Econometric Issues in the Analysis of Linked Cross-Section Employer-Worker Surveys.

We consider the statistical problems involved in the econometric analysis of data from linked surveys of workers and employers. The context is a simple model of the incidence and duration of unemployement spells occurring between two waves of the UK New Earnings Survey, which have been linked to unemployment benefit records via the National Insurance number, and to Census of Production respondent firms via the Inter-Departmental Business Register.