1A History of Macroeconomics from Keynes to Lucas and Beyond is an ambitious book. First (and obviously), because the scope of the book is to address the evolution of modern macroeconomic theory from John Maynard Keynes’s General Theory to Michael Woodford’s Interest and Prices (2003), which involves investigating a seventy-year history of intricate debates, involving several dimensions (theory, methodology, policy) and hundreds of heterogeneous contributions from dozens of preeminent macroeconomists. Second, because it tries to analyze this convoluted history while addressing a broader audience than historians of (macro)economic thought. Indeed, the book “is primarily addressed to those macroeconomists, be they teachers or students, who feel the need to go beyond the technicalities that provide their daily bread and butter, and wish to ponder the origin of the kind of modeling with which they are familiar” (xiii).
2Michel De Vroey fulfils his ambitious promises. The book is at once, highly complex and enjoyably readable, deeply detailed but with a panoramic perspective, historically rigorous but never pedantically erudite. The reader never gets lost during this guided tour along the tortuous paths of the history, thanks to a good balance among three registers. First, a “static” (or “factual”) appraisal of the different theories, including the necessary pedagogical reminders (definitions, formal overviews of models, references); second, a well-organized exposition (with very helpful synoptic tables and figures) of the “dynamics” of the discipline, i.e. of the debates among and within different theories; third, a clearly identified personal assessment of these theories and debates, which gives to the book a provocative, lively and stimulating tone. At the same time, De Vroey openly encourages historians to provide alternative histories, and to explore episodes and issues which he has decided to leave aside. In short, the author provides “a” history of macroeconomics.
3The main claim of the book is made very explicit in its title: modern macroeconomics has undergone two important changes to purpose, method, conceptual apparatus, and results. The first change, which took place between the 1940s and the 1960s, was the development of the Keynesian paradigm, and the so-called neoclassical synthesis; the second change, which began in the 1970s and continues to the present, is the “Kuhnian revolution” set in motion by Robert Lucas, which overturned the Keynesian paradigm and led to the development of a new one. This new paradigm is labeled by De Vroey as “dynamic stochastic general equilibrium” (DSGE) macroeconomics. Note that this contemporary label (DSGE) is used to designate Lucas’s own approach and its developments, rather than as common today, exclusively to designate models à la Smets and Wouters (2003), the latter being labeled by De Vroey the “second generation of new Keynesians”. This choice makes sense (152-153) but potentially could be quite confusing for the reader.
4The book analyzes the Keynesian and Lucasian-DSGE paradigms following the Leijonhufvud (1994) “decision tree metaphor”. The idea is that (macro)economists face many “decisional nodes” (dilemmas about which questions to address, which concepts and methods to use): different responses to a decisional node create bifurcations in the development of the discipline. This process is seen as the outgrowths of the different branches of a tree:
Any major bifurcation on the tree, that is, a new research line, starts as an original contribution, which in the beginning is like a thin new branch on a tree. Its success hinges on the attention it receives. The original work must be considered sufficiently interesting to be elaborated on, and the ensuing chain of contributions building on each other is what makes the branch sturdy. (xvi)
5One can see the developments within the same scientific paradigm as a process of going back (“backtracking”) to a previous bifurcation of the tree:
Once mature, a research track may gradually lose its momentum: puzzles arise, objections are leveled, and doubt about its validity sets in. Leijonhufvud calls what occurs then ‘backtracking,’ that is, traveling back down the decision tree to an earlier bifurcation that at the time was neglected but now seems a viable and appealing alternative. (ibid.)
6Finally, Kuhnian revolutions such as those described in De Vroey’s book, can be seen, using this metaphor, as a process of backtracking toward previous, “major” bifurcations occurring in a fundamental (or “distant”) decisional node.
