Unemployment is perhaps the most pressing economic problem in Germany. Measured unemployment reached a postwar high of about 4 million people in early 1994, or about 10 percent of the labor force. This figure does not include about 1 million people who were kept off the unemployment rolls through special labor market programs in east Germany (job creation schemes, training programs, and early retirements). Unemployment and, to a lesser extent, the hidden unemployment embodied in active labor market programs bring a huge social cost. Both also bring large economic costs—the loss of the productive potential of people who would rather be working and fiscal costs.

Unemployment is perhaps the most pressing economic problem in Germany. Measured unemployment reached a postwar high of about 4 million people in early 1994, or about 10 percent of the labor force. This figure does not include about 1 million people who were kept off the unemployment rolls through special labor market programs in east Germany (job creation schemes, training programs, and early retirements). Unemployment and, to a lesser extent, the hidden unemployment embodied in active labor market programs bring a huge social cost. Both also bring large economic costs—the loss of the productive potential of people who would rather be working and fiscal costs.

High unemployment in east Germany was largely inevitable, as an obsolete industrial structure adjusted to new patterns of competition and as previously hidden unemployment emerged into the open. But west German unemployment, too, has been high and persistent since the early 1980s, except for a brief period around the time of the unification boom (Chart 3-1). West German developments are worrisome in their own right but also because of what they augur for east Germany: with western labor market institutions extended to east Germany, will a similar rate of unemployment prevail there in the medium to long term, or, worse, to the extent that unemployment displays persistence, will the rate in east Germany be lastingly higher? Thus, although the most pressing part of the German unemployment problem is in east Germany, the west German experience holds important clues to eastern as well as western prospects, and it is on this that the present chapter concentrates.1

Chart 3-1.
Chart 3-1.

West Germany: Unemployment and Employment

(In thousands)

Sources: Statistisches Bundesamt, Volkswirtschaftliche Gesamtrechnungen; and Deutsche Bundesbank.

Labor Market Developments

Measurement Issues

The commonly reported unemployment rate in Germany consists of the ratio of unemployed job seekers to the civilian labor force; this is the rate used throughout this chapter, unless otherwise noted.2 Unemployed job seekers are defined as those who register as such with the Federal Labor Office, are available to take up a job, and currently work fewer than 18 hours a week.

The OECD-standardized unemployment rate for Germany (which is calculated by the OECD from labor force surveys) is typically much lower than the one based on national definitions; in 1993, it averaged 5.8 percent, compared with 7.3 percent for the commonly quoted rate. The OECD definition differs from the national one in three principal respects. First, the labor force is taken to include the military; however, this accounts for only a few tenths of a percentage point of the difference between the two rates. Second, the OECD defines as unemployed only those who have actively looked for work in the past four weeks. Third, and probably most important, the OECD defines as unemployed only those who did not work at all in the week prior to the survey. In the latter two respects, the OECD definition has the advantage that it does not count as unemployed those who are voluntarily unemployed or voluntarily working part time; but, conversely, it also eliminates from the definition of unemployment those who have been discouraged from actively seeking work and those who are working a few hours a week while looking for full-time work.

West German Employment and Unemployment

Unemployment in west Germany has registered three sharp rises since 1960 (Chart 3-1), only the first of which was reversed. Most worrisome was the third, which happened in the early 1980s: the long, albeit somewhat subdued, economic expansion of 1983-89 hardly lowered unemployment, which hovered around 8 percent—not far below its recent level. The unification boom produced a short-lived reduction in unemployment, and with the 1992-93 recession unemployment rose again, to record levels.

The picture for employment, however, is very different. Paradoxically, employment grew little in the 1960s and 1970s (allowing for cyclical variations) and then rose rapidly during the 1980s and early 1990s, before turning down again since 1992 (Chart 3-1). A total of 2 1/2 million jobs were created between the peak employment years 1980 and 1992.

At the same time, the labor force, which had hardly grown in the 1960s and 1970s (by less than 1 1/2 million in 20 years), increased by 3 million people between 1980 and 1992. This trend almost entirely reflected a rise in the working-age population, owing first to the coming of age of the German baby boom (1980-88) and later to net immigration of more than half a million people a year (1988-92). Age-adjusted participation rates for the total population have remained rather stable: female participation rates have risen, but male rates have fallen, reflecting longer schooling and later entry into the job market. In addition to the growth of the (resident) labor force, a large net inflow of commuters from 1989 onward—including many from east Germany—dampened the rise in resident employment: the turnaround in the net number of commuters between 1989 and 1992 was close to half a million.

East German Employment and Unemployment

As in other planned economies, open unemployment was essentially unknown in the German Democratic Republic: enterprises simply employed the entire available labor force, whether it was productive or not. With the opening of the east German economy to market forces, this previously hidden unemployment emerged into the open in 1990. Employment was initially supported by liberal provisions for short-time work (subsidized by the Federal Labor Office), but they expired at end-1991, and, despite increases in the numbers involved in other active labor market programs, unemployment rose sharply (Chart 3-2). The number of unemployed, however, has risen only slightly further since the end of 1991, and employment appears to have stabilized.

Chart 3-2.
Chart 3-2.

East Germany: Labor Market Developments

(In millions)

Source: Deutsche Bundesbank, Monthly Report, and Deutsches Institut für Wirtschaftsforschung, Sozialprodukt und Einkommen-skreislauf 1/1989 bis IV/1993.

Nature of Unemployment in West Germany

The OECD-standardized unemployment rate for west Germany is low compared with most other industrial countries. At 5.8 percent in 1993, it compares very favorably with an European Union (EU) average of 10.7 percent and ranks as the third lowest among the 17 industrial countries regularly monitored by the OECD (OECD, 1994a). Also on the positive side, the economy succeeded in creating a large number of jobs during the 1980s, and labor force growth is expected to be slower in the future, between -0.4 percent and 0.8 percent annually between 1995 and 2005, compared with an annual average of over 3 percent in the 1980s (OECD. 1994b). Nevertheless, west Germany has not been an exception to the Europeanwide trend of rising unemployment over the past three decades, nor is its level of unemployment—much less the level in east Germany—considered acceptable. The puzzle of why the west German economy did not create sufficient jobs to accommodate the growing labor force in the 1980s casts an especially long shadow over prospects for east Germany.

Cyclical Unemployment Versus the NAIRU

In principle, unemployment can be decomposed into a cyclical and a “structural” component, with some interactions between the two, which are considered below. The structural component can be defined as that rate of unemployment that is consistent with stable wage-price inflation (the nonaccelerating inflation rate of unemployment, or NAIRU) and can be viewed as an “equilibrium” rate of unemployment. In other words, given a relation between unemployment and changes in inflation, only at the NAIRU is inflation stable and, other things being equal, does unemployment remain unchanged?3 The significance of the distinction lies in the fact that, by definition, unemployment can be reduced—for in stance, by expansionary demand management policies—to the NAIRU, but not beyond it, without putting upward pressure on inflation.

There is little doubt that some of the current unemployment in west Germany is cyclical. The recession that began in mid-1992 was accompanied by a rise in the unemployment rate from 5.5-5.7 percent to a peak of 8.4 percent (May 1994). But the levels of unemployment observed around 1991 were exceptionally low. The persistence of unemployment rates well above 7 percent from the end of 1982 to the end of 1989, during seven years when GDP was growing at an average of 2 1/2 percent a year, is prima facie evidence that a good deal of unemployment during the 1980s, at least, was not related to cyclical weakness. Indeed, all measures of inflation began to rise between 1987 and 1989, when unemployment was still above 7 percent, and the further declines in unemployment around the time of unification were accompanied by rising inflationary pressures (Chart 3-3). Casual observation would thus suggest rather strongly that by the end of the 1980s the west German NAIRU was not much below 7 percent. Indeed, a number of studies that have addressed this issue formally conclude that the NAIRU displayed a clear upward trend from 1970 onward—although there is little unanimity on its current level, with estimates in the studies ranging from 6 percent to 9 percent for the mid-1980s and early 1990s (Coe, 1985; Franz and König, 1986; Burda and Sachs, 1987; Elmeskov, 1993).

Chart 3-3.
Chart 3-3.

West Germany: Unemployment and Inflation

Source: Deutsche Bundesbank, Monthly Report

The Beveridge curve (Chart 3-4) also suggests that the NAIRU has risen. The curve relates the level of vacancies to the level of unemployment: since a cyclical increase in unemployment is normally accompanied by a fall in the vacancy rate, movements along the curve can be interpreted as cyclical movements, and shifts of the curve are indicative of a rise in structural unemployment. Chart 3-4 is strongly suggestive of a shift in the German Beveridge curve around 1983. Various studies have investigated this issue formally, concluding that the curve has indeed shifted (Budd, Levine, and Smith, 1987; Börsch-Supan, 1991; Franz, 1991).

Chart 3-4.
Chart 3-4.

