Linking Trade and Technology Policies: An International Comparison of the Policies of Industrialized Nations (1992)

PAUL R. KRUGMAN

In a broad sense, the relationship between technology and trade has been a central theme of international economics since the early nineteenth century. The basic Ricardian model of comparative advantage takes as its starting point international differences in productivity across industries, which is to say differences in technology. Empirical work confirms that countries tend to be net exporters in industries in which they have relatively high productivity. So one could say that the study of technology and international trade is virtually the same thing as the study of international trade in general.

In practice, however, discussions of technology and trade usually focus on a narrower issue: trade in so-called high-technology products. A high-technology industry may be defined conceptually as one in which knowledge is a prime source of competitive advantage for firms, and in which firms invest large resources in knowledge creation. Operationally, high-technology industries are usually defined by above-average spending on research and development, above-average employment of scientists and engineers, or both. Several basically similar classifications of high-technology industries are in circulation; Table 1 lists the U.S. Department of Commerce set of high-technology industries, which is pretty representative.

The purpose of this paper is to describe trends in high-technology trade, and to try to assess what those trends mean. The paper is in four parts. The first part asks why it is meaningful to study high-technology industries and why they should be a focus of special concern. The second part describes the actual trends in high-technology trade. The third part offers some hypotheses regarding the causes of these trends. The last part of the paper presents a summary and some tentative conclusions.

The focus of this paper is somewhat U.S.-centered. This emphasis reflects in large part availability of data. It also makes some analytical sense, however, since the story of international competition in high technology over the past generation is largely the story of the erosion of U.S. dominance.

It is surely fair to say that most observers place more stress on competition in high-technology industries than the sheer size of those industries would warrant in itself. The U.S. loss of most of the semiconductor memory market to Japan is a famous, much-emphasized story, even though memories are only a part of the semiconductor industry, and the semiconductor industry itself is by no means among the nation's largest. Europe's Airbus challenge to Boeing has attracted at least as much attention as Europe's far larger program of subsidized agricultural exports, which competes with U.S. producers just as directly. Most observers, in other words—myself included—feel that there is something special and important about high-technology industries.

It is widely suspected that high-technology industries are particularly likely to generate positive external economies, both within particular high-technology sectors and for the economy as a whole. Thus, the social return to resources placed in those sectors exceeds the private return; and to the extent that international competition leads a country to shift resources away from high technology, such competition can reduce that country's welfare.

The arguments for external economies within high-technology sectors, and those for spillovers from high-technology sectors to the economy as a whole, are somewhat different. In the next section, we will go over the argument for within-sector externalities at some length before turning briefly to the much more diffuse question of spillovers to other sectors.

It is a familiar observation—although not an observation popular among traditional theorists of international trade—that local, regional, and perhaps national advantages in particular industries are not necessarily the result of underlying differences in primary resources. Instead, advantage is often created through a process of positive feedback.^1,2 This process has recently been emphasized as a source of international competitive advantage by Porter (1990); although Porter's cases are new, his conceptual framework is essentially that introduced a century ago by Alfred Marshall.

Figure 1—a triangle that resembles Porter's ''competitive diamond,'' but with a somewhat different grouping of factors—illustrates schematically how a local or national advantage in a particular sector can be self-reinforcing. At a conceptual level, there are two kinds of external economy: market size effects, and pure informational spillovers. In practice, the two types of external economy tend to interact and to be hard to distinguish.

Market size effects act on both the labor market and on suppliers of intermediate goods (including capital equipment). A strong local or national industry, by providing a large market for labor with the right specialized skills, helps to encourage workers to acquire those skills (or to encourage workers with those skills to migrate into the relevant location). That strong

Figure 1 Self-reinforcing advantage in high technology.

industry also supports local or domestic suppliers of inputs. In turn, the availability of skilled labor or inputs helps make an industry internationally competitive, completing the loop.

When there are a number of firms in an industry, there may also be direct technological spillovers: firms may be able to learn from each other either through personal contact or by reverse-engineering each others' products. Again, the availability of a common knowledge pool helps make the industry competitive, completing a second loop.

Finally, technological spillovers and market-size effects surely interact in a mutually reinforcing way. Ideas spread best when there is a pool of highly skilled people able to appreciate them; and the process of technology diffusion often takes place between firms and their suppliers or customers rather than directly between rivals. Conversely, the skill of the labor force comes partly from knowledge that spreads informally rather than from formal training; and the strength of input suppliers rests in part on their access to the latest knowledge.

