Navigation – Plan du site

AccueilNuméros13-3The Computerization of Economics....The Computerization of Economics:...

The Computerization of Economics. Computers, Programming, and the Internet in the History of Economics

The Computerization of Economics: Three Lessons for Economics

Introduction to the Special Issue
L’informatisation des sciences économiques : trois leçons pour les sciences économiques
Marcel Boumans, Cléo Chassonnery-Zaïgouche, Pierrick Dechaux et Francesco Sergi
p. 637-655

Texte intégral

1 In 1994, the Journal of Economic Perspectives published an article by William Goffe titled “Computer Network Resources for Economists”, starting with the statement: “[t]he Internet, the large and very rapidly expanding computer network, is revolutionizing research” (Goffe, 1994, 97). Goffe’s article aimed at providing the first “guide” for economists to understand the principles and mechanisms behind the functioning of the Internet, and it showcased a series of possible uses of this new technology for economic research, teaching, and professional networking. The article mirrored Goffe’s own personal effort to disseminate online resources for economists, and it paralleled a more collective endeavor that led, eventually, to the establishment of REPEC and of Federal Reserve Economic Data (FRED).1 Thirty years later, the Journal of Economic Literature hosts an article titled “Generative AI for Economic Research: Use Cases and Implications for Economists” (Korinek, forthcoming), emphasizing again how a new technology—this time, artificial intelligence tools like ChatGPT—“has the potential to revolutionize research”. The article aims at illustrating possible applications of AI that are “useful for economist”, for instance in producing “significant productivity gains” for research (ibid., 1).

  • 2 The more general view of computers as “institutions” rather than technological devices is well know (...)

2 Both episodes illustrate the past and current optimistic engagement of the economic profession with new computer technologies. This special issue provides the reader with a historical reflection about such engagement, documenting several episodes about the way computerization changed economics. The articles gathered in this special issue document the opportunities, constraints, and challenges that the dissemination of computers (computerization) brought (and continues to bring) to economics. In this introduction, we use the term computer as a denominator to capture not only the “machine” but also to include computer programs, software, computer languages, computer communication technologies, etc. Such a history questions the optimistic, somehow naive view of computerization enabling the “progress” of economics in a universal and unidirectional way. By contrast, this special issue considers failures, U-turns, and struggles in the way economists engaged with computers. This focus enlightens how the adoption of new technologies (whether hardware, software, or combinations of new data and of new computational methods) was dependent on intellectual and practical determinants within the social context of the practice of economics. When we consider these social dimensions of computers and analyze computers as institutions, a different “image” of economists’ relation with computers emerges, one that could be useful to reflect on the current evolution of economists’ practices.2

3“Computerization” designates the historical and social process resulting in a change of practice, not only in terms of the use of a new technology, but also a change in the character of economic research itself. Not only did economists begin to use computers for conducting their activities, computers themselves became a crucial, distinctive, and necessary element of economists’ practices, associated with specific fields of inquiry and skills. Each article of this special issue focuses on a historical case study of this process, overall documenting the computerization of economics for a variety of contexts: different sub-disciplines of economics (from experimental economics to macroeconomics), different periods and countries, different types of hardware (from analogue mechanical computers to electronical mainframe computer and personal computers) and different computer-related technologies (from programs and software to time-sharing and networking technologies). These case studies investigate both the “ordinary” (the daily routines of the “average” economist) and the “extraordinary” (such as the creation of new computer technology and the invention of new practices).

4This introduction points to the commonalities across the case studies, suggesting three general lessons that economists and historians of economics can draw from the study of the computerization of economics. Firstly, computerization reshapes everything that economists do: it reshapes both the disciplinary boundaries of economics (research questions, theoretical and modeling strategies, the methodologies for empirical work, etc.) and the structure and role of the economic profession (the forms of expertise, the division of labor within the profession, its relation to private corporations). Secondly, computerization is not about pre-made technology entering the field of economics—or, in other words, computerization does not constitute some “exogenous technical progress” or some deus ex machina; quite the opposite, computerization is an active, endogenous area of practices within economics, insofar as economists attempt—and sometimes, although not always, succeed in—reshaping computer technology for their own purposes. Finally, documenting computerization requires reconsidering historiography, especially in terms of sources and perspectives, gearing historians towards a more careful examination of economics as a practice. This leads to a renewed image of what (and who) matters in the evolution of economics.

1. Reshaping the Discipline

5Computers are not mere “instruments” for economists to import, that is, they are not simple means to a given, pre-established end, that economists would passively “adopt”. Economists actively shape the computerization of economics. Only this “appropriation” makes computerization possible and, sometimes, successful; this process is also what makes the historical study of computerization interesting and relevant.

  • 3 Morgan (2012) discusses these ontological implications with regard to modeling. Stapleford (2017) d (...)

