[This article is part 4 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. For previous articles on Austrian epistemology/methodology, see herehere, and here.]

 

 

“Statistics is a method for the presentation of historical facts concerning prices and other relevant data of human action.  It is not economics and cannot produce economic theorems and theories.”-Ludwig von Mises[1]

 

How many of you reading have been in a discussion on economics where the other individual throws out statistics to validate his or her claim? Depending on the topic — for example, minimum wage — this happens many times. Since I have studied Austrian Economics there is a saying I have become fond of regarding statistics which goes: Austrians have strong words for those who use statistical analysis without explaining qualitative, cause-effect explanations. So, what is the Austrian view on statistics?  To be clear, we don’t “oppose” anything at all, as Austrians. Austrians say that statistical analysis can’t yield correct predictions except by accident, because no theoretical model can generate valid predictions from historical data, because there are no quantitative constants in human action. Which means this is specific to economics, so while it is appropriate in the natural sciences where causal factors can be isolated in laboratory conditions, the actions of human beings are far too complex for “numerical” treatment of them as passive non-adaptive subjects. Instead, one should isolate the logical processes of human action.

 

All you can hope to establish with statistical analysis is correlation, and in the real world even that is a great feat in the face of so many complicating factors. But even if a correlation is discovered consistently enough to imply a likely causal relationship, the direction of the causal arrow still needs to be determined, and each of its endpoints. That is, you can’t hope to discover via this analysis whether a later event was caused by the earlier one, or whether they are both caused by some, perhaps unknown third factor.

 

This doesn’t even begin to expose the problem. Collecting all the relevant data is indeed a difficult undertaking. But then you face the task of assigning the appropriate relevance to each fact, and this is more than difficult. You will necessarily do this according to whatever model you’re using to analyze it, and this is why the data are ultimately useless at informing theory. Theory precedes understanding, understanding of “the data,” that is. The data cannot precede theory, nor can understanding be derived from “the data” sans theory. To do either is to work in the shadow of assumptions and biases you have not acknowledged and which will lead you blind.

 

 

Austrian economics works on first principles, cause and effect. You give me statistics, and I can give one or more causal factors which would impact those statistics. The quantitative relevance of each factor is unknowable, however. For instance, if you were to show me that unemployment has gone up, I might talk about minimum wage laws, payroll taxes, regime uncertainty, and so on. Your response may include outsourcing, cash hoarding, etc. However, there’s nothing in the statistics to let this dispute be resolved. Only theory can establish a causal link between two (real or hypothetical) economic observations. However, more importantly, the only way to adopt these points meaningfully is to use deductive axiom(s) that explain why those factors matter. Arguments against Austrian Economics have to attack the logical basis behind its corollaries, rather than attacking straw man arguments and misrepresentations.

 

 

In a similar vein, you might say “Austrian Fail!  Minimum wage went up and unemployment went down!” In which case you’ve misunderstood Austrian theory:  Austrians don’t claim increasing the minimum wage will cause unemployment to go up, only that increasing minimum wage will cause unemployment to be higher than it would otherwise have been, but that employment levels are still affected by other factors.

 

Austrians would argue against increasing the minimum wage because praxeology tells them that future unemployment with a higher minimum wage will be greater than future unemployment with a lower minimum wage, not because they predict future unemployment rates to rise as a result of the increase, other things being equal. You can never observe both of these states because only one will occur in reality.  We must rely on the deductive logic of action to figure out the proper course.

 

 

Despite what has been explained above, there are those who would argue that Austrians have placed themselves in an unchallengeable position, that they have rigged the game to win. Sure, this is a valid concern. Let’s look at the minimum wage example from above. Let’s say you are a supporter of a competing economic school of thought or a “newbie” to economics in general and we are discussing minimum wage. If the unemployment rate goes up after a minimum wage increase, to you your Austrian counterpart can claim victory. On the other hand, if the unemployment rate decreases afterward, the Austrian can still claim victory because Austrians can assert that the unemployment rate would be lower still, had the minimum wage not been increased. No matter what the end result is, the answer seems rigged to assert that something would have been better applying Austrian principles, or if an Austrian understanding of the problem had been applied, that any bad result would have been worse otherwise. Either way, the answer comes off to never be wrong and unchallengeable, right?

 

 

Not really. One Austrian might say, “See? Just as predicted!” But a more thoughtful, and more nearly correct Austrian might say, “Not so fast, there are other influences on unemployment — we also have this increase in the capital gains tax and this onerous regulation making it more difficult and/or expensive to enter into or stay in business. They all applied pressure in the same direction and we can’t say a priori which of them had the most important influence, only that it would be lower than it is now if they had not also raised the minimum wage.”

