The GPT That Dares Not Speak Its Name

President Obama has been a little slow in building a narrative of the Whys and Wherefores of the situation he unexpectedly inherited.  So it was a step in the right direction last week when he said at his press conference, “[T]his crisis didn’t happen overnight and it didn’t result from any one action or decision. It took many years and many failures to lead us here. And it will take many months and many different solutions to lead us out.”


Not that it’s easy to build political consensus about a crisis. People are still arguing about the causes and cures of the Great Depression, after all. But it is critically important to make a beginning:  the midterm elections are barely eighteen months away. So I was glad to see both Paul Krugman and Alan Greenspan last week take a crack at explanation.


Greenspan, writing in the Financial Times, traced the origins of the system that broke down to “the extraordinary risk management discipline that developed out of the writings of the University of Chicago’s Harry Markowitz in the 1950s.”  That skein of work had won several Nobel Prizes, Greenspan noted, for Markowitz and others. (The story of the first 35 years or so was beautifully told by Peter Bernstein in 1991 in Capital Ideas: The Improbable Origins of Modern Wall Street.) The discipline had been gradually embraced over a half century, the former Fed chairman noted, not just by academia but by a large majority of financial professionals and regulators around the world. Then in 2007, the risk-management structure cracked – overwhelmed, on the one hand, by the complexity of the instruments, and, on the other, by insufficient capital reserves put by against loss.


Krugman, not surprisingly, zeroed in on the deregulation of financial markets that followed the election of Ronald Reagan in 1980, after which the financial sector began to grow rapidly. “Old-fashioned banking was increasingly replaced by wheeling and dealing on a grand scale,” he wrote in The New York Times last Friday. The size of the financial sector soon doubled – even in the go-go years of the 1960s, the finance and insurance industries had amounted to no more than 4 percent of the American economy, but by the eve of the current crisis they had grown to 8 percent of GDP – largely on the strength of the Markowitz-inspired process known as securitization, turning illiquid debt instruments such as mortgages into tradeable securities.


Myself, I thought of Nathan Rosenberg.


Not that Rosenberg, 81, ever worked for a bank of any sort, a hedge fund, an insurance company – or even a think-tank. He is a retired Stanford University economist, the author of several books, including Inside the Black Box and How the West Grew Rich (with L.E. Birdzell), one of the nation’s most distinguished historians of technology. There is not a rapacious bone in his body.  But it was Rosenberg who, in 1962, clearly formulated an idea that, once it had passed through a great many hands, was used to justify policies in the Nineties and early Oughts that have proved to be ruinous, at least in the present day. By that time, Rosenberg’s idea had acquired a name – General Purpose Technology, or GPT.


And while no one has done more to hone our sense of how economic growth unfolds to national advantage than Rosenberg, it is also true that, once everyone understands how a policy trick works, the trick stops working. (In central banking that’s called Goodhart’s Law.)  The great thing about Rosenberg’s discovery – it is not too much, I think, to call it that – goes a long way towards illuminating what those wild and crazy bankers and their political enablers were thinking among themselves – to the extent they were thinking – as they levered up the Anglo-American financial system to nosebleed levels.


Let me explain.  Consider, as did Rosenberg, in “Technological Change in the Machine Tool Industry, 1840-1910,” the history of the turret lathe.


In 1820, no machine tool industry existed in the United States (or, for that matter, anywhere in the world). There were plenty of machines, of course, but they were home-made affairs, built on an ad hoc basis by those who would ultimately use them.  Forging and cutting metal into the precise shapes required for looms, gears, boilers and axles wasn’t easy. Not surprisingly, factories specializing in a final product – in textiles, especially – had the best machine shops. Lowell Mills, in Lowell, Mass., and the Amoskeag Manufacturing Co., in Manchester, New Hampshire,  began selling textile machinery to other firms and then various other sorts of machinery as well:  steam engines, turbines, mill machinery and, most important, the cutting, drilling and shaping tools they used to make their own machines.


Between 1840 and 1880, a vast wave of mechanization occurred. The Lowell Machine Shop became independent of its mills in 1845. Soon it was producing steam locomotives for the new railroads. The Baldwin Locomotive Works, in Philadelphia, emerged from a textile-printing firm to become the nation’s largest engine-maker. Arms manufacturers invented a series of lighter tools for making interchangeable parts that could be assembled smoothly into weapons – jigs, taps and gauges, turret lathes, milling machines, precision-grinders. Whole new industries emerged:  shoe machinery, locomotives, dynamos, bicycles, sewing machines, typewriters, and, of course, the machine tool industry itself, heavily concentrated in the Connecticut and Ohio River valleys, providing lathes, planers, drillers, borers, grinders, millers and shapers to all the rest.