6The view of computers as instruments is quite widespread. It usually emphasizes the technological aspect of the change introduced by the arrival and the development of computers. According to this view, computerization allowed economists to perform a larger set of computational operations (manipulating data, estimating models, solving models, simulating models, etc.) faster, cheaper, and to a larger scale (more data, more complex models, etc.). The contributions to this special issue show that computers actually changed the practice of economic research more profoundly: they changed economic methodology—that is to say, they transformed economic research—and also economic epistemology—in particular, new phenomena came to be studied. Moreover, computers had ontological implications, that is, implications for the way economists think about their materials and subject matters and so the objects that they think exist in the world.3

7Practical consequences of computerization appear trivial, plain to see to anyone wandering around an economics department today. In the daily life of economists, the main distinctive activities are conducted through computers—whilst they were not a few decades ago: collecting, storing, and analyzing empirical data; building mathematical models; reading and writing research articles, reports, and preparing lectures and communications to peers and students alike. Taking a longer view, one would notice, for instance, how experimental economists went from “conducting pen-and-paper experiments in ad hoc locations” to “purpose-built” laboratories, in which computers are a distinctive, essential feature of the experimental activity (Andrej Svorenčík, this issue). Similarly, in the early years of econometrics, calculations were carried out by humans (“human computers” or “computors”), with the help of small electronic calculators; from the 1950s, analog and then electronic computers transformed the practice of econometrics (Chung-Tang Cheng, this issue).

8The contributions to this special issue highlight that computerization brings more substantial practical changes. One key example of such changes concerns the specific practice of crafting codes, packages, software, online platforms, and datasets. In most cases analyzed, these tasks are undertaken by some economists specializing in designing these new computer technologies. Although these activities provide key contributions to the profession as a whole, it has not always been acknowledged as such. Programming deserves a more careful study by historians, as an integral part of the practice of economics, on equal footing with formalizing mathematical models, performing statistical or econometric analysis, or conducting experiments.

  • 4 A somehow related, but distinct literature has focused more precisely on the trans-disciplinary rel (...)

9The consequences of computerization encompass changes in research topics, in theoretical approaches, in the methodology and in the epistemology of economics. The existing literature in the history of economics engaging with computerization had already investigated some of these transformations. In Machine Dreams. Economics Becomes a Cyborg Science, Philip Mirowski (2011) argued that the development of computers (and of computers science) provided a new intellectual blueprint for economics’ representation and understanding of markets.4 In a later book with Edward Nik-Khah (Mirowski and Nik-Khah, 2017), this observation was extended to the appraisal of the transformations of microeconomics and specifically of the development of new fields, such as information economics or market design. Roger Backhouse and Beatrice Cherrier (2017) provided an overview of the crucial role played by computers in what they call the “applied turn” in economics. Here, computers were notably crucial for redefining the boundaries between theoretical and “applied” work. Another stream of work highlighted how new methodologies and new epistemologies (especially new forms of proofs) emerged from the use of computers—for example in the case of computer simulations in various contexts (e.g., Fontana, 2006) or the use of multiple regression analyses in courtrooms and policy (e.g., Chassonnery-Zaïgouche, 2020).

10Compared to this existing literature, the contributions to this special issue highlight somehow different practical and intellectual effects of computerization. We suggest that these can be grouped into three different topics: the effect of computerization in reshaping the role of “computation” in economics (1.1); the effect of computerization on the division of labor within the economic profession, and the reallocation of credit and prestige to certain activities (1.2); and, finally, how computerization moves the boundaries between academia, policy-making, and private business (1.3).

1.1 Redefining Computation

  • 5 Between 1988 and 1992, the journal was actually published under the title Computer Science in Econo (...)
  • 6 The development of the field of computational economics provides a promising case for further resea (...)

11In 1988, the first issue of Computational Economics hailed the birth of “a new fully fledged discipline” in economics and management, soon to be nurtured by a new society.5 This new discipline was later defined by its founders as a field providing “a new methodology for solving economic problems with the help of computing machinery” (Amman, 1997, 103). The creation of this new field is already a telling example of the way computerization modifies intellectual boundaries of economics.6 Moreover, in its “Editorial Foreword” to the first issue of Computational Economics, Hans Amman noted:

In recent years … there has been a tremendous proliferation of computers inside and outside the academic world. One of the reasons is the introduction of relatively cheap microcomputer systems and due to a greater availability of these computers, the possibilities for economic and management research have expanded considerably. Computers are no longer just an aid for research, but are now making it possible to push the boundaries of science further forward. (Amman, 1988, 1, our emphasis)

  • 7 Lejeune (2021) has documented a similar process in the computerization of the field of medieval his (...)
  • 8 On human computers and mechanical calculators, see Light (1999) and, for economics, Cheng’s contrib (...)