West Germany: Beveridge Curve

Sources: Bundesanstalt für Arbeit and Statistiches Bundesamt.

A further indicator of cyclical unemployment, the Okun curve, relates the level of unemployment to measures of capacity utilization. To the extent that unemployment is accompanied by underutilization of capacity, it can fairly be called cyclical. Chart 3-5 shows the relation between unemployment and capacity utilization in manufacturing. The chart again strongly suggests a noncyclical rise in unemployment between the recovery periods 1975-79 and 1982-90. Studies by Schultze (1987) and Jaeger and Parkinson (1990) formalize this conclusion by using the relation between cyclical unemployment and measures of capacity utilization to estimate the structural unemployment rate, which is found to be near 8 percent in the late 1980s.4

Chart 3-5.
Chart 3-5.

West Germany: Okun Curve

Sources: Bundesanstak fur Arbeit, Statistiches Bundesamt, and Deutsches Institut fur Wirtschaftsforschung.

NAIRU: Friction, Market Failure, or Hysteresis

Thus, although it is difficult to quantify the exact extent of cyclical unemployment, it seems unlikely to account for more than a small portion (perhaps 1-2 percentage points) of west German unemployment of about 8 1/4 percent of the labor force in 1994. At the same time, simply knowing that the NAIRU is high is of little help from a policy perspective, since the concept is compatible with a variety of models of equilibrium unemployment. It is useful to distinguish among three different types of equilibrium unemployment, all of which may underlie the NAIRU.

In a market-clearing framework, equilibrium unemployment is frictional: the unemployed are engaged in a job search or are choosing leisure over work, and unemployment is essentially voluntary. The equilibrium level of unemployment will depend on factors influencing the effectiveness of the job search (such as skill or regional mismatch or the efficiency of intermediation) and on factors influencing the choice between leisure and work (notably the level of unemployment benefits).

In a non-market-clearing framework, equilibrium unemployment can be involuntary, as long as some market failure stops wages from falling to market-clearing levels. For want of a better term, this chapter will refer to such unemployment as market failure unemployment. The most prominent contenders for the source of this market failure are efficiency wage theories and union power theories. In this framework, the equilibrium level of unemployment will depend in particular on the degree of unionization.

The difficulty with these concepts of equilibrium unemployment in west Germany is that none of these factors displayed the marked changes in the 1980s that would be required to explain the apparent rise in the NAIRU. Burda and Sachs (1987) conclude that frictional unemployment did not rise in the 1980s, because there was no increase in the generosity of benefits and only a slight increase in regional mismatch. Coe and Krueger (1990) allow for market failure unemployment as well as for frictional effects. They include in their estimation of equilibrium unemployment the age and sex composition of the labor force (the higher the proportion of the labor force that is prime age and male, the lower is frictional unemployment), the prevalence of apprenticeship programs (which again can be assumed to lower frictional unemployment), nonwage labor costs (which raise equilibrium unemployment if wages do not necessarily clear markets), the unionization rate, and unemployment insurance replacement ratios. They find that none of these factors explains the rise in unemployment in the 1980s and that, if the NAIRU were predicted on this basis alone, it would hover around 3 1/2 percent from 1973 to 1988.

If the structural determinants of equilibrium unemployment (frictional and market failure) have not changed and yet the estimated NAIRU has risen with actual unemployment, a third, history-dependent type of equilibrium unemployment, hysteretic unemployment, must be allowed for. In the (unlikely) case of pure hysteresis, the NAIRU is simply the actual level of unemployment: any reduction in unemployment will put upward pressure on inflation. In the less extreme case of persistence, a unique NAIRU exists, but the speed of adjustment toward it may be very slow. In both cases, unemployment will appear to be a simple autoregressive process—a hypothesis which a number of studies on west Germany are unable to reject (Jaeger and Parkinson, 1990; Carruth and Schnabel, 1990). The presence of hysteresis is of particular importance to the assessment of the current level of unemployment: since hysteresis turns cyclical into structural unemployment, the economy may well emerge from the current recession with a NAIRU even higher than was estimated in the late 1980s.

Two main theories have been put forward to explain hysteretic unemployment. The human capital theory of hysteresis stresses that it is especially difficult for the long-term unemployed to reenter the labor market, both because their human capital has depreciated during the spell of unemployment and because employers view long-term unemployment as a signal of low productivity. By contrast, the insider-outsider theory of hysteresis stresses that the employed, as insiders, have power over wages—either because of institutional arrangements such as unions or because of transaction costs—and that an increase in the numbers of unemployed outsiders may do little to moderate insiders’ wage claims.

Under both theories, the defining feature of hysteretic unemployment is a segmentation of the labor market between the “haves” and the “have-nots.” so that long-term unemployment comes to constitute a rising share of total unemployment and that the average duration of an uncompleted spell of unemployment lengthens. The west German experience of the 1980s exemplifies these developments. Long-term unemployment (defined as more than one year) rose, peak to peak, from 15 percent of total unemployment in 1979 to 28 percent in 1991. The share of unemployment of more than two years’ duration rose even more steeply, from 6 percent to 15 percent. The average duration of an uncompleted spell of unemployment rose from 7 months to 13 1/2 months over the same period.

Segmentation of the labor market is also evident from the fact that rising proportions of the unemployed, and high proportions of the long-term unemployed, display one or more of the following characteristics: they are unskilled, they live in the north of Germany, or they are older.

The unskilled constitute a disproportionate share of the total number of unemployed. The unemployment rates for specific skill groups are shown in Table 3-1. Even though the reported unemployment rate for the unskilled is probably an underestimate (see footnote 2 to Table 3-1), it is much higher than that for the skilled. It is true that, as is commonly observed in times of robust growth, the unskilled unemployment rate dropped substantially during the 1990-91 boom; but the fact that inflationary pressures mounted even as the unskilled unemployment rate was still above 8 percent supports the view that the west German labor market is strongly dualistic. The share of the long-term unemployed among the unskilled is also disproportionately high. In 1988, 37 percent of the unskilled unemployed had been unemployed for more than a year, compared with 28 percent of the skilled unemployed.

Table 3-1.

West Germany: Unemployment Rates by Skill Level1

(In Percent)

article image
Source: Staff estimates based on Bundesanstalt für Arbeit Arbeitsstatistik, and Mikrozensus (household survey) data as reported by the Statistisches Bundesornt.

Skill levels are defined as follows: vocational training includes betrieblicher Ausbildung and Berufsfachsschule; technical school is Fochschule; technical college is Fachbochschule; and university is Hochschule/Universitat. “No qualification” is a residual and includes persons who did not report their qualifications.

As the residual in unemployment, labor force, and population data, the group with no qualification is likely to be an overestimate; but the degree of overestimation is probably greater in the labor force data (because the question on qualifications is optional in the Mikrozensus survey) than in the unemployment data, which are taken from Labor Office records. The unemployment rate shown here is thus likely to be an underestimate.

In the same way as it is split by skill levels, the west German labor market also shows a pronounced north-south split. In 1991, unemployment rates across the Länder differed by almost 6 percentage points, with every one of the northern Lander registering a higher unemployment rate than any one of the southern Lander.5 Equally striking is the evolution of these unemployment rates over time (Chart 3-6). All the southern unemployment rates peaked in 1983, but the northern unemployment rates peaked between 1985 and 1988. Clearly the recovery took much longer to have an impact in the north than in the south, and countrywide inflationary pressures began to mount even as nearly all northern unemployment rates were at or above 10 percent.

Chart 3-6.
Chart 3-6.

West Germany: Regional Unemployment Rates

Source: Bundesanstalt für Arbeic

The wide dispersion of unemployment and the north-south divide existed in the late 1970s, with the variance of unemployment rates rising only a little between 1979 and 1991 (from 0.68 to 0.70). However, mismatch indices that take account of both unemployment and vacancy rates indicate a strengthening of the regional dimension of unemployment over the 1980s.6 Following Jackman and Roper (1987), two measures of mismatch can be calculated. With U and V referring to numbers of unemployed and vacancies, respectively, and the subscript i referring to regions,


which can be interpreted as the share of the unemployed who would have to move to achieve regional balance, and


which can be viewed as the potential employment gain from such movement. Both measures show a sharp increase between 1979 and 1991—Ml from 13.8 to 22.6 and M2 from 2.0 to 4.3.

Regional mismatch is sometimes taken as evidence that unemployment is frictional, since it makes job searches less effective. This view, however, is easier to defend when pockets of high unemployment are geographically dispersed. Paqué (1989) points out that findings of increased mismatch at the Land level in west Germany are at variance with the results of studies of mismatch at a more disaggregated level (the 142 labor districts), which show much less of an increase. He concludes that unemployment has turned from a “spot” issue into a “cluster” issue, and the latter is more troublesome since both trickle-down effects and labor mobility are likely to be more limited between “clusters” than between “spots.” The clear north-south split evident in Chart 3-6 suggests that clustering is a problem not only within Lander but also between groups of Lander. Overall, the picture that emerges is of a regional divide most probably brought about by structural change (the north being strongly dependent on heavy industry, such as mining, steel, and shipbuilding) and indicative not of increased length of job searches but of a marginalization of segments of the labor force.

