Since the history of econometrics became an epistemic object among historians of economics, intellectual demands for evaluating the evolution and dissemination of economic empirical tools subsequently emerged. In this book on the history of the empirical Cobb-Douglas production function (developed by Charles Cobb and Paul Douglas), Jeff Biddle provides a comprehensive and convincing case study in fulfilling these demands. Professor Biddle articulates that “this book is about the life of his [Douglas’s] most enduring creation, not the life of the creator” (p. 3); embedded in the neoclassical-econometric tradition, the lifestyle of the Cobb-Douglas production function is indeed lively and multifaceted and of great research value. While works by historians of economics often focus on the life story of economists, readers should welcome more historical contributions akin to this book, where the idea of human biography extends to the life story of non-human characters.
Using published books, reports, and journal articles as its principal sources, this book devotes a great deal of attention to documenting the historical dynamics of the Cobb-Douglas literature. Chapter 1 traces the emergence of the function in 1928 to various contemporary critiques, both friendly and unfriendly, by other economists. Chapters 2 and 3 examine those critiques from a broader perspective, including materials by, to name just a few, Douglas and his students, the Frischian Horst Mendershausen, Jan Tinbergen, and Cowlesians Jacob Marschak and William Andrews. After Douglas left academia, embarking upon a new career as a politician, the life of the Cobb-Douglas production function continued autonomously with these revisions and critiques onto its next episode. Chapter 4 demonstrates how the Cobb-Douglas function was popularized at the first stage, as shown in a survey of its use in econometrics textbooks, the Phelps Brown critique, and a history of the CES production function. Chapters 5 and 6 proceed to two crucial developments in the economics literature that were responsible for diffusions of the Cobb-Douglas function, one in micro-oriented agricultural economics and one in macro-oriented economic growth. These chapters further identify two allies, Earl Heady and Zvi Griliches, as facilitators of the Cobb-Douglas regression in each subfield (pp. 308–311). The book’s storyline is as neat as it is well-organized. The analytical style, far from being a simple review of the Cobb-Douglas literature, concentrates on the actual economic concept in operation, with supporting data behind every instance when constructing empirical knowledge. Readers are introduced to different historical set-ups and empirical practices arranged in a sensible order. Along with its clear writing style, this book is a solid and informative historical analysis for anyone interested in the history of twentieth-century economics.
This book carefully describes the empirical Cobb-Douglas production function as one of the “research programs” in the neoclassical “paradigm” that generates “heuristics.” Use of these reflect, albeit implicitly, a Lakatosian view in evaluating the history of empirical tools in economics. Based on the evidence demonstrated, readers will find no compelling reason to stand against this methodological framework. First, the Cobb-Douglas research program evolves through critiques via a learning-by-doing process, in which empirical questions shift progressively following the discovery of new theories or by replacing core assumptions. As chapters 1–3 have shown, when creating his input-output regression, what Douglas had in mind was testing the marginal productivity theory of distribution within the context of the works of John Bates Clark and Léon Walras. As the life of the Cobb-Douglas function continued, empirical questions shifted to the conditions of competitive or imperfect market equilibrium, and the marginal productivity theory was transformed into the theoretical hard core that could be measured rather than disproved. The production function was thereby transformed into a causal model of factor market equilibrium. Such a mode of progression displays the Lakatosian nature of the Cobb-Douglas program, exemplified by Douglas’s confidence in his results as argued in his 1947 AEA presidential address. Second, diffusions of the Cobb-Douglas production function hint at another Lakatosian element, that empirical research programs in economics enjoy higher (to co-opt Biddle’s term) flexibility or adaptability (p. 302). Through the case studies presented in chapters 4–6, readers are reminded that the regression framework erected by Douglas and his co-authors features multiple faces, and this standardized, neoclassical-econometric framework can be carefully transplanted to investigate other empirical questions. Efforts made by those allies, especially Heady and Griliches, indicate that the best way to keep an empirical research program alive is not to discard it immediately once anomalies occur but to live with the imperfect nature of empirical studies and to strive for revision of the program’s protective belt. To this end, the Cobb-Douglas research program is indeed a successful Lakatosian research program.
