46. The seemingly inexorable upward trend in unemployment since the early 1970s is arguably the most pressing economic problem in Germany. Since unification in 1990, labor market trends have weakened further, even abstracting from the unification-induced employment collapse in eastern Germany. Employment contracted by a cumulative 7 percent during 1991–97, and the unemployment rate rose to a postwar record of close to 12 percent at end-1997. Moreover, the recovery that began in 1993 has been unusual in that aggregate employment has contracted rather than increased after a lag.


46. The seemingly inexorable upward trend in unemployment since the early 1970s is arguably the most pressing economic problem in Germany. Since unification in 1990, labor market trends have weakened further, even abstracting from the unification-induced employment collapse in eastern Germany. Employment contracted by a cumulative 7 percent during 1991–97, and the unemployment rate rose to a postwar record of close to 12 percent at end-1997. Moreover, the recovery that began in 1993 has been unusual in that aggregate employment has contracted rather than increased after a lag.

II. Labor Market Trends Since Unification: A Vicious Circle Perspective25

A. Introduction and Summary

46. The seemingly inexorable upward trend in unemployment since the early 1970s is arguably the most pressing economic problem in Germany. Since unification in 1990, labor market trends have weakened further, even abstracting from the unification-induced employment collapse in eastern Germany. Employment contracted by a cumulative 7 percent during 1991–97, and the unemployment rate rose to a postwar record of close to 12 percent at end-1997. Moreover, the recovery that began in 1993 has been unusual in that aggregate employment has contracted rather than increased after a lag.

47. This chapter has two main themes: First, Germany’s aggregate labor market trends mask a marked disparity in disaggregated employment and unemployment developments. In particular, employment flows disaggregated by earnings and skill characteristics show that job losses in the lower half of the earnings/skill distribution more than accounted for the contraction in aggregate employment. Most of the employment losses in the lower half of the earnings/skill distribution were offset by large-scale job creation in the upper half. These disparate employment trends, which antedate unification, highlight the need to look beyond highly aggregated models of labor market behavior and the policy prescriptions based on these models.26

48. The chapter’s second theme is the interaction between adverse labor market trends at the lower end of the earnings/skill distribution and the spending and financing side of Germany’s comprehensive social insurance/protection system. The chapter develops a stylized labor market model illustrating how labor shedding at the lower end of the earnings/skill distribution due to a mismatch of labor productivity and labor cost can trigger a vicious circle of higher social spending, increases in social contribution rates, additional employment shedding at the lower end, and further increases in social spending. Moreover, in this model, an adverse exogenous shock to the finances of the social insurance/protection system that causes substantial increases in social contribution rates—as exemplified by the impact of German unification on social insurance finances or population aging—can also trigger a circular chain of declining employment and rising social spending.27

49. To put Germany’s labor market trends since unification into perspective, Section B lays out some stylized facts. The trends in aggregate output, employment, unemployment, and real wages suggest that labor market trends in western Germany have since 1991 shared many of the broad features of trends observed before unification, including: weak employment growth; a sharp increase in the level of the unemployment rate; rapidly rising labor productivity; and real wage growth that lags behind labor productivity growth. At the same time, the deterioration of labor market conditions has been heavily concentrated in the lower half of the earnings/skill distribution. The underlying labor market developments have been qualitatively similar in western and eastern Germany, although the employment contraction and the rise in unemployment have been much more pronounced in eastern Germany. Finally, while aggregate output growth since unification has been broadly similar in Germany and its EMU partners, employment has slackened markedly more in Germany, suggesting that slow aggregate demand growth cannot explain Germany’s adverse labor market trends.

50. Section C develops a simple analytical model to shed light on these stylized facts. The model suggests that labor market trends in Germany are partly explained by the interplay of four key factors: (i) labor productivity of lower-skilled workers lags productivity increases of the better-educated portion of the work force; (ii) wage bargaining leads to real wage increases that are similar across different segments of the skill distribution; (iii) social benefits of effectively unlimited duration provide a fallback position for workers priced out of the market; and (iv) the financing of social benefits requires the levying of social insurance contributions that fall proportionately on all workers across the different segments of the skill distribution.

51. In this stylized model setting, the employment opportunities of lower-skilled workers deteriorate over time because of the mismatch between productivity and labor cost at the lower end of the earnings/skill distribution, and the unemployment rate of these workers trends upward. These implications of the model are consistent with historical developments. Moreover, if perturbed by a shock that increases social contributions, the model would predict a circular chain of falling employment at the lower end of the skill distribution and further increases in social spending and contribution rates, which is consistent with the post-unification experience in Germany.

52. Section D draws attention to two additional dimensions of Germany’s labor market trends that are not captured by the stylized model—sectoral reallocation of labor and the duration of business cycle recessions. Sectoral data on employment losses/creation indicate that the shedding of lower-skilled workers took place in the manufacturing sector, mainly during the cyclical downturn, while employment creation almost exclusively occurred in the service sector. Moreover, job creation in the service sector in Germany appears to have been concentrated in the upper segment of the skills distribution. In view of the longer-term trend of a declining work force in the manufacturing sector, labor needs to be reallocated to the service sector. However, the pace of the sectoral reallocation of labor appears to have been insufficient—contributing to the observed upward ratcheting of the unemployment rate over business cycles (asymmetric hysteresis). Moreover, the duration of periods of deficient aggregate demand—defined as the number of years where the output gap is negative—increased substantially in Germany in the 1980s and 1990s, compared with the 1960s and 1970s, and may have constituted an additional factor accounting for asymmetric hysteresis in aggregate unemployment.

