1How should economists connect economic theories with observation and measurement? Amanar Akhabbar’s book *Wassily Leontief et la science économique* investigates one specific view about this fundamental issue. The narrative thread of the book is Leontief’s work and his views on the methodology and epistemology of economics: a 50-year long constant call for (and practice of) a closer articulation between measurement and theory. For Leontief, Akhabbar argues, economics should aim at building “empirical meaning” for theoretical concepts. In a nutshell, economics needs to establish unambiguous, direct relationships between measurement and theoretical concepts. The book permanently puts Leontief’s case in the context of the methodological debates of the time about theory and measurement and compares it with alternative views. This encompasses notably three key moments for the evolution of the discipline: the Vining-Koopmans (or “measurement without theory”) controversy in the 1940s (the book is mostly focused on this episode); the critique and crisis of (macro)econometrics in the 1970s; and finally, the most recent debates about contemporary dynamic stochastic general equilibrium (DSGE) macroeconomics.

2Given the ambition of the book, besides those readers interested specifically in Leontief’s thought, those who have a broader interest in the history of macroeconomics and in the methodology and epistemology of economics will find Akhabbar’s work highly stimulating and insightful. Similarly, although the book is published in French, its main findings will hopefully find an echo beyond the community of French-speaking economists and historians of economics.

3The main claim of the book is that Leontief held a distinctive view about the articulation between economic theory and empirical observation. Leontief’s “operationalist” view (as Akhabbar calls it) hinges on two ideas: (i) measurement and observation are providing economic concepts with meaning and purpose; (ii) all theoretical concepts should be (re)shaped in such a way that their formulation allows for direct measurement.

4Moreover, Leontief’s view was not just a “positive” methodology, but also a critical methodology. Indeed, Leontief’s distinctive articulation of theory and measurement was rooted in his conscious quest for a “third way”, which should be an alternative to both the Cowles Commission “hypothetico-deductive” (macro)econometric approach (14) and what Leontief called “the radical inductivism” of NBER (*ibid*.). This vision was embodied in Leontief’s work, particularly in his “input-output analysis” framework, for which he gained wide recognition (ultimately, the Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel in 1973). Indeed, the input-output analysis provides a distinctive way of combining and articulating statistics and data collection (the input-output tables) with mathematical modelling inspired from the general equilibrium theory (the Leontief model).

5The book follows Leontief’s intellectual journey starting from the 1920s, highlighting some crucial episodes: his early acquaintance, scepticism, and finally rejection of econometrics and macroeconometric modelling à la Cowles Commission (Chapter 1): the development of the “input-output” analysis (Chapter 2 and 3). Chapter 4 and 5 constitute the actual core of Akhabbar’s book, where he provides his most general discussion of the methodology underlying Leontief’s input-output analysis. Furthermore, Akhabbar illustrates the distinctive character of such a methodology through a discussion of the application of the input-output method to international trade and the resulting “Leontief paradox” (Chapter 6). The last chapter goes back to Leontief’s use of the input-output framework to discuss economic dynamics. Finally, appended to the book is a re-edition and a comment by Akhabbar of the French translation (by Marc Savona) of Leontief’s “Mathematics in Economics” (1954). This Leontief’s talk, given in 1953 to the American Mathematical Society, constitutes a genuine manifesto of his views about economic methodology.

6Chapter 1 begins by taking us back to 1927, when the young Leontief started working at the Kiel Institute for the World Economy. As it has been abundantly pointed out by historians, these inter-war years were the early times of new methods for economics: traditional economic analysis was combined with mathematics, models, national accounting, and probability theory, in what Alain Desrosières (to whom Akhabbar dedicated his book) called “a late encounter between statistics and economic theory” (2000). During this period (and actually for quite a long time after that), the study of the business cycle was the frontline for the struggles among different visions on the measurement-theory relation. Among other reasons, this relates to the fact that the collection and analysis of statistics was first organized on a large scale through institutes devoted to business cycle analysis—such as the National Bureau for Economic Research in the United States, but also at Harvard (where Warren Pearson developed his “barometers”) or in Moscow (where Kondratiev developed his analysis of the business cycle), just to name a few. Leontief participated to this dramatic renewal of methods; his econometric estimation of demand and supply curves, undertaken at the Kiel Institute, was published between 1929 and 1932 as a series of articles in German.

7Nevertheless, Leontief abandoned this research path right afterwards. This first chapter of Akhabbar’s book is entirely directed at explaining this turn, and the reasons and arguments for Leontief’s scepticism toward econometrics (51-53). In both his early writings (in particular his controversy with Ragnar Frisch, 45) and retrospective works (Appendix), Leontief’s main line of argument focused on the “tautological” character of structural econometrics à la Tinbergen. In his 1954 methodological manifesto, Leontief summarizes this concern as follow:

As long as a theoretical assumption is logically embedded in statistical analysis, it is almost impossible to verify this same assumption via a comparison with the results obtained by this same investigation (Leontief, 1954, 7)

8The structural forms of econometric models are one of the main targets of this criticism; according to Leontief, these are simply inferred from theory and not justified by any direct observation. A “vicious circle” is then prompted through the estimation of these forms (i.e. the quantification of coefficients that have no counterpart in any observable variable) and its use as a form of corroboration of the underlying theory. Henceforth, for Leontief, the pretence of econometrics to assess theories against observation is undermined by this logical flaw. He even characterizes the methodology of structural econometrics as “indirect statistical induction” or “indirect statistical inference”.

