Asymmetric volatility spillovers between UK regional worker flows and vacancies

This paper investigates volatility spillovers between UK regional job finding, job separation and vacancy rates. Employing a logistic smooth transition vector autoregression (VAR) to model the large nonlinear dynamic system, we use the methods of Diebold and Yilmaz (2012) to decompose the forecast error variances. Our approach is Bayesian. More specifically, we extend doubly adaptive elastic-net Lasso (DAELasso) methods for VAR parameter shrinkage into a nonlinear framework to allow for the possible regime changes. We find that for each variable, both the volatility spillovers to and from other variables are high, providing clear evidence for the close interdependence between UK regional labour markets. The pivotal role of London in generating and spreading changes in volatility is highlighted. Analysis of net spillovers shows that, in general, shocks to job separation rates tend to spread into job finding and vacancy rates. By contrast, vacancy rates are usually at the receiving ends of shocks transmitted from the job separation and finding rates. We further examine the shock propagation mechanism in more detail, such as the differences in spillovers between regions within the same regime, and that of the same region but in different regimes. Finally, we draw inferences that are of economic and policy importance.