53. The four key elements of the model provide the background for a brief discussion in Section E of the scope and limits of widely discussed policy options for addressing Germany’s labor market problems:

  • First, policies can try to tackle the root of the disparate labor market developments—lagging labor productivity growth at the lower end of the skill distribution by increased training and education to lift labor productivity at the low end of the earnings/skill distribution.

  • Second, the mismatch between labor productivity and labor cost at the lower end can be ameliorated by more wage differentiation.

  • Third, the incidence of nonwage labor cost at the lower end of the skill/earnings distribution can be lowered through targeted public intervention that subsidizes labor cost at the lower end, including the tapering of social contribution rates or explicit wage subsidies.

  • Fourth, incentives to seek work can be strengthened by limiting the duration of nonwork benefits and/or smoothing poverty traps.

54. The sectoral and business cycle dimensions of Germany’s labor market trends brings out two additional policy levers to address Germany’s labor market problem: (i) the fostering of a regulatory environment that promotes faster growth of service sector employment to speed up the absorption of workers laid off in the manufacturing sector; and (ii), given the greater difficulty of reallocating labor during a prolonged cyclical downturn, macroeconomic stabilization policies that keep periods of deficient aggregate demand (or negative output gaps) reasonably short would help mitigate hysteresis effects in unemployment.

55. This chapter’s main finding that Germany’s adverse labor market trends are largely concentrated at the lower end of the earnings/skill distribution underscores the potential benefits of a targeted approach to labor market reforms. Moreover, as pointed out by Coe and Snower (1997), policy measures in one area can have “complementary effects,” suggesting that successful labor market reforms should be combined in a comprehensive package. Finally, the stylized model developed in this chapter highlights an opportunity for labor market reforms to set in motion a virtuous circle between lower unemployment, improved financial position of the social insurance/protection system, lower social contribution rates, and still lower unemployment.

B. Labor Market Trends: Some Stylized Facts

56. Aggregate trends in the labor market of Germany and other industrial countries have been at the focus of an extensive literature and are therefore only briefly summarized here.28 During 1970–97, the level of employment in western Germany remained almost flat, while real GDP, and hence labor productivity, rose by some 80 percent (Figure II-1),29 The unemployment rate ratcheted upward from the early 1970s, while the employment rate (defined as the share of working age population aged 15–64 that is employed) remained broadly stable at around 68 percent. Aggregate real wages in western Germany also increased sharply, keeping pace with productivity gains until the mid-1980s. However, during 1986-97 average real wages lagged productivity gains cumulatively by 11 percentage points. Thus, with stable employment and lagging real wage growth, the labor share in national income has been declining since the 1980s.30

Figure II-1.
Figure II-1.

Germany: Aggregate Labor Market Trends, 1970–97 1/

Citation: IMF Staff Country Reports 1998, 111; 10.5089/9781451810325.002.A002

Source: IMF, World Economic Outlook.1/ For 1991-97. data refer to western Germany.2/ Defined as employed population aged 15–64 years old as a percent of the population of working age.3/ Average gross wages per employee deflated by GDP.4/ Wage share adjusted for relative shifts in dependent and self-employed work force.

57. Employment and unemployment developments in eastern Germany since unification were qualitatively similar to those in western Germany, although there were marked differences in the magnitude of the decline of employment and the level of the aggregate unemployment rate (Figure II-2).31 Unemployment in eastern Germany rose sharply after unification, and it has become entrenched at a high level as the convergence process in the new Lander has been much slower than anticipated following excessively rapid wage convergence in the early 1990s, The initial fall in employment in eastern Germany also reflected a sharp decline in labor force participation rates, owing partly to a normalization of the exceptionally high participation rates in the former German Democratic Republic and to higher hidden unemployment.

Figure II-2.
Figure II-2.

Germany: Aggregate Labor Market Trends in Western and Eastern Germany, 1991–97

Citation: IMF Staff Country Reports 1998, 111; 10.5089/9781451810325.002.A002

Source: IMF, World Economic Outlook.

58. Germany’s aggregate employment contraction since unification has been, however, overstated because the official employment statistics do not include the sharp increase in the number of “small-time jobs.” Small-time jobs are fully exempt from paying social insurance contributions if the work time is less than 15 hours per week and monthly pay does not exceed DM 620 (DM 520 in the new Länder) per month. According to estimates based on the employment survey data of the socio-economic panel, the number of small-time jobs amounted to 6.7 million in 1996, an increase of 37 percent from their estimated level in 199132 Simply adding the estimated number of small-time jobs—which is subject to a considerable margin of statistical uncertainty—to official employment numbers would imply a smaller employment contraction—for example, the cumulative decline in employment during 1991–96 of 5½ percent for western and eastern Germany would be reduced to 3 percent.

59. Since unification, Germany’s aggregate labor market trends have also been relatively weak compared to developments in most of its EMU partners (Table II-1). Cumulative employment losses in Germany during 1991–97 were significantly higher than in most other EMU partner countries except Italy and Portugal. The contrast in labor market developments between Germany and Austria is particularly noteworthy. Despite similar institutional arrangements in their respective labor markets and similar real output growth trajectories, aggregate employment in western Germany contracted much more sharply than in Austria. The job losses in Germany also contrast with developments in countries with more flexible labor markets, such as Ireland and the Netherlands.

Table II-1.

Germany: Cumulative Growth Rates of GDP, Employment, and Labor Productivity, 1991–97

article image
Source: IMF, World Economic Outlook, EU, European Economy, 1997; and staff calculations.

All EMU countries excluding Germany.

