What Theory Does Best

It was thirty-five years ago that young economists first explicitly tackled the mysteries of technical change, opening up a topic that until then had been carefully bracketed and set aside as a black box by macroeconomists preoccupied with the rhythms of the business cycle. One unassailable gain from that time is the notion of general purpose technologies (GPT).

Structural aspects of industrial revolutions that previously had been the province of economic historians, including Nathan Rosenberg, David Landes, and Paul David, were suddenly brought under the lens of models, measurement, and statistical assessment.

GPTs are novel possibilities – discoveries and inventions when they first appear – that open up myriad opportunities rather than offering one-time solutions. Classic examples of GPTs are the ones that Timothy Bresnahan, of Stanford University, and Manuel Trajtenberg, of Tel Aviv University, studied when they introduced the concept, in 1991, and five years later, examined their diffusion: the steam engine, electricity, the semiconductor.

A GPT is a technology that both finds itself adopted by an increasing number of user sectors, and fosters complementary advances in other sectors that raise the attractiveness of its adoption. Think of it, if you like, as the distinction familiar nowadays between an operating system and its many applications.  An extended comparison of the buildout of industries based on electricity and information technology, following Paul David’s landmark essay, The Dynamo and the Computer, is here.

Rosenberg, elucidating more clearly than any other the concept of technological interdependence at the heart of the GPT, illustrated its meaning in a famous article in 1963 by describing the vertical disintegration of the machine tool industry in the US in the nineteenth century. It began, he wrote, with companies that manufactured final products from scratch (steam engines, turbines, water frames). After twenty years or so, new firms hived off to make lathes, planers, and boring machines to sell to manufacturers of arms and railroad equipment. Another twenty years and a third generation of specialists went into business, creating turret lathes, milling machines and precision grinders that were soon employed in making sewing machines, bicycles and automobiles.

In Prometheus Unbound: Technological Change and Industrial Development in Western Europe from 1750 to the Present,  Landes remarked a few years later on the way that water power and mechanical looms in the late eighteenth century  stimulated the growth of the modern chemical industry in the early nineteenth

The transformation of the textile manufacture, whose requirements of detergents, bleaches, and mordants, were growing at the same pace as output, would have been impossible without a corresponding transformation of chemical technology. There was not enough cheap meadowland or sour milk in all the British Isles to whiten the cloth of Lancashire once the water frame and mule replaced the spinning wheel; and it would have taken undreamed of quantities of human urine to cut the grease of the raw wool consumed by the mills of the West Riding.

The GPT concept has obvious applicability in the present day – to big topics, such as the argument that Robert Gordon, of Northwestern University, has begun with various techno-optimists through the appearance of his book The Rise and Fall of American Growth: The US Standard of Living since the Civil War; and to small but highly consequential financial matters, such as the future of the smart phone, with its compass and accelerometer, as an instantiation of an unfolding GPT. (Did you notice those cardboard viewers that Google has been passing out to, among others, New York Times subscribers?)

But it was a very different application of the theory of GPTs that occurred to me the other day.

I was talking to a friend who had asked about the almost unfathomably great fortunes tossed up by the computer age.  Their relative magnitudes weren’t so different from the great fortunes of the past, conferred on entrepreneurs of textiles, railroads, oil, steel, chemicals, explosives, electricity, trolleys, automobiles, or finance, I said, reciting the roster of usual GPT suspects. We talked a little about the crazy big houses that such magnates built in different eras.

The next day I listened as another friend describe the controversy that has arisen over eighteenth–century slave master, college professor and Harvard Law School benefactor Isaac Royall Jr.’s place of honor at the law school. Similar reservations over loyal heroes have been expressed recently at Yale, Princeton, and Oxford.  Later I was surprised when it occurred to me that the first major GPT of modern times remains nearly invisible.

Was the Atlantic slave trade that flourished for 350 years after 1500 a GPT?  I called Stanley Engerman, at the University of Rochester, to ask his opinion. As described in a collection of essays prepared by former students, he is a man “who has spent as good part of his career exploring societies that espoused freedom, yet practices, and legally enshrined, extreme restraints on some individuals in the interests of improving the conditions of others.”

Engerman was co-author, with Robert Fogel, of the University of Chicago, of Time on the Cross: The Economics of American Slavery (Norton paperback, 2013)  an economic history classic which, despite a furious reception when it appeared in 1974, ushered in a new era of scholarship on slavery by putting economic analysis at the center of it. The authors argued that the enterprise was profitable in the main and conducted is ways more generally humane than was commonly believed. Engerman might have shared a Nobel Prize in 1994 with Fogel and economic historian Douglass North, had the Nobel committee been so inclined. His Slavery, Emancipation, & Freedom: Comparative Perspectives, three lectures given at Louisiana State University, in 2005, is a brilliant short gloss on the subject. Almost eighty, Engerman is preparing, with David Eltis, of Emory University, to close up the fourth and final volume of the Cambridge World History of Slavery. (It is volume two, covering the medieval period, that contained the surprises, he confided.)

“It was a very effective innovation,” Engerman said, after refreshing his memory by looking briefly at General Purpose Technologies and Economic Growth, a volume of essays edited in 1998 by Elhanan Helpman, of Harvard University.  The industry was about human labor and not machinery, he observed, but there was plenty of experimentation with the “performance characteristics” of slaves over the centuries: where they came from, how they were obtained, how they were treated after they were sold.  It was true, too, that many complementary innovations occurred in the ways in which slaves were deployed: as tending rice and tobacco crops in the Old South of the US; planting cotton in the New South; harvesting sugar throughout the Caribbean; farming coffee and building railroads in South America. Almost always the skills they had gained were mentioned when slaves were offered for sale. (A recent popular history, The Half Has Never Been Told: Slavery and the Making of American Capitalism, by Edward Baptist [Basic Books, 2014], begins with several pages of excellent maps.) Many improvements in husbandry and agriculture were developed by the slaves themselves.  As for “vertical dis-integrations,” meaning the practice of inventing new and related businesses, the US textile industry grew directly out of the application of slave labor to cotton agriculture in fertile lands west of the Allegheny Mountains. Perhaps someone will take up the challenge of integrating the Atlantic slave trade into the literature of GPTs. I hope so.

Whether they do any time soon, this strikes me as a pretty good example of what is most valuable about a good theory.  It make you see familiar phenomena in a different light.  I’ll cite just one example of the potential of the GPT interpretation to change conventional wisdom of the present day. A few years ago, economic historians Landes and Joel Mokyr, of Northwestern University, and economist William Baumol, of New York University, assembled an oversized volume of essays for the Kauffman Foundation, The Invention of Enterprise: Entrepreneurship from Mesopotamia to Modern Times (Princeton, 2010).  Slavery rates just two mentions in the text:  Roman slaves as entrepreneurs; and trade during the Middle Ages between Anglo-Saxon and Arab markets (the latter being one of the surprises contained in Volume 2 of the Cambridge World History). The essay on “Entrepreneurship in the Antebellum United States,” by Louis P. Cain, of Loyola University Chicago, doesn’t so much as note the practice.

That’s a mistake.  Slavery was not some aberration to be put aside out of embarrassment, but a central topic in the economic history of the New World.  A good theory helps to open your eyes.

Leave a Reply

Your email address will not be published. Required fields are marked *