VoxEU Column Europe's nations and regions Labour Markets

Labour-market institutions and worker flows: Comparing Germany and the US

Given the pressing need for labour market reforms in Europe, policymakers are looking to the Hartz I-IV reforms conducted in Germany in the mid-2000s for inspiration. To successfully apply their lessons one must understand why they worked. This column argues that the success of the Hartz reforms lay in improving matching efficiency between unemployed workers and vacancies – particularly effective in Germany where employment inflows are the main driver of labour market adjustment, in contrast to the US, where outflows play the primary role.

Reforming European labour markets ranks high on the current political agenda. Particular attention is being paid to the mid-2000s series reforms in Germany (the Hartz I-IV reforms) as a potential role model for other European countries. Before these reforms and the subsequent decrease in unemployment rates, politicians and economists had pointed to Germany as the ‘sick man of Europe’, a country characterised by high unemployment rates, low labour market turnover, and highly persistent of adverse shocks.

Pre-Hartz inefficiencies

What caused the poor labour market performance of Germany before the mid-2000s? To understand the recent decline in unemployment rates and the role played by the Hartz reforms one first has to examine their backdrop. There is no shortage of culprits accused in the literature. For example:

  • Ljungqvist and Sargent (1998) point towards the unemployment benefit system with high replacement rates linked to past earnings,
  • Blanchard and Portugal (2001) suggest a stronger bargaining position of workers in Europe, and
  • Bentolila and Bertola (1990) stress the role of layoff taxes and stricter employment protection legislation as driving factors of cross-county differences in average labour market flows.

These arguments are mirrored in ongoing debates about labour market institutions in Europe. For example, Bentolila et al. (2010) attribute high unemployment rates in southern Europe to the dual labour market and strict employment protection for senior workers. In the German case, Krebs and Scheffel (2013) and Krause and Uhlig (2012) view the reduction of replacement rates in the unemployment benefit system associated with step IV of the Hartz reforms as the main driver of the recent decline of German unemployment rates.  In contrast, Dustmann et al. (2014) point towards the decline in unionisation rates and wage moderation in Germany prior to the Hartz reforms as the main driver of the recent labour market success.

Why Hartz worked

In Jung and Kuhn (2014), we scrutinise the widely held belief that one (or combinations) of the above mentioned culprits – stricter firing protection legislation, higher unemployment benefits, or stronger unions – have been the key drivers of labour market rigidities in Germany from the 1980s to the mid-2000s. Based on new empirical evidence and the modern theory of search and matching, we find instead that inefficiencies in the matching process of unemployed workers to open positions are at the root of the problem. According to this view, rather than the cut in benefits (associated with reform step Hartz IV), the key to success of the Hartz reforms may have been in the reform steps Hartz I-III which were more directly aimed at improving matching efficiency by

  • Fostering alternative forms of employment (Hartz I),
  • Making marginal employment more attractive (Hartz II), and
  • Changing the organisational structure of the employment offices (Hartz III).

Indeed, increasing matching efficiency has been the explicitly stated goal of the Hartz commission’s mandate for designing the reform.

Our arguments are based on microdata from administrative records for the period of 1980-2004. Our empirical results have been recently confirmed by Hertweck and Sigrist (2012) using data from the Socio-Economic Panel (SOEP). Launov and Wälde (2013) refer to our arguments by exploiting the timing of the reform steps and the evolution of the unemployment rate from 2003 to 2005. Both studies find supportive evidence for a change in matching efficiency. Unfortunately, evidence on labour market dynamics in the post-reform period is still scarce. In ongoing work (Hartung et al. 2015), we provide new evidence on labour market dynamics to examine the channels through which the Hartz reforms improved Germany’s labour market performance from a macroeconomic angle.

The role of inflows

Why do our findings differ from previous research? Existing research on the differences in labour market dynamics between Europe and the United States has mainly focused on differences in average unemployment rates. Zooming in on the microstructure of worker flows, we confirm the fact that the share of workers going into (out of) unemployment each month is four (five) times smaller in Germany than in the United States for the 25 years from 1980 to 2004. This fact justifies the characterisation of Germany as a highly rigid labour market. However, we also show that differences in the labour market dynamics over the business cycle provide valuable additional information to identify the key cross-country differences in labour market institutions. A commonly held belief – which can be traced back at least to the work of Shimer (2005) – states that the outflows from unemployment to employment are the main driver of unemployment fluctuations, while inflows contribute relatively little. Figure 1 suggests that this U.S.[RB1] -based viewpoint cannot be easily applied to the German case. Despite stricter employment protection legislation, Germany (as well as many other European countries) uses mainly the firing margin to adjust to shocks over the business cycle. In contrast, the U.S. adjusts using the hiring margin. Inflows to unemployment (firings) react twice as much to business cycle shocks, and contribute 60% to the change in unemployment rates over the business cycle in Germany (but only 40% in the United States).

Figure 1. Business cycle component of EU rates (“firings”, red solid line) and business cycle component of unemployment rates (blue dashed line).

(a) Germany

(b) United States

This somewhat surprising fact is likely not a coincidence. As Figure 2 suggests, countries with lower average hiring rates tend to adjust to a negative shock by firing more, not by hiring less.

