VoxEU Column Labour Markets

Why unemployment benefits should be extended in recessions: US evidence

During the last recession the US experienced some of the highest unemployment rates in its history, prompting the government to dramatically extend the duration of unemployment benefits. Economists have since debated whether this action helped or hindered the recovery of the labour market. This column finds evidence of a ‘silver lining’ effect – extended unemployment benefits led people to submit fewer job applications, but this reduction increased the chances of each application being successful. Hence the overall impact of extending benefits on aggregate unemployment was rather small.

Watch Ioana Marinescu discuss her research on unemployment benefit extensions in the video above

During the last recession, the US experienced some of the highest unemployment rates in its history – as high as 10%. At this time, the US labour market did no better than traditionally weaker European labour markets. To help out the millions of unemployed Americans who were struggling with their finances, the US dramatically extended the duration of unemployment benefits, tripling them from six months to almost two years. In the years since, economists have debated whether this action helped or hindered the recovery of the labour market. Some studies found a small effect on the unemployment duration of benefit recipients (Rothstein 2011, Farber and Valletta 2015), concluding that the extensions were benign. But others argued that this small effect was deceptive because it did not take into account the negative impact of the extensions on job creation. Thus, Hagedorn et al. (2013) claimed that the extensions greatly increased aggregate unemployment.

In a recent paper (Marinescu 2015), I find no support for a negative impact of the unemployment benefit extensions on job creation. Overall, the impact of the extensions on aggregate unemployment during the Great Recession of 2007-09 was rather small, because the extensions reduced excessive competition for jobs. That is, the security of unemployment benefits led people to submit fewer job applications, but that reduction in turn increased the chances for each application to turn into a job. This is the ‘silver lining’ effect of unemployment benefit extensions.

The debate: How much did unemployment insurance extensions increase US unemployment?

It is well established in labour economics that unemployment benefits increase unemployment duration, as people receiving more generous benefits take longer to move into jobs (Meyer 2002, Schmieder et al. 2012). Some theories hold that the total impact of benefit extensions on aggregate unemployment can be even larger: job seekers who can afford to wait longer before taking on a job may demand higher pay, and firms that cannot afford to pay higher wages may then simply not create jobs. A controversial study (Hagedorn et al. 2013) claimed that this dynamic was taking place, with the unemployment benefits extensions causing a decrease in the number of available jobs during the Great Recession.

The debate about the effect of unemployment benefits was directly relevant to policy at this time, because the federally funded extensions first legislated in 2008 were set to expire at the end of December 2013. The Council of Economic Advisers and the Department of Labor issued a joint note in December 2013 arguing in favour of maintaining the longer benefits. The report pointed to the small unemployment effects of the extensions documented by Rothstein (2011), and criticised the model and empirics of Hagedorn et al. (2013). However, Congress did not accept the report’s arguments and let the benefit extensions expire on 1 January 2014.

The silver lining: Unemployment insurance extensions reduced the competition for jobs

I have revisited the impact of unemployment insurance extensions using data from the leading US online job board, CareerBuilder.com, which contains about one-third of all US job vacancies (Marinescu 2015). The key advantage of the data is to measure both the number of jobs and the number of job applications. The study is aided by the fact that, for reasons related to complex statutory conditions, benefits extensions began at different times in different US states. Because extensions depend on rules rather than policymakers’ discretion, the differences between states effectively constitute a quasi-experiment, giving us more confidence in identifying the causal impact of benefits extensions.

The extension in the duration of unemployment insurance did indeed dramatically decrease job applications. Figure 1 illustrates the average decline in applications for US states, relative to the time that each state imposed its largest increase in potential benefit duration. Overall, we can calculate that a 10% increase in benefit duration leads to a 1% decline in the total number of applications that job postings in a state receive (Marinescu 2015). The unemployed do indeed appear to search less hard when the duration of benefits increases.

