Diagnosing Labor Market Search Models: A Multiple-Shock Approach

We construct a multiple shock, discrete time version of the Mortensen-Pissarides labor market search model to investigate the basic model’s well-known tendency to underpredict the volatility of key labor market variables. In addition to the standard labor productivity shock, we introduce shocks to matching efficiency and job separation. We conduct two set of experiments. First, we estimate the joint probability distribution of shocks that simultaneously satisfy the observed data and the first-order conditions of the multiple-shock model, and then simulate its properties. Although the multiple-shock model generates significantly more volatility while preserving the Beveridge curve relationship, it generates counterfactual implications for the cyclicality of job separations. Using a business cycle accounting approach, we design the second set of experiments to isolate the sources of model incompleteness and show that the model requires significant procyclical and volatile matching efficiency and counterfactually procyclical job separations to render the observed data without error. We conjecture that the basic Mortensen-Pissarides model lacks mechanisms to generate sufficiently strong labor market reallocation over the business cycle, and suggest nontrivial labor force participation and job-to-job transitions as promising avenues of research.