Economics: An Art or a Science?

Many of Dani Rodrik’s arguments are standard defences of the economics discipline. What is new is the suggestion that economics can remain a science even if it does not follow the scientific method, explains Sinclair Davidson.

Economists get no respect. When our predictions aren’t wrong, our assumptions are fallacious, our views of humanity bleak, or we’re trying to impose our personal preferences on our fellow human beings. And that’s just what our friends say.

To counter these sorts of criticisms, Dani Rodrik—Ford Foundation Professor of International Political Economy at the John F. Kennedy School of Government at Harvard University—has written a defence of economics Economics Rules: The Rights and Wrongs of the Dismal Science.

Mind you, it isn’t clear if his is a defence of economics as a discipline, or academic economists as a profession. Most likely, it’s more the latter than the former. It is easy to confuse and confl ate the two concepts, as the university sector is the single largest employer of economists.

Rodrik begins, sensibly enough, by defining what ‘economics’ is. This is a variation of the old and rather unhelpful definition that economics is what economists do. On the one hand, economics is the study of how the economy operates. On the other hand, economics is a methodology employed in the social sciences. In this latter view, economics can be deployed to ask, and answer, many questions that are well beyond the study of the economy. It is here that Rodrik has gotten himself into some difficulty.

It is very interesting to understand, for example, why drug dealers live at home with their mothers, or show that teachers cheat in standardised testing, and the like. Who doesn’t like a clever answer to a difficult problem? Th e economic method provides deep and powerful insights into many areas of human life.

But—and this is an important but—the study of how the economy operates is what economists should be doing, fi rst and foremost. As Rodrik implicitly suggests, and as is reasonably well-known, economists are not actually very good at understanding how the economy operates.

He prefers to define economics as being a method within the social sciences—so a series of techniques and tools that can be deployed to the analysis of any problem within the social sciences. So economics is what economists do. Economists—strictly speaking, academic economists— become generalists with a specialist tool kit. Th is allows Rodrik to focus on the tools of the trade, or ‘models’ as they are known in economics.

When explaining models to my students, I like to show a photograph of the city of Melbourne and then a map of the city of Melbourne. I then pose the question: which of these two pictures would you find more useful when giving directions, as opposed to describing the city? While the photograph is a very realistic depiction of the city, the map is preferred for giving directions. Maps are models that most people understand and use with confidence. Nobody imagines that maps are realistic or represent an ideal reality. Economic models are like maps.

If only Rodrik would say that. Rather, he tells us that models are fables—simple stories that tell short but sharp messages. Yes, I can see that. Then he argues that models are experiments. No, I don’t think so. Experiments are mechanisms for testing the predictions or implications of models, they are not models themselves. While experimental environments can often be as contrived as models, there is an element of reality associated with experimental sciences that is not present in theoretical model building. That isn’t to suggest that real world results shouldn’t be incorporated into model building, but we’ll get to that later.

Model building raises the rather thorny problem of mathematics in economics. Rodrik doesn’t avoid the issue, but he does trot out the stock standard party line: mathematics is used to provide clarity and consistency in economics. Just because this argument is a cliché doesn’t make it wrong. But it doesn’t preclude the suspicion that most mathematics in economics consists of little more than intellectual onanism. Rodrik suggests that ‘math for its own sake doesn’t get you far in the economics profession’. That raises another question—what does he mean by ‘far’? Even if not far by Rodrik’s standards, it is certainly too far to add any value to society at large. The point remains, however, that mathematics is useful for the economics profession—even if to know that mathematical manipulations have no economic value; and aspiring economists need to learn and know the mathematics.

Surprisingly, he doesn’t say much about econometrics. A huge oversight given that he argues that economics is a science—even if only a social science—where hypotheses are formulated and tested. But the thing is so many economists don’t bother to test their own models or theories. To be fair, very oft en the data do not exist to test models—but by the scientific standard, models that cannot be falsified are neither theories nor scientific. Herein lies Rodrik’s challenge.

He argues that economics advances like a library—by adding to the collection of models, not by discarding falsified theories. Rodrik is well aware of the diffi culty this raises and poses the question himself: ‘What kind of a science has a model for everything?’ Can a collection of ad hoc models amount to a science? Clearly no. Yet Rodrik answers ‘Yes’. This is where he brings the mathematics story back into play—because models have clear, concise, and precise assumptions and implications, models have built in ‘teaching notes’ on how and when they should be applied.

That argument, however, is simply not convincing. Anyone who has ever attended an economics seminar, and observed economists arguing over precisely how clear and concise their underlying assumptions are, will know that Rodrik’s argument is more wishful thinking than reality. (See Rodrik’s Ten Commandments for non-economists—commandment 10 alludes to this very point).

Rodrik’s descriptions of economics and academic economists aren’t necessarily wrong. Many of his arguments are standard defences of the discipline. Despite his protestations to the contrary, most of his arguments are the ‘party line’. What is new is the suggestion that economics can remain a science even if it does not follow the scientific method. And while his argument fails, it is true that poor models are seldom discarded by the discipline. This makes economics an art and not a science—he seems to realise this:

As with fables, good judgement is indispensable in selecting from the available menu of contending models. Luckily, evidence can provide some useful guidance for sifting across models, though the process remains more craft than science …

Rodrik’s mistake is definitional. By defining economics as a collection of models that make up a method within the social sciences, he disconnects economics from its subject matter. An economics that primarily considers how the economy works must be anchored in reality.

Economists who have a series of hypotheses about how the economy works will quickly come to discard models that fail. For example, up until the 1980s, many within the economics profession believed that government ownership of the commanding heights could produce the same level of economic prosperity as a capitalist economy.

While the theory setting out why that could never happen was articulated in the 1930s, it was the practical failure of socialism that convinced everyone, including the economics profession, that capitalism was a superior economic system.

What is particularly disappointing is that Rodrik doesn’t cite many economists who have collectively made similar arguments. Names like Ronald Coase, James Buchanan, George Stigler, and Deirdre McCloskey are conspicuous by their absence.

Rodrik covers a lot of old ground. He does this well and, for the benefit of students, it is all in one place. He does have a very nice collection of commandments for both economists and non-economists. He provides an excellent guide to model selection. He is quite right to suggest that economics does not discard disproven, unproven, or even unprovable models.

He is wrong to suggest, however, that economics can then still claim to be a science.