The important point for economic policy is that when industrial advantage is the result of this kind of self-sustaining process, an industry in being is more than the sum of its parts. The external economies represent a hidden asset, a sort of invisible part of the nation's infrastructure. In effect, part of the industry's value lies not in the boxes in Figure 1 but in the arrows. This immediately suggests a potential role for government policy, as well as the possibility that foreign competition may sometimes have adverse effects. We will return briefly to this touchy issue later in this paper in relation to trends in high-technology trade.

How relevant is this discussion? The role of positive feedback loops in generating self-sustaining advantage is obvious in interregional trade. One need only think of the traditional economic geography of the United States, in which many industries were highly concentrated in one or two cities—autos in Detroit and Flint, furniture in Grand Rapids, rubber in Akron, hats in Danbury. In this sense, Silicon Valley is only a recent manifestation of a phenomenon with deep historical roots. To argue for the special importance of created advantage in high-technology sectors, however, one must make the case that such positive feedback is (i) relevant to international competition and (ii) especially important in high-technology sectors.

It is fairly common now to hear the argument that to the extent that there are external economies in high-technology sectors, they are global in scope rather than national—and therefore there is no real international conflict of interest. To the extent that knowledge spreads by reverse engineering, that can be done anywhere; to the extent that ability to compete depends on sophisticated inputs, in a world of low transport costs, fast travel, and high-speed communication, inputs can be acquired easily at great distances.

It is certainly true that one must be careful before assuming that external economies represent a source of national as opposed to global advantage. Within the United States, the localization of industry has by most measures been declining since the 1940s, suggesting that local feedback loops have increasingly become national instead; probably national loops are becoming international as well. Yet one should not carry the argument too far. In several important ways it seems likely that external economies continue to be strongly national and even local in scope—a point emphasized by Porter (1990).

The first point is that specialized labor markets remain local; even movement between cities within a country is costly, and international migration is limited. When executives from Silicon Valley or Boston's Route 128 are asked why their firms continue to locate there, they usually respond by first citing the availability of a pool of skilled labor; thus, the existence of these famous high-tech clusters, as well as less well-known ones like the group of optics-related firms in Rochester, can be attributed to a highly localized kind of external economy.

The relationship between a domestic industry and a domestic supplier base is also far from gone, in spite of global communications and low transport cost. There are intangible costs to transactions at a distance; in some ways the move to modern management systems based on just-in-time inventory and production has increased the premium placed on proximity, so that in the electronics industry in particular there has been a discernible trend for firms to move production back from low-wage offshore sites to home locations close to suppliers and customers. Perhaps the best evidence of the continuing importance of local markets is the wide disparities that still exist between domestic and overseas market shares. Table 2 shows the share of U.S. high-technology sales in the domestic and foreign markets from 1980 to 1987. While some movement toward internationalization is visible, U.S. high-technology firms still have nearly eight times as high a share of the domestic market (which consists to an important extent of other high-technology firms) as they do of the foreign market.

So one may argue that external economies at the national level remain an important determinant of competitive advantage in high-technology industries. But is high tech special in this respect? It is clearly not unique. To take only the most obvious example, international trade in financial services is dominated by New York and London; there is no question that the dominant role of these centers is the result of self-reinforcing advantages rather than basic resources. One can, however, argue plausibly that the knowledge intensity of high-technology industries probably makes external economies more important there than in the average industry. The need for a highly skilled, specialized labor force is greater than in the average industry; close contact between suppliers and customers is more important when technology changes rapidly; knowledge spillovers are greater because there is more knowledge to spill.

Economic history also suggests that there is a life cycle to the location of industries, which reinforces the suggestion that external economies are particularly important in high tech. The characteristic pattern—illustrated by the joint history of autos and of Detroit—is that an emergent industry first develops a local focus, frequently as a result of accident or personality. Then, as the technology stabilizes, production begins to move away from that focus to save on production or transportation costs. High-technology industries are characteristically, though not always, in that first stage of rapid innovation.

It is widely believed that high-technology sectors are important, not only because of the income they generate directly, but because they yield external economies to other sectors. For example, it is often remarked that semiconductors are a "technology driver" for many other sectors, or the "crude oil to technology," phrases that are presumably meant to imply that a country that has a strong semiconductor sector will have higher productivity, other things equal, than one without.

