Editor's note: This story was originally published in our April 1994 issue, and has been reposted to highlight the long history of Nobelists publishing in Scientific American.
The real wage of the average American worker more than doubled between the end of World War II and 1973. Since then, however, those wages have risen only 6 percent. Furthermore, only highly educated workers have seen their compensation rise; the real earnings of blue-collar workers have fallen in most years since 1973.
Why have wages stagnated? A consensus among business and political leaders attributes the problem in large part to the failure of the U.S. to compete effectively in an increasingly integrated world economy. This conventional wisdom holds that foreign competition has eroded the U.S. manufacturing base, washing out the high-paying jobs that a strong manufacturing sector provides. More broadly, the argument goes, the nation’s real income has lagged as a result of the inability of many U.S. firms to sell in world markets. And because imports increasingly come from Third World countries with their huge reserves of unskilled labor, the heaviest burden of this foreign competition has ostensibly fallen on less educated American workers.
Many people find such a story extremely persuasive. It links America’s undeniable economic dificulties to the obvious fact of global competition. In effect, the U.S. is -- in the words of President Bill Clinton -- like a big corporation in the world economy and, like many big corporations, it has stumbled in the face of new competitive challenges.
Persuasive though it may be, however, that story is untrue. A growing body of evidence contradicts the popular view that international competition is central to U.S. economic problems. In fact, international factors have played a surprisingly small role in the country’s economic diffculties. The manufacturing sector has become a smaller part of the economy, but international trade is not the main cause of that shrinkage. The growth of real income has slowed almost entirely for domestic reasons. And -- contrary to what even most economists have believed -- recent analyses indicate that growing international trade does not bear signifcant responsibility even for the declining real wages of less educated U.S. workers.
The fraction of U.S. workers employed in manufacturing has been declining steadily since 1950. So has the share of U.S. output accounted for by value added in manufacturing. (Measurements of “value added” deduct from total sales the cost of raw materials and other inputs that a company buys from other firms.) In 1950 value added in the manufacturing sector accounted for 29.6 percent of gross domestic product (GDP) and 34.2 percent of employment; in 1970 the shares were 25.0 and 27.3 percent, respectively; by 1990 manufacturing had fallen to 18.4 percent of GDP and 17.4 percent of employment.
Before 1970 those who worried about this trend generally blamed it on automation -- that is, on rapid growth of productivity in manufacturing. Since then, it has become more common to blame deindustrialization on rising imports; indeed, from 1970 to 1990, imports rose from 11.4 to 38.2 percent of the manufacturing contribution to GDP.
Yet the fact that imports grew while industry shrank does not in itself demonstrate that international competition was responsible. During the same 20 years, manufacturing exports also rose dramatically, from 12.6 to 31.0 percent of value added. Many manufacturing firms may have laid off workers in the face of competition from abroad, but others have added workers to produce for expanding export markets.
To assess the overall impact of growing international trade on the size of the manufacturing sector, we need to estimate the net effect of this simultaneous growth of exports and imports. A dollar of exports adds a dollar to the sales of domestic manufacturers; a dollar of imports, to a first approximation, displaces a dollar of domestic sales. The net impact of trade on domestic manufacturing sales can therefore be measured simply by the manufacturing trade balance -- the difference between the total amount of manufactured goods that the U.S. exports and the amount that it imports. (In practice, a dollar of imports may displace slightly less than a dollar of domestic sales because the extra spending may come at the expense of services or other nonmanufacturing sales. The trade balance sets an upper bound on the net effect of trade on manufacturing.)
Undoubtedly, the emergence of persistent trade deficits in manufactured goods has contributed to the declining share of manufacturing in the U.S. economy. The question is how large that contribution has been. In 1970 manufactured exports exceeded imports by 0.2 percent of GDP. Since then, there have been persistent deficits, reaching a maximum of 3.1 percent of GDP in 1986. By 1990, however, the manufacturing deficit had fallen again, to only 1.3 percent of GDP. The decline in the U.S. manufacturing trade position over those two decades was only 1.5 percent of GDP, less than a quarter of the 6.6 percentage point decline in the share of manufacturing in GDP.
