A Regionally Focussed Survival Analysis Into Re-employment Rates in New Zealand

Abstract

I apply survival analysis techniques to analyse regional variations in the probability that a jobseeker will transition from unemployment into employment. Controlling for variations in individual characteristics, I aim to identify whether an individual’s job search is affected by the New Zealand region in which they reside. Statistics New Zealand’s IDI dataset is used to link 2013 census responses to MSD benefit spell and residential address data. Unemployment spells are identified using administrative Unemployment and Jobseeker Support benefit spell records. The IDI provides a rich dataset on individual characteristics and allows for individual benefit spells to be tracked over long periods with precise information on regional assignment and spell dates. MSD data in the IDI allows for the isolation of specific benefit types ensuring that the identification of unemployed individuals is robust. Earlier New Zealand datasets were not capable of achieving this. Unemployment spells that begin during the full four calendar months either side of the 2013 census are tracked until data is unavailable after July 2020. Regional variation is commonly observed in foreign literature; however, the New Zealand literature is scarce. I conclude that regional variation does exist in New Zealand. Jobseekers in the Southland and Canterbury regions are most likely to transition into employment. Contrarily, Wellington and Northland exhibit the lowest transition probabilities. Jobseekers in these regions are on average 10% less likely to find work at any spell duration compared to a jobseeker in Auckland. Findings in this research are robust to various specifications and robustness models.