Introduction
Widespread use of the term historical epistemology can be traced to the founding of the Max Planck Institute for the History of Science in Berlin in 1994, which adopted it to describe “a central research goal” of the institute.Footnote 1 Though “historical epistemology” had rarely appeared in English prior to that date, within fifteen years, the term had clearly taken off: three separate conferences were held on the topic from 2008 to 2009 (at Columbia, Leuven, and the Max Planck Institute itself), followed by the creation of numerous special issues, seminars, and research networks on historical epistemology across Europe and the US over the next decade (Vagelli Reference Vagelli2024: 41–43).
From the outset, though, the origins and aims of this new research program were unclear.Footnote 2 Lorraine Daston, the director of one of the departments at the Max Planck Institute, described it as “the history of the categories that structure our thought, pattern our arguments and proofs, and certify our standards of explanation” (Reference Daston, Chandler, Davidson and Harootunian1994: 282).Footnote 3 But this loose description left open questions about what made historical epistemology distinctive within the history and philosophy of science and to what historical traditions it relates. Daston herself confessed that she was “not sure as to the provenance of the name,” though she cited the philosophers Ian Hacking and Arnold Davidson as exemplary practitioners (Reference Daston, Chandler, Davidson and Harootunian1994: 283).Footnote 4 In doing so, she evoked an intellectual lineage that led back to the Institut d’Histoire des Sciences et des Techniques at the Sorbonne, where Gaston Bachelard, his successor Georges Canguilhem, and Canguilhem’s student Michel Foucault approached epistemology empirically through the history of science. For Bachelard, épistémologie examined the ways in which humans generate knowledge through instruments, concepts, and methods, a history that he saw as marked by ruptures and discontinuities (Chimisso Reference Chimisso2001). The term épistémologie historique (historical epistemology) gained currency in French when the philosopher Dominique Lecourt (Reference Lecourt1969) used it to characterize Bachelard’s work in a text that offered a Marxist reading of Bachelard. As this tradition evolved over the course of the twentieth century, it influenced, among others, the work of both Hacking and Davidson, whose engagement with Foucault and French historical epistemology left a mark on an entire generation of historians of science, Daston being one of them.Footnote 5
The director of another department at the Max Planck Institute, Jürgen Renn (1994), however, explicitly linked his research agenda to that of the American philosopher of science Marx Wartofsky. In the early 1970s, Wartofsky had drawn on Marxist philosophy to argue that social practices create the conditions of possibility for how we perceive the world and thus for knowledge itself (Wartofsky Reference Wartofsky1973, Reference Wartofsky1983). Since social practices are historical in nature, Wartofsky labeled his research program “historical epistemology” with little reference to the French tradition. Renn followed Wartofsky’s approach, for example, by historicizing the knowledge and organizing the large-scale infrastructures that govern the “Anthropocene” (Reference Renn2020).
Adding to this proliferation of meanings, Hans-Jörg Rheinberger, a later director at the Max Planck Institute, published a book-length introduction to historical epistemology in 2007 (translated 2010). Without tying it to a specific school of thought, he identified the historicization of scientific knowledge as a general theme in twentieth-century intellectual history. He included the French lineage, but not Wartofsky, while also discussing at length phenomenological and hermeneutic thinkers, including Edmund Husserl, Ernst Cassirer, and Martin Heidegger, as well as classic figures in the history and philosophy of science, such as Ludwik Fleck, Karl Popper, Thomas Kuhn, Paul Feyerabend, and Bruno Latour. Tellingly, the 2008 conference at the Max Planck Institute edited by Uljana Feest and Thomas Sturm (Reference Feest and Sturm2011) was teasingly titled “What (Good) is Historical Epistemology?”
If the various research programs at the Max Planck Institute did not form a single, coherent set, however, they at least shared what Wittgenstein would call a family resemblance. In contrast to the frequent separation of the history and philosophy of science in the late twentieth century, all of these approaches brought historical and philosophical concerns together. More particularly, all were rooted in the diverse efforts to bring down to earth Kant’s disembodied conception of epistemology as “transcendental philosophy.” Kant was concerned “not so much with objects as rather with our way of cognizing objects,” that is, with the structures that made any cognition possible (Kant Reference Kant and Pluhar1996: 64). However, whereas Kant focused on pure a priori conditions of knowledge, grounded in innate human cognitive structures that were logically independent of experience (“causality,” “modality,” etc.), various scholars in philosophy and the emerging social sciences, such as Karl Marx, Edmund Husserl, Jean Piaget, and Michel Foucault, sought either to complement or to supplant that pure a priori with an historicized one. They examined structures that made human knowledge possible but which were also subject to change: institutional settings, material conditions, the tools and infrastructures of research, power dynamics, perceptual norms, modes of reasoning, among others. The term “epistemology” in historical epistemology thus refers less to theories of knowledge, as in most contemporary analytical philosophy of science, than to what Rheinberger calls “the process of generating scientific knowledge and the ways in which it is initiated and maintained” (Reference Rheinberger2010: 3). The diversity of these processes, and the variety of ways to probe and illuminate them, gave rise to the diversity of accounts that could reasonably be described as “historical epistemology.” These accounts ranged from microstudies of laboratory processes to longue durée narratives that crossed multiple centuries and encompassed broader geographies.Footnote 6 Historical epistemology, then, can denote a wide range of approaches on a diversity of scales that in some way or another explore the historical conditions of the possibility of knowledge.
Given that variety, this Element brings together three authors who, socialized into the history of economics from different angles, have developed their own styles of research and writing, shaped by different historical inspirations.Footnote 7 Rather than opting for joint authorship throughout this Element, let alone presenting a top-down account of the historical epistemology of economics, we chose to present a series of short, accessible studies that showcase the variety of narratives that we believe historical epistemology can offer to the history of economics. Each section is authored individually, preserving our distinct voices, but grouped around a shared interest in three “themes” in historical epistemology. As we think our approaches are mutually enriching rather than competing, we have complemented each section with a brief commentary by one of the other contributors. These comments draw connections to the broader theme of the section and open discussion about how the section exemplifies a case of historical epistemology. In this way, the resulting text reflects a back-and-forth dialogue between the authors that may inspire further ways to historicize economic knowledge production. Thus, rather than providing an exhaustive account of the historical epistemology of economics, we have designed this Element as a primer with hands-on examples illustrating its potential for creating novel historical narratives.Footnote 8
History and Economic Knowledge
The first theme illustrates three ways in which historical epistemology stands apart from other forms of the history and philosophy of economics. The key idea is both simple and powerful: knowledge production is historically specific. Instead of tracing ideas along arcs of scientific progress, oblivion, and rediscovery, we focus on the detailed historical settings in which economic knowledge takes shape. Even if more focused, this gives us another kind of breadth because we take into account factors that might not initially seem central, or might be contingent, and yet turn out to be crucial for understanding how economic knowledge becomes possible in the first place.
Historical specificity has never been foreign to the history of economics. There is no historical writing without some attention to what is broadly called context – a notion which, upon closer look, serves confusing and even contradictory purposes. It can be a guide to the proper interpretation of ideas or a decorative backdrop, add a human touch to the making of economics, prop up idealized or hagiographic portrayals of thinkers, or function as a critical tool to challenge overly universal claims. All of these uses can, in their own way, be valuable for the agenda of one or another practicing economist; yet in each, context is something external to economic knowledge.Footnote 9
By contrast, when we speak of the historical specificity of knowledge production in historical epistemology, we do not mean something added to the independent body of economic knowledge, but a condition of its very possibility. Without its specific historical conditions, a given theory or method might not have come into existence, might not have drawn attention, might not have become effective, or, conversely, might have been more effective. As each section exemplifies, historical epistemology is not a mere history of theories of knowledge. It uncovers the contingent conditions behind what presents itself as autonomous scientific knowledge by emphasizing that knowledge production is inseparable from its social, economic, cultural, institutional, and technological dimensions. In Lorraine Daston’s words, it is “history that would pose transcendental questions in a highly particularistic mode” (Reference Daston2000: ix). In this vein, Tom argues that the recent rise of the use of “big data” was not merely the technological mining of an information-based economics but the response to business opportunities in platform capitalism. Till shows how the rise of mathematical purism in the early postwar years was not merely a resurrection of the works of Walras but can be read as a strategy for managing the social anxieties provoked by the Cold War. And Verena, treating input–output tables as artifacts, shows that what seemed like a simple application of Leontief’s big innovation in Italy was actually shaped by the specific situation of the local research team, who had to create the table’s allure of completeness and policy relevance on their own.
Unlike other histories of these epistemological shifts, our accounts treat factors that do not feature in the narratives told by economists themselves. Historical epistemology proceeds with caution regarding the categories and self-descriptions employed by economists, even if those categories can become objects of historical explanation. What economists say they do is too often limited by the institutional demands to justify their knowledge and is thus rarely a reliable guide to understanding what they actually do. Historical epistemology seeks to move beyond merely interpreting or analyzing methodological disputes among past economists, demonstrating greater autonomy from preexisting narratives about economics as written by economists. The advantage of placing the economist’s self-understanding in brackets, as it were, is that it leads to novel narratives, otherwise invisible, about what knowledge performs and fails to perform.
Without supplanting more traditional modes of intellectual history that emphasize the intentions of historical figures, historical epistemology offers an alternative viewpoint. By focusing less on (economists’) justification of scientific beliefs and more on the historical phenomena that correlate with those beliefs, historical epistemology allows the history of economics to serve new ends. Within this framework, scientific knowledge appears not only as a set of true or false representations but also as a window into the world in which that knowledge became effective. In part, we thus take the historical study of economic knowledge as a means to understand broader historical transformations that are occurring alongside the practice of science itself. Such an historical inquiry moves in the opposite direction of the more familiar approach to intellectual history, in which historical “context” serves to illuminate the meaning or significance of particular knowledge claims. This situates historical epistemology within the field of critical studies of science, without aligning it with any particular heterodox position in economics. It prompts us to question what may appear as the linear and harmonious development of ideas, uncovering instead the often messy transformations and stabilizations of knowledge regimes in response to changing historical sites.
Practices and Economic Knowledge
In the last decades of the twentieth century, “practice” became a popular buzzword across the humanities and social sciences, including in the study of science.Footnote 10 The “practice turn” had many inspirations – from Marx to Wittgenstein to Pierre Bourdieu – and found attention from scholars across multiple disciplines and sites, including philosophers and historians at Stanford (e.g., Ian Hacking, Nancy Cartwright, Peter Galison, and John Dupré), sociologists and historians at Edinburgh’s Science Studies Unit (such as Barry Barnes, David Bloor, and Steven Shapin), as well as anthropologists and sociologists pursuing ethnographic studies of scientific labs (such as Karin Knorr Cetina, Bruno Latour, and Steve Woolgar). What was meant by “practice” could vary, but it broadly signaled attention to action, to what people did. Although we sometimes place “practice” in opposition to “theory,” the most rigorous reflections on practice (especially those indebted to continental philosophy) have noted that even theorizing involves action: reading, talking, scribbling on paper, typing, debating at chalkboards, and so forth. Indeed, although the archetypical theorist might be the scientist sitting quietly and thinking, thinking itself is a kind of action (albeit an interior one) and can only become the possession of a community, part of the collective endeavor we call science, through expression. Action, in other words, is central to the constitution of theoretical knowledge as well.Footnote 11
To attend to practice in the history of economics, therefore, means to attend to what economists actually do. If they work with data, how do they gather it, find it, compile it, manipulate it, analyze it, and represent it? If they consider themselves pure theorists, whom do they talk to, what do they read, how do they brainstorm, how do they represent, test, and communicate their ideas? Quite naturally, attention to practices draws us to material culture, to the artifacts economists use and the physical environment in which they work. Pursuing the history of economics as a history of practices demands avoiding the lure of talking merely about ideas; it pulls us back to the concrete, to the bodies and actions of economists and to the environment and tools that make those actions possible.Footnote 12 To study practices thus connects closely to the emphasis on context from our first theme (“History and Economic Knowledge”), but it pulls us to think about that context in specific ways, not the “ideas of the time” but to the actions that created artifacts (working papers, articles, data sets, etc.) or helped to disseminate practices (presentations, lectures, and seminars) and thus constituted economics in particular times and places.
Describing practices, though, entails more than cataloging components. A collection of movements becomes a practice only insofar as it can be rendered intelligible, that is, insofar as it can be seen as aiming at a goal or following a norm (consciously or unconsciously, successfully or not).Footnote 13 Crafting a history of practices thus requires interpretation, recognizing how certain practices fit together, perhaps only temporarily, or how they worked at cross-purposes, or could be recombined into new configurations. Tom’s examination of forecasting in the early twentieth century, for example, shows how material technologies and institutions briefly created space for certain forecasting practices that ultimately could not be sustained. Till’s account of struggles over the meaning of honesty, personal conviction, and dogmatism in postwar East German economics likewise depicts the fragmentation of practices initially thought to be compatible (open debate, honest speech, and party loyalty). And Verena’s close study of “precision” in postwar economic planning shows that beneath a universal valorization of precision lay two competing conceptions and sets of associated practices.
Strikingly, all three of these short studies also show how clashes between practices were equally clashes between competing personae: for Tom, between the academic scholar and the commercial forecaster; for Till, between alternative visions of what it meant to be a scientific economist in Communist Germany; for Verena, of what constituted the ideal econometrician. Yet on reflection, these ties are not surprising, for practices both shape persons directly (as repetition shapes dispositions) and indirectly (by rewarding certain skills, abilities, and values). Thus, if historical epistemology pulls us to consider historical specificity, especially in the form of practices and material culture, it likewise asks us to consider how that specificity cultivates particular kinds of persons.
Persons and Economic Knowledge
Many popular accounts present the history of economics as shaped by great thinkers and their ideas. In these narratives, economic knowledge is of an immaterial quality, diffusing over decades and centuries to exert an elusive influence on later thinkers. Individual economists take center stage as the bearers of economic knowledge, as heroes or villains. They are characterized primarily by their thinking, the ingenuity of their ideas, their political–economic allegiances, and sometimes also by their personalities or charisma. In contrast, from an historical epistemological perspective, economists are less bearers of economic knowledge than participants in its structures, dynamics, and performances (Rheinberger Reference Rheinberger2010: 3). Following the epistemologist Ludwik Fleck, they could be viewed as the “personified intersection of various thought collectives as well as of various lines of development of ideas and as a personal center of new ideas” (Fleck Reference 75Fleck1979: 118). This perspective does not deny the role of persons but instead allows us to investigate how specific practices of knowledge production were accompanied by particular images of the economist. It also enables us to highlight the various ways in which economists downplayed the role of lived experience in the formation of their knowledge and stylized certain personal traits as foundational to their ideas.
Histories of scientific personae and scientific selves provide a starting point for considering the role of persons in an historical epistemology of economics. These histories delineate the skills, knowledge, authority, values, and virtues that make a “good” economist. As Lorraine Daston and H. Otto Sibum put it, constructing a persona is not a matter of fulfilling a “role.” Rather, “personae are as real or more real than biological individuals, in that they create the possibilities of being in the human world, schooling the mind, body, and soul in distinctive and indelible ways” (Reference Daston and Sibum2003: 4). Two of the most substantial changes in twentieth-century economics – its mathematization and formalization – came with a depersonalization of authorship (Düppe and Weintraub Reference Düppe and Weintraub2014a). The name of the author no longer conferred authority to knowledge; instead, authority came primarily from technique and method. This separation of author and person often meant putting aside the struggles, tensions, and imperfections of research while cultivating new scientific personae and virtues. Like other postwar social scientists, economists self-consciously turned from gentlemen-amateurs into craftsmen-scientists, from sole theorists into the team players of a self-proclaimed “technical,” tool-based profession.
The depersonalization of authorship in formal and technical economics did not reduce scientific personae to impersonal qualities.Footnote 14 Economists emphasized, for example, how model construction relied on the modeler’s creativity and tacit knowledge; how measurement involved rules of thumb and the economist’s judgment; and how forecasting relied on the expert’s ingenuity and intuition. These capacities resided within individual economists and were situated somewhere between trained skills, faculties gained through experience, and personal traits (Morgan Reference Morgan2012, Boumans Reference 72Boumans2015, Maas Reference Maas2018). Moreover, in economics as in other fields, informal communication, personal contact, and mentoring were considered vital for imparting techniques, as were qualities such as charm, sociability, and enthusiasm (Kaiser Reference Kaiser2005). The epistemic values of modern economics such as mathematical rigor or formal precision were presented as actual virtues in the sense that following them would make economists “better” persons – which often meant they would be able to find common ground in mathematical expression and technical reasoning despite different theoretical, intellectual, and political allegiances (Düppe and Weintraub Reference Düppe and Weintraub2014b).
In his contribution, Tom goes beyond the narrower boundaries of the economics discipline to investigate the evolving (gendered) qualities that made both “good” field agents and “good” respondents in household surveys. Shaping their experiences and subjectivities was a crucial part of the effort to control the personal and idiosyncratic elements in economic research and governance. Verena, in turn, zooms in on the difference between household accounts as part of everyday personal bookkeeping and the account books of “good housewives,” which met the criteria for proper, depersonalized data sources in studies of household finance. These stories demonstrate the active, productive element of economic knowledge making through figures and virtues that hinge on the polysemy of “good.” Finally, Till illustrates how certain characteristics can be virtues in one setting and vices in another. Qualities of courage, loyalty, and perseverance made the post-Keynesian economist Sidney Weintraub valuable in a community pushed to the margins of the discipline. Yet by showing that the stabilization of these traits had devastating consequences in other areas of Weintraub’s life, Till presents one way of reintroducing “the personal” into the history of economics.
