Computers: Waiting for the Revolution

1.

In early January, the Bureau of Labor Statistics reported data, including strong job growth in December, suggesting that the economy again grew rapidly in the last three months of 1997. This followed eighteen months in which the real Gross Domestic Product (discounted for inflation) had already grown by an annual rate of 4 percent, and unemployment had fallen to a quarter-century low of 4.7 percent. The output of goods or services per hour of work (known as productivity) had risen by 2.25 percent in the preceding twelve months, and nearly 2 percent a year since 1996, well above its historically slow annual growth trend of 1 percent since the early 1970s (though after revisions it will probably have slowed significantly in the fourth calendar quarter of 1997). All the while, consumer inflation had fallen to less than 2 percent, and Washington was exulting because rising federal tax revenues had nearly eliminated the budget deficit well ahead of schedule.

Short spurts in growth occurred before, in the 1970s and the 1980s, but they quickly subsided. But now, with inflation, unemployment, and the deficit reduced beyond expectations, the enthusiasm that typically accompanies even modest improvements in the economy is at a higher pitch than usual. To many newspaper and magazine writers, business consultants, and some economists, the surprisingly strong performance of the economy since 1996 suggests that the long-awaited “computer revolution” has at last taken hold. They argue that, because of the spreading use of computers, the economy has finally passed through a threshold which will enable it to grow rapidly once again after twenty-five years of historically torpid growth—thus, the “new economy.”

Vice President Al Gore is among the most ardent supporters of this vision. Last summer, he announced that he, like Alexis de Tocqueville a century and a half earlier, had taken a tour of the nation “to see an old economy once again giving way to a new one.” Gore and many others credit the computerization of business and society at large with a transformation that will rival the passage from an agricultural to a manufacturing age in the nineteenth century. It is “an economy driven by information, research, knowledge, and technology,” Gore went on. “During these three months, I have seen the new economy unfold in mammoth airplane hangars in Seattle, in the sunny cornfields of Iowa, on the streets of inner-city Detroit, in the high-tech labs of New England, and even in the far reaches of cyberspace…. I have seen the new economy, and I am here to tell you that it works.”

To such enthusiasts, the numbers speak for themselves. Where only about 40 percent of businesses invested in computers in 1977, according to one sample, more than 80 percent do so today. One out of two American workers is now likely to use a computer, compared to one out of four in mid-1987.¹ In particular, the enthusiasts say the proliferation of computerized information technology, which makes unprecedentedly large amounts of information inexpensively and instantaneously accessible, should dramatically raise the productivity of most American workers. “The single greatest driver of growth and creator of wealth over the next two decades,” writes consultant Michael Moynihan in a recent book called The Coming American Renaissance, will be “the information superhighway.”

The normally sober Economist claims nothing less than that information technology represents “a change even more far-reaching than the harnessing of electrical power a century ago.” The Boston office of a US consulting company is discarding all of its paper-work and building a computerized Knowledge Exchange, “a vast on-line database containing the company’s accumulated wisdom, available…anywhere in the world seven days a week” on every worker’s PC. “Where once greater distance made communications progressively more expensive and complicated,” writes The Economist, “now distance is increasingly irrelevant.”² Business Week concludes that the computer is “a transcendent technology—like railroads in the 19th century and automobiles in the 20th…. Forget 2% real growth. We’re talking about 3%, or even 4%.”³

It all sounds plausible. But it may surprise many readers of today’s financial pages that, despite the rapid growth of the last year and a half, the expansion of the economy that began in the early 1990s has remained the slowest in the post-World War II period. The expansion hasn’t even reached the slow pace of growth of the 1970s and 1980s. Historical comparisons should be made over comparable business cycles, which include a recession as well as a recovery. As of the end of the third quarter of 1997, the average annual growth rate of real Gross Domestic Product (discounted for inflation) has been only 2.2 percent since the last business cycle peak in 1990.⁴ By contrast, the long business cycles of the slow-growing 1970s and 1980s produced average annual rates of growth of about 2.75 percent and 2.95 percent respectively.

