A Dynamic Approach to the Study of Unemployment Duration

The aim of this paper is to analyze the influences of labor market
structure and of the individuals characteristics on the unemployment
duration using the survival analysis.
The changing labor market affects job opportunities. We propose
introducing the changing labor market structure into event history models
for transition rates from unemployment to employment. Estimating the risk
models with time-varying covariates we calculate the risk of leaving the
unemployment.
To avoid a linear dependency of covariates, we try to find measures for
the theoretically important macro effects such as unemployment rate,
proportion of registered vacancies of all jobs, proportion of employed in
public sector, income per capita, private consumption, level of
productivity, number of firm failures, proportion of bad debt in the
banking sector. Using factor analysis with principal factoring we find
orthogonal factors explaining most of the variance in the time series.
The original unemployment episodes are split into subepisodes every month
so that the factor scores could be updated in each episode for all the
unemployed every month. We find that the more favorable the business cycle
is, the easier is to move up in the labor market. The positive effects of
changes in the labor market structure lead to increasing opportunities for
people to find jobs.
Hazard models with time-varying covariates will comprise other observable
characteristics of individuals such as gender, age, place of residence,
education level, individual labor force experience, year of entry in the
labor market too.
To estimate the models, we use data from the labor offices in Poland