During the Great Recession, the unemployment rate increased from just over 5% in 2008 to 10% at the beginning of 2010. By the beginning of 2015, it had almost returned to the 2001 - 2007 average 1. Good news, right? However, most of this decline in the unemployment rate can be accounted for by the normalisation of short-term unemployment rates (i.e. people who have been unemployed for less than 26 weeks). By 2014, the short-term unemployment rate had returned to the 2001 - 2007 average 2. The long-term unemployment rate (i.e. people who have been unemployed for more than 26 weeks), on the other hand, remains just less than double the pre-recession level 3. On current projections, it will end up taking a decade or so from the beginning of the Great Recession before long-term unemployment returns to the pre-recession average. For policymakers, that's not good. It represents roughly 2.7 million people in the United States who are chronically out of work. If we're faced with another unemployment shock like the Great Recession, we want to know in advance why long-term unemployment is so sticky, so that we can implement the appropriate policies to get people back into work.
One of the theories why this is happening is that recessions induce a compositional shift in the demographics, occupations and regional distribution of the unemployed. The idea is that the type of person who is unemployed during a recession is different from the type of person who is unemployed during an expansion. Darby, Haltiwanger, and Plant (1985) argue that the people who end up unemployed during a recession are people who would have a low probability of finding a job no matter when they were unemployed, such that much of the increased unemployment rate can be explained by an influx of people who deflate the average job-finding probability. This theory doesn't have very much empirical support. This paper proposes a model of labour market dynamics (movements among employment, unemployment and non-participation) and for each demographic-occupation-region group, they calculate from Bureau of Labour Statistics data the pre-recession long-term unemployment rate. Holding those within-group unemployment rates fixed at pre-recession levels, they then allow the composition of the labour market to evolve over the Great Recession and find almost no evidence that compositional shifts can explain why long-term unemployment rose so high, nor why it's taking so long to come down.
With that theory not holding water, another one is becoming popular (and this is my personal favourite), which says that there is a form of duration dependence in labour markets. That's econo-jargon for saying that unemployment is somewhat self-fulfilling. The longer you are unemployed, the more your skills atrophy, the more your professional networks erode, and the more that employers look disfavourably at your resume for those reasons, and so you stay unemployed. Under this theory, the long-term unemployed are basically 'unlucky', and get left behind as job vacancies go to the short-term unemployed and new entrants to the labour force. There is some good evidence to support this.
This paper is a very interesting experiment. The authors sent fictitious resumes to actual job postings in 100 U.S. cities, taking care that the randomly-generated names were not racially-informative (since we have a whole body of research on racial discrimination in labour markets). For each job applied to, except for administrative / clerical jobs, 2 male and 2 female resumes were sent and 2 of those were 'low quality' while 2 were 'high quality'. For each resume, employment status and the length of current unemployment spell was randomised. The results show that job-finding probabilities attenuate really quicky as the duration of unemployment goes up. It drops from almost 30% for a duration of 0 months to 8% for a duration of 8 months. Another paper by the same authors takes a more structural approach to this same question. It introduces a model of labour market dynamics. Within that model, it replaces the observed job-finding probability (calculated again from Bureau of Labour Statistics data) with a job-matching function, calibrated in the pre-recession period, that depends on the duration of unemployment. The idea is to test whether a model that holds fixed the structure of job-matching and only allows variation by duration of unemployment can explain what happened to long-term unemployment in the Great Recession. The answer? Kinda. It explains a good chunk of what happened (see Figure 9). So did the other component of their paper, which is to allow movement in and out of the labour force itself. There's still a part of the increased long-term unemployment rate that can't be explained by anything described above. The mystery of long-term unemployment continues, although we know much more now than we used to.
Questions: Why do you think the long-term unemployment rate is so high? Do you think there's merit to the idea that unemployment is self-fulfilling? Any thoughts about what policies should be implemented to prevent long-term unemployment from becoming such an issue in the first place?
[1]
[2] [3]