60. Disaggregating labor market trends by skills indicates that labor shedding and rising unemployment in western Germany was heavily concentrated in the lower portion of the skill distribution. In particular, workers with low skills have borne the brunt of the employment contraction, even though they represent a declining proportion of the labor force (Figure II-3). During 1976–95, employment of lower-skilled workers in western Germany fell by about 45 percent, while employment of higher-skilled employees increased by about 120 percent. Over the same period, the unemployment rate of workers with lower skills increased from 6 percent in 1975 to 20 percent 1995, while the unemployment rate for workers with higher skills rose from 2 percent to about 4 percent although their share in the labor force had increased markedly. The position of less skilled workers deteriorated particularly sharply since unification. During 1991–1995 employment of workers with lower skills fell by a cumulative 10 percent compared with a cumulative decline in aggregate employment of only 2.5 percent. In eastern Germany, the adverse labor market trends for less skilled labor were even more pronounced, with the unemployment rate of workers with lower skills rising to 45 percent in 1995 relative to an aggregate unemployment rate of 14 percent in the same year.

Figure II-3.
Figure II-3.

Germany: Unemployment Rates and Employment by Skill Level, 1976–95 1/

Citation: IMF Staff Country Reports 1998, 111; 10.5089/9781451810325.002.A002

Sources: Reinberg (1997); and staff estimates.1/ Data refer to western Germany.

61. Measures of skills based on schooling characteristics provide, however, only broad indicators of the actual skill distribution. For example, the characteristic “medium skills” covers persons with an apprenticeship education, a group that accounts for some 70 percent of the employed. However, apprenticeship programs can range from very advanced to elementary.33 Additional evidence on the incidence of employment losses across the skill distribution can be won by considering employment changes in different earnings brackets during 1991–95 (Figure II-4). Most strikingly, in western Germany, employment for those in the lower half of the earnings distribution declined by 2.9 million during 1991–95, while employment in the upper half of the earnings distribution increased by 2.7 million.34 In eastern Germany, employment losses were even more concentrated in the lower earnings brackets, although this partly reflects the lower wages paid in eastern Germany.

Figure II-4.
Figure II-4.

Germany: Cumulative Employment Changes in Different Earning Brackets, 1991–95

(In thousands)

Citation: IMF Staff Country Reports 1998, 111; 10.5089/9781451810325.002.A002

Source: Reinberg (1997).

C. A Stylized Model of the Labor Market

62. This section develops a stylized model to interpret the concentration of adverse employment trends in the lower part of the earnings/skill distribution. An intuitive discussion of the model is provided in this section, while the formal model is presented in the appendix. The model tries to bring out the interplay of four key factors: (i) labor productivity of lower-skilled workers lags productivity increases in the upper part of the skill distribution; (ii) wage bargaining equalizes real wage increases across the skill distribution and introduces a wage floor; (iii) social benefits of unlimited duration provide a fallback position for unemployed workers; and (iv) social benefits are financed by proportional social insurance contributions.

63. In the model, workers are ordered by their productivity from the lowest to the highest skill level, and the productivity differential between higher and lower skilled workers widens over time. Thus, labor productivity (or the marginal product of workers) in the lower portion of the skill distribution is assumed to increase at a slower pace than in the upper portion of the distribution. This stylized assumption can be motivated by skill-biased technological progress, reflecting, e.g., increased demand for skills at information handling or the ability to learn on the job (learning to learn).35

64. Workers are paid their marginal product. At the same time, wage bargaining partners are assumed to have a “preference for equality,’ resulting in a effective wage floor. The assumptions of broadly constant wage differentials between the high and the low end of the distribution and an effective wage floor can be motivated by the implications of the wage setting approach pursued by the social partners.36 Evidence of stable wage differentials generated by the collective bargaining system in Germany is provided by the similar rates of increase in wages at different skills and earnings deciles (Figure II-5). Moreover, data on the shape of the distribution of hourly wages in Germany reported in OECD (1994) suggest that it is characterized by a distinct wage floor.37 Recent empirical work by Fitzenberger and Franz (1997) indicated that increased wage differentiation could substantially lower unemployment among the lower skilled. In particular, these authors’s calculations suggest that equalization of the 1991 unemployment rates across lower and medium skills (based on the skill categories used above) in western Germany would, e.g., have required an additional wage dispersion of 15 percent between the median wages of the lower skilled and of those with medium skills.

Figure II-5.
Figure II-5.

Germany: Wage Differentiation 1/

Citation: IMF Staff Country Reports 1998, 111; 10.5089/9781451810325.002.A002

Source: Statistisches Bundesamt; United States Bureau of Labor Statistics; and Institut fur Weltwirtschaft Kiel.1/ Data refer to western Germany.2/ D5/D1 denotes ratio between the fifth and the first deciles of the male earnings distribution. Similarly, D9/D5 denotes ratio between the ninth and the fifth deciles of the male earnings distribution.

65. Unemployment benefits are assumed to be a proportion of the net minimum wage. Any increase in the level of unemployment benefit is financed by corresponding increases in social contribution rates. With increasing social contribution rates the difference between average and minimum wages shrinks and the wage distribution narrows. An increase in the duration of entitlement to unemployment benefits, which represents an increase in the discounted value of the future stream of benefits, implies the same outcome.

66. In practice, unemployed workers in Germany have recourse to three types of social benefits: unemployment compensation (Arbeitslosengeld) funded by social contributions; unemployment assistance (Arbeitslosenhifle) and social assistance (Sozialhilfe) both funded by general taxation.38 The replacement ratio of unemployment compensation for claimants with children (without children) is 67 percent (60 percent) of previous net earnings. For those who exhaust unemployment compensation or do not qualify for it, means-tested unemployment or social assistance is available for an indefinite period.

