We have long supposed and have been frequently told that America's economic troubles stem from our declining productivity growth. Yet the basis for this belief is now eroding. Early in 1991 we learned that during the 1980s our manufacturing productivity grew at 3.6 per cent per year, an "unbelievable revival" in the words of The New York Times. And this revival is as mysterious, as opaque to economic science, as the decline that it seems to have superseded. If the gloomy consensus is now giving way, it is certainly not to optimism, but to confusion.
Most economists had accepted the fact of productivity decline without ever having settled the question of how and why it happened. Around the presumed fact of decline, and despite the acknowledged dissensus as to its cause, liberal economists in particular built a core of policy doctrine. They now must face the question whether that policy doctrine stands independent of the changing evidence about productivity or falls victim to it.
Moreover, if productivity decline does not explain our economic problems, what better explanation is there? I will argue that we can find much of the explanation in an alternative picture of the American economy that brings into sharp resolution the different paths taken by American industries in response to the challenge of international trade.
Most economists believe that economic growth and prosperity depend fundamentally on productivity. Productivity, by definition, is the ratio of physical outputs -- the goods and services that make up the gross national product -- to the physical inputs, the land, labor, and capital goods used in production. Usually economists simplify the computation by focusing on "labor productivity" -- the ratio of total outputs to the total hours of work. Productivity growth as conventionally measured is the change in that ratio from year to year.
When labor productivity growth slows down, as it apparently did after 1973, the cause may be a decline in the ability or effort of workers. Or it may be equally a decline in the quality of the machines, or the land, or the fuels with which labor works, so that less output is forthcoming for a given labor input. But the cause must, by mathematical definition, reside in one of these two sources. For researchers, therefore, the problem is to identify the specific culprits and to measure each one's contribution. And for policy makers, the problem is equally well-determined. Just as a decline in factor quality must be the cause, improvements in factor quality must be the cure.
Until quite recently, neither party took this simple mathematics seriously. Reagan Republicans talked about productivity, but emphasized cures -- from Kemp-Roth to capital gains -- that always coincided with the tax agendas of their rich friends. Claims that these policies work accompany arguments that we need more tax cuts for the same clientele. And Brookings Democrats, until recently the dominant strain, emphasized sound budget policies and higher rates of savings and investment, despite the lack of evidence connecting these instruments to the productivity slowdown.
Now, an emerging third view does emphasize the importance of factors of production: the quality of labor and capital, the mechanisms for bringing them together, the level and quality of research and development, and infrastructure outlays as well as the supply of savings. Those who hold this view are just as passionate about the key role of productivity as are their Reaganite or Brookings colleagues. As William Baumol and associates have written: 'Tor real economic miracles one must look to productivity growth. And economic miracles it has indeed provided." Or as Paul Krugman writes, "Productivity isn't everything, but in the long run it is almost everything." Since Krugman teaches at MIT and Baumol's new book, Productivity and American Leadership, is published by the MIT Press, I shall call this the "MIT View." Yet another MIT product associated with this view is Made in America, the widely publicized 1989 Report of the MIT Commission on Productivity.
The MIT view underpins a new political agenda of the broad liberal center. The conclusions of Governor Mario Cuomo's competitiveness commission were close to MIT's. To judge from George Bush's State of the Union address, which calls for "investment in America's future -- in children, education, infrastructure, space and high technology," the President even holds some of the same beliefs.
The MIT view offers an appealing combination of fiscal rectitude and social compassion. Balanced budgets are good things, it argues. But if we are prepared to tax ourselves, or cut the military, or cut "entitlements," we can and should spend more on education, roads, bridges, housing, and scientific research. So long as they do not come at the expense of private saving, these investments will work to cure our productivity deficit and to close the productivity growth gap between ourselves and the Germans, the Japanese, and the newly industrializing countries. Faster productivity growth will cure our competitiveness problem, restore our trade balance, end the sell-off of our national assets, and permit sustained real growth of, say, 3 percent per year or better without inflation. Not bad work -- if you can get it.
