The Macroeconomic Roots of High European Unemployment:

The Impact of Foreign Growth

Dean Baker and John Schmitt

October 15, 1998

Revised June 21, 1999

Paper prepared for conference on "Creating Competitive Capacity: Reassessing the Role of the U.S. and German Labor Market Institutions in the New Economy," Washington, DC, October 23-25, 1998.  This is the English version of the article that will appear in the December, 1999 issue of WSI Mitteilungen Paper

In 1994, economists in the United States were virtually unanimous in believing that the unemployment rate could not fall much below 6.0 percent without triggering an inflationary spiral. The Phillips curve regressions that were used to determine the non-accelerating inflation rate of unemployment (NAIRU) in the United States appeared to provide solid empirical support for a NAIRU that was in a range from 5.8% to 6.5% (Weiner 1993, Congressional Budget Office 1994, and Gordon 1982). The unemployment rate has now been below 6.0% for more than four years and below 5.0% for the last two years. Instead of rising, inflation, by every measure, is lower today than it was in 1994. In short, the history of the last few years has completely disproved the accepted view within the economics profession.

As a result of the Federal Reserve Board's decision not to use monetary policy to prevent the unemployment rate from falling below 6.0%, approximately 4.0 million more workers now have jobs than would otherwise have been the case. In addition, standard estimates of the Okun relationship between GDP and unemployment imply that the U.S. economy has produced more than $400 billion in additional output over the last four and a half years because the unemployment rate was allowed to fall below the accepted measures of the NAIRU.

The declines in unemployment have been especially dramatic for traditionally disadvantaged groups. As the overall unemployment rate fell from about 6% to just over 4%, the unemployment rate for African-American adults, for example, fell from 10.1% to just 6.6%. For African-American teens the unemployment rate over the same period fell from over 35% percent to as low as 20%.

The U.S. experience of the last four years is an important point of departure for any discussion of European unemployment because it demonstrates the limits of the profession’s understanding of the determinants of unemployment. The evidence placing the NAIRU in the 5.8-6.5% range for the United States was certainly stronger and more widely accepted than the evidence for any particular measure of the NAIRU for any of the European countries that are currently suffering from high unemployment. The enormous gains that resulted from a U.S. monetary policy that was not guided by the accepted view of the NAIRU points to the enormous stakes involved in accurately determining the barriers to reducing unemployment in Europe.

The next section of this paper examines the theoretical and empirical problems with the widely accepted view that high unemployment in Europe is primarily attributable to the structure of the labor market. The following section outlines the case that high unemployment is primarily attributable to contractionary macroeconomic policies, the impact of which is quickly transmitted across national borders. The last section is a brief conclusion.

Every nation in western Europe has experienced a dramatic rise in its unemployment rate between the 1960s and the 1980s and 1990s. The (unweighted) average unemployment rate for these countries in the 1960s was 1.9%. In the 1980s it had risen to 7.4%, and thus far in the 1990s the average has been 8.8%. While this rise in unemployment is clearly the largest problem facing the economies of western Europe, no generally accepted explanation has emerged. A Keynesian macroeconomic explanation of this increase would require evidence that these economies have typically been demand constrained. This would mean showing that the unemployment rate has been responsive to changes in aggregate demand, for example, an increase in government spending leading to a fall in unemployment. Explaining a simultaneous rise in unemployment across countries would require an explanation of how contractionary policies can be transmitted across borders. In the case of Europe the obvious mechanism is exports. Since such a large portion of GDP crosses national boundaries, the impact of a contraction in one nation is quickly felt elsewhere in Europe. The degree of economic integration within Europe also largely negates the benefits of expansionary policies in a single country since much of the benefits of any increase in demand will be felt elsewhere.

By contrast, a microeconomic explanation for a simultaneous increase in unemployment requires that either a common set of factors altered the labor markets in each nation at approximately the same time, or that a common shock hit the European economies and their labor markets were ill-equipped to adjust to it. The first sort of explanation would look to evidence that the European labor markets were more rigid in the 1980s and 1990s than in the 1960s. For example, the finding that government unemployment benefits were more generous on average in the later period would provide evidence in support of this theory. The second explanation would require a clear identification of what the shock was, and then an explanation of why Europe’s labor markets were unable to adjust. Both sorts of microeconomic explanations rely on rigidities in the labor market to explain European unemployment. The major difference is that the first view explains current high levels of unemployment by claiming that labor markets have become more rigid. The second view attributes the rise in unemployment to an underlying change in the structure of the economy, which leads to unemployment because labor markets are rigid. In both cases, the policy prescription is the same, reduce rigidities in the labor market.