12The case of computational economics and the remarks by Amman point to the first fundamental intellectual transformation brought by computerization: the redefinition of the role of computation (or “calculation”)—that is, the various ways of manipulating informational content about economic systems, either in the form of statistical information (“data”) or in the form of mathematical specifications of the behavior of economic agents.7 In this respect, computerization changed the technological scope, i.e., the possibility of manipulating larger informational sets, faster and cheaper. However, if this was the whole point of the story, then computerization would just entail more efficient “research aid” provided by computers to economists—more efficient than, say, mechanical calculators operated by human computers.8 As noted by Amman, there is more than this at stake with computerization: shaping new types of computations and adding research questions. However, these changes happened way before the 1990s and the emergence of a specific field (“computational economics”).

13Since the 1960s, the arrival of mainframe computers granted a more preeminent and central role to computation in economics, especially under the specific form of handling statistical information about the economy—as documented notably by three contributions to the special issue: Laetitia Lenel’s work on the early days of forecasting at the National Bureau for Economic Research (NBER) in the US, Cheng’s analysis of the early days of econometrics at the Department of Applied Economics at the University of Cambridge, and Pedro Duarte and Francesco Sergi’s history of the rise of macroeconometric modeling at Data Resources Inc. (DRI), a private forecasting firm based in Lexington, Massachusetts. In all these three cases, the collective practice of economics crucially relied on new ways of manipulating statistical information: new ways of collecting it, storing it, analyzing it, mobilizing it for different purposes. These three different cases reveal how, since their arrival, computers have been taking center-stage, and not an ancillary or “aid to research” position: it is, precisely, the computerization that reshaped the ways of handling statistical information. This historical process is still ongoing, as highlighted by two other contributors to the special issue: Julien Gradoz, in his study of the use of scanner data for calculating price indexes, emphasizes how the current state of computers still raises challenges about the issues of handling large informational sets (the so-called “big data”); and Nik-Khah, in his study of the role of “tech economists” within digital platform firms, stresses how this specific context brings new opportunities for collecting and using data about market behavior.

14A second way in which computerization changed computation in economics is by establishing new ways of manipulating informational content about economic systems. This change is key in two other contributions of the special issue. Cherrier, Aurélien Saïdi, and Sergi study “Dynare”, a set of programming routines that allows macroeconomists to simply handle the computation of the equilibrium and dynamics of large scale dynamic stochastic general equilibrium models. Finally, in Svorenčík’s article, one learns how the introduction of computers and new software allowed experimental economists to modify the way information circulates across the participants to an experiment, fundamentally changing the nature of the experiments conducted by economists.

1.2 New Divisions of Labor and “Hidden Figures”

15Computerization changed the division of labor within the profession, and eventually the allocation of prestige and recognition. As emphasized by Cheng’s article about econometrics, this change is particularly relevant during the “transition” periods between different technological configurations—different types of computers, or software, and their embeddedness with economists’ activities. A generic line that could be drawn is between the early decades of the history of computers, dominated by mainframe computers, and the more recent decades, characterized by the spread of personal computers.

  • 9 The expression “hidden figures” comes from Margot Lee Shetterly (2016)’s book on African American f (...)

16The early period entailed a clear division of labor between economists and other specialized professional figures (such as computer scientists, electrical engineers, and mathematicians) that operated computers and designed computer programs. In this period, economists usually stood at a distance from computers, physically and intellectually, delegating this part of the job to others. With some exceptions though: a few economists engaged with these new computer tools, especially by developing programming skills, effectively constituting a specific professional profile—one that we could call “programming economists”. Documenting the computerization of economics brings thus to uncover both “programming economists” and several other professionals, such as human computers, computer operators, and software developers. These are “hidden figures”, whose contributions to economics has been forgotten or rarely recognized.9 By focusing on these figures we can improve our understanding of research practices in economics (see e.g., Boumans and Duarte, 2019).

17The various contributions to the special issue present histories animated by such hidden figures. Cheng, for instance, depicts vividly the key role of programmer Lucy Joan Slater (a mathematician with no background in economics), in developing the econometric research program of Cambridge Department of Applied Economics. Slater’s contribution was well-known and recognized by her colleagues and contemporaries; she was praised by Richard Stone as the “dea ex machina” (goddess behind the machine), who “ought to have had a share in the [Nobel] prize” (Cheng, this issue). Duarte and Sergi, in their contribution, document in quite some detail the daily work done by computer operators at DRI, based on the recollections of DRI employees and customers. Duarte and Sergi’s article also highlights the role of Harvard economist Otto Eckstein, the founder of DRI, and the first macroeconomist who became a millionaire thanks to macroeconomics. As they argue, Eckstein’s contribution to macroeconomics, so far pretty much neglected, was key to the evolution of the discipline, insofar as it marked the dissemination of large-scale macroeconometric modeling expertise beyond academia. The case of DRI also provides some elements about a local group of “programming economists” (Robert Hall, Edwin Kuh, Phillip Cooper, among others), who had been instrumental in developing new programming tools for econometrics and macroeconomics.

