By Jake Bradley
http://d.repec.org/n?u=RePEc:iza:izadps:dp15017&r=dge
There has been a substantial body of work modeling the co-movement of employment, vacancies, and output over the business cycle. This paper builds on this literature, and informed by empirical investigation, models worker and firm search and hiring behavior in a manner consistent with recent micro-evidence. Consistent with empirical findings, for a given vacancy, a firm receives many applicants, and chooses their preferred candidate amongst the set. Similarly, workers in both unemployment and employment, can evaluate many open vacancies simultaneously and choose to which they make an application. Business cycles are propagated through turbulence in the economy. Structural parameters of the model are estimated on U.S. data, targeting aggregate time series. The model can generate large volatility in unemployment, vacancies, and worker flows across jobs and employment state. Further, it provides a theoretical mechanism for the shift in the Beveridge curve after the 2008 recession – a phenomenon often referred to as the jobless recovery. That is, persistently low employment after the recession, despite output per worker and vacancies having returned to pre-crisis levels.
What attracted me to this paper is the promise to explain the job-less recovery. I understand that the mechanism at play is that half of job matches are with workers who currently have a job. In a sense, a recovery is happening with workers reallocating to more efficient jobs, and thus little change in the number of workers. Fine. But then, why did this not happen in previous recoveries? What changed? The paper left me hanging there.
NEP-DGE Blog 
Key to the mechanism in the paper is that to understand the movement in unemployment one also needs to understand the distribution of productivity amongst the employed. In a standard search model, following a shock, unemployment moves extremely quickly to its new steady state. However, the distribution of productivity amongst the employed is far more sluggish. Thus, a deep and prolonged recession (as in 2008) will look very different from other recent recessions.