The Productivity Paradox : Computer Revolution Fails to Light Fire Under Output

Wells Fargo & Co., once a leader in implementing the high-speed communications system known as the Pony Express, has more experience than most companies in managing technological change.

But nothing could have fully prepared it for the onslaught of new technology that has transformed the financial services industry over the last two decades. Automated teller machines, unheard of as recently as the early 1970s, are now the centerpiece of Wells Fargo’s retail banking operation. Checks once painstakingly verified and logged by hand now whir though giant “reader/sorters” at the rate of several thousand per minute. Loans are made only after consulting sophisticated data bases.

Wells Fargo has invested close to $1 billion on these and other new technologies over the last decade and the payoff seems obvious. The bank now handles more than twice as many checks, keeps its branches open longer and offers a far broader array of services than it did a decade ago--all with only 15% more people.

Viewed from another angle, however, many of the apparent efficiencies at Wells Fargo and other banks are illusory. Banks can handle checks more cheaply, which encourages people to write more checks, even though computer-based automated debit systems should be making checks obsolete. If all banks use computer systems to evaluate investments, everybody’s cost rise, but the number of good investments does not.

Indeed, the economists who make their living analyzing business efficiency have consistently found that the information technology revolution has had no measurable impact on the overall productivity of American industry. Although individual firms such as Wells Fargo might be able to point out specific benefits, these efficiencies have failed to filter through to the economy as a whole.

“The cost of information technologies has fallen 15% a year for 20 years, and that’s a remarkable opportunity,” says Gary Loveman, an economist at the Harvard Business School. “You would expect to find lots of evidence of the economic benefits and major, first-order improvements in productivity. But you don’t.”

Although productivity--measured as output per worker--remains higher in the United States than in other countries, it has simply failed to improve significantly as a result of the computer revolution.

Computer professionals have greeted these findings with a revealing mixture of derision and alarm. Some industry executives and analysts say that what’s come to be known as the “productivity paradox” reflects little more than an obtuse academic devotion to irrelevant statistical measures. The greater convenience now afforded Wells Fargo customers, for example, doesn’t necessarily show up in productivity numbers.

Yet others worry that the lack of productivity gains from computer technology could be a factor in the severe slump now gripping the computer industry. And even those who believe that the issue has been overblown reluctantly agree that the productivity paradox sheds light on some important issues about how computers are used and what they can reasonably be expected to accomplish.

“We tend simply to automate what was done by hand,” says Paul R. Jones, head of computer operations for Unocal. “We need to ask: Why are we doing this faster? Do we need to do it at all? What we really need to be doing is re-engineering the business. That’s where we’ll see productivity gains.”

To be sure, part of the explanation for the debate over computer efficiency lies in the way productivity is calculated and defined. While there are several different ways of measuring productivity, all basically seek to gauge the value of output for a given level of input--be it labor or cash for equipment.

The problem is that while the input part of the equation is easy to measure--a certain number of people are employed at a given salary, a certain number of dollars are spent on computers--the output side can be difficult to evaluate in traditional terms.

“The types of things that information technologies make possible--variety, quality, timeliness--are exactly the things that don’t show up in productivity statistics,” says Eric Brynjolfsson, an assistant professor at MIT’s Sloan School of Management. “The productivity gains are there, but we’re not measuring them.”

The banking industry, Brynjolfsson says, is a perfect example. Customers now have access to ATMs, hundreds of new financial products, far more detailed account statements and other services that would have been unthinkable in the pre-computer era.

“It’s possible that adding products doesn’t help anything,” Brynjolfsson says, “but if consumers want something, there is probably some value to it.”

There are numerous examples of computers producing seemingly substantive benefits that are almost impossible to measure. What’s the value of a research report that features elegant page layouts and snazzy graphics rather than simple typewritten text? Or how about the ability to create a three-dimensional computer model of a building design, rather than rely on drawings? Or the opportunity to watch any of 50 television stations at the touch of a button?

“The concept of productivity came out of a world organized around industrial manufacturing,” says Stewart Alsop, editor of the P.C. Letter. “But what we should measure now is quality of life and quality of society, how you organize your life and how you interact with other people.”

Most economists, however, believe that the measurement question is only part of the story. There are a host of other, less-comforting explanations for the productivity paradox.

One important element is the tendency of firms in a market economy to “compete away” the benefits of innovation. In consumer goods businesses, for example, one company might gain a temporary edge by developing a computer program that enables it to better target its market.

But such an effort will be quickly matched by competitors. While the technology investment might be a wise one for all companies, there is no net effect on the economy: costs stay the same, and output stays the same since the innovation does not expand the market.

