Journal Issue

Interview with Olivier Blanchard: European unemployment has origins in shocks, institutions, and interactions between them

International Monetary Fund. External Relations Dept.
Published Date:
January 2000
  • ShareShare
Show Summary Details

IMF Survey: European unemployment is commonly perceived to have been rising steadily. True?

Blanchard: Nearly, but not quite. Until 1970, European unemployment was very low. Teams of U.S. economists crossed the ocean to try to understand why Europe had 2 percent unemployment, while U.S. unemployment remained at 5 percent. The talk was of the “European unemployment miracle.” Then in the late 1960s, European unemployment began to rise—slowly at first, and then, with the two oil shocks, very quickly. By the end of the 1970s, unemployment was very high, and it continued to rise for most of the 1980s. In the 1990s, it more or less plateaued—increasing a bit, but not much.

The “not quite” refers to the past three years, when most European countries saw a decline in unemployment. Spain’s unemployment dropped from 22 percent to around 15 percent. French unemployment went from 12.5 percent to under 10 percent. There’s still a long way to go, but that’s clear progress. In addition, the Netherlands and Ireland saw their unemployment rates drop to less than 4 percent—basically no unemployment at all.

IMF Survey: But European unemployment has remained a hot topic. What characteristics differentiate it, for example, from the U.S. experience?

Blanchard: In the mid-1970s, the initial focus was, quite naturally, on oil shocks. Oil prices eventually came down, but European unemployment remained very high, while it declined in the United States. So oil price shocks became a less convincing explanation for what was happening.

Many politicians and economists then shifted the blame from shocks to institutions. European labor markets are organized differently from the U.S. labor market: there is more social insurance, more employment protection, and a larger tax wedge between take-home pay and the cost of labor to firms. It became very tempting to say, well, these institutional differences explain why Europe is not doing well. The old argument that a generous welfare state makes people lazy resurfaced and provided the underlying theme for a number of major studies in the early 1990s.

Sure, European institutions are different. But they were different in the 1960s and the 1970s when unemployment was low. So the simple answer that “the welfare state did it” has difficulty explaining why it worked for so long and why it suddenly became the trigger for higher unemployment.

IMF Survey: But the United States did seem better able to digest the shocks of the 1970s and 1980s.

Blanchard: The theme of my lectures at the IMF—which reflects, I think, a growing consensus among researchers—is that an explanation for high European unemployment must be constructed on three pillars: shocks, changes in institutions, and interactions between shocks and institutions.

Adverse shocks, the first pillar, have been important. In retrospect, the oil shocks were partly a smoke screen. Oil price shocks did trigger the recessions of the mid-and late 1970s, but the major event was happening behind the scenes: a slowdown in the rate of total factor productivity growth—essentially a slowdown in the underlying rate of technological progress.

From the end of World War II until the early 1970s, Europe grew at an amazing rate, fed initially by postwar reconstruction and then, at least in part, by a process of catch-up with the United States. Real wages went up about 5–6 percent annually without putting pressure on costs. But in the 1970s, the growth rate of total factor productivity decreased. By the end of the 1970s, we were on a path where wages could rise only 2–2½ percent if they were to be consistent with the evolution of factor productivity.

This major change in the economic environment required major adjustments in expectations. For quite a while, though, workers continued to bargain, and firms negotiated, on the basis of the earlier, stronger growth. Amid oil shocks, changes in relative prices, problems of measurement, and two recessions during this period, it was difficult to know just what was going on. But when the smoke cleared, it was clear that lower factor productivity growth was a new fact of life and that there had been an effective wage explosion relative to what should have happened. That shock, which was not obvious at the time, now looks very, very big.

But there were other shocks, too. Real interest rates moved as they have never moved, at least in recent memory. In the 1970s, inflation exploded; central banks increased nominal rates but typically by much less, so real rates plunged, often becoming negative. Then, in the 1980s and 1990s, central banks tightened money, and there were high real interest rates due to German reunification at the beginning of the 1990s. Interest rates went up and inflation went down, leading to very large positive real interest rates.

