The Macroeconomist as Scientist and Engineer

Mankiw’s thesis is that macroeconomics has evolved through the efforts of two types of macroeconomist – “those who understand the field as a type of engineering and those who would like it to be more of a science” – and that one recurring theme in the field’s history is the interaction, “sometimes productive and sometimes not,” between the two (Introduction). Engineers are “first and foremost, problem-solvers,” while “the goal of scientists is to understand how the world works.” Mankiw is explicit that this is not a story of “good guys and bad guys” or “deep thinkers and simple-minded plumbers” – both mindsets are intellectually demanding – but he states his own view that “the discipline would advance more smoothly and fruitfully if macroeconomists always kept in mind that their field has a dual role.”

The subfield “was born not as a science but more as a type of engineering,” created to address the Great Depression, in which U.S. unemployment reached 25 percent in 1933 and real GDP was 31 percent below its 1929 level (The Keynesian Revolution). Keynes’s General Theory became the focal point of the response, but Mankiw notes that reading it today is “both exhilarating and frustrating” because, despite its scope, “it somehow seems incomplete as a matter of logic.” A generation of economists – Hicks (1937) and then Modigliani (1944) – built the simpler IS-LM model to formalize Keynes’s argument, and applied econometricians such as Klein built large policy models (the Wharton, DRI, and MPS models) that, while differing in detail, shared an essentially Keynesian IS-LM-Phillips-curve structure. As a matter of science, Mankiw judges the Keynesian revolution “a remarkable success”; as a matter of engineering, Keynes and the early Keynesians were “heavily involved in offering policy advice,” with Tobin, Solow, and Eckstein working at the Council of Economic Advisers and the 1964 Kennedy tax cut emerging “in many ways” as a direct product of the Keynesian consensus.

The first wave was monetarism, led by Milton Friedman, whose permanent-income hypothesis undercut the Keynesian consumption function and whose work with Schwartz traced business-cycle instability to “inept monetary policy” rather than to private “animal spirits,” implying that policymakers should follow simple rules (The New Classicals). Friedman’s 1968 AEA address (with Phelps 1968) also attacked the Phillips-curve tradeoff, arguing it would not hold in the long run once expectations adjusted. The second wave was the rational-expectations revolution: Lucas’s (1976) “Econometric Policy Evaluation: A Critique” argued Keynesian models were useless for policy analysis because their estimated relationships would break down under a different policy regime, Lucas (1973) proposed a business-cycle theory in which monetary policy matters only by surprising people, and Sargent and Wallace (1975) concluded that systematic stabilization policy is “doomed to failure” because rational agents cannot be systematically surprised. The third wave was the real business cycle theory of Kydland and Prescott (1982) and Long and Plosser (1983), which, unlike its new classical predecessors, dropped any role for monetary policy and attributed fluctuations to technology shocks and intertemporal substitution in consumption and leisure. Mankiw notes that proponents of this tradition have since “backed away” from the real-versus-monetary distinction and increasingly describe their work under the label “dynamic stochastic general equilibrium” theory.

The first wave was general-disequilibrium theory (Barro and Grossman 1971; Malinvaud 1977), which took wages and prices as exogenously fixed and analyzed the “Keynesian regime” in which both goods and labor markets show excess supply. The second wave used rational expectations in non-market-clearing models, partly to answer Sargent and Wallace by showing systematic monetary policy could still stabilize the economy under staggered contracts (Fischer 1977; Taylor 1980), though this work rested on a labor-contract assumption “hard to square with microeconomic principles.” The third wave sought microeconomic reasons for sticky wages and prices – menu costs, efficiency wages, near-rational price setters – with Mankiw (1985) and Akerlof and Yellen (1985) showing that firm market power can make a sticky-price equilibrium privately rational but socially costly, Blanchard and Kiyotaki (1987) attributing part of the gap to an aggregate-demand externality, and Ball and Romer (1990) showing strong complementarity between real and nominal rigidities. Mankiw argues the three waves are “more related and complementary than they seemed at the time”: the third wave’s market-power/efficiency-wage mechanisms help explain why the “Keynesian regime” of the first wave is not just one possible outcome but the typical one.

