[I tried to post this as a comment, but the blog wouldn’t let me (too long?) so I’m submitting it as a new post.]

Thanks, Lee, for your kind remarks. My debate with Robert Taylor was on New Year’s Day, so naturally I was in recovery mode, and even slower-witted than usual. I failed to make some key points.

I don’t dispute that it’s possible to formulate a useful deductive system in which conclusions can be spun out from axioms. Euclidean geometry is an obvious example. The theorems follow from the axioms. Furthermore, we feel intuitively that the theorems are true, that is, that they tell us something about the physical structure of objects in space. Yet, as we now know, Euclidean geometry is not literally true of the physical world. In other words, if we take Taylor’s example of the Pythagorean theorem, the conclusion of this theorem is not true of real triangles in space (it is slightly off, because of the curvature of space).

Yet the Pythagorean theorem is true in two senses: 1. The theorem does follow from the axioms, and is therefore true in Euclid’s world, much as ‘elves have a characteristic smell, different from that of hobbits’ is true in Tolkien’s world; 2. The theorem applies to real-world triangles of manageable size with very close approximation. It’s therefore fine for building bridges.

Now, we can assume that ‘Man acts’ is true, and derive from this certain propositions (which then help to define what we mean by that assertion). We can logically derive from these propositions some conclusions about the way humans will act. I don’t deny that we can do this, and that the results will sometimes be illuminating. For example, if we assume that business firms always do what they perceive as maximizing their financial returns, we can make predictions about the behavior of firms which will very often be true (and where not true may draw our attention to some special circumstance, which may be helpful to our analysis of what is going on in the behavior of firms). Furthermore, the assumption that business firms always do what they perceive as maximizing their financial returns is not arbitrary, since we know that competition favors the survival of firms which do better at maximizing their returns. So, I agree that there’s a lot of mileage in the view that important steps in economic reasoning may be made by applying a Pure Logic of Choice. And, of course, virtually all economists would agree with this, and many of them, never having met a Misesian, would wonder why I am saying anything so trite.

However, Mises and Rothbard maintain that all of economic theory is derived by deduction from axioms. In other words all of economic theory is derived from self-evident axioms about human action, and not in the least from empirical observation!

Here we should note the historical fact that Rothbard was a huge influence on the libertarian movement. To my mind, as well as being an entirely lovable person, for I met him several times and had a number of long conversations with him, he was a wonderful pamphleteer and propagandist. He was no great shakes as an economist, and an abysmal thinker about economic epistemology. Despite his shortcomings, he was a huge improvement over the very poor philosophizing of Ayn Rand, which threatened to engulf libertarianism in the 1960s and 1970s. A lot of what Rand said was entirely correct, but totally unoriginal with her, and this ninety-five percent of her output was simply the common heritage of classical liberalism since Locke’s Second Treatise. Whenever Rand came up with an original idea, she was not only wrong but most of the time almost unbelievably sloppy in her reasoning. Rothbard was a scholar who knew quite a bit about the history of liberal thought. He had read and understood, not only Locke’s Second Treatise, the fountainhead of libertarianism, not only Mill and Spencer, but numerous other thinkers, some of them still under-examined, like Bastiat, Molinari, La Boettie, Lysander Spooner, and Franz Oppenheimer. Rothbard had his limitations but he would never blunder as badly as Rand, who was essentially a creature of Hollywood. Rothbard helped a lot of people transition out of Randism into something more defensible, something with more potential for serious elaboration.

Today this ideological background is less important. Rothbard’s following has largely evaporated. The fact that he quit the libertarian movement in the last few years of his life has something to do with it. Today the issue of ‘praxeology’ is hardly ever raised. Economists arguing for the free market simply don’t address it. It still attracts a few enthusiasts like Robert Taylor and animates some of the output of the Mises Institute. But, for example, if you take the work of a capable and effective popularizer of the economic case for the free market like John Lott, you don’t find any appeals to aprioristic reasoning. It’s all empirically based, this is what you expect, and it would be a distraction for anyone to suggest that it all follows from ‘Man acts’.

Mises was a Kantian. He believed that there are synthetic a priori propositions, and that the essential truths about purposive action were among these propositions. I think there are no synthetic a priori propositions. Like Hume and the logical empiricists, I think that what is analytic is a priori and what is a priori is analytic. But more narrowly, though I’m always prepared to be surprised, I just haven’t seen a convincing candidate for a synthetic a priori proposition.

If a statement tells you about the world, it’s an empirical claim and can’t be deduced from self-evident axioms. If something’s deduced from self-evident axioms, it doesn’t tell you anything about the world (except that part of the world which is the theory in question), though it may be useful if you can make the assumption that the world complies with the axioms. I am entirely prepared to accept that we can construct a body of praxeological theory from axioms and that this might include a great deal of economic theory. But in applying this theory to the world in order to draw conclusions about the world, we would be taking a step beyond the pure theory, and this would make our conclusions fallible. The empirical is always fallible in a way that deduction isn’t.