7The most fundamental decisional node in the book (although not the only one) is what De Vroey calls the “Marshall-Walras divide”. Hence, his reading of the evolution of macroeconomics is basically a matter of a bifurcation from this node: Keynesian macroeconomics represents the Marshallian branch of the tree, Lucasian-DSGE macroeconomics the Walrasian branch (encompassing Walras’s vision and neo-Walrasian approach à la Arrow-Debreu-McKenzie). However, the dimensions within this divide go beyond the traditional conflict about how to address complexity, i.e. the conflict between partial and general equilibrium. Indeed, in De Vroey’s conception, the Marshall-Walras divide involves methodological priorities (external or internal consistency), the use of mathematics and microfoundations, price setting, and coordination among economic agents. De Vroey proposes a very peculiar understanding of what is meant by Marshallian or Walrasian along all those dimensions. Certainly, his vision could be criticized in many ways, and surely will be: still, it is a complex vision, subtle, and far from being binary or simplistic. Most importantly, the book shows that such a vision of the Marshall-Walras divide is an operational, useful framework for building a global perspective on the history of macroeconomics. Concerning this use of the Marshall-Walras divide, note that A History of Macroeconomics from Keynes to Lucas and Beyond is very close to De Vroey’s previous book (De Vroey, 2009, published only in French). However, I think that the content and the insights in the new book are richer and more convincing. My feeling is that this is due precisely to a substantial refinement of the conception of Marshall-Walras.
8The book is in three parts. The first two address the two paradigms in the history of macroeconomics, the Keynesian and the Lucasian-DSGE. Both parts are built following the perspective of the Leijonhufvud decision tree: each part starts by characterizing the “original contribution” and the “decisional node” it took, then follows the “ensuing chain of contributions building on each other” leading to a “sturdy branch” (a paradigm). Finally, “puzzles, objections” and the “backtracking” processes are analyzed. The third part of the book provides a three-step “broader perspective”. First, it develops a summary of the role of the Marshall-Walras divide in the history of macroeconomics (chapter 19); second, it accounts for some debates in the wake of the 2008 crisis (chapter 20); third, it offers some considerations about the future of the discipline (chapter 21). I will not comment in detail on this third part (which is quite short compared to the other two) because the first point (the Marshall-Walras divide) is a summing up, and because the second point, however interesting and stimulating, goes beyond the fundamental scope of the book (if we follow the principle that “history can be written only after the dust has settled”, 386).
9Part I accounts for the development of Keynesian macroeconomics and neoclassical synthesis (chapters 1 and 2), and the partial “backtracking” made by alternatives or critical views within this paradigm (chapters 3 to 7).
10The starting point of the book is Leijonhufvud's (1968) distinction between Keynesian economics and the economics of Keynes. Chapters 1 and 2 address this transformation. In chapter 1, De Vroey makes three basic claims. He characterizes Keynes’s contribution as an attempt to explain involuntary unemployment (seen as a real world phenomenon) otherwise than by wage rigidity, in order to justify stabilization policies. Second he claims that Keynes developed his explanation in a Marshallian framework. Third he claims that such a framework is methodologically incompatible with the notion of involuntary unemployment, except for the “trivial case” of ad hoc wage rigidity. According to De Vroey, this is the main reason why Keynesian macroeconomics à la Hicks (1937) reverses the perspective in Keynes’s original purpose, introducing nominal wage sluggishness as the major explanation for involuntary unemployment. Chapter 2 starts by reinforcing this conclusion, showing how Hicks’s model became the IS-LM model, the hallmark of the Keynesian paradigm. De Vroey then discusses two powerful extensions of the IS-LM model: a qualitative one proposed by Franco Modigliani, and a quantitative one proposed by Lawrence Klein. The subsequent development of the Keynesian paradigm consists of the introduction of the “Phillips curve” which provides both a theoretical account of price-wage formation (Lipsey, 1960), and also a clarification of policy issues (Samuelson and Solow, 1960). The last peculiar characteristic of the Keynesian paradigm is the notion of neoclassical synthesis, i.e. an “eclectic view” of the discipline which splits scopes and methods between Keynesians (short term, Marshallian approach) and Classics (long term, Walrasian approach). In De Vroey’s view, this is indeed not an actual synthesis in the sense of a convergence between two different approaches toward a unified vision. According to the usual metaphor, neoclassical synthesis was merely agreement about a legitimate bifurcation in two branches of the tree. Chapter 3 presents the Keynesian contributions challenging this loose notion of a synthesis. These contributions are a “backtracking” toward the Marshall-Walras divide, aimed at producing a real synthesis between the two approaches. Roughly, this was an attempt to express Keynesian concepts (involuntary unemployment, sluggish prices) within a Walrasian framework. Such attempts were proposed especially by Klein (1950) and Patinkin (1956). Following the decision tree metaphor, Patinkin’s approach was not very fruitful in terms of developments apart from his convoluted influence on Leijonhufvud (1968), and Clower's (1965) disequilibrium theory or “non-Walrasian equilibrium theory” provided another misleading label. These two alternatives research lines within the Keynesian paradigm are analyzed in chapters 6 and 7. De Vroey, while commenting the work of Patinkin in chapter 3, provides a clear explanation for this failure:
My inquiry in this book has led me to believe that any attempt at a Keynes-Walras synthesis is doomed to fail at least in so far as its Keynesian component has involuntary unemployment, defined as individual disequilibrium, at its core. (84)
11Chapters 4 and 5 discuss the criticisms of Keynesian economics made by Milton Friedman and Edmond Phelps. Chapter 4 provides a detailed account of Friedman’s work, as a methodologist, as a monetarist, and as an “ideologist”. This allows De Vroey to introduce some considerations about the frontiers between theory and “ideology” (understood as a “vision held about the ideal way of organizing a society in its economic dimension” (87)). Basically, De Vroey claims that Friedman’s distinction between positive and normative economics is inconsistent with the state of affairs in macroeconomics: indeed, producing policy advice is a constitutive scope of the discipline. On this premise, De Vroey derives the following methodological roadmap:
If the ideological dimension cannot be erased, what must be done is to circumscribe its impact. That is, the profession should strive for a situation where it would not matter whether a given argument is underpinned by an ideological motivation because methodological criteria exist to gauge its validity whatever this motivation. (88)
12Building on this idea, De Vroey suggests three methodological criteria to “circumscribe” the impact of ideology:
The use of mathematical modeling as opposed to modeling in prose is a first step in this direction. Yet, more is needed, namely, making the view adopted on the matter of the relationship of models and reality explicit, defining guidelines for econometric work, and, finally, enunciating the standards to be respected for sound theory construction. All these choices are definitely a matter of convention and may always be impugned through a scientific revolution. (ibid.)
13It is important to quote these considerations in extenso, because they constitute an essential ingredient of the personal assessments made by De Vroey, and not only about Friedman but also about Lucas, the RBC (real business cycles), and contemporary debates.
14Chapter 5 presents the debate on the Phillips Curve and the “natural rate of unemployment”. This new concept was formulated simultaneously (and independently) by Phelps (1967; 1968) and Friedman (1968). According to De Vroey, Friedman (1968) was the more discussed and influential of the two at the time: however, his criticism was formulated within a Marshallian framework, and consequently, within the same branch of the decision tree as the Keynesian paradigm. Thus, Friedman’s work was not a scientific revolution although it demolished some pillars of the Keynesian paradigm. This is the general assessment (in chapter 8) which concludes this first part of the book. At the end of the 1960s, both the open criticisms (mainly, Friedman) and the alternative lines within the Keynesian paradigm (Patinkin, Clower, etc.), constituted only a partial backtrack. Indeed, these contributions did not go back as far as the fundamental decisional node, i.e. the Marshall-Walras divide. According to De Vroey, this is what distinguishes Friedman from Lucas, and what distinguish the debates within the Keynesian paradigm from a change in the paradigm.
15Part II of the book investigates the Kuhnian revolution set in motion by Lucas (chapters 9 to 11), the reactions to it (chapters 12 to 14), and finally, its transformation into a “sturdy branch”, thanks especially to the RBC approach (chapters 15 to 17) and the “second generation” of new Keynesians, or the “new neoclassical synthesis” (chapter 18).