Moreover, the raw value of the trade deficit overstates its actual effect on the manufacturing sector. Trade figures measure sales, but the contribution of manufacturing to GDP is defened by value added in the sector -- that is, by sales minus purchases from other sectors. When imports displace a dollar of domestic manufacturing sales, a substantial fraction of that dollar would have been spent on inputs from the service sector, which are not part of manufacturing’s contribution to GDP.
To estimate the true impact of the trade balance on manufacturing, one must correct for this “leakage” to the service sector. Our analysis of data from the U.S. Department of Commerce puts the figure at 40 percent. In other words, each dollar of trade deficit reduces the manufacturing sector’s contribution to GDP by only 60 cents. This adjustment strengthens our conclusion: if trade in manufactured goods had been balanced from 1970 to 1990, the downward trend in the size of the manufacturing sector would not have been as steep as it actually was, but most of the deindustrialization would still have taken place. Between 1970 and 1990 manufacturing declined from 25.0 to 18.4 percent of GDP; with balanced trade, the decline would have been from 24.9 to 19.2, about 86 percent as large.
International trade explains only a small part of the decline in the relative importance of manufacturing to the economy. Why, then, has the share of manufacturing declined? The immediate reason is that the composition of domestic spending has shifted away from manufactured goods. In 1970 U.S. residents spent 46 percent of their outlays on goods (manufactured, grown or mined) and 54 percent on services and construction. By 1991 the shares were 40.7 and 59.3 percent, respectively, as people began buying comparatively more health care, travel, entertainment, legal services, fast food and so on. It is hardly surprising, given this shift, that manufacturing has become a less important part of the economy.
In particular, U.S. residents are spending a smaller fraction of their incomes on goods than they did 20 years ago for a simple reason: goods have become relatively cheaper. Between 1970 and 1990 the price of goods relative to services fell 22.9 percent. The physical ratio of goods to services purchased remained almost constant during that period. Goods have become cheaper primarily because productivity in manufacturing has grown much faster than in services. This growth has been passed on in lower consumer prices.
Ironically, the conventional wisdom here has things almost exactly backward. Policymakers often ascribe the declining share of industrial employment to a lack of manufacturing competitiveness brought on by inadequate productivity growth. In fact, the shrinkage is largely the result of high productivity growth, at least as compared with the service sector. The concern, widely voiced during the 1950s and 1960s, that industrial workers would lose their jobs because of automation is closer to the truth than the current preoccupation with a presumed loss of manufacturing jobs because of foreign competition.
Because competition from abroad has played a minor role in the contraction of U.S. manufacturing, loss of jobs in this sector because of foreign competition can bear only a tiny fraction of the blame for the stagnating earnings of U.S. workers. Our data illuminate just how small that fraction is. In 1990, for example, the trade deficit in manufacturing was $73 billion. This deficit reduced manufacturing value added by approximately $42 billion (the other $31 billion represents leakage -- goods and services that manufacturers would have purchased from other sectors). Given an average of about $60,000 value added per manufacturing employee, this figure corresponded to approximately 700,000 jobs that would have been held by U.S. workers. In that year, the average manufacturing worker earned about $5,000 more than the average nonmanufacturing worker. Assuming that any loss of manufacturing jobs was made up by a gain of nonmanufacturing jobs -- an assumption borne out by the absence of any long-term upward trend in the U.S. unemployment rate -- the loss of “good jobs” in manufacturing as a result of international competition corresponded to a loss of $3.5 billion in wages. U.S. national income in 1990 was $5.5 trillion; consequently, the wage loss from deindustrialization in the face of foreign competition was less than 0.07 percent of national income.
Many observers have expressed concern not just about wages lost because of a shrinking manufacturing sector but also about a broader erosion of U.S. real income caused by inability to compete effectively in world markets. But they often fail to make the distinction between the adverse consequences of having slow productivity growth – which would be bad even for an economy that did not have any international trade -- and additional adverse effects that might result from productivity growth that lags behind that of other countries.
To see why that distinction is important, consider a world in which productivity (output per worker-hour) increases by the same amount in every nation around the world -- say, 3 percent a year. Under these conditions, all other things remaining equal, workers’ real earnings in all countries would tend to rise by 3 percent annually as well. Similarly, if productivity grew at 1 percent a year, so would earnings. (The relation between productivity growth and earnings growth holds regardless of the absolute level of productivity in each nation; only the rate of increase is significant.)