18 The dissemination of personal computers since the mid-1980s brought a new era, in which the operation of computers seemingly no longer required the intervention of other professional figures: economists can use the computer directly, on their own, to perform their own desired tasks. However, this change actually does not correspond to the disappearance of “programming economists”: there are, still, a few economists specializing in developing new computer technology. The figures revolving around the Dynare team (Laffargue, Juillard, and others), as highlighted in Cherrier, Saïdi, and Sergi’s article, are a case in point in the domain of macroeconomics.

1.3 Shifting Academic Boundaries

  • 10 At this time, the computer industry often spurred from the pre-existent mechanical calculator indus (...)
  • 11 See Renfro (2011) for the development of econometric software.

19Ownership and property rights constitute a crucial aspect in the general history of computers. Although deeply embedded with government initiatives and government procurements, the development of the electronic computer in the immediate post-war has rapidly resulted in the constitution of a large private industry;10 the rise of the personal computers in the 1980s has amplified this phenomenon, also leading to a greater international dissemination of the computer. By contrast, computer codes and software have been, from the start, a battlefield for ownership rights, between the computer industry and a growing community of users developing software.11 Ultimately, this historical process resulted in the current dichotomy between the free software movements and the business of proprietary software.

20Consequently, financial costs related to the ownership and use of computers have played a significant role in determining the paths of the computerization of economics. More specifically, as emphasized by several contributions to this special issue, computerization has set novel paths in the boundaries between academia, policymaking institutions, and private business. In this issue, Duarte and Sergi, Nik-Khah, Gradoz, and Cherrier, Saïdi, and Sergi show that economists have built new relationships across these spheres in order to support their use of computers. In the case of the development of large-scale macroeconometric models of the 1960s and 1970s, macroeconomists have turned this practice, initially developed within academia (and with support from some policymaking institutions, like the Federal Reserve Board), into a flourishing consulting business (Duarte and Sergi, this issue). This allowed them to acquire powerful computers, to establish large databases, and to develop software required to popularize this approach. In the 1990s, macroeconomists seem to have followed a different path (as documented by Cherrier, Saïdi, and Sergi, in this issue). Teaming-up with policymaking institutions (French economic planning, the International Monetary Fund and, later on, central banks), they were able to develop and maintain a new computer-tool (Dynare), which became essential to the simulation and estimation of large scale dynamic stochastic general equilibrium models. However, this time, they did not turn this development into a business, but made their computer package freely available to everyone. Dynare’s case is therefore one clearly siding with the free software movement. More recently, in microeconomics, and especially in market design (Nik-Khah, this issue), academic economists have been taking up key positions in tech firms, where they are able to devise computer-tools for testing theories and collecting data. Finally, statistical institutes (Gradoz, this issue) have been reluctant to exploit scanner-data for building price-indexes, for two reasons: on the one hand, these data retain a key economic and strategic value for private business, which requires for public statistical official to enter difficult negotiations; on the other hand, the computer programs that allow to exploit these data are mostly owned by private tech firms, such as Google, which requires to enter into collaboration spurring financial and ethical issues.

2. Technology as a Source of “Progress”?

21Several contributions to the special issue highlight the role of computers in the success or failure of some methodologies, modeling approaches, and research questions. The issue of success is well-known, since narratives from economists quite often emphasize that there is “progress” in economics, and it comes from technological developments. One prominent advocate of this view is Robert Lucas:

  • 12 Lucas did not change his mind about progress in economics. In his address to the History of Economi (...)

One would expect developments to arise from two quite different kinds of forces outside the subdisciplines of monetary economics or business cycle theory. Of these forces the most important, I believe, in this area and in economics generally, consists of purely technical developments that enlarge our abilities to construct analogue economies. Here I would include both improvements in mathematical methods and improvements in computational capacity. (Lucas, 1980, 697)12

22 The idea of progress (that is, some cumulative upgrade of economic knowledge) is obviously a contentious idea per se (see e.g., Boehm et al., 2002). Thus, the easy association between computerization and the “progress” of economics should be treated with caution. However, one may agree, more prudently, that computerization has brought changes in economics and that, indeed, some of these changes entailed, for instance, expanding research fields, building new types of models and extending their range of application, and applying new econometric techniques. We could label such fruitful collaborations between computers and economists “progress” or “computer boosts”. In this special issue, several of these examples are documented in more depth and with precision—making explicit why exactly a given computer technology or a given computer tool provided “progress”. That is, what kind of issues, constraints, or difficulty (theoretical, computational, or econometric) economists try to address through computers.

  • 13 For a general analysis of the role of “tractability” in the history of economics, see Cherrier (202 (...)