Although the microeconomic explanation for high European unemployment is widely accepted among economists and in policy circles, it actually rests on very little evidence. The claim that European labor markets are more rigid at present than they were thirty years ago is questionable. That the degree of rigidity has increased sufficiently to explain a rise in unemployment of the magnitudes observed is almost implausible on its face. The alternative explanation—that a shock has occurred that is of sufficient magnitude to explain this rise in unemployment—also seems questionable. The evidence for each of views is discussed briefly below.

While European labor markets clearly have more rigidities than the U.S. labor market, it is not all evident that they have more rigidities in the 1990s than they did twenty or thirty years ago. Labor unions, the most obvious source of rigidities, are weaker in nearly every European country in the 1990s than in the 1960s. The proportion of the work force that is covered by union contracts has declined in most European countries. In some cases, Austria and France, for example, the decline has been quite sharp, with the percentage of wage and salary earners represented by unions falling by more than ten percentage points between 1970 and 1990. In many nations, the legal protections for unions have been weakened as well (OECD 1994b, pp. 9- 23). The evidence indicates that if any change has indeed occurred, it is that labor unions now produce fewer rigidities than they did in the 1960s.

The situation with government supports in the labor market is mixed. Legislated minimum wages show no clear trend relative to average wages over this period (OECD 1994b, pp.46-51). Government-mandated vacation time has increased in many countries and the length of the standard work week has become shorter in some countries. These trends, however, have been a feature of the entire postwar period, as workers have sought to ensure that a portion of the gains from higher productivity accrue in the form of more leisure. The data do not suggest that the trend towards a shorter statutory work week has accelerated in the last two decades. These restrictions on work time, however, have been at least partially offset by other changes that allow greater flexibility. Specifically, most European nations have adjusted their labor laws to facilitate part-time work, temporary employment, and other arrangements (OECD 1994b, pp. 93-100).

Many European nations have had restrictions on dismissals that require advance notice and severance pay. But these restrictions have been in place for decades and in many cases have been weakened in recent years (OECD 1998, p. 18). The movement in marginal tax rates on labor has not shown any clear trend. Some rise took place from the late 1970s to the mid-1980s, but marginal tax rates have fallen in the 1990s, so that they are not very different from the levels in place at the end of the 1970s (OECD 1994b, pp. 240-244).

More generous unemployment benefits could explain some of the increase in European unemployment, but no simple relationship exists over time between changes in benefit levels and changes in unemployment. While benefits were in general higher in the 1990s than in the 1960s, much of the increase took place in the 1970s, long before the large increase in European unemployment. For example, in both Belgium and the Netherlands virtually all of the rise in the generosity of benefits took place before 1977. In the case of Germany, the rising generosity of benefits explains none of the increase in unemployment, since benefit levels have actually become slightly less generous over the last three decades. The same is the case in Italy, where benefits are virtually non-existent. In these countries, at least, high unemployment benefits can not explain rising rates of unemployment.

The magnitude of the increase in European unemployment rates is far larger than what can plausibly be supported by the limited evidence of increased labor market rigidities over the last three decades. Madsen (1998) applied a model developed by Layard and Nickell (1986), which explains movements in unemployment rates largely by changes in labor market rigidity, to the unemployment path in OECD nations over the period from 1960 to 1993. The study found that the coefficients of the variables measuring the generosity of unemployment benefits were insignificant and that the model could explain less than 20% of the increase in unemployment in the OECD over this period.

Recent OECD analyses of changes in the structural unemployment in the 1990s underscore the implausibility of the rigidities story. As Table 1 demonstrates, according to the OECD, the non-accelerating-inflation-rate of unemployment rose in most European countries in the 1990s. In some instances, the increases were large. Between 1990 and 1997, the estimated NAIRU, for example, increased 5.8 percentage points in Finland, 3.5 percentage points in Sweden, and 2.7 percentage points in Germany (OECD 1998, p. 9). (See Table 1.) Yet, by the OECD's own assessment (OECD 1998), the 1990s were a period during which nearly all European nations reduced labor market rigidities. Since labor market rigidities have not increased during the 1990s, they cannot possibly explain the dramatic increases in unemployment over the last decade.