As a matter of science, Mankiw judges the new Keynesian program broadly successful – “we understand how markets interact when there are price rigidities, the role that expectations can play, and the incentives that price setters face” – while cautioning that “as a participant, I cannot claim to be entirely objective.” As a matter of engineering, “the judgment here must be less positive.” He also notes a sociological point: despite academic hostility between the camps, new Keynesians, like earlier Keynesians, were far more likely than new classicals to take leave from academia for government policy jobs (he lists Fischer, Summers, Stiglitz, Yellen, Taylor, Clarida, Bernanke, and himself), which he reads as evidence the split is “fundamentally” one between “pure scientists and economic engineers” rather than between political left and right.

Mankiw treats the public dispute between Solow and Lucas – Solow’s 1980 AEA address calling it “foolishly restrictive” for new classicals to assume away wage/price rigidities, and Lucas’s 1984 remark (via Klamer) that Solow had “never tried to come to grips with any of these issues except by making jokes” – as reflecting genuinely differing views of the field’s goal, with “each have a point”: new classicals offered more analytic rigor, while Solow objected to the “patent lack of reality” of market-clearing assumptions. He argues the resulting “vitriol among intellectual giants,” while attention-grabbing, “was not healthy for the field,” and that it led many younger economists in the 1990s to avoid taking sides by moving into long-run growth research instead – one of several reasons (alongside Romer’s 1986 growth papers, new Penn World Table data, and the long 1990s expansion) that growth, rather than business cycles, dominated 1990s macroeconomics.

The new synthesis – named by Goodfriend and King (1997) and most fully developed in Woodford’s (2003) treatise – takes the dynamic-stochastic-general-equilibrium tools and optimization-based microfoundations of the new classical models, and combines them with the new Keynesian device of monopolistically competitive firms that change prices only intermittently (yielding a “new Keynesian Phillips curve”), producing a dynamic general equilibrium system that deviates from Pareto optimality because of sticky prices. Mankiw notes this synthesis has already been applied to monetary-policy research (Clarida, Gali, and Gertler 1999; McCallum and Nelson 1999).

Mankiw offers a deliberately “less sanguine” alternative reading: that the new synthesis is “not so much a synthesis as a truce between intellectual combatants, followed by a face-saving retreat on both sides,” since its core feature – a dynamic general-equilibrium system with nominal rigidities – “is precisely what one finds in the early Keynesian models” of Hicks and the postwar model-builders. On this reading, the new classicals “promised more than they could deliver” (they aimed to discard Keynesian theorizing entirely, but instead helped build tools now used for sticky-price models resembling what they campaigned against), while new Keynesians “can be criticized for having taken the new classicals’ bait” and pursued a research program that, per Krugman’s (2000) assessment which Mankiw endorses as containing “some truth,” can explain how price stickiness could happen but has not produced “useful predictions about when it happens and when it does not.”

Mankiw’s primary evidence is Laurence Meyer’s memoir of his 1996-2002 Federal Reserve governorship, which Mankiw reads as showing that “recent developments in business cycle theory, promulgated by both new classicals and new Keynesians, have had close to zero impact on practical policymaking” – Meyer’s analysis “shows no traces of modern macroeconomic theory” and “would seem almost completely familiar to someone…schooled in the neoclassical-Keynesian synthesis that prevailed around 1970.” He stresses this is “typical,” not idiosyncratic to Meyer, among economists who have held top central-bank positions, and attributes it not to policymakers’ ignorance – Fed and CEA staff are drawn from top research universities – but takes the lack of practical uptake as itself “prima facie evidence” that the research has limited use for engineering purposes, even while it may succeed as science.