A point I didn’t bring out in my conversation with Robert Taylor is that even if we had a complete body of praxeological theory which we then set about applying to economic life, there would be the question of whether we had applied it accurately. This is more serious than you might think. Take the example of identifying whether something is money. (Jeff Hummel made this point and others like it back in the 1970s, at a time when we were all much more preoccupied with Mises-Rothbardism than anyone is today.)

Although there are strong tendencies for one good to emerge as the single monetary unit, these tendencies do not fully carry through. Today we know that bank deposits, coins and bills, gold and silver, and bitcoin, all function as money. Other entities, like commercial bills, could also be included. The central notion of money is that it is a good which you acquire because it will enable you to exchange for other goods. It follows that goods can have more or less moneyness. Diamonds have more moneyness than machine tools. Once a good has a lot of moneyness, it automatically tends to accumulate more moneyness, but in practice this is not going to culminate in just one good having all the moneyness.

The point of this for praxeological reasoning is that identifying and measuring what is actually functioning as money can be a tricky empirical exercise. So even with our complete praxeological theory, we would still have to do a lot of empirical work to explain, say, what happened to the US money supply in 2008. There would still be empirical issues about whether this or that asset was functioning as money.

However, I do not think we have, or are going to have that complete praxeological theory to start with. For example, praxeology holds that if someone prefers a to b and b to c, they will prefer a to c. This certainly looks convincing as a likely empirical generalization, but is it apodictically true? Obviously not. We might say that someone who prefers a to b and b to c but chooses c when confronted by a choice between a and c is somehow muddled. Yes, but people can be muddled. Even if we never observed a case of someone being muddled, we would still have to say that the very idea of someone being muddled is conceivable, and so the claim that no one is ever muddled would be an empirical claim. But it is false. People actually are sometimes muddled. Psychology experiments have documented that some kinds of muddle are actually quite prevalent. (I have elsewhere argued against the usual conclusion from this fact, the conclusion that people are often irrational. I maintain that people are never irrational—here Mises was right—though they often make mistakes.)

Exhibiting non-transitive preferences (preferring c to a) is just one example; there are many others, such as counting sunk costs. How does the praxeologist handle cases like these? He can accept them as allowed within his theory of human action, or he can outlaw them as contrary to the basic conception of action. One way to do the latter is to say that if someone preferred a to b and b to c, and was then observed to pick c over a, he must have changed his preferences. But then, if economics is to make empirical predictions about the world, it becomes useless, since any false prediction we make can be accommodated by saying that people’s preferences unaccountably changed during the real-life process we were analyzing. In order to keep economics as capable of making falsifiable predictions, we therefore have to add the premiss (not a fundamental component of praxeology) that people’s preferences are stable. Now we have something empirically falsifiable, and therefore empirically useful, but we have introduced something not derivable from ‘Man acts’.

So what if we take the other route and say that we allow people to have non-transitive preferences, and generally to be muddled? Then we can’t get the most elementary market mechanisms started. We will be unable even to get to the downward-sloping demand curve. Not much, if anything at all, can be spun out from the action axiom if we allow that at any point people can be muddled.

If we turn some of the aprioristic claims into empirical generalizations, then we can make progress. We can say, for instance, as a first stab at it, that people are not muddled all the time, and when dealing with a group of people some of the muddles will cancel out, and if the consequences of the muddles become too painful people will pay more attention and reduce their muddles. We can also observe, as Becker did, that since people’s incomes are limited, the downward sloping demand curve is automatically generated as highly probable for any given good (for example, it’s just not possible for someone to adopt the rule to buy more at higher prices for all the goods they buy; their limited income won’t let them do it). So even if individuals’ buying behavior were totally random (I admit that’s impossible for other reasons) we would still observe downward-sloping demand curves in the vast majority of cases, probably all. The downward sloping demand curve, which we couldn’t derive by deduction from actions, starts to look pretty good as an empirical generalization. Is it a law? Yes, or at least it’s a putative law, just like a putative law in physics or chemistry, provided we specify the circumstances of its application stringently enough.

Much more could be said at this point, but to me one of the most interesting aspects of market behavior is that markets behave much more in accordance with textbook models than they ‘ought’ to. We know that real human beings have a tendency to count sunk costs, yet industrial investment goes on as if they didn’t. A study was done of drivers’ responses to different and changing gas (for Brits: petrol) prices at different gas stations. Their behavior was observed and they were questioned as to what they knew about prices at different gas stations. It was found that they knew very little about prices at different stations, yet their actual behavior was quite close to what we would expect if they had had a perfect knowledge of the prices at all the local stations. Of course, we can come up with explanations for these phenomena, and these explanations tell us a lot about the way markets work. But we have come a long way from the Misesian attempt to move from ‘Man acts’ by strict deduction to ‘Consumers buy from the cheapest seller’.