Figure 2. Data and fitted regression line from a linear regression of the log EU rate (‘firings’) volatility on the log UE rate (‘hirings’) and a constant.

Source: Data for OECD countries from Elsby et al. (2013). Norway and Sweden as two outliers are dropped.

Why does the nature of the cyclical behavior of firings matter for the current debate about the labour market reforms in Germany? If any of the above mentioned ‘culprits’ were quantitatively important, they would imply – at least viewed through the lens of the basic search and matching model – the opposite behavior, i.e. more rigid countries would rely more on adjusting through hirings, not less. The reason for this fact is simple. Higher benefits, stronger unions, or higher layoff taxes are institutions that make the surplus of being employed relative to being unemployed smaller. But, as is well known from the debate about the so called ‘Shimer puzzle’ (Shimer 2005) and (one of) its resolutions (Hagedorn and Manovskii 2008), a small average match surplus tends to increase the hiring rate volatility in the basic model by making the percentage deviation of profits over the cycle large. All the traditional explanations therefore fail to account for the business cycle dynamics.

We show that inefficiencies in the process of matching unemployed workers to open positions are a consistent explanation both for the differences in the average transition rates as well as for the cyclical adjustment process.  A lower matching efficiency reduces unemployment-to-employment transitions, increases the average search duration, and makes unemployment less attractive. As a result, the steady-state surplus of matches increases. This increase makes it less likely that an adverse idiosyncratic shock hitting a particular worker-firm match leads to a separation. This results in fewer transitions from employment to unemployment, which explains the low average inflow rates in Germany. The longer search duration implies that unemployed workers are less affected by cyclical productivity changes because their chances of reemployment are smaller. On the contrary, already-employed workers are directly affected. As a result, the worker surplus – the difference between the value of employment and unemployment – becomes more volatile. Because the EU rate is a function of the surplus, this mechanism links steady-state UE rates and EU rate volatilities in line with the data (cp. Figure 2).

A lower matching efficiency in Germany compared to the U.S. is commonly attributed to two sets of factors. A first set captures the credential-based occupational structure that limits the range of jobs applicable for an individual worker. The second set includes the institutional setup and the physical search technology. Examples are the equipment and services offered at employment offices, measures of active labour market policies, delay induced by involving work-councils, or details of the application and hiring process.

It is the second set of factors that might have important consequences for policy reforms. The Hartz IV reform, i.e. the reduction in unemployment benefits, has received much attention in the public debate. Our research provides a rationale for why the reform steps associated with Hartz I-III – designed explicitly to improve the efficiency of the matching process – may have been the key to the success.


While changing the benefit system, the bargaining power of unions, or the employment protection legislation is politically costly, changing the efficiency of a bureaucracy like a labour agency might be difficult but less politically controversial. Our research suggests that – at least for the period of 1980-2004 – differences in matching efficiency have been an important determinant of cross-country differences in labour market flows between Germany and the U.S. This suggests that a reform that was explicitly designed to foster efficiency along that dimension might have actually contributed to the ‘job market miracle’ in Germany.


Bentolila, S and G Bertola (1990), "Firing costs and labour demand: how bad is eurosclerosis?", The Review of Economic Studies 57.3: 381-402.

Bentolila S, P Cahuc, J J Dolado and T Le Barbanchon (2010), "Two-tier labour markets in the great recession: France vs. Spain" (No. 5340). Discussion paper series//Forschungsinstitut zur Zukunft der Arbeit.

Blanchard, Olivier, and Pedro Portugal. "What hides behind an unemployment rate: comparing Portuguese and US labour markets", The American Economic Review (2001): 187-207.

Dustmann, C et al (2014), "From Sick man of Europe to economic superstar: Germany's resurgent economy", The Journal of Economic Perspectives 28.1: 167-188.

Elsby M W L, B Hobijn and A Şahin (2013), "Unemployment Dynamics in the OECD", Review of Economics and Statistics 95.2: 530-548.

Hagedorn, M and I Manovskii (2008), "The cyclical behavior of equilibrium unemployment and vacancies revisited", The American Economic Review, 1692-1706.

Hartung B, P Jung, and M Kuhn (2015), "The effects of the Hartz Reforms on Labour Market Mobility in Germany", Working Paper, University of Bonn.

Hertweck, M S and O Sigrist (2012), "The aggregate effects of the Hartz Reforms in Germany", No. 532. SOEPpapers on Multidisciplinary Panel Data Research.

Jung, P and M Kuhn. (2014), "Labour Market Institutions and Worker Flows: Comparing Germany and the US", The Economic Journal.

Krause, Michael U., and Harald Uhlig. "Transitions in the German labour market: Structure and crisis", Journal of Monetary Economics 59.1 (2012): 64-79.

Krebs, Tom, and Martin Scheffel (2013), "Macroeconomic evaluation of labour market reform in Germany", IMF Economic Review 61.4: 664-701.

Launov, Andrey, and Klaus Wälde (2013), "Estimating incentive and welfare effects of nonstationary unemployment benefits", International Economic Review 54.4: 1159-1198.

Ljungqvist, L and T J Sargent (1998), "The European unemployment dilemma", Journal of Political Economy 106.3: 514-550.

Shimer, R (2005) "The cyclical behavior of equilibrium unemployment and vacancies", The American Economic Review: 25-49.

4,094 Reads