The alternative microeconomic explanation for high European unemployment is not that rigidities increased, but rather that the underlying economic environment changed. According to this line of argument, rigid European labor markets may have been well adapted for the early post-war period, but they were unable to adjust to the more turbulent environment of the last twenty five years. The shocks that are potentially to blame for this turbulence are the oil price hikes and the productivity slowdown of the 1970s, the spread of computer technology, and the increase in global competition. Each of these will be briefly discussed below.

The 1970s hit European economies with two major oil price shocks. At the same time, productivity growth rates slowed across Europe and the rest of the OECD. Many models have attempted to blame the rise in European unemployment in the 1980s on the inability of European labor markets to adjust to these shocks. According to this view, workers resisted the decline in real wages necessitated by the rise in oil prices or, alternatively, workers resisted the deceleration in real wage growth necessitated by the deceleration in productivity growth. In either case, workers' ignorance or stubbornness led them to press for excessive wage demands that, in the end, drove unemployment up. Lindbeck and Snower (1988) have offered an influential variation on this process: given strong unions and legal protections against dismissal, powerful "insiders" (the employed) can continue to secure substantial wage gains with relatively little risk of unemployment. As a result, even a large pool of unemployed "outsiders" does little to moderate wage demands in the face of adverse shocks.

Whether or not these models can explain the rise in unemployment in the 1980s, they offer little help in coming to grips with high unemployment in Europe at the end of the 1990s. The productivity slowdown was a quarter of a century ago and the last major oil shock was twenty years ago. Furthermore, world oil prices have long since returned to their pre-shock levels. Any negative impact that these shocks had on the unemployment rate in the 1980s should have been reversed by the 1990s. Given that the world economy has not suffered any new, comparable, shocks, and given the reversal of the 1970s oil price rise, we should have seen a fall in the unemployment rate in the 1990s across Europe. That this has not occurred suggests that a slow labor market adjustment to the 1970s shocks was not the cause of the increase in European unemployment in the 1980s.

If wage demands had been excessive, the capital share of income and rates of return to capital should have fallen, or at least remained constant, over the 1980s and 1990s. The large increase in almost every European nation since 1979 in both the capital share of corporate income and in the return to capital (see Table 2) argues against the view that excessive worker demands have generated high unemployment. Instead, the profit data suggest that workers, in aggregate, have made significant concessions to capital over the last two decades, casting significant doubt on the idea that excessive labor power lies at the root of European unemployment.

The third and forth possible shocks, technology and trade, can be treated together as relative demand shocks. The basic argument is that either technology or trade has led to a significant increase in the relative demand for more highly skilled workers and significant decrease in the relative demand for less skilled workers. In the case of technology, the claim is that the spread of computers and other information technologies have radically altered the production process, increasing the productivity of well-educated workers that have mastered the technology, but decreasing the productivity of less- educated workers. In the case of trade, the claim is that large numbers of less-skilled workers in manufacturing are being displaced by low-wage workers in developing nations. Since highly-educated workers do not face the same sort of competition, this leads to a relative decline in the demand for less-skilled workers.

In both scenarios, the decline in the relative demand for less-skilled workers should have led to a fall in the relative wages for the less-skilled. The United States has experienced a large decline in the relative wages of the three-quarters of the work force without a college degree, as predicted by a relative demand shift view. In Europe, no comparable decline in the wages of less-skilled workers occurred. In the standard view, the failure in Europe of relative wages of less-skilled workers to fall during the 1980s and 1990s is then the "cause" of high European unemployment.

Regardless of the evidence for technology or trade as the major factor behind the change in relative wages in the United States, this sort of relative demand shift explanation is inconsistent with basic features of European unemployment. If the problem is that wage rigidities are preventing a fall in the relative wages of less-skilled workers, then the rise in unemployment should be concentrated among the less skilled. This, however, has not been the case. Table 3 shows that the ratio of the unemployment rate of less-educated workers to that of the college-educated workers is no higher in Europe than it is in the United States. Less-educated workers in Europe are actually less likely to be unemployed, relative to better-educated European workers, than are their less-educated U.S. counterparts. These data indicate that labor market rigidities have not raised the relative unemployment rates of less-educated workers in Europe; instead, some other force appears to have raised the unemployment rate of all European workers relative to rates in the United States. Employment rates by educational level reinforce the message of the unemployment data. The decline in employment rates in European economies cuts across all skill and education levels (see Nickell and Bell (1995), Card, Kramer, and Lemieux (1996), and Krueger and Pischke (1997)).