Eclectic, not especially ideological, based on a coherent and comprehensible argument, the MIT view offers refreshing relief from the stagnant passivity of the Brookings Democracy on policy matters. For these reasons, and others, it is now ascendant, especially in the Democratic Party. But is it right? I am skeptical, and this article suggests several reasons why policy makers should worry when economists come to a policy consensus about issues that they themselves admit they do not understand. I propose instead to look at distinct industrial sectors, and to find therein alternative ways of understanding productivity and arriving at policy remedies. I also want to hone an argument over the particular interventionist policies that the MIT view suggests.
Limits of the MTT View
The MIT view shares with the Brookings view the conviction that productivity growth has declined. Efforts to find the undetected declines in the quality of our inputs have been extensive -- to locate the absenteeism, alcoholism, sloth, accelerated or energy-induced obsolescence of machinery, to measure the paper entrepreneurialism, short-focused management, the collapse of innovation, the resources wasted in mergers, acquisitions, Liar's Poker, golden parachutes, poison pills and what-have-you. But they have not succeeded. Edward Denison, the most systematic and persistent researcher on productivity, wrote in 1979, "What happened, to be blunt, is a mystery." In 1990 Paul Krugman summed up the progress since then: "Why did it happen? And what can we do about it? The answer to both is the same: We don't know."
Agnosticism on this point, indeed, is now ingrained in the economics profession, to the point where many who were once fascinated by the productivity phenomenon have now given up hope that a coherent explanation will ever emerge. Even if some new and compelling explanation were now offered, it seems unlikely that it could be accepted. Orthodoxy itself would be at stake; too many high authorities in our discipline have decided, to their own satisfaction, that the problem cannot be solved. But might the observations themselves be wrong? There are enough weaknesses in the mainstream treatment of productivity growth to warrant at least raising the question.
To begin with, the MIT view stresses the long-run character of the productivity problem. Baumol begins his new book with a chapter entitled "Why the Long Run?" Krugman similarly opens his discussion with a figure showing the decline in productivity growth, decade by decade, since 1899 (see Figure 1).
The point of the emphasis on decade averages is, quite deliberately and without apology or even explanation, to call attention away from annual data, which might lead the naive observer to the believe that short-run phenomena -- the recession phases of the business cycle -- are really the source of the evil. Figure 2 illustrates how this impression might arise.
The figure shows that in the recession-free 1960s and from 1982 until the 1989 slowdown, output growth stayed comfortably ahead of the growth of labor hours. Is it possible that our problems in the 1970s and early 1980s were due to the four recessions between 1970 and 1982? Baumol, Krugman, and the economics establishment recoil from this possibility, but they do not confront it directly and do not offer, to my knowledge, any persuasive test that refutes it.
From a policy standpoint the short-run/ long-run issue is crucial. For if productivity declines are largely cyclical -- the result of increased fluctuations in economic output -- then the "supply-side" policy discussion of savings rates and literacy and even infrastructure fail as explanations. One might talk instead about the need for low and stable interest rates, and for an incomes policy to tie wage increases to real productivity growth, and of the "demand-side" effects of expanding public investment and the social welfare system, all of which were hallmarks of the policy that kept us free of recession from 1960 through 1970. And to be perfectly nonpartisan, one might also ask whether those Reagan deficits, which got us out of recession and productivity slump after 1982, were so bad after all. These are thoughts congenial to both left and right, but that the liberal mainstream, needless to say, does not like.
A second weakness turns up when the argument shifts away from the historical decline in U.S. productivity growth rates, to the low average rate of U.S. productivity growth in comparison with other industrial countries. Here the hard evidence, even from the best sources, takes a most uncomfortable turn.
Baumol and associates, using data compiled by Angus Maddison, make two discoveries. First, while U.S. productivity growth is assuredly low by world standards, it seems to have been low by world standards at least since 1880! One is entitled to wonder about the meaning of a statistic that speaks so ill of U.S. economic performance over the whole of the American century. Oddly enough, a low average rate of measured productivity growth does not necessarily entail a loss of trade competitiveness or technological leadership in world product markets. Baumol and associates find, to their surprise, that U.S. shares of the high technology markets -- those industries where productivity growth is crucial -- have been rising, not falling, for the last decade. All in all," they say, "that is still not too bad a record," though it may be "as surprising to the reader as it was to the authors." But this surprise could reflect a problem, not in productivity, but in the way we conventionally measure productivity.