Such differences between current and past performance have huge consequences. If there had been an annual rate of growth of 2.95 percent in the 1990s, the economy would have generated another $400 billion or so in goods and services in 1997 alone, and gradually lesser amounts in each of the preceding years. Similarly, by 1997, federal tax revenues would have been higher in each year, with perhaps $80 to $100 billion in additional federal tax revenues in 1997 alone, easily pushing the budget into surplus last year. In the 1950s and 1960s average annual rates of growth, of course, reached 3.5 percent and 4.5 percent.

There may be a computer revolution, but it has not yet transformed the economy to the extent the visionaries have promised, and in my view there are reasons to wonder whether it ever will. The enthusiasts surely underestimate just what it would take to return America to a fast rate of growth. Some argue, for example, that low inflation and interest rates and a subsiding federal deficit mean that the expansion is now so well-balanced that the economy can avoid a recession indefinitely. This overlooks the historically high levels of stock prices and consumer borrowing. Even so, unless real GDP grows significantly faster than 3 percent a year for quite a few years to come, the slow-growth era that began in the early 1970s will not have ended. If real GDP grows at an annual rate of 2.5 percent without a recession until 2010, which is what most economists think is the best possible rate of growth, the rate of growth between 1990 and 2010 will have been lower than it was between 1973 and 1990. The annual rate of growth over those years, which included the two most severe recessions of the post-World War II period, was 2.6 percent, nearly a full percentage point below the average annual rate of 3.5 percent since the Civil War.⁵

With the unemployment rate at a quarter-century low of 4.7 percent, how can this be? Labor productivity, or, roughly, real GDP per hour of work, is the heart of the matter. It is the source of any nation’s economic growth and rising standard of living. But the recent 2.25 percent rise in productivity is not nearly enough to raise productivity growth above the average of 1 percent growth per year since 1973. In fact, the growth of productivity in the 1990s expansion has been even weaker than during the slow-growth cycles of the 1970s and 1980s. By comparison, productivity grew at a rate of nearly 3 percent between 1948 and 1973 and more than 2 percent a year on average in the hundred years after the Civil War.

Nor is a 2.25 percent spurt in productivity growth during a single twelve-month period by any means unprecedented in the slow-growth era that began in 1973. Productivity rose faster in 1986, for example, which was also late in the cycle of expansion when productivity growth normally slows down. (The average annual growth rate over the two years, 1985 and 1986, was nearly 2 percent.) There were equally rapid rises in the late 1970s as well.

One major consequence of such slow productivity growth has been to keep wages and salaries from rising the way they once did. The one especially bright spot in employee wages—a 1.5 percent rise after inflation in the average wage of production and non-supervisory employees in 1997—is certainly welcome but so far not more than a minor blip in a long downward slide. Despite the recent rise, which is the result of the falling unemployment rate and a higher federal minimum wage, average hourly earnings, when discounted for inflation, remain about $1 an hour below the 1973 rate of nearly $9, and even remain below the high in the mid-1980s of more than $8 an hour. Broader measures of compensation, which include all workers and also take account of corporate benefits such as pensions and health insurance, are rising no more quickly in the current expansion of the economy than in the Seventies and Eighties.⁶

The fall in the rate of inflation to below 2 percent is also cited as evidence that a technological revolution is underway, making business so much more efficient that it need not raise prices. But some conventional economic explanations readily account for the low inflation rate. For one thing, much of the recent economic growth has been caused not by productivity gains but by new jobs and by the greater number of hours worked by those already employed. Workers are “coming out of the woodwork” as jobs become available, so the pressure to raise wages and therefore prices has been dampened. The work force seems to be both larger and more flexible than was previously realized, which is another way of saying that the unemployment rate was really higher than reported, and may still be. Two decades of surplus labor have also kept workers too docile to demand serious wage hikes. We are also all aware of what has happened to the power of organized labor, which once represented one third of all workers and now represents about one tenth of them.

In addition, prices of imported goods, which account for nearly 15 percent of what Americans buy, have been falling for several years, and are currently well below their 1992 level. Just as important but rarely commented on is that there may be an oversupply of services in the US. Consider the large numbers of health care and financial institutions, in addition to retail outlets such as coffee shops and department stores. This highly competitive environment keeps companies from raising prices rapidly. The problem for future rates of growth is that none of these conditions is likely to be permanent. The unemployment rate cannot fall indefinitely, for example. This is why few economists believe that a rate of growth of more than 2.5 percent or slightly higher at best can be sustained indefinitely.⁷

2.