Together, our studies showcase variations in the historical epistemology of economics organized around the three themes of historical specificity, practices, and personae. Each section embodies what we consider the distinctive mark of work in historical epistemology: a focus on the historical conditions of possibility for knowledge. But our topics are neither exhaustive nor representative of the history of economics as a whole; rather, our chosen studies reflect the interests and expertise of the three authors. We hope to inspire other historians of economics to explore some of the rich work referenced above on the historical epistemology of science and to bring this mode of analysis to their own projects and concerns.
THEME I HISTORY AND ECONOMIC KNOWLEDGE
1 Big Data Come to Economics Thomas A. Stapleford
In March 2005, Yahoo CEO Terry Semel warned his top executives that they risked being outflanked by companies better prepared to exploit the evolving digital world (Delaney Reference Delaney2006). Semel was an old-media executive, a former cochair of Warner Brothers, and he had positioned Yahoo as an internet-based media conglomerate rather than a tech-focused firm like its rival, Google (Malone Reference Malone2005). But despite that reputation, Semel understood the risks that Yahoo faced, and he agreed to a proposal from chief data officer Usama Fayyad to hire top academics who could pursue “fundamental research” in areas such as internet searching, user interaction, and machine learning. While many of Yahoo’s hires were in familiar fields like computer science or media, the company also did something more rare: It lured about a dozen young economists from top graduate schools (Delaney Reference Delaney2006).
The move was unusual enough to warrant a lead article in the Wall Street Journal. Few tech companies at the time employed economists. Microsoft had over 700 researchers in 2006; none of them were economists. But both Fayyad and his hiring targets recognized a new research opportunity created by the scale, speed, customizability, and data collection of e-commerce. As the Journal explained:
For economists, Web operations are data-rich fantasy lands where they can observe in real-time the behavior of millions of consumers in varied marketplaces far more effectively than ever before…. Until now, university economists typically have had to rely on limited sets of historical data to test their theories. At Yahoo, they can tweak market conditions and watch what happens to the company’s roughly 500 million monthly visitors.
Online markets could be a playground for economics research.
One of Fayyad’s early hires was Michael A. Schwarz, a Russian immigrant with a doctorate in economics from Stanford University’s Graduate School of Business. Schwarz had been an assistant professor at Harvard for several years before being hired by Yahoo, where Fayyad’s team asked him to improve Yahoo’s auction process for search ads. Yahoo used a different search algorithm than Google, but it also followed a different auction process that Yahoo’s leadership suspected was costing the company millions in lost revenue (Delaney Reference Delaney2006). Collaborating with two other young economists who had studied at Harvard during his tenure, Michael Ostrovsky and Benjamin Edelman, Schwarz and his colleagues created path-breaking work on the economics of ad auctions. Edelman, Ostrovsky, and Schwarz (Reference Edelman, Ostrovsky and Schwarz2007), published in the American Economic Review, would later win a 2013 prize from the Game Theory Society for “best paper at the interface of game theory and computer science in the last decade”; a separate article by Edelman and Ostrovsky (Reference Edelman and Ostrovsky2007) became a prize-winning paper in management science. But it was a working paper by Ostrovsky and Schwarz that best exemplified the research possibilities of a major internet platform. “Reserve Prices in Internet Advertising Auctions: A Field Experiment” proposed to test the effects of setting auction reserve prices according to a game-theory model (Ostrovsky and Schwarz Reference Ostrovsky and Schwarz2009). Previous smaller-scale experiments and anecdotal evidence offered ambiguous results: reserve prices tended to discourage bidders, and some scholars argued that they had negligible net effects. Ostrovsky and Schwarz used Schwarz’s unprecedented access to Yahoo’s auction process to conduct a controlled, large-scale experiment that yielded a clear conclusion: theoretically optimal reserve prices could indeed raise revenues substantially.
Yahoo credited Schwarz’s research with earning the company hundreds of millions of dollars,Footnote 15 and it exemplified for both economists and tech companies the potential benefits of mutual collaboration. If an internet company hiring economists was rare in 2006, it soon became commonplace: in 2018, economists Susan Athey and Michael Luca concluded that “the number of [new] positions available for economists in tech companies likely exceeded those at policy schools.” Amazon alone had “hired more than 150 PhD economists in the past five years,” and now had “several times more full-time economists than the largest academic economics department” (Athey and Luca Reference Athey and Luca2019: 210, 209). Schwarz himself moved from Yahoo to Google, and today is Chief Economist at Microsoft; his paper with Ostrovsky about the experiment appeared in online proceedings for the Association for Computing Machinery and was finally published in the Journal of Political Economy (Ostrovsky and Schwarz Reference Ostrovsky and Schwarz2011; Reference Ostrovsky and Schwarz2023).
From one perspective, the articles by Schwarz, Edelman, and Ostrovsky are simply recent additions to a longstanding field of economic research. Ostrovsky and Schwarz, for example, present their experiment as the latest contribution to economic studies of auctions dating back to the early 1980s (Ostrovsky and Schwarz Reference Ostrovsky and Schwarz2009: 2). On a longer timeframe, Mirowski and Nik-Khah situate this same research within economists’ attention to “information,” a tale they trace back to at least the socialist calculation debates of the 1930s (Mirowski and Nik-Khah Reference Mirowski and Nik-Khah2017). Historical epistemology, however, offers an alternative lens, one that attends to the practices, instruments, and institutions that create the “conditions of possibility” for a science (Foucault Reference Foucault and Faubion1997: 326). Here, new chronologies emerge in which continuities of citations, terminology, or even theory are broken by ruptures at other levels. Mirowski and Nik-Khah, for example, also identify discontinuities in economists’ treatment of information, including a rapid increase in computing technology that allowed economists to simulate complex auction designs, quite literally turning auction analysis into a project of “information processing” (Mirowski and Nik-Khah Reference Mirowski and Nik-Khah2017: 142–43).
If we look at Ostrovsky and Schwarz (Reference Ostrovsky and Schwarz2009) from that angle, it serves as a marker for a much broader break in the historical epistemology of modern economics: a new set of practices, artifacts, and institutional realignments that are logically linked together and that collectively create new ways of doing economics and of being an economist. To borrow Michel Foucault’s metaphor of archaeology (Reference Foucault and Smith1982), we can see in Ostrovsky and Schwarz the effects of corresponding changes in a series of other strata. Without resorting to monocausal accounts of these transformations, we can nonetheless recognize their logical relations and see how they form a framework stabilizing a particular way of life and scientific practice for economists (Davidson Reference Davidson and Davidson1997).
Economists themselves have heralded the emergence of new methods, research goals, and activities over the last two decades tied to computing technology. Often identified under the rubric of “big data” (Einav and Levin Reference Einav and Levin2014; Varian Reference Varian2014), these phenomena encompass the massive quantity of data available but also the overwhelming volume of data production, expanding storage capacity, and the increasing speed with which data can be analyzed, condensed, and visualized through algorithms. That growth rests on a technical foundation – advances in digital storage, transmission, computing power, and software techniques – but it is equally a product of new business practices, government practices, and social behavior in which vast amounts of information are recorded, transmitted, stored, shared, and analyzed (Kitchin Reference Kitchin2014; Schneier Reference 81Schneier2015). Though these technologies and practices have their own heterogeneous timelines, they have coalesced to enable important shifts in economics research, including the work of Schwarz and his colleagues. Together, they are opening new vistas: As Stanford economist Susan Athey recently proclaimed, “the combination of ML [machine learning] and newly available data sets will change economics in fairly fundamental ways” (Athey Reference Athey, Agrawal, Gans and Goldfarb2019: 508).Footnote 16
In a longer article (Stapleford Reference Stapleford2021), I examined three methodological shifts tied to the growth of big data and machine learning techniques: (1) new efforts to build economic models from data through machine learning rather than a priori designs based in economic theory; (2) decreased attention to probability theory in favor of tests of predictive fit from data science; and (3) the rise of experimentation through digital systems (such as Ostrovsky and Schwarz’s experiments with ad auctions for Yahoo). Here, I would like to focus on how these methodological shifts are also tied to new forms of life for economists. These changes have created new priorities in economics, notably access to large-scale data (to bypass probabilistic analysis and provide sufficient grounds for model-building), a grasp of data science tools and techniques, and access to manipulable digital systems for experimentation. In turn, those priorities are reshaping how economists are educated, how they relate to other disciplines, and where they work.
Proponents of big data in economics stress the need to develop new skills and to reconsider how students are trained (e.g., Einav and Levin Reference Einav and Levin2014: 715; Varian Reference Varian2014: 3, 24–25; Athey and Luca Reference Athey and Luca2019: 227–28). The basic techniques and tools of large-scale data analysis – new forms of databases, machine learning algorithms, strategies for boosting the out-of-sample stability for trained models, and so on – are shared and discussed independently of particular applications. In this respect, they function much like familiar economics toolkits such as mathematical probability or regression analysis. But there is a catch – precisely because traditional econometrics began by specifying an economic model, economic theory was intertwined with statistical methods from the outset. That was why economics departments taught econometrics themselves rather than just having students take general courses on statistics. Now, however, methods and disciplinary theory can increasingly be divorced: “My standard advice to graduate students these days,” Varian reports, “is go to the computer science department and take a class in machine learning” (Varian Reference Varian2014: 3). Stanford’s Economics department, for example, now counts a substantial number of computer science and statistics courses focused on data science as economics electives for undergraduates, and doctoral students in certain fields (such as “Econometrics for Causal Inference”) are explicitly directed to grad-level versions of those courses.Footnote 17
To the extent that these methods become central to a given economist’s research, the reliance on general data science techniques begins to elide the boundaries between economists and other social scientists working with large-scale data, especially insofar as economists are using machine learning to build predictive models from data rather than starting with a priori economic theories. Among the proponents of machine learning within economics, even those like Athey, who have argued that standard approaches need to be modified to handle the causal analysis and policy aims of economics, see the “economics” contribution coming from inferential techniques, not micro or macroeconomic theory. Naturally, such techniques will be of interest across the social sciences. Thus, Athey foresees a “substantial increase in interdisciplinary work,” where “computer scientists and engineers focus on algorithm design, computational efficiency, and related concerns” while “academics of all disciplines” shape these tools to “facilitate[] measurement and causal inference” (Athey Reference Athey, Agrawal, Gans and Goldfarb2019: 538). Looking at the rise of joint degree programs in computer science and economics, Athey suggests that “economic education” will focus on teaching students “how to use data to answer questions” (and notably not on general theories of human behavior). Though data science currently draws largely from computer science and engineering, she “predict[s] that these programs will over time incorporate more social science, or else adopt and teach social science empirical methods themselves” (Athey Reference Athey, Agrawal, Gans and Goldfarb2019: 541, emph. added). Note that rather than economic theory, it is “social science empirical methods” (namely a concern for causal prediction) that Athey sees as being integrated into data science. Economic theory is not disappearing, and undergraduate curricula retain their standard courses (even at Stanford). But the boundaries are blurring in research areas that are more heavily reliant on machine learning and massive data sets.
Those economists who adopt the research methods of the data revolution are also finding new institutional homes: Silicon Valley firms where they can combine access to massive amounts of data with the prospect of running rapid experiments by manipulating their firms’ digital platforms. Writing in 2018, Athey and Luca suggested that the boundaries between academia and business would become increasingly blurred as research and intervention are intertwined:
Microsoft, Google, Yahoo!, Facebook, Amazon, eBay, Yelp, Uber, and other companies have all hosted faculty during sabbaticals. Tenured faculty members have left academia for positions at Amazon, Google, and elsewhere. Practitioners have also transitioned into academia – for example, leaving Facebook and Microsoft for MIT and Stanford… As more PhD economists accept positions at tech companies, clearer paths for spending time (or re-entering) academia will likely appear….
Since that time, the job market for economics PhDs has shifted even more dramatically than Athey and Luca might have envisioned. In 2018, tech companies hired fewer than one in twenty economics doctorates from the top ten economics departments; as of 2022, the number was one in seven. Amazon’s employment of economists rose from 150 in 2018 to over 400 by 2022; in 2021, Uber hired one-fifth of Harvard’s graduating PhD cohort (Anonymous 2022).
Taken together, these shifts demonstrate that a new form of being an economist – a new set of methods, new forms and sites for working, and new sets of values – has taken root and grown rapidly. Although American economists have worked in corporations since the 1910s (Stapleford Reference Stapleford2017a; Reference Stapleford2017b; Reference 82Stapleford2020), the concentration of PhD economists in tech companies and the attraction of those companies as sites for novel economics research (and not merely lucrative applied economics) mark a distinctive new phase. This transformation, with its ties to new technologies, new practices for collecting, storing, and analyzing data, and new methodological tools, would not be immediately apparent by studying the history of economic ideas. Like members of any scholarly community, economists present their work as contributions to long-standing theoretical topics, just as Ostrovsky and Schwarz linked their research at Yahoo to much older studies of auction pricing. The new configurations of methods, research sites, training, and priorities reveal themselves (and their logical interconnections) only when we follow historical epistemologists in attending to the relationships among practices, artifacts, and institutions that underlie the generation of knowledge.
Comment Till Düppe
Data were big sixty years ago when they were first fed into computers, and they are still big today. It is not size itself that makes them big, but the cultural eye through which they appear so. As Christine Borgman has shown in her study on large data practices, data require a “knowledge infrastructure” to evoke the enthusiasm they do (Reference 72Boumans2015). Thus, a history of the use of data in economics told in terms of the size of data would make for a rather thin reading.
What Tom shows is that what makes data seem big since the New Millennium is that they are fed into the big business of e-commerce and platform economies. It’s the business opportunities data have opened and the new role economists could play in this business that has changed and affected economists’ “way of life.” Of course, causation goes both ways, as economists’ new role is not only to conceptualize but also to “design” markets. This two-way causation might seem blatantly obvious, but it’s easy to overlook when looking at economists’ own narratives of what happened. They may speak of scientific advances by the use of new technologies that help make economics more “empirical.” This language has been taken up by many scholars in the history and philosophy of economics when speaking of “the empirical turn.” Their accounts track the development of ideas over time, either reinforcing or challenging the histories that economists give in their own papers. For example, whereas Ostrovsky and Schwarz (Reference Ostrovsky and Schwarz2009) date the origin of auction price research to the 1980s, Mirowski and Nik-Khah (Reference Mirowski and Nik-Khah2017) place it within a longer and broader history of “information” in economics.
Tom, however, by looking at the history of research practices, embeds Ostrovsky and Schwarz in a very different timeline, with a major rupture in the mid-2000s that sets Ostrovsky and Schwarz apart from previous studies in the same field. Tom’s story is not in conflict with the economists’ narratives or that of Mirowski and Nik-Khah. They are looking at ideas; he is examining research practices. Instead, Tom’s approach shows how historical epistemology can reorient our attention to alternative levels of analysis.
To underline the novelty, a tentative comparison may help. What Tom shows with reference to Foucault, how institutions cause epistemological rupture, reminds me of institutionalist economic historian Douglass North (Reference North1968). Using data sixty years ago in a new systematic way, North argued that technological innovations in seventeenth-century ocean shipping, notably a fast, light Dutch ship called the flute, could have led to increased productivity much earlier. The problem was that they couldn’t be used because pirates were still around, and older ships were needed to transport arms and soldiers. The same applies here with Ostrovsky and Schwarz. Having been out there since the 1980s, it was the rise in e-commerce that created the serendipitous, that is, historically specific conditions where several dormant factors fell into place: the theory of market design, the technology of machine learning, the decline of theory-heavy economics, the rise of natural experiments, and a more open-minded disciplinary politics in economics.
This change could not have come from within the discipline alone, that is, from a mere drive to advance economics. What the heterodox community had bemoaned for decades with very little success (the withdrawn, mathematics-heavy, unrealistic economist) was realized in less than a generation within a changed institutional landscape. Epistemological tastes are the result of the historical situation that allows them to thrive, which in Tom’s case in this section means: they are opportunistic.
2 Mathematical Purism Soothing Cold War Fears Till Düppe
In the spring of 1949, Gérard Debreu traveled from France to the US, still by boat and train. A young mathematician of twenty-seven, he felt uncomfortable speaking English, little eased by his shy character. Though uncertain of what lay ahead, in hindsight it seemed that he arrived at the right place at the right time. He was soon offered a research assistant position at the Cowles Commission for Research in Economics at the University of Chicago, the center that hosted a remarkable number of future Nobel laureates, including Debreu. He arrived in the US at what seemed a serendipitous moment, but, as we shall see, it proved rather difficult.
Debreu had a contact in Chicago, someone who not only spoke French but also shared his mathematical language: André Weil in the mathematics department. Weil, who had been in Chicago since 1947, was, like Debreu, a graduate of the École Normale Supérieure in Paris and one of the founders of the French group of mathematicians publishing as Nicolas Bourbaki. Debreu, a 1945 graduate, had studied under Bourbaki member Henri Cartan. The 1950s would see Bourbaki’s approach gain increasing influence within US mathematics departments, and Debreu’s arrival in the US heralded the introduction of this axiomatic style to American economics. It continues to set the standard in advanced microeconomics textbooks and remains one of the most arcane forms of reasoning in economics.
