The two German labour market miracles: blueprints for tackling the unemployment crisis?

Bonin, Holger

INTRODUCTION

At a time when the unemployment crisis is hardly improving in the United States and most of the European Union, many decision-makers looking for policies to boost employment are turning their attention to Germany, a country where the labour market has shown remarkable strength in a highly unfavourable global economic environment.

Real German GDP dropped 6.6% during the great recession of 2008-2009. Thus the decline in global trade in the aftermath of the financial crisis hit the economy hard, compared with the average OECD country. Yet despite of what has actually been the strongest negative output shock after World War II, the unemployment rate in Germany only rose by 0.5 percentage points--the mildest increase among all OECD member states. Many international observers have called the surprisingly moderate employment response to the great recession Germany's labour market miracle.

This success story, however, may distract from the greater labour market miracle that becomes apparent considering the country's long-term labour market development. At the turn of the century, Germany still showed typical symptoms of Eurosclerosis--low job creation in spite of overall economic growth. Structural unemployment had increased with each recession since the 1970s and high shares of long-term unemployment prevailed.

But what The Economist then called the 'sick man of Europe', the sick man has remarkably well recovered. Two facts may illustrate the labour market turnaround. In April 2012, the number of employed reached 41.4 million, a gain of 1.5 million compared with the average employment of 2000, and a record high in post-reunified Germany. In the same month, Germany's seasonally adjusted unemployment rate according to Eurostat declined to 5.4%, which was about half the euro area average (11%), and also well below the US unemployment rate. In 2000 the German unemployment rate of 7.5% was much closer to the euro area average (8.5%) and clearly above the US rate (4%).

This paper reviews the factors that can explain the exceptional performance of the German labour market since 2005. I will argue that progress was achieved through a series of labour market reforms which were initiated already before the great recession, in conjunction with a long period of considerable wage moderation. Owing to these structural improvements, the German labour market miracle might well develop into a lasting phenomenon. The favourable pre-recession state plays a major role in explaining why Germany's labour market fared so well during the great recession. In contrast, discrete policy measures and traditional German labour market institutions that facilitate buffering demand shocks within firms were less important for this other labour market miracle.

The remainder of the paper proceeds as follows. The next section explains the labour market reforms Germany adopted prior to the great recession and highlights the core success factors. The subsequent section focuses on the peculiar response of the German labour market in the aftermath of the financial crisis. The last section concludes with a reflection on what countries struggling with soaring unemployment may learn from the German case.

THE STRUCTURAL MIRACLE

Germany's social policy reforms of 2002-2005

In the period following reunification, the structural problems in the German labour market became ever more apparent. (1) High barriers to (re-)enter the labour market existed especially for the less qualified who were made obsolete by technological change. Generous unemployment benefits provided social protection and limited income inequality, but also fostered labour market segmentation. In East Germany, the registered unemployment rate quickly doubled and got stuck around 20%. But also in the West, in spite of a long phase of moderate economic growth, unemployment grew and became increasingly persistent. As a result, in 2003 Germany was one of the very few OECD countries with higher unemployment than close to the peak of the world recession in 1993. Moreover, the German unemployment rate exceeded that in most advanced OECD countries. Regular employment was in decline and long-term unemployment on the rise.

In a first response to the labour market challenges, Germany's decision-makers disposed of a core piece of social legislation of the 1960s. The Employment Promotion Act had put emphasis on individual employment creation schemes and training programmes aimed at adjusting the skills of the unemployed to the supposed demand of employers. However, growing and persistent skills mismatches and especially the failure to handle the transition crisis on the East German labour market showed that these instruments had become inapt in face of growing global competition and fast technological change. The new Social Code gave priority to job placement over all other active measures, and endowed local employment offices with more autonomy. However, these adjustments did not bring the desired effects. Quite on the contrary, in 2002 a scandal jolted the Federal Labour Office--the employment placement statistics had been sugarcoated.

The incident triggered the launch of the so-called Agenda 2010 by then-Chancellor Gerhard Schroder. The outcome of the Agenda, worked out in part by an independent expert commission, was a sequence of laws whose final stage came in effect on 1 January 2005. The whole body of legislation, also known as the Hartz Reforms, constitutes the largest labour market and social welfare policies amendment in the post-war history of Germany. Three overriding principles of the reform package are worth pointing out, for they illustrate the change of paradigm in German social policy-making.

First of all, under the guiding principle of 'supporting and demanding' the reforms established a push and pull approach. The innovation for Germany was the latter part of the principle. The reforms put more pressure on the unemployed to engage in job search, in exchange for the active services of the labour agencies and passive income support. 'From now on, we will not allow anybody to sit back and do nothing, we will sanction those who reject reasonable work ...', this quote from Chancellor Schroeder's government declaration about the Agenda 2010 on 14 March 2003 shows the new attitude. At the level of the public employment services, the change is reflected in the introduction of personalized integration contracts that define each job seeker's obligations and activation plan, but also point to the sanctions (in the form of temporary benefit cuts) looming for non-compliers.

Secondly, the reforms were carried by optimism as regards the proper functioning of labour market forces. This perspective helped tackling the unemployment problem as a labour supply problem. The decision-makers in favour of stimulating labour market participation incentives for the less qualified and of limiting early retirement incentives were confident that employers would respond by creating the low-wage jobs and employment opportunities for the elderly that did not exist at the outset of the reform.

