The fact that the discipline of economics hasn’t helped us improve our predictive abilities suggests it is still far from being a science, and may never be....Over time, the question of why economics has not (yet) qualified as a science has become an obsession among theorists, including philosophers of science like us...What is economics up to if it isn’t interested enough in predictive success to adjust its theories the way a science does when its predictions go wrong?--Alex Rosenberg and Tyler Curtain, in the New York Times.
Sometimes I receive the following back-handed, compliment-question, "Why do you do history? You might make a decent real philosopher." A part of the answer is that the second-order stories we tell ourselves -- often handed down by mentors and supervisors, and senior peers -- about where we come from and what we do often are just as interesting and important as the first-order activities; they may also influence us and others in ways that are often hard to spot. This by way of introduction because in what follows I primarily challenge the conceptual oppositions (and associated historical myths) in Rosenberg and Curtain's Opinionater piece (allowing that the genre they are writing in need not require scholarly precision). Perhaps, my challenge allows some clarity about the first-order issues to emerge.
The main stated point of Rosenberg & Curtain (hereafter RC) is that "the task of the Fed’s next leader will be more a matter of craft and wisdom than of science." Surprisingly enough, given that RC are philosophers, they spend very little words on conceptualizing the nature or origin and causes of such "craft and wisdom," even though at the end of their piece they boldly assert that "the Fed chairman must, like a first violinist tuning the orchestra, have the rare ear to fine-tune complexity (probably a Keynesian ability to fine-tune at that)." We are not told why of all the crafts and skills, the fine-tuning violin is the most appropriate exemplar for a Fed chairman. Even if we grant the fruitfulness of the tuning metaphor, a fine-tuning violinist possesses a skill that may not require (much) wisdom; she tunes an orchestra that is oriented toward a common goal with skilled performers. The Fed-chairman deals with a more heterogeneous population with ends that are -- I hope -- not unified.
In fairness to RC, most of their piece is focused on a self-described "obsession:" is economics 'science?' So, let's turn to that first.
As an aside, professional economists are far more familiar with the contrast between positive and normative economics (also due to J.N. Keynes and again promoted by Friedman). So much so, that a generation ago, knowledgeable commentators could discuss the lost art of economics. RC never pause to consider what caused this (partial) loss.
Instead, RC worry quite a bit about the fact that economics is not predictive. The word 'predict' and its cognates is repeated like a mantra in the piece. (They even offer an argument that goes from the unpredictability of scientific/technological innovation to economic unpredictability.) Just for the record: it is false to say that economics is not predictive. Economics is very good at predicting in certain domains. (The predictive consequences of, say, rent-control are very well understood; there are also by now lots of robust predictions in experimental economics.) Even macro-economic predictions can be very impressive: between 1971-2008, the Dutch central planning agency (CPB) predicted Dutch growth with a 0% average error margin in growth. (Over time the errors in direction canceled out.) At one point, the modelers had fantastic reasons to think that they had succeeded! (Recall this post with more analysis.)
It is an open question to what degree economic models can ever yield proper macro-predictions in environments unlike the great moderation; as regular readers of this blog know, agreeing with RC, I am on the side of the pessimists (uncertainty, reflexivity, etc.), but decreasing cost of computing may be a game-changer for some domains that seem intractable now.
The first step to wisdom has to be to understand the limitations and potential capacities of one's tools.
RC address this point in their piece as follows, "Fixing bad economic and political institutions (concentrations of power, collusions and monopolies), improving good ones (like the Fed’s open-market operations), designing new ones (like electromagnetic bandwidth auctions), in the private and public sectors, are all attainable tasks of economic theory." Oddly enough, they pretend that no branch of economics is engaged with this -- conveniently ignoring public choice theory, New Institutional Economics, Agent Based Modeling, and what I call 'institutional game theory' (here), etc.-- and assert that it has been primarily "social and political philosophers" that reflect on institutional design. (They name Hobbes, Hume, Rawls, and Nozick along the way.) As regular readers may know, some of my best friends are philosophers (and I have even offered my own approach to rethink the ethos of institutions), but (really) we're barely in the game (although some formal philosophers are). Rather than focusing on a gotcha, I offer this point because it goes to the heart of the problem with the piece. RC recognize that "None of our models of science really fit economics at all." But they fail to reflect on the harms produced by bad images of science -- the unintentional legacy of philosophy of science -- in scientific practice (and to which they contribute). One important constructive role we philosophers of science can play is to facilitate a better image of science for the policy sciences.
I start with a point of agreement with RC: ought we really want predictions from the art of economics? Agreeing with RC, I think not. The primary function of economic models and expertise in a policy context is to help decision-makers think about trade-offs, relevant constraints, and help explore alternative policy possibilities (or even make a conversation among warring political elites possible); in a broader context economists can help us rethink (local) institutional design. Clearly, some knowledge of causal consequences is relevant for all of this, but having access to expected pattern-margins is far more important than precise prediction(s).