Yet there was something paradoxical about Debreu “applying Bourbaki” to economics. The Bourbaki group aimed to render mathematics self-contained, with a focus on the foundations of mathematics rather than its applications in scientific theories.Footnote 18 Their preoccupation with mathematical proofs served no explanatory, predictive, or otherwise scientific purpose. Looking for a job at the time of postwar reconstruction, Debreu sought to get out of this intellectual cocoon. He became a research assistant to an engineer-economist at the École Polytechnique, Maurice Allais. But there, Debreu later lamented, he was frustrated by his mentor’s lack of rigor. At Cowles, by contrast, his purism was again very welcome. “Whereas before I was in a group which felt mathematics went too far and points of rigor were not terribly important, at Cowles I came to think, very quickly, that full understanding of a problem required no compromise whatsoever with rigor” (Debreu in Weintraub Reference Weintraub2002: 153). How come? Why was the Cowles Commission so receptive to a level of mathematical purity that even scientific interests could not spoil?
The situation after the war in the US was quite different from that in France. In August 1949, the Soviet Union detonated its first nuclear bomb, fueling widespread fears, notably among scientists, of an all-destructive World War III. These anxieties were stoked by the belief that knowledge could, and had, leaked from academia into communist circles, a notion vigorously promoted by Senator Joseph McCarthy. Fear of communists in strategic university positions was at its peak upon Debreu’s arrival in the US. European émigré scientists were viewed with particular suspicion, as many had initially shown sympathy for socialism and the proactive role of scientists in supporting the state’s promotion of the social good. The atmosphere of McCarthyism was especially pronounced at the “radical” University of Chicago, as Herbert Simon, a research consultant at the Cowles Commission, later recalled “by 1948, Communists and supposed Communists were being discovered under every rug (…). Any graduate of the University of Chicago, with its reputation for tolerance for campus radicals, was guaranteed a full field investigation before he could obtain a security clearance” (Simon in Mirowki Reference Mirowski2001: 246).
Debreu was navigating dangerous waters when, after his arrival, he began addressing problems in Walrasian economics, then still laden with socialist ballast. Oskar Lange, a professor in Chicago’s economics department until 1945, was still vividly remembered on campus. Lange had applied Walrasian economics to economic planning in a theory he called “market socialism” (Lange and Taylor Reference 78Lange and Taylor1964). Given that Walras’s concept of “perfect competition” did not imply rivalry but rather that “prices are given” (meaning no market agent can influence prices), his framework could, in theory, apply if the state sets prices and adjusts them in response to observed gluts or shortages. Without shedding this historical baggage, Walrasian economics could have been easily dismissed as politically unacceptable during McCarthyism and might never have become the hallmark of Cowles’s economics in the two decades following Debreu’s arrival. What if not Debreu’s commitment to mathematical purism could provide distance from such historical connotations of Walrasian economics? The inscrutability of mathematics, which may be its primary social function, protected against unwanted questions from those sympathetic to McCarthy’s cause.
This might have been true for outsiders, but not for insider economists. In the University of Chicago economics department, McCarthy’s influence found an academic representative in Milton Friedman, who had been teaching economic theory since 1946. Friedman, who considered himself a statistical economist, was politically critical of the Cowles Commission’s use of mathematical economic theory. As he later explained: “Mathematicians … tend to be favourable to central planning … When they enter a field like economics, they carry over the belief that all problems have clear-cut solutions and that they are competent to find them” (Friedman and Friedman Reference Friedman and Friedman1998: 262). Friedman’s position is essentially the same as what Friedrich Hayek had advocated in the years before and after 1945 in response to Oskar Lange and other left-leaning economists (Hayek Reference Hayek1949). Against the prevailing sentiment of the time, Hayek highlighted the conflict between the values of a liberal society and the belief in the all-powerful capabilities of science. Hayek, too, had been at the University of Chicago since 1950.
The surprising fact that Debreu might have slowly come to understand at Cowles is that some of his colleagues were indeed vulnerable to this critique! While the researchers at Cowles were not politically unified, some came from distinctly left-leaning backgrounds. Lawrence Klein, for instance, had previously been a member of the Communist Party, and immigrants like Jacob Marschak, a former Menshevik, carried with them the political influences of their early socialization. Many were enthusiastic about developing methods of planning, an attitude fostered by the patron of the Cowles Commission, the officially independent think tank for the Air Force, the RAND Corporation (short for Research and Development). The Cowles Commission was only formally hosted at the University of Chicago, but its research was funded largely by RAND and similar national laboratories, where military-planned, state-funded secret science was conducted. These laboratories were still inspired by the successful wartime collaborations between the military and scientists. Economists, physicists, and engineers worked hand in hand to secure the triumph of science over totalitarianism. The optimization techniques of operations research, for example, emerged to tackle problems such as submarine search, materials allocation for tank production, decoding secret radio messages, and figuring out the optimal height for bomb detonations.Footnote 19 The US did not win the war by letting the forces of free competition decide which weapons to produce or how scientists were distributed across military posts. Scientists contributed to winning the war because military actions were planned and thus also required tools for planning. Why should the techniques developed in this way not have a place in a peaceful society? Most of the economists at Cowles would have nodded without hesitation.
In short, Cowles’s economists might have been in a serendipitous position, one might say in retrospect. However, they were, more than others, exposed to the value conflict of early Cold War society, the conflict between scientific optimism and liberal values. As vital as Debreu’s Bourbakism was for Cowles in maintaining the contradictory links with the academic, military, and political spheres, it was equally important for Debreu himself to avoid being drawn into them. From his own point of view, this simply meant he could appreciate the fact of “being left alone”: “I was left alone to do my work during the five years from 1950 to 1955 – a marvelous opportunity that I tried to use fully” (Debreu in Feiwel Reference Feiwel and Feiwel1987: 256). In the profound ideological warfare of the 1950s, Debreu was not in control of the channels that led from the pure to the applied. Limiting himself to the pursuit of mathematical rigor, to Bourbaki’s purism, meant, first and foremost, not raising suspicion.
Hence, Debreu remained discreet in all respects. He never took a stance on Cowles’s internal affairs concerning the importance of mathematical economic theory, despite having plenty of opportunities to do so.Footnote 20 Debreu embodied mathematical rigor but did not propagate economists’ appreciation for it. The same holds true for his relationship with RAND. Unaware of what was going on, he later said that he was ignorant of what his colleagues did “during the summer.”
Some of the mathematical economists I knew spent a significant part of the summer at RAND. I did not do that and that may be due to some extent, but not entirely, because I was not a U.S. citizen, and RAND was doing a number of things for the army. (…) I do not know who from Cowles went to RAND in the summer.
Some went, Debreu knew. But he preferred not to ask who went. Who knows what they did there? Better not to ask. In this attitude, he resembled what his Bourbaki mentors had published in 1949 more explicitly about the mathematician’s responsibility for scientific applications, an article that Debreu surely read word for word.
Why do applications (of mathematics) ever succeed? Why is a certain amount of logical reasoning occasionally helpful in practical life? Why have some of the most intricate theories in mathematics become an indispensable tool to the modern physicist, to the engineer, and to the manufacturer of atom bombs? Fortunately for us, the mathematician does not feel called upon to answer such questions.
Bourbaki’s purism served to relieve the sense of responsibility weighing upon economists in the early Cold War.
Debreu’s attitude is thus distinct from several others in the Cowles Commission who indeed acted as Cold War warriors, such as Charles Hitch and Kenneth Arrow, but also Herbert Simon and Oskar Morgenstern. However, even if from today’s perspective these economists were vital for the emerging identity of the economics discipline as a modeling science, they tended to be skeptical of disciplinary boundaries and theoretical dogmatism, particularly of Walrasian economics. On the contrary, their, say, cybernetic activism was a transdisciplinary affair.Footnote 21 Unsurprisingly, it is also these economists that are today again mentioned when it comes to propagating nondeterministic ideas open to methods and perspectives from other sciences.Footnote 22 It was the inward-oriented attitude of scholars like Debreu, hiding behind higher mathematics in difficult times, that became a model for later economists shying away from other social scientists.
The diverging intellectual attitudes of these two types of economists became manifest in the single most important collaboration between Debreu and Arrow (Reference Arrow and Debreu1954), resulting in an article for which both would receive separate Nobel Prizes. The evolution of their article can be reconstructed letter by letter and shows to what extent the two authors were after different goals (Düppe Reference Düppe2012, Düppe and Weintraub Reference Düppe and Weintraub2014a, Reference Düppe and Weintraub2014b). Personally, Arrow was extroverted, outspoken, and would say immediately what came to his mind, exactly the opposite of Debreu. Regarding their motivations, Arrow wanted to know more about the welfare implications of general equilibrium theory, and Debreu was grateful for the opportunity to use Bourbaki-style topology. Debreu would note mathematical flaws, which Arrow showed to be economically trivial, while Arrow would point to economic flaws, which Debreu ruled out mathematically. Debreu would suggest a deletion in case of a “controversy about the interpretation of the text,” and Arrow would write the introduction and the historical note. While Debreu continued pursuing yet more purity in economics, Arrow continued pursuing his eclectic interests in economics as an applied science. A few years later, as circumstances changed, their differences would have become more apparent than they were in this “serendipitous” moment of the early 1950s. It would have been impossible for them to join forces again.
Much has been written about the role of economists in shaping the early Cold War’s Zeitgeist.Footnote 23 Leaving grand narratives aside, what the case of young Debreu reminds us, is that it was also a time of social hysteria that exposed scholars to contradictory demands.Footnote 24 The military entanglements of economists that ran counter to the public image of economic knowledge challenged intellectual integrity and increased the desire to relinquish responsibility. The rise of mathematical formalism in Walrasian economics occurred partly because it could meet these exact needs, offering both discursive coherence and interpretive indifference. It was not the implied set of beliefs that made this theory attractive, but the possibility of referring to a formal social structure without going into the details of the meaning of its elements. This need for protection helps explain the allure of intellectual purism, which facilitated the transformation of economics into a science that prioritizes rigorous modeling over practical relevance (e.g., Gordon Reference Gordon1976). In other words, economics is not only a form of seeing society but also acts as a buffer against it, making it more bearable. Considering how much fear permeated early Cold War US society, the choice of methods in economics can be seen as a symptom of larger struggles in cultural history.
Comment Verena Halsmayer
In contrast to historical accounts that trace concepts of economic equilibrium over time, Till highlights how the particular situation of the early Cold War shaped the postwar ascendancy of mathematical purism.Footnote 25 Situating economists’ methodological choices within the larger settings of cultural history, he zooms in on anxieties fostered by the Cold War and shows how they turned Bourbaki’s purism into a refuge from ideological warfare. The case of Gérard Debreu exemplifies how time, place, and person can align and make the persona of the mathematical general equilibrium analyst a suitable way of life. In this vein, Till’s study shows how the historical conditions of possibility for knowledge formed the identities and self-understandings of the economics discipline – here, particularly by elevating the virtues of rigor and detachment.
From within the frame of historical specificity, Debreu’s case illustrates how an accepted persona legitimizes the channeling of attention and promotes both knowledge and ignorance (here, ignorance of economic theory’s political associations as well as its social and political settings). Debreu happened to have “the right stuff.” The philosopher Isabelle Stengers has argued that having the right stuff to be a scientist means, first, “having faith that what a scientific question doesn’t make count, doesn’t count; a faith that defines itself against doubt” (Stengers Reference Stengers and Stephen2018: 36). For Stengers, this purposeful ignorance is a defining feature of a “scientific ethos,” marked by its gendered, masculinized character. Although historical specificity is not central to her argument, her perspective serves as a reminder that scientific personae are inherently collective and political and thus historically situated.
Studying systems of equations as expressions of the lives of mathematical economists in specific historical circumstances can further elucidate that having “the right stuff” and identifying with it are two different things. Scientific personae instill an ethos in those who embody them, but this does not necessarily unfold as seamlessly as it appeared, at least from the outside, in Debreu’s case. Till’s study provides a starting point for also studying the doubts, struggles, and defiance of economists grappling with prevailing personae, which might be particularly prevalent among less prominent economists and other knowledge brokers (Forget Reference Forget2020). In this vein, we might also ponder how to account for those in the history of economics who were excluded from embodying these personae at various points in history, those deemed not to have “the right stuff” in the first place.
3 Situating Input–Output Tables Verena Halsmayer
Folded inside the back cover of a 1953 monograph by the Mutual Security Agency is a large, unwieldy poster featuring a table. The rectangular table of roughly 60 columns and 200 rows is ordered by groups of industries, from “agriculture” to “paper” to “transportation,” each assigned a numeric label. Industries listed on the left differentiate further into products. “Food drink and tobacco processing,” for instance, includes industry 14, “Sugar and confectionery,” which comprises “refined sugar,” “cocoa, chocolate, candy,” and “molasses and other products.” Each row is devoted to one such product and spans the columns denoted with the same set of industries ordered by their numeric label. The boxes at the intersections of rows and columns contain mostly dashes; only some of them are populated with numbers. The intersection of the row for “molasses and other products” and the column for “animal products and forage,” for instance, lists 1.679 million lire. These boxes provide the money value of goods that are produced in one industry and used in another. The final columns and rows provide the relevant totals. In addition, the table includes columns for the “final demand” listing, for instance, “exports,” “households,” and “government.” Accordingly, the poster’s title identifies the table as presenting the “interindustry flow of goods and services by producing industry and purchasing industry.” This input–output table was the main result of a two-year empirical research project on “The Structure and Growth of the Italian Economy” (Mutual Security Agency 1953). To economists and planners at the time, such a table presented both a compact survey of the economy and a tool that would allow intervention into the material world of production. Indeed, the Italian poster was received as a “visible token of a real victory over scanty and intractable data won by ingenuity and hard work,” as a reviewer of the book celebrated in Econometrica (Evans Reference Evans1955: 110). Input–output tables and their rendition of the economy as a whole gained momentum in the postwar decade as a scientific as well as governmental triumph.
The Italian table came together in the context of larger Cold War planning efforts directed at European reconstruction. The Mutual Security Agency administered both economic and military assistance to what the US saw as “friendly countries.” Its Special Mission in Italy carried out a research program to provide information for assisting the country’s recovery and supporting its defense effort. Its research contributed to the endeavor of the Organization for European Economic Co-operation (OEEC) to increase total national outputs by 25 percent between 1951 and 1956. By making visible the “structural” interrelations of a national economy, input–output tables were used to assess the effects of various policies. In the Italian case, researchers checked, for instance, the implications of the OEEC’s growth goal for the production capacities and import requirements of the various sectors and for employment and hours worked (Mutual Security Agency 1953: 3). The project contributed to the OEEC’s experimentation with techniques to find out about how to further economic growth in poorer areas such as Romania, Turkey, or, for that matter, the Italian Mezzogiorno. The methods that were found to be conducive to growth later became blueprints for decolonizing countries, reinforcing a notion of development that involved little more than increases in national product. The OEEC was just one place where comprehensive input–output tables afforded the impression of being able to look at an economy’s structure in one huge yet compact foldout chart. Enabled by the development of computing machinery, input–output tables became one of the essential components of planners’ toolboxes in the 1950s and 1960s (Lahr and Dietzenbacher Reference Lahr and Dietzenbacher2004). Eventually, in 1968, input–output methods were standardized in the United Nations System of National Accounts.
As with any artifact in the history of the sciences, an input–output table allows for various strategies of historicization, each of which makes the table appear in a different light (Halsmayer Reference Halsmayer2018). Frequently, input–output analysis is investigated as a contribution to economic theory, in particular when it comes to the work of Wassily Leontief, the Russian émigré Harvard economist. The focus on theory sometimes leads to looking for possible origins of ideas and influences. Leontief himself mentioned the iconic tableau économique as a precursor of input–output analysis. From such a perspective, the idea of the economy as a system of production is kept constant over the decades; it just required specific circumstances (wartime economizing, state planning endeavors, and computer development) to be realized as a proper empirical program. The Italian table, in such an account, would appear as a neglectable empirical realization of Leontief’s theoretical input–output framework. Historical epistemologists such as Georges Canguilhem have criticized the figure of the precursor as a “false historical object” (Canguilhem Reference Canguilhem and Gutting2005: 206). Different from the scientist’s object, the historian’s is of a cultural object with its contextualized movement. Such a perspective requires paying attention to the situated meanings of scientific concepts in their concrete circumstances, for instance, in a particular experimental setting. Moreover, following Canguilhem, it attends “to the nonscience, to ideology, to political and social practice” that inform scientific practice and contribute to its situational specificity (Canguilhem Reference Canguilhem and Gutting2005: 204). From this perspective, an input–output table turns into the (preliminary) result of an activity that is set at a specific time and place. What appears as irrelevant in the history of input–output as theory takes center stage here.
Taking seriously the situational specificity of input–output as a practice of measurement also brings into view, for instance, the many (often female or otherwise obscured) empirical workers who are involved in the tedious and time-consuming making of numbers. It sheds light on the institutional settings and financial backings that were necessary for large-scale empirical research. In the case of the Italian table, the project was conducted by a team of twelve researchers headed by Vera Cao Pinna, at the time the foremost input–output expert in Italy (Corsi and Zacchia Reference Corsi, Zacchia, Madden and Dimand2019). The Italian staff collaborated with the mission of the Mutual Security Agency, Italian government agencies, and private firms, not only with regard to research funding but also to compile data.