 

Austrian economic propositions can’t be validated or falsified by observation. Note that the proposition is only that given an increase in the minimum wage, unemployment will be higher than it would otherwise have been, and we can’t make an observation and go, “Ah ha! unemployment is not higher than it would otherwise have been!” This is because we don’t know what it would have been.

 

So, you can’t test them empirically, because they aren’t hypothetical propositions. They are deductions from the action axiom, so if they’re wrong, you have to find the flaw either in the premise (the action axiom) or in the deductive logic used to arrive at the conclusion.

 

 

And this is far more amenable to error correction than the alternative. No non-Austrian ever has to go back and say “I was wrong — the data have disproved my theory” because he can always find data to vindicate his theory. Whereas I have to admit I was wrong, like, once a week when I make an assertion and another Austrian points out the error in my logic — because it doesn’t matter what data I can dig up. If my logic is wrong, then at best my proposition is “not true” — it may not be false either, but I haven’t demonstrated that I’m correct, and no Austrian will take my claim seriously unless I can make a logically valid argument for it.

 

One should look at events as though they needed to be explained by a theory, rather than be used to test a theory.

 

You don’t test the theory, anymore than you can test that a triangle has three sides and the same internal angle sum (180 degrees) regardless of how you arrange it. The theory explains the data — it is the confluence of many separately derived effects together that produce a result.

 

Still not convinced? If not, it is no surprise. As with empirical observation, statistics is seen as the norm. Lets go over a few possible objections to what we have discussed so far.

 

 

  • “If statistics do not work for real-world scenarios, as they cannot assimilate the complexities of any given situation, how can anything be inferred from something as non-specific and as varied as ‘human action’? Isn’t the Austrian view on statistics and dismissing it out of hand because the real world is “too complex” tantamount to dogma?”

 

How can anything be said about falling objects? There are feathers falling from 5 feet and moons “falling” perpetually. How can anything be inferred from something as non-specific and as varied as vectors and scalars? Fundamental physics has no problem doing it. There’s a difference between the logic of something and the application of that something’s logic to generate more specific predictions.

 

For example, the logic of praxeology generates the entire body of economics and economics informs all of the more specific methods that entrepreneurs use [which may include econometrics/statistics].

 

The difference is that we are talking about the form of human action. For instance, regardless of who you are, you still must select a goal and use scarce means to obtain it. There is no alternative to this. What your ends are is a subjective preference, and that there is a one is undeniable (denying it would imply a preference to deny it to obtain some end goal).

 

You can use statistics to bear out human preferences, at least as much as you can infer the preferences of any individual. You can say such things as 10% of the population doesn’t like ice cream, and that can be very accurate. Not to mention what percentage is lactose intolerant, whether they “like” it or not. You can do studies and find out what percentages like each flavor, or dislike each flavor. Maybe more than 50% don’t like pistachio flavor, while for 5% it’s their favorite, and still 99% like vanilla and don’t mind eating vanilla if that’s what’s being offered (there’s a reason sometimes bland but always very standard things are called “vanilla”).

 

But this does not mean you can predict the preference of a given individual at a given time (even those who say pistachio is their favorite will sometimes eat a different flavor just because). This demonstrates class probability vs case probability. Class probability can be measured accurately. Case probability is always uncertain and a gamble.

 

Statistics, in this sense of class probability, can be very useful for business and entrepreneurship (two different things, by the way), but, as we will find out in another section, that does not make it economics.

 

The problem with the use of statistics in economics has nothing, or almost nothing, to do with complexity. I mean, complexity is often why you use statistics in the first place. And Austrian economists do use statistics, just not when they’re “doing economics.” Economists like to comment on history and current events, and when telling those stories it’s often useful to include some quantitative information about recent relevant trends. So they’re useful for the narrative, but they can’t inform the theory.

 

Lastly, the action axiom is both logically true and applies to the real world. It is a necessary truth tied to reality. It is important to recognize that Mises and Rothbard [2] (correctly) stressed the nature of methodological dualism in that the realm of our inquiry is the logical derivation of the things we know to be true through self-reflection and verbally building up the chain of reasoning where each step along the way is very important. So in that sense, being “dogmatic” is the correct and thoughtful approach.

 

So the “theory” that the Austrian school is interested in is best characterized as a “causal-realist” approach. As Hoppe says, economic theory may not tell us “much” but what it DOES tell us is of utmost importances since we’re talking about apodictically true knowledge about the real world and not contingent hypotheticals that could be true today but false tomorrow and vice versa.

