1Harro Maas says at the outset of his introduction to Economic Methodology that it is unusual in not being written from the perspective of the philosophy of science. It is clear enough what he means by this: the book isn’t organized in terms of the standard philosophy of science headline topics of theory, argument forms, rationality, explanation, causation and so forth. But in a more substantive sense he’s wrong to say this, because he’s written just the sort of book we need for teaching philosophy of science (as applied to a specific science). Its principal virtues are close attention to history, with selection of episodes guided by milestones in the evolution of the discipline’s methodology, and an equally consistent focus on practice rather than theoretical doctrine.
2In consequence both of its approach and its execution, this is a terrific book, which philosophers of economics and methodologists should read even if they’re not preparing a class for which they might want to prescribe it. It tells a coherent, critical but respectful, history of the discipline that—finally!—most economists will recognize as an accurate profile of their actual methodological preoccupations.
3It is not obvious that this is what we’re going to get from the first post-introductory chapter, which is about Mill and the tension between deductive and inductive reasoning in political economy. Here we go again, one might think, expecting the next chapter to be about rationality. But Maas is teasing us—Mill’s high level of abstraction, it turns out, is an effective foil for the rest of the book, to be contrasted with attention to the actual products that economists produce and the tools by which these products are made.
4Maas pivots off the induction / deduction contrast by showing the reader its expression in the contrasting philosophies of economics promoted by Jevons, who sought to bring calculus to bear on economics and to make it a practical policy tool by modeling it after engineering, and the Austrians (here, Menger and Robbins), who aimed at deriving generalizations of sweeping scope from purportedly intuitive first principles. Jevons is often given short shrift in methodological literature and pedagogy because he was not philosophically sophisticated like Menger or Robbins (or Keynes). But his general orientation is much closer to that of the typical contemporary economist, who tries to fashion solutions to real business or policy problems using technical modeling tools. Among the first generation of marginalists, Jevons is just the one with whom to start.
5This leads Maas to focus, in his next chapter, on the rise of statistical modeling in economics before World War II. He takes the reader on a quick tour through the career of Tinbergen and early business cycle modeling. Statistics and business cycles, before general equilibrium, Weberian ideal types as precursors to representative agents, or pre-Samuelsonian utility theory! —I was cheering as I read this chapter, for this is where the real engine room of the modern discipline was built. Maas’s context as a Netherlander is helpful here, as the reader is introduced to Tinbergen’s role in fostering a uniquely Dutch attitude to economic technocrats as umpires of politics, in whose terms politicians are expected to couch rival policy proposals. (That is, they must indicate which variables in the national forecasting model they will manipulate.) Of course, those who think that a simulacrum of the Dutch approach has poisoned the roots of participatory democracy in the European Union will blame Tinbergen’s influence instead of celebrating it. (That is not my view.) All sides in the contemporary debates over Europe’s governance structures and economic institutions will benefit from learning this history.
6Chapter 5 takes us to the mathematization of economics, again by the distinctive route we’ve been started on—through Keynes’s quarrel with Tinbergen’s econometrics, rather than up the usual street past the houses of Pareto, Hicks, and Samuelson. And again it’s a sound choice to skip the famous statues on the ceremonial boulevard and instead watch economics develop its modern character on the factory floor. Providing a rigorous basis for national accounts and for the decisions of economic engineers—especially national treasuries and central banks—is the set of preoccupations that made economics so different from the other social sciences in its sociological dynamics, and in its self-conscious occupancy of the frontier between academia and practical affairs.