In the third place, the reforms made a leap towards evidence-based policymaking. One result of the statistics scandal was that academic researchers gained better access to registered individual labour market records, in order to evaluate the causal effects of active labour market policies, as well as of the enacted reforms, based on methodologically sound control group designs. Since then probation periods for new measures and reshaping of active instruments according to the results of sound programme evaluation have become a regular practice in Germany.

Successful cornerstones of the Hartz reform

It appears that among the very huge number of changes initiated during the period 2002-2005, two measures are essential for the reforms' long-term success: a reorganization of the unemployment benefit system that strengthened labour supply incentives, and a reorganization of labour market services that strengthened the efficiency of active policy measures.

Reorganization of the benefit system

The reorganization of the unemployment benefits system created stronger employment incentives through a move from status protection to basic income support. (2) It merged two distinct systems supporting the long-term unemployed into one.

One of the past systems was organized (if not paid for) by the Federal Labour Office and provided unemployment aid, a non-means-tested payment which, after expiration of unemployment insurance benefits, replaced a constant fraction of former net wage income for an unlimited amount of time. The other system was under the responsibility of the municipalities. It provided only for the subsistence level of consumption. This social welfare scheme took care of needy individuals without unemployment benefit claims due to an insufficient record of prior employment covered by social insurance. Most of the individuals under the social welfare scheme were not registered unemployed, not asked to engage in job search, but also excluded from the integration measures of the Federal Labour Agency. If at all, the social welfare recipients could benefit from communal labour market polices whose scope varied greatly between municipalities.

The merger of the two systems abolished unemployment aid. Now all needy unemployed receive the same means-tested basic income support. While the new uniform transfer is labelled as 'unemployment benefits II', it is essentially identical to the previous tax-financed social welfare benefits.

A second element of the reorganization shifted the hitherto divided responsibility for the labour market integration of the unemployed without access to unemployment insurance benefits to one single authority--the so-called job centres. Two approaches how to operate these compete. Most local authorities established a working partnership with the local branches of the Federal Labour Agency. But some municipalities use the option to operate the job centres by themselves.

The merger of the two systems had positive effects. First of all it lowered the reservation wages of those eligible to unemployment benefits. The move from earnings-related unemployment benefits to basic income maintenance implies that many unemployed former high-wage earners are worse off compared with the old system. They seem to adapt their labour supply behaviour in order to avoid threatening income loss if becoming long-term unemployed (Clauss and Schnabel, 2008). This can explain why a substantial increase in the outflow rates from short-term unemployment began to show just toward the end of 2005. At this time the first cohort of unemployment benefit recipients affected by the reform were coming close to the expiry of their insurance benefits.

A stylized fact illustrates the progress achieved. The Federal Labour Agency (2011) reports that the monthly outflow rates from short-term unemployment were about 12% in December 2008 and June 2011, that is, before and after the great recession. This compares with outflow rates of around 8%-9% in the pre-reform period 2000-2004. Consequently, the average duration of unemployment when re-entering regular employment has declined by more than a quarter. Shorter unemployment duration reduces the stock of unemployed for any given inflow rate into unemployment. Fahr and Sunde (2009) confirm the improved matching efficiency, in the sense of speedier return to reemployment, also from a macro perspective.

A second effect of the reform is that it brought a very large number of long-term unemployed in need of activation and placement measures back to focus. In fact, very quickly almost all of former social welfare recipients of working age became classified as employable (which is defined as being capable of working at least three hours a day). More than one half of these were also classified as unemployed, and thereby became customers of the new job centres. (3)

As formerly hidden unemployment turned into open unemployment, the number of registered job searchers in Germany reached an all-time high of 5.3 million in February 2005. The unemployment rate jumped to 12.7 % compared with 10.8% in the last pre-reform month, that is, December 2004. The apparent worsening of the labour market data, while in line with the intentions of the reformers and only a statistical artefact, left an immensely negative impression of the policy change on the public.

The new job centre customers felt the pressure of the 'demanding' policy and had to prove job search effort or even to take up one of the newly created so-called 'l-Euro jobs', a form of public employment rewarded with only a symbolic wage and thus mimicking workfare. At the same time, the 'supporting' part of the policy, for example, good quality counselling or training, did not reach many of the long-term unemployed. The imbalance of demanding and supporting may explain why the sanctioning mechanisms designed by the architects of the reform have never been used to the full possible extent. Decision-makers also have not closed the loophole to deregister from unemployment by starting minor employment (so-called 'mini jobs') while receiving supplementary welfare benefits.

These factors may explain why progress on long-term unemployment is much weaker than on short-term unemployment. Nevertheless, the monthly outflow rates from long-term unemployment have improved. Since 2005 they have been in the range of 3%, while they were in the range of 2% in the period 2000-2004. In June 2011, those municipalities that operate jointly with the Federal Labour Agency recorded less than 900,000 long-term unemployed, compared with the maximum of 1.7 million in early 2006 when the statistical reclassification effect described above unfolded its full impact. (4)

Yet in April 2012 the share of the long-term unemployed among the unemployed was 35.5 %, little less than the rate at the start of 2000 (37.3 %). Altogether, the reorganization of the unemployment benefit system has reduced the size of long-term unemployment much more by reducing the probabilities for short-term unemployed to become long-term unemployed, than by raising the probabilities of basic income recipients to regain employment.