Situational specificity is not simply of social and political but also of epistemological relevance (Lampland Reference Lampland2023). Regarding input–output, the economist Martin Kohli (Reference Kohli, Klein and Morgan2001) has shown how the practical contingencies of measurement at the Harvard Economic Research Project (HERP) in the early 1950s essentially framed the input–output framework as it was later standardized. Akin to Kohli’s work, examining the practices that transform economic abstractions into whatever actors treat as “instruments” of measurement and intervention also allows us to account for the frictions and failures of input–output work: how staff struggled with large amounts of data, worried about missing data, adapted classifications to available data, and made judgment calls about including, excluding, and triangulating numbers (Tooze Reference Tooze2001: 203–7, Tinn Reference Tinn2018, Halsmayer Reference Halsmayer2024: chapter 3). In contrast to the smooth and effortless impression conveyed by input–output tables and the associated computer metaphors, much had to come together to realize these tables and to make them seem plausible. The clean appearance of a ready-made table hides as much as it shows: Neither empirical research nor the economic world it sought to portray worked like a frictionless flow of information. The same held for the work of the Italian researchers. Even though they had personal connections to Leontief and the HERP, their measurements were not simply a smooth application of Leontief’s method.
Testifying to the effort required to turn an economic abstraction into a recognized depiction and effective tool, Cao Pinna emphasized that empirical work “cannot be rigidly programmed in advance” but “calls for long and patient research work, analysis, integration and coordination of available data, performable only through stages of successive approximation” (Mutual Security Agency 1953: 139; cf. Tinn Reference Tinn, Evans and Johns2023: 139). Her team shaped and reshaped the theoretical schema, the available data, the relevant computing techniques, and more. The Italian table was based on the latest available surveys – conducted in 1937 and 1939, more than a decade earlier. Researchers updated data, homogenized formats, prepared calculations, triangulated missing figures, excluded data, and eventually calculated the so-called “technical coefficients” featured in the table. As a historian’s object, the Italian table came together in a process entangling all kinds of social, cultural, political, economic, technical, and epistemic factors, which ultimately shaped its specific setup.Footnote 26
Like its sibling metric, the gross domestic product (GDP), input–output relied on a set of rules for turning scarce and scanty data into relatively consistent numbers. Again, this did not mean that data could simply be filled into a fixed form. Nevertheless, such rules did foster the belief in the global comparability of economic structures and national performances, promoting transfers of growth strategies aimed at “international development.” The historian of technology Honghong Tinn has laid out how input–output tables participated in the associated hierarchies of reasoning: Was input–output particularly useful for so-called “underdeveloped” and “developing” countries where data were scant? Or was it particularly inadequate because “less mature” economies would did not conform to the imagery of the economy as a computer and “deliberations about such economies merely required human reasoning instead of sophisticated machines” (Tinn Reference Tinn, Evans and Johns2023: 140)? Diverging assessments on these questions also related to the notorious problems of the comparability of metrics that emerged from locally situated data practices. Despite the UN’s standardization, input–output frameworks differed across countries, which made the comparison of economies’ structures and strategies for growth problematic, even for those who advocated their use in the first place.
Comment Till Düppe
The shift in perspective that Verena illustrates in this section – from long-term continuities of ideas to their historical specificity – recalls an image used by Reuben Hersh, a creative historian of mathematics, more than thirty years ago (Reference Hersh1991). Trying to capture what mathematicians do, he distinguished the “front” and the “back” of mathematics, such as in a restaurant or a theater where the front is open to the public, while the back remains hidden: the kitchen versus the serving area, the stage versus backstage. The front is “formal, precise, ordered, and abstract,” while the back is “fragmented, informal, intuitive, and tentative” (Reference Hersh1991: 127 f.). The same seems to apply to the Italian team of economists producing an early version of an input–output matrix: complete, simple, useful in the front, while improvised, incomplete, and approximated in the back. One strategy of historicization would be to look at the final product in the front that the scientist, like a chef or a director, wants us to perceive: their novel improvement of an old theory, such as the Tableau économique. Historical epistemology, instead, invites us look behind the scenes and observe the frictions, failures, and adjustments made to ensure that the final result appears seamless and, in Verena’s case, relevant for policymakers.
What is striking in this shift is that what is essential to one perspective is accidental to the other, and vice versa. As the quality of the ingredients may be essential for the serving area, the backup schedule for orders might be key to success when running the kitchen. While in the front the structure of the input–output table seems essential, the gender of those collecting them becomes essential to those producing the table. And, not to forget, Verena shows that this shift in perspective makes visible what remains otherwise imperceptible: how the tool of input–output tables changed the language of the Italian policymakers what development policy should achieve.
What is self-evident in almost any professional domain meets particular resistance in science, where the ideal of anonymous, universal truth hinders us from seeing science as a product of humans muddling through. As rigid as scientific methodology may appear, it inevitably involves, like any other human art, an element of pretending.
THEME II PRACTICES AND ECONOMIC KNOWLEDGE
4 Forecasting Markets Thomas A. Stapleford
As World War I drew to a close, an economist in Cambridge, Massachusetts, stood peering down at a wooden box with a glass top and an interior light. On the glass surface, lit from beneath, were two thin sheets of tracing cloth, commonly used by draftsmen and artists to make copies. Each cloth had its own graph, drawn in ink, representing fluctuations in a financial or industrial data series, such as interest rates on bonds, production of pig iron, or the quarterly total of new business permits. The X-axis for each graph was marked off in years beginning in 1903, with each year divided into quarters; the Y-axis depicted the value of the data series at each quarter. Each data series had been “normalized” by converting its values into standard deviations from the arithmetic average of the entire series. (Thus, if the interest rate on railroad bonds in March 1910 was two standard deviations below its average for the entire period, the Y-value on that date would be plotted as -2.0.) Normalization ensured that two data series had a similar scale on the Y-axis.
Because each graph had been drawn on a separate sheet of translucent tracing paper, by placing the two pages on top of one another on the glass surface of the box (lit from below), the two lines could be seen simultaneously and then rearranged by shifting one of the papers. Initially, the two graphs were aligned so that identical dates on the two charts overlapped. But the economist then began to move the top graph to the left and right, attempting to align the fluctuations on the graphs as closely as possible. By doing so, the observer could judge visually if the two series were correlated, either directly or with a time lag. Having obtained the best match possible, the economist then wrote down a qualitative evaluation of the direction and level of correlation (positive or negative; high, moderate, or low) and the best time lag, as well as the consistency of that lag. The same process was then repeated for 2 different time series – 20 overall, yielding 190 distinct comparisons – with each comparison made independently by 3 different observers, each using an illuminated box and tracing paper.
The visual comparisons of these 20 data series were just one step in the creation of a new “Index of General Business Conditions” by Harvard economist Warren Persons. Persons described his handiwork in 1919 for the second issue of The Review of Economic Statistics, a new journal established by Harvard’s Committee on Economic Research (Reference Persons1919: 117–27). This initial qualitative analysis of charts, Persons explained, “yields valuable results” but also remained “dependent upon the judgment of the observer,” whose “personal equation, preconceived notions, or theoretical bias influence his opinion” (Persons Reference Persons1919: 121). Accordingly, the comparisons that had shown the most promise were subject to the then-novel techniques of statistical correlation analysis. Based on the computed correlations and lags, Persons grouped the twenty series into five groups, which he then further reduced to a final three. These three groups showed a distinct chronological order in their fluctuations from 1903 through 1914, with Group A preceding Group B, which in turn preceded Group C. Taken together, the three series appeared to offer a reliable, forward-looking guide to the general cyclical fluctuations of American business.
The Harvard Index was both an exemplar of its time – the 1920s saw a boom in economic forecasting, underwritten in part by economists – and a site of multiple tensions. Persons, for example, contrasted the initial visual comparison of graphs (which was distorted by a “personal equation” and “preconceived notions”) with the more automatic, algorithmic analysis of statistical correlation that generated the final three groups of data series. Yet the visual comparison had been essential to whittle down the possible permutations to a manageable set for correlation analysis. Nor was Persons ever able to rid himself of the “personal equation”: as Laetitia Lenel (Reference Lenel2021) has shown, the predictive power of the three series broke down within a few years, and Persons and his colleagues began supplementing their statistical analyses with qualitative predictions and explanations drawn from regular correspondence with Federal Reserve officials, bankers, and other businessmen. Persons himself increasingly lost confidence in the applicability of probability theory to time-series data, and as Lenel rightly points out, the entire episode complicates any simple story about the adoption of probability-based statistical inference in economics.
But there were deeper and broader tensions as well. Epistemological practices (such as sliding graphs over an illuminated box or calculating correlation coefficients) are inescapably entangled with other practices, artifacts, and institutions. Most directly, for example, the creation of the Index depended upon certain material technologies, such as the availability of translucent cloth (used in art and drafting) and electric lighting. The data series themselves were products of government agencies, trade associations, and financial publications, all of which, for their own purposes, had rendered different business transactions quantitatively legible. But the epistemological practices underlying the Index were also a product of certain absences, albeit ones only recognizable in hindsight. Persons pioneered the elaborate visual comparison of charts (190 different combinations!) to determine the most likely candidates for correlation analysis because calculating coefficients for all the data series on all possible time lags would have been too computationally demanding. Today, of course, that task would be trivial. Epistemological practices emerge from both the positivities and gaps of their environment.
Because epistemological practices take form within a web of other practices, institutions, and artifacts, though, they can be either supported by those surrounding elements or torn apart as the web tugs in different directions. That is especially true when we think about people’s motivations. Epistemological practices take on meaning as they are embedded within higher-order practices; that is, when they are subsumed within a larger project. That project can be truncated at a relatively low level (“I am comparing these data series in order to construct a reliable predictor of business cycles,” Persons might have explained if asked). But it can also be pushed further. Repeatedly asking, “But why are you doing that?” can drive the explanation to higher or broader goals, in turn revealing latent tensions or ambiguities that in turn can rebound back on the lower-level practices. The Harvard Index of Business Conditions illustrates precisely this issue.
As James C. Bonbright (an economist and professor of finance at Columbia) pointed out in 1926, “You can study the [business] cycle simply in order to play with it; or you can study it in order to relieve unemployment and misery; or you can study it in order to make money by playing the stock market” (Wolfe et al. Reference Wolfe, Phillips, Bonbright and Day1926: 238). Those three potential larger goals (understanding the factors driving business cycles, mitigating their excesses, or making money from predictions) could all be connected to the Harvard Index, but they also pulled the enterprise in different directions. Harvard economics professor Charles Bullock (1869–1941) created the Harvard University Committee on Economic Research in 1917 to bolster Harvard’s otherwise lackluster reputation in the field, and the Committee established the Harvard Economic Service (HES) to improve forecasting techniques, the analysis of economic statistics, and the quality of economic data more generally. All of these could be construed as strictly scholarly, scientific goals, and indeed The Review of Economic Statistics (under Persons’s editorial leadership) became a major academic journal. But the HES was also very much a business, equally dedicated to selling economic information and analysis. Bullock offered the Review for free to “a limited number of libraries and scientific investigators,” but everyone else had to purchase a hefty package subscription ($100) that included a monthly statistical supplement, upgraded to a Weekly Letter in 1922 (Bullock Reference Bullock1919). Bullock hired salesmen to boost subscriptions, sent direct mail advertising (440,000 pieces in 1922 alone), and pitched the service to prominent Harvard alumni at invited dinners. By the mid-1920s, the HES had almost 2,400 subscribers. At its height, the HES employed a staff of forty, including fourteen economists and statisticians (Friedman Reference Friedman2014: 141–44). The business side of the HES was not a mere side hustle that could be detached at will; it underwrote the staff whose work made the labor-intensive forecasts (and its scientific research) possible.
The character of the Review as a scientific journal (with its norms of public information) lay in tension with Bullock’s other aim of attracting subscribers to the associated Weekly Letter, whose members paid for the private expertise of the Review editors. Unlike Babson or other commercial forecasters, HES shared its methodology openly, but even when the Index itself was produced mechanically according to published methods, Persons emphasized that its interpretation required expert analysis of changes in “underlying economic conditions” to reach final conclusions (Persons Reference 80Persons1920: 40; Friedman Reference Friedman2014: 129, 138). Here the “personal equation” that had troubled him earlier reared its ugly head. The importance of expert judgment only grew as the predictive power of the Index itself began to wane and the HES relied more and more on its network of informants. Yet as the HES abandoned the centrality of the statistical methods published in its Review, its ability to hold together its scientific and commercial aims began to falter. Writing privately in 1924, Wesley C. Mitchell complained that “just so far as the Harvard Group is relying on general information or broad observations or tips or ‘hunches’ they are precisely in the same position as any other group which makes no pretensions to scientific method.”Footnote 27 Mitchell had given high praise to “the scientific side” of the Review’s initial work, whose only blemish (in Mitchell’s mind) was Persons’s love of “puzzling mathematical problem[s]” over refining basic data (Mitchell Reference Mitchell1919: 873, 875). It was thus distressing to see Persons now giving shorter shrift to both data refinement and mathematical analysis in favor of the “tips or ‘hunches’” associated with commercial news services. Though the HES’s diminished reliance on correlation analysis may have been driven by the failure of those methods, that paradoxically undermined some of its academic credibility. Harvard began to feel the same way, and in 1928, the university divested itself from the HES. Tellingly, Warren Persons left that same year “to become a vice president of the National Investors Corporation, which had been organized by a former business manager of the Harvard Economic Service,” and he eventually founded his own consulting company (Friedman Reference Friedman2014: 155–56).
Though the tension between academic and commercial research formed the most prominent challenge for the HES, its basic aim (to predict the movements of business cycles) likewise ran at cross-purposes to another application: using the same information to mitigate those very cycles. Already in 1924, John Maynard Keynes had cautioned Bullock that the growing power of organized monetary policy, itself relying on the same statistics but aiming to dampen cyclical fluctuations, would disrupt the previous structural relationships the HES had identified. It had been a mistake, Keynes warned, to treat the business cycle like “a natural phenomenon such as the tide.”Footnote 28 Privately, Bullock agreed, and Lenel argues this was part of the HES’s decision to prioritize consultation and discussion with leading bankers and Federal Reserve officials rather than relying on correlations in historical data (Lenel Reference Lenel2021: 153). Since one rationale for studying business cycles was to change those cycles, the application of business cycle research could potentially undermine the very predictive power of that research.
By the early 1930s, the HES was struggling. It failed to forecast the Great Depression, an error that Bullock (ironically) blamed on having not followed a strictly “mechanical reading” of the service’s own statistics, and the divestiture by Harvard stripped it of a key distinguishing mark against other commercial forecasters (Lenel Reference Lenel2021: 156–57). In 1935, it shut down and handed the Review over to the Harvard economics department. In many respects, the collapse of the HES reflected the general failure of economists-turned-forecasters in the interwar era; Irving Fisher famously suffered a similar fate. But for historical epistemology, the HES also illustrates the dialectic between epistemological practices and the broader webs in which they are embedded. The HES’s techniques were made possible by the presence of specific material technologies and necessitated by the absence of others. The group had sought to exploit the teleological flexibility of business cycle research: it could be the foundation for a scientific journal and the grounding for a commercial newsletter. It could be read by academics and housed in a university, but sold to businessmen who hoped to profit from knowing the future and to government officials who hoped to alter that future. Its methods were rule-governed quantitative calculations (and hence could be offered as a contribution to economic science) and private judgments (and hence marketable expertise). The epistemic limits of its quantitative calculations pushed it closer to commercial expertise but likewise undermined its status as a scholarly enterprise, even though Persons and Bullock were undoubtedly correct that statistical correlations alone would be insufficient for forecasting. That the HES ultimately failed to hold its competing aims together shows how the flexibility of lower-level epistemic practices can be both a strength and a weakness for the production of knowledge.
Comment Verena Halsmayer
One of the central themes in historical epistemology is the role of practices in understanding the production, transmission, and effects of scientific knowledge. The premise is that all research, from empirical measurement to theoretical thought, is work that involves effort, instruments, and skills (e.g., Kaiser Reference Kaiser2005; Klein Reference Klein2003; Warwick Reference Warwick2003). While the history of economics may lack instruments such as particle accelerators or bubble chambers, it does involve a variety of “paper tools,” including models, diagrams, and, for that matter, indexes. Rather than tracing the development of disembodied ideas, historical epistemological studies of economics have focused on the practices of constructing, manipulating, and employing such artifacts. Tom’s treatment of the Harvard Index, informally known as the “Harvard Barometer,” exemplifies the significance of such an analysis. In the first place, he demonstrates how forecasting practices are entangled with a variety of other practices and technologies. Economists rely on the knowledge and the tools at their disposal: Rather than picking and choosing from a variety of existing theories, they use what is at hand and disregard what is out of reach. From this perspective, fixing together and working with an index depends as much on the situational enabling conditions as on the “personal equation” – the never trivial role of nontextual elements such as intuition, skill, and judgment.
One strength of focusing on practices is that it acknowledges the conflicts between different sets of practices and the internal tensions they might contain. Tom highlights that a specific research practice can be motivated by different endeavors, serve different goals, and can be interpreted in different ways at a given moment. In the case of the Harvard Economic Service, the flexibility of its calculative practice led to tensions between competing scientific and commercial aims, which eventually resulted in its discontinuation. By contrast, in other cases, the multivalent meanings of business cycle research helped stabilize its practices and institutions. At the National Bureau of Economic Research, as the historian Eli Cook has argued, specific calculation techniques aligned the Bureau’s visualization of the business cycle with diverse agendas (2017: 261). Here, a lower-level epistemic practice permitted different political references, from bankers and managers to corporate liberals and left-leaning technocrats. While there was no political harmony, different factions could still agree on the same economic phenomenon, and, in this sense, shared a way of seeing.