Figure 1. The impact of potential unemployment benefit duration on job applications

However, this dramatic reduction in job applications was not accompanied by a parallel reduction in the number of available jobs (Figure 2, which is constructed in the same way as Figure 1). This combination of facts implies that each individual job application had a higher probability of success. That is, reduced competition for jobs meant that job-seekers needed to submit fewer applications to successfully find employment. These results show that it is not possible to understand the overall effects of unemployment policies by considering only the negative direct impact on unemployed benefit recipients. Instead, one must also consider the indirect ‘silver lining’ effects of the policy on the labour market itself.

Figure 2. The impact of potential unemployment benefit duration on job vacancies

In the Great Recession, the two effects almost counterbalanced each other, so that extending unemployment benefits had little effect on aggregate unemployment. Indeed, one can quantify the size of the effect using the theoretical model of Landais et al. (2014). Using this method, I calculate that a 10% increase in unemployment benefit duration increased aggregate unemployment by only 0.6% during the Great Recession in the US. Without taking the ‘silver lining’ effect into account, the impact would be nearly twice as large.

Extending unemployment benefit duration during a recession is a good policy

What policy lessons can be drawn from the US experience? In order to go from empirical results to conclusions about the desirability of a policy, it is necessary to make some theoretical assumptions about how the labour market and unemployment insurance work. Landais et al. (2014) present a theoretical model that relies on the ‘silver lining effect’ that I documented in the data. Consider the difference between boom and recession. In a boom, too many employers are chasing too few job seekers, so the labour market is ‘tight’ (i.e. a low number of applications relative to the number of vacancies) and hiring is too difficult. In a recession, the inverse situation occurs: too many job seekers are chasing too few job openings, labour market tightness is low, and finding a job is difficult. Adjusting unemployment insurance is a means of smoothing cycles in labour market tightness, i.e. of dampening the swing from boom to recession that might otherwise happen.

According to the theory, whether unemployment benefits should be more or less generous during a recession depends on the impact of unemployment benefits on labour market tightness. My empirical results and the theory taken together lead us to conclude that unemployment benefits should be more generous during a recession. If unemployment insurance is made more generous during a recession, the many unemployed job seekers are encouraged to search less hard, reducing competition for jobs (as documented in Marinescu 2015) and helping job seekers find a job more easily. If it is made less generous during a boom, the few unemployed job seekers are encouraged to search harder, decreasing labour market tightness and helping employers hire more easily.

Was it a good idea for the US to increase the duration of unemployment benefits during the Great Recession? Despite the fear that more generous unemployment insurance would increase aggregate unemployment, this effect was likely very small. Moreover, unemployment insurance had additional economic benefits that we have not previous discussed: it supported the consumption of the unemployed, especially as credit had dried up following the Global Crisis of 2007-08. Taking into account the impact on the labour market and the positive impact on consumption, there are good reasons to believe that extending unemployment benefits during a recession is a good policy choice.


Farber, H. S. and R. G. Valletta (2015), “Do Extended Unemployment Benefits Lengthen Unemployment Spells? Evidence from Recent Cycles in the U.S. Labor Market”, Journal of Human Resources 50(4), pp. 873–909.

Hagedorn, M., F. Karahan, I. Manovskii and K. Mitman (2013), “Unemployment Benefits and Unemployment in the Great Recession: The Role of Macro Effects”, NBER Working Paper No. 19499. National Bureau of Economic Research.

Landais, C., P. Michaillat and E. Saez (2014), “Optimal Unemployment Insurance over the Business Cycle", CFM Discussion Paper No. 2013-03.

Marinescu, I. E. (2015), “The General Equilibrium Impacts of Unemployment Insurance: Evidence from a Large Online Job Board”,

Meyer, B. D. (2002), “Unemployment and Workers’ Compensation Programmes: Rationale, Design, Labour Supply and Income Support”, Fiscal Studies 23(1), pp. 1–49.

Rothstein, J. (2011), “Unemployment Insurance and Job Search in the Great Recession”, Brookings Papers on Economic Activity 43(2), pp. 143–213.

Schmieder, J. F., T. von Wachter and S. Bender (2012), “The Effects of Extended Unemployment Insurance Over the Business Cycle: Evidence from Regression Discontinuity Estimates Over 20 Years”, The Quarterly Journal of Economics 127(2), pp. 701–52.

8,189 Reads