A third dimension along which the west German labor market has become increasingly segmented is age. In contrast to many other countries where unemployment is concentrated among the young (a problem that is avoided in Germany thanks largely to the famous apprenticeship system), west German unemployment is increasingly concentrated among older groups. As shown in Chart 3-7, the unemployment rate for those aged over 50 rose sharply between the cyclical peak of 1979 and the trough of 1983 and then continued to rise through 1989. From 3.5 percent in 1979, the unemployment rate for this group had reached 7.5 percent in 1990, well above the rates for other age groups.

Chart 3-7.
Chart 3-7.

West Germany: Age-Specific Unemployment Rates

(In percent of labor force)

Source: Bundesanstalt für Arbeic

In summary, there is evidence of an increasingly segmented labor market in west Germany, which has probably been a major factor behind the rise in structural unemployment. The unskilled, those in the north of Germany (most probably those with skills that have been devalued by structural change), and older groups of workers constitute marginal labor markets with far worse prospects than workers in the primary labor market. Whether these groups have become marginalized as a result of their unemployment, as the human capital theory of hysteresis would have it, is debatable. In the words of Paqué (1990), “the incidence of long-term unemployment is so strongly related to identifiable structural characteristics that it would be far-fetched to place much explanatory weight on processes of endogenous de-qualification or demotivation.” Instead, those who are marginalized in the labor market would appear to be those who started out with low human capital, at least as viewed from the perspective of a changing production structure. Even the high rate of unemployment of older workers may reflect the devaluation of skills in the course of structural change, as well as the need for worker flexibility and adaptability in the growing service sector. Either way, it would seem that groups of low-productivity workers have been priced out of the labor market. The next section describes the wage bargaining system and examines whether it might have contributed to labor market segmentation.

Wage Bargaining System

Key Institutional Features

Collective bargaining is enshrined in the German Constitution and in the wage contract law (Tarifver-tragsgesetz) of 1949. According to the principle of Tarifautonomie, it takes place without the intervention of the Government. The partners to the negotiations are 19 national unions organized along sectoral lines and without occupational distinctions, 16 of which are members of the umbrella organization, the Deutscher Gewerkschaftsbund (DGB), and 1.000 employers’ associations, which together form the Bundesvereinigung Deutscher Arbeitgeberverbände (BDA). On the employee side, only unions can enter into collective agreements; employers can negotiate as associations or individually.

The outcomes of the wage negotiations are a set of “tariff wages” by job grade (minimum basic wages) and provisions (typically in separate multiyear agreements called Manteltarifverträge) on working time, holidays, employment protection, training and retraining, and so forth. The wage contract law specifies that deviations from tariff wages are permitted only “in favor of the worker”—a provision known as the “favorability principle” (Günstigkeitsprinzip). Thus, although there are no further formal negotiations at the firm level, employers are free to set wages above tariff wages, and most employers actually pay wages some 10-20 percent above tariff wages. A second, informal round of wage negotiations does take place at the firm level between management and the legally mandated Workers’ Councils (Betriebsräte); the Works Constitution Act (Betriebsverfassungsgesetz) of 1972 prohibits industrial action during this second round. However, in the great majority of cases wage structures are simply adjusted by the percentage increase in tariff wages. Profit sharing is little used in Germany. A few wage agreements provide for payment of bonuses at the employer’s discretion, but in practice these bonuses have tended to become set amounts unrelated to profits.

Wage negotiations take place at the regional level, but national demands are publicized by the unions in advance of the wage round, and wage contracts are typically identical nationwide within a sector.7 In addition, wage agreements in different sectors typically turn out to be very similar; in particular, the metalworking industry (represented on the workers’ side by IG-Metall with its 3 1/2 million members, amounting to one quarter of all union members in Germany), the public sector, and construction are the leading sectors, and the behavior of the other sectors has sometimes been characterized as “convoy behaνior” (Geleitzugverfahren).

Union membership (the proportion of employees who are unionized) and union coverage (the proportion of firms covered by collective bargaining agreements) in Germany are quite different. Overall, about 40 percent of employees are union members—a proportion that, in contrast to many other countries, has remained rather stable since the late 1970s. But the applicability of the collective wage agreements is much wider than data on union membership might indicate. All employers who are members of an employers’ federation are bound to pay at least the union-negotiated wage: they cannot hire union members below tariff wages, and morale considerations typically dictate that union and nonunion members be paid the same. Even if this were not the case, nonunion members can claim tariff wages through the courts on the basis of the “equal treatment” provisions of the labor law. It has been estimated that traditionally in west Germany’s manufacturing and financial sectors no less than 80 percent of employers, employing 90 percent of workers in these sectors are members of employers’ associations (Soltwedel. 1988).

Even if a firm and its workforce are agreed that wages should fall below tariff wages, perhaps to safeguard employment or the firm’s very existence, the “favorability principle” has been interpreted by the courts to mean that this agreement can only be made as a formal exception to the collective agreement. Thus, unless the collective agreement explicitly states otherwise, the firm must ask its employers’ federation to negotiate its individual case with the union. Soltwedel (1988) suggests that such requests meet with little sympathy because “the intra-marginal members of the [employers’] cartel would prefer endangered firms to go bankrupt.”

In addition, even employers who are not members of employers’ federations may become subject to the collective bargaining agreements as a result of “declarations of general validity” (Allgemein-verbindlichkeitserklärungen). According to the wage contract law, in a sector in which more than half the employees are employed by firms that are party to the collective agreements, either bargaining party may request that the Minister of Labor declare the collective agreement to be binding on all employers in the sector, regardless of whether they are members of an employers’ association. The only—ill-defined—condition such a declaration must satisfy is that it be “in the public interest.” Recently these declarations have been little used (about 2 percent of all contracts in 1993, estimated to cover a similar or marginally higher proportion of the workforce), and, when they have been, it has typically been for contracts relating to working conditions rather than to wages.

Aggregate Real Wages

The extent of the centralization of wage bargaining is generally thought to affect the degree of wage moderation, with important effects on overall employment (Calmfors and Driffill, 1988). The relation between centralization and average real wages is thought to be hump shaped: both decentralized and centralized systems may be compatible with wage moderation, while intermediate systems—industry-level bargaining, in particular—may be less so.8

The theoretical basis for this relation is that increasing union power at higher levels of centralization is offset by the fact that a number of externalities become internalized, as unions take into account the effect of their demands on other parts of the economy. Key among these externalities are price externalities (wage increases raise prices for other people), fiscal externalities (if wage increases give rise to unemployment, the corresponding unemployment benefits are partly funded by other people), unemployment externalities (higher unemployment makes it more difficult for other people to find jobs), and envy externalities (wage increases reduce other people’s welfare to the extent that they care about relativities). In addition, centralization may improve coordination and prevent the excessive wage increases that can ensue under imperfect information when each union tries to prevent its own wage increase from falling short of the national average, so that each has an incentive to err on the side of excess (Bhaskar, 1990).

The German system, while an example of industry-level bargaining, appears to reap many of the benefits of centralization and is generally ranked just behind Austria and the Nordic countries in its degree of centralization. Two factors justify this view of the German system. First, it can be characterized as “pattern bargaining” (Flanagan, Soskice, and Ulmann. 1983), in which leading sectors (and, within sectors, leading regions) set the standard for wage increases throughout the economy. Pattern bargaining is by no means identical to centralization, because the leading sector may take only its own interests into account. But compared with pure industry-level bargaining, some externalities are internalized, since the interests of the leading sector are affected by the wage increases that it knows will follow elsewhere in the economy. Second, cooperation between employers and unions in the different sectors, institutionalized in Germany under the umbrellas of the DGB and the BDA, also helps to internalize externalities and overcome the informational problems that might lead each union to demand excessive wage increases.

Wage Relativities

Although centralization appears to be conducive to overall real wage moderation, it also reduces the ability of wages to respond to conditions in different parts of the labor market. The literature has recently paid considerable attention to wage dispersion, although it has not always distinguished clearly between dispersion across firms (including across groups of firms by sector or by region) and differentiation across categories of workers (including by skill level or age). This chapter will refer to the former as “wage dispersion” and the latter as “wage differentials.”

There is evidence that the centralized systems that tend to produce real wage moderation also tend to produce low wage dispersion across firms (Freeman, 1988; Rowthorn, 1992), as might be expected since uniform wage contracts leave less scope for “tailoring” to the conditions of individual firms. Although the correlation between centralization and wage dispersion makes it difficult to separate the effects of each of these on employment, the same studies suggest a weak link between higher wage dispersion and higher employment across countries.