Nevertheless, the sudden upturn in productivity in 1996 and 1997, coming so late in an economic expansion, is worthy of attention. It is certainly reason to examine more closely what the supporters of the new-economy thesis are saying. Most of these analyses are limited to short, largely anecdotal newspaper and magazine pieces. But a few books treat the subject at length. Michael Moynihan’s The Coming American Renaissance is designed for easy reading by businessmen, but when he avoids the hyperbole so common to such books he makes some worthwhile points. The most important of these, I think, is that America’s market size and mobile labor and capital markets put it in a better position than most other economies of the world. One important reason America grew more rapidly than did its overseas competitors since the mid-nineteenth century was the continent-wide, single-currency market that could support unrivaled economies of scale and efficient domestic trade. Despite the European common market and the growth of Asian economies, this remains an American advantage. In addition, there are still regions in the US, such as the Southwest, where costs are relatively low. These regions provide places for business to expand economically within the country; and such possibilities are typically not as available to the chief competitors in Europe and Japan.

But these factors alone will not transform the American economy. Moynihan believes the “information superhighway” will stimulate rapid growth again in America, and here Moynihan only repeats what so many others have said before him, and for so long. Electronic communications will enable business to relocate to some of the low-cost geographical regions he mentions. The availability of inexpensive information will give more people with entrepreneurial ideas access to commercial markets. The Internet will become a vast commercial market that is more efficient than the old-fashioned department store or shopping mall. “Networks multiply knowledge and power as though a man in the industrial age has access to all the mills and turbines and cotton gins in the world,” writes Moynihan.

Such statements need closer analysis than they are usually given. If everyone has access, then why is it more valuable to one person than another? Only so many new ideas will prevail in the market, no matter the access to information. And all such assertions beg an important question. Why hasn’t productivity risen already in response to these changes? They have, after all, been going on for at least a decade, and often longer. Business has increased its investment in computers by more than 30 percent a year since the early 1970s, but the rate of growth of productivity has fallen from 2.85 percent a year between 1947 and 1973, to about 1.1 percent a year since 1973.

Frances Cairncross, the author of the Economist article cited earlier, readily admits in The Death of Distance that there has been no major payoff as yet from the revolution she foresees. “For the moment,” she writes, for example,

on-line commerce is tiny. Many companies say that thousands of people look at their wares every day but few buy. Even companies that have succeeded with on-line retail marketing are still mainly minnows.

In fact, Cairncross writes, the growth of the Internet is already slowing down. Nevertheless, Cairncross believes that the falling cost of telephone calls, the hundreds of potential choices of cable-TV channels, and the Internet itself have made communications so much cheaper and more efficient that productivity will ultimately rise rapidly as a result.

Cairncross says we are too impatient about the impact of computers; she claims it even took the automobile a long time to affect the national economy. One of the tantalizing possibilities for the future, she says, is making access to the Internet as easy as turning on the television set. Dozens of companies are working on developing such capacities. For example, Compaq Computer, Intel, and Microsoft announced a joint venture in January in which the Internet, presumably tied to a TV monitor, would be available over normal telephone service.

This would probably raise the impact of information technology, but by how much we have no idea. Whether such innovations will literally transform the economy is merely speculative. Cairncross offers no concrete way of analyzing such possibilities, nor do most of the other new-economy enthusiasts. And doubts come quickly to mind, prodded by the largely barren commercial results of the Internet so far. People will have only so much time to make use of the multitude of new services. How will they watch all those new TV channels and surf the Net at the same time? Will at-home computer shopping really keep people from the less efficient malls, which are after all the town centers of our time, where people mingle and meet and observe one another, not to mention the appeal of the physical goods themselves? As Cairncross herself notes, consumers spent only about $43 billion on mail-order shopping last year in the US, compared to $2.4 trillion at retail stores.

