David Warsh: Economic models and engineering
SOMERVILLE, Mass.
The best book about developments in the culture of professional economics to appear in the last quarter century is, in my opinion, The World in the Model: How Economists Work and Think (Cambridge, 2012), by Mary S. Morgan, of the London School of Economics and the University of Amsterdam. The best book of the quarter century before that is, again, according to me, An Engine, Not a Camera: How Models Shape Financial Markets (MIT, 1997), by Donald MacKenzie, of the University of Edinburgh.
Both books describe the introduction of mathematical models in the years beginning before World War II. Both consider how the subsequent use of those techniques has changed how economics is done by economists. Morgan’s book is about the kinds of models that economists devise experimentally, not those that interest MacKenzie most, models designed to be tested against the real world. A deft cover illustrates Morgan’s preoccupation, showing the interior of a closed room with only a high window. On the floor of the room are written graphic diagram of supply and demand. The window opens only to the sky outside, above the world itself, a world the model-builder cannot see. The introduction of statistical inference to economics she dealt with in The History of Econometric Ideas (Cambridge, 1990).
I remember the surprise I felt when I first read Morgan’s entry “Economics” in The Cambridge History of Science Volume 7: The Modern Social Sciences (Cambridge, 2003). She described two familiar wings of economics, often characterized in the 19th Century as “the science of political economy” and “the art of economic governance.” Gradually in that century they were relabeled “positive” economics (the way it is, given human nature) and “normative” economics (the way it ought to be).
Having practiced economics in strictly literary fashion during the modern subject’s first century, Morgan continued, economists in the second half of the 19th Century began adopting differential calculus as a language to describe their reasoning. In the 20th Century, particularly its second half, the two wings have been firmly “joined together” by their shared use of “a set of technologies,” consisting mainly of mathematics, statistics and models. Western technocratic economics, she wrote, had thereby become “an engineering science.”
I doubted at the time that it was especially helpful to think economics that way.
Having read Economics and Engineering: Institutions, Practices, and Cultures (2021, Duke), I still doubt it. That annual conference volume of the journal History of Political Economy appeared earlier this year, containing 10 essays by historians of thought, with a forward by engineering professor David Blockley, of the University of Bristol, and an afterword by Morgan herself. Three developments – the objectification of the economy as a system; the emergence of tools, technologies and expertise; and a sense of the profession’s public responsibility – had created something that might be understood as “an engineering approach” to the economy and in economics, writes Morgan. She goes on to distinguish between two modes of economic engineering, start-fresh design and fix-it-up problem-solving, noting that enthusiasm for the design or redesign of whole economies and/or vast sectors of them had diminished in the past thirty years.
It’s not that the 10 essays don’t make a strong case for Morgan’s insights about various borrowings from engineering that have occurred over the years: in particular, Judy Klein, of Mary Baldwin University, on control theory and engineering; Aurélien Saïdi, of the University of Paris Nanterre, and Beatrice Cherrier, of the University of Cergy Pontoise and the Ecole Polytechnique, on the tendencies of Stanford University to produce engineers; and William Thomas, of the American Institute of Physics, on the genesis at RAND Corp. of Kenneth Arrow’s views of the economic significance of information.
My doubts have to do with whether the “science” of economics and the practice of its application to social policy have indeed been in fact been “firmly joined” together by the fact that both wings now share a common language. I wonder whether more than a relatively small portion of what we consider to be the domain of economic science is sufficiently well understood and agreed-upon by economists themselves as to permit “engineering” applications.
Take physics. In the four hundred years since Newton many departments of engineering have been spawned: mechanical, civil, electrical, aeronautical, nuclear, geo-thermal. But has physics thereby become an engineering science? Did the emergence of chemical engineering in the 1920s change our sense of what constitutes chemistry? Is biology less a science for the explosion of biotech applications that has taken place since the structure of the DNA molecule was identified in 1953? Probably not.
Some provinces of economics can be considered to have reached the degree of durable consensus that permits experts to undertake engineering applications. I count a dozen Nobel prizes as having been shared for work that can be legitimately described as economic engineering: Harry Markowitz, Merton Miller and William Sharpe, in 1990, for “pioneering work in financial economics”; Robert Merton and Myron Scholes, in 1997, “for a new method to determine the value of derivatives”; Lloyd Shapley and Alvin Roth, in 2012, “for the theory of stable allocations and the practice of market design”: Abajit Banerjee, Esther Duflo and Michael Kremer, in 2019, for “their experimental approach to alleviating global poverty”; and Paul Milgrom and Robert Wilson, in 2020, for “improvements to auction theory and inventions of new auction formats.”
This is where sociologist Donald McKenzie comes in. In An Engine Not a Camera, he describes the steps by which, in the course of embracing the techniques of mathematical modeling, finance theory had become “an active force transforming its environment, not a camera, passively recording it,” but an engine, remaking it. When market traders themselves adopted models from the literature, the new theories brought into existence the very transactions of which abstract theory had spoken – and then elaborated them. Markets for derivatives grew exponentially. Such was the “performativity” of the new understanding of finance. After all, writes Morgan in her afterword, hasn’t remaking the world been the goal of economic-engineering interventions all along?
Natural language has a knack for finding its way in these matters. We speak easily of “financial engineering” and “genetic engineering.” But “fine-tuning,” the ambition of macro-economists in the 1960s, is a dimly remembered joke. The 1942 photograph on the cover of Economics and Engineering – graduate students watching while a professor manipulates a powerful instrument laden with gauges and controls – seems like a nightmare version of the film Wizard of Oz.
John Maynard Keynes memorably longed for the day when economists would manage to get themselves thought of as “humble, competent people on a level with dentists.” Nobel laureate Duflo a few years ago compared economic fieldwork to the plumbers’ trade. “The scientist provides the general framework that guides the design…. The engineer takes these general principles into account, but applies them to a specific situation…. The plumber goes one step further than the engineer: she installs the machine in the real world, carefully watches what happens, and then tinkers as needed.”
The $1.9 trillion American Rescue Plan Act that became law last week, with its myriad social programs, is not founded on what “the science” says. It is an intuition, an act of faith. Better to continue to refer to most economic programs as “strategies” and “policies” instead of “engineering,” and consider effective implementations to be artful work.
David Warsh, an economic historian and a veteran columnist, is proprietor of Somerville-based economicprincipals.com, where this column first appeared
Copyright 2021 by David Warsh, proprietor