67. Generous unemployment benefits, in particular long and indefinite entitlement periods, were found to be empirically significant in affecting the length of unemployment spells. Hunt (1995) examined the disincentive effect associated with the unlimited duration of unemployment assistance in Germany by comparing the exit probability from unemployment (hazard rate) to a control group unaffected by the extension of the duration of unemployment benefits. The level of benefits was not found to affect the probability of exiting from unemployment. However, longer unemployment benefit durations increased unemployment durations. Nonetheless, in this empirical work, the duration of benefits accounted for less than one third of the difference between unemployment spells in the United States and Germany in the 1980s.

68. The finding that older individuals and those who exit the labor force experience longer unemployment spells highlights the interrelation between the various elements of Germany’s social insurance/protection system. In particular, some employers, seeking to reduce their workforce in response to structural change and unification shocks, may have encouraged older workers to leave the work force via unemployment. The combination of severance pay, high unemployment insurance benefits, long entitlement periods, and the payment of both pension and health insurance contributions for the unemployed by social insurance made unemployment a bridge to early retirement. Thus, some reduction in the workforce was achieved at the expense of the social benefits system.

69. In the model, the social insurance/protection system is financed through flat contribution rates levied across the entire income distribution—there is no lower threshold for, and no upper cap on, contributions.39 The incidence of social contributions falls on workers, and the combination of a wage floor and increasing social contribution rates can price workers out of the labor market at the lower end of the wage distribution.

70. Endogenizing tax/contribution rates by specifying a government budget constraint that links tax revenues and social expenditures generates a dynamic process that can be vicious or virtuous. Higher unemployment requires additional spending on social benefits. To finance these social benefits, higher contribution rates must be levied. The gross minimum wage rises with the higher tax burden, further increasing unemployment. The lower skilled lose their jobs, pushing up average productivity and average wages of the employed. Since the minimum wage is linked to the average wage in the model, the minimum wage is increased further, resulting in still more unemployment and higher taxes/contributions.

71. In this model, German unification can be represented as an adverse exogenous shock to the social contribution rate. After unification, western Germany’s social institutions were extended to eastern Germany’s workers, and the sudden increase in the number of beneficiaries in the social insurance system required higher contributions and taxes. Social security contribution rates rose by 6 percentage points during 1990–97.

D. Sectoral Reallocation of Labor and Business Cycle Durations

72. Two additional dimensions of Germany’s labor market trends that are not captured by the stylized model in Section C are the sectoral reallocation of labor from industry to the service sector and the longer duration of business cycles during the 1980s and 1990s. Since at least the early 1970s, manufacturing employment (including construction) in Germany has been on a downward trend relative to total employment (Figure II-6).40 A decline in manufacturing employment would not be problematic if service sector employment expanded sufficiently to absorb employment losses in the manufacturing sector. Indeed, service sector employment grew by a cumulative 60 percent during 1970-96. This was, however, insufficient to absorb the heavy losses in manufacturing employment. In fact, movements in the loss of jobs in manufacturing and in the overall unemployment rate are closely linked (Figure II-6). In each recession since 1970, manufacturing employment declined sharply and the aggregate unemployment rate ratcheted upwards. Thus, from a sectoral angle, the combination of massive labor shedding in the manufacturing sector and sluggish employment growth in services largely accounted for the upward ratcheting of the aggregate unemployment rate in Germany.41

Figure II-6.
Figure II-6.

Germany: Sectoral Employment Trends, 1970–97 1/

Citation: IMF Staff Country Reports 1998, 111; 10.5089/9781451810325.002.A002

Source: Statistisches Bundesamt Wiesbaden.1/ Data refer to western Germany.

73. Data available for the period 1991–95 shed light on the skill composition of sectoral employment changes (Table II-2). Employment shedding in the secondary sector (manufacturing including construction) in both western and eastern Germany was heavily concentrated in the lower and medium skill categories. For example, the secondary sector in western Germany lost 1.4 million jobs during 1991-95; most of the workers affected by job losses had low or medium skill characteristics.42 In contrast, employment in the tertiary sector (services) increased by some 1.3 million in western Germany, keeping employment losses to only 0.2 million. Moreover, the increase in tertiary sector employment was concentrated in the higher-skilled and higher-paid job categories.

Table II-2.

Germany: Cumulative Employment Changes by Sector and Skill, 1991-95

article image
Source: Reinberg (1997) and staff calculations.

No formal qualifications.

Medium qualification: vocational and apprenticeship education.

University education or equivalent degree.

74. The sectoral data on employment losses/creation disaggregated by skill and earnings characteristics add an important sectoral facet to Germany’s labor market dynamics. Employment shedding typically takes place in the manufacturing sector among the lower skilled/paid during a cyclical downturn. At the same time, employment creation during the cyclical downswing is concentrated in the service sector among the higher-skilled/better-paid. Moreover, in view of the long-term de-industrialization, this is not just a cyclical phenomenon that involves a spell of temporary unemployment and reabsorption during the recovery. Labor laid off in the manufacturing sector needs to be permanently shifted to the service sector. Moreover, as already discussed in Chapter I, the duration of recessionary periods—defined as periods where the output gap is negative—lengthened substantially in Germany in the 1980s and 1990s compared to the 1960s and 1970s (Figure II-7). Extended periods of deficient aggregate demand can slow the labor reallocation process, and the unemployed may experience a deterioration in their skills and/or work ethic, a phenomenon particularly relevant for the lower skilled. The combination of slow sectoral reallocation of labor and prolonged periods of negative output gaps may have contributed to the observed upward ratcheting of the aggregate unemployment rate during business cycles, a phenomenon that is sometimes referred to as “asymmetric hysteresis” in unemployment.43

Figure II-7.
Figure II-7.