In a truly centralized system, low wage dispersion arises both because low-productivity firms pay relatively high wages and because high-productivity firms pay relatively low wages (both compared with the outcome under a decentralized system, at least in the short run). If these firms are thought of as constituting two separate labor markets, excess supply of labor will arise in the low-productivity labor market and excess demand for labor will arise in the high-productivity market. If labor can move freely between the two markets (for example, if the only difference between the two markets is the technology the firms use), the result for total employment is ambiguous: it may be higher or lower than with decentralized wage setting. However, if labor mobility between the two markets is limited (perhaps because the two markets require different skills or because they are geographically separated), total employment will be lower than if both markets cleared.

In an intertemporal setting, such wage equalization has a further effect, one that has formed the explicit basis for the solidaristic wage policies of unions in Nordic countries (see, for example, Flanagan, Soskice, and Ulmann, 1983) and that has been used to justify rapid convergence of east German wages to west German levels. The exclusion of low-productivity firms from the market and the high profits earned by high-productivity firms will strengthen incentives for productivity enhancements and structural change. But as Calmfors (1993) points out, even if the capital stock is younger, on average, with wage equalization, it may also be smaller: incentives to invest will be dampened by the fact that wages are foreseen to grow at the economywide rate of productivity growth, even though the productivity of a particular vintage of capital will remain unchanged. Overall, it is not clear that even the intertemporal effects of wage equalization among firms are by themselves beneficial.

Moreover, the German wage bargaining system does not generate true wage equalization because of the presence of wage drift at the firm level: firms are free to pay wages higher, but not lower, than those agreed centrally. Thus, when excess demand arises in the high-productivity labor market, wages there will rise, eliminating the already limited opportunities for labor to move to this market and thereby the outside chance that total employment might be higher than under flexible wages. The intertemporal benefits of wage equalization, too, will be muted in the presence of wage drift, since the high profits that high-productivity firms make under a policy of wage equalization will be eroded. Thus, on balance, there is considerable doubt as to the beneficial effects of wage equalization among firms even in fully centralized systems, and even more doubt about centralized systems with wage drift, such as the German one.

At the same time, it seems plausible—as suggested, for instance, by casual observation of the U.S. and European wage structures—that centralization will be accompanied by low wage differentials across workers. Such a result would not necessarily follow from the institutional setup (since a fully centralized system can still set widely different wages for different categories of workers) but from the unions’ objectives. There is much evidence that unions are averse to income inequality, perhaps because union members—and workers in general—are disportionately in the lower end of the productivity distribution (Freeman, 1980). Because of the limited or nonexistent mobility of workers between categories (such as those based on skill or age), the potential employment-increasing effects of wage equalization across firms do not arise in this case: squeezing wage differentials must have unambiguously negative effects on employment.

Wage Developments

The theoretical considerations outlined above suggest that the level of real wages in aggregate should be less of a concern in Germany than in true industry-level bargaining systems. Nonetheless, a number of studies in the 1980s (Artus, 1984: Burda and Sachs, 1987) concluded that unemployment in west Germany in the early to mid-1980s, in the aftermath of the oil shock, was at least partly traceable to excessively high labor costs, especially in manufacturing: actual wages were estimated to exceed warranted wages by up to 25 percent. However, this aggregate “wage gap” diagnosis has lost some of its appeal in the wake of subsequent wage moderation. Labor cost growth lagged productivity growth significantly from about 1984 onward (Chart 3-8)—despite the fact that the wedge between total labor costs and take-home pay rose from 78 percent of net wages in 1984 to 86 percent in 1993. While the endogeneity of productivity means that the development of unit labor costs can only be suggestive of a declining wage gap, Landmann and Jerger (1993) estimate, on the basis of trend rather than actual productivity growth, that the wage gap had disappeared by 1987-91.9

Chart 3-8.
Chart 3-8.

West Germany: Wages and Productivity

(1979 = 100)

Source: Deutsche Bundesbank.1GDP per employed person (in 1991 prices).2Gross income from dependent employment per employee (divided by GDP deflator).

However, even if real wages are not excessively high in aggregate, wage dispersion or wage differentials—in the terminology of the previous section—may still be insufficient to allow full employment. The former would explain the concentration of unemployment in structurally similar regions (given less-than-perfect mobility of workers between regions or sectors), and the latter the concentration of unemployment among the low skilled. Because it is impossible to establish empirically how much wage differentiation is “adequate.” an investigation of this proposition must rely on comparisons with other countries or with Germany itself in other periods, and the conclusions can only be suggestive.10 The Appendix to this chapter gives a more detailed review of the relevant empirical studies for Germany.

Cross-country studies have found that west Germany has a moderate degree of wage dispersion across sectors, one that is much lower than that of the United States but generally higher than that of the Nordic countries. Although many authors (for example. Bell and Freeman, 1987) rightly urge caution in interpreting results for the United States as a benchmark, and although U.S. labor markets are by no means free of rigidities the argument in the previous section suggests that the United States, with its much more decentralized wage setting, should come closer to market-clearing wage dispersion than Germany docs. These results suggest, therefore, that German wage dispersion may be insufficient.

During the 1980s, dispersion in west Germany rose across sectors (Chart 3-9), though much more weakly within manufacturing than economy wide. Comparing changes in dispersion across countries in the 1980s, most studies find that wage dispersion rose more slowly in west Germany than in other countries, particularly the United States. To the extent that advanced economies were subjected to similar forces of structural change in the 1980s—technological change and increasing competition from developing countries in labor-intensive sectors—this comparison suggests that German wages reacted less flexibly to external shocks than did those of many other countries.

Chart 3-9.
Chart 3-9.

West Germany: Coefficients of Variation of Wages1

Source: Statistichcs Bundesamt.1Wages are measured as gross income from dependent employment divided by number of employees.2Ten sectors.3Thirty-two subsectors.

West German wage dispersion across regions fell between the 1979 and 1991 peaks of the economic cycle (Chart 3-10) despite the growing differences in regional labor market conditions. Indeed, the annual growth of wages between 1979 and 1991 varied by only 0.6 percentage point across the regions, with wages in the northern Lander rising by 4.1—4.6 percent a year and those in the southern Lander by 4.6—4.7 percent a year.11 These observations are again suggestive of insufficient wage flexibility.

Chart 3-10.
Chart 3-10.

West Germany: Coefficients of Variation of Regional Wages1

Source: Statistiches Bundesamt.1Average hourly earnings of blue-collar workers in industry (11 Länder).

The issue of wage differentials between workers is perhaps even more important in Germany, given the concentration of unemployment among the low skilled. Definitional problems complicate crosscountry comparisons of wage differentials to the point of irrelevance. However, two main studies have examined trends in wage differentials in west Germany by focusing on the relations between the 90th, 50th, and 10th percentiles of the earnings distribution. OECD (1993) finds that earnings at the 90th percentile of the wage distribution rose slightly (1 percentage point) relative to median earnings between 1981 and 1990, but that earnings at the 10th percentile rose substantially (4 percentage points) relative to the median over the same period (Table 3-2). Abraham and Houseman (1993) also conclude that wage differentials in west Germany narrowed during the 1980s.

Table 3-2.

West Germany: Ratios of Percentiles of the Earnings Distribution1


article image
Source: OECD (1993); GSOEP stands for German Socio-Economic Panel data.

Earnings are gross monthly earnings plus benefits (calculated as one twelfth of 13th and 14th month pay, holiday allowances, and Christmas allowances) of full-time, full-year workers. Data are taken from social security records or from the GSOEP, Waves 1—8. 90/50 indicates the ratio of the 90th percentile to the median, 10/50 the ratio of the 10th percentile to the median.

Narrowing wage differentials at the bottom of the pay scale appear to be related primarily to a reduction in education-skill differentials. Table 3-3 suggests that between 1979 and 1991, when most sub-sectors of manufacturing (22 out of 32) saw a widening of wage differentials between high-skilled and medium-skilled workers, the majority (23 out of 32) also saw a reduction in differentials between medium-skilled and low-skilled workers. Several studies that have estimated standard earnings functions on panel data also suggest some narrowing of education and training differentials in the second half of the 1980s, with the major change a decline in the premium for holding a vocational qualification (typically a completed apprenticeship).

Table 3-3.

West Germany: Ratios of Earnings of Different Skill Groups in Manufacturing Industry1

(Ratios of hourly earnings)

article image
Sources: Statistisches Bundesamt; and IMF staff estimates.

Skill levels are the groupings defined in collective bargaining agreements: high skill, low skill, and unskilled—corresponding to Leistungsgruppe 1, 2, and 3 (master craftsman, completed apprenticeship, and unskilled), respectively.

