Economic Methodology
a Chicago perspective
Some time ago, I came across a talk by Roderick Long, a libertarian philosopher and blogger sympathetic with the Austrian approach to economics. In it he criticizes the argument in my father’s essay "The Methodology of Positive Economics," which defends the use in economics of unrealistic models such as perfect competition on the grounds that the test of a model is not its descriptive accuracy but its ability to make correct predictions.
Roderick, if I understand him correctly, imagines that all that is going on in the Chicago approach is blind curve fitting, looking for patterns in the observed data and assuming that those patterns will continue. A body of data can be fitted with an infinite number of different curves, a set of facts with an infinite number of theories, making the probability of picking the right one by chance low — so the theorist does not pick his theory at random. The test of whether he has done a good job of figuring out what simplified model includes the important factors and excludes the unimportant ones is the ability of the model to make correct predictions.
Crucial to this view of the process is the distinction between explaining facts you already know and predicting facts you do not know, a point emphasized in my father's essay but, I think, missed in Roderick's lecture. Explanation of known facts can be blind curve fitting but if that is all it is, predictions of facts that did not go into constructing the model are unlikely to be correct, increasingly so over time.
Roderick offered an elaborate philosophical explanation1 of why my father rejects what Roderick views as the correct approach to doing economics, the a priori approach associated with Ludwig Von Mises and some of his followers. There is a much simpler explanation. The problem with that approach, at least in its extreme version, is that pure a priori argument is unable to predict anything of economic interest. If one is completely agnostic about the facts, including both utility functions and production technologies, any physically possible pattern of human behavior is consistent with the theory. As I put it long ago in my Price Theory, explaining why the assumption of rationality is empty unless combined with some knowledge of what humans value:
Why did I stand on my head on the table while holding a burning $1,000 bill between my toes? I wanted to stand on my head on the table while holding a burning $1,000 bill between my toes.
I conclude that the correct way of doing economics combines a priori theory with evidence. You form plausible conjectures on the basis of theory and evidence, where part of forming them is deciding what simplifications, what unrealistic features of the model, assume away inessential complications while retaining the essential features of what you are trying to understand. You find out how good a job you have done by using the conjectures to make predictions and testing them. An added benefit of that process, as I discovered in the course of revising what became my first published journal article in economics in response to an initial rejection, is that finding real world predictions of your model may force you to think through the model itself more carefully.
That is the Chicago School methodology as I understand and practice it.
I have sometimes challenged supporters of the extreme a priori approach to describe any real world prediction that can be derived from economic theory with no additional information. One response I sometimes get is that if you raise the minimum wage, all else held constant, employment of workers previously receiving the minimum wage must decline.
My rebuttal used to be to posit a situation where consumers do not like to buy goods produced by very low paid workers, have no easy way of knowing what wage each employer pays, and so are reluctant to buy goods that they know were produced by unskilled labor. After the minimum wage is raised consumers know that all goods are produced by workers making what they consider a tolerable wage. Their demand for goods made by unskilled labor increases, increasing employment for the workers who make such goods. Economic theory does not tell us what tastes consumers will have so cannot eliminate that possibility, however implausible it may seem.
That made the point that the conclusion depends on factual assumptions as well as theory but I now have a more interesting example of circumstances where raising the minimum wage increases, instead of decreasing, employment for low skill workers.2 Suppose the employers of unskilled labor are monopsonists, each the only employer of such labor in the relevant market. It pays each to hold down the number of workers he hires in order to hold wages down, just as a monopolist holds down the amount he sells in order to hold price up. Increasing the minimum wage to what it would be in a competitive labor market results in the monopsonist hiring more workers, since doing so no longer increases the wage he must pay.
My previous example showed that it was logically possible for the prediction to be false. This one shows that there are plausible real world circumstances — perhaps not in America today but in some past market societies — where it would be false. That was a fact that had not occurred to the people who offered that response to my challenge or to me.