In contrast to the almost self-evident case for external economies within high-technology sectors, intersectoral external economies are harder to argue forcefully. The market-size effects associated with localized high-technology industries are not visible for interindustry effects; there is a Silicon Valley, but not an Everything Valley.

The best argument for strong spillovers from high-technology sectors to the economy at large is that the lines of communication between domestic firms give countries with strong high-technology sectors a head start in introducing applications of new technology. This is easiest to think of in the case of consumer electronics, in which new developments in semiconductors may be crucial to new product development. It is hard to see compara-

ble links for other high-technology sectors or other parts of the economy, however; on the whole we must say that the case that high-technology sectors generate strong returns over and above their direct return is at best unsupported by the evidence.

The beginning of the 1990s represents a particularly difficult time to assess trends in international competition and trade. The reason is that during the past decade a volatile international macroeconomic and financial environment threw up so much dust that it is difficult to detect any underlying trends. For example, from 1980 to 1986 U.S. exports of high-technology products grew at an annual rate of less than 6 percent, while Japan's high-tech exports grew at an annual rate of 13 percent; yet from 1987 to 1988 the U.S. rate was 24 percent. Clearly not all of this change represented a reversal in the long-term competitive trend, but it is difficult to tell what, if anything, happened to that trend.

The most important source of confusion about the 1980s is, of course, the rise and fall of the dollar. Figure 2 illustrates the extent to which that rise and fall affected the competitive position of major competitors. In 1980

Figure 2 Ratio of German to U.S. hourly compensation costs in manufacturing. SOURCE: U.S. Department of Labor (1989, p. 575, Table 149).

German manufacturing workers received an hourly compensation that was 25 percent higher than that of their U.S. counterparts. By 1985 the rise of the dollar had shifted the German rate of compensation to 25 percent below the American; by 1988 the dollar's plunge had more than restored the German premium, bringing the German rate of compensation to a level 30 percent higher than the U.S. rate.

The consequence of these huge changes in comparative labor cost has been roller-coaster behavior of trade in manufactures. Table 3 shows U.S. trade balances in high-technology and low-technology manufactures. During the 1980s, the traditional U.S. surplus in high technology disappeared, while the deficit in other manufactures ballooned; then both balances began to recover—a recovery that has without doubt continued beyond the dates covered in this table. One cannot infer from the decline in the high-technology surplus over this period that the United States was experiencing a fundamental decline in its high-technology competitiveness—nor can one refute this proposition. The point is that the financial shocks obscure the evidence.

Nonetheless, it is possible to discern, albeit rather dimly, some broad underlying trends in international high-technology competition. Perhaps the best way to highlight the nature of the trend is to place it in opposition to a straw man representing conventional wisdom about the competitive position of the United States. The conventional wisdom, which one still finds in many popular discussions, sees a U.S. economy that is suffering a progressive process of deindustrialization due to imports of manufactured goods; U.S. industry supposedly cannot compete with low-cost imports of traditional manufactures. The only available strategy is to compensate with growing exports of high-technology products. For a while the United States was able to offset its growing trade deficit in low-tech manufactures with a growing surplus in high-tech goods; the risk is that the United States will lose what remaining edge it has, and that high-technology industry will go the way of apparel and textiles.

The reality is quite different, and in some ways nearly the reverse. De-

industrialization, a "hollowing out" of the U.S. manufacturing base by imports was never a possibility that made sense: given the dominant role of manufactured goods in U.S. trade, one could always count on the dollar eventually to fall to a level that made U.S. production competitive in world markets. And in practice, the negative impact of trade deficits on the size of the U.S. manufacturing base has never been extensive and has been diminishing in the last few years. Table 4 shows the U.S. trade balance in manufactures as a percentage of manufacturing value added, for selected years. Before the 1980s the number was very small, so that trade had no significant deindustrializing effect before 1983 or so. At the trade deficit's peak it was more than 14 percent of manufacturing value added; by 1990 it had again fallen to about 9 percent. Only part of that 9 percent represents manufacturing value added shifted abroad, since some of the cost of imports and exports consists of inputs that the manufacturing sector purchases from other sectors. Thus, at a rough estimate the U.S. manufacturing sector in 1990 was only about 5 percent smaller than it would have been had the U.S. economy been closed to foreign competition.