Germany: Business Cycle Durations, 1964-97

Citation: IMF Staff Country Reports 1998, 111; 10.5089/9781451810325.002.A002

Sources: IMF, World Economic Outlook;1/ “Booms” are defined as periods with positive output gaps. “Recessions” are defined as periods with negative output gaps.

E. Policy Options

75. The stylized model provides a convenient template for discussing broad policy options grouped around its four key elements: (i) increased training and education that could lift labor productivity at the lower end of the skill distribution; (ii) more flexible wage bargaining to ameliorate the mismatch between labor productivity and labor costs of the lower skilled; (iii) enhancing job search incentives through lowering reservation wages; and (iv) policy measures to reduce nonwage labor costs targeted at the lower paid/lower skilled.

76. The importance of enhanced education and training to upgrade the skills of workers within or outside the context of active labor marker policies (ALMPs) has been a longstanding theme in policy discussions.44 In the long term, a permanent reduction in the demand for low skilled workers requires (at a fixed real wage) a shift in the skill distribution of the labor force toward skilled workers. Some of the unemployed may not have the financial resources to acquire adequate training and education. At the same time, the potential public cost of a large-scale education and training program aimed at the unskilled could be quite high. Although no specific estimates are available for Germany, those derived for the United States may well illustrate the magnitude of these cost.45 However, as pointed out by Nickell and Bell (1996), the skill deficiencies at the lower end of the labor market in Germany appear to be significantly less pronounced than in the United States, which would lower this cost estimate. Moreover, there exist several other less expensive options to sharpen incentives to acquire skills. For example, making unemployment assistance conditional on participation in training would increase the incentives to acquire skills. Moreover, existing education programs could focus more directly on the acquisition of marketable skills.

77. Active labor market policies (ALMP) generally target the unemployed and aim at increasing the employability of workers by providing training, job brokering and placement services, and direct job creation in the public sector. Additional training of the existing work force can fulfil two basic functions. First, it can maintain or increase the skills of the unemployed in times of recession. As such, training reduces the number of discouraged workers and exerts a positive effect on the aggregate supply of labor. Second, training can help overcome structural imbalances in the labor market by adjusting labor supply to demand. And third, re-training may facilitate the sectoral reallocation of labor. Empirical research has, however, detected only weak favorable employment effects of ALMPs. Calmfors and Skedinger (1995), e.g., used pooled time-series and cross-section data from 24 Swedish regions and found that the effect of training on the total jobless rate was positive or negative depending on the equation specification. At the same time, training programs were found to have a slightly more favorable effect on the total jobless rate than public sector job-creation schemes.

78. Greater wage differentiation to accommodate the widening of skill differentials would provide more scope for reabsorbing the unemployed subject to two qualifications. First, even if tariff wages reflected market pricing of labor, generous social benefits could still foster a reservation wage that keeps unemployment high among the lower skilled. And second, increased wage flexibility at the lower end of the wage distribution could result in real wage levels for the lower skilled that are deemed “too low” to allow workers to support themselves and their families.46

79. There is some evidence of increased relative wage flexibility. In an attempt to accommodate productivity differentials among workers, recent collective wage agreements in the western German chemical industry incorporated a provision that allowed new recruits to be paid (for one year) wages 5–7½ percent below tariff wages and 10 percent below tariff wages if they were recruited from long-term unemployment. Further flexibility of the collective wage bargaining system at the firm-level could allow wages to react to local labor market conditions and allow workers freedom to agree on more flexible remuneration methods and working time arrangements. Enterprise bargaining could also accommodate productivity differentials between workers by allowing greater differentiation in tariff wage increases across skill levels. Profit-sharing could also permit a lower base wage at times of downturn which would limit labor shedding.

80. Reforms of the German unemployment benefit system to increase work incentives could take two general dimensions—lower benefits and/or reduced durations.47 Reductions in unemployment compensation are not considered a high priority for two reasons. First, replacement ratios in Germany are not particularly high by European standards.48 According to OECD data, the net benefit replacement ratios for single-earner households with no children are the lowest in the EU except for Ireland and Italy. Second, social assistance replacement ratios form a floor for unemployment benefits. Therefore reform may need to focus on the shortening of benefit durations and, in particular, on the reform of open-ended unemployment assistance.

81. Social security contributions in Germany have increased substantially since the 1970s and raised labor costs, exerting downward pressure on labor demand. To expand the demand for labor, especially unskilled labor, non-wage costs could be reduced. The potential employment impact may be significant given that the demand elasticity for unskilled labor seems to be substantially higher than for skilled.49 One option would be to taper social contribution rates at monthly earning levels above DM 620 to reduce nonwage labor cost for these workers. However, this approach would weaken the relatively close link between social contributions and insurance benefits—a key characteristic of the design of Germany’s social insurance system. Another option would be to directly subsidize wage income for low-paid workers (through what is referred to in Germany as combi-wages).50 Supplementing low wage income by a subsidy could provide incentives for employers to create additional low-skilled jobs where currently wage rates exceed marginal productivity. For the employee, combi-wages would reduce welfare traps and could provide incentives to accept work in the lowest pay grades of collective agreements. Such wage subsidies should be strictly targeted at the low end of the wage scale, but would also need safeguards against abuse by firms and employees. Some abuse can be discouraged by limiting the subsidies to one or two years and by paying only part of the wage-productivity difference. There is, however, the danger that tariff 1 partners might, at the expense of the public finances, raise wages in the lowest pay grades in response to a combi-wage subsidy.

