Job Flows, Worker Flows and Churning in South Africa
Author | Andrew Kerr |
DOI | http://doi.org/10.1111/saje.12168 |
Published date | 01 January 2018 |
Date | 01 January 2018 |
JOB FLOWS, WORKER FLOWS AND CHURNING
IN SOUTH AFRICA
ANDREW KERR*
†
Abstract
How large are worker flows in a country where only 43% of the working age population are in
employment? Rather surprisingly, the answer is that worker flows in South Africa are relatively
large. In this paper, worker and job flows are estimated using anonymised IRP5 tax certificate
data from the South African Revenue Service. The data used in this paper is from the 2011–
2014 tax years and contains information on more than 12 million individuals and nearly
300,000 firms. The main finding of the paper is that worker flows are substantial, around 53%
per year, or 58% when employers classified as engaged in “public administration” are excluded.
One interpretation of this finding, as well as the patterns of separations and hires by firm growth
rates, is that the labour market is not as rigid as had previously been thought. Worker flows are
declining in firm size and in median earnings in the firm, they vary substantially by industry and
are very low in the public sector. There is heterogeneity in worker experiences – while flows are
high on average there are many workers that have very stable employment but those workers in
the bottom quintile have worker flows more than three times those in the top quintile. Heteroge-
neity and persistence in worker flows and churning in firms are important – some firms have
high levels of churning whilst others have much lower levels. Measurement error in the data
about period employed from the IRP5 tax certificates is a concern but as far as can be ascertained
this does not seem to be responsible for creating higher worker and job flows.
JEL Classification: J21, J23, J63
Keywords: Churning, job and worker flows, labour regulation
1. INTRODUCTION
Worker and job flows are a pervasive feature of labour markets. Every year a substantial
fraction of the workforce either start working for a new employer or leave an old
* Corresponding author: 3rd floor, School of Economics Building, Middle Campus, Cape
Town 7701, South Africa. E-mail: andrew.kerr@uct.ac.za
†
DataFirst, University of Cape Town.
I thank the South African Revenue Service (SARS), UNU-WIDER, and the South African
National Treasury for support. The opinions expressed in this paper do not reflect the views
of SARS, UNU-WIDER or the National Treasury. Without implicating any of them in the
conclusions drawn in this paper, I thank Liz Gavin, Kholofelo Makua, and Michelle Smit
from SARS; Aroop Chatterjee, Chris Axelson, Wiaan Boonzaaier, Marle Van Niekerk, and
LandonMcMillan from the National Treasury; as well as Channing Arndt, Lawrence Edwards,
Friedrich Krueser, Tasha Naughtin, Vimal Ranchhod, Neil Rankin, Nicola Viegi, and Martin
Wittenberg for helpful comments, suggestions, and advice on using the data. The suggestions
and comments of the two anonymous reviewers were also extremely helpful in revising the
paper. All output used in the paper has been checked by the National Treasury to ensure that
it does not compromise the anonymity of any firm or worker.
V
C2017 UNU-WIDER. South African Journal of Economics published by John Wiley & Sons Ltd on behalf of
Economics Society of South Africa.
This is an open access article under the terms of the Creative Commons Attribution-NonCommercial License, which
permits use, distribution and reproduction in any medium, provided the original work is properly cited and is not
used for commercial purposes. doi: 10.1111/saje.12168
141
South African Journal of Economics Vol. 86:S1 January 2018
South African Journal
of Economics
employer. Simultaneously firms are contracting or expanding the number of workers
they employ. Patterns of hiring and firing are complex, with shrinking firms hiring and
growing firms firing workers (Hamermesh et al., 1996). But how large are worker flows
in a country where only 43% of the working age population are in employment? Rather
surprisingly, the answer is that worker flows in South Africa are relatively large.
In this paper the extent of job flows, worker flows and churning in South Africa is
documented using a new source of administrative data derived from anonymised tax
records. A better understanding of job and worker flows is important in South Africa
because these shed light on labour demand, labour reallocation and the effects of the reg-
ulatory environment on the labour market, which are not well understood and which are
critical to understand if South Africa’s extremely high rate of unemployment is to be
lowered.
New data made available by the South African Revenue Service (SARS) make the
measurement and analysis of job and worker flows possible in South Africa. These data
are obtained from the IRP5 tax certificates, which are a census of all worker-firm spells
for which a worker earned more than ZAR2,000 per year in Pay As You Earn (PAYE)
tax registered firms. At the time the data was analysed it included individual and firm
identifiers, individual earnings and period of employment in each tax year, along with a
firm four-digit industry code. Future releases of the data may include worker age and
gender and further details about the firm that can be merged in from the company tax
data held by SARS.
This matched firm-worker data can be made into a panel of workers and then also of
firms. This firm dataset is comprised of tax-paying, and thus formal, firms. The exclusion
of firms not registered for PAYE tax is not as much of an issue in South Africa as in other
developing countries because South Africa has a smaller informal economy relative to
these countries (Magruder, 2012). The total number of individuals in the data used in
this paper who are employed in March 2011 and March 2014 is approximately 9.5 and
10.1 million, compared to the estimate from the South African Quarterly Labour Force
Survey (LFS) of approximately 13.1 and 15 million from the first quarters of 2011 and
2014. The tax data thus cover between 67% and 72% of all employment in the country,
a very similar number to the 70% of the employed in the LFS data who consider the
business they work in formal. Any worker not working in a PAYE registered firm would
be excluded from the IRP5 data.
In this paper the extent of worker and job flows is estimated using the SARS IRP5
data. Worker flows are substantial, around 53%, or 58% if public administration is
excluded, which is higher than most European countries but lower than in the United
States and the few developing countries for which worker flows have been estimated
using administrative data (Davis and Haltiwanger (1999), Bellmann et al., 2011). These
relatively high worker flows are perhaps surprising given the reputation of the South Afri-
can labour market as extremely rigid (see Go et al., 2009). The finding in this paper of
extensive worker flows when using the IRP5 tax data is consistent with previous work
using worker-level panel data by Banerjee et al. (2008), and contradicts the assertion of
Go et al. (2009) that the high level of worker movement between the informal sector,
the formal sector, unemployment and not economically active estimated by Banerjee
et al. (2008) was due to more fluid transitions between non-employment and the infor-
mal sector. Even in the SARS IRP5 data of individuals working for PAYE tax registered
firms there are still substantial worker flows.
142 South African Journal of Economics Vol. 86:S1 January 2018
V
C2017 UNU-WIDER. South African Journal of Economics published by John Wiley & Sons Ltd on behalf of
Economics Society of South Africa.
To continue reading
Request your trial