Mankiw identifies two problems with the “self-congratulatory view” that Kydland and Prescott’s (1977) time-inconsistency work explains the rise of independent, inflation-targeting central banks and the resulting good inflation outcomes: first, the observed institutional changes are “at best loosely connected” to the theoretical issues – an independent central bank is not a rule-bound one, and even inflation targeting is “closer to a statement of intentions…than it is a commitment to a policy rule” (citing Bernanke’s “constrained discretion”); second, Ball and Sheridan (2005) find across a large country sample that adopting inflation targeting does not help explain the move to low, stable inflation, since monetary policy improved in both adopting and non-adopting countries. As a direct illustration, Mankiw cites Alan Greenspan – whom Blinder and Reis (2005) call “the greatest central banker who ever lived” – who explicitly rejected policy rules in favor of discretion (“rules by their nature are simple, and when significant and shifting uncertainties exist…they cannot substitute for risk-management paradigms”), yet produced low and stable inflation throughout his tenure, which Mankiw reads as proof, “contradicting Kydland and Prescott,” that central banks “can produce desirable outcomes while wielding substantial discretionary powers.”

Drawing on his own tenure as CEA chairman, Mankiw reports that the model the Council of Economic Advisers actually used to quantify the employment effects of the Bush tax cuts was the commercial macroeconometric model built by Macroeconomic Advisers (the firm Laurence Meyer ran before joining the Fed), a model that predated Mankiw’s arrival, had been used by both parties “for almost two decades,” and – like the structurally similar Federal Reserve FRB/US model – is a “direct descendent” of the 1960s Klein-Modigliani-Eckstein modeling tradition, with “minimal influence” from new classical or new Keynesian research. He also notes the policy’s short-run rationale, as President Bush articulated it (“When people have more money, they can spend it on goods and services…it means somebody is more likely to be able to find a job”), is “quintessentially Keynesian” in the old IS-LM sense, even though the tax design’s long-run rationale (reducing the double taxation of corporate capital, moving toward consumption taxation) drew on an older public-finance literature (Diamond and Mirrlees 1971; Atkinson and Stiglitz 1976; Feldstein 1978; Chamley 1986) unrelated to the new classical/new Keynesian debates.

Mankiw argues that despite forty years of new classical and new Keynesian research, “the basic theory taught to undergraduates is some version of aggregate demand and aggregate supply, and the basic theory of aggregate demand is the IS-LM model” in every leading intermediate textbook (Gordon; Hall and Taylor; Dornbusch and Fischer a generation ago; Blanchard; Abel and Bernanke; and Mankiw’s own text today), all written by economists trained in the Samuelson-Solow MIT tradition. He treats Robert Barro’s 1984 intermediate textbook – which foregrounded new classical theory and relegated Keynesian models to brief, late coverage – as “the exception that proves the rule”: it was read and admired by many macroeconomists but adopted by “many fewer” for their own students, so “the new classical revolution in pedagogy that Barro hoped to inspire never took off.” He contrasts this with the rapid, wholesale adoption of Samuelson’s Keynesian 1948 textbook, concluding that, unlike the original Keynesian revolution, “the ideas of new classicals and new Keynesians have not fundamentally changed how undergraduate macroeconomics is taught.”

Mankiw closes by invoking Keynes’s (1931) hope that “if economists could manage to get themselves thought of as humble, competent people on a level with dentists, that would be splendid” – a vision of macroeconomics as routine, reliable engineering – and judges that the leading academic developments of the past several decades “bear little resemblance to dentistry,” having had little impact on either actual policymaking or undergraduate teaching. Yet he argues the same work “can be viewed more positively” as science: new classicals exposed the limits of large Keynesian models and highlighted expectations and policy rules, new Keynesians supplied better theories of why markets fail to clear, and the resulting “tension between these two visions, while not always civil, may have been productive.” He ends by framing the emerging new synthesis as a “hopeful sign that more progress can be made on both fronts,” with “humble and competent” remaining the field’s aspirational ideals.