To sum up, microeconomic explanations for the rise in unemployment over the last two decades are far from compelling. Little evidence supports the view that European labor markets have become more rigid over last two decades. The major oil shocks took place 20-25 years ago and have largely been reversed. The deceleration in productivity growth has become a basic fact of the last 25 years of economic life, diminishing the plausibility that ongoing unwillingness to reconcile wage demands with slower productivity is still an active cause of current high unemployment. The substantial decline in labor's share of corporate income and the steep rise in the return to capital argue further against theories of unemployment that rely on excessive wage demands. Finally, relative employment and unemployment rates by education level are not consistent with a relative demand shock against less-educated workers. Both the timing across countries, and the distribution of unemployment across skill groups, suggest that high European unemployment rates may have its roots in the macroeconomy.

In this section, we outline an argument that macroeconomic factors are the cause of the rise in European unemployment. We also develop estimates of the potential impact of slower foreign growth on each nation’s domestic unemployment rate. These estimates indicate that, in the absence of offsetting factors, unemployment can spread easily across national boundaries. This discussion is intended to be suggestive, and to identify some of the issues that will have to be dealt with in subsequent efforts to test the hypothesis that macroeconomic factors explain high European unemployment.

The core of the macroeconomic explanation for higher European unemployment is quite simple. The European nations have relied on contractionary fiscal and monetary policy to bring down their inflation rates from the peaks reached in the late 1970s and early 1980s to the negligible rates of the last several years. These contractionary policies have both a direct effect on national output and employment and an indirect effect—through several channels—on foreign countries' output and employment. The first, and most direct, channel is through international product markets. The high degree of integration of Western European economies means that contractionary policies in one nation directly reduce demand for exports elsewhere in Europe. This effect will be especially important when the economies that pursue contractionary policies are the larger economies, such as France and Germany, which absorb a large share of European exports. The effect will have its greatest impact on smaller nations, such as Belgium and the Netherlands, which are most dependent on exports.

The second channel is through international capital markets. While the United States, Japan, and Germany have been able to maintain some independence in their monetary policy during the last two decades, other European nations, in practice, cannot maintain domestic real interest rates below those in Germany. As a result, the Bundesbank's decision to pursue contractionary monetary policy for most of the last two decades has effectively imposed the same policy on the rest of Europe. From the early 1990s, first the Maastricht treaty and then the creation of a European Central Bank, have formalized this arrangement.

The third channel is through international foreign exchange markets. In the integrated economies of western Europe, if one nation manages to sustain a rate of growth that is significantly higher than that of the rest of the continent, the fast-growing economy will inevitably run large trade deficit, as the growth of national imports exceeds the growth of national exports. Since no country can sustain a growing trade deficit indefinitely, the fast-growing country will either be forced to devalue its currency or to adopt more contractionary economic policies. The political and economic drive for a single currency, which has required relative stability in exchange rates across Europe, has generally ruled out devaluation as an option, with the result that contraction has been the route generally chosen. The ill-fated economic expansion in the early Mitterand years in France provides the model example of this scenario.

Unfortunately, the extent of the interdependence of the European economics also makes it difficult to test the impact of policy variables within each nation. Imagine trying to measure the economic impact of independent fiscal and (if it were possible) monetary policy operating in the state of Ohio, without reference to policies being pursued at the national level. An increase in the structural deficit in Ohio that was designed to stimulate economic activity in the state would almost certainly have only a limited effect on the state's economy. A large portion of the impact would be felt elsewhere in the form of higher "imports" from other U.S. states. A rise in interest rates specific to the state of Ohio, for example, probably would not have much consequence for economic activity within the state because firms and individuals could easily arrange to borrow from financial institutions in other states. That state-level fiscal and monetary policies would have little measurable effect on output and employment under such circumstances would hardly suggest that national fiscal and monetary policy would be similarly ineffective. In important respects, the relatively small, open, economies of the European Union are in a similar situation, making it difficult to gauge the importance of macroeconomic factors in explaining European unemployment.

Any effort to test a macroeconomic explanation of European unemployment also faces the standard problem of distinguishing between exogenous policy changes and endogenous economic factors. For example, the real value of a nation’s currency may decline in financial markets because of a conscious decision of its central bank or as a result of large trade deficit creating an oversupply of its currency. In the former case, the devaluation of the currency may be expected to spur future growth. In the latter case, the fall in the value of the currency may coincide with effects to constrict demand (thereby reducing the trade deficit), and therefore be associated with slower future growth.