9These considerations, Akhabbar argues, motivated Leontief’s quest for his own alternative approach to economics and a different method for connecting measurement and theory. He notably pursued two main principles for economic investigation: first, economic analysis should start (and not end) with undertaking the “painstaking” data collection work, which requires discipline and dedication in order to collect more and “better” data (i.e. data that will “better”, “more directly”, correspond to theoretical concepts). However, this footwork should be disciplined by theoretical guidelines, which should provide the organizing principles for gathering information through observation and measurement. To secure this use of the theory, the economist should engage with reformulating economic theory in order to bring its basic concepts closer to a form that allows direct measurement (54). The input-output analysis developed by Leontief resulted from this twofold ambition.

10Chapter 2 and 3 set the stage for understanding the methodological foundations of Leontief’s work on input-output from 1932 onwards at Harvard. Chapter 2 carefully summarizes the logic of the input-output analysis. Those readers who are not entirely familiar with Leontief’s work are not left behind, since the exposition in this chapter is highly pedagogical. Incidentally, those who are interested in integrating input-out analysis to their teaching of macroeconomics will find here some great material. That being said, the bottom line of the chapter is to illustrate how Leontief’s input-output reflected his distinctive methodology, as outlined in the previous chapter. On the one hand, the analysis is structured along the fundamental lines of the general equilibrium theory (i.e., a set of equations defining equilibrium prices and quantities for many interdependent branches and households in the economy); on the other hand, theory is disciplined and marshalled by the needs of data collection, which is actually the starting point of the analysis (and not its final stage). Finally, Leontief’s model is a computable general equilibrium model, and the whole apparatus was developed in the perspective of being integrated to calculators and computers, with the help of relevant applied mathematics. (The book deals very quickly with computerization, although it seems quite essential to Leontief’s project; besides, recent developments and debates about disaggregation of macroeconomic models—and the rise of agent-based modelling—could be purposefully nourished by a more in-depth investigation historical investigation of this aspect.) Chapter 3 develops the analysis of these three dimensions. It clarifies how Leontief’s re-elaboration of general equilibrium theory is sourced both in the Ricardian and the Russian tradition; it outlines how the technical coefficients are directly measured through ratios between (directly observed) quantities of inputs, and how this is distinct from econometric “indirect statistical inference”. The combination of these three dimensions makes the unique character of the input-output framework, which works both as a “measurement device” (organizing empirical observations through tables) and as a “mediator” between measurement and theory.

11After having built a vivid and clear picture of Leontief’s approach, Akhabbar develops in Chapter 4 and 5 a more systematic discussion of the unique character of this approach when compared to alternative methodologies. Chapter 4 takes us back to the Vining-Koopmans controversy “measurement without theory”. After a brief presentation of the main lines of the debate and the surrounding context (99-102), Akhabbar illustrates how Leontief’s position fits within this debate. Let me add here a disclaimer for those historians that are familiar with the Vining-Koopmans controversy: the author gives a very short and synthetic account of this debate, taking the risk of cutting short on some nuances in the arguments and positions. It is not unlikely that this specific part of the book would raise some scepticism, especially among those who know well the over-abundant historical literature on this episode. That being said, if one considers the overarching narrative of the book itself, the approximation seems reasonable, to the extent that it illustrates Leontief’s views, notably, his reject of NBER’s “radical inductivism”.

12Leontief agrees with Koopmans, against Vining, about the cornerstone role that economic theory should play in economics. Furthermore, both Leontief and Koopmans bear in mind similar conceptions of economic theory as mathematical modelling devices, in the style of neoclassical or general equilibrium approach (102-104). This agreement is also clearly reflected in the great success, within the members of the Cowles Commission, of Leontief’s intersectoral models embedded in the input-output analysis (104-109). Despite agreeing with Koopmans on the importance of mathematical modelling in clarifying theoretical debates and producing consistent analytical frameworks, Leontief expressed in several occasions his deep concerns about the path taken by Koopmans and the Cowles Commission. For Leontief, their views of economics had been drifting, indeed, towards a “theory without measurement”, an over-investment in mathematical rigour at the expense of empirical work. This is also an important point raised by Leontief (1954): economics suffered from a chronic lack of empirical foundations, which then failed to “suppor[t] the overachieving super-structure of economic theories” (112). These repeated criticisms by Leontief against the Cowles Commission led to harsh replies and attacks against some aspects of his work (109-112).