Here are my answers to Lee’s five questions

1. I don’t believe any significant number of economists will dispute that the minimum wage will increase unemployment, but they may still favor a minimum wage. Remember, that a minimum wage will both increase unemployment and increase the wages of those workers who keep their jobs or get new jobs. Some workers do benefit directly from a minimum wage. I have actually seen discussions where economists argue that if the increase in the wages of workers who keep their jobs exceeds the total wages lost by those who are made unemployed, this makes the case for the minimum wage! An economist (a leftist ideologue who happens to be an economist) may also take the view that it’s better for a worker to be on welfare than to be hired in a very low-wage job.

Common political views of the minimum wage are illustrated in the recent report of the non-partisan Congressional Budget Office on Obama’s proposed minimum wage increase. The CBO said the proposed increase would rescue 900,000 people from poverty, and would also cause an increase in unemployment somewhere between “slight” and one million, settling for 500,000 as a probable figure. This CBO analysis was taken by the vast majority of supporters of the increase as complete vindication of the measure: obviously 900,000 is bigger than 500,000! (The White House economic advisor objected and said there would be no increase in unemployment! But that’s the kind of nonsense you expect from someone in that position. White House strategists know that if anywhere there occurs an official endorsement of the claim that any of the administration’s measures will result in any unemployment whatsoever, this meme could mutuate into something fearsome that would come and eat them all up.)

To me the biggest tragedy of the minimum wage is that it takes the least productive and therefore most vulnerable workers and ensures that they will never get on-the-job training which would be one of their likely roads to improvement. Along with welfare and the public schools, it systematically creates a vast class of unproductive criminals. It also ensures that the snow doesn’t get cleared in the Loop every winter, that no one buses tables in diners any more, and that no one cleans your windshield or your tires when you stop for gas.

2. They didn’t necessarily react this way because they had been convinced by praxeological reasoning. It’s characteristic of all sciences that most scientists’ initial response to a challenge is to assume that the currently prevailing view is correct and a new challenge is pretty sure to be flawed (and a lot of the time they’re right; that’s why it’s the currently prevailing view). And, after all, you can give a loose, intuitive explanation of why a minimum wage will increase unemployment, and anyone can see the point of this, without accepting that it’s a matter of apodictic certainty. Remember, it’s not just the minimum wage. This is just one example of a price floor. Price floors have been observed for many different goods. I doubt that anyone would object to the statement that if you announce a minimum price for cement, above the market price, and enforce it, less cement will be purchased.

Some further comments on empirical challenges to the theory that minimum wage increases raise unemployment. 1. From what I’ve seen, the economists who challenge it usually say something like ‘Modest minimum wage increases will rescue people from poverty and will cause negligible increases in unemployment’. The implication is that negligible increases don’t matter. But the consequences for those who are hit are unusually severe and long lasting. A couple of thousand additional teenagers thrust into a life of drug dealing and gang banging is not negligible from where I’m sitting. For a certain percentage of them it will actually be a death sentence. We’re talking about marginal-quality workers, people whom it’s barely worth hiring. 2. Elasticity is always greater given more time, so you would need to look at a long enough time period to see the full effects. 3. The rise in unemployment might not show up in the statistics for total employment. Suppose McDonald’s hires fewer blacks from low-income backgrounds and replaces them with more middle-class students. Why would this happen? Because the middle-class students, many of whom can take the job or leave it, are induced to apply by the higher wage, and they on average possess positive attributes valued by the franchisee, who is, after all, paying more; the franchisee will be able to reject blacks from devastated neighborhoods he might have formerly taken a chance on. The middle-class students will on average be distinctly better at the job; they will find it less of a struggle to show up punctually and to be gracious to customers. If you’re paying $10 an hour, and you can get someone worth $10 an hour, you’ll go for it, even if you would have preferred to do what you’re no longer permitted to do: pay $8 an hour to someone worth $8 an hour.

3. Yes, just as physics is an empirical science and it’s reasonable to speak of laws of physics.

4. I don’t suppose they could ever be sure, but it doesn’t really arise since they usually do find clear enough results in this area. Most studies of the minimum wage look at things like differences between neighboring states with different state minimum wages. Actually, it’s remarkable to me that they can find anything at all, given how small these differences are. You might expect the outcomes to be swamped by market noise, but apparently they aren’t.

In cases where it’s not feasible to observe the effects, you would exercise your ingenuity to find more roundabout ways of testing the theory.

5. No, I don’t think so. We must distinguish between gut instincts about how the economy works and apodictic reasoning. People working in all the sciences are guided in what to look for by intuitive models which lack any formal standing. These intuitive models suggest; they don’t validate.