At a time when unemployment rates in France, Italy, the UK, and the US are stuck around 8%-9%, many are turning to the apparent miracle in the German labour market in search of lessons. In 2008–09, German GDP plummeted 6.6% from peak to trough, yet joblessness rose only 0.5 percentage points before resuming a downward trend, and employment fell only 0.5%. In August 2011, the standardised unemployment rate was about 6.5%, the lowest since the post-reunification boom of 20 years ago (Source: Bureau of Labour Statistics).
Reaction to this success has bordered on smugness in some quarters. Many have asserted that superior German labour-market institutions, particularly those facilitating reductions in hours per worker in a recession, were chiefly responsible for the miracle (Schneider and Graef 2010, Klös and Schäfer 2010, Möller 2010, German Council of Economic Advisors 2010). Yet these institutions have been in place for decades, so cannot explain why of five recessions since 1970, only the 2008–09 episode was not accompanied by plunging employment and a stubbornly higher unemployment rate. Recognising this, other German observers stress the continuing importance of reforms to the labour market enacted in 2003–05 (Gartner and Klinger 2010), and the absence of wage growth since 2001 (Boysen-Hogrefe and Groll 2010, Gartner and Klinger 2010).
In recent research (Burda and Hunt 2011), we show that one unnoticed explanation can explain a significant component of the reputed economic miracle. In the export-driven expansion of 2005–07, firms hired significantly less than that expected, given the extent warranted by GDP and wages. As evidenced by data on firm expectations and articles in the business press, firms had unusually low confidence the boom would last. We show that the missing employment increase in the 2005–07 boom was equivalent to 40% of the missing employment decline in 2008–09, and that more than half of the missing employment increase was attributable to firms’ pessimistic expectations. Firms hesitated to hire in the boom, as they feared it would not last, and when eventually the recession they feared arrived, they had less need to fire than in previous recessions.
Did flexibility in hours per worker shield the labour market?
American observers are rightly intrigued by the system of short-time work used in Germany and certain other European countries (Boeri and Bruecker 2011, OECD 2010). Under this system, a firm in financial difficulties can apply to the employment agency for approval of a package whereby the firm reduces workers’ hours and variable pay in proportion and refrains from layoffs, while workers are reimbursed by the employment agency for 60-67% of their lost pay. Abraham and Houseman (1993) showed this system leads to more cyclical hours per worker and less cyclical employment than in the US. Hours per worker indeed fell precipitously in the 2008–09 recession, by almost 4%. Yet the hours of short-time work compensated were no greater than in the 1974–75 recession.
There is another factor which helps account for the significant reduction of hours during what many now call the ‘Great Recession’. A new personnel management tool, working-time accounts, has been used increasingly by German employers to adjust hours per worker flexibly to short-term fluctuations. Employers may increase hours above standard hours without immediate additional pay, as long as hours are reduced in the future with no loss of pay so as to keep average hours at standard hours within a moving time window. The number of hours the employer owes the worker is tracked in a working-time account. The share of workers with such an account rose from 33% in 1998 to 48% in 2005 (Gross and Schwarz 2007), possibly permitting greater adjustment of hours per worker in the 2008–09 recession than in earlier recessions.
At the same time, an important finding of our research is that the fall in overall hours per worker in the Great (German) Recession was not surprising given its severity. It must have been the case that employers used working-time account balances to postpone or reduce their use of government short time, leaving overall hours flexibility little changed. On the other hand, working-time accounts probably played a role in stemming the employment decline by providing a disincentive to fire, since employers must compensate laid-off workers for any outstanding positive balance. Employer behaviour in cutting hours per worker was similar to the past, but they retained more workers on shortened hours. A theoretical model which differentiates between employment changes at the extensive and intensive margin can explain this behaviour, to the extent that underlying parameters of that model changed over the past 10-15 years.
Does wage moderation explain the employment pattern?
Worker bargaining power has declined in Germany since reunification. This is reflected by a decline in union membership and coverage of union agreements, increasingly decentralised wage bargaining, and union concessions such as the use of working-time accounts. One causal factor is likely the emergence of Eastern Europe as a competing location for manufacturing production. Since 2001, average hourly wages have stagnated in real terms. Assuming an ‘off-the-shelf’ estimate of labour-demand elasticity with respect to wages of 0.7 – we show that if wages had continued rising at the pace of the 1990s (1.12 log points per year), employment would have declined much more in the recession – by an amount equal to 20% of the missing employment decline.
A potential role for labour-market reforms
In 2003–05 a series of labour-market reforms was undertaken in Germany: more experience was required to qualify for unemployment benefits, the duration of benefits was cut for the older unemployed, the follow-on benefit was merged with the less generous welfare programme, the onus of job search was put on the unemployed person for the first time, sanctions for refusing job offers were enforced, private firms were recruited to help place the unemployed, the employment agency was reorganised and temporary work agencies deregulated. These reforms would have been expected to lower unemployment, and there is some evidence that they did. If their effects occurred with a lag extending into the recession, the reforms could have moderated the employment decline. There is some evidence consistent with this – the Beveridge Curve continued to shift in throughout the recession, while the share of long-term unemployed continued to decline throughout the recession – but it is impossible to judge the likely magnitude of the effect.