The economic impact of government deficits poses problems that are at least as thorny. In practice, economists have great trouble distinguishing between policy-driven changes in government deficits and endogenous changes due to higher unemployment and slower growth. Since the impact of changes in unemployment rates on deficits has probably not been constant in most nations over the last three decades, accurate measures of the impact of changes in fiscal policy would require assessment of the impact of institutional changes that have altered the relationship between unemployment and deficits. (The most obvious example of this problem is the unification of Germany.) In both the case of real exchange rates and the case of changes in structural government deficits, estimation of the impact of economic policy requires careful identification of policy changes before measuring any impact on unemployment rates.

The impact of policy variables almost certainly differs across nations in proportion to the share of GDP that is devoted to meeting domestic demand. For example, in the case of the United States, where domestic demand comprises almost 90% of GDP, the impact of larger budget deficits should be far larger than in Belgium, where the ratio is close to 50%. The same would apply to the impact of a reduction in short-term interest rates. By contrast, the impact of a policy- initiated currency devaluation should be expected to be roughly proportional to the ratio of total trade (exports and imports combined) to GDP. Any effort to test the effect of these policy variables on unemployment should include some scaling to take account of these differences across nations.

A final factor that should be explicitly incorporated into any analysis of international trends in unemployment is the direct impact of foreign growth. With individual European nations committed to contractionary policies for much of the last decade, and with those nations interested in using policy to promote faster growth effectively barred from doing so by international product, capital, and foreign exchange markets, each of Europe's relatively small, open, economies suffers from what appears to be an exogenous decline in foreign demand. From the standpoint of domestic employment, the impact of demand from domestic and foreign sources should be identical. This means that a decline in export demand should have exactly the same impact on domestic employment as a decline of the same magnitude in domestic consumption or investment.

Using this simple insight, we have constructed estimates of the impact of the deceleration over the last two decades in international growth rates on the domestic unemployment rate for each OECD country. Our purpose is to illustrate that general slower European growth rates in the 1980s and 1990s have had a large impact on the unemployment rates of individual European economies. Our estimates suggest that about 20-30% of the rise in European unemployment reflects the effects of slower international growth on the export demand facing Europe. Given the conceptual and econometric problems mentioned above with trying to use national monetary, fiscal, and exchange rate data to estimate the effects of national macroeconomic policy, we don't attempt to estimate the direct unemployment impact of macroeconomic policy on unemployment. We believe that the indirect macroeconomic effects that we estimate here, together with the direct macroeconomic effects that we describe, but do not estimate here, are the most important cause of the general rise in European unemployment in the 1990s.

Our basic estimation procedure is straightforward. For each country, we first estimated how a one-percentage-point change in the growth rate of GDP affects the national unemployment rate. Since a change in GDP should have roughly the same impact on employment whether the change is due to domestic demand or foreign exports, this estimate of the unemployment-output growth relationship should allow us to quantify the employment effects of export changes for each country.

Next, we identified each OECD country's top five trading partners (based on the average share of each country's export markets over the three years 1970, 1980, and 1990). We then used economic growth data for each of the country's trading partners to calculate "foreign growth rates" (weighted by each export market's importance to the corresponding national economy) for each OECD country in each year from the late 1970s through the late 1990s. (We also scaled the foreign growth rate for each country by the importance of exports in the country's overall output. So, even rapid foreign growth would still have only a relatively small effect on a closed economy, such as the United States, while a decleration in foreign growth could have a substantial negative effect on an export-intensive economy, such as Belgium.)

Table 4 shows the detailed results from these calculations. We use the estimated relationship between GDP growth and the change in the unemployment rate in each country to estimate the change in unemployment that would occur in each country if overall GDP growth were zero in a particular year. This "zero-GDP-growth change in unemployment" appears in column 1 of Table 4. The second column shows the average export share of GDP for the same countries. The third column shows the implied increase in the domestic unemployment rate that would result if all of the nation’s trading partners experienced zero GDP growth. Note that this is just the change in unemployment that would result from zero change in the overall GDP scaled by the share of exports in national GDP. The fourth column shows the "constant unemployment rate of growth" (CURG). The CURG is the rate of overall growth that is required to keep the national unemployment rate unchanged.

Table 5 shows the actual average growth rate (columns two and three) for each nation’s five leading trading partners during the 1980s and 1990s. Columns four and five show the predicted change in the unemployment rate for each decade, based on the gap between the actual foreign growth rate and the CURG. This difference is the expected change in the unemployment rate given the gap between the actual rate of growth of each nation’s major trading partners and the level of foreign growth that would have been required to keep the unemployment rate from rising or falling. For purposes of comparison, columns six and seven show the actual change in each nation’s unemployment rate over the two decades.