23In their contribution to this special issue, Cherrier, Saïdi and Sergi analyze, for instance, the case of rational expectations. Since the 1970s, the introduction of rational expectations in non-linear, large-scale macroeconomic models had raised a specific problem of recursiveness in solving and simulating this class of models within a given constraint in terms of computer power and cost—a specific case of a more general problem that the authors name the “tractability” of models.13 The article documents how several computer algorithms were designed to address this problem, focusing on the origins of one specific algorithm, designed by French macroeconomist Jean-Pierre Laffargue. Laffargue’s algorithm, together with its implementation throughout a “parser” (designed by Michel Juillard), became instrumental in solving, simulating, and estimating dynamic stochastic general equilibrium (DSGE) models in macroeconomics, thus contributing crucially to the successful dissemination of this specific kind of modeling approach in macroeconomics.

24In his contribution, Svorenčík documents how the arrival of computers and, more specifically, of computer networks in experimental economics’ laboratories, made possible to design new type of experiments—in particular more sophisticated auctions. Network-connected computers allowed to circulate instantaneously information across the participants (the experimenters and the experimental subjects alike), which was otherwise difficult or impossible with hand-run experiments. This provided an exceptional “computer boost” both to experimental economics as a methodology, and to market design as a research subject.

  • 14 Again, the standard narrative about macroeconomics provides a neat illustration of this idea: whils (...)

25However, a “computer boost” for one approach can equally mean a “computer bust” for others—that is, computers contribute to the “failure” of a given approach. Although this seems a symmetric question to the one about “computer boost”, “computer busts” rarely play any part in narratives of economists.14 In this special issue, it appears although more evident that “computer busts” do play a role, to a different extent, in the fate of some approaches. “Computer busts” should not be understood as related to “inadequate” computers or programmers, or “missed opportunities” by economists. Most often, “computer busts” are related to deeper divide between the computer and the type of research questions, methodology, or modeling technique.

26Lenel, in her contribution, documents how the NBER abandoned its forecasting efforts following the arrival of electronic computers in the 1950s. She highlights how electronic computers, bringing “more speed” for computation, were initially envisioned as a solution to problem of “timeliness” in the collection and handling of data for nurturing NBER economic forecast. And yet, these computers in fact slowed down and complicated forecasting routines—for instance, because of the time needed to perform programming and the scarcity of computer time—eventually contributing to putting an end to the NBER forecasting activities. Duarte and Sergi’s article recounts the fall of DRI, a private firm that provided macroeconomic and econometric consulting services to government agencies and private corporations. The authors document how DRI built its “hype”, in the 1970s and in the 1980s, combining large-scale macroeconometric modeling practices and a top-notch mainframe computer infrastructure, the time-sharing technology, and the production of software tailored for their customers’ needs. The dissemination of personal computers and, years later, the advent of the internet, reduced the popularity of this type of economic expertise. Finally, Gradoz’s article addresses a more contemporary issue, namely the use of scanner data by national statistical offices. This type of data could be thought of as providing the ideal base for building cost-of-living price indices, a type of price indices that entails a number of qualities and advantages over traditional cost-of-goods indices. However, as documented by Gradoz, scanner data raise a number of challenges (in terms of size and computational cost, in terms of collection, and in terms of treatment) that make it rather difficult to build cost-of-living price indices.

  • 15 A similar perspective has been taken in the history of statistics, for instance by Alain Desrosière (...)

27Computerization has thus, as one major intellectual consequence, to “boost and bust” research questions, methodologies, and modeling approaches. However, it is erroneous to consider this as a “fatality” (“It’s the computer, stupid!”, as summarized by Backhouse and Cherrier, 2017). It seems similarly erroneous to think about this dynamic as one of “progress” (or “non-progress”), as suggested by Lucas, since it reduces the evolution of knowledge to the evolution of technologies. To the contrary, a view of computers as institutions (putting at the center of the analysis their “embeddedness” with cognitive, social, financial, and even moral aspects) provides a more comprehensive and clear explanation of their ability to introduce change in economics.15 In short, computers are not dei ex machina, descending upon economists to “solve” their problems: quite the contrary, the ability of computers to “boost or bust” is embedded in the way economists “appropriate” them: that is, their ability to shape computers to purposefully serve their aims.

3. The Computerization of Economics as an Historiographical Challenge

28Given the key role of computerization in shaping the evolution of economics, it seems somehow paradoxical that this phenomenon had not yet been granted much attention by historians of economics. Although virtually all historians of economics interested in the second half of the twentieth century have already noticed the role of computerization, these mentions are often marginal remarks, footnote material, rather than central to the analysis.

  • 16 This is the case for all issues considered as “technical”, such as difficulties in solving mathemat (...)

29This marginality reflects how, in general, economists have covered computer-related issues in their publications: often incidentally and implicitly—by referring, with no further details or explanations, to how “difficult”, “easy”, “time-consuming”, or “costly” this or this computer operation has been, or how “precious” was the programming or computing work done by this or that research assistant. In short, constraints and opportunities created by computers are not explicitly discussed by economists: they are, most of the time, confined to the footnotes of economic articles and books.16

  • 17 When a new computer technique becomes “fashionable” within a field (like Dynare and the Newton-Raph (...)
  • 18 The existence of specialized outlets, like the Journal of Computational Economics, which provide co (...)