82. The sectoral dimension of the unemployment problem in Germany highlights another policy lever: the need to provide a regulatory environment to foster service sector employment growth. Specific policy measures in this area could include further streamlining of the approval process for new start-up businesses, relaxing licensing practices that curtail competition, and a further extension of shop opening hours, which remain among the most restrictive in Europe. Training of laid-off manufacturing workers aimed at developing the skills needed for service production may also be helpful.

83. The sectoral dimension also underscores the benefits of macroeconomic stabilization policies that keep recessionary periods reasonably short. Although asymmetric hysteresis in unemployment would imply that expansionary stabilization policies can not lower the structural rate of unemployment, shortening the duration of recessionary periods could reduce the “spillover” of cyclical into structural unemployment.

84. As discussed in Coe and Snower (1997)—policy measures of a broad package of reforms will have mutually reinforcing effects. Moreover, the model also indicates the possibility of triggering a virtuous circle between favorable labor market trends and the fiscal situation. Conversely, institutional rigidities reinforce and complement one another. For instance, employers’ social contributions discourage job creation and finance social benefits that discourage job search. The existence of these institutional complementarities lead Coe and Snower (1996) to suggest that the unemployment problem required a package of policies. Such a complementary package has a greater impact on unemployment than the sum of the individual measures.

85. Effective labor market reform would utilize the interrelations among the various policy measures. A comprehensive package of measures would have a greater impact on unemployment, and a virtuous cycle could be set in motion. For example, a shortening of the entitlement to unemployment benefits would increase job search; place downward pressure on wage bargaining, and cause unemployment to fall This fall would be amplified because the decline in unemployment further lowers government expenditure and contribution rates, which in turn reduce labor costs.

86. However, the design of such a package must also pay due attention to political institutions that underlie present labor market rigidities and therefore may hamper reform. Saint-Paul (1997) notes that from a politico-economic perspective, coordination failures are particularly important. Coordination failures arise when agents make decisions without taking into account the actions of others. Lobbyists or unions may create devices to protect their clientele from adjustment to negative shocks. However, if only one sector in the economy is flexible, it bears the entire adjustment cost, making this sector more likely to push for protection. The opportunity cost of introducing rigidities falls as the rest of the economy becomes less flexible.

APPENDIX II-1: A Model of Labor Market Trends and Vicious Circles

Skills and Productivity

87. There is a continuum of workers on the interval [0,1] ordered by their labor productivity. At time t, the productivity of worker i is

where dt is an increasing function of time. Output is produced with a constant returns to scale technology,

Yt = iεI (c + dti)di

where I denotes the set of employed workers.

Wage formation

88. Productivity is observable, and workers are paid their marginal product

wi, t = c + dti

A minimum after-tax wage wmin (1-τ) is set as a portion ρ ε [0,1] of the gross average wage wa.

wmin(1-τ) = ρwa

This relationship can be thought of as a “fair wage norm” or subsistence wage as a function of average productivity. Net minimum wages are assumed to be a function of the gross, rather than the net, average wage to avoid a drop in the subsistence wage whenever taxes increase. Assuming tax rates are unchanged, the ratio between the average and the minimum gross (and net) wage is constant. With increasing taxes, the difference between average and minimum wages will shrink, narrowing the wage distribution.

89. All workers with productivity higher than the gross minimum wage wmin will be employed. Denoting the marginal worker by i* (omitting the time index for simplicity), the average wage is determined as

wa = ½(wmax + wmin) = c + d(1+i*)/2

90. Combining this equation with the equation for minimum wages determines the marginal worker i*. All workers with an index above i* will be employed, and all others will be unemployed. The number of unemployed workers is simply given by i*:


For sufficiently small taxes, τ, such that (1-ρ-τ)>0, unemployment is an increasing function of (i) productivity, d; (ii) the subsistence parameter, ρ; and (iii) taxes, τ.

91. As productivity, d, increases, the ratio between the wage of the most productive worker and the wage of a less productive one increases. Hence average wages will also increase faster than the wage for the marginal worker. With the ratio Wmin/Wa constant at ρ/(l-τ), unemployment rises. As time progresses, the increase in productivity will widen the productivity distribution. However, since the wage distribution is not allowed to widen, unemployment will have a tendency to rise over time.

92. The parameter ρ captures the “preference for wage fairness”. Ceteris paribus, an increase in ρ is equivalent to a relative increase in minimum wages, and thus to a narrower wage distribution. The narrower wage distribution, however, can only be achieved at the price of higher unemployment. Higher taxes will be passed on to the gross minimum wage. Higher gross minimum wages push up the productivity threshold and thus increase unemployment.

Social insurance financing

93. The tax/contribution rate, τ, is endogenous owing to the government budget constraint. The only revenue source is labor income. The marginal product of labor is equal to the average product of labor, therefore labor income is equal to output. Total tax revenues are given by τY. The only expenditure item is social spending on pensions and unemployment benefits. Benefits are assumed to be a portion β ε [0,1] of the net minimum wage. The number of beneficiaries is the number of unemployed i*, plus the (constant) number of pensioners k. The government’s pay-as-you-go (PAYG) budget constraint is:


This gives the contribution rate τ as


which is equivalent to the traditional PAYG equation linking the equilibrium contribution rate, τ, to the effective benefit replacement rate, βρ, and the social insurance system’s dependency ratio (beneficiaries/contributors), (k+i*)/(1-i*). Taxes increase with the benefit parameter β, with the minimum wage parameter ρ, with the number of pensioners k, and with the number of unemployed i*.