As can be seen, for many countries, the change in unemployment due to the deceleration in the foreign growth rate was large relative to the actual changes in unemployment. In Germany, for example, the deceleration in growth among the major purchasers of German exports added 0.6 percentage points to the national unemployment rate over the course of the 1980s and 0.9 percentage points over the 1990s. These figures represent 20-30% of the total increase in unemployment over the two periods. Slower foreign growth rates also account for a significant portion of the increase over the same period in the unemployment rates of France, Italy, and Spain.

In some cases, actual foreign growth rates exceeded the rates necessary to keep unemployment constant. In these cases, the international economy helped to lower unemployment, all other factors held constant. The most notable case is the United Kingdom, where the export-weighted average-annual-foreign-growth rate was 2.8% in the 1990s—the highest rate in the OECD and 0.5 percentage points above the foreign growth rate necessary to keep unemployment from increasing. As a result, foreign growth lowered the United Kingdom’s unemployment rate 0.6 percentage points, all other factors held constant. The different growth experiences in the export markets for the United Kingdom and Germany suggest caution in making simple unemployment rate comparisons across the two countries. Exports are about equally important to national income in both countries—26.9% in Germany compared with 26.1% in the United Kingdom—but German export markets grew at only 1.8% per year in the 1990s, compared to a 2.8% per year rate for the United Kingdom. By 1997, the different growth rates in the two countries’ export markets accounted for 1.5 percentage points of the 3.0 percentage-point gap in unemployment rates.

Not surprisingly, foreign growth rates are much less important to the relatively closed U.S. economy. Foreign growth rates for the United States exceeded the CURG by 0.5 percentage points in the 1980s, but this pushed U.S. unemployment down by just 0.2 percentage points over the period. Foreign growth rates fell below the CURG in the 1990s, but this barely raised the U.S. unemployment rate (0.1 percentage points).

In light of the CURG analysis, the unemployment performance of four economies look particularly impressive. Denmark, Ireland, the Netherlands, and Norway all managed to lower their national unemployment rates in the 1990s, despite foreign growth rates below the foreign-growth level required to keep the unemployment rate constant.

The numbers in Table 5 represent the impact that the general deceleration in foreign growth rates would have had on domestic unemployment rates in each nation, in the absence of either offsetting policy, or endogenous economic developments. In principle, it should be possible to incorporate these predicted changes in national unemployment rates into regressions that include accurate measures of the key macroeconomic policy variables for each nation. This should provide for a better test of the hypothesis that the rise in European unemployment rates is primarily attributable to contractionary macroeconomic policy. Alternatively, the best test may require aggregating unemployment rates, interest rates, and deficits across countries, to construct measures for Europe as a whole. Only further research will construct effective tests of this view. But, the projections in Table 5 make clear that ignoring the spillover effects across nations means overlooking a major factor in determining national unemployment rates.

Even if generalized, contractionary, macroeconomic policy is responsible for high European unemployment, coordinated, expansionary, fiscal policy might still not succeed in lowering unemployment. As Blanchard and Summers (1986) and others have argued, the sustained, high, unemployment rates of the 1990s may have created conditions that make it difficult to reduce unemployment without igniting inflation, despite slack labor markets. Within this framework, the NAIRU could, for a number of reasons, have come to depend on the past levels of unemployment—a situation often referred to as hysteresis. Governments might respond to high unemployment rates, for example, by easing eligibility for, raising benefits of, or extending duration of unemployment benefits. This effort to ease the burden of unemployment could lower the incentives of unemployed workers to find new jobs, keeping unemployment from falling even after labor demand rose. Or, unemployed workers, particularly those out of work for extended periods, might lose their skills, work habits, or networks of contacts. The long-term unemployed may also face discrimination from employers. These and other factors may mean that expansionary macroeconomic policy will be relatively ineffective in lowering high unemployment once high unemployment has become an established part of the economic landscape. Even if labor demand improves, the large pool of unemployed workers will not easily rejoin the working world. Under these circumstances, the new labor demand will be translated primarily into higher pay for existing workers, a process that will, in turn, set off inflation.

Almost two decades of high unemployment in many European economies may well have raised the sustainable level of unemployment, but the question facing European governments is exactly how high the sustainable rate has risen in response to high unemployment. Ultimately, the economic importance of hysteresis effects is an empirical question and is closely related to estimates of the NAIRU. The earlier discussion of the United States is potentially instructive. While few would argue that the U.S. economy suffers from hysteresis, the vast majority of economists studying the United States, all of whom were fully aware of the flexibility of U.S. labor markets at the time that they derived their estimates, substantially overestimated the inflationary response of allowing unemployment to fall below the 5.8%-6.5% range.