Expectations in the 2005–07 expansion
Our account for the missing hires in the 2005–07 expansion relies on pessimistic firm expectations, and the increasing cost of adjusting employment at the extensive margin, which make expected future determinants of labour demand more important in the determination of current employment plans. In our paper, we first identify manufacturing as the industry with the missing employment increase in the boom and the missing employment decline in the recession. We then use data on firm expectations from the Ifo Institute to show that in manufacturing, but not other sectors, there was an anomalous gap in the 2005–07 boom between firms’ assessment of the current situation and their expectations for six months in the future. Figure 1 illustrates this, plotting the difference between the share of firms reporting their current situation is favourable versus unfavourable, and the difference between the share expecting an improvement in the subsequent six months versus a deterioration. Only in the uncertain period immediately following reunification were expectations similarly low compared to the assessment of the current situation. Our analysis of reports in the business daily Handelsblatt during the expansion confirms that firms were uncertain the boom would last.
Figure 1. Manufacturing firms’ business assessments
Note: Current refers to the difference between the share of firms reporting their current situation is good rather than poor; Expect refers to the difference between the share of firms reporting they expect their situation in six months to be more favourable rather than less favourable.
Source: Ifo Institute, Munich.
We use the Ifo data to generate counterfactual expectations based on the historical relation between firms’ assessment of the current situation and expectations, and we use the counterfactual expectations to predict what the employment increase in the boom would have been had firms formed their expectations as in the past. We find that pessimistic employer expectations in the boom explain 56% of the missing employment increase, and hence 23% of the missing employment decline in the recession.
It would be heartening were institutions facilitating cuts in hours per worker able to eliminate employment declines in a recession entirely, but this does not appear to be the case. Rather, the extraordinary performance of the German labour market in 2008–09 can most clearly be tied to a possibly one-off event with less favourable social welfare implications. In other words, firms had less need to lay off than in a typical recession, because an unusual lack of confidence in the preceding boom had made them reticent to hire.
Some of the dampening of the employment cycle may also be due to firing disincentives implicit in the system of working-time accounts, rather than to the 2005–07 firm pessimism we were able to quantify. Other countries may want to examine this labour practice, as well as the labour-market reforms enacted in 2003–05.
In addition, growing concerns about shortages of skilled labour could reflect increasing marginal costs of changing the extensive margin, which would also lead to a more sluggish, drawn-out reaction to expected future determinants of labour demand. At the same time, by 2009 there was evidence in the business press that firms were confident that the (export-driven) recession would be relatively brief. It is not entirely clear that this will be repeated in future recessions.
Abraham, Katharine G and Susan N Houseman (1993), Job Security in America: Lessons from Germany. Washington, Brookings Institution.
Boeri, Tito and Herbert Brücker (2011), “Short-Time Work Benefits Revisited: Some Lessons from the Great Recession”, Institute for the Study of Labour Discussion Paper 5635, April.
Boysen-Hogrefe, Jens and Dominik Groll (2010), “The German Labour Market Miracle”, National Institute Economic Review, 214:R38-R50.
Burda, Michael, and Jennifer Hunt (2011), “What Explains the German Labor Market Miracle in the Great Recession?” CEPR Discussion paper 8520, August.
Gartner, Hermann and Sabine Klinger (2010), “Verbesserte Institutionen für den Arbeitsmarkt in der Wirtschaftskrise”, Wirtschaftsdienst, 90(11):728-734.
German Council of Economic Advisors (2010), Jahresgutachten 2009/10: Die Zukunft nicht aufs Spiel setzen. Paderborn: Bonifatius.
Gross, Hermann und Michael Schwarz (2007), “Betriebs- und Arbeitszeiten 2005: Ergebillionisse einer repäsentativen Betriebsbefragung”, Beiträge aus der Forschung: Sozialforschungsstelle Dortmund, 153:1-174.
Klös, Hans-Peter and Holger Schäfer (2010), “Krisenmanagement über Variationen des Arbeitsvolumens?”, Arbeit, 2-3:132-146.
Möller, Joachim (2010), “The German labour market response in the world recession – de-mystifying a miracle”, Zeitschrift für Arbeitsmarkforschung, 42(4):325-336.
Organisation for Economic Cooperation and Development (OECD) (2010), Employment Outlook 2010: Moving Beyond the Jobs Crisis, Organisation for Economic Cooperation and Development.
Sachverständigenrat zur Begutachtung der gesamtwirtschaftlichen Entwicklung (2010), Jahresgutachten 2009/10: Die Zukunft nicht aufs Spiel setzen Paderborn: Bonifatius GmbH Buch-Druck-Verlag.
Schneider, Stefan and Bernhard Graef (2010), “Germany’s jobs miracle: Short-time work, flexible labour contracts and healthy companies”, Deutsche Bank Research Briefing, 27 April.