94. As productivity, d, increases over time, unemployment will increase (as described above). As the number of beneficiaries increase, and the number of contributors declines, the tax rate will also increase, causing a further hike in unemployment. With a hike in the number of pensioners, k, the tax rate increases. This raises the minimum wage, pricing lower productivity workers out of the labor market. As a result, unemployment increases further, pushing the required tax rate even higher, and triggering another round of labor shedding at the lower end of the earnings distribution

Work incentives

95. In this framework, the parameter ρ summarizes wage setting, or labor supply, behavior. Benefits, β, and the “preference for wage fairness,” ρ, are assumed to be independent. However, in a richer model, the two parameters would be related: “wage fairness” for those employed can be thought of as underpinned by generous benefits to those unemployed. Low productivity workers priced out of the labor market remain unemployed, so in a multi-period model, the present value, rather than the level of current benefits would provide the support for limited wage differences. A decrease in the level of benefits, or shorter duration of benefits would presumably contribute to the choice of a wider wage distribution (a lower ρ).


  • Booth, Alison L., and Dennis J. Snower, 1996, “Introduction: Does the Free Market Produce Enough Skills?” in: Acquiring Skills: Market Failures, Their Symptoms, and Policy Responses, edited by Booth, Alison L., and Dennis Snower, pp. 116 (Cambridge: Cambridge University Press).

    • Search Google Scholar
    • Export Citation
  • Calmfors, Lars, and Per Skedinger, 1995, “Does Active Labor-Market Policy Increase Employment? Theoretical Considerations and Some Empirical Evidence From Sweden,” Oxford Review of Economic Policy, Vol. 11, No. 1, pp. 91109.

    • Search Google Scholar
    • Export Citation
  • Coe, David T., and Dennis Snower, 1997, “Policy Complementarities: The Case for Fundamental Labor Market Reform,” IMF Staff Papers, Vol. 44, No. 1, pp. 135.

    • Search Google Scholar
    • Export Citation
  • Commission of the European Communities, 1993, Taxation, Employment and Environment-Fiscal Reform for Reducing Unemployment,” (Brussels: DG for Economic and Financial Affairs).

    • Search Google Scholar
    • Export Citation
  • Carlin, Wendy, and David Soskice, 1997, “Shocks to the System: The German Political Economy Under Stress,” National Institute Economic Review, pp. 5776.

    • Search Google Scholar
    • Export Citation
  • Deutsches Institut für Wirtschaftsforschung (DIW), 1997, “Erwerbsstatistik unterschätzt Beschäftigung um 2 Millionen Personen,” DIW Wochenbericht 38/97, pp. 68994.

    • Search Google Scholar
    • Export Citation
  • Fitzenberger, Bernd, and Wolfgang Franz, 1997, “Flexibilität der qualifikatorischen Lohnstruktur und Lastverteilung der Arbeitslosigkeit: Eine ökonometrische Analyse für Westdeutschland,” Discussion Paper No. 97-32 (Mannheim: Zentrum für Europäische Wirtschaftsforschung (ZEW)).

    • Search Google Scholar
    • Export Citation
  • Freeman, Richard, 1995, “The Limits of Wage Flexibility to Curing Unemployment,” Oxford Review of Economic Policy, Vol. 11, No. 1, pp. 6372.

    • Search Google Scholar
    • Export Citation
  • Gordon, Robert J., 1997, “Is There a Trade-Off Between Unemployment and Productivity Growth?,” in: Unemployment Policy: Government Options for the Labour Market, edited by Snower, Dennis J., and Guillermo de la Dehesa, pp. 43366 (Cambridge: Cambridge University Press).

    • Search Google Scholar
    • Export Citation
  • Heckman, James, 1993, “Assessing Clinton’s Program on Job Training, Workfare and Education in the Workplace,” NBER Working Paper Series, No. 4428.

    • Search Google Scholar
    • Export Citation
  • Hunt, Jennifer, 1995, “The Effect of Unemployment Compensation on Unemployment Duration in Germany,” Journal of Labor Economics, Vol. 13, No. 1, pp. 88119.

    • Search Google Scholar
    • Export Citation
  • Jaeger, Albert, and Martin Parkinson, 1994, “Some Evidence on Hysteresis in Unemployment,” European Economic Review, 38 (2), pp. 32942.

    • Search Google Scholar
    • Export Citation
  • Layard, Richard, Stephen Nickell, and Richard Jackman, 1991, Unemployment: Macroeconomic Performance and the Labor Market (Oxford: Oxford University Press).

    • Search Google Scholar
    • Export Citation
  • Ministry of Labor and Social Affairs, 1997, Statistisches Taschenbuch ‘97 (Bonn: Ministry of Labor and Social Affairs).

  • Nickell, Stephen, and Brian Bell, 1996, “Changes in the Distribution of Wages and Unemployment in OECD Countries,” American Economic Review, Papers and Proceedings, May, pp. 302308.

    • Search Google Scholar
    • Export Citation
  • Organization for Economic Cooperation and Development, 1994, The OECD Jobs Study: Facts, Analysis, Strategy, Part I and II (Paris: OECD).

    • Search Google Scholar
    • Export Citation
  • Phelps, Edmund S., 1997, Rewarding Work (Cambridge: Harvard University Press).

  • Reinberg, Alexander, 1997, “Bildung zahlt sich immer noch aus,” IAB Werkstattbericht No. 15, pp. 125.

  • Saint-Paul, Gilles, 1996, Dual Labor Markets: A Macroeconomic Perspective (Cambridge: MIT Press).

  • Saint-Paul, Gilles, 1997, “High Unemployment from a European Perspective,” in: Unemployment Policy: Government Options for the Labour Market, edited by Snower, Dennis J., and Guillermo de la Dehesa, pp. 5473 (Cambridge: Cambridge University Press).