Concerns about international competitiveness, as opposed to low productivity
growth, correspond to a situation in which productivity growth in the U.S. falls to 1 percent annually while elsewhere it continues to grow at 3 percent. If real earnings in the U.S. then grow at 1 percent a year, the U.S. does not have anything we could reasonably call a competitive problem, even though it would lag other nations. The rate of earnings growth is exactly the same as it would be if other countries were doing as badly as we are.
The fact that other countries are doing better may hurt U.S. pride, but it does not by itself affect domestic standards. It makes sense to talk of a competitive problem only to the extent that earnings growth falls by more than the decline in productivity growth.
Foreign competition can reduce domestic income by a well-understood mechanism called the terms of trade effect. In export markets, foreign competition can force a decline in the prices of U.S. products relative to those of other nations. That decline typically occurs through a devaluation of the dollar, thereby boosting the price of imports. The net result is a reduction in real earnings because the U.S. must sell its goods more cheaply and pay more for what it buys.
During the past 20 years, the U.S. has indeed experienced a deterioration in its terms of trade. The ratio of U.S. export prices to import prices fell more than 20 percent between 1970 and 1990; in other words, the U.S. had to export 20 percent more to pay for a given quantity of imports in 1990 than it did in 1970. Because the U.S. imported goods whose value was 11.3 percent of its GDP in 1990, these worsened terms of trade reduced national income by about 2 percent.
Real earnings grew by about 6 percent during the 1970s and 1980s. Our calculation suggests that avoiding the decline in the terms of trade would have increased that growth to only about 8 percent. Although the effect of foreign competition is measurable, it can by no means account for the stagnation of U.S. earnings.
A more direct way of calculating the impact of the terms of trade on real income is to use a measure known as command GNP ( gross national product). Real GNP, the conventional standard of economic performance, measures what the output of the economy would be if all prices remained constant. Command GNP is a similar measure in which the value of exports is deflated by the import price index. It measures the quantity of goods and services that the U.S. economy can afford to buy in the world market, as opposed to the volume of goods and services it produces. If the prices of imports rise faster than export prices (as will happen, for example, if the dollar falls precipitously), growth in command GNP will fall behind that of real GNP.
Between 1959 and 1973, when U.S. wages were rising steadily, command GNP per worker-hour did grow slightly faster than real GNP per hour -- 1.87 percent per year versus 1.85. Between 1973 and 1990, as real wages stagnated, command GNP grew more slowly than output, 0.65 percent versus 0.73. Both these differences, however, are small. The great bulk of the slowdown in command GNP was caused by the slower growth of real GNP per worker -- by the purely domestic impact of the decline in productivity growth.
If foreign competition is neither the main villain in the decline of manufacturing nor the root cause of stagnating wages, has it not at least worsened the lot of unskilled labor? Economists have generally been quite sympathetic to the argument that increased integration of global markets has pushed down the real wages of less educated U.S. workers.
Their opinion stems from a familiar concept in the theory of international trade: factor price equalization. When a rich country, where skilled labor is abundant (and where the premium for skill is therefore small), trades with a poor country, where skilled workers are scarce and unskilled workers abundant, the wage rates tend to converge. The pay of skilled workers rises in the rich country and falls in the poor one; that of unskilled workers falls in the rich country and rises in the poor nation.
Given the rapid growth of exports from nations such as China and Indonesia, it seems reasonable to suppose that factor price equalization has been a major reason for the growing gap in earnings between skilled and unskilled workers in the U.S. Surprisingly, however, this does not seem to be the case. We have found that increased wage inequality, like the decline of manufacturing and the slowdown in real income growth, is overwhelmingly the consequence of domestic causes.
That conclusion is based on an examination of the evidence in terms of the underlying logic of factor price equalization, first explained in a classic 1941 paper by Wolfgang F. Stolper and Paul A. Samuelson. The principle of comparative advantage suggests that a rich country trading with a poor one will export skill-intensive goods (because it has a comparative abundance of skilled workers) and import labor-intensive products. As a result of this trade, production in the rich country will shift toward skill-intensive sectors and away from labor-intensive ones. That shift, however, raises the demand for skilled workers and reduces that for unskilled workers. If wages are free to rise and fall with changes in the demand for different kinds of labor (as they do for the most part in the U.S.), the real wages of skilled workers will rise, and those of unskilled workers will decline. In a poor country, the opposite will occur.