For the most part, such books as Moynihan’s and Cairncross’s are not so much hardheaded analyses as expressions of faith, almost religious in nature, which often sarcastically chide those who resist the messages of true believers. But some academic scholars argue that the impact of computers will simply need a longer time to take effect. In a widely cited paper, the economic historian Paul David compares the proliferation of computers to the evolution of electricity.⁸ Such radically new technologies, he says, spread only slowly over several decades as people learn how to use them effectively and as resistance to discarding older, costly technologies diminishes.

David ignores, I think, how many other factors contributed to the value of electricity over time, from the new assembly-line mass production methods to highly useful home appliances such as washing machines, phonographs, and radios. Even the railroads were not alone in raising America’s productivity in the nineteenth century, but they did so in conjunction with many other forces that were revolutionizing industry. The Nobel laureate Robert Fogel claims that in a period of great economic vitality other forms of transport, such as a network of canals and rivers, may have made almost as great a contribution to growth as railroads. Moreover, the many allusions to economic development in the nineteenth and early twentieth centuries usually ignore the fact that productivity rose more rapidly over this period when the transformation was allegedly taking so long to permeate the nation than it has risen for the last twenty-five years.

In view of the claims now being made for the computer, it is worth recalling the arguments of Alvin Toffler’s 1980 book The Third Wave, which in many ways is the model for today’s new-economy optimism.⁹ Toffler has considerable talent for spotting a trend. He correctly claimed in 1980 that the industrial era would be replaced by a “de-massified” era, in which traditional mass production would be pushed aside by a more flexible kind of production that targets smaller “niche” markets. He also foresaw the advent of personal computers and the potential importance of genetics in improving agricultural and other products, though he has so far been wrong about the rewards from outer space and undersea exploration. “The dawn of this new civilization is the single most explosive fact of our lifetimes,” he wrote in The Third Wave, claiming that the end of the standardized age would radically change society. “It is an event as profound as that First Wave of change unleashed ten thousand years ago by the invention of agriculture, or the earthshaking Second Wave of change touched off by the industrial revolution.”

What Toffler did not account for was the slowdown in productivity growth already underway when his book was published, and the resulting pressure on the federal budget as well as the American standard of living. In a more recent book, Creating a New Civilization, written with his wife Heidi, and including a foreword by Newt Gingrich, he again betrays little awareness of the slowing in rates of economic and productivity growth. Books like The Coming American Renaissance, The Death of Distance, and The Third Wave uncritically assume that progress is linear. Impressed with change for its own sake, they typically presume that any technical innovation today will have the equivalent impact on the economy as did innovation in the past.

Cairncross, for example, is much taken with how information technology has made distance inconsequential in communications. But what about the steam engine, which reduced transoceanic and transcontinental travel from a few weeks to a week or less in the nineteenth century? Or the telegraph, to cite just one other of many such inventions? In the 1840s, it took ten days to send a one-page message from New York to Chicago. In the 1850s, after the telegraph was invented, it took a few minutes. “The next quarter century will see the fastest technological change the world has ever known,” writes Cairncross, but she presents no serious evidence for such a sweeping pronouncement.

3.

Why haven’t computers dramatically augmented productivity? Some analysts believe that the enormous power of computers is often superfluous. For example, law briefs are now much thicker and more detailed than they once were, thanks to word processing and the greater ease in obtaining supporting documents. But this does not mean that law is being practiced any more efficiently; perhaps just the opposite is true. Communications may now be more frequent among employees on a computer network’s e-mail, but it may be largely extraneous. Research into a variety of sources is now easier thanks to the Internet but word processors have not made novels or news reports any better; nor have faster calculations made economic forecasting more accurate. One can wonder whether all those urgent cellular phone calls from restaurants and cars have made business more efficient. The cheapness of telephone calls from anywhere may even keep people from doing something more productive, just as the distractions of television may keep people from seriously informing themselves.

Another cause for the disappointing contribution of computers to productivity may be that business managers don’t fully understand how costly computerization really is. Some observers contend that managers often underestimate the costs and downtime involved in educating employees, managing more complex technologies, and constantly fixing and updating them. Therefore, they overestimate the productivity of their computer investments and may invest too much in them.¹⁰

In contrast to the fuzzy optimism of much recent discussion, Daniel Sichel’s The Computer Revolution tries to define the upper limits of the potential contribution of computers to economic growth, and in doing so he sets a standard for rational discussion that has largely been missing to date.