7This does not mean that Samuelson is ignored; his attitude to the value of highly abstract formal modeling is indeed the central topic of Maas’s chapter 7. There, Samuelson is nicely defended from Daniel Hausman’s (1992) allegations that toy models such as overlapping generations are exercises in empirically frivolous mathematical play. Echoing recent suggestions from practitioner methodologists Robert Sugden (2002) and Edward Leamer (2012), Maas locates the practical significance of conceptually stripped-down models in their status as thought experiments. However, we get to Samuelson not through reflections on behaviorism and positivism—which, although relevant features of his peripheral intellectual landscape, were not at all central motivating forces for him—but through a superb chapter 6 on the Cowles Commission, Koopmans, and structural modeling of national economies. Once again the history of Dutch economic institutions provides a concrete and illuminating platform. This delivers a major bonus: Milton Friedman’s (1953) famous argument for instrumentalism about assumptions, which has struck most philosophers of science as borderline crazy (see Mäki, 2009), is set in its actual dialectical context within economics. Friedman was opposed to the idea that economists could profitably find a surrogate for experimentation in econometrics, and his stalking horse for this skepticism was the failure of the structurally rich Klein model of the US economy to predictively out-perform simpler models that were self-evidently naïve in their assumptions. In denying that the empirical applicability of assumptions ever matters at all Friedman rather ludicrously over-extended his point, which in any event Maas (and I) think was incorrect; but Maas’s reader will much more charitably understand what Friedman’s point really was through being given its context within the practical concerns of the discipline.
8A theme stressed by Maas throughout his book is that economists engage in pretext when they imagine that their cultivation of technical expertise as modelers can lift them above politics and normative judgment altogether. He diagnoses Friedman’s unhappiness with the Klein model, in conflict with Cowles, and his disagreements with Samuelson, as expressing ideological differences: those who think that economies can’t be accurately structurally modeled are less likely to advocate policy interventions to improve their efficiency or fairness. Based on his chapter 9 about the whole-economy simulation models used by modern central banks, Maas seems to think that economists, and those who use their work in crafting policy, should best self-consciously acknowledge that technical expertise does not transcend normative judgment, but rather presupposes it. This is because there is no purely algorithmic way of assessing a model’s empirical adequacy; one must bring to bear technical understanding of the model but also appreciation gathered from reflective experience of how the real forces in the real economy play out on the ground. Value judgments are both inputs to and elements of such appreciation. Maas might think that the history of Dutch policy-making provides a stock of examples of judicious balancing of this kind; if so, and if he’s right, then students have good reason to want to delve into his case history sketches in more detail. It is a deeply happy outcome when one’s students see the value of responsible historicism.
9In chapter 8 Maas devotes considerable space to consideration of two cases from experimental economics. Happily, he doesn’t choose experiments intended to examine utility or response functions of ecologically isolated individuals—which are properly the province of psychology—but instead dwells on experiments intended to reveal properties of markets. His critical discussion wisely focuses on two leading methodological issues raised by economic experiments generally, namely, their artificiality if they’re confined to the lab and not complemented by work in the field, and problems associated with scaling so that laboratory effects can be considered relevant to drawing conclusions about the real markets that experimental designs aim to simulate.
10Maas has thus produced the best secondary source available for teaching economic methodology. Despite his calling it an introduction, it would be entirely suitable for a graduate class, indeed would perhaps work better with students who have enough acquaintance with economics to appreciate its more subtle lessons.
11That said, I have two minor objections to claims made by Maas that it would be nice to see corrected in subsequent editions. In Chapter 5 he says that “With the rise of subjective probability theory, the distinction between risk and uncertainty that Keynes saw as fundamental was about to lose its significance” (70). In fact, as recently emphasized by Binmore (2009), the founder of subjective probability theory, Savage (1954), explicitly reproduces Keynes’s distinction in arguing that the theory should be expected to fail in “large worlds” where important probabilities cannot rationally be quantitatively estimated. I also wish that Maas had not recycled the myth that the Allais paradox “remains particularly tenacious” and “cannot be resolved within the paradigm of present-day economics” (125). In fact, the weight of experimental evidence suggests that the choice behavior that generates the paradox falls to low frequencies that are compatible with generally expected stochasticity of choice if payoffs are both real and of magnitudes with which subjects likely have experience, rather than hypothetical and fantasy-sized (Conlisk, 1989; Harrison, 1994; Burke et al., 1996; Fan, 2002; van de Kuilen and Wakker, 2006). Furthermore, if subjects presented with Allais’s choices interpret the interaction strategically—which, in light of the strangeness for most people of being given such choices, would hardly be an outlandish thought—then the Allais choice pattern doesn’t necessarily violate expected utility theory in the first place (Kadane 1992).