It is widely believed that until the 1980s the stability of the overall U.S. trade balance in manufactures was the result of a simultaneously growing surplus in high technology and a deficit in other goods. This is certainly true of the nominal trade balances. Virtually all of the growth in these imbalances, however, can be tied to inflation and the general growth of the U.S. economy. Table 5 shows the relevant trade balances in nominal terms and as a share of gross national product (GNP); these latter numbers show little pronounced trend, and were in any case quite small.

If one looks at the composition of output, furthermore, one sees if anything a somewhat worse U.S. comparative performance in high technology than in other goods. Table 6 compares the U.S. share of world high-technology output with its share of manufactures in general; both shares have declined since 1980, but the decline has actually been faster in high technology—or, to put it differently, the distinctive U.S. specialization in high

technology has eroded. Table 7 offers a similar form of evidence, comparing the high-tech share of output in the United States, Japan, and Europe; this share has risen for all three, but much more slowly for the United States and Europe than for Japan; again, one sees U.S. distinctiveness eroding. Finally, Table 8 shows measures of U.S. and Japanese "revealed comparative advantage" in high-technology products—the ratio of their export shares in that sector to all manufactured exports. Again the evidence suggests that since 1980 the U.S. position has become less distinct.

Putting all of this together, one may offer a hypothesis that is pretty much the opposite of the common view that the United States is now competitive, if at all, only in high-technology industries. In fact U.S. manufacturing in general continues to be able to sell both in the domestic and the international market; the soaring trade deficits of the 1980s were an aberration due to a strong dollar, and the subsequent several years have been marked by a widespread export revival across a broad spectrum of industries. The terms of competition, however, have gradually changed. In 1970 the United States, with a dominant position in advanced technology, was able to compete internationally despite high wages relative to other nations. At present, the United States sells goods that are no more advanced and

sometimes less so than those produced by other countries; it is able to do this because U.S. wages are not much higher and often less than those of our competitors. In other words, the United States remains able to compete; but it has shifted its manufacturing downscale relative to other advanced nations, Japan in particular.

The interesting question is whether this process of relative decline shows any signs of ending. Here the wild currency fluctuations of the 1980s make it difficult to arrive at an answer, because it is impossible to separate long-term trends from short-term events. It is possible to find pieces of evidence that point in either direction. On one side, U.S. manufacturing has experienced an impressive revival of productivity growth. Table 9 compares U.S. productivity performance with Germany, Japan, and an average of trading partners. Since 1980 the historical trend of U.S. relative decline has slowed or even reversed, thanks to a remarkable burst of productivity improvement

in the United States. This could represent a one-time gain as U.S. firms have grown leaner and meaner, or it could in part represent a more basic turnaround in the U.S. ability to apply new technology; it is still too early to tell.

On the other side, studies of the relationship between trade and exchange rates have shown little sign of a change in basic trends. Lawrence (1990), who earlier showed that the United States appears to need a persistent depreciation of the dollar in real terms to make up for lagging trade performance, finds no sign that this fact has changed in the past decade.

At this point one is forced to rely on impressions. And here it is hard not to feel that, whatever the numbers say, the United States has continued to slide in relative terms, and Japan to rise. As recently as the early 1980s it was common for economists to argue that the United States retained a dominant position in advanced technology and that despite Japanese inroads in a few areas, U.S. dominance remained largely intact. Few economists would now make the same statement; even if the measured high-technology share has remained fairly stable, there is a deeper underlying erosion.

In a broad sense the picture of high-technology trade over the past 20 years is one of gradual erosion of U.S. preeminence, with the main beneficiary being Japan. This paper is not the place to attempt a comprehensive review of explanations for these trends; all that will be attempted is a quick presentation of the main issues.

In general one may identify two main kinds of explanation for the relative decline of the United States and rise of Japan. One explanation stresses aggregate inputs, especially capital and highly educated labor. The other stresses differences in the competitive environment, such as the alleged closed nature of the Japanese market.

Extensive attention has been given to the difference in the cost of capital between the United States and Japan. A high cost of capital discourages firms from making all kinds of long-term investments, perhaps including in

particular the willingness to accept low returns during the initial development phases of new technologies. For technical reasons (involving international differences in taxation and financial structure) it is difficult to make firm estimates of the cost of capital; Table 10 reports several representative estimates, all of which suggest a substantially higher U.S. cost.