The recent experience of blacks in the United States may be even more relevant for Europe. Hysteresis is a supply-side phenomenon, related to generosity of unemployment benefits, atrophy of skills, lack of connections to the formal labor market, and employer discrimination. All of these factors loom large in the labor market for black Americans, who have been disproportionately affected by welfare reform, long-term unemployment, and labor-force withdrawal. In addition, black workers still face substantial employer discrimination (Holzer 1996). Yet, the sustained, low, unemployment of the last three years has reconnected many marginalized black workers to the labor market, with the unemployment rate for black adults falling from above 10% to below 7%, and the unemployment rate for black teens falling 15 percentage points, from 35% to 20%. As the New York Times observed recently, the overall unemployment rate for black Americans is now lower than the overall European unemployment rate.

Table 6 demonstrates that conventional estimates of the NAIRU for European economies may also be off the mark. The first column of Table 6 gives the OECD's estimates for the NAIRU for 19 OECD countries in 1997. The next three columns show the standardized unemployment rates for the same countries in 1997, 1998, and for the most recent month or quarter available (generally early 1999). In 1997, the actual unemployment rate in nine of the countries was below the OECD's corresponding estimate for the NAIRU. In 1998, the actual unemployment rate was below the estimated 1997 NAIRU in 13 countries. The most recent available unemployment rate was below the estimated 1997 NAIRU in 14 countries. The last three columns of the table report annualized changes in the consumer price index for the same countries in 1997 and 1998, and then show the International Monetary Fund's (IMF) forecast for inflation in 1999. Despite a sustained, European-wide, tendency for the unemployment rate to be below the estimated NAIRU, the data show no signs of accelerating inflation.

This paper has briefly examined various explanations for the increase in European unemployment over the last three decades. It argues that the evidence for a microeconomic explanation for this rise in unemployment, based on labor market rigidities, is weak. Little or no evidence supports the view that European labor markets are more rigid in the late 1990s than they were in the 1960s. The alternative microeconomic explanation—that high European unemployment reflects inflexibility in the face of relative demand shocks—is not consistent with the patterns of relative employment and unemployment by skill levels. The increase in unemployment has not been concentrated among the least skilled, as this theory would predict. Instead, workers across all skill and education levels have experienced a rise in unemployment.

As an alternative, this paper argues that we can best understand European unemployment as a macroeconomic phenomenon, with the impact of contractionary policies transmitted across national borders. Specifically, the disinflationary policies of the 1980s and 1990s have had a predictable effect on unemployment. We constructed a set of estimates of the predicted impact of foreign growth on national unemployment rates. These estimates suggest that slow foreign growth was a major factor in the rise of unemployment rates across Europe in the last two decades, explaining in many cases 20%-30% of the actual rise in unemployment. These estimates do not include the direct, national, effects of contractionary macroeconomic policy, and lead us to conclude that macroeconomic forces are the principle cause of high European unemployment.

While we did not develop a full model to explain the rise of unemployment in Europe, the most perfunctory examination of monetary policy would suggests an obvious villain. Even after a recent cut in interest rates, the real short-term interest rate set by the European central bank still stands at more than 1.5%. For much of the last decade, the various European central banks were holding real short-term interest rates in the range of 3.0%-4.0% in order to bring inflation rates down to the levels required by the Maastricht accords. That persistently high real interest rates led to slow economic growth and high unemployment should come as no surprise to macroeconomists. In striking contrast, the Federal Reserve Board, under the leadership of Alan Greenspan, allowed the real interest rate to fall to zero in 1992 and to remain at zero for nearly two full years. This was in response to a much less serious problem of unemployment and excess capacity than is currently confronting Europe. The United States has benefited enormously from the Federal Reserve Board’s willingness to ignore the consensus on NAIRU in the conduct of monetary policy.

The evidence for the high European NAIRUs claimed by the OECD and other economists is not nearly as strong as it was for the range previously accepted in the United States. Europe's weak recovery since about 1997 has further undermined the credibility of conventional estimates of the NAIRU for many European economies. Inflation appears almost uniformly low and stable across the OECD. This recent evidence suggests that the NAIRUs held up as a constraint on macroeconomic policy lack a solid basis in either theory or evidence.