16The starting point of this second part of the book is Lucas’s contribution to macroeconomics. Again, De Vroey's reading of it is combined with the idea of a radical “backtrack” to the Marshall-Walras divide: “Lucas [made] macroeconomics neo-Walrasian, while traditional Keynesian macroeconomists followed Keynes by adopting a Marshallian methodology” (174). The analysis of the Lucasian revolution, which is in three steps, adopts this line: chapter 9 describes Lucas's “theoretical journey”, i.e. the evolution of his work during the 1970s; chapter 10 describes Lucas’s “methodological breach” with respect to the Keynesian paradigm; and in chapter 11, De Vroey provides a personal assessment of Lucas's methodology. The depiction of Lucas’s “theoretical journey” constitutes a narrative on the emergence of the main concepts of the Lucasian-DSGE paradigm, and Lucas’s fierce contrast with the concepts of the Keynesian paradigm. De Vroey emphasizes five main changes introduced by Lucas to macroeconomics: a new scope (the study of business cycles in general but excluding “dramatic events” such as the Great Depression), a new purpose (a “better-founded discussion of economic policy” (152)), a unified framework (the microfoundations of macroeconomics within a neo-Walrasian general equilibrium model), a new concept of equilibrium (intertemporal, dynamic equilibrium), and a new set of methodological rules for model building. This last is discussed in detail in chapter 10. De Vroey’s account of Lucas’s methodology brings to light an interesting tension between (in short) two methodological rules. On the one hand, Lucas claims that the assumptions made in the theory are not propositions pertaining to the real world but simply modeling devices to ensure the internal consistency of models. On the other hand, Lucas supports the view that macroeconomics should be an applied field, producing policy recommendations based on modeling. Thanks to unpublished material, De Vroey is able to suggest a possible underlying solution to this tension: the notion of an analogy between real world phenomena and “phenomena” produced by the model.
17These insights into Lucas’s methodology, in my view, constitute the real core of the book for at least two reasons. First, and this point mostly concerns the audience of the historian of macroeconomics, because these three chapters suggest a very comprehensive view of a surprisingly neglected topic (Lucas’s methodology), and do so by drawing on new material (especially drafts and other unpublished writings, collected by the Lucas Archives at Duke University). This makes those three chapters the cornerstone for further historiographic research, a thesis about recent macroeconomics that historians will certainly passionately debate. Second, and this will be of most interest to macroeconomists, because Lucas’s contribution represents the “fulcrum” of the (r)evolution of the discipline from Keynesian macroeconomics to its today standards. The influence of Lucas’s contribution on contemporary macroeconomics is often evoked by macroeconomists themselves to explain “why” they practice macroeconomics as they do (see e.g. Chari, 1998). However, this appeal to Lucas’s methodology more often constitutes a petitio principii than a real historical perspective on Lucas: these three chapters provide a very useful analysis of which aspects of Lucas’s contribution have been relevant and influential (and which have not).
18Chapters 12 to 14 investigate different attitudes to Lucas’s methodology which can be summarized roughly as either a total rejection or a partial acceptance. Chapter 12 is about total rejection but from two very different perspectives. On the one hand, Christopher Sims suggests a research line (the VAR approach) diverging from both the Keynesian and Lucasian-DSGE paths; on the other hand, Keynesians such as Robert Gordon, Arthur Okun and James Tobin, attempt to impose methodological and empirical “damage control” (to use De Vroey’s words) on the Phillips Curve and the notion of involuntary unemployment. Notably, chapter 13 deals with the work of the “first generation of new Keynesians”. This heterogeneous group is identified with the contributors to a collected volume edited by Mankiw and Romer (1991)—which gathers four distinct approaches (implicit contracts, staggered wages, efficiency wage, and menu costs). The first generation of new Keynesian writings is characterized by De Vroey as a partial acceptance of Lucas’s methodology (this includes, mainly, the idea of microfoundations) but also a defense of some Keynesian notions (mostly unemployment, wage sluggishness, and the non-neutrality of money). The same idea of a partial acceptance of Lucas’s methodology is applied to the three contributions (Diamond, Hart, Robert) examined in chapter 14. In this case, the distinctive element is rejection of the Walrasian auctioneer and development of alternative models of trade technology (search behavior, strategic behavior).
19Chapter 15 presents the RBC approach. It starts by discussing Kydland and Prescott's (1982) seminal paper, and then proceeds to address subsequent developments (especially related to calibration, and reinterpretation of the neoclassical growth model) which led to the RBC baseline model (King, Plosser and Rebello, 1988). De Vroey emphasizes the convergences and divergences between the RBC and Lucas’s methodology. The important insight in this chapter (and developed and completed in chapter 17) is on the divergences (the convergences seem quite evident). The RBC model clearly differs from Lucas's method in its the empirical method and its real explanation of the cycle. There are also two major inconsistencies with respect to Lucas’s methodology: a “poor explanation” of the cycle, and violation of the “non-exploitation” principle related to policy conclusions. The first problem arises from the “distance” between the explanans (RBC model) and the explanandum (fluctuations in the market economy). According to De Vroey, RBC models are nothing more than a “colossal ‘as if’ story” (274). Hence, this bold methodology should be handled with more care than what is to be found in RBC economists’ writings. This should apply especially to its application to “explanation” of the cycle and to the policy conclusions (following the Lucas’s “non-exploitation” principle).