Indeed, do we really know that productivity growth, the "rate of growth of output per unit of input," has declined? Or more precisely, do we correctly infer that real outputs have fallen off in relation to the real inputs used to produce them? Or did something else happen instead?
Correcting for Quality
Suppose -- just suppose -- that, instead of an undetected decline in the quality of our inputs, there was an unmeasured but resource-using improvement in the quality of our outputs. In that case, the mathematically measured result, declining productivity growth, would be exactly what, in fact, was observed. But its meaning and importance as a social phenomenon would be altogether different.
Consider several examples of unmeasured quality change. An airline is an industry whose standard product is the passenger-mile. Labor units per passenger-mile change only when the speed or size of the aircraft increases, or when the crew required for safe and successful operation declines. Are we then to say that there has been no productivity improvement in airlines since the introduction of the jumbo jet? The new airplanes are better than the old along a host of other dimensions that are not quite captured in the standard measure of output. New jets are cleaner, quieter, easier to navigate, and much more reliable. Passenger-miles per worker or any other standard measure of output volume do not reflect this change.
Measures of "total factor productivity" can capture improvements in fuel economy. To get at the other qualitative changes, one can compare the price of apparently equivalent aircraft, new and old, controlling for differences in measurable operating expenses. The remaining difference will require an economic valuation of quality improvements. But this type of comparison, while obvious after the fact, is far too complex to be carried out systematically in the construction of output measures in the national income accounts. And even total factor productivity is not measured reliably on an economy-wide basis.
Take an even more dramatic example. A computer today does exactly the same thing as a computer fifteen years ago. It performs mathematical computations. But the machines we call computers share little with their predecessors of fifteen years ago. Card readers? Tape punches? It feels strange below the age of forty to have memories of such dead technologies and to be working with "High Resolution Monitors" and "80386 processors" that did not exist at that time. No matter how one measures the "output of computers" or of "computers per worker," or even "value of computer sales deflated by the falling price of computers, per worker" one cannot hope to capture all of the changes that make the machines we work on today better than those of just a few years back.
The more one considers quality change, the more one realizes that it is the characteristic way in which technologies diffuse. New products are rare; old ones are continually being revised, A Teddy Ruxpin is neither a new nor a bigger product. It is not more teddy bear. It is a teddy bear with a cassette player in its back. If you add the capital and labor required to put a cassette player into a teddy bear's back, and count the output as one teddy bear, you will conclude that Teddy Ruxpins are extraordinarily low-productivity teddy bears. But you will be missing something that even your children can tell you about.
What applies to goods applies to services. Health care is a case par excellence of new services without comparable antecedents. Many newborns have grown to adulthood thanks to more intensive (and expensive) medical care that employs new technology and new human skill. In standard economics, this shows up as a decline in measured productivity rather than an improvement in unmeasured quality.
One might ask why, in searching the world for things that have not been detected, one should prefer to hunt for unmeasured output quality improvements over unmeasured input quality declines. One answer is that the formal search for the latter has been extensive but unsuccessful, while the formal search for the former, though not so extensive, is proving extraordinarily successful. In recent years there has been an outpouring of fruitful and important effort, which now includes many disparate studies and an important book by Robert Gordon, The Measurement of Durable Goods Prices. Unmeasured quality change is pervasive.
The Good Old Seventies
Most economists recognize that unmeasured quality improvements are a major distortion of our economic accounts. But did an increase in product quality cause a decrease in measured productivity growth during the benighted 1970s -- that terrible decade when Jimmy Carter was President and everything about our economy was going to hell?
Could be. The 1970s were a strange time, when many disconcerting changes took place. These included large changes in the composition of industrial output, particularly an explosive growth in computers and consumer electronics. U.S. trade shifted toward manufactured goods. And female labor force participation and employment grew sharply. All of these changes tended to generate unmeasured improvements in the quality of outputs.
First, the initial explosive growth of the computer and electronics industries in the 1970s led to an acceleration of quality improvement in both computers themselves and the industries where they and electronic components were used. None of this was measured at the time. According to estimates made well after the fact, adjusting output for quality in the computer sector alone would have considerably reduced the apparent fall in manufacturing productivity growth from 1972 through the end of the decade. This was an event without dear parallels either before or later.