The narrowing of wage differentials at the bottom of the scale is of particular concern in light of the high unemployment among unskilled workers. Industrial countries have in recent years seen a shift in the demand for labor toward high-skilled workers, which has tended to widen wage differentials in most of them (see Katz and Murphy, 1992; Davis. 1992). Tellingly, the OECD (1993) study finds that, among 17 OECD countries examined, west Germany is the only one that shows a pronounced narrowing of wage differentials at the bottom of the distribution during the 1980s. Although the unique German system of vocational training may have played an important part in limiting the supply of the unskilled, high and persistent unemployment among the unskilled suggests that the narrowing of wage differentials was not purely an equilibrating response and that a lowering of wages at the bottom of the scale would benefit employment.

Key Labor Market Policies

The segmented nature of the labor market and the compression of wages at the lower end of the scale point to three policy areas that are likely to be of particular importance: the wage bargaining system and income support for the unemployed (which affect the lower part of the wage scale) and employment protection (which erects barriers between those with jobs and those without). This section reviews policies in these areas.

Recent and Potential Innovations in Collective Bargaining

Both the extension of west German labor market institutions to east Germany and the recession in west Germany have thrown into sharp focus the biases of the German collective bargaining system against low-productivity firms and workers. A measure of decentralization, within the framework of collective bargaining, was achieved in the 1980s with increased devolution of authority over industrial relations to the firm-level Works Councils (Streeck, 1988). But the debate over how to make the system more flexible, while preserving its essence, has now come to center stage. Although Tarifautonomie places the responsibility for the bargaining system squarely with the social partners, the Government retains a special responsibility to influence the debate both by rhetoric and example.

The rapid convergence of eastern wages to western levels is an extreme example of the way a centralized bargaining system tends to equalize wages between different firms. The convergence has been rationalized in part by the intertemporal arguments outlined above: higher wages in east Germany generate pressures for productivity growth and speed the “creative destruction” of the inefficient inherited production structure. However, the resulting very high levels of unemployment have brought a search for ways in which the wage bargaining system can accommodate less productive firms. Four main ways of doing this have emerged so far, although only two are consistent with the essential collective nature of the German system.

Perhaps the method that has attracted the most attention has been the introduction of “opt-out” clauses (Härteklausel) in wage contracts in east Germany. These clauses allow below-tariff wages to be paid if a firm and its workforce agree that there is a financial crisis. Such clauses allow the individual employer seeking an exemption to bypass the employers’ federation. However, the regional-national union continues to be responsible for negotiating the exemption on behalf of the workforce. These clauses have been very little used, partly because the unions may be reluctant to allow the setting of low-wage precedents, especially when other avenues of survival (such as the seeking of subsidies or government guarantees) are still open to the firm. Placing the authority to negotiate exemptions at the level of the firm’s Works Council might make such clauses more useful.

However, two other methods of paying below-tariff wages have emerged in east Germany. Employers have been pulling out of, or not joining, employers’ federations: it is estimated that a quarter of employment in east Germany is in firms that are not members of employers’ federations. In addition, even some firms that are members of employers’ federations have been paying below-tariff wages, with the tacit agreement of their workforce, thus avoiding the need to seek union agreement on the use of an opt-out clause. Overall, it is estimated that 36 percent of east German firms pay below-tariff wages (as of early 1994). Of course, both these developments, if sustained, would represent a major departure from the collective bargaining system. But both the Government and the social partners have tolerated them as an appropriate response to a temporary and highly unusual situation in east Germany.

Finally, there have been moves in both west and east Germany toward greater flexibility at the firm level on points other than wages, in particular on working time. In the 1994 wage round, the metal-working and chemical sectors incorporated in their collective agreements a provision for working-hour “corridors.” In the chemical industry, for instance, the uniform 37 1/2-hour work week has been replaced by a “corridor” of 35—40 hours: workers who work less than 37 1/2 hours receive commensurately less pay, and workers who put in more time receive standard, rather than overtime, pay for the additional hours. In addition, in a number of sectors, firms were given greater freedom to compensate overtime with additional free time rather than with extra pay. Both these developments provide firms with better opportunities to minimize their labor costs while continuing to pay tariff wages.12 Further devolution of responsibility to the firm level for nonwage elements promises a better reconciliation of flexibility at the firm level with the principle of collectively set wages.

Two further methods of injecting greater flexibility at the firm level within the confines of collective wage bargaining might be considered. The first is greater reliance on wage drift. Indeed, greater wage drift may well arise if the real wage moderation demonstrated in the 1994 wage round persists into the economic recovery. However, centralized unions will always have strong incentives to legitimize themselves by seeking substantial nominal wage increases. And these incentives become all the more problematic in the German environment of low inflation, where zero or negative nominal wage increases might well be called for if wage drift were to be substantial (Calmfors, 1993).

A second potential avenue for greater flexibility is profit sharing, which has been much discussed in Germany but little implemented. A base wage lower than current wages would both allow the survival of lower-productivity firms and reduce the marginal cost of labor more generally. However, as with recourse to wage drift, the benefits of profit sharing depend on unions moderating their wage demands: otherwise profit sharing becomes merely an additional claim on existing profits. Thus, although both wage drift and profit sharing appear to promise greater firm-level flexibility, this promise cannot be fulfilled without a public debate and a wholehearted commitment by the unions and their members to the notion of holding back tariff wages.

The search for ways to accommodate productivity differentials across firms within a collective bargaining framework has been accompanied by a similar search for ways to allow for greater productivity differences across workers. Structural change in east Germany, which has rapidly devalued human capital there, and the persistence of low-skilled unemployment in west Germany have added urgency to this search. Although some have argued that employing a willing worker who would otherwise be unemployed at a below-tariff wage is by definition a development “favorable to the worker,” the “favorability principle” has been interpreted to mean that such arrangements are illegal unless explicitly provided for in the collective agreements. However, in 1994 for the first time, the collective agreement in the chemical industry incorporated a provision that allows new recruits to be paid (for one year) wages 5-7 1/2 percent below tariff wages and 10 percent below tariff wages if they were recruited out of long-term unemployment. Whether this example will be followed elsewhere remains to be seen. Equally important would be a greater differentiation of tariff wage increases among skill levels. In both cases, the Government as employer could set potent precedents.

Income Support for the Unemployed

Regardless of the form of the wage bargaining system, the income level available to the unemployed plays an important role in determining net reservation wages and hence—once account is taken of the wedge between net wages and labor costs—in setting a lower bound on the productivity of those employed. The unemployed in Germany have recourse to three different kinds of assistance: unemployment benefit (Arbeitshsengeld), unemployment assistance (Arbeitslosenhilfe), and—together with others whose income is inadequate—social assistance (Sozialhilfe).

Unemployment insurance is compulsory for all employees except civil servants and soldiers. Unemployment benefit and unemployment assistance are paid to those who have contributed to the unemployment insurance scheme for at least one year out of the three years preceding unemployment, who have involuntarily lost their jobs, and who are seeking work. Those who voluntarily quit their jobs must wait 12 weeks before drawing benefits. Those who turn down one job offer that the Labor Office considers acceptable (zumutbar) have their benefits interrupted for up to 12 weeks, and those who turn down two such offers generally have their benefits terminated.

Both unemployment benefit and unemployment assistance are exempt from tax and are calculated as ratios of the worker’s previous net earnings, excluding the typical 13th month wage. (Thus the replacement ratios typically quoted, and given below, are somewhat overstated: the official replacement ratio of 67 percent, for instance, translates into about 62 percent of total annual earnings.) The maximum net earnings to which the replacement ratios are applied are (for married persons with two children) about DM 5,070 a month in west Germany and DM 4.130 a month in east Germany, or close to twice average net earnings. The reference period for the calculation of the wage that serves as the base for unemployment benefits was three months until recently, but to combat collusion between employer and employee it was lengthened to six months, effective January 1994.

Unemployment benefit is generally payable for one year; for workers older than age 42, the period of eligibility rises with employment tenure and with age, up to a maximum of two years and eight months at age 54 or over. Until recently, unemployment benefit was payable at the rate of 68 percent of previous net earnings for a claimant with children and 63 percent for a claimant without children; from January 1994, these replacement ratios have been lowered to 67 percent and 60 percent, respectively. These ratios are not out of line with those in other OECD countries, though it might be noted that they are still above those that prevailed in Germany through the early 1970s (63 percent for a person with children).

After unemployment benefits have expired, unemployment assistance is currently available without time limit to those in need. It is also payable to most people in need who lost their jobs and were not entitled to unemployment benefit, but for them the period of eligibility, which until recently was also unlimited, was reduced to one year with effect from January 1994. Unemployment assistance was until recently payable at the rate of 58 percent of previous net earnings for a claimant with children and 56 percent for a claimant without children; from January 1994, these ratios were lowered to 57 percent and 53 percent, respectively. In principle, the wage to which these replacement ratios are applied is the potential wage rather than the actual previous wage, and, as the spell of unemployment lengthens and it becomes evident that no job with similar wages is available, the Labor Office is supposed to reduce the basis for the replacement ratio in line with their perception of the wage the unemployed person could potentially earn. Recipients must have no other means of support; if the recipient has other income or his or her spouse has income above DM 800 a month, the amount of unemployment assistance is reduced one for one.