Under such circumstances, European economies should have little reason to treat such estimates of NAIRU as a constraint on macroeconomic policy. European nations have an enormous amount to gain, and very little to lose, by following the same path pursued by the Federal Reserve Board. The adoption of the Euro and the change in governments across the continent remove many of the obstacles to international macroeconomic coordination that may have existed previously. The current political situation offers an opportunity for a coordinated expansion that may finally bring an end to the period of high unemployment across Europe. It would be tragic loss if, at this historical moment, the new governments in Europe could not even show as much courage as Alan Greenspan.

The authors acknowledge helpful comments from Robert Blecker, Eckhard Hein, Hartmut Seifert, and participants at the "Creating Competitive Capacity" conference and a seminar at EPI.

Blanchard, O. and L. Summers. 1986. "Hysteresis and the European Unemployment Problem." in S. Fischer (ed.) NBER Macroeconomics Annual 1986. Cambridge, MA: MIT Press.

Card, D., F. Kramarz, and T. Lemieux. 1996. "Changes in the Relative Structure of Wages and Employment: A Comparison of the United States, Canada, and France." National Bureau of Economic Research Working Paper 5487. Cambridge, Massachusetts: National Bureau of Economic Research.

Congressional Budget Office. 1994. The Economic and Budget Outlook: An Update. Washington, D.C.: Congressional Budget Office.

DiNardo, J. and J.S. Pischke.1996. "The Returns to Computer Use Revisited: Have Pencils Changed the Wage Structure, Too? National Bureau of Economic Research Working Paper 5956.

Gordon, R.J. 1994. "Inflation and Unemployent: Where Is the NAIRU?" Paper presented at the Board of Governors of the Federal Reserve System Meeting of Academic Consultants, December 1, 1994, Washington, D.C.

Krueger, A.B. and J.S. Pischke. 1997. "Observations and Conjectures on the U.S. Employment Miracle," NBER Working Paper No. 6146. Cambridge, Mass.: National Bureau of Economic Research.

Holzer, H. 1996. What Employers Want. New York: Russell Sage Foundation.

International Monetary Fund. 1999. <http://www.imf.org/external/pubs/ft/weo/1999/>.

Layard, R. and S. Nickell. 1986. "Unemployment in Britain," Economica, vol. 53, pp. S121-169.

Layard, R., S. Nickell, and R. Jackman. 1993. Unemployment: Macroeconomics and the Labor Market. Oxford: Oxford University Press.

Lindbeck, A. and D. Snower.1988. The Insider-Outsider Theory of Employment and Unemployment. Cambridge, Mass.: M.I.T. Press.

Madsen, J.B. 1998. "General Equilibrium Macroeconomic Models of Unemployment: Can They Explain the Unemployment Path in the OECD." The Economic Journal, Vol. 108 pp.850-867.

Mishel, L., J. Bernstein, and J. Schmitt, 1998. The State of Working America 1998-9. Ithaca, N.Y.: Cornell University Press.

Nasar, S. 1999. "Where Joblessness is a Way of Making a Living," New York Times, May 9, p. 5.

Nickell, S. and B. Bell. 1995. "The Collapse in Demand for the Unskilled and Unemployment Across the OECD," Oxford Review of Economic Policy, vol. 11, pp. 40-62.

Nickell, S. 1996. "The Low-Skill Low-Pay Problem: Lessons from Germany for Britain and the U.S.." Policy Studies. Volume 17, pp.7-21.

Organization for Economic Cooperation and Development. 1994a. The OECD Jobs Study: Facts, Analysis, Strategies. Paris: OECD.

Organization for Economic Cooperation and Development. 1994b. The OECD Jobs Study: Evidence and Explanations. Paris: OECD.

Organization for Economic Cooperation and Development. 1998. "The OECD Jobs Strategy: Progress Report on Implementation of Country-Specific Recommendations." Economics Department Working Papers No. 196.

Organization for Economic Cooperation and Development. 1998. <http://www.oecd.org/
news_and_events/new-numbers/sur/sur0499a.pdf.

Schmitt, J. and L. Mishel. 1998. "The United States in Not Ahead in Everything That Matters," Challenge, November-December 1998, pp. 39-59.

Sarantis, N. 1993. "Distribution, Aggregate Demand and Unemployment in OECD Countries. The Economic Journal. Vol. 108, pp.459-467.

Weiner, S.E. 1993. "New Estimates of the Natural Rate of Unemployment." Federal Reserve Bank of Kansas City Economic Review, (Fourth Quarter), pp.53-69.