Bretton Woods, Past and Present: 3. Models in Economics


I cannot resist but to start quoting Mary Morgan’s second entry to the second edition of the New Palgrave: “Modeling became the dominant methodology of economics during the 20th century.”

If models were not a “recognized category in discussions about methodology” in the late 19th century, they have become central to economics after the 1930s. Models are “complex objects constructed out of many resources” that come from disparate sources and serve as measuring instruments (as argued by Marcel Boumans in his 2005 book How Economists Model the World into Numbers), and are autonomous agents that mediate theories and the real world (as argued by Mary Morgan and Margaret Morrison in their chapter to the 1999 edited volume Models as Mediators). And, returning to the Palgrave entry, models function in different ways for different purposes:

(1) Models serve to fit theories to the world: they are here often seen as econometric objects, with well-defined statistical and mathematical properties, used in empirical work. Models are here measuring instruments used as investigative devices.

(2) Modeling as a way of economic theorizing, increasingly by using mathematics in economics. Mary Morgan explores here the idea of Tjalling Koopmans that economic theory is a sequence of models: they help us “arrange and record our logical deductions in such a manner that any particular conclusion or observationally refutable implication can be traced to the postulates on which it rests” and each one seeks “to express in simplified form different aspects of an always more complicated reality” (Koopmans 1957, quoted by Morgan)

(3) More generally, models are tools or investigative instruments. Here we have econometric and mathematical (geometric or analytical) tools, but also simulation tools and even hydraulic analogues of an economy. The gist here is that models are tools that allow economists to explore the world by using and manipulating them, more than on building such tools.

In the video that follows, we have eminent economists and a journalist pondering what is a good model in economics, how they contribute (or not) to the progress of economics, and if they have a role to play. What strikes is the close association of models with economic theories, and at some points it seems that people are in fact talking about theories, not models understood as those complex mediating objects or measuring instruments that we just discussed.

After Philippe Aghion’s (Harvard University) statement that we can see what are the good models by those that manage to survive over time, we come to what we can call traditional views on models. Bradford DeLong (University of California, Berkeley) recapitulates Milton Friedman’s position that basically says that good models are those that predict well, not necessarily those that are realist descriptions of the world. And the tension between Friedman’s position and the opposite view, that models ought to be “realistic,” informs several of the following comments. We have considerations that good assumptions are important, that models are always in a context, and that they are glasses (used to integrate theory and “empirics”) or serve as benchmarks (and, thus, are not directly testable because they are unrealistic representations). DeLong’s next piece stresses the different roles that models can play, closely related to the three roles emphasized by Mary Morgan.

It is clear that given that modeling has become central to economics since the 1930s, models are central to better understanding the world in which we live, which was identified as a major understanding of progress in economics, as Floris illuminated us with his video and post, with the very interesting points raised by Yann. Aghion stresses that models allow economists to make progress by better understanding reality, getting rid of bad and less general models along the way. In fact, these economists, who I believe are quite representative of a larger population, do not fully articulate the functions and purposes models have. So it is not surprising that they have no historical (and sociological) understanding on how modeling evolved in economics – it is fair to counter-argue that they were not explicitly asked about this. It is in this sense that I see some confusion of models with theories and both as products that economists always offered to capture the complexity of the world. We are then left with James Galbraith’s (University of Texas at Austin) assertion that economics ought to go beyond modeling, which can serve as a hindrance to the evolution of this science and a religious cult of autistic scientists. As we can see, besides being tools, models are indeed a source of passionate discussions related to knowledge production and the evolution of the community of these thinkers.

Share your perspective