Sichel was formerly with the Brookings Institution and is now on the economics staff of the Federal Reserve in Washington. In order to estimate the potential contribution of computers to US economic growth and productivity, he calculates how much investment has been made in computers and what rate of return such investments earned. The logic behind such an assumption is fairly straightforward. If computers are valuable in the economy, especially as computer power has become inexpensive, business will increasingly invest in them. The share of the nation’s stock of capital invested in computers, peripheral equipment, and software is a guide to just how much such high technology contributes to the economy’s output—that is, its growth and productivity. The economist Robert Solow, a Nobel laureate, has written recently that Sichel “tells the best story you are likely to hear” about what many now call the productivity paradox.¹¹ Solow himself defined this paradox when he wrote several years ago that “you can see the computer age everywhere but in the productivity statistics.”

The first issue Sichel raises is how to determine the size of the returns on these computer investments. Precise estimates of the returns on the millions of computers bought by businesses are obviously difficult to make. But many economists believe the returns on computers could not remain significantly above the average for all other investments for very long. If returns were above average, business would invest in more computers until the entrepreneurial opportunities are sufficiently exploited to bring returns down near the norm for most other investments. Ignoring these excess returns would be, many economists say, like leaving money on the table.

On the other hand, some economists maintain that returns on computer investments may be persistently above the norm because opportunities in the computer field are expanding so much faster than the ability of businesses to keep up. Some studies of computer investments among individual firms, though subject to the limitations of such selective samples, suggest that the returns have remained above average.¹²

Sichel has estimated the likely contribution of computers to the economy’s productivity under both hypotheses. Purchases of computers during the last few years come to nearly 20 percent of capital investment each year. But according to Commerce Department statistics, the nation’s accumulated investment in computers and peripheral equipment comes to only about 2 percent of all the existing investment in the factories, machines, and other equipment that make up the nation’s total capital stock.¹³ Business investment in software is not counted in the GDP; and software is one of the fastest-growing industries in the nation. Sichel finds that if it were included in GDP, it would produce a one-time boost of about 0.75 percent above its current level. But it would add only about 0.1 percent a year to the rate of GDP growth, no more than about $7 billion at current levels.

The costs of computer-related labor must also be included in any calculation of the contribution of computers to the economy because such costs are an additional investment. Adding together the investment in computers and his estimates of investment in software and labor, he finds that if computer investment earns only normal returns, that investment has contributed no more than 0.31 percent a year to the nation’s growth. Computers, moreover, are outmoded quickly, so taking into account depreciation (the charge for the wearing out of an investment) reduces the contribution further. With depreciation, Sichel figures that the contribution of computers to growth is no more than 0.15 percent a year. This amounts to only a small share of the average 2.5 percent annual growth each year over the past quarter-century.

When Sichel goes on to make estimates on the assumption that computer returns are unusually high, he finds that such returns are still not sufficient to transform the economy. If returns are as abnormally high as some studies have suggested (and Sichel presents a strong argument why they are probably not), he calculates that the contribution of the computer revolution to growth is no more than 0.61 percent a year, and 0.51 percent a year after depreciation. This is significant, but even this generous estimate is not adequate to put the economy back on its former fast track.¹⁴

Sichel also casts much-needed light on the claims made by many, including Moynihan and Cairncross, who say the government is inaccurately measuring the pace of economic growth and productivity advances, particularly because it has overestimated the increases in the Consumer Price Index. If the government’s estimate of the rise in the Consumer Price Index is overstated, the estimate of the annual value of the nation’s goods and services has been reduced too much for inflation. Both real economic growth and therefore productivity will have been stronger than has been reported. But even members of the Boskin Commission, the congressionally appointed committee that claimed last year that the CPI is overstated by 1.1 percent a year (which this writer has argued is exaggerated), concede that inflation was about as overstated before as after 1973. So, even if it is adjusted for the CPI, the shortfall in productivity growth would have remained as great.¹⁵

Another argument of such writers as Cairncross, one partly related to the CPI, is that the productivity of some of the newer services of the 1980s and 1990s, such as medical procedures, are particularly difficult to measure. Thus they claim that the rate of productivity is being understated today. But Sichel calculated the proportion of these hard-to-measure service industries in the economy, and he takes account of the degree to which some economists think they may be mismeasured. He finds that even if we accept the mismeasurement hypothesis at face value, overall productivity would be understated by only 0.25 percent a year.