Second, and more broadly, the recessions and changing patterns of trade in the 1970s shifted the composition of U.S. manufacturing output toward traded goods, where competition demands quality upgrading. Expanded, trade also increased technology transfer, as well as foreign direct investment. Recessions cleared out the least efficient enterprises in many manufacturing sectors, from textiles to automobiles to steel. The subsequent expansions and export booms replaced these enterprises with growing output in the advanced technology export-oriented sectors, where unmeasured quality change is more important Expansion also brought more rapid quality changes to the output of the traditional sectors. Steel nowadays has become a custom-made product, and steel firms compete on the basis of their ability to match a precise metallurgy to the customer's needs; these services do not show up in traditional measures of steel output.
The third factor, largely but not entirely outside the industrial sector, was the effect of the rise in female labor force participation after 1970. This rise added new service output to the national economic accounts, in branches of activity whose measured productivity levels are low. Thus the rising share of female workers lowered the growth rate of measured productivity overall. But this calculation neglects work that the women in question were doing before out of the paid labor force. And so it overlooks the productivity improvement that results from their leaving inefficient homework for more productive employment. Once again, we have real productivity gains but measured productivity slowdown.
All in all, the current mainstream discussion seems to have left hanging questions in at least three directions. First, it is unable entirely to lay to rest whether the decline in U.S. productivity growth is long- or short-run in nature, and thus whether it is attributable in the main to structural ("supply-side") or cyclical ("demand-side") factors. Second, it is unable to link a record of low relative productivity growth at the national level to poor technological performance of industries. And third, it has not, at least not yet, responded to the possibility that the real change of the last twenty or so years was not an unmeasured decline in the quality of inputs, but rather an unmeasured improvement in the quality of outputs.
For these reasons, the self-confident policy assertions of the new liberal center appear to lack any very firm logical foundation. For if we do not really know whether the critical issues lie on the side of supply or demand, how can we confidently assert that increasing investments in the supply side (education, infrastructure, or for that matter such right-wing variations as cuts in the capital gains tax rate) will have a productivity effect? And if we do not know how industrial performance is linked to national average productivity growth, how can we assert that raising private capital formation or national research and development efforts will cure our comparative productivity problem? Indeed, how do we know that "productivity crisis" is the appropriate point from which to begin a discussion of American economic leadership and of the economic policy questions that we face?
An Alternative Picture
Productivity crisis or no, America's economic problems are real. We have seen the collapse of certain heavy industries (steel), severe erosion in others (autos), a sharp decline in high-wage, blue-collar employment, a flood of low-wage imports, and a decline in the trade balance at high employment, an increase in wage and income inequality, rising poverty, and a deepening of fundamental social problems -- all within the past decade. If the productivity lens fails to focus clearly on these problems, assuredly we must find something else.
Suppose our goal is a large and growing number of jobs that pay high wages. Suppose further that one could show that high-wage jobs are to be found mainly -- and not by accident -- in industries that do well in international trade. Then, if we examine groups of industries for characteristics that lead to success in trade, we might be able to identify the main ingredients of a high-wage strategy for the American manufacturing economy as a whole. In this deceptively simple way, we might sidestep altogether the wreckage of traditional productivity analysis. Provided, of course, that the critical linkages, between industry traits and trade performances, and between trade and wages, actually exist.
My colleague and co-author Paulo Du Pin Calmon and I have sought to make some sense out of the structure of American industrial wages. We have found that one can learn a great deal about industrial performance from wage data, for the simple reason that industries which are doing well tend to share their good fortune with their workers, while industries that are doing poorly do not. By analyzing wage patterns, we classified the whole structure of American industry into six distinct groups (see Figure 3), plus a dozen special cases or "outliers," each of which has its own distinct pattern of wage change. The trade performance of these different groups, taken as a whole, is very distinctive and covers the full range of American industry.
The United States has best held its ground in what we call the "Advanced Technology" cluster, which actually combines industries with three distinctive characteristics: land-intensity (food products); energy-intensity (chemicals, soaps, paints); and reliance on labor skills and technological advantage (aircraft, communications, Pharmaceuticals, special industrial machinery). These industries have benefited decisively over the years from the position of the United States as a country of cheap land, low energy taxes, and above all preeminent science, technology, and advanced training.