5 Debating Honesty Till Düppe
March 26, 1957. The debate had been dragging on for months. This was the fifth meeting where we, the party members of the new economics institute at the Academy of Sciences in East Berlin, discussed the same issue over and over again. The party, naturally, was the Socialist Unity Party (SED), and nearly everyone at the institute was a loyal member. The subject was a small booklet written by the esteemed director of one department, Friedrich Behrens, and his promising student, Arne Benary. They rightly argued that under socialism, as Marx envisioned, the role of the state would gradually wither away, giving space to decentralized, bottom-up planning similar to what happened in Yugoslavia. Workers would learn to self-govern. We had discussed an early draft already last year, approvingly, and the booklet was printed and ready to be shipped out. But after the Hungarian uprising in the fall of 1956, the party decided not to distribute it. Since then, one meeting after another was scheduled to discuss whether their ideas should be dismissed as “revisionist,” that is, they were not true to Marx, essentially no better than social democrats. Strange things happened in these meetings, and hardly anyone wanted to give in. But this meeting at the end of March was different. Two men arrived unannounced, their identities undisclosed, and they conspicuously recorded each member’s stance on an official resolution condemning the ideas as revisionist. Everyone understood the stakes. How did it get this far?
Before it all began, we were all deeply devoted to the virtue of honest scientific debate. This was not only a reflection of our scientific ethos but also a mandate of our party’s constitution. “Party members are obliged to develop self-criticism and criticism from below, to fearlessly expose shortcomings …, to fight resolutely against any attempt to suppress criticism and to replace it with whitewashing and adulation.”Footnote 29 It may seem odd today, but we took these lines very seriously. Epistemologically, socialism was to create better conditions for science to thrive, as only in socialism could knowledge be free from class interests. Under the party’s guidance, society was expected to be free from dogmatism. The 20th Party Congress in Moscow the preceding year, in February, when Khrushchev broke with Stalin, made us hope these promises would come true. Fighting dogmatism, we thought as economists, also meant ending Stalin’s rigid centralism. Behrens and Benary’s booklet pointed exactly in this direction. The first discussions we had last summer over its draft were vivid. How refreshing it was to read lines like:
The view that the state can do everything, and that every affair, even the most private, has to be managed and controlled by the state, is not socialist, but ‘Prussian’, that is Junker-monopolistic. Socialist, that is, Marxist-Leninist, is the view of the dying out of the state as the socialist relations of production become stronger, and the capitalist threat becomes ineffective; but this means that the self-government of the working masses by the state must find its complement in the self-government of the economy.
We truly believed this was the right path, especially given the reforms in Yugoslavia, Poland, and Hungary. But everything changed on November 4, 1956, when Soviet tanks crushed Hungary’s democratically inclined government. Our First Secretary, Walter Ulbricht, feared a similar scenario in the GDR. To solidify his power, he initiated what we now recognize as a “campaign” against various individuals and groups advocating for democratic reforms, labeling them “revisionist.” Our institute was one of his targets.
And so it happened that on December 15, the CC Secretariat (the executive committee below the Politburo) decided not to ship the booklet because it “currently presents the grave danger of politically disorienting the reader.”Footnote 30 Naturally, we thought of censorship and were perplexed about how to behave. The CC Secretariat was informed about the questions we asked ourselves:
How should the scientist behave if the results of his investigations contradict the views of the party leadership or party decisions? How does a scientist behave when his research results cannot be published due to the political situation? … Can there be a scientific debate if the scientist cannot publish his research results? What role and tasks does the party apparatus have in connection with the above questions?Footnote 31
Ulbricht understood that the integrity of his relationship with the academic intelligentsia was at risk. To maintain scientific credibility, he wished to beat Behrens and Benary at their own game: an open scientific debate. On January 15, the CC Secretariat thus decided that the booklet should be published in a more specialized journal, including counter statements from economists known to be critical. For the general reader, their ideas would appear as revisionist, a verdict supported by scientific consensus. But how about us, the supposed new elite of academic economists? Hence, we, too, had to “debate” the issue. “Since the views contained in the booklet are widely used in academia and other circles,” the minutes of the CC Secretariat stated, “it is necessary … to have a debate about these false theories.”Footnote 32 The party wanted to “nudge” us to a scientific consensus, as one would say today.
What we did not know, but suspected, was that our discussions were being orchestrated by the Ministry of State Security (Stasi). The Stasi aimed to stage a “show debate,” similar to the well-known show trials. Just as the trials were designed to appear just, so was our debate meant to appear scientific. More concretely, the Stasi launched what they dubbed “Operation Theory” with the goal of sowing division, and thus to steer us into a position to publicly label Behrens and Benary’s ideas as both dogmatic and revisionist.Footnote 33 The goal of the operation was “to isolate Comrade Benary and to support the work of the party-group such that Benary, with our help, would be ultimately removed from the institute.”Footnote 34 The two main Stasi informants at the meetings were the Contact Person (KP) Karl Bichtler, codenamed “Bischak,” party secretary of the institute, and the Secret Informant (GI) Günther Ulisch, codenamed “Walter,” assistant of the head of the institute. After each meeting, they met with the Stasi officer, prepared detailed reports, and strategized on how to direct the discussion. We sensed something was off, but it was not until recently, through an historical article that utilized these reports to reconstruct the debate, that we learned the full extent of the Stasi’s involvement in manipulating our discussions (Düppe Reference Düppe2021).
The Stasi began by investigating Benary’s private life. They interrogated former colleagues, his neighbours, surveilled his apartment, read his letters, and recorded his phone calls. They even considered recruiting a “slim” woman to gather information from him. But they found little that could discredit him personally. In our first meeting on January 23, none of us, except of course Bichtler, supported the claim of the booklet being “revisionist.” And this despite the fact that in this and subsequent meetings, one or another member of the CC Secretariat had been oddly present. Behrens boldly confronted them by saying they had no expertise to judge the book’s ideas. It was basic research, he argued. The first meeting ended in favor of Behrens and Benary.
Pressure increased when some days later Ulbricht publicly denounced Benary at the thirtieth Plenum of the CC in front of the entire East German public. He dedicated an entire section to the fight against dogmatism and claimed that Benary was “at war with Marxism” (Ulbricht Reference Ulbricht1957: 50). He associated dogmatism with the “simple reproduction of dogmas without connection to life and praxis” (Ibid.: 72). The connection to life and praxis, of course, meant that ideas should be relevant to the latest party decisions. He thus claimed political authority over what counted as scientific, and our further meetings would decide not only about our approval of Ulbricht’s economic policy but also about the future of the professional identity of academic economists. After Ulbricht’s speech, all eyes were on us.
With Ulbricht’s backing, the Stasi hoped to undermine the consensus at the institute. The plan for the upcoming meetings, unknown to us, was as follows:
Next week, another party gathering on the booklet will take place. … this meeting will serve to show the other comrades that the theories are wrong and can lead to dangerous political conclusions. To this end, the existing agents (GI and KP) must take an offensive position and, above all, stress the danger of such ideas in order to isolate Behrens and Benary. In the second meeting, which would take place in two to three weeks, the GI and KP should openly address the fact of Behrens and Benary being against the party and call for removal from the institute and initiation of a party trial.Footnote 35
But this was not to happen. We did not bow to Ulbricht’s authority. In the meeting on February 15, Benary presented a statement accepting that he might have made theoretical mistakes but refused the label of revisionism. How could he, because this would have meant giving up loyalty to the party! Or should he have instead accepted the label out of loyalty to the party? He really was in a quandary but tried to be as frank as possible: “Nobody will expect me to condemn right now as false all my theoretical ideas that I have developed over a long period of time. Above all, nobody will be served if I engage in ‘general self-criticism’ without being really convinced.”Footnote 36 How right he was, everyone agreed.
In the next meeting, on February 27, the counter statements to be published together with the booklet circulated. Behrens was enraged. He denied that his book was revisionist because he did not touch Marx’s basic principles. He called the articles written against him a “defamation” and put in words what everyone thought: that “at present a campaign is started against him … They try to plant things onto me … What happens has nothing to do with a scientific dispute.”Footnote 37 This was Behrens’s last intervention, as he did not come again. He must have given up his belief in the rules of the game.
Then came the meeting on March 13. It felt like a showdown. Bichtler presented all sorts of discrediting evidence against Behrens and Benary. We knew this information came from Kurt Zieschang, one of Behrens’s assistants, as Zieschang had presented the same rumors in the department heads’ meeting just before our meeting. What we suspected, and what was confirmed by the recent article, is that Zieschang had cooperated with the Stasi. He had been interrogated in three separate meetings during February, in which he divulged all sorts of true and half-true information. He might have been under personal pressure, though we do not know for certain. He claimed, for example, that Behrens and Benary had contacts with the enemy, particularly the West German SPD, and said, “Behrens wants to represent his own doctrine, as every bourgeois scientist wants to have a theory associated with his name.”Footnote 38 Most importantly, he spoke of a “program” which he may have simply invented, as nobody else confirmed its existence:
Some weeks ago Zieschang walked into Benary’s room, Mann was also present. Comrade Mann had several pages in his hand and behaved very secretive. Zieschang, who wanted to read them, did not get them, so he tore them from Comrade Mann. However, they were immediately taken away from Benary, who pocketed them. Zieschang said he had recognized some sort of program and read slogans such as ‘freedom of expression, freedom of the press’ and the like.Footnote 39
The Stasi knew that these claims went too far even for our democratic taste. Democratic reform, we always thought, had to come from below, from the economic base, not the political superstructure. Benary, following Bichtler’s presentation, denied the existence of this counterrevolutionary program. Indeed, all of this seemed so construed and our reactions were rather startled. One of us felt offended and said that he “refuses constructing artificially political factions.”Footnote 40 Others excused Benary by pointing to the general confusion in the institute after the Twentieth Party Congress, arguing that it was wrong to judge comrades one-sidedly and that any personal allegations should be thoroughly checked. When Bichtler proposed that Benary be removed from the institute, nobody said a word, as he reported later: “Not a single comrade said anything about the Contact Person’s demand to remove Benary from the institute. It was agreed to hold another group meeting next week and continue the discussion.”Footnote 41
It was in this last meeting on March 26 that Bichtler proposed a formal resolution about the revisionist character of Behrens’s and Benary’s ideas. It was then that the two unknown men, who everyone knew to be officers of the Stasi, were present. We knew what was at stake. Our careers. Nobody voted against the resolution. The agreement had been obviously forced. Even the informant of the Stasi reported that “those from Behrens’s group had only agreed on the resolution because they saw no other way out for themselves.”Footnote 42 Saving our careers, we knew our decision was the end of Behrens and Benary’s careers. Behrens was stripped of his higher functions, including his position as head of the Central Statistical Office, and Benary was moved from academia to industry. The goal of the Stasi was achieved. Formally. A forced consensus appears better than tanks shooting into masses, but both ultimately were state force.
And how fake the resolution was! It claimed to be the result of “thorough, severe, and serious discussion.”Footnote 43 It declared Behrens’s and Benary’s ideas revisionist because of a general disregard for the party’s leading role. Even worse, a month later, we published a second statement in our name, expressing self-criticism about the spirit of our scientific debate. We regretted that Benary had spread a “petty-bourgeois atmosphere” in the institute, fostering an “individualist work ethos.” Behrens’s notions of self-governance were deemed too “theoretical” and too far removed from practice. Our mistake was that we did not thoroughly discuss them, we wrote, thus treating them “dogmatically.” In other words, we agreed to equate dogmatism with anything not in line with the party! “It is imperative that the party groups watch over an open, factual, undogmatic and comradely atmosphere that includes, however, ruthlessness against deviations from Marxism-Leninism. Petty-bourgeois sensitivities must be persistently fought. The conviction must be created that we can only live up to our political-ideological demands if all important issues are discussed to the end.”Footnote 44
After all, Ulbricht not only made his political opponents appear illegitimate for being against Marx, but he also won the epistemological battle over the meaning of dogmatism and scientificity. If party decisions define that which is “practically relevant,” everything departing from it can be dubbed as dogmatic and bourgeois intellectualism. Despite obvious oppression, the party managed to maintain the facade of a democratic, science-driven state devoid of censorship.
Ironically, some years later, after the building of the wall in 1961, Ulbricht fostered reforms that every one of us recognized as being influenced by Behrens’s and Benary’s ideas. While this rekindled hope among many academic economists, the long-lasting effect of the “revisionism debate” on the profession was profound. Today, many of us regret, and are ashamed of, what we did. We missed a chance to establish a better, truly democratic socialism. Never again will someone write as freely and sincerely as Behrens did. In the wake of the debate, a culture of self-censorship took root, stifling intellectual discourse. The promise that socialism would create better conditions for science to prosper was defeated.
It is rare that historical documents allow historians to get that close to how people think on their feet. The secret police’s acute observations and several interviews I conducted with participants and relatives provided a rare glimpse. Since my historical labor in reconstructing who said what was unavoidably one of empathy, the idea of referring to “them” as “we” came quite naturally as I wrote this section. The difference between the formal historian’s voice and the imagined voice of the party group members seemed little more than a formality. Well aware of the risk of becoming overly sympathetic and of the possible bias of my sources, I consciously allowed myself to adopt the fictitious “we” since their perspective is key to understanding how the regime balanced power and knowledge.Footnote 45 It was their subtle inner moral conflict between maintaining honesty and avoiding dogmatism that was “micro-managed” by the regime. Moreover, it was through this inner struggle that the personae of the East German economist under socialism took shape.
Comment Thomas A. Stapleford
Till’s reconstruction of the quandary faced by East German economists under the rule of the Socialist Unity Party, grounded in his archival research and personal interviews, engages with practices on two levels: the practices of economics in Cold War East Germany and the practice of writing history itself.
Regarding the former, it would be easy to tell this story as a conflict of ideas: What constituted a genuine “scientific debate”? What counted as “dogmatism” or “revisionism” (i.e., opposition to Marx)?Footnote 46 A more traditional history of ideas would take precisely that approach. But Till draws our attention to the practices and artifacts that undergirded this debate. The economists initially crafted a discussion grounded in a text (the party constitution) that promoted “criticism from below” and decried “whitewashing,” which to the economists meant that their institute could form its own independent judgments about how to interpret Marx. Meanwhile, however, a second set of practices was in operation with its own material technologies: spies, secret reports, interrogations, public condemnations, Stasi agents, and the whole apparatus of a police state. Till shows us that the debate about how to make a collective scientific judgment was over before it began because it was already structured by the practices defining “good science” according to party leaders, viz., a judgment that accorded with the desires of party officials. The drama in Till’s story lies in the economists’ gradual realization of hidden practices that had been in operation all along.
That drama is heightened by Till’s choice to narrate this episode as a fictionalized, collective reckoning with the past by the institute economists, which forces the reader to think of the interacting practices of memory among agents and historians. We are shown both how the institute’s members saw this situation at the time and what they subsequently learned; as the narrative proceeds, we see how they remain haunted by the past. Should the economists have continued to advocate for Behrens and Benary as the machinery of the state closed in around them? The dilemma is left unanswered and ambiguous. So, too, is the accuracy of the account itself. The economists’ reflections are fictional – Till’s own creation, based on his sources – but how much trust can we place in the sources themselves, documents written by Stasi agents or recollected long after the events? Till’s practice of weaving archival evidence together with fictional commentary from anonymous members who are never named highlights the inevitable uncertainty and ambiguity of history itself.
6 Planners’ Precision Verena Halsmayer
On Sunday morning, October 13, 1963, Pope Paul VI opened the annual study week of the Académie Pontificale des Sciences in the Apostolic Palace in the Vatican. While previous study weeks focused on the natural sciences, this one, for the first time, addressed one of the new mathematical social sciences. Its title: “Le rôle de l’analyse économétrique dans la formulation de plans de développement.” The pope welcomed some of the most prominent representatives of econometrics, many of whom (and all of them male) worked as government advisers.Footnote 47 Paul VI praised econometrics as a new science that would contribute significantly to the “orderly progress of human civilization” (Pontificia Academia Scientiarum 1965: XXXVII–XXXVIII). In contrast to earlier beliefs, the chancellor of the academy continued, it was now obvious that the free play of individual decisions did not guarantee favorable results for the collective. Rather, historical experience showed the need to set common goals and to work out possible ways to achieve them using “instruments of knowledge and control” (Pontificia Academia Scientiarum 1965: first page of the preface, unnumbered). This reformulation of the old liberal juxtaposition of individual and general interests in terms of a technical problem heralded a week of discussions on the governmental potential of econometric knowledge.
Perhaps contrary to what one might expect, the study week was no peaceful get-together of satisfied economists who had finally made it to the ranks of proper science. The proceedings of the study week, more than 1,500 pages strong, document controversial debates about whether and how econometrics could make good on the claim of contemporary social science “to speak precision to power” (Porter Reference Porter2006). Among the participants were several self-identified “planners” and “social engineers” who worked in both academia and governmental bodies. Several were associated with European social democracies. Their contributions to the Vatican’s proceedings offer some indications of a specific culture of doing economics that was not only confident about its general, rather abstract, governmental importance but also claimed direct relevance to policymaking. Other participants vigorously contested such involvement in political decision-making processes. The proceedings of the study week do not contain any serious discussion of the social and political values supported by econometric knowledge. They did record, however, scattered remarks on the implicit values guiding the practices of econometric research.Footnote 48
Whether politically interventionist or not, practices of knowledge-making never take place in value-free spaces. After all, the very objects of scientific knowledge are based on ideas of what a specific community can, wants to, and should know. Moreover, the related research practices come with preferred methods that make a body of knowledge “scientific.” This essay focuses on differences in epistemic values among the participants of the Vatican study week, especially their diverging notions of “precision” and what kind of econometric precision was warranted. Epistemic values not only shift over time, they also differ within disciplines and, as is the case here, subfields. What was this “precision” that seemed so constitutive for this specific culture of doing economics?
