The Austrian school of economics is in a permanent “We told you so” mode these days. In this very readable post, Roderick Long gives the Austrian economist viewpoint of the current financial meltdown. After offering a detailed explanation of why it is regulation that got us into this mess (most of his points I agree with, some I am skeptical), he asks the million dollar question:
Okay, some will say, maybe it was government, not laissez-faire, that got us into the mess; but now that we’re in it, don’t we need government to get us out?
His answer is that the government cannot get us out.
There’s just not much the government can do that will help (apart from repealing the laws, regulations, and subsidies that first created and then perpetuate the mess – but that would be less a doing than a ceasing-to-do, and anyway given the incentives acting on government decision-makers there’s no realistic chance of that happening). The bailout is just diverting resources from the productive poor and middle-class to the failed rich, which doesn’t seem like a very good idea one either ethical or economic grounds. The only good effect such a bailout could possibly have (at least if you prefer costly boondoggles without piles of dead bodies to costly boondoggles with them) is if it convinced the warmongers that they just can’t afford a global war on terror right now – but there’s no sign that they’re being convinced of anything of the sort.
So what should the government do? Nothing, is his answer.
I don’t know if I agree with him. Even if he is right on the economics, there are political consequences to doing nothing, and not all of them are good for freedom or economic health. Furthermore, not beng a trained economist, I cannot fully measure the accuracy of his claims or pass a judgement on whether the Austrian approach to money is superior to, say, that of the more mainstream Chicago school of economics. The fact that the Austrian economists have — unlike everyone else — always used verbal, imprecise and non-mathematical arguments, does not exactly inspire scientific confidence.
However, I do recommend that you read the whole thing.
Thanks for the plug. But I’m obviously going to disagree with the association of “verbal” with “imprecise” and of “mathematical” with “scientific.” Whether mathematics is appropriate to a subject matter depends on the subject matter; math is quite appropriate to the subject-matter of physics, say, because there are quantitative constants in physics (while explanations in terms of agents’ beliefs and desires would not be appropriate in physics, since the entities that physics deals with don’t have beliefs and desires). By contrast, there are no quantitative constants in the subjects that economics deals with; instead, the constants are found in the structure of belief-and-desire explanations. Hence on the Austrian view it is unscientific and imprecise to apply mainstream mathematical methods to economic subject-matter (just as it would be unscientific to explain the behaviour of neutrinos using biological or psychological categories), while explanations in terms of beliefs and desires are fully scientific and fully precise.
I disagree there. First of all, while economics is closely connected with complex human beliefs and desires, the predictions it is supposed to make (credit, the markets, profits, money, growth) concern exact numerical quantities. So it is wrong to say that there are no quantitative constants in the subject. Moreover, anything that can be said, can be said mathematically. Even the purely verbal arguments in your post at Art of the Possible can be rephrased mathematically. Math does not imply precise equations, there are several ways to deal with behavorial aspects.
There are several advantages to writing any scientific theory, from the behavorial to the psychological, in as close to a mathematical manner as possible. Even if the subject matter involves entities with beliefs and desires, the consequences of these beliefs and desires are quantities, whether or not they can be modelled by a nice function. One, such a practice enhances the ability of the theory to make predictions. Not only does it improve the structure of logical reasoning, it also allows modelling by functions, which, while they may not accurately reflect the real world (because of the complexity of human behavior) nonetheless allow a glimpse into what one may expect. Two, such a rewriting, in the case of Austrian economics, would allow a better comparison with other theories, such as the Chicago school. Anything that can be said, can be said mathematically; the only problem is that the functions are often too complex, thus sometimes giving no tangible benefit to doing so. Ultimately though, as computing ability progresses and we are able to deal with more complicated functions without using techniques that cost accuracy, everything, from behavorial economics to psychology will move to a more mathematical basis, and so must Austrian economics. In many cases, the mathematical tools are already there in a form suitable for making quantitative predictions.
You might also wish to go through this article by Bryan Caplan where he says that Austrian economists have overstated their differences on some issues with other schools of thought. A prime reason, I think, is their fixation on verbal reasoning.
First of all, while economics is closely connected with complex human beliefs and desires, the predictions it is supposed to make (credit, the markets, profits, money, growth) concern exact numerical quantities.
“Supposed to make” according to mainstream economics — but not, of course, according to Austrian economics, where all rankings are ordinal rather than cardinal, and predictions of exact numerical quantities are impossible.
Moreover, anything that can be said, can be said mathematically.
You’ll have a hard time convincing me of that. In fact, I think that very statement is a counterexample to that statement.
while they may not accurately reflect the real world (because of the complexity of human behavior) nonetheless allow a glimpse into what one may expect
On why Austrians aren’t interested in unrealistic but predictive models, see my piece on abstraction in economics.
You might also wish to go through this article by Bryan Caplan
I know Bryan’s article well, but I’m not convinced; I’ll see you that article and raise you this one.
Thanks for the articles. I read them and while they did not change my mind, I have a better idea now on the philosophical points where we differ. I will not attempt to convert you to my more positivist worldview; but will just give a brief explanation of the role of mathematics in social sciences (like economics) where the underlying entities are complex and hard to quantitatively analyze.
I maintain that anything that can be said, can be said mathematically. As to expressing this particular statement mathematically, its fairly trivial. Once we allot variables for the different actions, the only thing to make precise is what one means by ’said mathematically’. That has, however, already been done in formal mathematical logic.
To give another (trivial) example, abstracting an object (like a horse) is simply passing from the set of all objects of that type to the set of common properties that these objects share.
