Job polarization and labour supply changes in the UK

Giulia Montresor, Navarra Center for International Development and Matthias Parey, University of Essex

(Project no. 1000198)

A growing number of studies have documented the polarisation of employment across many developed countries over the past two decades.

This paper aims to explore the determinants and consequences of employment polarization in the UK during the most recent decade through a threefold analysis. The first purpose of the study is to disentangle the causal effect of "routinization" from changes in the composition of local labour supply due to immigration surges and higher educational achievement.

We adopt a spatial analysis approach to provide first empirical evidence on how local UK labour markets have experienced job polarization to different degree, depending on their different level of local routine-intensity, graduate workers' share and immigration inflows. Have the areas with higher graduate workers' share registered the same employment decline in routine occupations? Have the areas that received less immigration inflows experienced the same significant increase in service employment? Did non-graduate natives filled those jobs? Results will be very useful in advising labour market policy making.

As regards the details of empirical implementation, we will be using the LFS to analyse job polarization over  time using econometric models.

A number of estimation problems arise due to the presence of unobserved time-varying characteristics correlated with both our dependent and explanatory variables.

We can solve this issue by using  the lagged values of our variables of interest sufficiently back in time. We therefore rely on census data since 1971 to construct local shares of routine occupation employment, graduates and immigrants' concentrations.

These variables will serve as instruments for my econometric analysis. For  this, we need access to individual-level micro-level data detailed by occupation, industry, country of birth, education, gender from censuses 1971 onwards. Data will be aggregate up to TTWA-level.