13Chapter 5 develops further on Leontief’s methodological vision, which is interpreted as a form of operationalism. In contrast with the Cowles Commission hypothetico-deductive method (where it is admitted that some concepts, axioms, and assumptions are not observable), Leontief championed an approach to theory constrained by a criterion of direct observability. In other words, no theoretical concepts should be admitted into economic analysis before a “factual inquiry” could bring to its direct observation and/or measurement. For Leontief, the development of economics in the 20th century was unfortunately characterized by the opposite process, with theoretical concepts outnumbering observations and measures. This view, as mentioned by Akhabbar in the introduction of his book, echoes a famous metaphor of economics as a collection of “empty boxes” by John Harold Clapham (1922). For Leontief, this state of affairs is an obstacle to building a scientific method for economics. As he argues in his methodological manifesto:

Economics, mathematical economics, in particular, acquired very early in its development the attitudes and manners of the exact empirical sciences without really having gone through the hard school of direct, detailed factual inquiry. Possibly it will do us good to be sent down in order that we may catch up with the experience we have missed. (Leontief, 1954, 222)

14Leontief conceives his own input-output framework as such a “catch up with experience” (*ibid*.), “filling the boxes” of the general equilibrium theory.

15In chapter 6 we continue the chronological route, following Leontief’s work in the after-war. Chapter 6 focuses in particular on the “internationalization” of the input-output analysis, which Leontief conducted relying on a growing volume of available international data. This brought him to formulate, in the late 1950s, the observation that came to be well-known as the “Leontief’s paradox”. This is based on Leontief’s input-output tables, where one can easily observe how the U.S. economy exports labour-intensive goods, despite being a capital-abundant country. This observation qualifies as a paradox since it contradicts the basic predictions of the Heckscher-Ohlin theory. The discussion of the paradox by Akhabbar emphasizes two ideas. First, the paradox (or more precisely, its making) can be replaced in the perspective of Leontief’s critical methodological stance against the hypothetico-deductive method (as developed more in abstract in chapter 5). Second, the reactions to the paradox reveal the divide between Leontief and other economists. Indeed, as Akhabbar ironically notes, there is a paradox within the paradox: despite Leontief’s forceful demonstration, the Heckscher-Ohlin theory continued to constitute a benchmark for international trade theories—though Leontief’s paradox stimulated countless departures, amendments and reappraisal. The discussion of these reactions to the paradox are particularly interesting in that it highlights the limitations (or inconsistencies) of an ideal vision of economic methodology as truly based on the sequence “theory, model, econometric corroboration” (145-150). These remarks introduce a few more general themes that are key to the conclusive remarks of the book (cf. *infra*).

16Finally, chapter 7 focuses on what is perhaps a less well-known aspect of Leontief work, namely his ambition to build an analysis of economic change, i.e. a framework able to analyze variations of prices and quantities of goods and, therefore, cycles and growth. In short, this could be broadly understood as an attempt as developing dynamic input-output models, where the change in technical coefficients triggers changes in prices and quantities. This final chapter of the book is quite short and descriptive, but it provides some interesting insights about how this part of Leontief’s work could fit with other more well-known strains of the business cycle literature, including the real business cycle approach pioneered by Finn Kydland and Edward Prescott.

17The book features no less than three conclusive chapters (“Synthesis”, “Epilogue”, and “Post Scriptum”). Besides the standard summing-up, the reader will find here some very stimulating insights, which open several interesting leads.

18The “Post Scriptum” briefly develops the historiographical perspective of the author. As the book tries to highlight, the historical evolution of economics went hand in hand with the transformation of its methodological principles, notably the changing or opposing views about the articulation of statistics, observations, modelling and economic theory. Yet, historians of economic thought typically focus on the evolution of theories alonethough there are notable exceptions. Akhabbar calls for a shift towards a type of history of economics that builds a more comprehensive picture. Akhabbar’s view fits well with several other recent contributions that make similar claims (see for instance Boumans and Duarte, 2019, for the history of macroeconomics).

19Akhabbar’s “Epilogue” is a provocative take on the “What happened to macroeconomics?” (193) question, where Akhabbar tries to put in perspective the broader evolution of the field since the 1970s—with respect to the rise of the new classical/DSGE macroeconomics. Indeed, no matter how celebrated or purposeful Leontief’s input-output framework has been, it is hard to deny that macroeconomics has taken a different path. Expect a few exceptions (such as for instance Anne Carter, to whom *Œconomia* devoted its first special issue in 2011), Leontief’s methodology has had few followers within academic ranks. Akhabbar points out two major differences to explain this fact. On the one hand, there is a sharp contrast between the use of disaggregated, multi-sector macroeconomic models (Leontief) and the macroeconomics of the “representative agent” characteristic of the DSGE approach (but also the neoclassical growth models *à la *Solow). On the other hand, there is an opposition between Leontief’s operationalism and the hypothetico-deductive method, which has been common to both the old and new neoclassical synthesis in macroeconomics. There is, however, a subtler thread in this conclusive discussion, raising doubts about how Leontief’s own view could be understood as contributing to the rise of the neoclassical synthesis (old and new).

20The reader interested in more contemporary questions will find no definitive answers in *Wassily Leontief et la science économique* (since that is obviously not the purpose); nevertheless, he will certainly appreciate the many stimulating insights that this well-documented book has to offer.