The unlimited duration of unemployment assistance is a peculiarly generous feature of the unemployment insurance system. In 1989, the German replacement ratio for a single person in his or her third or subsequent year of unemployment was by far the highest in the OECD, exceeding the ratio in the next highest nation, the Netherlands, by 12 percentage points (OECD, 1991). Proposals to limit unemployment assistance to two years (first considered and then rejected in late 1993) were recently put forward as part of the draft 1995 budget. This measure would appear to hold considerable promise to reintegrate the long-term unemployed into the job market, both directly by raising incentives to work and indirectly by reducing insider power over wages through an increase in the search intensity of the unemployed.

Finally, social assistance (Sozialhilfe) is available for an unlimited period to anyone living in Germany whose income is inadequate to meet basic needs, although social assistance recipients who are able to work are required to register with the Labor Office and to accept appropriate job offers. The payments consist of cash allowances, contributions to housing and heating costs, and special one-off payments for necessary purchases. The amounts paid depend on family size and structure and on the ages of any children. Table 3-4 shows some typical amounts of social assistance and their relation to average earned incomes of the lowest-paid group of blue-collar workers in west and east Germany for varying household structures. The difference is large only for childless people in west Germany and for single childless people in east Germany. For all others, the ratio of social assistance to average earnings is above 70 percent, and in east Germany the ratio is at 87 percent or above for people with two or more dependents. Social assistance is indexed not to prices but to wages, suggesting that the poverty line is defined in relative rather than absolute terms.

Table 3-4.

Social Assistance and Average Earnings1

(Deutsche mark per month)

article image
Sources: Presse- and Informationsamt der Bundesregierung, Sozialpolitische Umschau, Nr. 11/1994 (January 1994); and IMF staff estimates.

As of July 1992. Earnings are for blue-collar workers of Leistungsgruppe 3 and include child and rent allowances.

Ratio of social assistance to average earnings (in percent).

Assumes nonearning spouse.

Decisions about the poverty line and about the respective importance of absolute and relative poverty are ones on which nations can have very different views. However, in making these decisions it is important to bear in mind their consequences for the labor market. Table 3-4 suggests that social assistance may provide very high replacement ratios for many low-skilled workers, as a result of which work may not be worthwhile. Conversely, since the level of social assistance is one factor affecting wage setting. Table 3-4 can be interpreted as confirming that social assistance effectively sets a floor on wages economywide. Using the figure in Table 3-4 for social assistance for a west German with two dependents (approximately DM 2,150 a month) as the reservation net wage, and adding social security contributions of about 50 percent of net wages, suggests that the effective minimum labor cost in Germany may be DM 3,200 (about US$2,000) a month, or over DM 20 (about US$13) an hour.13 These minimum costs may exclude significant numbers of the unskilled or low skilled from (legal) employment and prevent some growth in sectors dependent on low-skilled labor, such as personal services.

Employment Protection

Segmentation of the labor market between the employed and the unemployed is often supported by employment protection for those with jobs. Such protection both increases insiders’ power by raising transaction costs and reduces firms’ willingness to hire by raising the fixed costs of employment. That there are serious obstacles to hiring in Germany is suggested both by employer surveys and by the fact that the German economy operates with considerable amounts of overtime even in periods of recession (Franz and König, 1986). Indeed, employment protection legislation in Germany is relatively severe and ill defined, and both the direct costs of dismissals and the uncertainties surrounding the courts’ interpretation of the relevant legislation have probably contributed to a reluctance to hire: a hiring mistake, whether in assessing an individual recruit or the appropriate level of employment overall, can be very costly.

Employment contracts in Germany are typically of indefinite duration. Dismissal protection applies to all permanent employees in firms with six or more workers (excluding apprentices), after a probationary period of six months. A notification period between four weeks and seven months is required, depending on the worker’s seniority and age.

Individual dismissals have to be “fair,” but criteria for fairness are only partially laid down in legislation. An employee may be dismissed for personal reasons, in particular, ineptitude; but in many cases personal reasons are not recognized as acceptable reasons for dismissal if the employer could fill the gap by other means, such as temporary reorganization or an increased workload for the remaining staff. Employees may be dismissed for redundancy only as a last resort—if the alternatives are “intolerable”—and the employer must use “social” criteria in choosing the redundant workers. It has been up to the courts to define both the “tolerable alternatives” to dismissals and the social criteria to be used in redundancies. In the latter case, criteria that have been adduced include period of employment, age, health, family responsibilities, spouse’s income, and wealth. The burden of proof regarding the fairness of a dismissal rests with the employer, and employees who contest a dismissal must be allowed to continue in employment until the case is decided. If a dismissal is found to be unfair, the employee either remains in the job or is paid compensation equivalent to at least 12 months’ pay (15-18 months’ pay for older workers with longer tenure in the firm).

Overall, for strictness of protection, Grubb and Wells (1993) rank Germany fifth among 11 OECD countries, behind 4 southern European countries (Portugal, Spain, Italy, and Greece). But perhaps even more important than the rules laid down in the law is the discretion of the courts in the area of employment protection, which has created considerable uncertainty for employers. The Deregulation Commission recommended in 1991 that the definition of the social criteria used in selecting employees for dismissal should not be left to the courts, but this recommendation has not been implemented.

Collective dismissals at firms employing more than 20 employees are subject to even more stringent conditions. If more than five employees are to be dismissed within 30 days, the Labor Office must be notified of the dismissal and can delay it by up to two months. More important, under the Works Constitution Act, if more than 20 percent of the workforce or more than 60 employees are to be dismissed, management and the Works Council must agree on a “social plan” that stipulates compensation for those workers who are to lose their jobs. If the parties cannot agree on a social plan, the case goes to binding arbitration. Since 1985, new firms have been exempt from the social plan requirement for the first four years of their existence. It has been estimated that payouts under social plans average four to six months of wages for a worker with average blue-collar industrial earnings. The Deregulation Commission had recommended that compensation under social plans be limited to damages directly related to the dismissal, but this proposal was not implemented.

The restrictions on dismissals also apply to workers on fixed-term contracts during the term of their contract, but the expiration of such a contract provides the employer an opportunity to reassess the employment decision. Fixed-term contracts are normally permitted in German law only under specified circumstances (such as for seasonal work in agriculture) and are not normally renewable. The Employment Promotion Act of 1985 made fixed-term contracts possible in any sector and under any circumstances, extended the maximum duration of fixed-term contracts from 6 months to 18 months (and 24 months for new small businesses), and made it possible to renew such contracts once. This law was due to expire in 1995, but its validity has recently been extended to the year 2000.


This chapter has suggested that only a small part of the German unemployment problem is cyclical in origin. More important, the west German labor market has become increasingly segmented, with a sharp division between the primary labor market and a marginal, high-unemployment, labor market consisting of the low skilled, those whose skills have been devalued by structural change (especially in the north), and older people. Economic recovery holds little hope for this secondary labor market, where a large amount of unemployment is likely to remain even when the primary labor market has tightened sufficiently to raise fears of rising inflation. The marginalization of people in the secondary labor market is worrisome in itself but also augurs ill for east Germany.

Both theoretical considerations and empirical studies of wage differentiation suggest that the German system of collective bargaining, which has served the country well in terms of overall wage moderation, has also resulted in wages that make it difficult for low-productivity firms and workers to participate in the labor market. The challenge now is to find, within the framework of centralized bargaining, ways in which relatively low-productivity firms can be assured better chances of survival (a challenge made especially pressing by the situation in east Germany) and ways in which wages at the bottom of the scale can react to the excess supply of low-productivity workers. Some progress has already been made in these respects, but much more is needed. Reform of unemployment insurance (especially the recently proposed limit on the duration of unemployment assistance), a reexamination of social assistance, and a reduction in the courts’ discretion in the area of employment protection would support such efforts by reducing reservation wages and lowering barriers between the two labor markets.

Appendix: Wage Relativities in West Germany—A Review of Empirical Studies

An examination of whether wage relativities are “adequate” in Germany is complicated by the fact that there is no clear standard for what constitutes “adequate” differentiation. Two approaches to this problem have been taken in the literature, one based on cross-country studies and the other on time-series analysis. Because both require something of a leap of faith when using them as a benchmark for what constitutes “adequate” differentiation in a particular country or time, their results can only be suggestive.

Cross-country studies (Bell, 1986; Freeman, 1988; Rowthorn, 1992) have compared levels of wage dispersion across sectors and concluded that Germany has a moderate degree of it, one that is much lower than that of the United States but generally higher than that of the Nordic countries. Bellmann and Möller (1993) and Zweimüller and Barth (1994), in recent studies that control for human capital factors in an attempt to isolate sectoral effects, come to a similar conclusion comparing Germany with Sweden and the United States, and with Austria, Norway, and the United States.14 To the—admittedly limited—extent that the United States, with its famously flexible labor markets, can be taken as a reference for what constitutes “sufficient” wage dispersion, these results suggest that German wage dispersion is insufficient.