It is likely that information technology yields what economists call “social” returns—i.e., financial advantages that accrue even to those businesses which don’t invest in computers.¹⁶ If the Internet someday replaces a large proportion of retail shopping, for example, this might yield large social returns for makers of consumer products. Some argue that the existence of the computer stimulates innovation as well. But it is unlikely that such social returns are nearly as high as they are for, say, corporate investment in research and development, which may regularly produce new ideas for many other industries. What Sichel’s analysis reveals, nevertheless, is that such social returns would have to be very large to make an impact. When railroads were America’s supposedly “transcendent” technology, they made up 15 percent to 18 percent of the nation’s capital stock, far more than the 3 percent or 4 percent of capital stock now made up by computers and software. Sichel’s may not be the last word. Social returns may yet surprise us and more precise methods of measuring the size of investments in computers will probably be found. But any serious discussion of the contribution of computers to the economy must take account of his analysis.

4.

What is holding US productivity back? Economists have come up with no generally accepted answers. But one question that should be more deeply explored, and may have a part in the productivity slowdown, is the actual impact of the great variety of products so widely praised today. We should keep in mind that by standardizing products, the mass production and distribution revolution that has brought about dramatic productivity growth since the nineteenth century rapidly also removed human workers from the production and distribution processes. The assembly line, new mass production machinery, distribution outlets like department stores, and steamboats, railroads, trucks, and planes all meant that far fewer hours of human labor were needed to make, sell, or transport a product. Consumers who had owned almost nothing now eagerly bought more efficiently produced, if standardized, goods and services. Productivity grew as fewer hours of work were required to perform a given task.

This did not reduce the demand for workers in the overall economy, as was feared then and is feared again today. Rising productivity increased profits, and the nation was able to invest in new industries, businesses, products, and ideas. It also enabled business to pay higher wages to the workers who were retained; and they in turn could spend their money on still more goods, such as chewing gum, cigarettes, and bottled beer, among countless new products, and on the services of highly efficient railroads and retail outlets like supermarkets. These businesses would then hire new workers.

Today’s consumers, who by the 1960s or 1970s owned most basic goods, demand variety, quality, and constant novelty. Businesses tailor products to smaller, less homogenous markets, a practice which they call customization. The competition to meet these needs in the US and from equally competitive and often lower-wage companies abroad is ever more intense. The number of new food and household products introduced each year, for example, has increased fifteen- and twentyfold since 1970. The Gap retail chain revamps its product line every six weeks, and changes its advertising frequently as well. Nike introduces a new sneaker model every six weeks. Several hundred basic car models are now available; and the number of mutual funds and insurance products has vastly increased. Cairncross writes that even “tangible goods have more and more knowledge imbedded in them: a washing machine, for example, might include software that calculates the correct water temperature; a videocassette recorder might include software to control the timer or other functions. Intangible inputs now account for 70 percent of the value of an automobile.”

All this requires greater use of the one characteristic that machines cannot replace: human imagination. The modern economy, I would argue, may be returning to a high-technology version of a crafts economy, based on worker skills, thinking, and inventiveness, rather than on the muscle of large-scale factories and distribution networks. In an information age created by computerization, having information—for example, about what products customers are buying daily, or even hourly—is no longer as valuable because so many others have it as well. What becomes increasingly valuable is what is done with that information. What new products, marketing campaigns, or store layouts, for example, will improve sales? New technological and managerial methods help, but the new economy may simply not be able to remove human beings from the production process as rapidly as the old standardized economy of the mass production age. Productivity can still rise, but it is worth exploring whether the growing need for human imagination might inevitably hold back its rate of growth compared to former times.