More effective labour market services and instruments

Another key factor behind the substantial improvement of exit rates from unemployment is the better performance of public employment services and the labour market instruments they are using, (5) The reorganization followed the principles of New Public Management by introducing more flexibility and responsibility at the subordinate levels in the Federal Labour Agency. Managers of local labour offices and individual case workers have gained wider discretion over the instruments they want to use. In exchange, they have become subject of much stricter monitoring, with accountability based on quantitative targets.

The idea to modernize public services through the introduction of competition with private entities that provide quasi-public goods, however, did not work well. One example is the attempt to outsource placement activities by setting up so-called Staff Service Agencies ('Personal Service Agenturen'), temporary work agencies for the hard-to-place unemployed. These agencies never really took off and the concept was quickly abandoned. Another example is vouchers for private education and placement services giving the unemployed the possibility to decide against corresponding public services after using these to no avail. The literature could not substantiate any positive impact of the vouchers on individual employment probabilities. One lesson possibly to be learned from this experience is that successful private business models for integrating and placing the difficult-to-integrate and difficult-to-place unemployed are very hard to find.

The better organized public employment services are also using a better mix of active policy measures. The reforms introduced effective new tools, whereas policies with too bad a track record got reduced or even abandoned. Furthermore, a better fit of workers and programmes has helped improving the employment impact of active policy measures.

Considering the newly introduced successful instruments, a shared aspect is that they are directed at the integration into employment. This means that they are designed such as to avoid locking the participants in positions outside the regular labour market. A good example of such an instrument is the 'Ich AG', literally Me, Inc., a start-up subsidy that caught a lot of public attention (and an 18% share of active labour market policy spending) when it was launched in 2003. (6) This subsidy secured income for unemployed agents who set up a small-scale start-up on the basis of a business plan that had to be approved by the local chambers of commerce. According to the evidence, this measure has been highly effective with respect to income and employment outcomes even in the long run. Moreover, a study of heterogeneous treatment effects shows that the start-up subsidies unfolded their best effects on the disadvantaged groups in the labour market (Caliendo and Kfinn, 2011).

A second example of an active integration policy is wage subsidies given to employers who hire unemployed with potential productivity shortfalls. The reforms of 2003-2005 loosened the eligibility criteria of such integration subsidies. Employers deciding to employ an elderly unemployed, for example, could now receive up to one half of the new hire's remuneration for a period of up to two years. Despite some effort to prevent revolving door effects and free riding, the substantial windfall gains typical for wage subsidy schemes have occurred and probably hampered the efficiency of the policy. Yet the data indicate a remarkable 20-50 percentage points higher probability of employment post-treatment, if one compares the treated with comparable untreated individuals (Boockmann et al., 2012).

A way how public employment services managed to achieve better focus of their active labour market measures is customer profiling. The current practice is that caseworkers use standardized criteria in order to classify the unemployed after an initial interview serving to identify personal strengths or weaknesses that may facilitate or impede employment. The classification system defines four types of unemployed. The two types at the top and bottom are excluded from active labour market measures. Agencies expect that the 'market clients' at the top do not require special assistance to find employment, and that the 'supervision clients' at the bottom would not regain employment even if they were given the chance to participate. The 'activation clients' (ranked second, in terms of labour market prospects) and the 'support clients' (ranked third) receive counselling and targeted measures focusing on activating job search, respectively more intense measures such as training.

The classification system not only concentrates resources on the right groups of unemployed. The underlying idea of specific measures associated with specific groups has also helped re-designing measures to better satisfy the needs of their target group. A good case in point is the adaptation of training measures, which according to a huge body of empirical studies had no or only moderately positive effects before the reform. The reforms introduced the idea of explicit cream skimming. Training programmes have become available only to those unemployed with an estimated 70 % chance of obtaining employment post-treatment. This approach immediately raises efficiency as it concentrates the measure on a much smaller but also much better fitting group of people. Another, indirect effect strengthens the positive results. In equilibrium training providers must adapt their products in order to match the 70% individual target success rate at the company level.

The impact of the policy change as regards training the unemployed confirms the expectations. The share of unemployed engaging in training has declined by a wide margin, training measures have become much shorter, and they reach the target group earlier during an unemployment spell. Finally, training seems to be more effective than before, although to some extent the advancement is due to the systematically smaller locking-in effects that are associated with the shorter duration of the programmes (Bonin and Schneider, 2006).

A slightly different way of interpreting the changing practice with regard to training measures is that public employment services put under stricter efficiency control give up measures with a poor track record. The fate of job creation schemes as a substitute for market employment may illustrate this point. It was obvious from a huge body of studies that the job creation schemes widely used before the reform reduced the chances of the participants to return to the regular employment by a very significant margin. Following the reform the job creation measures faded. Agencies concentrated the instrument on the hard-to-place unemployed, that is, the supervision customers according to the new job seeker classification. As a result, the negative treatment effects got smaller but did not turn into positive. The resources spent of this activation policy thus declined more and more, and finally, in April 2012, the instrument vanished altogether from the toolbox of public employment services.

Labour market deregulation?

