Revolution Watch: An Exchange

To the Editors:

Jeffrey Madrick’s description of the productivity disappointments of the past and the present [“Computers: Waiting for the Revolution,” NYR, March 26] is right on the mark. His skepticism about the capacity of computers to boost productivity growth in the years to come is understandable, given the developments of the past quarter-century. But his dour forecast for the future is too bleak. He implicitly assumes that the best way to predict the near future is to extrapolate from the recent past. That’s a respectable way to make a forecast, but not always the best way.

Sometimes a careful observer can see something happening before it shows up in the official statistics. This is such a moment. Thanks to the maturation of computer technology and American management practices, we are indeed on the verge of a significant upturn in productivity growth. It is easy, as Mr. Madrick demonstrates, to find examples of doubtful benefits from computerization or computer advances that seem to carry us two steps forward and one, and sometimes even two, steps backward. One can also identify counterexamples where computers have transformed the way we work (compare the time it takes to order a book from Amazon.com to the time it takes to find it in the local bookstore?) or to do things that we don’t yet know how to fully value (what’s it worth to diagnose a tumor early enough to remove it before the cancer spreads?). The question is: Which set of examples tells us more about the future? Perhaps Mr. Madrick is right, and we will continue to waste money by using personal computers as overpowered electric typewriters or by inflicting poorly designed systems on American workplaces. But people aren’t stupid: They eventually learn from the mistakes and successes of others, and replicate more of the successes than the mistakes.

As Mr. Madrick notes, economic historian Paul David has drawn an elaborate comparison between the early days of electricity and our present productivity paradox. He says it takes a long time for human beings to realize the productivity potential of radically new technologies. Mr. Madrick dismisses Mr. David’s argument by mentioning all the other factors that contributed to a productivity surge in the 1920s. Other factors were important; the 1990s are not the 1920s. Mr. David’s analysis is best seen as a metaphor. He figures that electricity added five percentage points to manufacturing productivity in the 1920s. Mr. Madrick would agree that if computers help us to eke out just one-half percentage point faster productivity growth, boosting productivity growth from the recent 1.1 percent a year to around 1.6 percent, we will have turned a significant corner in US economic history.

Mark Twain understood. In A Connecticut Yankee in King Arthur’s Court, a New England machinist is magically transported back to sixth-century England. He brings with him the miracles of nineteenth-century engineering. But the knights eventually rebel, destroying the telephones, phonographs, and typewriters. New technology, Twain tells us, won’t change people’s lives permanently until the culture and mindset are ready for the change. “Old habit of mind is one of the toughest things to get away from in the world. It transmits itself like physical form and feature,” he observes.

Dan Sichel, to whom Mr. Madrick refers approvingly, does argue that the US will have to invest more in computer technology to produce a historic change in productivity that rivals the contribution of the railroads, for instance. Railroads accounted for roughly 12 percent of the national stock of plant and equipment at the end of the last century, Mr. Sichel has told us. Current investment in computer equipment amounts to only 2 percent or 3 percent. But that shortchanges the significance of the computer era. It leaves out the massive investments in software and telecommunications gear, and the potent microprocessors that control cars, televisions, heating and cooling systems, and a myriad of other devices. Add it all up, and it accounts for 10 percent or so of the national capital stock.

It is noteworthy that among those who share our view is Federal Reserve Chairman Alan Greenspan. With increasing conviction, he argues that we are now witnessing the early stages of a technology-generated upturn in the productivity trend. As he has observed, when businesses raise inflation-adjusted wages (as they’re beginning to do) without raising prices and manage to maintain fat profit margins, productivity must be improving. The big difference between Mr. Greenspan’s opinions and those of other economists is that his matter. Other mainstream economists have been warning for well over a year that the economy has been growing too fast, that the unemployment rate has fallen too low, and that inflation is just about to show itself. Mr. Greenspan has resisted this pressure, precisely because he sees what Mr. Madrick doubts. And Mr. Greenspan’s recent track record is pretty darn good.

Bob Davis
David Wessel
The Wall Street Journal
New York City

Jeff Madrick replies:

Mr. Davis and Mr. Wessel, among the nation’s best business reporters, claim with uncharacteristic immodesty that a careful observer can see something before it shows up. Observers have been seeing this “something” for nearly two decades, however, and it has yet to arrive. For all their protestations, the authors’ analysis amounts only to a profession of faith in technology with little serious attention paid to how such technologies differ in their impact on the economy, and how economic times themselves may fundamentally change. Mr. Davis and Mr. Wessel are the ones who are merely extrapolating from the past to predict the future.
Their basic argument is that “people aren’t stupid,” and won’t use computers for long if they waste money. This betrays a misunderstanding. In fact, computers can be used quite rationally to make it easier to differentiate products, make them more marketable through design and packaging, improve services, and in general keep up with the competition to retain or improve market share, among other uses. All these may increase revenues. But they do not necessarily raise productivity in doing so. They may even occasionally reduce productivity because such competitive, computer-related strategies often require more input from people—more hours of work. You increase productivity by reducing the number of people used to make goods or services. Computers simply may not do that as rapidly as did traditional mass production and earlier technologies. This is the “something a careful observer” would do well to see before others do.

As for Paul David’s widely cited paper on how long it took electricity to affect productivity, I and many others do not believe it makes a convincing case. I don’t know whether Mr. Wessel and Mr. Davis realize that the key statistical evidence for David’s argument—which purports to show the relation between electricity and productivity—is based on remarkably few observations. Mr. David, an accomplished economic historian, readily admits that his paper depends on conjecture, whether it is understood as “metaphor” or as historical fact. The growth of productivity in this period was related to many other factors besides electricity: the extraordinary innovations of Henry Ford which led to true mass production, for example.

Dan Sichel has not left out software from his computations, as Mr. Wessel and Mr. Davis should know. He does not believe you can accurately compute the capital stock of software, though this has not deterred Wessel and Davis from rather causally doing so. But he has cleverly and painstakingly included measured software spending (as opposed to the stock of capital) in his analysis of the degree to which it could affect productivity, and it cannot have affected it very much. Mr. Davis and Mr. Wessel would also include telecommunications investment in the computation so as to reach 10 percent of total capital stock. This mystifies me. Surely, most such long-term capital stock consists of telephone wiring and switching equipment, not hot new computerized systems. By the way, Mr. Sichel believes railroad capital likely came to 18 percent of the total, and no less than 14 percent.

As for Alan Greenspan, he has produced little serious evidence of a productivity revolution. He often cites the productivity record for nonfinancial corporations because it has risen faster than broader measures. But many economists doubt that this category of measurement is any more accurate than the others commonly used, and it may be less accurate. It shows no productivity slowdown since the early 1970s, for example, which is quite at odds with almost all other measures of business productivity. (And to get a bit technical, it is based on the income rather than product accounts, which are highly inflated recently by capital gains, which have nothing to do with the output of goods and services.)

I doubt the cautious Mr. Greenspan has kept interest rates down because of his unsupported conviction about productivity. Rather, inflation is almost negligible these days partly owing to falling import prices and relatively low wages. Computers have been around for a long time and productivity has grown at historically slow rates throughout this period. The burden of proof is on those who insist the revolution is right around the corner.