would surely prove that something was wrong with the hypothesis as it stood. But it would not prove that the hypothesized relationship between the specified events could never be observed. It would merely show that considering and controlling in one’s observations only what up to now had been actually accounted for and controlled, the relationship had not yet shown up. It cannot be ruled out, however, that it might show up as soon as some other circumstances have been controlled.“[3]


Therefore, again, if the result fits the prediction, we figure that it must have been validated even if other explanations are available for the same result, as there always will be. If it doesn’t fit, we can always find an excuse, some complicating factor (it always comes down to an assessment of relevance, which is a quantitative question that can never finally be answered). If all the pertinent factual questions are agreed upon, but if our theories are different, we’ll end up assigning a different relevance to each of those facts and we will  have no way to resolve such disputes. Thus, in the end no observation can ever validate one theory or falsify another to anyone’s satisfaction, and no school of thought that teaches otherwise has any real way to correct errors. The problem with empirical observation as a tool for economic analysis is that you and I can look at the same situation with the same agreed-upon set of facts and never be able to resolve a dispute about what caused it, except by discovering which theoretical analysis, if any, is correct.


Empricism works in the physical sciences because we can do experiments, but we can’t do experiments with things like the economy: There’s no control group for the economy. Economic causation cannot be shown through empiricism, if you even want to call historical study of correlations empiricism. Empirical economists are more akin to the astrologers of old, finding patterns that fit the king’s wishes. 



Jeff Peterson II

We the Individuals




[1] In Defense of Extreme Apriorism By Murray N. Rothbard


[2] Is Austrian Economics a Cult?  By Walter Block


[3] Economic Science and the Austrian Method by Hans-Hermann Hoppe; Chapter 2 On Praxeology and the Praxeological Foundation of Epistemology, Section 1