Determinants of Long-Term Unemployment Percentages
The evolution of long-term unemployment percentages depicted in Figures 8.1 and 8.2 present a vivid picture of high levels and/or large increases for OECD-20 countries, as well as for CEE-FSU countries. This appendix reports the findings of regression equations intended to identify the determinants of these percentages. While the analysis was undertaken for both groups of countries, most of this appendix will focus on the OECD-20 countries. The relevant historical periods for the CEE-FSU countries are so short that little can be inferred from the regressions.
For 18 of the OECD-20 countries there were at least 15 years of data on the percentage unemployed 12 months or longer. 1 The specification used to explain movement in these percentages included three explanatory variables: the current year's unemployment rate, the unemployment rate lagged one year, and a linear time trend. It is well understood that equilibrium unemployment duration increases when the unemployment rate increases. However, the short-run effect of higher unemployment on the long-term percentage also depends upon the underlying average duration of unemployment and the speed with which unemployment increases. In countries with long duration, higher unemployment can initially reduce the percentage unemployed 12 or more months because the new unemployment inflow adds so many new spells of short duration.
The lagged unemployment rate was included in the regression equations because long-duration spells tend to persist even when unemployment decreases. Unlike the current unemployment rate where the expected sign on the coefficient is ambiguous, the lagged unemployment rate is expected to have a positive coefficient.
Finally, a trend was included to test for developments in the long- term unemployment percentage after controlling for effects of the current and lagged unemployment rates. A positive coefficient would indicate a secular increase in the percentage of long-duration spells. While