That this specialization proceeded by spin-off had long been noted, Rosenberg wrote.  Less commonly recognized was the underlying process he called technological convergence – the process by which a solution to a problem in one industry is quickly employed to solve roughly similar problems in other industries. He illustrated it with the story of a series of lathes. Thomas Blanchard’s stocking lathe of 1818, invented to carve gunstocks, which previously had required whittling, boring and chiseling by hand, was quickly put to work making hat blocks, wheel spokes, oars, sculptured busts and shoe lasts. Stephen Fitch’s turret lathe of 1845, with its cluster of tools held on a vertical axis, made it possible to perform a sequence of operations without removing the piece from the lathe. Fitch built the first one to make parts for government pistols, but soon the principle was adapted by other hands to make components for watches, sewing machines, typewriters, locomotives, bicycles and, eventually, automobiles.  Sewing machines provided a further example: mechanized stitching gave rise to huge industries of mass-produced boots and shoes, ready-to-wear clothing, and was employed in countless other ways, manufacturing everything from tents and sails to harnesses and books.


Rosenberg concluded:

The interesting thing about the group of industries discussed here is that they were all dependent in their development upon technological changes dealing with a limited number of processes and that the solution to problems posed by these processes eventually became the specialized function of a well-organized industry. A question of more contemporary interest is whether similar technological convergences are occurring in twentieth century conditions; whether, for example, the chemical and electronics industries are playing the same roles for production and transmittal that machine tools played at an earlier stage of our history.

Rosenberg’s paper, which appeared in the Journal of Economic History in 1963, became an underground classic among those who thought about the economics of specialization, along with essays by Adam Smith, Allyn Young and George Stigler, but it otherwise lay little read and unattended by economists concerned with macroeconomics. Its gist, on the other hand – that certain sectors contained the possibility for growth so rapid that they would cause compositional changes in the economy itself – gradually became part of the folklore of post-World War II America.  The point was neatly encapsulated in the patronizing advice a businessman gives his partner’s 22-year-old son in the 1967 movie “The Graduate:” “I just want to say one word to you – just one word – ‘plastics.’”


There continued to occur such “revolutions,” (that being the word most often used to describe the advent of transformative new technologies) – microprocessors one decade, software the next, the Internet and the World Wide Web the decade after – and gradually economists took note. In 1990, Harvard Business School guru Michael Porter published The Competitive Advantage of Nations, with its emphasis on the advantages derived from the clustering of related industries, including financial services.  The following year, Stanford economist Paul David gave a major push to such thinking with “The Computer and the Dynamo,” a paper comparing the history of the diffusion of the electric motor between 1880 and 1930 with that of the computer. And in 1995, Manuel Trajtenberg and Stanford University economist Timothy Bresnahan introduced a formal model to describe various “general purpose technologies” as engines of growth – the steam engine, electricity, the laser, the computer, the Internet.


Which brings us back to the current crisis. Until recently, the American and British banking and financial services industries were understood, however imperfectly, both by their practitioners and those who sought to regulate them, as embodying a GPT for the twenty-first century, a worthy successor to the Internet technology boom that sent American servers, routers and fiberoptic cable, and British media and advertising,  around the world. To them , that compositional shift of financial services to 8 percent from 4 percent indicated success, not excess (though perhaps today no one doubts that it went too far)


If there is a good history of the modern financial industry, chronicling the development of its many recondite risk-management techniques, on the model of, say, Janet Abbate’s Inventing the Internet, or M. Mitchell Waldrop’s The Dream Machine: J.C.R. Licklider and the Revolution that Made Computing Personal, I don’t know it. The closest probably is still Ross Paving Wall Street: Experimental Economics and the Quest for the Perfect Market. Equally interesting, though heavier on the sociology of finance than the underlying economic rationale, is An Engine Not a Camera: How Financial Models Shape Markets, by Donald MacKenzie. Soon the histories of credit default swaps will begin rolling in, led by Gillian Tett’s Fool’s Gold. (Matt Taibbi has a boisterous version of the many of the same events, The Big Takeover, in the current Rolling Stone.) But capturing the broad outlines of the story of the rise of modern finance, perhaps from the moment when the old Chicago Butter and Egg Board reorganized itself as the Chicago Mercantile Exchange and set out to create the first financial derivatives, trading futures on currencies, Treasury bills and interest rates, will be the work of many years.


In the meantime, though, to understand the difference of opinion that has developed between the administration and its liberal critics over how to go about recapitalizing the banking industry, it helps to recognize that, from the quarterdeck at least, the question is as much one of industrial policy as public finance. For whatever combination of political and strategic reasons, the administration wants to keep the industry more or less intact.  The critics would prefer to break it up to some degree.  It is crucial that policymakers and their interpreters begin to fill in some of the blanks.


Hard as it is to see in this climate of fear and recrimination, the US must remain among the world’s leaders in banking and finance. It is a matter of national security – as is the even more urgent matter of raising the industry’s safety standards. The technologies of risk management that began with the work of Harry Markowitz are here to stay – option pricing, dynamic hedging, risk arbitrage, auction design and all the rest. Risk managers must be reined in, however, tethered, harnessed, contained, made to serve the public purpose, after having so grandly betrayed it. “Masters of the universe” no longer:  financial engineering is the general purpose technology that dares not speak its name.