European unemployment.

Blanchard, Olivier J.

The unemployment rate in the EC increased sharply in the early 1980s, rising from 5.7 percent in 1979 to 10.9 percent in 1985. Why it increased is not mysterious: it was the result of a general shift toward anti-inflation policies, adopted first in England and then a couple of years later on the Continent. Indeed, the increase in unemployment was associated with a large decrease in inflation, from 10.7 percent in 1979 to only 5.5 percent in 1985. The mystery came later, in the second part of the 1980s, as unemployment remained high and no longer was associated with disinflation. Today, the EC unemployment rate stands close to 11 percent, and inflation remains at about 4 percent, only 1.5 percentage points down from 1985.

Lawrence H. Summers and I first took up this issue in 1986.(1) We were struck by the failure of traditional theories to explain continuing high unemployment. On the one hand, the fact that high unemployment no longer was associated with disinflation suggested an increase in the equilibrium rate of unemployment. But, reviewing the list of "usual suspects," we found little evidence of major increases in either structural change and reallocation activity, or union activity, or generosity of unemployment benefits, or explicit and implicit labor costs.(2) On the other hand, if continuing unemployment was caused by a continued lack of aggregate demand, and thus was far above equilibrium unemployment, why was disinflation coming to an end? To cut the Gordian knot, Summers and I suggested an alternative answer: unemployment remained high mainly because it had been high for so long. Put another way, continued high unemployment had led to high equilibrium unemployment.

Our basic theory was simple, even simplistic. We viewed wage bargaining as a process that demonstrated that employed workers cared about their own employment prospects but did not care much, if at all, about the unemployed. Thus, after a sequence of adverse shocks leading to higher unemployment, those who still had jobs chose wages low enough to keep them employed, but not so low as to create jobs for the unemployed. The implication was straightforward: once unemployment was high, it showed no tendency to decrease, and remained high until some favorable shocks came along. To capture this dependence of equilibrium unemployment on history, we called our theory a "hysteresis" theory of unemployment.(3)

The theory made a very relevant point, namely that the unemployed are not present at the bargaining table.(4) But it explained too much, and it explained it too easily.

It explained too much because, most of the time, unemployment appears to return to some stable value: the European experience of the 1980s is the exception rather than the rule. One must go back to the Great Depression to find unemployment behaving similarly.

It also explained persistence too easily, by in effect assuming away, rather than explaining away, the potential effects of unemployment on wage determination. There were many issues left unresolved. Even if the employed did not care about the welfare of the unemployed per se, shouldn't high unemployment lead them to be more careful in their wage demands, so as to decrease the probability of finding themselves unemployed? And, why didn't firms rely on the presence of a large pool of unemployed workers to extract wage concessions from their employed workers?

These questions led us to think harder about why the unemployed had so little effect on wage outcomes. In a series of papers, we explored the connection among the level of unemployment, the proportion of long-term unemployed, and wage determination. We argued that a higher unemployment rate typically is associated with a higher proportion of long-term unemployed. And, we argued that the long-term unemployed have a weaker effect on wage determination than other workers do, for two reasons: they search less as time passes, and they search less effectively. The long-term unemployed also tend to lose skills and work ethic. They become less employable, or are perceived as less employable by employers. In either case, this makes them less of a threat to those who are still employed, and thus decreases the effect of unemployment on wage determination.(5)

Our theory, so extended, lost the most striking implications of the initial model: the strict hysteresis property of unemployment. Equilibrium unemployment was no longer purely dependent on history. After a series of adverse shocks to unemployment, and thus an increase in the proportion of long-term unemployed, the pressure on wages was weak, but no longer zero. Unemployment returned, albeit slowly, to a stable long-run value. But something also was gained from the extensions. The theory now potentially could explain why the European experience of the 1980s was different from the general postwar experience. It implied that the slower the speed at which unemployment returned to normal, the longer the period of high unemployment and the higher the proportion of long-term unemployed. Thus, the theory implied, the problem of Europe in the second half of the 1980s was that if had accepted high unemployment for too long, weakening the forces that usually reduce unemployment.