Not so successful is the second group, "Heavy Industry," the highest-wage of all the clusters. This grouping includes automobiles, its satellites (engines, glass), and several industries whose wages have been linked to the automotive sector for historical reasons, including construction equipment and sawmill products. While wages are high in this sector, they are dwarfed by the costs of capital and materials.
Until the 1980s the U.S. showed a comparatively strong trade performance in Heavy Industry because rapidly growing exports of engines, equipment, and lumber products, especially to the Third World, offset the imbalance in automotive trade. After 1983, however, the trade performance of this group of industries declined precipitously, reflecting both the collapse of exports to developing regions and a rapidly increasing foreign share in the U.S. automobile market.
Next there is a group of industries we call "General Industry," which includes a variety of metal trades, household appliances, printing and publishing, and general industrial machinery. These industries share a comparatively strong, though declining, trade position, and a comparatively high, though declining, position in the wage structure. They epitomize the core of U.S. industries that do not enjoy strong comparative advantages or government support for their markets, but that have nevertheless not yet been overrun by imports.
We have lost ground most conspicuously in industries that are labor- and materials-intensive. These include garment trades, where imports now overwhelm exports, and cotton textiles, whose technological and market renewal has distinguished its wage performance from that of the garment trades -- though not by much. They also include two types of "Light Industry": the production of personal consumers' durables, such as watches, sporting goods, toys, jewelry, and electronic components, and construction-related products, such as fabricated metals and ceramics. That we have lost competitive ground in these areas can hardly be surprising, since cheap labor and cheaper materials (as opposed to cheap energy, land, and knowledge) are the hallmarks of production in the developing regions.
Beyond these six groups of industries lie a number of distinct special cases. These are characterized by industry-specific historical events (oil, steel) or by uniquely strong trade performance (cigarettes), or by an industrial transformation that changed the nature of the work force and through those compositional changes the average wage rate (the explosion of computers, the decline of beef).
It is, or has been until very recently, a widely held view among economists that the internal industrial wage structure has remained stable in the United States over a long period of time, despite the large changes that have occurred in our pattern of trade. Figure 4 and 5 show that this is not the case. When industries are grouped according to the pattern of the wage changes that have occurred, the data reveal a very substantial realignment of average industrial wage rates. Wages in the advanced technology group began the period just slightly over the average, and have been rising relative to that average ever since. Wages in heavy industry rose relative to average industrial wages in the mid- to late 1970s, and then began to slide through the 1980s. Wages in the general industry, light industry, and garment sectors have been falling, while among relatively low-wage groups cotton textiles is distinguished by a modest recovery of its wage position after the mid 1970s.
What has brought about this realignment? Before 1970, industrial trade performance mattered little to wages. But since the floodgates of trade in manufactured products swung open in 1970, U.S. wages have come to depend on whether the industries in the relevant wage group are faring well or poorly in the trade contest. Now, it appears, the effects of trade performance on wage rates are nearly pervasive. Indeed, outside of heavy industry one might say that patterns of trade now determine the wage rate, while the resistance even of heavy industry to this rule is gradually eroding. In general, as Figure 6 makes very clear, export success means high and rising wages.
Lessons for Policy
The alternative picture of the economy I have sketched has implications for policy that contrast sharply with established opinion. The MIT view, which is based on the idea that there have been unmeasured declines in the quality of our inputs, says that to improve quality we need more and better capital, and better if not more labor.
- To get more capital investment, according to the MIT view, we need more savings. Thus the emphasis on savings incentives and on reduction of government dissaving -- the budget deficit. To get better capital investment, we need more expenditure on research and development. And perhaps fewer resources should be devoted to the satisfaction of environmental, safety, and other regulatory goals that use capital and impose burdens on business. To be fair, the liberal mainstream (or "MIT") position on regulation is not monolithic. Many who agree in general that regulations impose "burdens" also believe that the social benefits often exceed the economic costs. And some recent authors, such as Michael Porter of the Harvard Business School whose 1990 book, The Economic Competitiveness of Nations, comes with a full complement of MIT endorsements, have caught on to the pro-competitive effect of regulatory standards.