Reflecting broader debates about planning in the 1960s, the study week’s “plans de developpement” did not, unlike earlier discussions, refer to socialization of the means of production. Neither did they relate to a comprehensive political program, not even to concrete policy measures. Instead, “planning” referred to the development of techniques for organizing decision-making. In this way, the study week followed a shift in discourses about planning from a primary focus on the very possibility of planning and its relation with democratic ideals in the 1930s (today commonly related to figures such as Barbara Wootton, Karl Mannheim, or Friedrich Hayek) to an emphasis on the improvement of planning technologies. In an industrial, mass consumption society, the necessity of planning was widely acknowledged as the “rational organization” of economies. Whether centrally planned economies, Western macroeconomic management, or “development planning” – at the center was the technical design of the procedures that shaped policy measures. In this sense, planning encompassed a variety of practices stretching from the making of new data to the coordination of decision-makers, from modeling sets of possible long-term goals to establishing strategies for achieving them. In effect, “good planning and bad planning,” as one of the participants in the Vatican, the British econometrician Richard Stone, noted, was a matter of the “functional and unfunctional design” of decision-making processes (Stone Reference Stone1965: 26).
While the analysis of the goals and instruments of economic interventionism was by no means new, its framing in terms of econometric planning models brought forth a new problematization (Morgan Reference Morgan, Giraud and Duarte2020). From the 1960s onward, different institutions – from planning offices and public administration to central banks and businesses – increasingly used macroeconometric models as decision-support systems, mostly as instruments for forecasting and evaluating different strategies (Boumans and Duarte Reference Boumans and Garcia Duarte2019). Above all, they engaged the problems of national economic management on a scale previously neither analytically nor experimentally manageable without sufficient computing capacities. Like other simulation sciences, this type of econometrics enabled the design of different possible worlds in the distant future and the creation of different paths to reach them from the present. When it came to planning models, the related techniques and procedures were not merely thought of as resources for designing action (as might be the case for the kind of “planning” done at the Cowles Commission) but to be actually employed as mediums for organizing decision-making. Labeled “development programming,” “planning in stages,” or “decision modeling,” such model-guided processes came with a different relation between economic knowledge and its object: models did not provide instructions on how to intervene in an economy external to their analysis but were used directly to shape decision-making processes as part of this economy. These different uses were also the bone of contention at the Vatican, not least because they brought forward different roles for the econometrician as a planner and diverging notions of “precision” as an epistemic value.
The precision advocated by econometric planners extended the kind of precision to which econometric analyses had been committed since the 1930s. The founding documents of the Econometric Society from 1930 referred to the exemplars of biometrics and psychometrics; economic questions were now equally subjected to scientific, that is, mathematical-statistical, treatment. Earlier approaches to political economy such as institutionalism or the quantitative research related to the Harvard Barometer now turned into purely “descriptive” undertakings; the search for patterns in data sets was replaced by the systematic analysis of regularities (Desrosières Reference Desrosières1998: 316–17). By constructing mathematical control systems, measurement processes should exhibit as little variance as possible in repeated measurements. After all, the guiding maxim of such research was not empirical accuracy but mathematical precision: the aim was to integrate non-knowledge ever more precisely in the form of error terms by refining estimation models (Wise Reference Wise1995, Morgan Reference Morgan, Daston and Lunbeck2011: 307–8, Biddle and Boumans Reference Biddle and Boumans2021). In the first place, however, such econometric knowledge had “explanatory” and “forecasting” purposes, as the French physicist and economist Maurice Allais put it when he took a stance against econometric planning in the Vatican (Dupont-Kieffer Reference Dupont-Kieffer2019: section 3). He thought econometrics had merely indirect governmental relevance in that it could be used for policy recommendations.
Other participants in the study week argued for a direct involvement of the econometrician in policymaking. Also here, models were the central instruments that would ensure precision. But the planners’ precision did not merely serve analytical and forecasting purposes. Primarily, it related to the goal of a rational design and orderly implementation of political decision-making – which equipped the term with a different meaning. The Vatican presentation of Richard Stone, which he had already published as a progress report of the Ford Foundation-funded Cambridge Growth Project (CGP), illustrates the kind of precision that econometric planners aimed at. As the first speaker in the study week’s program, he talked about the CGP’s Programme of Growth, which built on the Cambridge Multisectoral Dynamic Model of the British economy. In contrast to existing growth policies, which were based on Treasury’s short-term projections, the CGP presented a desirable long-run growth rate and various paths along which this rate could be approximated via a take-off within the next ten years. The central contribution of the project, as Stone emphasized, was the linkage of the modeling and the planning processes.
In his presentation, Stone delineated how the project updated the imagined long run with current trends and iteratively aligned the two. One of the flowcharts he presented showed “the model in its environment.” Boxes labelled as “theories,” “facts,” “objectives,” “model,” “controls,” “policy” (meaning administrative procedures), “plan,” and “experience” were interconnected to the model via feedback loops. The box “events” presented a noncontrollable component that influenced the system from the outside (Stone Reference Stone1965: 9). Under the motto solvitur ambulando, the econometrician’s task was to construct a prototype model and adapt it to the constantly changing environment. In a mixture of organicist and mechanist metaphors that was characteristic of information science and cybernetic approaches, the project described its model as an organic entity. The structure of the model changed in the course of an iteration process, thereby integrating different bodies of knowledge and non-knowledge, varied goals and interests in a procedural manner.
This form of planning tied the epistemic value of precision to the adequacy of the model for the purpose at hand. Most importantly, a model had to be able to integrate heterogeneous planning components – such as data, experts, mathematical equations, policy goals, institutions, national accounts, legal frameworks, and bureaucratic procedures – into a working control system (van den Bogaard Reference Van den Bogaard1998). As media for decentralized decision-making, such models did not yield optimal solutions but produced “workable compromises,” as another participant of the study week, the Norwegian econometrician Ragnar Frisch (Reference Frisch1965: 1219), put it. A model’s precision increased not by refining measurement techniques but through trial and error. Ideally, goals and plans were modified according to changing circumstances and new data collections. As part of the economy, this form of planning made not only administrative structures but also economic structures more efficient.Footnote 49
If the precision of a model was about its capacities to enable adequate intervention through its local and situational adaptability, what did that mean for the role of econometricians? What were the characteristics, dispositions, and skills that made them good econometric planners?Footnote 50 Back at the community meeting in the Vatican, admittedly an ironically apt setting for discussing democratic and decentralized decision-making, Frisch spoke of economists’ “steering work” (Reference Frisch1965: 1198) and their role as “social engineers” (1202). In his version of model-guided iterative decision-making processes, the econometrician was expected to “help the politicians work out an economic system built upon a set of incentives” (1198). Neither he nor the others addressed directly the planner’s virtues, but his notes let us speculate about a specific persona tied to the planners’ notion of precision.
According to Frisch’s recounting of his own experience, the econometrician translated between model and policymaker (and hereby the electorate) in a series of interviews. “You start with a first and tentative formulation of the policy maker’s gross preference function” (Frisch Reference Frisch1965: 1223). Decision-makers were presented with a simplified version of the model and supposed to utter their wishes and adapt them, in a series of steps back and forth, to the alternatives that the computer model was able to calculate and evaluate. Akin to the process described by Stone, here, both political goals and instruments had to be framed precisely. Adaptability was therefore not only required of the model but also of decision-makers. After all, “meaningful” preferences were only those that fit the model in the first place.
Initially, Frisch neutralized the role of the econometrician as a mere mediator between model and decision-maker. Simplifying and interpreting the model on behalf of the econometrician was framed as a didactic activity, simply serving a learning effect. Frisch did not address the role of an econometrician’s prestige (which he enjoyed as a widely known professor and Nobel Prize winner), the authority with which an economist was able to claim knowledge, or the role of geopolitical strategies and power in political decision-making. Neither did Frisch recount the specific settings of the interviews, the aspects that factored in political decision-making, the specific characteristics of administrative procedures, or the role of bureaucratic tools. What he did emphasize was the problem of non-knowledge on behalf of the decision-maker. In effect, it was necessary to “make him give up his own ideas.” Comparing the econometric planner with a psychiatrist, he was not only the one with the proper knowledge but ultimately also responsible for, if necessary, brutally instilling the right ideas into a recalcitrant policymaker. The problem was that decision-makers were not econometricians; ideally, they would be computers, machines for rational decision-making based on the econometrician’s model.
Econometricians’ views of their role in policymaking were not descriptions of what happened. But they did shape planners’ involvement and actions. Investigating the specific intellectual and ethical disciplines that framed econometric planners’ selves opens up the history of economics to larger questions of the functional aspects and political significance of epistemic values, as discussed in the next, third part of this Element. Remarkably, a joint final statement of the pontifical study week’s proceedings, signed by all the participants, only highlighted the least common denominator, mathematical rigor, as the main strength of econometric research. The fundamental question discussed during the study week, whether and how econometrics could provide interventionist knowledge, was omitted in the closing notes. That the “precision” of those participants who insisted on the strengths of models as direct tools for planning had become the point of contention in a sub-disciplinary clash led to its downgrading in the historical record. Following up on notes such as Frisch’s is one way to investigate the varied meanings, values, and virtues of postwar mathematical economics.
Comment Thomas A. Stapleford
Verena’s analysis of the 1963 Vatican study week on econometrics and planning illustrates how attending to different levels of disagreement in economists’ practices can unearth splits that may or may not map onto discontinuities at other layers. Verena begins by acknowledging that “planning” was a broad and ubiquitous term in the early 1960s. It could refer to comprehensive, centralized planning in socialist economies or looser macroeconomic planning for capitalist ones, or even developmental planning for those seeking to move from one economic system to another. Planning thus seemed to denote only what the Pontifical Academy called the use of “instruments of knowledge and control.” Nor indeed was there any consensus about how effective or refined these instruments might be in the near future. In that context, it is no surprise that participants differed on many points despite avoiding sustained arguments about the ultimate aims of planning.
Rather than following these familiar debates about political economy (e.g., Goodwin Reference Goodwin1998; Fourcade Reference Fourcade2009), however, Verena has focused on the subtle splits around another ubiquitous term: precision. For some participants, valuing precision meant valuing the extent to which an econometric model could capture, in close detail, the various features of an economy. This was what we might call representative precision. Other participants, such as Richard Stone, used precision to identify the practical utility of a control system for policy purposes. This was policy precision: a precise planning process was one which allowed the planner to maximize rational control along desired lines. In that approach, the model itself might be modified iteratively (as Stone suggested) through a trial-and-error process; it was precise not necessarily because it represented the economy at a high level of detail but because it enabled precise control.
These competing notions of precision were tied to competing goals for econometricians (build the most representative model or the most practically useful model) and hence different conceptions of the ideal econometrician. For Ragnar Frisch, a proponent of policy precision, the ideal econometrician should mediate between the model and the politician, posing the politician’s desires to the model and reporting back what the model said, and then using that process to refine the model itself. Yet, as Verena notes, this was both a simplification and a distortion. The model, after all, was a creation of the econometrician, and thus another way to understand the process was one of education: the successful econometrician could reshape the policymaker’s understanding of the world, its salient variables, and its functional levers. If policy precision entailed building effective planning models, Frisch was implicitly suggesting that making precise (effective) models also required making precise (effective) policymakers, and that the ideal econometrician could do both.
THEME III PERSONS AND ECONOMIC KNOWLEDGE
7 Keeping the Books Verena Halsmayer
Rent, health insurance, old-age insurance, telephone, light, record player, two pillows, newspaper, and piano lesson. These items appear on a page from five account books of a German upper-middle-class household, cowritten by Lore Seglitz and her husband Hans-Joachim. Their books span almost seven decades, from August 1939, a few months after their wedding, to the widower’s passing in March 2004. Like all the other entries, the page of January 1964 records various kinds of expenditures and compares them with the available income. As on each January page, there is an annual balance for the previous year. Inserted scrap papers list detailed expenses for groceries. Their totals were transferred to the book under the category of “household expenses.”Footnote 51
Household accounts have had a long career as particularly promising sources for social and economic studies, at least from eighteenth-century statistics to today’s financial anthropology. Typically, at the center of research with account books is what they record: prices, quantities, and types of products. Economic historians argue that because documents such as the Seglitzs’ books were not made for the purpose of historical transmission but had rather practical, existential functions, they could be treated as “relics” in line with artifacts such as clothes, tools, and administrative files (Fischer Reference Fischer2011: 65). For more than a century, empirical research has fostered the argument that “meticulously kept household accounts provide an almost unexploited material for exact economic research” (Bücher Reference Bücher1906: 700, cited in Fischer Reference Fischer2011: 66, translation and emphasis VH).
From the perspective of historical epistemology, this notion of research “material” invites closer attention. Treating something as material, as recent work in the history of science has highlighted, assigns it a specific function in the research process: The material (such as data, observations, or sources) is decidedly different from the object of research. The latter consolidates a specific research interest, a specific problematization. The material, in turn, is chosen for its usefulness and appropriateness regarding that very object (Hoffmann and Hagner Reference Hoffmann, Hagner, Hagner and Hoffmann2018: 7). In this sense, everyday writing practices such as listing expenses, budgeting, and bookkeeping become subsidiary activities in economic knowledge-making. In what follows, I will focus on the differences between qualities that made accounts proper research material on the one hand and the characteristics of books as part of an everyday, personal, sometimes idiosyncratic writing practice such as the Seglitzs’ on the other. Not least, this difference illustrates how research shapes and reshapes the world it seeks to investigate.
Quantitative research with account books in the late eighteenth century stretched from political–arithmetical analyses of the body politic as a whole to investigations of the cost of living of the poor; approaches ranged from the investigation of a handful of peasants’ books to standardized questionnaires. With the rise of the “social question” and the associated search for scientifically founded policymaking, surveys became the prime strategy to find out about the living conditions of urban industrial workers and peasants. Large-scale enquêtes involved intensive monographic investigations and extensive statistical studies – both of which used household accounts as their prime research material (Desrosières Reference Desrosières, Herrnstadt and Renard2025). As survey researchers developed different methodological approaches, they also differentiated between found and prompted accounts.Footnote 52 To them, the advantage of found accounts was that scientific deliberation had played no role in keeping them. They served those who kept them as means of economic self-assurance (as in the case of the Seglitzs’) and financial justification toward the authorities.
True to these traditions of social research, the economic historian Toni Pierenkemper (Reference Pierenkemper1988) related his late-1980s discussion of household accounts as a source to those of his nineteenth-century “precursors.” His survey sketched the development of quantitative methods as circling around a set of problems that beset their material base from the start. Among them were (1) the number of accounts (not large enough for proper statistical analysis), (2) the composition of accounts (not all parts of the population kept books, as few were familiar with the necessary cultural techniques), and (3) the time span of bookkeeping (most books were not kept long enough to capture seasonal or lifetime fluctuations). In this sense, found account books came with a set of standard problems for the quantitative researcher that consisted, among others, in issues of comparability, standardization, and weighting. One thing account books had to lose in order to qualify as quantitative research material was the idiosyncrasies of their writers and their specific writing practices.
Research with account books, whether directed at living standards or, increasingly from the 1920s onward, at consumption and consumer behavior, intersected with efforts to rationalize household practices. As in the US (Stapleford Reference Stapleford2004: 95), some German social reformers held that if scarce resources of the lower (middle-) classes did not increase to a considerable extent, they should at least be taught to economize. Moreover, this would make the economic activities of poor(er) families legible and manageable, especially those of working-class families. Not only social statistics and Staatswissenschaft embraced the household and the work of the “housewife” as quintessential for the national economy. In the 1920s, with renewed focus on male breadwinner models, the middle-class household became a popular site of economic hope. Research-based guides for rational homemaking were part of a wider endeavor of reforming life, including home economics education and household consulting services.Footnote 53 Prominent economists and social reformers such as Henriette Fürth and Alice Salomon championed the importance of bookkeeping (Taratko Reference Taratko2022: 766). In their publications, household budgets appeared as a means for both national economic reconstruction and self-fashioning. In the 1920s and 1930s, budgets were a “surprise hit genre” (Wilkowske Reference Wilkowske2017). Upper-middle-class women in both the US and the Weimar Republic enthused over (note)books with structured pages of ruled paper containing tables or matrices.
Reformers such as Erna Meyer, one of the best-known German household experts in the interwar period, found most popular books too simple to capture everyday household and life management (Taratko Reference Taratko2022: 763). The demands that she and her colleagues developed for proper bookkeeping went beyond a specifically formatted paper tool to encompass a whole set of practices. Meyer’s best-selling guide, published in Schriften des Vereins für Sozialpolitik alongside works of Gustav Schmoller and Max Weber, presented strategies of self-improvement based on meticulous daily recordkeeping. The “housewife” should stick to a systematic and continuous practice of making notes; ideally daily, like brushing teeth.Footnote 54 At the center of this discipline was the account book or a system of books to track expenses and compare them with the available funds. Registering monetary data and organizing it in tables brought the ordinary events of everyday life into a concrete order and allocated them to specific categories of life and expenses. It was a rationalizing tool against what Meyer presented as objectionable forgetfulness: forgetting singular expenses as well as a moral absent-mindedness (Wimmer Reference Wimmer2011: 135–36).