Nine to Five Economists
A very long time ago I had a friend and colleague whom I concluded was only an economist in working hours. Having come across another example more recently, I thought it would be worth explaining the concept.
Suppose there is some issue on which economics implies a straightforward conclusion, but an implication short of a proof. Minimum wage laws provide a simple example. The conclusion that raising the minimum wage will reduce the employment of those currently receiving a minimum wage is a straightforward implication of the assumption that demand curves, in this case for a particular sort of labor, slope down. It is however possible, as I have just demonstrated, to construct a model of the labor market which yields the opposite conclusion.
The response of a real economist will either be "raising the minimum wage almost certainly reduces employment for low skilled workers" or, less likely but not impossible, "one of those odd models might possibly be right, I will look for a natural experiment by which I can test it."
The response of someone who is only an economist in working hours is to believe whatever he would believe if he was not an economist. If that requires him to believe that raising the minimum wage will not reduce employment for low skilled workers and someone points out the inconsistency with standard economic analysis, he will justify himself on the grounds that the economic argument is not actually a rigorous proof, so its conclusion could be false. For an example on the other side of the political fence, consider a conservative economist who wants to support trade restrictions.
In either of these cases, there may be a way for a real economist to support the policy conclusion he wants. If the supporter of a higher minimum wage is unwilling to claim, implausibly, that employers of low-skill labor are mostly monopsonists, he can argue that the loss of employment to some current minimum wage workers is more than balanced by higher wages to others so that the total earnings of low skilled workers go up instead of down. If that is his argument he will want to look for evidence on the relevant elasticities. He may support his point by observing that losing a ten dollar an hour job is not really a net cost of ten dollars, since it increases leisure by an hour for every ten dollars lost, but he will be bothered by the possibility that losing a ten dollar an hour job removes the first step leading to a fifteen or twenty dollar an hour job.
In the trade case, the argument might be that although free trade produces net benefits for Americans for familiar economic reasons, the gains go to richer people than the losses, so a gain in value measured in dollars leads to a loss in value measured in utility. A real economist will want to make some effort to find out if the claim is true, to estimate the income distribution of gains and losses and their relative size. If he cannot find economic support for his position, he may agree in private that the effects of trade restrictions are negative but support them in public as a way of getting the rust belt votes needed to elect politicians who will do other things that, in his view, more than compensate for the loss.3
The simple test is whether an economist's views tend to diverge from those of his ideological allies when the ideology clashes with the economics. If they do he is a real economist. If they do not, he is only an economist in working hours.
I came across a different example of the same pattern accompanying my daughter on her college search. Wandering around the economics department of one of the colleges to get a feel for the place I spoke with three faculty members, none of whom appeared to be an economist in my sense of the term, someone for whom economics was part of how he viewed the world. My daughter, having audited an econ class, told me that a student had made a comment which an economist should have responded to with some version of "that sounds plausible but is wrong because." The professor simply let the comment go.
A long time ago I commented to a graduate student at Chicago that it seemed to me that there were a lot of economists who did not really believe in economics; it was what they did in working hours, not how they thought about the world. His response was that some of his fellow graduate students had noticed that when visiting other schools. I do not know enough about economics department to say which ones currently are in which category. Chicago, so far as I can tell, is still a place where some of the economists believe in economics.4 One simple test would be to have lunch with members of the department, perhaps also with their graduate students, and see what they talk about. A few years ago I did the experiment, repeatedly, at George Mason. It had, at the time, three different campuses with parts of the economics department in them. The one I was visiting easily passed.
My point is not that people should believe in economics because it is true, although I think it largely is, only that economists should believe in economics. I am not inclined to take theology very seriously but, if I did take a course in it, I would expect to learn more and have a more interesting time if the professor was a believer than if he were an atheist.
Getting Disagreement Wrong
An article on my father by Steven Pearlstein starts with a puzzle. Most of the articles published on Milton Friedman after his death agree that he was a great economist; many compare him to John Maynard Keynes, another great economist. But Friedman and Keynes held different, indeed inconsistent, views; an important part of Friedman’s accomplishment was to undo Keynes' accomplishment. If Keynes was right, how can Friedman be a great economist? If Friedman was right, how can Keynes be?