30One reason for such a footnoting is that the computer aspect of economists’ practices is indeed not easily communicable to other economists (and, eventually, to a wider audience), especially when technologies are new. Therefore, exposing extensively computer-related aspects of one’s research would be somehow detrimental to the communication of its results, at least as long as computer-related aspects are perceived by the audience as “geeky”.17 Thus, the absence of such careful discussions does not mirror the actual importance of computers in the production of research results.18 This implies that to unveil and understand this importance, historians need to rely on other sources than just published articles.

31Inquiring about what exactly economists do with computers requires turning towards some distinctive written sources. Before the 1990s, computers are mostly mainframe computers, that is, large infrastructures used collectively and owned by large private corporations, universities, or public institutions. Thus, precious information about their operation can be found in these organizations’ archives, or in the personal archives of individuals that were high placed within the organization (as done, for instance, in Lenel, Svorenčík, and Duarte and Sergi, this issue). The advent of the internet, in the mid-1990s, provides historians with another type of distinctive written sources to investigate the history of computerization: personal and institutional web pages, online repositories, and even online forum and mailing lists archives can help documenting the use of computers and the dissemination of software (as done notably by Cherrier, Saïdi, and Sergi, this issue).

32However, these written sources need to be complemented for several reasons. First, the practice of computers does not always leave written traces (records can, for instance, be “deleted” as they are disregarded). Second, historians of economics might not have the necessary technical knowledge (and, most often, they have no direct experience) to properly understand the functioning of past computer technologies and programming languages. Finally, several aspects of computer practices, especially those pertaining to programming, are most often not put in writing. For these reasons, oral history proves to be unvaluable for writing the history of computerization. As displayed by several contributions to this special issue (notably Svorenčík, and Duarte and Sergi), interviews give access to the context of computerization, enlightening the questions, problems, routines faced by those engaging with computers, as well as clarifying the decisions and choices they made.

33Even if they are not the first to see the need for a reconsideration of historiography because of the introduction of new technology (see Davis, 1997 for an earlier attempt)—nor for a reconsideration of the historiography of economics per se (see e.g., Stapleford, 2017 and Düppe and Weintraub, 2019)—the contributions to this special issue provide solid examples about the relevance of such a reconsideration. They constitute, in this respect, a “twin” contribution to the previous special issue published by Œconomia this year, showcasing how quantitative methods can be fruitfully applied to the social studies of economics (Goutsmedt et al., 2023).

34Changes in historiographical perspective allow thus to unveil the mechanisms behind the production of economic knowledge, shedding new light onto the evolution of economic thought. This special issue draws attention, precisely, to computerization as a driver of this evolution, while putting a strong emphasis on the “endogenous” nature of computerization—that is, on economists’ practices reshaping and adapting computers to the field of economics.

We would like to thank the editor-in-chief of Œconomia, Jean-Sébastien Lenfant, who trusted and supported this project. We owe a great debt of gratitude, for encouraging us at an early stage of this project, to the participants to the October 2021 REhPERE workshop held at University Paris 1. The special issue has benefited enormously of the workshop, gathering all the authors, which was held at the Université Paris Est Créteil in May 2022. We gratefully acknowledge the financial support received to organize this event, both from the Laboratoire interdisciplinaire d’études du politique Hannah Arendt (LIPHA) and from the History of Economics Society (HES) New Initiatives Fund. We would like to address a special thanks to Hamida Berrahal, from LIPHA, for her contribution to the organization of the workshop.

Haut de page

Bibliographie

Amman, Hans M. 1988. Computer Science in Economics and Management: A New Discipline! Computer Science in Economics and Management, 1(1): 1.

Amman, Hans M. 1992. A Message from the Managing Editor. Computer Science in Economics and Management, 6(1): iii.

Amman, Hans M. 1997. What Is Computational Economics? Computational Economics, 10: 103-105.

Backhouse, Roger E. and Beatrice Cherrier. 2017. “It’s Computers, Stupid!” The Spread of Computers and the Changing Roles of Theoretical and Applied Economics. History of Political Economy, 49(Suppl.): 103-126.

Boehm, Stephan, Christian Gehrke, Heinz D. Kurz, and Richard Sturn (eds). 2002. Is There Progress in Economics? Knowledge, Truth and the History of Economic Thought. Cheltenham: Edward Elgar.

Boumans, Marcel, and Pedro Garcia Duarte, 2019. The History of Macroeconometric Modeling: An Introduction. History of Political Economy, 51(3): 391-400.

Chassonnery-Zaïgouche, Cléo. 2020. How Economists Entered the ‘Numbers Game”: Measuring Discrimination in the US Courtrooms, 1971-1989. Journal of the History of Economic Thought, 42(2): 229-259.

