Aki Kangasharju

Returns to scale in a matching model: evidence from disaggregated panel data

The returns to scale in the matching function play an important role in models with endogenous search effort. Due to positive externalities, increasing returns to scale in matching can support high or low activity equilibrium in the labour market. In this study, we examine this issue using panel data from Finnish employment offices. The study finds that the results from the Cobb–Douglas and translog specification are qualitatively different. The CD specification of the matching function exhibits constant returns to scale. The translog specification, in turn, exhibits increasing returns to scale. The elasticity estimate for returns, using the preferred specification and minimum and maximum sample values for job seekers and vacancies, fall in the range of 1.1 to 1.6.

Matching Inefficiencies, Regional Disparities, and Unemployment

In this paper we apply a stochastic frontier approach to examine how matching efficiency and regional differences in structural factors contribute to regional and aggregate unemployment. Our results suggest that there would be a substantial decline in aggregate unemployment if (i) all local labour offices operated with full efficiency or (ii) they shared the same structure of job seekers and vacant jobs as the most favourable office. In the former case an increase in hirings would lower the average unemployment rate by 2.4 percentage points. In the latter case the decrease would be 1.4 percentage points. Further, we find that fixed effects are positively c0orrelated with both a more favourable structure and higher efficiency. This suggests that the fixed effects may capture some part of time invariant features in the structure and efficiency. Thus, the role of structural factors and efficiency in regional unemployment disparities may be higher than estimated.

Births-Deaths Nexus of Firms: Estimating VAR with Panel Data

This paper investigates dynamic interrelationships between the births and deaths of firms at the level of the 88 Finnish subregions during 1989–1993. The births-deaths nexus of firms is analysed using vector autoregressions (VAR) and an instrumental variable estimator. The main findings are: first, that the firm births equation has a two-year lag structure, which means that firm deaths cause firm births, and firm births cause subsequent births following two- year lags; and second, that according to the firm deaths equation, which has a one-year lag structure, firm births do not cause firm deaths, but firm deaths cause subsequent deaths. A further finding is that the use of the ordinary least-squares estimator produces inconsistent, and clearly different results from those produced by the instrumental variable estimator.