A remarkable aspect of the German approach for animating the labour market is that the reforms have relied little on labour market deregulation. In particular, there have been no major institutional changes in order to promote numerical or wage flexibility in Germany.

As regards numerical flexibility, the traditional system of employment protection legislation remains basically intact. Firing costs are high by international comparison. Dismissals on grounds of personal misconduct and layoffs for operational reasons are generally forbidden, and a clear-cut severance payment scheme is lacking. The main reform in the system was a change in the coverage of small establishments in 2003, which raised the employment protection legislation exemption threshold from 5 to 10 workers, but this moderate change probably has had no significant effect on job flows (Bauer et al., 2007).

In this context, the amendments promoting two forms of 'atypical' employment that could be used to circumvent employment protection, so-called mini-jobs and temporary work, are often discussed as drivers of Germany's post-reform employment growth.

Mini-jobs, which pay incomes of up to 400 euros per month and are exempt from payroll and income taxes, were intended to provide stepping stones to regular employment for the long-term unemployed. It turns out, however, that a very vast majority of these jobs is occupied by second-earners, students or retirees attracted by a zero marginal tax rate. At the same time, it appears that the mini-jobs have crowded out regular low-productivity jobs, which makes it actually harder for the less-skilled to obtain full-time employment. It is therefore quite clear that this instrument has not contributed to the unemployment turnaround.

In the period 2003-2010, the incidence of temporary work roughly doubled, and the rise is observed throughout the economic cycle (Spermann, 2011). The pattern suggests a response to the Hartz reforms which abolished certain restrictions on synchronization, re-assignment and the maximum duration of temporary employment. Still the regulation of temporary agency work in Germany remains rather strict. In the wage dimension, an 'equal pay for equal employment' principle prevails that drives up labour costs. In addition, since January 2012 a sector-specific minimum wage is in place. Such regulation keeps the sector at an employment share of 2 %-3 %, too small a figure to explain much of recent employment growth.

The example of the new minimum wage for temporary agency work illustrates that decision-makers have not made Germany's labour market more flexible with regard to wages. In fact, the expansion of minimum wages appears as a response to the downward pressure on reservation wages stemming from the unemployment benefits reform (cf. section 'Reorganization of the Benefit System'). A distinctive feature of the minimum wages is that they are determined by collective agreements and installed on application of the bargaining unions and/or employer organizations. As bargaining takes place at the sector level, the minimum wages form a sketchy and diverse pattern. Some of the stipulated minimum wages are in the range of 7-8 euros per hour and thus comparable, for example, with the UK or Canadian minimum wage levels. But some of them are in the range of 11-13 euros and thus appear anything but low if one supposes a social protection target. (7)

In line with the principle of evidence-based labour market policy-making, the government ordered a comprehensive assessment of the minimum wage legislation impact. In summary, the currently available eight sector studies indicate no adverse employment effects. Yet this observation needs to be treated with caution. In some cases, the introduction of the minimum wage did not bite, since wages below the minimum wage did not exist. In other cases, employers have been creative in avoiding looming labour cost increases. Labour standards or unpaid working hours go up, wages just above the minimum wage grow less than normal, or jobs are assigned to uncovered sectors.

In the realm of the collective bargaining system, one notable innovation is a weakening of the principle of 'one establishment--one collective agreement'. The admission by court ruling of different agreements within a firm for different occupations has introduced some competition among unions. So far, however, only few occupational unions have emerged. They typically represent small occupations with strong bargaining power in product markets with weak competition and therefore, if anything, drive up labour costs. On a larger scale, the requirement to represent a relevant share of workers for obtaining collective-bargaining capacity still prevents stronger competition through the arrival of smaller unions.

Yet while the regulatory framework for the collective bargaining system seems rather rigid, the practice of industrial relations in Germany is generally highly cooperative, which helps achieving flexibility where needed. This brings us to the favourable auxiliary conditions that promoted strong employment growth in the aftermath of the Hartz reforms.

Favourable auxiliary conditions

Certainly the most important supporting factor of the structural reforms is the restoration of Germany's international competitiveness, which was achieved through a remarkably long period of annual wage growth staying below the productivity growth generated by technical progress. As a result, unit labour costs basically moved sideward during the period 1994-2005 (see Figure 1). During the same period, unit labour costs in OECD-Europe increased by 38 % and in the US by 17%. This has given Germany a boost in exporting capacity and thereby in labour demand.

[FIGURE 1 OMITTED]

To a large degree, this achievement is related to prudent behaviour of trade unions and employer organizations. The social partners responded to the pre-reform labour market sclerosis (and the emergence of Eastern Europe as a competitor) with moderate collective agreements aiming at employment security. In the same spirit, they have established more flexibility at the decentralized level. Often 'opening clauses' are written into the collective agreements that allow individual employers and works councils to go below the collective wage agreement, if this preserves employment in the establishment. As trade unions need to approve this kind of decentralized wage setting, they have maintained a certain control over the degree of wage flexibility within the collective system. Nevertheless, they could not stop a rise of decentralized wage bargaining through shop agreements often that are used in larger firms.