Yet, against RC, I hold that one can be properly scientific in all kinds of ways without having access to precise predictions in the domain with, say, the highest social needs. So I accuse them of a philosophical mistake here. Such mistakes are not innocent, they can infiltrate the sciences. Some such infiltration story is, in fact, suggested by RC:
The irony is that for a long time economists announced a semiofficial allegiance to Karl Popper’s demand for falsifiability as the litmus test for science, and adopted Milton Friedman’s thesis that the only thing that mattered in science was predictive power. Mr. Friedman was reacting to a criticism made by Marxist economists and historical economists that mathematical economics was useless because it made so many idealized assumptions about economic processes: perfect rationality, infinite divisibility of commodities, constant returns to scale, complete information, no price setting.
Let's ignore the oddity of how the demand for 'falsification' (as litmus test) gets transformed into the embrace of 'predictive power.' RC are writing for a popular audience, after all, and one can kind of figure out how the right sort of predictions are necessary precondition for falsification. To be clear: contemporary economists are not in the business of producing falsifiable models. The action in what passes for mainstream theoretical/mathematical economics (not 'pure' theory) is to take institutions as given, and explore minor variations in policy changes. (I have blogged scathingly about this, and the kind of technocrats it breeds.) Yet, in the quoted passage RC also create new myths: (a) the silly claim that according to Milton Friedman "the only thing that mattered in science was predictive power;" (here's Friedman: "A hypotheses is important if it “explains” much by little, that is, if it abstracts the common and crucial elements from the mass of complex and detailed circumstances surrounding the phenomena to be explained and permits valid predictions on the basis of them alone.") [This is not to deny that Friedman emphasized predictions.] (b) The idea that Friedman was defending "mathematical economics;" while Friedman made some contributions to mathematical economics, he loathed much what passes for mathematical economics in his own time. His economics was primarily data-driven (see this article) and primarily oriented toward guiding the art of economics. (As Mark Blaug reported to me, Friedman also taught economics by having students discuss newspaper articles in class.) The art of economics disappeared from mainstream economics in the period of the displacement of Friedman's work from the technocratic center. (Friedman wasn't the only source of a possible art of economics, so the larger story is more complicated.)
It is convenient to RC, who are writing for a popular audience, that Milton Friedman is still a famous economist and associated with an ideological orientation. But in their eagerness to rely on the Meme of Friedman-as-villain, RC engage in some revisionary history and are blind to Friedman's living legacy in economics, even when it stares them in the face:
At this point it is a craft, to be executed with wisdom, not algorithms, in the design and management of institutions. What made Ben S. Bernanke, the current chairman, successful was his willingness to use methods — like “quantitative easing,” buying bonds to lower long-term interest rates — that demanded a feeling for the economy, one that mere rational-expectations macroeconomics would have denied him.
Well, before Bernanke became chairman he was the world's expert on the Great Depression of the 30s. One way to understand his contribution to that technical, field of study, is to see him bring the latest economic, technical tools to re-evaluate and empirically refine the framework he inherited from (yes) Milton Friedman's and Anna Schwart'z great study. As it happens, Bernanke was as prepared for the 2008 crisis as any person could be because he had been spending most of his adult life studying the relevant problem(s) before being confronted by Lehman's collapse. Most importantly, at some point in 2008 he recognized that the lessons learned (in part because of his work) from the collapse of the banking sector during the 1930s were relevant to the problems he faced.
It is no surprise that economic history has a robust institutional presence within economics in lots of countries (and hopefully returns to prominence Stateside). This is a good thing because history is the science par excellence that cautions against hubris--the one virtue that RC explicitly identify as belonging to the craft. (Violinists are really irrelevant as exemplar here.) One need not be a friend of Friedman's political views, to recognize that his branch of historical-data-driven economics (once associated with the NBER) was oriented toward informing and improving the art of economics. (We also need not be blind to the limitations of this approach.)
This, in turn, gets me to the point of this post. RC take for granted the contrast between science and wisdom/craft. They paint a picture of expertise in which 'science' is possible without the virtues and the ability/capacity to apply the sciences properly. (This is compatible with the claim that some virtues are instrumental in science.) In the piece RC do not spell out their picture of science, but they do embrace Kuhn's account of "scientific paradigm shifts," and this helps us put a finger on the core problem here.
Kuhn's philosophy of science embraces scientific unanimity around a shared paradigm as intrinsic to science. In this alluring picture, the role of science is to deliver unequivocal answers to well-posed questions. This picture of science was adopted by the economists, including Friedman's best friends, and helped facilitate a disciplinary orientation around a search for tools that deliver unequivocal answers (recall this post, and see this paper (or lecture)), that also ended up marginalizing Friedman's role in the profession. (Yes, we all wish to be so marginal!)
In the Kuhnian picture, the scientific virtues are instrumental to the end of theory-generated-answers. If one's expertise consists in having the right answers, then a bias toward hubris is (given human nature) inevitable. It also entails that the means of using science in one's art is relegated to the fringe of science, if not removed from it altogether (sometimes even lodged within philosophy). So, one important task for philosophers of science is to promote a picture of policy science in which the art of economics is internal to what it means to be a good scientist.
Recent Comments