Worker flows in South Africa from 2008 to 2019: Evidence from matched quarterly labour force survey

Thesis / Dissertation


Permanent link to this Item
Journal Title
Link to Journal
Journal ISSN
Volume Title
This study examines the extent, cyclicality, and heterogeneity of worker flows for the working-age population of South Africa from 2008 to 2019. Worker flows are measured using quarterly panel data at the individual level, which is constructed by matching individual records between consecutive quarters of the cross-sectional data of the Quarterly Labour Force Survey (QLFS) from 2008q1 to 2019q2. This study expends a significant amount of effort in proposing and implementing a method for evaluating the quality of different matching procedures for the QLFS, which has never been accomplished systematically in previous literature. Using the matched QLFS panel, we find that the average quarterly worker flow rate was 46.26% over the period 2008 – 2019, which is large when compared with other developing countries. This is surprising in a labour market considered by many as rigid. Gross worker flows are procyclical. But its procyclicality is driven by the procyclical worker flows in the informal sector. In contrast, worker flows in the formal sector are acyclical. There is substantial heterogeneity in worker flows across different firms, jobs, industries, and sub-population groups. A series of descriptive analyses, complemented by a regression analysis of separation probabilities, reveal that worker flows are much larger for some groups (e.g., those who are younger or have lower levels of education), firms (e.g., those that are smaller in size or private), jobs (e.g., those with fixed-term contract relative to permanent contract), or industries (e.g., construction, agriculture, and domestic services) when compared to the rest. Therefore, although substantial gross worker flows in South Africa imply that the labour market as a whole is not as rigid as most people think, certain sub-sections of the labour market are definitely more rigid than others. Despite a myriad of data challenges and obstacles, we believe that this study has managed to capture the true extent and pattern of worker flows in South Africa. Measurement errors may exist in our estimates, but a robustness check shows a high degree of consistency between our worker flow estimates and those in the past literature, implying that measurement errors are moderate.