Cherrier, Beatrice. 2023. The Price of Virtue: Some Hypotheses on How Tractability Has Shaped Economic Models. Œconomia. History/Methodology/Philosophy, 13(1): 23-48.

Davis, John B. (ed.). 1997. New Economics and Its History. Annual supplement to History of Political Economy, volume 29. Durham: Duke University Press.

Desrosières, Alain. 2008. Pour une sociologie historique de la quantification. L’argument statistique. Paris: Presses de l’Ecole des mines.

Düppe, Till, and Ivan Boldyrev (eds). 2019. Economic Knowledge in Socialism, 1945-89. Annual Supplement to History of Political Economy, volume 51. Durham: Duke University Press.

Düppe, Till, and E. Roy Weintraub 2019. A Contemporary Historiography of Economics. London: Routlegde.

Fernandez-Villaverde, Jésus. 2010. The Econometrics of DSGE Models. SERIEs, 1(1-2): 3-49.

Fontana, Magda. 2006. Computer Simulations, Mathematics and Economics. International Review of Economics, 53(1): 96-123.

Goffe, William L. 1994. Computer Network Resources for Economists. Journal of Economic Perspectives, 8(3): 97-119.

Goutsmedt, Aurélien, François Claveau, and Catherine Herfeld. Quantitative and Computational Approaches in the Social Studies of Economics. Œconomia. History/Methodology/Philosophy, 13(2): 146-162.

Hicks, Mar. 2017. Programmed Inequality. How Britain Discarded Women Technologists and Lost Its Edge in Computing. Cambridge: MIT Press.

Iacono, Suzanne and Rob Kling. 1988. Computer Systems as Institutions: Social Dimensions of Computing in Organizations. International Conference on Information Systems (ICS) Proceedings, 32: 100-110.

Korinek, Anton. Forthcoming. Generative AI for Economic Research: Use Cases and Implications for Economists. Journal of Economic Literature. https://www.aeaweb.org/articles/pdf/doi/10.1257/jel.20231736 [retrieved 23/10/2023].

Lejeune, Edgar. 2021. Médiévistes et ordinateurs. Organisations collectives, pratiques des sources et conséquences historiographiques. PhD Dissertation, Université Paris Cité.

Light, Jennifer S. 1999. When Computers Were Women. Technology and Culture, 40(3): 455-483.

Lucas, Robert E. 1980. Methods and Problems in Business Cycle Theory. Journal of Money, Credit, and Banking, 12(4): 696-715.

Lucas, Robert E. 2004. Keynote Address to the 2003 HOPE Conference: My Keynesian Education. History of Political Economy, 36(Suppl.): 12-24.

Mirowski, Philip. 2011. Machine Dreams. Economics Becomes a Cyborg Science. Cambridge: Cambridge University Press.

Mirowski, Philip and Edward Nik-Khah. 2017. The Knowledge We Have Lost in Information: The History of Information in Modern Economics. Oxford: Oxford University Press.

Morgan, Mary S. 2012. The World in the Model: How Economists Work and Think. Cambridge: Cambridge University Press.

Renfro, Charles G. 2011. Econometrics and the Computer: Lover or a Marriage of Convenience? History of Political Economy, 43(Suppl.): 86-105.

Sergi, Francesco. 2020. The Standard Narrative about DSGE Models in Central Banks’ Technical Reports. The European Journal of the History of Economic Thought, 27(2): 163-193.

Shetterly, Margot Lee. 2016. Hidden Figures. The Untold Story of the African American Women Who Helped Win the Space Race. London: William Collins.

Stapleford, Thomas. 2017. Historical Epistemology and the History of Economics: Views Through the Lens of Practice. Research in the History of Economic Thought and Methodology, 35A: 113-145.

Tulbure, Narcis. 2020. Data Poverty in a Thoroughly Planned Economy: Statistics and Data in Socialist Romania. Studii şi materiale de istorie contemporană, 19(19): 149-181.

Haut de page

Notes

1 See https://ideas.repec.org/history.html [retrieved 09/11/2023].

2 The more general view of computers as “institutions” rather than technological devices is well known in the history of computing and information technologies (see Iacono and Kling, 1988 for a review). This special issue brings this perspective into the history of economics.

3 Morgan (2012) discusses these ontological implications with regard to modeling. Stapleford (2017) discusses them in terms of “phenomenotechnique,” a term coined by Gaston Bachelard: Scientific practice “evinced a dialectical relationship in which engagement with empirical phenomena transformed the ideas of scientists, who, in turn, reified their theories in new instruments and techniques, using them to create new phenomena.” (Stapleford, 2017, 132)

4 A somehow related, but distinct literature has focused more precisely on the trans-disciplinary relation between economics and computer science, highlighting, for instance, the emergence of the field of “economic cybernetic” in the context of economic planning in Socialist economies (see for example Düppe and Boldyrev, 2019; Tulbure, 2020).