“A great deal of uses of information technologies simply reallocate things in the same way that advertising does,” Harvard’s Loveman says.

And while this is neither good nor bad in itself, Loveman points out that it undermines the common notion among policy-makers that greater use of technology will in itself improve economic performance and raise the standard of living.

Outright mismanagement of information technologies is also a factor in postponing--if not ultimately eliminating--the benefits of information technology. Few companies will admit their failures, but many have spent a lot of money on the wrong equipment.

Blue Cross/Blue Shield of Massachusetts, for one, lost many of its largest accounts--including the state of Massachusetts--in part because of a disastrous computer project that went $80 million over budget and, after numerous delays, never performed up to expectations. The company recently agreed to have an outside specialist, Electronic Data Systems, manage all its computer operations--an arrangement that is becoming increasingly popular in corporate America.

Such problems are expected to diminish as users gain experience and new, more flexible desktop computer systems take over tasks once performed by big, complicated mainframe systems. Alsop of the P.C. Letter says the benefits of the personal computer revolution--as opposed to other earlier advances in information technologies--have only begun to be felt in the last couple of years, and therefore do not show up in productivity statistics for the mid-1980s.

More troubling are the numerous situations where computers lead people to do things of dubious value that they wouldn’t do at all if it weren’t so easy. Lawyers, for example, are now tempted to word-process 40-page briefs where they might have settled for 10 typewritten pages in the past. Reading the extra pages takes more of everyone’s time, and the impact on the outcome of cases is likely to be marginal.

Similarly, the popularity of financial spreadsheets--the second most widely used of all personal computer programs--has led countless managers to spend time analyzing arcane financial scenarios that might be best ignored. And while electronic mail makes business communications efficient, it may also increase time spent sending and reading unnecessary information, such as office gossip.

Even those who remain deeply confident in the long-run benefits of information technologies agree that computer power is too often used in the wrong ways. As Esther Dyson, publisher of the influential industry newsletter Release 1.0, wrote in a recent article on the productivity issue: “Pointless processes are automated but not redesigned or streamlined, as companies try to adopt technology without disruption. A little disruption--rethinking of business processes and policies--is key to making new technology useful.”

Both academics and farsighted industry officials, in fact, are increasingly looking at how firms should be redesigned to function more productively in the information age. At MIT’s Sloan School, a new discipline called “Coordination Science” has been established to examine how computer and communications technologies might alter and improve the basic organizational structures that have been in place for more than a century.

“We’re entering a new era of how economic activity is organized, and it could be as important as the Industrial Revolution,” says Thomas Malone, who heads the new MIT program.

It is difficult to see what these organizational changes might look like, but a historical example unearthed by Stanford University economics professor Paul A. David offers a good illustration. David studied the impact of electrification, one of the most critical technological innovations of the late 19th and early 20th centuries, and the analogies with the current situation are illuminating.

As with computers, the coming of electricity had some immediate and highly visible benefits, such as electric mass-transit systems, but it did not bring about the overnight productivity improvements that many had expected from such a breakthrough.

In factories, electricity replaced the existing power sources--steam boilers and water wheels--relatively slowly. More important, it served at first as a simple substitute in the existing factory design, which was based on huge, rotating shafts that powered individual machines via a series of belts. The belt-drive system--dictated by the centralized nature of the power source--required a compact, multistory factory layout in order to keep all machines in reasonably close proximity to the shafts.

Replacing steam-driven shafts with electrically driven ones, David says, had only a minimal impact on the efficiency of factories. The real advances did not come until companies began to exploit the underlying advantage of electricity: It allowed machines to be powered individually by their own motors.

Freed from the tyranny of the energy-dissipating shafts and belts, industrial engineers could then devise new factory layouts that were cheaper, easier to modify for new products and more appropriate for material handling. Major productivity benefits could thus be realized, but it took decades for this revolution to spread throughout the industrial economy.

In today’s economy, some newer industries--unburdened by entrenched management practices and investments in old systems--have demonstrated the productivity-enhancing possibilities of new information technology. Federal Express and other overnight package-delivery firms, for example, pioneered the use of specialized hand-held computers to coordinate and track parcel deliveries in ways that would have been impossible before.

Information technology companies themselves, not surprisingly, tend to be rather effective users of computer systems. And there are many other businesses, from tiny desktop publishing companies to specialized investment management firms, that would not even exist without the new technologies.

In the long run, most economists agree that the question is not whether information technologies will improve productivity and thus help raise the standard of living, but when they will do so, and by how much. But, they add, realizing the promise sooner rather than later will require both sellers and buyers of information technologies to stop assuming there are benefits, and begin thinking about how those benefits should be achieved.