These swings in interest rates had an effect on unemployment. Lower borrowing costs partly offset higher labor costs in the 1970s, leading to more capital accumulation and less unemployment than might have been otherwise. Conversely, in the 1980s and 1990s, higher borrowing costs added to unemployment. In effect, some of the increase in unemployment that would have happened in the 1970s was deferred to the 1980s and early 1990s.

Changes in institutions are the second pillar, and the question here is how much of the increase in unemployment came from an increase in the welfare state. Thanks to work done at the Organization for Economic Cooperation and Development (OECD) and elsewhere, we now have good measures of the generosity of unemployment benefits and of the degree of employment protection across both countries and time.

Unemployment benefits did become a bit more generous in the 1960s and in the 1970s, but not by much. Since then, there has not been much action in average replacement rates (the ratio of unemployment benefits to take-home pay), but the most extreme distortions have been reduced, and the most generous replacement rates have decreased.

Employment protection was also strengthened in the 1970s and early 1980s, as, under pressure from labor, governments often tried to prevent firms from laying off workers on a large scale. But again, the increase was incremental; employment protection had been high before. And, while progress has happened at a glacial pace, employment protection is less stringent now than it was in the mid-1980s. This doesn’t quite fit the evolution of unemployment.

In short, the welfare state did not suddenly come into being in 1975, and everyone did not stop working because it became attractive to be unemployed. Of course, you do find instances where institutions have had perverse effects. In the late 1980s, France introduced a guarantee of a minimum income level, whether or not people work. With the minimum income level set at roughly half the minimum wage, there seemed to be sufficient incentive to look for work. The problem, however, was that often the unemployed lost housing subsidies and lunches for their kids if they returned to work. So the incentive to return was weak. It is an example of bad design and presents a problem for France at this juncture. Unemployment is coming down, but as many as 1 million of the roughly 2.5 million unemployed may not have much financial incentive to take the jobs that are coming on line.

But for every case like this, many others have been improved. Indeed, given the urgency, the French government is working on this one as well. Blaming the increase in the welfare state for the increase in unemployment reflects more ideology than fact.

IMF Survey: But in terms of institutional changes, can’t changes in the environment also make institutions less relevant to the needs of the job market?

Blanchard: That’s the perfect transition to the third pillar: the interactions between shocks and institutions. It could be that institutions stayed the same, but a change of circumstances made them less appropriate—like winter clothes in the summertime.

A popular hypothesis—articulated, for example, by the OECD’s jobs study of the early 1990s—is that while Europe was a quiet place—growing fast but without much reallocation—it was okay to have high employment protection and generous unemployment benefits. If a firm does not want to lay off in the first place, firing costs are irrelevant. If people do not become unemployed, unemployment benefits may not be very important, and so on. Then, the argument goes, the environment changed and became more turbulent. Europe now needs a lot of reallocation, the old labor market institutions are standing in the way, and this leads to higher unemployment.

This all sounds plausible but it has empirical and conceptual problems. The empirical problem is that the measures of turbulence we have constructed—be it relative changes in employment across firms, sectors, or regions—show little evidence of an increase in turbulence. This could be a problem with our measurements, but I believe the measures deliver the right message: Sure, the world is changing, but it has changed in the past as well. Maybe “always changing” is the nature of the change. Perhaps the high-tech sector and the new economy of the past five years or so are indicative of more intense structural change, but this clearly does not account for the change in unemployment since 1975.

The conceptual problem is that if you need to reallocate labor but are prevented from doing so by high firing costs, it may be very bad for efficiency and growth. But it may not lead to high unemployment; the result may be that no one moves and no one is unemployed. Indeed, the empirical evidence is that high employment protection does not systematically lead to higher unemployment; it leads to a longer duration of unemployment, but lower flows through unemployment. The result for the unemployment rate is ambiguous.

