Editorial

It is with great honour and pleasure that I am introducing this special issue of the International Journal of Microsimulation, devoted to the work of Gunnar Eliasson around the MOSES (MOdel of the Swedish Economic System) microsimulation architecture.

Arguably, dynamic microsimulation has three parents: Guy Orcutt and Barbara Bergmann in the U.S., and Gunnar Eliasson in Europe. In the late 1950s, Guy Orcutt had the original idea of using representative distributions of individuals, households, or firms as inputs for large-scale computed models - “a new type of modelling approach” that would emphasise their heterogeneous characteristics and decision making, allowing to represent and analyse not only economy-wide averages, but the entire distribution of relevant outcomes (Orcutt, 1957; see also Watts, 1991). Working with a team of doctoral students and programmers at the University of Wisconsin - Madison, he developed his first “microanalytical model” (Orcutt et al., 1961), later morphing into the DYNASIM model he developed at the Urban Institute and Yale (Orcutt et al., 1976). At the same time, Barbara Bergmann at the University of Maryland was working on her “U.S. Transactions” model (Bergmann, 1973). Bergmann is most renown for her work as a pioneering feminist economist, but her contributions were also foundational for the emerging field of microsimulation modelling (Olson, 2007). Meanwhile, in Stockholm Gunnar Eliasson – then Chief Economist at the Federation of Swedish Industries – was in talks with IBM to develop a computer model of the Swedish economy. It was Kenneth Arrow who put him in contact with Bergmann and Orcutt (Eliasson, 2018), and he quickly realised that it was Bergmann’s model that was closer to the specification he had in mind, namely a complete model of the economy encompassing not only individuals and households, but also firms. The model was thoroughly described in a lengthy 1976 paper (Eliasson, 1976), with results being published as Eliasson (1977). For their nature emphasising interactions between economic agents, Bergmann’s and Eliasson’s models can rightly be considered as ante litteram data-driven macro agent-based models (Richiardi, 2014; Richiardi et al., 2024). The pioneering role of Orcutt, Bergmann and Eliasson can be appreciated in a joint edited volume (Bergmann et al., 1977). 50 years on, the DYNASIM model is still developed at the Urban Institute, while the MOSES architecture, after having proved fruitful for decades, is dormant but ready to be revived (the pseudo-code for the model is made available for download as part of this special issue).

Aside some prestigious hits however (Orcutt publishing in the Review of Economics and Statistics, Bergmann in Econometrica, Eliasson in the American Economic Review), much of this early work is documented only in reports, working papers, technical notes, or book chapters that are increasingly difficult to access. Such was the norm at the time, with much less emphasis than today on journal publications. One aim of this special issue is therefore to save this important material from fading away, and publish it in journal format for the first time. The fact that Gunnar himself managed to curate the special issue, selecting the articles, abridging them wherever appropriate, adding clarifications here and there, and writing an introductory article where he reflects on his 50-year experience, summarises the main contributions of his modelling effort and places MOSES in the context of the history of economic ideas, is outright exceptional. We are extremely grateful to him for this effort, which we are proud and honoured to host on the pages of the journal.

An outline of the special issue is presented in Table 1. The first article is novel and contains, as already mentioned, Gunnar’s own appreciation of the MOSES experience (the pseudo-code for the last version of the model is attached as additional material to this article). This is followed by 9 other articles, only some of which were originally accessible, either in working paper format or as book chapters.¹ The first of these articles (Eliasson, 1976) is an abridged version of the original model description paper. The subsequent articles describe applications and developments of the model.