20Chapter 16 describes the criticisms leveled at RBC theorists, their reactions, and the way eventually they were able to integrate some of these critiques into their framework. This leads directly to the conclusions in chapter 18, which closes De Vroey's history. In chapter 18, he exposes the developments made by the “second generation” of new Keynesians (the DSGE approach in contemporary terminology). This development is characterized as a “New Neoclassical Synthesis”, a true synthesis in the sense of a real integration of two frameworks (the RBC and the “first generation” new Keynesians). Within the new synthesis, the RBC modeling methodology continues to be a “sturdy branch” but there is a partial backtracking in respect of the role of perfect competition, money neutrality, and price-wage rigidity.
21My review began with a quite enthusiastic assessment of the book—especially with respect to its ambitious purpose. I conclude with some personal disappointment. As I have already said, in the introduction, De Vroey is quite explicit about the exact scope of his book:
I have chosen to give more emphasis to theoretical aspects than to empirical ones. My work is internal history and leaves aside most of the contextual dimension. I deal neither with pre-Keynesian macroeconomics nor with heterodox theory. Moreover, although macroeconomics as it is understood at present encompasses both growth and business fluctuations, I will say nothing about the former (xvi).
22Hence, among other elements, empirical aspects—which I interpret here in a broad sense to include all the measurement/observation procedures pertaining to real world or models—should be excluded from De Vroey’s main picture. However, empirical aspects occupy a great deal of space in this book. Also, they are not integrated into De Vroey’s narrative as a generic element, aimed simply at giving some intuition of the complete picture (as in the case for instance of the contextual dimension). Empirical aspects play an active role in De Vroey's history narrated in this book. For instance, they are crucial for determining the development of the two paradigms: indeed according to De Vroey, the Keynesian and Lucasian revolutions owed much of their success (in the sense of Leijonhufvud's decisional tree) to the development of a consistent empirical framework:
While models à la Lucas could mobilize only a tiny fraction of macroeconomists, Kydland and Prescott’s applied research program provided the bread and butter to legions of macroeconomists (both top-notch and more run-of-the-mill) for more than a decade. This is what it takes to have a successful revolution. Therefore, I like to regard the relationship between Lucas, and Kydland and Prescott, as mirroring that between Keynes and his immediate successors such as Hicks, Klein, and Modigliani. What would have happened to The General Theory if it had not been transposed into the IS-LM model and if Klein had not extended it into an econometric framework? (281)
23Another example of the importance of empirical aspects in De Vroey’s account of the history of macroeconomics is the separation made between ideology and theory. I recalled above that the “guidelines for empirical work” in De Vroey’s conception, are a main requirement for establishing such separation. That means that empirical aspects underlie most of his assessments of various theories throughout the book.
24To sum up, the book assigns a crucial role to empirical aspects while refusing to discuss (i) how they evolved, and (ii) how this evolution interacted with the theoretical aspects. In short, changes in empirical methods, as they are represented in the book, look much like “exogenous technological progress” in neoclassical growth models which in my view, is a disappointing picture. A better one would have been a kind of Leijonhufvud decision tree, specific to empirical aspects. The evolution of econometrics in general, and macroeconometrics in particular, could have been appraised using this same framework, although it would have required reconsideration of the decisional node and especially, abandonment of the Marshall-Walras divide as the fundamental bifurcation. This would be an absolute requirement since Lucas’s emphasis on empirical work represents his “major departure from the Walrasian approach” (346). What seems even more important is the idea that the decision tree of empirical developments overlaps the decision tree of theoretical development, i.e. the two aspects interact very closely, and they influence each other—which seems quite clear from De Vroey’s remarks discussed above. Hence, instead of providing a separate analysis of the decision tree of theoretical developments of macroeconomics, De Vroey would have gained from parallel consideration of the decision tree of empirical techniques, and the interactions between them. This would have provided a more complex and complete perspective, which would have enhanced the interdependencies among the empirical tools, theoretical apparatus, and modeling practices. In short, De Vroey would have gained from promoting more of a Walrasian than a Marshallian spirit in his approach to the history of recent macroeconomics.