    • Search Google Scholar
    • Export Citation
  • Sneessens, Henri, 1993, “The Skill Composition of Labor Demand and Supply,” in: Growth and Employment: The Scope for a European Initiative, edited by Drèze, J.H., and E. Malinvaud.

    • Search Google Scholar
    • Export Citation
  • Stocker, Thomas M., 1993, “Empirical Approaches to the Problem of Aggregation over Individuals,” Journal of Economic Literature, Vol. XXXI, No. 4, pp. 182774.

    • Search Google Scholar
    • Export Citation
  • van der Willigen, Tessa, 1995, “Unemployment, Wages, and the Wage Structure,” in: United Germany: The First Five Years, edited by Corker, Robert, and others, pp. 2140 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

Prepared by Albert Jaeger, Kornelia Krajnyak, and Catriona Purfield.


In an earlier staff analysis, van der Willigen (1995) noted that Germany’s unemployment problem was overwhelmingly concentrated among the lower-skilled. A number of other authors have recently stressed the need for a disaggregated analysis of labor market trends; see, for example, Saint Paul (1996). Stoker (1993) provides a survey of issues that can arise in highly aggregated macroeconomic models that disregard heterogeneity.


Some simulation evidence on this second type of vicious circle dynamics in the context of an aggregate model of the labor market is provided in Chapter III.


These aggregate trends in output and employment are often contrasted with those in the United States, where real GDP during 1970–97 expanded by a similar cumulative amount as in Germany. However U.S. employment rose by some 50 percent and, conversely, labor productivity cumulatively grew by 25 percent. The sharp contrasts between employment and labor productivity performance in Germany and the United States would be even starker if employment were measured in hours worked instead of persons; see, e..g, Gordon (1997).


The lagging of aggregate real wage growth behind labor productivity growth since the mid-1980s has often been interpreted as ruling out an aggregate “wage gap” diagnosis for Germany’s adverse labor market trends; see, for example, van der Willigen (1995). SM/97/206, Chapter II, reports some evidence for a persistent aggregate “wage gap” during the 1990s.


Labor market developments in eastern Germany and the convergence process were reviewed in SM/97/206, Chapter III.


Carlin and Soskice (1997) estimate that about 40 percent of apprenticeships, mainly served in the artisan sector (Handwerk), are associated with a lower level of skill acquisition.


Mikrozensus data published by the Ministry of Labor and Social Affairs (1997) indicate that similar patterns of job losses and creation by earnings brackets occurred in western Germany throughout the 1980s.


Phelps (1997, pp. 64–78) discusses other economic factors that can create widening gaps between wages in the lower and upper half of earnings distribution including import competition and rising real interest rates.


See van der Willigen (1995) for a detailed description of the German wage bargaining system.


One example is the hourly wage distribution for German workers aged 25 years and over; see OECD (1994, Part II, p. 39). Wages for young workers in apprenticeship positions can, however, often be far below those of adult workers.


See van der Willigen (1995) for a detailed description of the German unemployment compensation system.


In practice, social contributions are subject to upper ceilings and, as mentioned in Section B in the context of the recent surge in the number of small-time jobs, to a lower threshold of DM 620.


Compared to the United States, deindustrialization was particularly severe in western Germany. While manufacturing employment increased between 1976 and 1997 in the United States, it fell by some 25 percent in western Germany during the same period. In eastern Germany, manufacturing employment in eastern Germany declined by about 65 percent during 1991-1996.


A bivariate linear regression of the aggregate unemployment rate on manufacturing employment explains 86 percent of the variation in aggregate unemployment for western Germany during 1970-96.


Also, available data on job losses/creation by earnings brackets suggest that job losses in manufacturing within a given skill category occurred primarily in the lower-paid earnings brackets of that skill category.


Jaeger and Parkinson (1994) define “asymmetric hysteresis” as permanent upward shifts in the structural rate of unemployment that are due to cyclical labor market slack. By contrast, the definition of “symmetric hysteresis” would also allow for permanent downward shifts in the natural rate of unemployment during cyclical labor market booms.


See, e.g. the collection of articles edited by Booth and Snower (1996), Chapter 7 of the OECD’s Jobs Study (1994), and the many measures to increase the employability of the unemployed listed in Germany’s National Employment Program.


For example, Heckman (1993) calculated that assuming a real rate return of 10 percent on resources spent on education and training, a US$2 trillion (some 25 percent of GDP) investment would be needed just to bring real earnings of the low-skilled back to their relative position in the late 1970s.


E.g., Freeman (1995) reported that the sizeable reductions in pay for the less-skilled in the United States led to a substantial lowering of family incomes at the lower end of the earnings distribution and may have encouraged substitution of illegal activities for work in the formal sector. He noted that in 1993 the equivalent of 6.6 percent of the U.S. male work force was either imprisoned or otherwise subject to the criminal justice system and that this ratio had risen drastically since the widening of wage inequality began in the late-1970s.


See, e.g., Chapters 8 and 9 of the OECD’s Jobs Study (1994).


The replacement ratios for unemployment insurance and assistance were lowered in 1994 to 60 and 53 percent, respectively, for individuals without children.


Sneessens (1993) reports that most estimates of own wage elasticities for blue-collar workers range between 0.3 and 1.2 while the elasticity of white-collar workers is substantially smaller.


Among others, Phelps (1997) recommends the payment of employment subsidies as an effective means to improve the employment opportunities of low-productivity workers.

Germany: Selected Issues and Statistical Appendix
Author: International Monetary Fund