I felt, however, that even this extended theory was still lacking in important ways. In particular, the framework was built on the assumption of collective bargaining, while many, it not most, wages actually are determined outside of collective bargaining. It did not include an explicit description of the flows between unemployment and employment, but such a description is needed to derive the relationship of the proportion of long-term unemployment to the history of the unemployment rate. These were some of the motivations for the research I then undertook with Peter Diamond. That research was not aimed primarily at explaining European unemployment. Rather, its purpose was to take a fresh look at unemployment and wage dynamics, starting from an explicit description of job creation, job destruction, and of the process by which workers and jobs meet and match. In a series of papers, we explored, both analytically and empirically: the implications of fluctuations in job creation and destruction for fluctuations in unemployment and vacancies; the relationship of the flows of workers among unemployment, employment, and nonparticipation in the labor force to the level of aggregate unemployment; the efficiency of the matching process; and, the relationship of wages to both unemployment and vacancies.(6)

More importantly for the topic at hand, we used this framework to return to the issue I had first tackled with Summers: the relationships among unemployment, long-term unemployment, and wage determination. We explored the implications of an assumption we called "ranking": that, other things equal, employers prefer to hire those unemployed workers who have been unemployed for the least amount of time.(7) We showed that, under that assumption, the proportion of long-term unemployed increases rapidly with the level of unemployment: the higher the unemployment rate, the higher the chance that a long-term unemployed worker will be the job applicant with the shortest duration, and thus the lower the chance that the worker is hired. We also show that, as a result, the effect of high unemployment on wages is reduced. In bargaining for wages with firms, those who are employed realize that, if they were to become unemployed, they would be hired ahead of those currently unemployed. The long-term unemployed are no competition, and thus are no threat to the employed workers.

Do I feel I have uncovered the key to European unemployment? No. There are clearly other reasons why unemployment in Europe cannot return to the levels of the 1960s, when high growth was consistent with high levels of workers' protection. I feel, however, that this line of research has uncovered a key. Going beyond the specific mechanisms I have described, economies will adapt to high unemployment. The unemployed workers adapt in various ways: returning to live at home, finding occasional undeclared employment, and so on. The political system responds by providing the unemployed with enough to live on. Unemployment in Spain would not be above 20 percent today if many of the unemployed had starved. But, while this adjustment is socially desirable, it also weakens the forces that would naturally return the economy to lower unemployment. My research suggests two major conclusions for Europe today: first is the scope for a large decrease in unemployment. Some of the factors I have identified, such as the disenfranchising of the young, may be hard to reverse; but most, including the decrease in search intensity, are likely to disappear as labor market conditions improve. Second, market forces alone would take a very long time to increase employment unless there is an explicit change in policies aimed at increasing employment.(8)

1 O. J. Blanchard and L. H. Summers, "Hysteresis and European Unemployment," in NBER Macroeconomics Annual 1986, S. Fischer, ed. Cambridge: MIT Press, 1986, pp. 15-77.

2 Very useful country-by-country analyses, focusing on those factors, were presented at a conference in England in 1985, and published in The Rise in Unemployment, C. Bean, R. Layard, and S. Sickell, ads. Oxford: Basil Blackwell, 1987. This set of studies convinced us to look for an alternative explanation.

3 Hysteresis is related in many ways to multiple equilibriums. We explored these connections in O. J. Blanchard and L. H. Summers, "Beyond the Natural Rate Hypothesis, "American Economic Review 78, 2 (1988), pp. 182-187.

4 This was also the central point of the work by A. Lindbeck and D. Snower, summarized in their book, The Insider-Outsider Theory of Employment and Unemployment, Cambridge: MIT Press, 1988.

5 Two papers that discuss these channels and their implications are "Getting the Questions Right--and Some of the Answers" in Europe's Unemployment Problem, C. Bean and J. Dreze, eds. Cambridge: MIT Press, 1991, pp. 66-89; and "Wage Bargaining and Unemployment Persistence," Journal of Money, Credit and Banking 23, 3, 1 (1991), pp 277-292.

6 O. J. Blanchard and P. Diamond, "The Beveridge Curve," Brookings Papers on Economic Activity 1 (1989), pp. 1-76, "The Cyclical Behavior of the Gross Flows of U.S: Workers," Brookings Papers on Economic Activity 2 (1990), pp. 85-143, and "The Aggregate Matching Function" in Productivity, Growth, and Unemployment, P. Diamond, ed. Cambridge; MIT Press, 1990, pp. 159-201.

7 O. J. Blanchard and P. Diamond, "Ranking, Unemployment Duration, and Wage Determination," Review of Economic Studies, forthcoming.

8 O. J. Blanchard and P. A. Muet, "Competitiveness Through Disinflation: An Assessment of French Macroeconomic Policy Since 1983," Economic Policy 16 (April 1993), pp. 12-56.