The high U.S. cost of capital may, in turn, be explained by low national savings rates. The United States has consistently had somewhat lower savings rates than other advanced countries, while Japan has saved more; during the 1980s the U.S. rate plunged to only 3.6 percent of GNP, less than half the average of nations in the Organization for Economic Cooperation and Development, while Japan saved 17.8 percent of GNP.

Trends in human capital can also help explain the trends in high-technology trade. A useful indicator—albeit a problematic one, as we will see in a moment—is the employment of scientists and engineers. Table 11 shows the number of scientists and engineers engaged in R&D per 10,000 workers in the labor force, for the United States, Japan, and Germany. The erosion of the distinctive U.S. position is evident.

The problem with this measure is that it does not clearly distinguish between supply and demand. Japan may be employing more people in R&D because its high-technology industries are successful, rather than the other way around. One can try to get around this by appealing to evidence on the quality of basic education—which is sufficiently poor in the United States to explain just about any pattern of decline. It is also possible to argue, however, that Japanese high-technology success is at least in part generated by government policies and by the advantages of a de facto closed domestic market. Given the description earlier in this paper of how advantage can be created in high-technology industries, one certainly cannot rule out such an explanation. A closed market can in principle allow a country to break into industries in which the self-reinforcing advantages of established competitors would otherwise block entry and can conversely deny foreign

rivals the opportunity to establish the virtuous circles that allow an industry to succeed.

All of this is very iffy. The best one can do without getting down to detailed cases is to point out that the evidence does suggest in a circumstantial way that Japanese high-technology markets are indeed remarkably closed to imports. Table 12 shows the share of the domestic market for high-technology products accounted for by imports (import penetration) for the United States, Germany, and Japan. One might have expected that import penetration of the Japanese market would have been somewhat greater than that of the still considerably larger U.S. market, and that Japan would have shared in the trend toward internationalization. In fact, however, Japan has consistently remained nearly self-sufficient in high-technology products. At least on the face of it, the caricature of Japanese high-technology industries as being protected from imports, but free to export if they succeed in acquiring a competitive advantage, appears consistent with the numbers.

This brief overview of technology and trade has made three main points. The first is that the special emphasis many observers place on international competition in high-technology industries makes considerable sense. The

qualification to this is that it is not the technology per se that makes the industries special, but the likelihood that such industries are characterized by strong external economies that give rise to self-reinforcing advantage. From this perspective, the financial services sector is as worthy of special concern as microelectronics.

Second, there has been a long-term trend of U.S. industry ''downscale'' and of Japanese industry "upscale." That is, what was once a distinctive U.S. position of dominance in high-technology industries has eroded over time, and Japan has begun to show the kind of distinctive pattern that the United States no longer has. This trend is not, however, associated with wholesale U.S. deindustrialization.

Third, the sources of this trend are ambiguous. Aggregate factors such as the cost of capital and the supply of highly educated labor have moved in a direction that helps explain the trends in high-technology competition, but the circumstantial evidence is also consistent with stories that emphasize market access and government action.

This overview, then, leaves the most crucial issues—how does policy affect the relationship between technology and trade, and what should be done differently—open for further discussion.

Arthur, B. 1990. Positive feedbacks in the economy. Scientific American 262:92-99.

Economic Report of the President. 1991. Washington, D.C.: U.S. Government Printing Office.

Lawrence, R. 1990. U.S. Current Account Adjustment: An Appraisal. Brookings Papers on Economic Activity 2:1990.

National Science Board. 1989. Science & Engineering Indicators--1989. NSB 89-1. Washington, D.C.: U.S. Government Printing Office.

National Science Board. 1991. Science & Engineering Indicators--1991. NSB 91-1. Washington, D.C.: U.S. Government Printing Office.

Porter, M. 1990. The Competitive Advantage of Nations. New York: Free Press.

Poterba, J. 1991. Comparing the cost of capital in the United States and Japan: A survey of methods. Federal Reserve Bank of New York Quarterly Review Winter 1991: pp. 20-32

Survey of Current Business. Various issues. U.S. Department of Commerce. Washington, D.C.: Government Printing Office.

U.S. Department of Commerce. 1988. United States Trade: Performance in 1987. Washington, D.C.: Government Printing Office.

U.S. Department of Commerce. 1989. United States Trade: Performance in 1988. Washington, D.C.: Government Printing Office.

U.S. Department of Labor. 1989. Handbook of Labor Statistics. Washington, D.C.: Government Printing Office.