These are simple examples, and it might not be immediately clear why it is of any advantage to formulate things in a different (though equivalent) manner. The point is, as the phenomena studied increase in complexity, it is difficult, if not impossible to pin them down quantitatively using mere verbal arguments. Mathematics is quite simply a much better language for dealing with quantities. Even many philosophical essays — like yours on abstraction — are, I feel, better understood from a mathematical set-theoretic viewpoint. And even when a thing cannot be analyzed mathematically at present, stating whatever is known in clear mathematical terms often clarifies where the trouble (or complications) lies.
The Austrian approach seems to be to simply abandon any attempt at quantization and to simply deal in ordinals and use verbal logic. That to me illustrates a misunderstanding of mathematical abstraction (to be fair, Friedman may have made some of the same mistakes). Mathematical abstraction is not the same as mathematical idealization. Expressing something mathematically is different from modelling it mathematically. The latter requires a reasonable (in a computational sense) function that closely resembles what we want. For instance, modelling large scale human behavior, is what some AI researchers and some behavorial economists are trying to do. On the other hand, stating how the mind works mathematically (without any attempt at simplicity) involves understanding how the neurons of the brain work, how they interact and send impulses (in essence writing the equations of physics in this enourmously complex setup). The whole point of modelling is to reduce a complex phenomenon to computationally viable equations. It does not imply denying the existence of other factors as you seem to suggest. When things have been mathematically stated from the very beginning, and appropriate experiments conducted, it is often possible to give quantitative upper bounds on the accuracy lost in making the idealization. Even if this is not the case, it always makes sense to have two mathematical versions of the theory; the ‘complete’ one with all the variables, and an imperfect one whose equations can be solved.
Understanding, or even modelling reasonably, the human mind may seem like an enourmously difficult task, and it is. Indeed, for some disciplines the underlying complexity is so high and the lack of viable models so great that it might seem that nothing is really gained by restating it mathematically, so why bother? After all, verbal arguments have the advantage of greater plausibility to the human mind (most people are used to English words and are less comfortable with propositional logic or probabilistic equations). But even there, it is critical to recognize that the obstruction to mathematical formulation is simply a result of lack of data or current computational capabilities, both of which will improve as time goes on. It is necessary to recognize verbal arguments when dealing with quantities for what they are, a compromise forced by lack of knowledge. As time passes and knowledge increases, things will move to a more mathematical footing. To give a not so relevant example, medicine earlier used to be mostly ad-hoc and empirical but is now often done by analyzing the properties of the individual molecules mathematically and combining this with our knowledge of biochemistry.
That is why I am disturbed by your assertions that predictions of exact numerical quantities are ‘impossible’ or that ‘it is unscientific and imprecise to apply mainstream mathematical methods to economic subject-matter ‘. The human mind and its beliefs and desires stem from the brain which is an extremely complex object but not in any fundamental way impossible to mathematically analyze. Besides (and more pertinently), economics is not about analyzing the human mind but about the effects of large scale human behavior on economically relevant quantities. When one is dealing with many people together, the complexities tend to balance out. It is a bit like studying gases in physics. Analyzing the motion of 10^23 molecules by solving all the equations is practically (but not theoretically!) impossible; however statistical mechanics does precisely that by using probabilistic techniques of dealing with many things together. No, I am not claiming that large-scale human behavior is comparable in simplicity to Boyle’s Law, but simply pointing out two things: (i) There is no theoretical basis to saying math is not applicable to economics, (ii) Even if there are practical constraints to effectively modelling it today, at least attempting to formulate the assumptions and the theories mathematically is neither unscientific nor futile.
The proper approach for Austrian economists would be to formulate their theories in mathematical language, even if they choose not to model it by simple functions. Precise cardinals are not a must; inequalities are perfectly ok in the absence of requisite information (or fast enough computers) as long as it is recognized that there is no conceptual obstruction to doing so. Among other things, it will allow much easier comparison with other economic theories and (when the time is ripe) allow a better, more scientific rewriting of its conclusions. And yes, this may mean abandoning some of that ‘anything other than ordinals is unscientific’ dogma.
I doubt if I managed to change any of your philosophical viewpoints, but here’s a prediction: In fifty years, Austrian economics will be either restated in fairly mathematical terms, or it will be pretty much dead. I hope (and predict) it will be the former.
That was a very persuasive argument! But still leaves the practical situation the same: It might be 3-4 generations more before some theorem is formulated to express all the empirical observations of Austrian economics.
I think the Austrians’ disdain for maths shows intellectual integrity. Much of mainstream economics uses statistics to prove its theorems, which is a far cry from the kind of experimental data used by scientists. No wonder they are often wrong.
There is that other bad side effect that people in power grab any set of statistical data that supports their pet schemes – central banking being the best example.
Pramod , thanks for your comment.
“It might be 3-4 generations more before some theorem is formulated to express all the empirical observations of Austrian economics. I think the Austrians’ disdain for maths shows intellectual integrity. ”
I do not necessarily disagree with your statement. But as I wrote, my main objection to Prof. Long was to his assertion that ‘in the Austrian view, it is unscientific and imprecise to apply mainstream mathematical methods to economic subject-matter ‘. That has a certain air of finality to it, don’t you think?
“Much of mainstream economics uses statistics to prove its theorems, which is a far cry from the kind of experimental data used by scientists.”
Actually that is not entirely true. Scientists also use statistical methods to deal with experimental data. The intellectually honest position would be to state clearly the assumptions made in any work of mathematical economics, and to make an effort to quantize the effect of these assumptions. Most statistical methods are sufficiently mature today so as to imply clearly the degree of their confidence. I have not read enough (mainstream) economics paper to pass a judgement on whether most papers are intellectually honest, but it won’t surprise me if many are not.
You are right about people in power (or positions of influence) misusing statistical methods. The fault there lies not with the methods but with the (mis)usage. That’s a perfect example of intellectually dishonest work.