 

 

  • “Sure, collecting all the relevant data is indeed a difficult undertaking, but then again, so is understanding “human action,” which by its nature is extremely situation- and individual-specific. By comparison it is much easier to calculate statistics like average debt per household, unemployment, etc.”

 

Why would a scientist want to take the much easier route if their goal is understanding? And how do those calculations that aren’t predictable constants further that goal? The problem has nothing to do with collecting the data, and nobody’s saying the statistics or the data is useless — they just can’t inform economic theory. Theory says “maximum price controls promote shortages” without reference to any particular good, market segment, production technique or anything else. The historical data can tell you there was a shortage of widgets, and that it was preceded by a price control on one of the raw materials required to produce widgets, so with the aid of economics you can tell a cohesive historical narrative, but the data didn’t tell you anything you didn’t already know, or couldn’t already deduce, about economics — nor could it ever.

 

Speculators, entrepreneurs, investors, historians, policy analysts… they all make good use of the data and various ways of looking at it including but not limited to statistics… but all of them are applying economic theory to interpret the data — they can’t get valid theoretical economic propositions out of the data.

 

 

  • “Is there something wrong with including practical, real world examples too?”

 

Including them in what? History books? Of course not. Policy analysis? Obviously not. Economic theory? For reasons other than mere illustration, yes. For reasons we have gone over.

 

 

  • “What informs us on how humans will generally act anyway?”

 

Psychology, anthropology, all kinds of things. Economics is only concerned with the fact that humans do act, and the logical implications of that fact. The only aspect of human behavior that economics concerns itself with is that we have subjectively chosen ends and use reason to choose the best means to those ends. An economist doesn’t need to know I collect comic books, let alone why I collect them or how they make me feel. The whole body of Austrian economic theory is deduced from that proposition, that we have ends and use means to realize them. Its statements are no more hypothetical than the Pythagorean theorem, and no more empirically falsifiable, and for the same reasons. If any of them are wrong, you have to find the flaw in the logic. The good news is that unlike the other guys, if, while using the Austrian method I make an incorrect economic assertion, somebody is usually able to show me my error to my own satisfaction which can never happen to any economist who admits empirical data into his theory, because he can never fail to find what he’s looking for in the effectively infinite set of numbers.

 

  • “So nothing can be proven wrong?”

 

Hypothetical propositions can be disproven empirically. Austrian propositions can only be disproven logically.

 

 

  • “People will differ on how statistics are collected and what they mean, but they will also disagree on first principles and deductive logic. For a theory to be valid, it must be falsifiable. “

 

So what if they do disagree? If they can deny the action axiom without proving themselves wrong in the act, they win. If they claim deductive logic is an invalid way to find truth, they’ve already lost so hard there is little to say to them. Hypotheses need to be empirically falsifiable. A priori theoretical propositions need to be logically proven, and fail when they are logically disproven or when any error is found in the presented proof.

 

  • “How does one judge whether or not the course taken was correct?”

 

If by “course taken” you mean “conclusions deduced,” the chain of reasoning is sometimes long, but not usually very complicated — if you deduce correctly from a true premise, your conclusion will also be true. If you want to find a conclusion to be untrue, all you have to do is find the error in the logic used to arrive at it or the original premise. Remember the scope of inquiry in economics isn’t “reality.” The data are useful, and even useful to economists when doing something besides developing theory, such as writing on economic history or current events. But there is nothing you could ever find in the data that could legitimately inform economic theory.

 

Statistics are simply data points. They aren’t very good ones, either, as any comparison of them relies on the assumption that they have not been “massaged” to fit a narrative, or that the situations being compared do not differ. In other words, if I have a situation where unemployment rises, this is a fact to be explained, not a fact that proves anything.

 

Most other schools of thought fall into a correlation=causation fallacy here, or a post hoc ergo propter hoc fallacy. The Austrian school, in contrast, starts with deducing the laws of economics logically, and then proceeding to explain events. Only with the natural sciences, where controlled experiments exist and are repeatable (and assumptions of constancy are introduced), can hypothetical natural laws be deduced from data. There is no “control set” with human beings; only deduction from true axioms, like the action axiom, can lead to knowledge of human behavior. You can’t say Y happened after X, so therefore because of it, but that’s exactly what every other school of thought does, while we sit here explaining the actual, logical causal relationships.

 

 

Jeff Peterson II

We the Individuals

 

 

[1] Human Action by Ludwig von Mises

XVI. PRICES, 5. Logical Catallactics Versus Mathematical Catallactics

https://mises.org/humanaction/chap16sec5.asp

[2] Theory and History: An Interpretation of Social and Economic Evolution by Luwig von Mises