Another factor keeping average unit labour cost down is the rise of low-wage employment sectors without relevant social partners, neither on the employee nor on the employer side. This development shows in particular in the post-reform period 2005-2007 when unit labour costs dropped by more than 4%. The increased pressure on the unemployed forced agents into these jobs which they would have considered paying less than their reservation wages before. Yet the strong decline of unit labour costs in this period is not only due to a composition effect. It is also due to especially moderate collective agreements. Indeed, additional wage restraint is a rational answer to the reforms. The reduction of expected unemployment benefits lowered the value of the outside option for the incumbent workers.

In addition to the return of international competitiveness, several impulses to labour supply coming from policy areas other than labour market policy in a strict sense are contributing to boosting employment figures in Germany. (8)

One especially important issue is the growing labour supply of the elderly associated with pension policies. The progress of Germany with regard to labour force participation rates of individuals aged 55-65 is impressive in international comparison. While the country's participation rate of 43% in 2000 was below the EU average, the current rate of above 63 % is among the top in Europe. The development also does not show signs of a diminishing trend. Germany in terms of labour market integration of the elderly may soon catch up to the US or even the Japanese levels.

Several factors are behind this special labour market turnaround. First, a reform of early retirement, which introduced pension cuts close to actuarially fair, lead to a marked increase of the effective retirement age. Next, a gradual shift of the standard retirement age from age 65 to age 67 has started to become effective and will push old-worker participation rates in the decades to come. A third factor is equalization of male and female standard retirement age. The improved dynamics in the labour market for elderly workers through the pension reforms helped achieving the almost immediate success of the activation measures within the Hartz reforms that cut early retirement incentives, notably the shorter unemployment benefit duration for agents aged 55 and older (Dlugosz et al., 2009).

Additional labour supply incentives are coming from the family policy domain. Reconciliation of family and work, especially for mothers, is an acknowledged primary target of social policies for families in Germany. Ongoing expansion of day-care centres for children younger than 3 reflects this target. Legislation stipulates a 'right to day-care' from 2013, which requires an ambitious development programme to meet the estimated demand for care, that is, a supply quota of 40% for children below kindergarten age. Beninger et al. (2010) estimate that the expansion could raise the labour force participation rates of the targeted mothers by up to 5 percentage points. In view of the traditionally low labour market attachment of German women, this would be quite a remarkable advancement.

The mothers who are predicted to benefit most from cheaper day-care are the well-educated. This points us to a third kind of labour supply impulse, which is related to education policies. In Germany upskilling is underway through growing participation rates in tertiary education. As agents with tertiary educated exhibit systematically stronger labour force attachment than agents with secondary education, a composition effect will raise labour supply in the long term. This effect is strengthened by some recent effort to shorten tertiary education beginning to pay.

In this context, it is worth noting that this development could also pose a challenge for Germany. The expansion of tertiary education implies fewer workers with an apprenticeship training degree who are still the backbone of the German workforce. Thus some projections foresee labour shortages in the medium skill range. There would be scope to refill the pool from the about 10% of a cohort that leaves school with inadequate cognitive and non-cognitive skills. A massive intervention system serving to facilitate transitions from school to apprenticeship to work does not seem very effective. Rather, changes in the primary school system, which is known to inhibit inter-generational mobility far more than in other countries, would be necessary. But in contrast to labour market policies, set at the Federal level, school policies are decided at the state level, which impedes synchronized reform effort. At this margin, in view of the, by international standards, low youth unemployment rate, Germany perhaps simply does not yet suffer enough to make rethink occur.

THE ANATOMY OF THE GREAT RECESSION MIRACLE

The different elements gathered in the previous chapter make it clear that the German labour market had been successfully reactivated by structural progress when it was hit by the great recession in the fourth quarter of 2008. In fact, some observers have argued that panic-proof employment in the course of the crisis primarily reflects a lagged labour market response to the reforms. They claim structural decline in unemployment released by the improvement in institutions was extending well into the recession, which then worked against the employment decline associated with the macroeconomic shock (Gartner and Klinger, 2010). One piece of evidence supporting this claim is decline in long-term unemployment despite the recession. Another piece of evidence is that the Beveridge curve appears to be located further to the left in the period 2008-2011. The sideward move, suggesting decline in structural unemployment, clearly dominates a slight bulge in the curve attributable to the cyclical fluctuation.

However, while it is a well-known fact that the adjustment of an economy to structural changes at the micro level may require quite a long 'medium run' (Blanchard, 1997), a lagged response to the labour market reforms hardly tells the full story. Probably several others factors contributed to the German employment miracle during the great recession, although the specific effect of each factor is difficult to ascertain. (9)

A first argument relates to the period of wage moderation before the great recession (recall Figure 1). Using dynamic simulations based on an empirical model of demand for total hours worked and employment, Boysen-Hogrefe and Groll (2010) conclude that wage moderation since 2001 was gradually increasing the equilibrium level of employment throughout the great recession, which cushioned the adverse employment effect of the demand shock. Moreover, it appears that the wage moderation effect has had the strongest positive employment impact among a number of rivalling factors. Note that the interpretation of the German case is consistent with an empirical observation stemming from international comparison. Countries with slower unit labour cost growth before the crisis showed systematically less employment decline during the crisis.

