The South African labour market during COVID-19

Image: Ketut Subiyanto on Pexels.

Long before COVID-19 came to dominate our society, it had been well established that the labour market was one of the key institutions for understanding poverty and inequality dynamics in South Africa. Formal sector employment remains the primary pathway to achieve upward mobility for the majority of South Africans, while losing a job could have dire implications for the well-being of one’s family. It was thus concerning that the official unemployment rate had slowly but steadily been increasing, and stood at over 30% prior to the lockdown. And things have only gotten worse since then.

Over the past few months, several Saldrupians have been involved in research on employment dynamics using the first two waves of the NIDS-CRAM survey, and a total of five related papers have been produced.

Two papers were authored by Vimal Ranchhod and Reza Daniels, who profile the dynamics in the labour market between February and April, and then again for the period from April to June. Two similar but slightly different papers (see the Wave 1 paper and the Wave 2 paper) were authored by Saldrupians Rocco Zizzamia, Josh Budlender and Ihsaan Bassier, together with their co-author Ronak Jain. They also took their labour market analysis further by mapping from labour market dynamics to poverty dynamics. The final paper was written by Gabriel Espi-Sanchis, Murray Leibbrandt, and Vimal Ranchhod. They utilize the fact that the NIDS-CRAM sample was drawn from the long-running NIDS sample, and link data from individuals in the panel from 2008 to 2017 to the data for the same individuals in 2020. This allowed them to investigate how historical experiences in the labour market might correlate with differential degrees of vulnerability in the current moment. What follows is a brief synthesis of the key findings from these papers.

Before we begin, it is important to contextualise the moments in time which the data are referring to, in relation to the stages of the pandemic and related lockdown. The first wave of NIDS-CRAM was implemented in May and June, and asked questions about employment and earnings in February and in April. Thus, Wave 1 allows us to compare changes from February, i.e. before the lockdown and pandemic, to Stage 5 of the national lockdown which was enforced in April. Stage 5 entailed severe restrictions on people’s physical mobility and on business activities. Wave 2 was conducted approximately two months later, and asked about respondents’ labour market experiences in June. This coincided mostly with Stage 3 of the national lockdown, which entailed some restrictions on business operations and mobility, but was much less stringent than Stage 5. These three data points help us to tell a story about the time path of labour market experiences as we work our way through this pandemic.

The major finding from Wave 1, an overwhelming one, has been that millions of South Africans lost their jobs.

Between February and April, the percent of employed adults in the survey decreased from 57% to 48%. This had severe implications for welfare, which is corroborated by other research showing dramatic increases in self-reported levels of hunger.

A second substantial observation was the increase in the fraction of people who were on a temporary leave of absence from work.

Under normal circumstances, this is usually a very small percent of employees, most of whom are on vacation and still receive their remuneration. During the lockdown, this group of absentees increased dramatically. Some of these employees received full pay, but many received reduced pay, and a substantial fraction were put on unpaid leave. Thus, the conventional statistics do not adequately reflect the degree of hardship in the labour market. Different authors used different ways to try to address these issues, and this leads to slightly different estimates. Nonetheless, the overall pattern remains clear that if one classifies these furloughed workers as ‘not-employed’, then the fraction of adults in employment decreased further by about 10 percentage points in April.

The third point of consensus from these studies is that the hardship was disproportionately borne by already vulnerable groups in the labour market.

Low-wage earners were much more likely to lose their jobs. In terms of demographic groups, this included women, African/Black people, and people with at most a high school education. These patterns mirror the responses to a question on who is capable of working from home, where we found that White people and more highly educated people had substantially greater possibilities to continue working from home. The net result of this heterogeneity is that the losses in employment would both widen existing inequalities and amplify any impacts on poverty.

These dynamics were not only related to contemporary characteristics. By analysing the long-term experiences of individuals, we established that job losses under the lockdown were strongly correlated with people’s employment histories.

For example, 86% of the historically stable employed retained employment between February and April, compared with only 72% of the transient employed and 67% of the persistent non-employed who were employed in February. This feeds into a much deeper issue relating to long-term implications of the pandemic, and how it might adversely impact on people’s employment trajectories over their life course going forward.

By June, when the economy had started to open up again, we do see a substantial reduction in the fraction of people who were furloughed in April. A majority of these individuals had returned to work and were earning some wage income. But the ‘bouncing back’ was not complete. For one thing, the aggregate level of employment had not substantially increased. Put differently, the easing of the lockdown did not generate a return to the pre-pandemic levels of employment. Secondly, a substantial minority of people who were furloughed in April report having lost their jobs by June.

Nonetheless, the limited ‘bouncing back’ in June was correlated with a reversal of the patterns of poverty-enhancing job-losses that were observed between February and April. Low-wage workers were relatively more likely to return to work and earn an income. In addition, the roll-out of varying forms of state provided assistance also started to have a significant impact. The poverty analysis suggests that this combination of labour market dynamics and state support reduced the adult poverty rate between April and June by approximately 3 to 6 percentage points. This means that somewhere between 1 to 2 million individuals had escaped from poverty in this period.

Looking over the research collectively, it is clear that the past six months has been a period characterized by extreme volatility, uncertainty, and hardship.

At present, South Africa is in Stage 1 of the lockdown and almost all restrictions have been lifted. The pandemic seems to be dissipating, and the worst – from a health perspective – seems to have passed for now.

What remains uncertain are the longer-term prospects for the labour market. Will we return to our pre-pandemic levels of employment, which were already rather dire, or will we shift to a new steady state with substantially fewer jobs available? This depends to a large extent on the different processes that are unfolding concurrently in the economy. The longer-term dynamics will depend on a multitude of factors, many of which are beyond the control or influence of local institutions. How severe and long with the global recession be? What is going to happen to international trade and travel, the price of commodities, and will we see significant changes in market structures and supply chains?

As is usually the case, we have more questions than answers. With time and more data, we can (and will) answer some of these questions. Despite this lack of certainty, a few general points are worth emphasising. First, the labour market remains crucial to understanding welfare in the South African context, and is perhaps even more important at present than it was before the pandemic. Second, the labour market does not exist in a vacuum. It is inter-linked with many other spheres of society, and these are worth considering when one starts to think of potential interventions. Third, the present moment has generated a potential rupture in the way our labour market functions, some of which may be temporary but some of which will be long-lasting. It is worth bearing in mind that our pre-pandemic unemployment rate of 30% is a complete global and historical outlier. Perhaps it would be worth thinking hard about how we can do things differently and better, rather than simply getting back to ‘normal’.