Another methodological account adopted by this book is the distinction between the rhetoric of expert judgment and mechanical objectivity in statistical inferences (pp. 84–85). A representative contrast is Ragnar Frisch’s bunch-map analysis as used by Mendershausen versus the statistical stability test used by Martin Bronfenbrenner. This episode depicts how statistical economists in the early years lacked a credible criterion in legitimizing their empirical evidence. While the neoclassical-econometric program underscored the active role of economic theories in guiding regressions, the bunch-map analysis, due to its atheoretical nature, conflicted fundamentally with Bronfenbrenner’s statistical method. Lacking an objective standard in evaluating statistical validity did sometimes complicate the matter in the 1930s but only adds to the fascination of the story in retrospect. Without the Douglas-Mendershausen debate, historians of economics would lose two intriguing pinhead graphs that characterize, in my opinion, the most unique element in the story of the Cobb-Douglas production function.
From a historical perspective, this book primarily contributes to our current understanding of the history of econometrics from the 1920s to the 1960s. On the one hand, this book documents how econometricians under the neoclassical-econometric program applied firm-level or industry-level data to solve their own theoretical puzzles. It seems to me that the development of the neoclassical-econometric program following the early demand studies (Morgan Reference Morgan1990, part II) could be separated into two interwar literatures: one from the US on the empirical production function using firm-level or industry-level data, and one from the UK on the empirical family-expenditure function using household-level data. In the former, Douglas’s The Theory of Wages (Reference Douglas1934) was aimed at the Clarkian marginal productivity theory, whereas in the latter, studies on Ernst Engel’s law (e.g., Allen and Bowley Reference Allen and Bowley1935) targeted the Hicks-Allen value theory (Hicks and Allen Reference Hicks and Roy1934a, Reference Hicks and Roy1934b) for econometric analysis (Cheng Reference Cheng2021, ch. II). Such developments in parallel indicate a potentially fruitful map of interwar econometrics calling for more dedicated historical examination. In both cases, with different theoretical targets, empirical relations, such as the Cobb-Douglas regression and Engel’s law, constantly act as a spur driving econometric practices forward. A similar case could also be found in the story of empirical consumption function inspired by John Maynard Keynes’s General Theory (Thomas Reference Thomas1989, Reference Thomas1992; Chao Reference Chao2019). Connecting this book’s contribution to histories of econometrics using household-level data would give historians a deeper understanding of the formation of interwar econometrics through various lives and faces of empirical relations.
On the other hand, this book illustrates the detailed dynamics of econometric issues raised by various attempts at the estimation of production function. The controversy at the first stage concentrated on the validity of time-series and cross-sectional studies, intertwined with differences in data availability and statistical methods. Since then, the progression of the Cobb-Douglas program has depended on other external technological factors. A critical transition in this case was the new estimation method using panel data that emerged in the 1950s, exemplified by the works of Irving Hoch, Yair Mundlak, and Griliches (pp. 196–209). To this end, the transition story presented in this book is sufficiently informative but, however, not exhaustive. One aspect that could perhaps be considered is the material conditions in the 1950s that allowed large-scale regressions to be performed by those early-career scholars. Since execution of a panel regression requires more computing power, readers may be interested in learning exactly what technological innovations made the new econometric estimation available.
A potentially interesting but absent element in this book is how the Cobb-Douglas research program intersects with the rise of the “Chicago” approach in empirical microeconomics, as characterized by the uses of “less formalistic” theoretical economic models and reduced-form regressions by Chicago affiliates since the 1950s (Biddle and Hamermesh Reference Biddle, Hamermesh, Backhouse and Cherrier2017, p. 39). It may be worth examining whether the Cobb-Douglas tradition endured among Chicago affiliates or students due to two relevant facts. First, Douglas’s coworker H. Gregg Lewis remained as the spokesperson of econometrics at Chicago until the 1970s, and Lewis had a prominent impact on doctoral training (Becker Reference Becker1976; Biddle Reference Biddle and Samuels1996). Second, Chicago graduate students such as Griliches and Hoch were active in developing the Cobb-Douglas program in the 1950s (pp. 194–204). While the linkages between Douglas, Lewis, and Chicago-trained graduates have still not been systematically explored in the current history of econometrics literature, future research devoted to these histories may be helpful in generating a more coherent picture of the consolidations of the “Chicago” approach as well as the neoclassical-econometric program in the postwar period.