Another aspect to consider is the peculiar nature of the economic crisis itself. First of all, the global financial market turmoil hit the German economy comparatively little. Germany benefitted from a conservative banking system characterized by a large number of small, semi-public local banks that are fairly disconnected from the international capital markets. In addition, the conservative attitude of German consumers as regards portfolio strategies paid off. As the rate of stock ownership is very low by international standards, private wealth was not affected much by the losses on international stock markets. Finally, the country never got anywhere near a bursting housing market bubble.

As regards the real economy, the crisis did not deteriorate confidence that the domestic fundamentals were in very good shape. Firms generally interpreted the recession correctly as an export crisis not rooted in lack of their own competitiveness. They also anticipated that the collapse of global trade that originated in a very high degree of uncertainty abroad could not persist for too long. This optimism was backed up by strong internal reserves they had built during the upswing before the crisis, not the least due to the decline in unit labour costs associated with the labour market reforms. It is likely that consumer confidence also benefitted from the reforms because of their positive effect on employment propensities. Furthermore, while real hourly wages stagnated, the number of wage earners was on the rise.

As a result of these special conditions, output decline hit mostly firms in the manufacturing sector, whereas the providers of consumer goods and services were only mildly affected. A consequence of the special shock was a rather unusual regional dispersion of the recession. Output decline was especially strong in the thriving south of Germany where a large share of export-oriented firms is located, and where local labour markets had shown some signs of labour shortages prior to the crisis. In some areas, unemployment rates had been at or close to the natural rate right before the recession began. Another consequence was that the demand shock affected especially highly trained and highly productive employees- the typical workforce of a typical German medium-sized exporting 'global champion'.

The particular labour market stance of the exporting manufacturers can explain why firms absorbed the shock more intensely by labour hoarding than could be predicted considering the experience from past recessions (M611er, 2010). A destruction of highly specialized firm-specific human capital would have been expensive. Moreover, the labour shortages and strong competition in search for apt professional workers many firms had experienced on tight labour markets before the crisis probably raised the expected rehiring costs.

Massive labour hoarding was achieved through adjustments both at the working hours and at the labour productivity margin. According to a decomposition of the cyclical response, Germany's first ever decline in productivity per hour, by almost 1%, buffered about 1.1 million person equivalents, whereas the reduction of working time, by almost 4%, buffered about 1.4 person equivalents (Mo11er, 2010). The consequence of massive hoarding was a sharp jump in unit labour costs during 2009 (by 6.6%, see Figure 1). Yet firms were willing to bear the cost of the option to keep their workers. As employers did not detect structural weaknesses and were optimistic about a quick recovery, the discounted price of the alternative, that is, fire now and rehire and retrain later, was higher than the immediate losses incurred by maintaining an underutilized workforce.

It also helped that firms generally could afford the costs of the hoarding strategy, as they had accumulated profits from the export-driven expansion of 2005-2007. Related to the cyclical experience in the exporting sector, Burda and Hunt (2011) have put forward a somewhat unusual explanation for the labour hoarding pattern. They note that employment growth during the precrisis boom was low considering the output expansion and moderate wage growth, since pessimistic employers convinced that the upswing would not last long were reluctant to recruit. As a result, when the expected recession arrived (if more severe and fuelled by unexpected reasons), exporting firms were not burdened with an inflated workforce and therefore had to fire less than one would normally expect considering the great recession and soaring labour costs.

While labour hoarding thus was rational to cope with the temporary demand shock, German labour market institutions facilitated the response. It is important to note that the relevant practices supporting short-term work in firms experiencing business difficulties were in place already before the reform. In a setting with high employment protection hampering external flexibility, institutions geared to achieve internal flexibility are a rational substitute.

In this context, many observers have focused on so-called short-work plans ('Kurzarbeit'), subsidies for firms faced with slack demand. Employment agencies reimburse up to 67% of the pay workers lose when their employers shorten working weeks in order to avoid layoffs. Brenke et al. (2011) claim that the increase in German unemployment would have been twice as large in 2009 without the increased use of short-term work plans. But the evidence also suggests that the instrument did not play a stronger role than in previous recessions (Boysen-Hogrefe and Groll, 2010). Like in any other recession since the 1970s, the average reduction in working hours of employees on short-time-work plans equalled roughly one third. And the short-time employment share of 3.8% (equivalent to 1.5 million workers) at the height of the recession in May 2009 was smaller than would predict from past experience given the size of the negative demand shock.

In particular, it appears that the discrete changes in the short-term scheme introduced as a response to the great recession did not really contribute to heighten the impact of the instrument. One amendment was the extension of the maximum duration of subsidized short-term work to 24 months. But as the economic crisis lasted much shorter and the upswing started already in the second half of 2009, this extension was not of much practical use. Another amendment was an increased subsidy payment provided that workers on short-term work used their spare time to engage in occupational related training. However, take up rates in this programme were extremely low, as one would expect given that structural problems related to qualification did not feed the demand shock to the manufacturing firms.

Considering the evidence, it appears that the true innovation as regards internal flexibility was not in the public but in the private domain. The personnel management tool of work-time accounts, currently available in about one half of German firms, unfolded its capacity to cushion cyclical fluctuations. Work-time accounts, often introduced and regulated by collective agreements, allow firms (or rather individual workers) to save overtime hours during booms in an account without adjusting wages. They can later use the working time stored in the accounts to shorten working hours during downturns without cutting wages.

