Random and stock-flow models of labour market matching - Swedish evidence

In this paper we estimate aggregate matching functions taking advantage of a rich data base that enables us to compute observations on the variables in the matching function at (virtually) any frequency to assess the importance of the time aggregation problem. We also generate stocks, outflows and inflows of vacancies and job seekers to shed light on the importance of stock-slow matching. Finally, we assess the contribution of labour market programme participants to matching. <p> Our evidence rejects random matching. More precisely, we find that a non-trivial fraction of new job seekers match instantly (within the first week), that stocks of "old" vacancies and job seekers do not contribute significantly to matching and that the inflow of vacancies matches with the lagged stock of job seekers. Our results also suggest that labour market programme participants contribute to matching to a lesser extent than openly unemployed job seekers. <p> We also find that the use of lagged stocks as right-hand side variables in matching functions (i.e., ignoring the within-period inflow of job seekers and vacancies) gives lower estimates of matching elasicities and that this is more pronounced the lower the measurement frequency.