Studies tracing changes over time have found that west German wage dispersion between sectors was stable or slightly increasing over the first half of the 1980s (Bell, 1986; Gundlach, 1986; Freeman, 1988; Rowthorn, 1992; also see Chart 3-9 of this chapter). Bellmann and Möller (1993), in a study controlling for human capital factors, also conclude that wage dispersion across sectors rose in the 1980s.

However, west German wage dispersion, even though rising in the 1980s, may still not have risen enough from the point of view of employment creation. Again, cross-country studies have been adduced to explore how much of an increase is enough. Bell (1986), Freeman (1988), and Rowthorn (1992) all provide comparable estimates for the rise in wage dispersion in west Germany and the United States, They all find wage dispersion rising considerably less fast in west Germany than in the United States and other countries. Of course, using the United States as a benchmark here requires not only the assumption that its labor markets are more flexible but also the assumption that the changes in the structure of advanced economies in the 1980s were very similar. Because the main forces that induced structural change in these economies were probably similar—technological change and increasing competition from developing countries in labor-intensive sectors—this assumption may serve as a first approximation.

One further clue as to whether wages have been sufficiently flexible across firms comes from the dispersion of wages across regions. Gundlach (1986) found that within most sectors the dispersion of wages across regions narrowed between 1973 and 1985 (see Chart 3-10 of this chapter).

Definitional problems complicate cross-country comparisons of wage differentials to the point of irrelevance. However, both OECD (1993) and Abraham and Houseman (1993) have examined trends in west German wage differentials by focusing on the relations between the 90th, 50th, and 10th percentiles of the earnings distribution, using social security data and data from the German Socio Economic Panel. Both studies find a significant narrowing of wage differentials, particularly at the bottom of the scale (see Table 3-2 of this chapter).

Both the Abraham and Houseman study and, to a lesser extent, the OECD study are open to the criticism that they do not take sufficient account of cyclical considerations as possible explanations for the changes in German wage differentials. The structure of earnings—especially the monthly earnings investigated in these studies—is not unrelated to the phase of the economic cycle. Monthly hours, and hence monthly earnings, of the low paid are more cyclically sensitive than those of the higher paid; for this reason alone, narrowing differentials of monthly earnings can be expected in a rising phase of the economic cycle. Also, recessions are normally associated with a relative lowering of earnings at the bottom of the scale, both because of slacker competition among firms for labor and because low-skilled workers are more likely to be laid off in a recession; a recovery would narrow wage differentials on this count as well.

While the 1983-90 period in Germany can be characterized as a long, slow recovery, two factors mitigate against a cyclical explanation of the narrowing of wage differentials at the bottom of the scale. First, the fragmentary data available for 1979-81, years when output is estimated to have been above potential, suggest that wage differentials narrowed during the 1980s even when comparing two cyclical peaks (around 1980 and around 1990). Second, despite a clear, though momentary, cyclical slowdown in 1987, the trend toward narrowing wage differentials continued unabated.

In many industrial countries, the 1980s saw increases in wage differentials by age. The evidence for this is mixed in west Germany. Based on aggregate data. Abraham and Houseman (1993) document a tendency for age-related earnings differentials to narrow slightly between the mid-1970s and 1988, with workers aged 15-19 gaining on those aged 20-29 throughout this period and workers aged 20-29 gaining on all older groups at least from 1983 onward. Similarly, they find a falling coefficient on age in their earnings functions between 1983 and 1988, but Bellmann and Möller (1993) find a rising coefficient on experience between 1979, 1984, and 1989.15

With age differentials playing a minor role at most, the narrowing of wage differentials at the bottom of the pay scale appears to be related primarily to a reduction in education-skill differentials (see Table 3-3 of this chapter). Abraham and Houseman (1993), using social security and GSOEP data classified by occupational qualification, conclude that there was a slight narrowing of the differentials between more and less educated workers from the mid-1970s to 1988. Both Abraham and Houseman (1993) and Bellmann and Moller (1993) also estimate standard earnings functions on panel data, using education and training dummies as well as age and other data. Both sets of results suggest some narrowing of education and training differentials in the second half of the 1980s, with the major change a decline in the premium for holding a vocational qualification (typically a completed apprenticeship).

Should such a narrowing of west German wage differentials be a source of concern from the point of view of employment, or was it an equilibrating response to changes in the supply and demand of different categories of workers? The study by Abraham and Houseman (1993) is the only one to address the issue of relative supply and demand for different categories of workers, but it finds declines both in the relative demand for low-skilled labor (using a shift-share analysis of the structure of output and thus ignoring another powerful source of the same effect, skill-biased technical change) and in the relative supply of low-skilled labor. It is not clear which of these declines was larger. It is striking, however, that among 17 OECD countries examined in OECD (1993), west Germany is the only one that shows a pronounced narrowing of wage differentials at the bottom of the distribution during the 1980s. Moreover, the high rates of unemployment for low-skilled workers strongly suggest that the narrowing of wage differentials was not merely an equilibrating response.


  • Abraham, Katherine G., and Susan N. Houseman, “Earnings Inequality in Germany,” NBER Working Paper No. 4541 (Cambridge, Mass.: National Bureau for Economic Research, 1993).

    • Search Google Scholar
    • Export Citation
  • Artus, Jacques R., “The Disequilibrium Real Wage Rate Hypothesis: An Empirical Evaluation,” Staff Papers, International Monetary Fund, Vol. 31 (June 1984), pp. 249 –302.

    • Search Google Scholar
    • Export Citation
  • Bell Linda A., “Wage Rigidity in West Germany: A Comparison with the U.S. Experience.” Federal Reserve Bank of New York Quarterly Review, Vol. 11 (1986), pp. 11 –19.

    • Search Google Scholar
    • Export Citation
  • Bell Linda A., Richard B. Freeman “Does a Flexible Industry Wage Structure Increase Employment? The U.S. Experience.” NBER Working Paper No. 1604 (Cambridge, Mass.: National Bureau for Economic Research, 1987).

    • Search Google Scholar
    • Export Citation
  • Bellmann, Lutz, and Joachim Moller, “Institutional Influences on Inter-Industry Wage Differentials,” Regens-burger Diskussionsbeiträge. No. 257 (1993).

    • Search Google Scholar
    • Export Citation
  • Bhaskar, V., “Wage Relativities and the Natural Range of Unemployment.” Economic Journal, Vol. 100 (1990).

  • Börseh-Supan, Axel, “Panel Data Analysis of the Beveridge Curve: Is There a Macroeconomic Relation Between the Rate of Unemployment and the Vacancy Rate?” Economica, Vol. 58 (1991). pp. 279 –97.

    • Search Google Scholar
    • Export Citation
  • Budd, Alan, Paul Levine, and Peter Smith, “Long-Term Unemployment and the Shifting U-V Curve.” European Economic Review, Vol. 31 (1987). pp. 296 –305.

    • Search Google Scholar
    • Export Citation
  • Burda Michael C., and Jeffrey D. Sachs, “Institutional Aspects of High Unemployment in the Federal Republic of Germany.” NBER Working Paper No. 2241 (Cambridge, Mass.: National Bureau for Economic Research, 1987).

    • Search Google Scholar
    • Export Citation
  • Calmfors, Lars, “Centralization of Wage Bargaining and Macroeconomic Performance: A Survey.” Seminar Paper No. 536 (Stockholm: Institute for International Economic Studies, 1993).

    • Search Google Scholar
    • Export Citation
  • Calmfors, Lars, John Driffill, “Bargaining Structure, Corporatism, and Macroeconomic Performance,” Economic Policy, Vol. 6 (1988), pp. 13 –61.

    • Search Google Scholar
    • Export Citation
  • Carruth, Alan, and Claus Schnabel, “Empirical Modelling of Trade Union Growth in Germany, 1956-86: Traditional Versus Cointegration and Error Correction Methods.” Weltwirtschaftliches Archiv, Vol. 126 (1990), pp. 326 –46.

    • Search Google Scholar
    • Export Citation
  • Coe, David T., “Nominal Wages, the NAIRU, and Wage Flexibility,” OECD Economic Studies, Vol. 5 (1985), pp. 87 –126.

  • Coe, David T., Thomas Krueger “Why Is Unemployment So High at Full Capacity? The Persistence of Unemployment, the Natural Rate, and Potential Output in the Federal Republic of Germany,” in German Unification: Economic Issues, IMF Occasional Paper 75 (Washington: International Monetary Fund, 1990).

    • Search Google Scholar
    • Export Citation
  • Davis, Steven J., “Cross-Country Patterns of Changes in Relative Wages,” NBER Macroeconomics Annual (Cambridge, Mass.: MIT Press, 1992).