The qualities that made a good (middle-class) housewife resembled those that made account books good material for social research. In her life’s endeavor to study the black box of consumption, Meyer demanded that accounts should be exact and that there should be a specific consistent method of bookkeeping (Taratko Reference Taratko2022: 762). Among other problems listed earlier (that link more strongly to late twentieth-century empirical methodology), Pierenkemper mentioned that the correctness and honesty of household accounts were dubious – due to laziness, convenience, or polishing (especially when it came to entries such as tobacco and alcohol). This was also one of the problematic aspects of private bookkeeping that Meyer sought to confront. She did so by supplying the rational homemaker with an accounting infrastructure that was intended to help them not only live more rationally but also provide correct numbers and useful governmental knowledge about the smallest entity of the nation-state.
Books like those of Lore and Hans-Joachim Seglitz echo household experts’ guidelines. After all, most of such more detailed, longer-term budgets were written by women with a corresponding training or profession as accountants, office staff, or paralegals. Lore Seglitz attended the Höhere Töchterschule in Chemnitz and worked as an executive secretary in a large construction company. Her husband was an economist, visible in the prints on the back of the scrap papers used to sum up daily expenses.Footnote 55 However, regardless of whether the writers studied bookkeeping in school or practiced a corresponding profession, keeping a book did not mean simply following rules. Rather, individual procedures evolved over decades from developing, training, and stabilizing a specific writing practice. In the Seglitzs’ books, the writing remained relatively compact over time, though some categories were added (e.g., pocket money: “Taschengeld H.” and “Taschengeld L.”) and some were abandoned (e.g., alcohol).
Not only were categorizations diverse and changed over the course of an account book’s lifetime. Household accounts also differed from guidelines in detail and comprehensiveness. Meyer recommended a breakdown of expenses as detailed as possible to enable precise allocation and warned against using overly comprehensive categories such as “miscellaneous.” In contrast, the Seglitzs’ books, though meticulous, involve large categories such as “household money.” The Seglitzs’ accounts are part of a rare archival collection that keeps a variety of household books, among other types of life writing (Gerhalter Reference Gerhalter and Feigl2016). Against the expert focus on completeness, many books in the Sammlung Frauennachlässe contain breaks and gaps. Preprinted forms were filled differently than intended or not used at all. Lists were incomplete, records sporadic, items missing, pages empty, tables crossed out, and lines ignored or overwritten. Sometimes, notes seem made just to fill space rather than record actual expenses. In this way, these books support both economic historians’ worries that found account books were not fully trustworthy and social scientists’ concerns with expenditure surveys: the expenses that people self-reported in preprinted lists often diverged from the information gained in interviews (Stapleford Reference Stapleford2012).
Frequently bemoaned by social researchers, account books contain not only breaks and gaps but also excesses that go beyond their seemingly trite format. Books in the Sammlung Frauennachlässe abound with additional entries, marginalia, scribblings, and memos that seem, at first sight, unrelated to household budgets. Despite, or perhaps even due to, their tabular forms, account books offered opportunities for life writing. It seems that they prompted writers not only to fill the empty boxes with monetary transactions but also to tell specific kinds of (everyday) stories. The Seglitzs’ books are a case in point. The page of September 1939 begins with the note “War!!!”; the page of May 1945 includes “May 8: Cease-fire. The War is lost.” In August 1995 the script changes; a scribbling at the top of the page reads “Yesterday, my beloved Lorelein passed away.” The following month lists expenses for the funeral, “chapel and cremation,” “pastor,” and “wreath.”Footnote 56 The entries are in the same format as all the other items of expenditure. Such personal notes appear as insertions that complement and exceed the sparse form. They do not close any gaps on the page or start a longer narrative. They are, probably spontaneous, parts of a writing practice that followed specific, both given and idiosyncratically developed, lines. These notes exceed the form of account books, yet they are linked to economic life, where the boundaries to the noneconomic are precarious and hardly existent.
Compared to the requirements of quantitative studies for good research material, the Seglitzs’ account books would only partly qualify. Yet their supposed weaknesses make them promising material for understanding everyday economic practices and the relevant knowledge, which was personal, subjective, and also noneconomic. The difference between the demands of research vis-à-vis its material and the characteristics of household accounts highlights how research practices shape things, even if these things remain physically compact. The lists of expenses of a midcentury “housewife” have different functions, purpose, and relevance than the same documents in the archive or in the eyes and hands of an economist (cf. the other examples given in Hoffmann and Hagner Reference Hoffmann, Hagner, Hagner and Hoffmann2018, 7). The contrast indicates the transformative efforts necessary to turn material from everyday calculation and writing practices into research material.
Social and economic researchers find or collect household accounts and turn it into research material through deliberate action: from a specific research question to the (necessarily) selective analysis. What gets lost, namely the gaps, breaks, and excesses, is also a result of actively demarcating an “area of the economic principle and the ways of life at large,” as the national accountant Simon Kuznets (Reference Kuznets and Seligman1933: 209) put it. Like other empirical researchers involved in making data, Kuznets cautioned against the difficulties of drawing the line between the economic and the noneconomic. Investigating the practices of home economists, social reformers, and economists of consumption clarifies both the epistemic and governmental problems associated with household budgets and the attempts to control this material. Questions about how scholars’ guidelines and tools were used by others, whether they were used at all, which purposes they served in everyday life, what kinds of rationality they fostered in daily practices, and how they were beset with errors, resentment, or joy challenge historical-epistemological inquiries. Attending to the characteristics of account books and their transformation into research material opens a path to investigate whether and where a line between the economic and the noneconomic was drawn in everyday life and how this line changed when bits and pieces of this life turned into material for social and economic research.
Comment Till Düppe
It is, of course, a perfectly legitimate scholarly question to ask what exactly people consumed, in what quantities, how that changed over time, and how their “financial literacy” evolved. The availability of “high frequency” data, as we might now classify a personal housekeeping book, naturally excites researchers, as such data promise answers to questions that are otherwise approached only through empirical estimation or theoretical speculation. Today’s technologies, which track nearly every step and transaction we make, only fuel this excitement.
Treating the historical source in this way, as Verena suggests in this section, ignores the first-person perspective of Lore Seglitz and her husband actually keeping these books over an astonishing period of sixty-five years! The first question that comes to my mind when thinking of their daily writing practice, possibly at a specific time of the day, is about the sheer dedication and discipline this represents. It is, very literally, life writing. Thus, thinking of Lore Seglitz doing this makes me want to speak with her! I would want to understand the kind of intimacy she had with these books. Did she show them to others? Did she encourage others to do the same? I would ask how she got the idea. Did she undergo the education Verena mentioned? Did she see it as appropriate to the social status she represented? How did the practice evolve over time as she lived through the major events of contemporary history? I would want to better understand the personal value she attached to the books. Was her discipline related to the anxiety of one day not having enough? Or to the anxiety of one day being held accountable by someone? In short, what did it mean to her, on a personal level, to be “rational” in this very specific, enduring way? And in preparing for my conversation with her, I would of course be especially interested in what Verena described as those marginal comments about life events that fall outside the categories that she had created to structure her consumer existence.
To write down every expense, week after week, over an entire adult life is not well captured by the term “data collection.” It became part of who Lore Seglitz was. The economic knowledge that emerges from such lifelong writing inevitably carries, as all knowledge does, an existential dimension. Or one might go further, saying that it was her writing practice that made her into what she was and how she saw herself. Keeping the books would then be considered a case of what Michel Foucault called a “technology of the self” – cultural practices taken up voluntarily to shape one’s self according to an ideal and thus create one’s personae (Foucault et al. Reference Foucault, Martin, Gutman and Hutton1988). Historical epistemology opens space for these questions by pointing to the everyday practices of producing economic knowledge that Verena highlights in this section.
There’s no doubt that Lore Seglitz’s meticulous recordkeeping, whatever it meant to her from her first-person point of view, contributed to the production of economic knowledge to be exploited by economists for their own interests. Scientific creativity, you might say, is just that, finding potential evidence at places one might not expect it. Still, if you asked me, it remains a form of appropriation and assimilation.
8 Personae and Survey Data Thomas A. Stapleford
Data are the heart of empirical economics, yet that data often depend on other people, artifacts, and institutions. For economists, that means gathering data requires a different ingenuity and resourcefulness than used to build models or statistical analyses. When Robert Chapin tried to compile family expenditure data from “workingmen’s families” in New York City in 1907, for example, he quickly learned the value of a keen social awareness and adroit questioning, especially around sensitive topics:
When the item of ‘liquors’ … was touched upon, the wife, who was answering most of the questions, emphatically exclaimed, ‘Nothing.’ At this, the father, sitting silently and letting the wife do the talking, turned his head outside of the window towards the street. However, when I touched again upon the items of beer and whisky, looking straight into the father’s face, I emphasized the words ‘drinks away from home.’ Here the father could no longer restrain himself, and said, ‘About 10 to 15 glasses of beer a day and a glass of whisky’.
Chapin’s story reminds us that below economic theorizing, data analysis, or even statistical concepts lies another layer in the historical epistemology of economics: creating data in the first place (cf. Didier Reference Didier and Vari Sen2020; Gitelman Reference Gitelman2013; Schlicht et al. Reference Schlicht, Ledebur and Echterhölter2021). A key theme in that history is the challenge of control, of the need to reshape respondents, field agents, or the world itself in order to render it visible in the categories economists seek.
Consider the evolution of data collection at the US Bureau of Labor Statistics (BLS). When American state governments first established bureaus of labor statistics in the 1870s, they created the institutional foundations for a field of knowledge, which did not yet exist and which accordingly had no clearly defined features. The first thing everyone tried was mailing out questionnaires, but these, the novice statistician quickly learned, had serious weaknesses. Most would not return at all, and many others would arrive filled with inaccurate or confused responses. “It seems easy to ask questions,” declared Charles Pidgin, chief clerk at the Massachusetts bureau and author of an early statistician’s handbook, “and it seems strange that persons receiving schedules by mail do not fill them out properly and return them when requested. When applied practically, it becomes hard to ask even the simplest question and secure uniform and correct answers” (Pidgin Reference Pidgin1888: 1).
For those agencies with the resources, the obvious solution was to send a representative – the field agent – to gather information directly, and field agents quickly became the preferred method for gathering complex data, especially where respondents might be unreliable. Field agents were thus the critical link between the aspiring statistician and a world of latent economic information. In these early decades, bureaus depended on the charisma and social skills of their staff. The commissioner of the Massachusetts bureau, for example, emphasized his agency’s reliance on “tact,” “discretion,” and “moral suasion” to obtain results,Footnote 57 and Pennsylvania’s commissioner concluded that “patience, tact, intelligence, and insistency” often had “more to do with the successful collection of statistics than all other elements combined.”Footnote 58 Nowhere was this more true than when gathering information on family finances from individual households, where both reluctance to share private information and lack of accurate records combined to create endless hazards. Thus, a 1920s description of an ideal field agent for household expenditure surveys declared that “kindness and a knack at making friends, ability to state her mission in perfectly simple, impressive and appealingly argumentative language, and pleasing personality are the basic requirements of a good agent.”Footnote 59 Likewise, draft instructions for a proposed study in 1931 exhorted each agent to “use patience, perseverance, tact, and skillful and ingenious questioning” to elicit financial details, and to “determine by experimentation the best order in which to take up questions so as to avoid tiring the housewife so quickly that the schedule cannot be completed accurately.”Footnote 60 As the gendered pronouns in these descriptions implied, the BLS typically hired women for expenditure surveys in the early twentieth century, in part (one suspects) because they were thought to be more effective dealing with fellow women (housewives) and in part because the emphasis on “kindness,” “patience,” and conversational acumen resonated with twentieth-century conceptions of feminine characteristics.Footnote 61
Yet the vision of the field agent as a savvy and charismatic interrogator who used “ingenious questioning” and careful “experimentation” sat in tension with another priority: the desire of top staff to oversee the entire data collection process. That desideratum became increasingly important as academically trained social scientists gained control of government statistical agencies during the interwar period and began to treat survey methodology itself as a kind of science, subject to rigorous experimentation and testing. As such, they looked for proven ways to improve the quality and reliability of their survey projects.
Part of the new strategy involved training to overcome any deficiencies in a field agent’s personality or social expertise. Agents participating in a housing survey in the late 1940s, for example, received a three-page sheet describing “Interview Technique[s].” Some were basic social skills adapted for an effective sales pitch: “Look at respondent”; “Talk slowly”; “Don’t pause between sentences”; and “Sell, don’t order.” Agents were told to take control of the conversation and to have “a ready answer for every objection,” with the staff providing seven common objections and possible responses.Footnote 62 Likewise, the training regimen for field agents in the 1950 BLS expenditure survey included a battery of practice interviews plus a cartoon film on “Some Helpful Guides for Interviewing” (Lamale Reference Lamale1959: 63, 65). To their chagrin, however, social scientists found that training in charisma or conversational acumen had its limits. After several decades of research, two experts in the 1970s concluded that although “rapport” was universally regarded as critical to the success of any survey interview, the concept could not be rigorously defined for experimental purposes and hence could not be studied effectively (Goudy and Potter Reference 76Goudy and Potter1975).
Unable to manufacture rapport efficiently, statisticians relied on a different tactic: control as much as possible. That tendency had been evident earlier: in 1929, for example, the sociologist George Lundberg noted that “most field work requires resourcefulness, initiative, and tact.” Nevertheless, he insisted that “the most important criterion of [field agents’] desirability is that they should be willing to follow instructions explicitly” (Lundberg Reference Lundberg1929: 144–45). As the decades progressed, BLS staff developed evermore extensive practices to ensure field agents behaved as top staff envisioned, delineating the survey process in detail and creating new methods of surveillance. Whereas draft instructions to field agents from the 1920s totaled twenty-eight pages, the 1930s version ran nearly three times as long. Meanwhile, the staff created an extensive system for training and monitoring field agents, including exams, closely guided reviews of completed “schedules” (survey forms), and regular “check-interviews” in which supervisors revisited homes in order to double-check an agent’s report.Footnote 63 For the 1950 expenditure survey, basic training for field agents took a full seven days (Lamale Reference Lamale1959: 65). Indeed, following the rules had become the major desideratum: as the BLS explained, “the techniques for hiring and training the interviewers” for the 1950 expenditure survey “were designed to reduce the interviewer variability as much as possible” by ensuring that agents would follow instructions precisely (Lamale Reference Lamale1959: 29). In 2003, the rationalization of survey work reached its next natural stage with the introduction of the Computer Aided Personal Interview, a software package that permits direct data entry and guides field agents through the interview process (U.S. Bureau of Labor Statistics 2007: 3).
This emphasis on rule-following led BLS staff to envision expenditure surveys as double experiments: studies of the economy and studies of how to collect data. If the accuracy of a survey depended solely on the formal procedures and tools used to gather data (survey forms, instructions to field agents, etc.), then one could test the efficacy of different procedures experimentally. Though BLS staff had always learned from past experiences and used previous surveys as models, that viewpoint became far more formal and deliberate after social scientists took over the bureau in 1933. Thus, for example, the 1940s staff described innovative survey schedules as “experimental in character” and explained that “in a very real sense, each survey constitutes a methodological test for the one that follows it” (Brady and Williams Reference Brady and Williams1944: 333). In turn, that led to new practices. Previously, BLS staff rarely archived any survey documents beyond a few blank schedules and (occasionally) the completed forms. The 1930s brought a radical leap in archival recordkeeping, as staff preserved instruction manuals, memos to and from the field, minutes and memos from the planning and tabulation processes, a broad range of sample forms, and extensive documentation of methodological discussions.Footnote 64 Not surprisingly, BLS publications on the next major expenditure survey (1950) included a volume on methodology that offered an historical overview and detailed the current state of the art (Lamale Reference Lamale1959). Simultaneously, the bureau began running small-scale, experimental surveys so that staff could test “the phrasing of the questions, the selection of details to be recorded, and the inclusion of probes and check questions” (Brady and Williams Reference Brady and Williams1944: 332–33; Lamale Reference Lamale1959: 6, 19). In later decades, the BLS established a Statistical Methods Division to oversee research on data collection, and by the early 1990s this had been supplemented by a Collections Procedures Research Laboratory that performed “cognitive testing” of questions and survey forms on small groups of subjects (Dippo Reference Dippo1987; Jacobs and Shipp Reference Jacobs and Shipp1993: 71).
From the late nineteenth century to the early twenty-first century, the ideal BLS field agent evolved from a charismatic, “ingenious questioner” who could extract sensitive financial data to a closely monitored rule-follower, guided first by extensive instructions and eventually by a software system. That evolution was accompanied by a corollary growth in survey methodology as an experimental science, one that relied for its success on rigid repetition of plans designed by top staff. Indeed, this change allowed staff to take a new perspective on errors in basic survey data: whereas earlier statisticians had focused on making each response as accurate as possible, later BLS statisticians relied on their experiments to identify systematic biases, assuming that unsystematic errors in individual responses would cancel each other out at the aggregate level (Stapleford Reference Stapleford2012: 179–80).
Nonetheless, statisticians’ inability to codify “rapport” signaled a critical weakness: how could you ensure a recalcitrant public would cooperate with surveyors’ requests? The response rate for the BLS expenditure survey in 1972–1973 was nearly 90%; three decades later it had fallen to roughly 70%–75% (U.S. Bureau of Labor Statistics 2007, 5). Today, continued declines in response rates are contributing what some former officials characterize as a crisis in US economic data; response rates for the BLS expenditure survey, for example, are now down to nearly 40 percent. Economists may soon face a choice: find ways to reshape Americans’ willingness to participate in government surveys or, once again, reach for alternative strategies, such as leveraging “blended data” from a combination of public and private sources (Auerbach et al. Reference Auerbach, Bowen, Citro, Pierson, Potok and Seeskin2024, 59–61). In either case, constructing basic economic knowledge will require continuing to reshape the actions of noneconomists to render the world more legible.