It is an interesting question but the author gets the answer wrong. He concludes that both were right. Keynes' version of economics was correct for the forties and fifties, Friedman’s for the seventies and eighties when the Keynesian model "had played itself out."
That is a claim that neither Keynes nor Friedman would have taken seriously. Keynes titled his magnum opus “The General Theory of Employment, Interest and Money” not “The Theory of How Employment, Interest and Money Will Work from 1930 to 1960.” Part of the work that earned Friedman his Nobel was A Monetary History of the United States (coauthored with Anna Schwartz), in which he demonstrated that the Keynesian analysis of the Great Depression, a centerpiece of the Keynesian view of economics, was based on a historically mistaken account of what actually happened. It is an odd view of science in which the historical facts about the 1930’s changed between 1940 and 1970.
Pearlstein claims that economics is less of a science than physics, hence its truths more temporary. Yet the history of physics offers the same puzzle. Newton was a great physicist. Einstein was a great physicist. Part of Einstein’s accomplishment was to show that Newtonian physics was, in certain fundamental ways, wrong.
Newton was wrong, wrong not only now but then, but Newtonian physics provided the foundation of ideas on which later generations of physicists, including Einstein, built. Keynes was wrong, but his attempt to make sense of what he believed happened during the Great Depression provided a theoretical foundation on which later theorists, including Friedman, could build. Hence Friedman’s comment on Keynes: “In one sense, we are all Keynesians now; in another, no one is a Keynesian any longer"—misquoted by Time Magazine as “We are all Keynesians now.”
Friedman and Keynes are not the only example in economics of such a pattern. David Ricardo was one of the most important figures in the history of economics, arguably its first great theorist. He starts his book by saying some admiring things about Adam Smith and then demonstrating that Smith’s view of how prices are determined could not be true.
Past posts, sorted by topic
My web page, with the full text of multiple books and articles and much else
A search bar for text in past posts and much of my other writing
I got this point from a article by David Card and Alan Krueger, later discovered it had been made much earlier by George Stigler. The Card and Krueger article is mostly empirical — the theoretically interesting part starts on page 791.
On methodology, Long and Friedman are both wrong. A priori, in the Rothbardian version, is absurd and not really worth addressing. Friedman's perspective is more interesting. The philosophy of economics (and social science more generally) literature has advanced a lot since th 50s! Here are a few problems with his methods piece.
1. Unrealistic Assumptions Are Not Harmless
Critics: Uskali Mäki, Tony Lawson, Nancy Cartwright
• Critique: Friedman claimed that unrealistic assumptions are acceptable if the model predicts well. But critics argue that assumptions matter because they can lead to faulty reasoning or harmful policy.
• Mäki’s View: The “as if” methodology undermines our grasp of real causal structures.
• Lawson’s View: Economic theories should reflect ontological realism—identifying actual causal mechanisms, not just formal structures that generate accurate forecasts.
2. Predictive Power Is Not the Only Scientific Virtue
Critics: Daniel Hausman, Wade Hands, Kevin Hoover
• Critique: Friedman prioritized predictive success as the ultimate test of a theory. Critics argue this is too narrow. Science also values explanation, causal insight, and internal coherence. We care about causal mechanisms, in addition to causal effects.
• Hausman’s View: A theory’s value comes from its ability to contribute to understanding, not just prediction.
• Hands and Hoover: Relying solely on predictive accuracy ignores how models can mislead when assumptions diverge from reality.
3. Friedman’s View is Internally Inconsistent
Critics: Alex Rosenberg, Don Ross
• Critique: Friedman’s own examples (e.g., billiard players behaving as if they solve complex equations) contradict his broader claim. If we say agents behave as if they optimize, but they clearly don’t, then the assumptions don’t really help us understand anything.