    • Search Google Scholar
    • Export Citation
  • Elmeskov, Jorgen, “High and Persistent Unemployment: Assessment of the Problem and Its Causes,” Working Paper No. 32 (Paris: OECD Department of Economics, 1993).

    • Search Google Scholar
    • Export Citation
  • Flanagan, Robert J., David W. Soskice, and Lloyd Ulmann, Unionism, Economic Stabilization, and Incomes Policy: European Experience (Washington: Brookings Institution, 1983).

    • Search Google Scholar
    • Export Citation
  • Franz, Wolfgang, Arbeitsmarktökonomik (Berlin: Springer, 1991).

  • Franz, Wolfgang, Heinz König, “The Nature and Causes of Unemployment in the Federal Republic of Germany Since the 1970s: An Empirical Investigation,” Economica (Supplement 1986), pp. S219 –44.

    • Search Google Scholar
    • Export Citation
  • Freeman, Richard, “Unionism and the Dispersion of Wages.” Industrial and Labour Relations Review. Vol. 34 (1980).

  • Freeman, Richard, “Labour Market Institutions and Economic Performance.” Economic Policy, Vol. 4 (1988), pp. 64 –78.

  • Grubb, David, and William Wells, “Employment Regulation and Patterns of Work in EC Countries,” OECD Economic Studies, Vol. 21 (1993), pp. 7 –58.

    • Search Google Scholar
    • Export Citation
  • Gundlach, Erich, “Gibt es genügend Lohndifferenzierung in der Bundesrepublik Deutschland?” Die Weltwirtschaft (Tübingen), Vol. 1 (1986), pp. 74 –78.

    • Search Google Scholar
    • Export Citation
  • Jackman, Richard, and Stephen Roper, “Structural Unemployment,” Oxford Bulletin of Economics and Statistics, Vol. 49 (1987), pp. 9 –36.

    • Search Google Scholar
    • Export Citation
  • Jaeger, Albert, and Martin Parkinson, “Testing for Hysteresis in Unemployment: An Unobserved Components Approach,” Empirical Economics, Vol. 15 (1990), pp. 185 –98.

    • Search Google Scholar
    • Export Citation
  • Katz, Lawrence F., and Kevin M. Murphy, “Changes in Relative Wages, 1963-1987: Supply and Demand Factors.” Quarterly Journal of Economics, Vol. 107 (1992), pp. 35 –78.

    • Search Google Scholar
    • Export Citation
  • Landmann, Oliver Jerger, Jürgen “Unemployment and the Real Wage Gap: A Reappraisal of the German Experience,” Weltwirtschaftliches Archiv, Vol. 129 (1993), pp. 689 –717.

    • Search Google Scholar
    • Export Citation
  • OECD, Employment Outlook (Paris: Organization for Economic Cooperation and Development, 1991).

  • OECD, Employment Outlook Employment Outlook (Paris: Organization for Economic Cooperation and Development, 1993).

  • OECD, Employment Outlook (1994a), Quarterly Labor Force Statistics, No. 1 (Paris: Organization for Economic Cooperation and Development, 1994).

    • Search Google Scholar
    • Export Citation
  • OECD, Employment Outlook (1994b), Employment Outlook (Paris: Organization for Economic Cooperation and Development, 1994).

  • Paqué, Karl-Heinz, “Wage Gaps, Hysteresis and Structural Unemployment: The West German Labour Market in the Seventies and Eighties.” Working Paper No. 358 (Kiel Institute of World Economics, 1989).

    • Search Google Scholar
    • Export Citation
  • Paqué, Karl-Heinz, “Unemployment in West Germany: A Survey of Explanations and Policy Options,” Working Paper No. 407 (Kiel Institute of World Economics, 1990).

    • Search Google Scholar
    • Export Citation
  • Rowthorn, Robert E., “Centralization, Employment and Wage Dispersion,” Economic Journal, Vol. 102 (1992), pp. 506 –23.

  • Schultze, Charles L., “Real Wages, Real Wage Aspirations, and Unemployment in Europe,” In Barriers to European Economic Growth: A Transatlantic View (Washington: Brookings Institution, 1987).

    • Search Google Scholar
    • Export Citation
  • Soltwedel, Rüdiger, “Employment Problems in West Germany: The Role of Institutions. Labor Law, and Government Intervention,” Carnegie-Rochester Series on Public Policy, Vol. 28 (1988), pp. 153 –220.

    • Search Google Scholar
    • Export Citation
  • Streeck, Wolfgang, “Industrial Relations in West Germany, 1980-1987,” Labour, Vol. 2 (1988), pp. 3 –44.

  • Zweimüller, Josef, and Erling Barth, “Bargaining Structure, Wage Determination, and Wage Dispersion in Six OECD Countries.” Kyklos, Vol. 47 (1994). pp. 81 –93.

    • Search Google Scholar
    • Export Citation

more detailed examination of cast German developments and prospects can be found in Chapter VI.


A second commonly reported rate uses only the dependent civilian labor force (excluding the self-employed); indeed this is the most frequently quoted rate for east Germany. This rate is not used in this chapter.


The NAIRU concept is useful as long as there is a relation between unemployment and changes in inflation, regardless of its theoretical underpinnings—be they new classical, where deviations in unemployment from the equilibrium rate are associated with errors in expectations, or derived from models of monopolistic competition, where changes in inflation result from inconsistent claims by wage setters and price setters.


Schultze (1987) uses the OECD-slandardized unemployment rate and on this (lower) basis estimates the NAIRU in 1983-87 at 6 1/2 percent.


For purposes of most regional comparisons in this chapter, the very smallest Lander are grouped with neighboring Lander, and Bavaria is divided into northern and southern Bavaria. Thus, the “northern” (really northwestern) Lander comprise Schleswig-Holstein (including Hamburg). Lower Saxony (including Bremen). North-Rhine-Westphalia. Rhincland-Palatinate (including Saarland), and Berlin-Brandenburg: the “southern” (really southeastern) Lander comprise Hessen, Baden-Wümemberg, northern Bavaria, and southern Bavaria. The “central” slates of Rhineland-Palaiinate and Hessen are allocated according to structural features: Rhineland-Palatinate shares an important concentration on heavy industry with the northern slates.


Vacancy rates need to enter into measures of mismatch because there is no regional imbalance—merely differences in (inefficiency of job search and intermediation—if high unemployment and vacancy rales coincide.


Streeck (1988) suggests that wage negotiations are formally conducted at the regional level largely so that (costly) strikes can be limited to “pilot areas.” To obtain uniform conditions across the country, the union relies on the incentives for the employers’ association to eliminate wage-based competition and thus to make all its regional affiliates sign similar agreements.


Centralized systems are often also associated with low wage dispersion, which has detrimental effects on employment. These effects are discussed in the next section.


The use of trend productivity growth circumvents the problems associated with the effect of wages on productivity through short-run fluctuations in employment but not the effect through capital accumulation. Indeed, Landmann and Jerger favor a capital shortage explanation of unemployment.


A further complication is that the structure of earnings is sensitive to the economic cycle. Monthly earnings especially are more cyclically sensitive for the lower paid than for the higher paid. But recessions are also normally associated with a relative lowering of hourly earnings at the bottom of the scale, both because of slacker competition for labor and because low-skilled workers are more likely to be laid off in a recession. For these reasons, comparisons highlighted in this chapter are peak-to-peak (1979-91) wherever possible.


All the Lander are considered separately in this comparison; the range quoted excludes Bremen, a northern city-state where wage growth reached 4,9 percent.


Greater freedom to reduce working hours also reduces the fixed costs of employment and should help to reduce employers’ reluctance to hire.


Social security contributions add up to about 40 percent of gross wages, of which half is paid by the employee (out of gross wages) and half by the employer (in addition to gross wages), Thus, even abstracting from income tax, the wedge between net wages and total labor costs is about 50 percent. Income tax currently payable on gross wages up to the amount of social assistance is to be abolished by order of the Constitutional Court.


Human capital factors can bias aggregate comparisons of wage dispersion. For instance, a country where highly paid workers (say, more educated workers) are distributed more unevenly across sectors will show a lower degree of wage dispersion than one with a more even distribution across sectors, even if in each sector the two countries’ pay scales are really identical. The most careful studies of dispersion, therefore, look not at average earnings in different industries but at the coefficients on industry dummies in standard earnings functions.


Nor is it clear how one should interpret trends in age differentials in Germany. Because the labor market situation is worse for older than younger people, the declining age differentials found by Abraham and Houseman might signal an appropriate responsiveness of wages to labor market conditions. However, more experienced people are also those most likely to be paid above-tariff wages, so that responsiveness in this area would not necessarily indicate a similar responsiveness in labor markets where tariff wages are binding.

The First Five Years: Performance and Policy Issues