- To get better labor inputs, in the MIT view, we need better education, particularly of our minority populations. And our labor productivity can also be enhanced by complementary investments in public capital, such as roads, rail, ports, and water systems.
While some of these remedies are sensible, the underlying intellectual construct, the emphasis on defective factors, is misleading. Our alternative proposals rest on an entirely different set of principles and place their priorities in a strikingly different way. Indeed, these proposals sweep away the false division between "macro-only" and structural policies. They suggest instead that policy must both promote rapid growth of the economy and actively support U.S. industrial success in world trade.
Trade Success We have observed that high and rising wages now depend on relative success in world markets. The advanced technology users in American industry do well in world trade and enjoy rising relative wages, when market conditions are right. Policy, therefore, should concern itself with the health of those markets.
First and perhaps foremost, given that most trade is between advanced countries, we must be concerned with assuring that Europe and Japan pursue pro-growth policies and that goods produced in the U.S., particularly food, chemicals, communications equipment, and aircraft, remain competitive in those markets. In the late 1980s some argued that U.S. exports could not expand because capacity was lacking and our advanced exporters had become uncompetitive. The export boom that followed dollar depreciation in 1989 and 1990 emphatically falsified this view. Now we must be concerned that dollar appreciation not wipe out the gains that have been made, and that slow-growth policies emanating from Frankfurt and Tokyo not undermine the recovery of our advanced exporting sectors. It is also obvious that U.S. efforts to lower the protectionist barriers of other countries in agriculture and high technology are not misdirected.
Growth and Development. Second, we need to take responsibility on a global basis for achieving adequate growth rates and stable development in the Third World. These measures would most strongly benefit producers of construction machinery, engines, and heavy vehicles that are the basic tools of development and that we once were able to export with great success to developing countries.
A resolution of the Latin American and East European debt crisis remains a pressing priority. And we urgently need a large-scale program for the reconstruction of Eastern Europe and the Soviet Union. As is by now dear, the free market alone cannot prevent political and social disintegration in the postcommunist world. Only a far-sighted and massive effort by the West seems remotely capable of achieving results on the scale required.
Regulation. Agreeing, to some extent, that government interference imposes burdens on industry, the MIT view shares the traditional conservative position that deregulation might help industries adjust to technology-based competition. Our re search suggests the opposite. Regulation and government involvement seem often to serve as standard-setting, reputation-building, and technology-forcing mechanisms. The advanced industries with the best wage and trade performance -- aerospace, electronics, food, and chemicals -- all have dose links to government, something that cannot be said of toys, sporting goods, watches, leather tanning, and garments. Cotton textile production is a case where the imposition of regulation, to clear the factory air of the dust responsible for brown lung disease, seems to have had a role in forcing recapitalization of the industry and an improvement in both its wage and trade performance. Automobiles are an opposite case. There successful American corporate resistance to government fuel efficiency, safety, and pollution standards may have played a role in the failure of the automotive industry to maintain its competitive position.
Industrial Organization. The MIT view calls for improvements in education and provides an economic rationale for educational aid to minority populations who have traditionally suffered from educational disadvantage. Increased spending on education is desirable for social reasons, but it may not have much effect on the competitiveness of the country as a whole. As an examination of the industries where the U.S. is most successful reveals, the well-springs of competitive success simply are not to be found in the public schools. Rather, they lie in blunter and more sweeping instruments: development of agriculture and energy resources, technology-forcing mechanisms, and wholesale subsidy of industrial research and development.
In the industries where the U.S. remains highly competitive, moreover, there is just no evidence of a labor problem. Indeed, we do best at training (or attracting) the highly skilled, creative, highly motivated talent that advanced technological systems require. We are not a bit short of aeronautical engineers, software designers, petroleum geologists, or top-flight machinists and other highly skilled technicians. It is possible that military cuts may eventually retard technical dynamism in aerospace, communications, and other sectors that have relied on the military-industrial policies of the Cold War. A remedy is to create and expand civilian projects -- in environmental sciences, medical research, transport, space, and other areas -- to assure continued strong public support for projects that link advanced technology to advanced manpower development. A civilian version of the Defense Advanced Research Projects Administration (DAR-PA), most recently and comprehensively proposed in legislation by Senators Jeff Bingaman, Ernest Hollings, and Sam Nunn has long been urged by technology policy advocates for precisely this reason.