Comment Verena Halsmayer
Tom’s piece showcases another way to reintroduce persons to the historical epistemology of economics by studying practices that contribute to economists’ knowledge but lie outside the conventional boundaries of the economics discipline. Historians of economics have long attended to how economists analyze and interpret data, as well as to the conceptual debates underlying the calculation of key statistics (e.g., Perlman Reference Perlman1987; Banzhaf Reference Banzhaf2001; Stapleford Reference Stapleford2011). The attention to practices in historical epistemology, though, also leads us to examine the nuts and bolts of how economic data are gathered, compiled, “cleaned,” tabulated, and transmitted (e.g., Stapleford Reference Stapleford2007; Didier Reference Didier and Vari Sen2020; Hirschman Reference Hirschman2021). At this level, we are often dealing with noneconomists, yet they, too, become entangled in the processes of constructing economic knowledge, alternately shaping and being shaped by the practices of economic analysis. Tracing the rationalization of survey techniques from the late nineteenth to the early twenty-first century, Tom points us to the evolving array of methodologies designed to control and shape the field agent in an effort to control and shape the respondent. The rationalization of the making of household finance data was intended to minimize the idiosyncrasies of individual agents while simultaneously cultivating new virtues, such as precision, patience, and adherence to procedure.
The shifting ideals of what makes a “good” field agent (from a socially adept, charismatic interrogator to a rule-following technician) illustrate the social and cultural embedding of empirical research: Just think of the various qualities attributed to female researchers that seemed to make them better suited for studying the household, and how these qualities were later devalued in the course of the rationalization of survey work.
At the same time, the transformation of the role of the field agent exemplifies how scientific knowledge actively intervenes in the world it seeks to represent, in shaping not only the data but also the (perceptions of the) people involved in its production (Hacking Reference Hacking1983). However, that process is not straightforward. Research work consists of a mélange of social, political, economic, cultural, and technical aspects that cannot be neatly disentangled. The interplay of these elements shapes and reshapes the very processes of data construction, making them difficult to control for individual actors. Zooming in on a specific data-gathering project might show how the research process did not go as planned, that tools and techniques failed, that field agents did not properly follow the rules, and that respondents behaved quite differently than they should. The “mangle of practice” might transform the participants themselves and run counter to the initial intentions of survey makers (Pickering Reference Pickering1993). Yet, from the longue durée perspective that Tom takes here, the productive element of economic knowledge-making takes on a more stable form: Creating new forms of visibility and legibility, the assumptions and strategies embedded in household survey practices shaped not only the data but also a broader understanding of economic life.
9 Heterodox Devotion and Personality Structure Till Düppe
In Reference Weintraub1978, E. Roy Weintraub wrote a short review in the Journal of Economic Literature about the proceedings of a conference organized by John Hicks on the microfoundations of macroeconomics. The event brought together economists from two emerging “schools of thought” in macroeconomics: the mainstream macroeconomists, who integrated the analytical tools of neoclassical economics, such as Tjalling Koopmans, Jean-Michel Grandmont, and Edmond Malinvaud, and the underdog post-Keynesian community, who considered themselves true to Keynes’s ideas, such as Paul Davidson and Geoffrey Harcourt. In his review, Weintraub questioned if the camps truly understood each other as they applied rather different research standards.
It was likely shortly after the publication of this review that Roy received the following undated note from his father, Sidney Weintraub. Sidney, who had recently launched the Journal of Post Keynesian Economics with Paul Davidson, was vital to the formation of the post-Keynesian community.
I would have returned your ‘paper’ but it would not be worth the postage. So I just tore it up. It might contaminate the place. I am old fashioned: I think patricide should be performed, if at all, privately … Remember me? Your father. Your psychiatrist must be a queer. I don’t want to hear from you again, except in the most impersonal terms. Considering your genius, you might consider changing your name to demonstrate what you can do on your own.Footnote 65
Historians of ideas might search for the reasons behind Sidney’s anger. They could point to the fact that a few years earlier, father and son had coauthored several articles, with Roy, a trained mathematician, focusing on the mathematics and Sidney focusing on the Keynesian ideas.Footnote 66 This changed when Roy began commenting on post-Keynesian ideas rather than just providing mathematical support. Roy’s review put him in competition with the community’s shared historical self-understanding and his father’s role in it. Sidney might have expected from his son a review resembling the arguments of Eichner and Kregel’s article of 1975 or also that of Harcourt’s of 1969, both equally published in the Journal of Economic Literature and crucial for building the history of the post-Keynesian diaspora.Footnote 67
This perspective might explain why Sidney was critical, but we need more to understand the intensity of his anger. If we had shown the note to any of his peers, they might not have been too surprised. They knew him as a devoted scholar who was true to his ideas. Galbraith described Sidney as a “quiet, gentle man” but an “untiring advocate of what he believed to be right” (Reference Galbraith1985: 508). His dedication to ideas was so strong that he would fiercely defend his position, even on seemingly minor points, as his friend Arthur Bloomfield observed: “Weintraub is regarded by some economists as rather aggressively controversial, argumentative, and often deliberately provocative … He feels deeply about what he believes is right, and makes no bones about saying so in vigorous terms. He is a man with a cause” (Reference Bloomfield1981: 297).
In an autobiographical essay, the virtues Sidney Weintraub himself attributed to his career were courage, loyalty, and perseverance (Reference Weintraub1985: 510). Courage meant probing novel ideas; loyalty meant pursuing Keynes’s intellectual mission; and perseverance indicated his ability to stick to his ideas despite major criticisms. He considered the fidelity to his ideas a virtue for being “without intimidation.” He saw himself as “seditious” by causing, like Keynes, upheaval among economists (Ibid.: 528). He was inclined to see resistance where other, less belligerent Keynesians might have seen friendly critiques in a shared conversation. When he did not achieve the success he expected, he would chalk it up to being strategically silenced: “In matters of philosophy and science,” he quotes Jevons, “authority has ever been the great opponent to truth. A despotic calm is usually the triumph of error” (Ibid.: 528). To his peers, his loyalty appeared as a willingness to “sacrifice” the career he could have had as a mainstream economist for the sake of truth (King Reference King1995). As his equally devoted student Paul Davidson put it: “Weintraub continually railed against the smug and complacent economics establishment who, he believed, ignored economic realities as they increased their professional prestige” (Reference Davidson1985: 539). This is what an entire community of Post-Keynesians would build their heterodox ethos upon: holding up “truth” while others stick to mere “power.” In contrast to the supposedly opportunistic attitude of technical neoclassical economists, Sidney Weintraub appeared as a courageous warrior of ideas.
But when his son Roy received the quoted note, he clearly did not read it as an expression of his father’s intellectual virtues. The note was simply shocking and deeply upsetting. Clearly, more than truth fueled his father’s doggedness. Indeed, over the years, partly in therapy, Roy came to recognize that his father’s intellectual tastes as an economist were a response to a specific personality structure marked by hypersensitivity to critique and a tendency toward black-and-white thinking. He came to understand his father’s devoted behavior as indicative of a borderline personality characterized by pronounced reactivity and alternating perceptions of others and oneself between idealization and devaluation. In a recently published interview, when discussing underrepresented questions in the history of economics, he referred to his father’s psyche:
I grew up (sigh) with a father (…) who saw the world (…) divided into ‘us’ and ‘them’. And ‘they’ would do everything possible to stop ‘us’ … In psychology, splitting by borderline personalities is associated with splitting off parts of one’s self that one cannot accept or dislikes and projecting them onto the other person. Therefore getting rid of them. I saw a great deal of that. And that was how my father conducted his academic life. It was very real and it was very ugly and it affected me because I was one of those he split from.
Roy learned to understand his father’s behavior as a reactive defense against a personal vulnerability that goes beyond the scientific sense of truth. From his perspective, the intense and clearly inappropriate fury of his father’s note is consistent with what psychologists describe as the defense mechanism of “splitting” (viewing people in extremes) and “externalization” (victimization and blaming external factors for one’s own problems). If the only alternative to idealizing himself was degrading himself, then for Sidney, accepting a position as a second-rank scholar would have been tantamount to giving up on himself. His fury was thus triggered by the fact that Roy did not unambiguously articulate in his review whether he was with “him” or “them,” leaving Sidney in the type of grey zone a borderline personality has difficulties navigating.
In other words, Sidney’s strong devotion to Keynesian ideas could be seen as a virtue by his peers but also as a defense mechanism in dealing with difficult emotions of rejection. These are not merely two independent observations. What is striking is that the reactive defense mechanisms Roy observed are consistent with the behavior that made Sidney a key figure among post-Keynesian economists. When David Colander asked: “Was Sidney upset about never being accepted into the mainstream”, rather than with an obvious “yes,” Paul Davidson’s wife responded: “I think he liked the fight, too” (in Colander and Davidson Reference Colander and Davidson2001: 97). Sidney might have sought conflict and the role of the underdog in order to create an emotional equilibrium. The heterodox identity of Post-Keynesian economics fit his personality structure.
Obviously, neither Roy nor we as historians are clinicians able to diagnose a third person, dead in this case, with a personality disorder. Should we, therefore, ignore his son’s point of view when thinking of Sidney’s role in the history of economics? Is the notion that a son constructs of his father’s personality structure an appropriate historical source? Note that we cannot insist on an actual diagnosis, since the concept of a borderline personality only emerged in the years after Sidney’s death. We might also not expect it to be a self-witness since the resistance against such a notion is part of the symptom of this personality (as it implies to accepting criticism). Furthermore, academic culture during the 1960s and 1970s was not yet ready to embrace mental health issues of professorial daily life.Footnote 68
Nevertheless, there are certainly important difficulties related to the use of family witnesses as a historical source. For one, Roy’s understanding of Sidney’s personality has been developed ex post, after the events took place. Neither Sidney nor Roy was aware of the psychological intricacies of their behavior at the time. Also, Roy’s later insight that emerged from therapy was itself part of the therapy process, as it helped him distance himself from his father’s destructive intimacy. It is not a disinterested fact. Plus, the consistency of Sidney’s personality with the post-Keynesian identity has no causal character in that it would apply to the entire community of post-Keynesian economists. The discursive identity of post-Keynesian economics might be compatible with other personality traits, and Sidney’s personality traits might be compatible with other discursive identities.Footnote 69 These difficulties certainly should prevent us from taking his memories at face value.
Even so, these difficulties should not prevent us from using the simple truth that emerges from the kind of biographical self-reflection practiced in psychotherapy: a preliminary result of an ardent effort of understanding oneself despite the odds of the false beliefs ingrained in our biographical system. Critical history and psychotherapy systematically confront one issue that is terribly underestimated in science and the institutions built upon it: The resistance against reality with which one is unable to make contact. Moreover, psychotherapy presumes all human behavior to be overdetermined to the extent that it can be explained in multiple ways (Sidney’s behavior could also be considered in terms of narcissistic traits). The truth of a notion gained in psychotherapy depends entirely on the quality of the experience it discloses in a personal life history. The point of view of family members, whether or not they understand economic ideas, is crucial for establishing this perspective, as they are part and parcel of a scholar’s struggle over self-esteem.Footnote 70 In short, we would fall short of the existential meaning of economists’ careers when excluding historical sources such as Roy’s understanding of his father’s personality structure.
Fortunately, psychotherapy is less stigmatized today than it was in the past. Professional life is seen as an aspect of personal growth, which can be hindered by mental health issues. Moreover, mental preconditions can influence professional practices, acting as a defense mechanism against underlying emotions. Personality disorders can be powerful forces in various professional settings even if they hinder personal development. Just as obsessive-compulsive behavior, for example, can promote success in professions that require narrow-mindedness, so can other behavioral patterns lead to success in specific communities. Yet, the sciences lag in understanding the importance of mental health for their profession, possibly due to a mistaken association of “smartness” with “health.” And so do the history of science and the history of economics in particular. And yet, for those interested in how real people and the power they exert function, how could one possibly avoid such issues? The belief in one set of economic ideas over another, which is the focus of the history of economics, is, apart from other things, also a matter of the different personality structures that bring about these ideas. Just as different communities require different behavioral patterns, as we highlighted in the second part, they also tend to attract different personality structures prone to those patterns. If unconscious motivations can explain career choices, they can also explain choices of disciplines, methods, or schools of thought – thus, topics in historical epistemology.
As a final remark, it is worth noting that Roy’s inner distancing from his father was the seed of the development of his scholarly profile as a historian of economics who resists being instrumentalized for the partial interests of a specific school of thought, particularly heterodox schools. In this sense, his work has been an important inspiration for what can now be labeled the historical epistemology of economics. Rather than promoting one economic idea over another, historical epistemology returns such ideas to the world where they are embodied, including, for example, into family relations.
Comment Thomas A. Stapleford
By delving into the psychology of Sidney Weintraub, Till shows how personality structure and behavioral inclinations may affect scholars’ epistemic judgments. Similarly, traditional biographies of scientists often emphasized how their personalities affected their work. The influence of the sociology of science in the 1970s and 1980s tended to undercut that approach: As the history of science became a history of communities, the effects of individual personalities drew less attention than broader and deeper forces of social relations, institutional structures, social norms, and so forth.Footnote 71 Yet Till’s contribution is not a reversion to an earlier form. First, this is not a morality play (where admirable qualities of scholars explain their success while vices lead to failure) but a more complex story in which the very traits that helped Sidney Weintraub to be celebrated in one community caused terrible damage to his family. Second, Till does not link Weintraub’s personality to specific scientific judgments but rather suggests that membership and leadership within a marginalized scientific community may be correlated with certain personality types. What we see here is a merger of biography and sociology rather than a bifurcation.
To make this argument, Till draws in part on a retrospective diagnosis of Sidney Weintraub developed by his son through his own psychotherapy. Till is not naïve about the challenges of this kind of evidence, but he rightly insists on the relevance of testimony from those who knew a person well, including family members. Indeed, the most striking moment from Till’s account is not the technical diagnosis that Roy came to accept (borderline personality), but the shocking note that Sidney sent to his son. By presenting the behavior, rather than just the professional judgment, Till allows readers to make their own assessment (what type of person would send such a note to their child?). In that respect, the details of the diagnosis become less important than Sidney’s actions themselves. The diagnosis becomes Till’s historiographic guide, in contrast to a monocausal explanation, leading him to ask questions about how such a person’s behavior might correlate with other aspects of their life.
The other unsettling aspect of Till’s account, of course, is that the very factors that led Sidney to behave viciously to his son might have been seen as virtues by his post-Keynesian colleagues. In this respect, Till’s story meshes well with a long trend of research in social epistemology insisting that individual vices can, in the right circumstances, be “virtues” for the communal pursuit of truth (Kitcher Reference Kitcher1990; Smart Reference Smart2018). Descriptively, of course, this may be true: Vices may lead to success in some situations, and what one community labels a virtue, another might consider a vice. One suspects that the case of the post-Keynesians was more like the former: few of Sidney’s colleagues may have found his note to Roy an admirable action, however, much they valued his strident attacks on Keynesian orthodoxy. Till’s narrative pushes us to grapple with the uncomfortable reality that battle lines and ethics do not always align neatly.
Acknowledgments
We are grateful to our series editors for inviting us to create this volume and guiding us through the writing and revision process. Their commitment and enthusiasm for the project have been invaluable. We also want to thank our anonymous referees, whose questions and suggestions have helped to make this a much stronger book. Verena’s research was supported in part by the Swiss National Science Foundation grant #P400PG_190987. Open access for this publication was made possible by the Franco Family Institute for Liberal Arts and the Public Good, College of Arts & Letters, University of Notre Dame. Finally, Tom thanks History of Political Economy for permission to include material from three essays previously published in that journal (Stapleford Reference Stapleford2012; Reference Stapleford2017a; Reference Stapleford2021).
Cléo Chassonnery-Zaïgouche
London School of Economics and Political Science
Cléo Chassonnery Zaigouche is Assistant Professor in the Department of Economic History at the London School of Economics and Political Science. Her main areas of research are the history of labour economics with a focus on inequalities, and the relation between economics, expertise, and policy.
Pedro Garcia Duarte
INSPER Institute of Education and Research
Pedro Garcia Duarte is Associate Professor at the INSPER Institute of Education and Research, Brazil, and is the Vice-President of the History of Economics Society (2025–2027). His research focuses on the technical transformations of economics in the twentieth century, particularly after World War II, with a special interest in the history of macroeconomics.
Catherine Herfeld
Leibniz University
Catherine Herfeld is Professor of Philosophy and History of Economics at Leibniz University Hannover, Germany. Her research focuses on the history and philosophy of scientific modeling, economic methodology, the history of rational choice theory, the transfer of models within and beyond economics, and questions concerning scientific progress in economics.
Harro Maas
University of Lausanne
Harro Maas is Professor Emeritus in History and Philosophy of Economics at the Centre Walras-Pareto for the History of Economics and Political Thought at the University of Lausanne. He publishes widely on the history of social quantification, visualization of the economy, and on the history of economic methods. He is a honorary member of the European Society of the History of Economic Thought.
About the Series
Over the past years, the history of economics has attracted a still growing audience of economists, historians, sociologists, and philosophers. This series targets this audience with learned essays on the state of existing scholarship and highlights new pathways of research, in particular for students at Master and PhD levels.