A new study by two economists at the Educational Testing Service, Anthony Carnevale and Stephen Rose, provides support for this possibility. The report, called “Education for What? The New Office Economy,” reclassifies government data on jobs according to function rather than industry, in order to find where the demand for new workers has come from since the 1960s. The surprise is that the net growth in demand has not been for workers in science and what we customarily think of as technology. Rising demand for computer workers has only just compensated for falling demand in other sectors of the economy.

The principal new demand, it turns out, is for people who work in offices. This group once made up 30 percent of all workers; now it is 40 percent. Moreover, a greater proportion of office workers are professionals today, and a lesser proportion are support staff like secretaries. More than half of these office workers also now have at least a bachelor’s degree. They include managers and executives, communications and financial experts, marketers, business consultants, lawyers, high-level sales people such as real estate and stock brokers, and scientists and engineers who work in management rather than in the classroom or the lab.

All this has occurred at a time when business has vowed to reduce its bureaucracy and eliminate superfluous layers of management. Nevertheless, these office workers are apparently needed—or thought to be needed. Just what functions they serve remains a good question. But one of these functions would seem to be to supply the ideas that, in Toffler’s fragmented, innovative, “de-massified” modern economy, are in ever greater demand. More often than the widely discussed technicians, the more educated office workers appear to be the source of business innovations, which include an increasingly rapid flow of new products and services, ad campaigns, marketing concepts, financing techniques, and managerial reforms. Of the 52 million office workers, nearly 10 million are now managers, another 4 million are management professionals such as accountants, and another 4 million are sales people and brokers in real estate, finance, and insurance. Most of these workers have some technical skills, such as using computers, but it is not technology that defines their jobs. About 1.5 million are science-based professionals, such as engineers, architects, and chemists, who also work in an office. Between 1979 and 1995, office work accounted for 59 percent of all new jobs, a much higher proportion than at any time in the past. “In the contemporary era,” Carnevale and Rose write,

cheap standardized products have been replaced by more customized and style sensitive products…. Heightened competition has required that firms be much more agile in terms of marketing and positioning their product…. Furthermore, as society has gotten richer, it has had to allocate a higher percentage of the labor force simply to managing its physical and monetary resources.

There may be other explanations for the prominence of office workers in the new economy. Cronyism may be part of it; decision makers may tend simply to hire more people like themselves.¹⁷ A more complex economy may also require more managerial workers to coordinate it. The causes of the rise of the office economy described by Carnevale and Rose certainly need further explanation, and the category itself includes so many different jobs and skills that it needs further analysis. But the authors make a suggestive case for the view that, more than ever before, office workers are needed for innovation.

What may lend some comfort to the promoters of the “new economy” is that Carnevale and Rose’s findings confirm that well-paid, relatively satisfying jobs are being created in great numbers, even in this slow-growth environment. Carnevale and Rose find that 33 percent of workers between thirty and fifty-nine now hold “elite” jobs, compared to only 28 percent in 1979. The average pay for these jobs has risen over these years to $47,000, though it has fallen slightly in recent years.

On the other hand, the report also provides evidence of a growing gap between America’s two tiers of workers. While a higher proportion than before are doing well, the remaining two thirds of US workers have either middle-income jobs, which pay on average $29,000 a year, or low-income jobs, which pay on average $19,000 a year. By contrast to the elite workers, the average income for both of these groups has fallen significantly since 1979. Carnevale and Rose provide us, I think, perhaps the most telling portrait yet of a divided America in the 1990s.

The study by Carnevale and Rose also raises questions about the type of education the economy now demands. The office economy may require communication skills, social ease, and basic reasoning abilities as much as, if not more than, technological expertise. But acquiring these skills today, however general they may seem, may only be possible through higher education, where students are exposed to a sophisticated culture, a variety of experiences, and varying disciplines that require analysis of facts and concepts. This study further emphasizes the need both to improve such education and to make it more widely accessible.

The US remains a complacent country. Little new money is being spent on education or on several other kinds of investment that will improve productivity over the next decade, whether in physical infrastructure, day care, or basic research. Self-congratulation about the current state of the economy is distracting the public and politicians from addressing what remains to be done to raise America’s rate of growth. As of today, it looks highly unlikely that the computer revolution alone will accomplish what is necessary.