As the great recession had been preceded by a boom, many exporting manufacturers managed to buffer employment by running down their positive balances in the working-time accounts- and probably by accumulating negative balances. According to the decomposition by M611er quoted above, the employment saving impact of the work-time accounts amounted to about two thirds of the short-term work impact. This meant that the effect of this business management instrument to preserve valuable human capital within firms was much larger than in earlier recessions.

Another employment protection effect probably came from the implicit disincentives to fire built into the tool. In case of a layoff employers should compensate the worth of the individual overtime savings account. In other words, a dismissed worker is entitled to receive a retrospective wage payment for the past overtime hours put in the account. Hence the positive amount of hours buffered during the pre-crisis boom brought about employment insurance at the individual worker level. This insurance mechanism only fails, if firms become bankrupt and have to dismiss their entire workforce. In case of bankruptcy, the worth of the savings times account is generally lost, as only a small share of employers buys insurance for their workers against this event. Not many firms hit by the great recession went entirely out of business, however, and therefore the individual firing insurance generated by the working time accounts was often in effect.

CONCLUSION: WHAT LESSONS FOR OTHER COUNTRIES?

The picture of the German labour market response in the world recession drawn in the previous section demystifies what some have called an employment miracle. A good deal of the development during 2008-2009 had less to do with superior institutions or discrete policy responses helping to cope with the great recession, but more with a lucky combination of circumstances: the export boom followed by an export shock on manufacturing firms with a specialist workforce, the sequence of pessimistic business expectations during the preceding boom followed by optimistic expectations during the recession, and the transition to a new labour market equilibrium with lower structural unemployment underway during the recession.

These factors appear unique and are impossible to copy elsewhere. In fact, one should not even conclude that Germany would cope with another great recession, this time perhaps associated with a euro crisis, as well as it did last time.

Still one stabilizing factor in the crisis has rightly caught the attention of decision-makers abroad--Germany's capacity to let working hours breathe. Reliance on high internal flexibility mirrors the country's low external flexibility associated with maintained high employment protection and a moderate employment share of atypical work forms. Working hour flexibility is not only supported by government-sponsored short-time subsidies. Cooperative industrial relations have backed the highly successful work-time accounts, and a high-quality occupational training system helps building up high stocks of firm-specific human capital.

The more valuable blueprints for countries struggling with structural labour market problems are perhaps to be found in the German labour market turnaround prior to the great recession. Decision-makers managed to put the country on a new equilibrium employment growth path, while maintaining a comparatively high level of social protection. In particular, Germany has only cautiously used a deregulation approach to increase external flexibility, but instead sought to maintain an internal numerical and functional flexibility model, which fits and promotes the country's highly qualified workforce.

A number of success factors might be suitable for other countries to copy. First, cooperative industrial relations with employers and unions giving employment stability high priority in their target functions can enormously enhance the effect of public institutions and policy reforms promoting employment. In particular, farsighted unions may choose to appropriate less than real labour productivity gains to ease labour market recovery. The German example is clear proof of the simple textbook model that in the medium run wage moderation pays off in terms of competitiveness and employment.

Secondly, as measures to retrain workers with obsolete qualifications are often ineffective, the public needs to accept a low-wage sector providing employment opportunities for these workers. (In a longer term perspective, one would of course prefer advancements in primary education.) The rapid decline in German unemployment figures was only possible by putting more pressure on agents at risk to become long-term unemployed, often due to qualification problems. This bold policy move generated strong labour supply incentives and again, a simple textbook model worked. Employers responded by creating appropriate new jobs. The same happened following Germany's structural changes to promote labour force participation of elderly workers and women. Decision-makers thus have good reasons to promote labour supply strategies.

Thirdly, efficient employment services matter. The German experience tells that a good way to achieve better market and costumer orientation is the introduction of clear quantitative targets, in conjunction with accountability at all levels of the public labour administration. Strict controlling also helps to realign the active labour market measures in use. Liable employment services have incentives to concentrate on tools with a proven effect. Rigid programme evaluation is a helpful complement to this management strategy. An evidence-based policy approach sets clear targets and allows pilots or trial periods to test new measures. Yet more importantly, it provides decision-makers with independent information on the effectiveness and efficiency of the tools they are using.

Finally, considering the political economy of the necessary structural reforms, achieving labour market turnaround is not easy. It probably needs an external push, like the scandal that unleashed the Hartz reforms in Germany, to move ahead. At present, the European debt crisis could be such a driver of structural reforms.

But in addition good political marketing is needed. This requires decision-makers who are endowed not only with a clear understanding of the fundamentals underneath their reform agenda, but also with the capacity to communicate the rationale of the reforms to the people who have to bear the burden of adjustment. In Germany, decision-makers failed on the communication side. The government of Gerhard Schroder was not re-elected, as his reform agenda did not unfold positive effects right away. But this was probably a price worth paying for a true German employment miracle.

REFERENCES

Bauer, T, Bender, S and Bonin, H. 2007: Dismissal protection and worker flows in small establishments. Economica 74(5): 804-821.

Beninger, D, Bonin, H, Horstschraer, J and Muhler, G. 2010: Wirkungen eines Betreungsgeldes bei bedarfsgerechtem Ausbau ffiihkindlicher Kindertagesbetreuung. Quarterly Journal of Economic Research 79(3): 147-168.