5 Between 1988 and 1992, the journal was actually published under the title Computer Science in Economics and Management, before the field “matured under the name ‘Computational Economics’.” (Amman, 1992, iii)

6 The development of the field of computational economics provides a promising case for further research, since none of the contributions to this special issue engages with it (although one can find some remarks about it in Cherrier, Saïdi, and Sergi, this issue).

7 Lejeune (2021) has documented a similar process in the computerization of the field of medieval history. Lejeune shows how computers introduced new ways to handle documents (sources), there transforming deeply medieval historiography and the research practices of historians.

8 On human computers and mechanical calculators, see Light (1999) and, for economics, Cheng’s contribution to this issue.

9 The expression “hidden figures” comes from Margot Lee Shetterly (2016)’s book on African American female mathematicians working for the NASA space programme. On the gender aspect in the history of computing, see Light (1999) and Hicks (2017).

10 At this time, the computer industry often spurred from the pre-existent mechanical calculator industry, as for instance in the case of IBM in the US or Bull in France.

11 See Renfro (2011) for the development of econometric software.

12 Lucas did not change his mind about progress in economics. In his address to the History of Economics Society, Lucas (2004, 22) reiterated the same message: “I see ... the progressive element in economics as entirely technical: better mathematics, better mathematical formulation, better data, better data-processing methods, better statistical methods, better computational methods.” Another example is the use of dynamic stochastic general equilibrium models in macroeconomics. The standard narrative about the success of this approach (Sergi, 2020) goes along these lines:

No matter how sound were the DSGE models presented by the literature or how compelling the arguments for Bayesian inference, the whole research program would not have taken off without the appearance of the right set of tools that made the practical implementation of the estimation of DSGE models feasible in a standard desktop computer (Fernandez-Villaverde 2010, 13).

13 For a general analysis of the role of “tractability” in the history of economics, see Cherrier (2023).

14 Again, the standard narrative about macroeconomics provides a neat illustration of this idea: whilst computer tools were crucial in fostering the DSGE approach (cf. supra), they did not play any role in the progressive dismissal of large-scale macroeconometric models of the 1970s, which, the narrative goes, progressively went out of fashion because of their outdated, “primitive” theoretical basis (Sergi, 2020).

15 A similar perspective has been taken in the history of statistics, for instance by Alain Desrosières (2008), substituting a history of “progress of measurement” throughout the evolution of statistical or econometric techniques with a “social history of quantification”, focused on the social process of establishing conventions about what should be measured.

16 This is the case for all issues considered as “technical”, such as difficulties in solving mathematical models, handling or collecting data, programming, and so forth. See for instance, the problem in the “tractability” of models—as documented by Cherrier (2023). It is also the case for issues considered trivially material, such as typing a manuscript, see for example the movement #thanksfortyping.

17 When a new computer technique becomes “fashionable” within a field (like Dynare and the Newton-Raphson algorithm in macroeconomics), putting it in the abstract can however become a sort of virtuous signaling. Nonetheless, because its usage is taken for granted, researchers are not expected to discuss its underpinnings. This can sometimes lead to blind uses of those techniques.

18 The existence of specialized outlets, like the Journal of Computational Economics, which provide confined spaces for discussing specifically computer-related techniques, does not solve this issue as it externalizes those discussions from the main community.

Haut de page

Pour citer cet article

Référence papier

Marcel Boumans, Cléo Chassonnery-Zaïgouche, Pierrick Dechaux et Francesco Sergi, « The Computerization of Economics: Three Lessons for Economics »Œconomia, 13-3 | 2023, 637-655.

Référence électronique

Marcel Boumans, Cléo Chassonnery-Zaïgouche, Pierrick Dechaux et Francesco Sergi, « The Computerization of Economics: Three Lessons for Economics »Œconomia [En ligne], 13-3 | 2023, mis en ligne le 01 septembre 2023, consulté le 06 mars 2026. URL : http://journals.openedition.org/oeconomia/15684 ; DOI : https://doi.org/10.4000/oeconomia.15684

Haut de page

Auteurs

Marcel Boumans

Utrecht University. m.j.boumans@uu.nl

Articles du même auteur

Cléo Chassonnery-Zaïgouche

Università degli studi di Bologna. cleo.chassonnery@unibo.it

Articles du même auteur

Pierrick Dechaux

REhPERE. pierrick.dechaux@gmail.com

Francesco Sergi

Université Paris Est Créteil, LIPHA. francesco.sergi@u-pec.fr

Articles du même auteur

Haut de page

Droits d’auteur

CC-BY-NC-ND-4.0

Le texte seul est utilisable sous licence CC BY-NC-ND 4.0. Les autres éléments (illustrations, fichiers annexes importés) sont susceptibles d’être soumis à des autorisations d’usage spécifiques.

Haut de page
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search