I find another explanation more plausible and better grounded in the empirical data—although, even there, there are still a few holes. One of the dramatic differences between the U.S. and the European labor markets is that even at the same unemployment rate, Europe has much longer duration of unemployment and much lower flows of workers through the labor market. For example, over the past 20 years, Portugal and the United States have averaged about 6 percent unemployment. But the average individual duration of unemployment in the United States has been about 2–3 months, while in Portugal it has been around 8–12 months. At the same time, flows in and out of unemployment have been about four times smaller in Portugal. Since the unemployment rate can be thought of as the product of flows times duration, overall unemployment rates have been roughly the same in both countries, but the two labor markets are completely different. In the United States you lose your job, you wait a few months, you get another one. In Europe, you are much less likely to lose your job but more likely to remain unemployed longer if you do. This is why Europe, even at the same unemployment rate as the United States, has a very high proportion of long-term unemployed.

The second point is that the long-term unemployed are different from the short-term unemployed. Some are different from the start, and that is why they are long-term unemployed. But some become different—long-term unemployment makes you lose morale, self-confidence, and skills. And for all these reasons, if you become long-term unemployed, your chances of finding a new job fall dramatically. In effect, you become irrelevant to the labor market. And this has an important macroeconomic implication: If the unemployment rate is high, but a high proportion of the unemployed are long-term unemployed, there will be little pressure on wages, which is the mechanism a market economy typically relies on in the medium run to return to lower unemployment.

In the United States, if unemployment reaches 10 percent, there is tremendous pressure on wages to come down. In Europe, if we go from 5 to 10 percent unemployed, a lot of the unemployed are going to end up long-term unemployed. And shocks are going to be larger and last longer, because Europe has a weaker feedback mechanism to translate high unemployment into lower wages and rising employment.

Caricaturing only slightly, I would say that the shocks that affected Europe and the United States in the 1970s and the 1980s may not have been that different. Differences in labor market institutions are the reason why their effects have been longer lasting in Europe.

We must avoid long-term unemployment traps. This means getting the currently long-term unemployed back into jobs and preventing new entrants from falling into the trap.

IMF Survey: What are the implications of your thesis, then, for European policymakers?

Blanchard: Prospects for the future strike me as bright—with a footnote. The effects of the slowdown in total factor productivity growth have long been absorbed, and, if the United States is any example, Europe may be on the verge of an increase in total factor productivity growth. Very high interest rates, I hope, are a thing of the past, so the cost of borrowing is lower. Thus, some key contributors to persistent unemployment are gone.

Macroeconomic fundamentals are good as well. The profit rate is as high as it has been in 40 years, and wage moderation continues. All of these factors point to the feasibility of high investment rates, sustained growth, and a continued reduction in unemployment. We are seeing all of this in the sharp reductions in unemployment in Spain and France.

Institutional reforms are still needed—some sooner than others—otherwise, lower unemployment will soon translate into higher inflation. We must avoid long-term unemployment traps. This means getting the currently long-term unemployed back into jobs and preventing new entrants from falling into the trap. A number of countries are making progress on both fronts. Unemployment insurance systems are changing, so that unemployment benefits are contingent on really trying to get a job, and benefits are lost if a job offer is not taken. Such a change is probably one factor—but by no means the only one—behind the dramatic fall in unemployment in the Netherlands. The United Kingdom has introduced a similar program, and France is on the verge of adopting it as well.

Employment protection provisions are costly. They lead to high unemployment rates among specific groups, such as the young and the old. Governments have worked mostly at the margin here, allowing, for example, fixed-duration contracts for new workers. These fixed-duration contracts make it easier for firms to hire new workers, but also make it very tempting to fire these workers at the end of the period before they come under employment protection. A better solution, from a political and an economic point of view, may be to simplify rather than drastically decrease the level of employment protection: offer automatic severance pay, with many fewer legal and administrative steps and delays.

The political economy of labor market reforms is a delicate one, but progress can be made, especially in an environment of sustained growth. The conditions for sustained growth are there, and I am optimistic about the future.

Professor Blanchard’s course at the IMF was based on his Lionel Robbins lectures. The full text of these lectures is available on his website (

Other Resources Citing This Publication