Blanchard, O. 1997: The medium run. Brookings Papers in Economic Activity 1997(2): 89-159.

Bonin, H and Schneider, H. 2006: Wirksamkeit der F6rderung der beruflichen Weiterbildung vor und nach den Hartz-Reformen. Wirtschaftspolitische Blatter 52(1): 161-172.

Boockmann, B, Zwick, T, Ammermfiller, A and Maier, M. 2012: Do hiring subsidies reduce unemployment among older workers? Evidence from two natural experiments. Journal of the European Economic Association 10(4): 735-764.

Boysen-Hogrefe, J and Groll, D. 2010: The German labour market miracle. National Institute Economic Review 214(1): 38-49.

Brenke, K, Rinne, U and Zimmermann, KF. 2011: Short-term work: The German answer to the great recession IZA Discussion Paper No. 5780, Bonn.

Burda, M and Hunt, J. 2011: What explains the German labour market miracle in the great recession? CEPR Discussion Paper No. 8520, London.

Caliendo, M. 2009: Income support systems, labour market policies and labour supply: The German experience, IZA Discussion Paper No. 4665, Institute for the Study of Labour, Bonn.

Caliendo, M and Kritikos, A. 2009: Die reformierte Existenzgrundungsforderung fur Arbeitslose--Chancen und Risiken. Perspektiven der Wirtschaftspolitik 10(2): 189-213.

Caliendo, M and Kunn, S. 2011: Start-up subsidies for the unemployed: Long-term evidence and effect heterogeneity. Journal of Public Economics 95 (3-4): 311-331.

Clauss, M and Schnabel, R. 2008: Distributional and behavioural effects of the German labour market reform. Journal of Labour Market Research 41 (4): 431-446.

Dlugosz, S, Stephan, G and Wilke, R. 2009: Fixing the leak: Unemployment incidence before and after the 2006 reform of unemployment benefits in Germany, ZEW Discussion Paper No. 11-013, Mannheim.

Eichhorst, W, Grienberger-Zingerle, M and Konle-Seidl, R. 2010: Activation policies in Germany: From status protection to basic income support. German Policy Studies 6(1): 59-100.

Eichhorst, W and Zimmermann, KF. 2007: And then there were four ... how many (and which) measures of active labour market policy do we still need? Applied Economics Quarterly 53(3): 243-272.

Fahr, R and Sunde, U. 2009: Did the Hartz reforms speed-up the matching process? A macro evaluation using empirical matching functions. German Economic Review 10(3): 284-316.

Federal Labour Agency--Bundesagentur fur Arbeit. 2011: Sockel- und Langzeitarbeitslosigkeit, Broschure der Arbeitsmarktbefichterstattung. Nunberg.

Gartner, H and Klinger, S. 2010: Verbesserte Institutionen for den Arbeitsmarkt in der Wirtschaftskrise. Wirtschaflsienst 2010(11): 728-734.

Jacobi, L and Kluve, J. 2007: Before and after the Hartz reforms: The performance of active labour market policy in Germany. Journal of Labour Market Research 40(1): 45-64.

Moller, J. 2010: The German labour market response in the world recession--De-mystifying a miracle. Journal of Labour Market Research 42(4): 325-336.

Rinne, U and Zimmermann, KF. 2011: Another economic miracle? The German labour market and the great recession, IZA Discussion Paper No. 6250, Bonn.

Spermann, A. 2011: The new role of temporary agency work in Germany, IZA Discussion Paper No. 6180, Bonn.

Wunsch, C. 2005: Labour market policy in Germany: Institutions, instruments and reforms since unification, Discussion Paper 2005-06, Department of Economics, University of St. Gallen.

HOLGER BONIN (a,b,c)

(a) Centre for European Economic Research (ZEW), P.O. Box 103443, D-68034 Mannheim, Germany. E-mail: bonin@zew.de

(b) University of Kassel, FacuLty of Economics and Management, D-34109 Kassel, Germany.

(c) IZA, Schaumburg-Lippe-Strasse 5-g, D-53113 Bonn, Germany.

(1) See, for example, Wunsch (2005) for a more detailed overview of the labour market development after German unification.

(2) See Eichhorst et al. (2010) for a detailed description of the changes in the German unemployment insurance system.

(3) The group of non-unemployed employable long-term unemployed includes participants in active labour market programmes, individuals in minor employment, individuals aged 58 and above, individuals in education and individuals who need to give care to children or relatives.

(4) A serious drawback is that due to the two competing administrative systems reliable time series information for the entire long-term unemployed does not exist.

(5) Jacobi and Kluve (2007) survey the empirical evidence on the effectiveness of labour market services and policies before and after the reforms. Eichhorst and Zimmermann (2007) is another excellent survey focusing on active labour market instruments.

(6) This subsidy scheme and an earlier one, the so-called bridging subsidy ('Uberbrfickungsgeld'), in 2006 were merged into one, less-well designed scheme, the so-called start-up subsidy ('Grundungszuschuss'). See Caliendo and Kritikos (2009) for a critical review.

(7) At least in the cases where employer organizations have supported the minimum wage, one may reckon that other aspects like impairing competition from outsiders play a role.

(8) See Caliendo (2009) for a review of the measures that have fostered labour supply.

(9) See Rinne and Zimmermann (2011) for a helpful survey.