• Ross’s View: Friedman selectively invokes realism—defending or ignoring it depending on convenience—undermining his claim to methodological consistency.
4. The “F-Twist” Is Methodologically Problematic
Critics: Alan Musgrave, Bruce Caldwell
• Critique: Musgrave labeled Friedman’s idea the “F-Twist”: that the more unrealistic the assumptions, the better, so long as the model predicts well. Critics argue this justifies poor theorizing under the banner of scientific rigor.
• Caldwell’s View: Methodological pluralism is better. Friedman’s monism leads economists to undervalue alternative approaches—historical, institutional, or behavioral.
I feel like you're talking past each other - or at least talking past the point of Mises with regard to apriorism. Your two examples are not examples of using empiricism to verify or falsify a claim of economic science or to make predictions; they are rather hypotheticals, the outcomes and causes of which are determined by your theoretical outlook, which change the prediction by changing the assumptions made. Your challenge that an economist must then determine which situation actually obtains in order to make a correct prediction is exactly in line with Mises' point about economics: it does not determine whether causes have the effects by examining whether the effects actually obtain through an empirical examination of the data, because this is impossible. It examines the effects under different hypothetical causal regimes, and then in application determines which set of causes are operative for the case under examination. The knowledge of the effects of these causes are given a priori in each case. Consider this statement from Human Action, which I imagine you are already familiar with:
"The imaginary constructions of praxeology can never be confronted with any experience of things external and can never be appraised from the point of view of such experience. Their function is to serve man in a scrutiny which cannot rely upon his senses. In confronting the imaginary constructions with reality we cannot raise the question of whether they correspond to experience and depict adequately the empirical data. We must ask whether the assumptions of our construction are identical with the conditions of those actions which we want to conceive."
From this point of view, your method of procedure here is exactly in line with Mises' comments about methodology; comments, by the way, intended as a description of how economic science operates in practice, not as a prescription for that practice as is often conveyed. Empirically speaking, I would say Mises is accurately capturing the procedures employed.
It is true that in practice we also blend theoretical and empirical aspects of the problem. Economics doesn't only examine statements like "Action is the effort to remove felt uneasiness", but also examines more concrete phenomena that are given in experience. This is true, but it does not alter the aprioristic character of *how* those concrete phenomena are examined; the procedure by which the effects of causes are determines is a theoretical one, one of applying the general principles of theory to a given case. This is how you determined that a monopsonist could pay wages lower than marginal productivity, or that buyers who dislike products made by unskilled workers would be less willing to pay for them.
Finally, a few quibbles regarding your examples.
First, for the monopsony example, I don't think Austrians usually dispute the theoretical claim that monopsony can be a case where minimum wages don't lead to unemployment. They usually argue that monopsony does not really occur in the free market, and cases would usually be caused by government regulation (in which case, wages could be increased by removing those regulations).
Second, the example which uses buyer preferences places two different kinds of products on the same demand curve. If buyers distinguish between products made by low-skill workers and other kinds of products, then ipso facto these are different goods which will demand different prices. Placing these different goods on the same demand or supply curve would be an error and lead to specious conclusions. Properly understood, we should say the minimum wage law would change what products are being offered. Ceteris are not paribus. I think every austrian agrees that a minimum wage, when is not above the marginal productivity of unskilled workers, will not cause unemployment. This is the primary thesis of the analysis of minimum wage viz-a-viz employment. Your example is a sneaky way of getting the marginal productivity of unskilled labor to change compared to the counterfactual; but in that case, I would say you haven't kept other things equal. Similarly, a minimum wage will not cause unemployment if a large increase in inflation or a large increase of productivity is parallel to the minimum wage. This doesn't conflict with the predictions of minimum wage analysis, it is simply changing the assumptions involved, and comparing the results under each assumption. In either case, you use a priori reasoning to know what will happen and how it can happen. In order to determine which situation actually obtains, you need to wait and see which set of assumptions obtains in experience, just as Mises emphasized.