Labor Markets and Communities. Outside of the industries where high technology, cheap energy, and abundant land confer comparative advantage, the U.S. needs to consider what alternative measures might be most effective. Can labor-enhancing policies really save our labor-intensive sectors by increasing the productivity of human capital?
Quite probably not. For we are not losing in the labor-intensive sectors to countries whose education systems compare well with ours. We are losing in toys and garments not to Japan, not even to Korea, but to Malaysia, Thailand, Brazil, the Dominican Republic. The critical variable here is, of course, not education and skills, but differences in wage levels that are as high as ten to one. Granted, U.S. educational standards in the overmatched sectors may be deplorable. Still, there is no strong reason to believe that better education in the U.S. can do much to offset the wage gap and no compelling reason to subordinate education policy to the pursuit of economical wills-o'-the-wisp.
Of course, industrial transitions will continue to be the order of the day, and the more rapid the better from the standpoint of the U.S. economic role in the world. But industrial transitions are hard on people, and these transitions will hit hardest at industries like garments and labor-intensive consumer products manufacturing whose employment is disproportionately female, low-wage, poorly educated, and immobile. The main need is for measures that make industrial transitions smoother for individuals and communities.
Adjustment assistance, long gone but not forgotten, is one possibility: direct payments to workers whose jobs are lost to trade. But while the income subsidies are useful, the cost of transition is not just lost incomes. It is rather in the decline of communities, the lost value of housing, the decay of infrastructure, the forced migrations that follow. The most urgent measures should aim to stabilize communities faced with economic transition, to maintain roads and schools, and to provide for new investment outreach so that our cities and towns can be born again. This is not a new problem. Indeed, the U.S. has a lot of experience, much of it successful, with this sort of redevelopment in the transformations of our Lowells and Pittsburghs. What is required is an orderly, and permanent, commitment to provide the resources.
Economic development in the world today is, broadly speaking, a polarizing force. The migration of basic manufacturing activities inevitably means sacrificing many middle-level jobs in favor of expansion at both the high end and the low. We cannot stop the processes of world development. But can we not reshape the social order that such patterns of development produce? This is the final, highly unfashionable note on which I will conclude.
Productivity and Social Policy
For several decades, a higher orthodoxy has held that we face a choice, a tradeoff, between economic performance and fairness, between efficiency and equity. In the name of an economic objective, namely the provision of capital to industry, we have sacrificed our urban centers, our public transport, our public schools, child care, health care, and social services. Under the strictures of Gramm-Rudman-Hollings, we continue to do so to this very day. It is clear that this strategy has not given us a better life. And it has not given us a more competitive economy either.
The MIT view has the considerable merit of challenging the tradeoff orthodoxy, by asserting that education, infrastructure, and other public functions are complements of, rather than alternatives to, successful industrial development and high productivity growth. This is a step in the right direction. But it risks being unsustainable, for as I have argued there may exist no dose and demonstrable connection between new public investments and international trade results. Indeed, the demand-side effects of any serious increase in public capital formation are virtually certain to make our overall trade performance worse, not better.
There is a another way to defend social outlay, which is to argue for decent social policy and public capital investments on their social merits. The Europeans, after all, did not build their welfare states for reasons of productivity or competitiveness. They do not now, except in limited cases, connect education or welfare or urban or housing policies to trade performance. Most such programs, with the exception of labor market policies, neither help nor hurt national productivity as a whole. Their logic is social and political.
If this is correct, we may seriously distort the proper role of education, social, and welfare policy in general, by making it too much the servant of business goals. We may end up having sacrificed independent judgment on what a school system, a child support system, or a housing or infrastructure program ought to do, in the service of economic objectives that these means cannot achieve. And it may be wiser, even from an economic standpoint, to divorce the issues of social welfare and competitiveness.
The country's salvation may well not lie in subordinating education, housing, health care, and child care to economic objectives. It may lie, instead, in taxing the successful members of the community to provide for the basic wants of the others. And in directing other instruments of policy, in particular policies of demand management, worldwide development, and technological standard-setting, to assure that there are enough prosperous and competitive members of the community to bear the burden.