All other things being equal, the rising wage differential will lead firms in the rich country to cut back on the proportion of skilled workers that they employ and to increase that of unskilled ones. That decision, in turn, mitigates the increased demand for skilled workers. When the dust settles, the wage differential has risen just enough to offset the effects of the change in the industry mix on overall demand for labor. Total employment of both types of labor remains unchanged.
According to Stolper and Samuelson’s analysis, a rising relative wage for skilled workers leads all industries to employ a lower ratio of skilled to unskilled workers. Indeed, this reduction is the only way the economy can shift production toward skill-intensive sectors while keeping the overall mix of workers constant.
This analysis carries two clear empirical implications. First, if growing international trade is the main force driving increased wage inequality, the ratio of skilled to unskilled employment should decline in most U.S. industries. Second, employment should increase more rapidly in skill-intensive industries than in those that employ more unskilled labor.
Recent U.S. economic history confounds these predictions. Between 1979 and 1989 the real compensation of white-collar workers rose, whereas that of blue-collar workers fell. Nevertheless, nearly all industries employed an increasing proportion of white-collar workers. Moreover, skill-intensive industries showed at best a slight tendency to grow faster than those in which blue-collar employment was high. (Although economists use many different methods to estimate the average skill level in a given industrial sector, the percentage of blue-collar workers is highly correlated with other measures and easy to estimate.)
Thus, the evidence suggests that factor price equalization was not the driving force behind the growing wage gap. The rise in demand for skilled workers was overwhelmingly caused by changes in demand within each industrial sector, not by a shift of the U.S.’s industrial mix in response to trade. No one can say with certainty what has reduced the relative demand for less skilled workers throughout the economy. Technological change, especially the increased use of computers, is a likely candidate; in any case, globalization cannot have played the dominant role.
It may seem difficult to reconcile the evidence that international competition bears little responsibility for falling wages among unskilled workers with the dramatic rise in manufactured exports from Third World countries. In truth, however, there is little need to do so. Although the surging exports of some developing countries have attracted a great deal of attention, the U.S. continues to buy the bulk of its imports from other advanced countries, whose workers have similar skills and wages. In 1990 the average wages of manufacturing workers among U.S. trading partners (weighted by total bilateral trade) were 88 percent of the U.S. level. Imports (other than oil) from low-wage countries -- those where workers earn less than half the U.S. level -- were a mere 2.8 percent of GDP.
Finally, increasing low-wage competition from trade with developing nations has been offset by the rise in wages and skill levels among traditional U.S. trading partners. Indeed, imports from low-wage countries were almost as large in 1960 as in 1990 -- 2.2 percent of GDP -- because three decades ago Japan and most of Europe fell into that category. In 1960 imports from Japan exerted competitive pressure on labor-intensive industries such as textiles. Today Japan is a high-wage country, and the burden of its competition falls mostly on skill-intensive sectors such as the semiconductor industry.
We have examined the case for the havoc supposedly wrought by foreign competition and found it wanting. Imports are not responsible for the stagnation of U.S. incomes since 1973, nor for deindustrialization, nor for the plight of low-wage workers. That does not mean, however, we believe all is well.
Some of those who have raised the alarm about U.S. competitiveness seem to believe only two positions are possible: either the U.S. has a competitive problem, or else the nation’s economy is performing acceptably. We agree that the U.S. economy is doing badly, but we find that international competition explains very little of that poor performance.
The sources of U.S. difficulties are overwhelmingly domestic, and the nation’s plight would be much the same even if world markets had not become more integrated.
The share of manufacturing in GDP is declining because people are buying relatively fewer goods; manufacturing employment is falling because companies are replacing workers with machines and making more efficient use of those they retain. Wages have stagnated because the rate of productivity growth in the economy as a whole has slowed, and less skilled workers in particular are suffering because a high-technology economy has less and less demand for their services. Our trade with the rest of the world plays at best a small role in each case.
The data underlying our conclusions are neither subtle nor difficult to interpret. The evidence that international trade has had little net impact on the size of the manufacturing sector, in particular, is blatant. The prevalence of contrary views among opinion leaders who believe themselves well informed says something disturbing about the quality of economic discussion in this country.
It is important to get these things right. Improving American economic performance is an arduous task. It will be an impossible one if we start from the misconceived notion that our problem is essentially one of international competitiveness.