In October 2020 the South African government launched a collection of public employment programmes, initially intended as a response to the Covid-19 pandemic. The initiative, called the Presidential Employment Stimulus, has been extended since then. The total budget allocation to March 2024 was R42-billion.

By December 2023 it had directly created 1.8-million jobs and livelihood opportunities. These have been mostly temporary jobs in public employment programmes such as school education assistants. It has also included financial support to various sectors.

An important question is how much the programme’s spending stimulates economic activity in local communities and nationally. That is, to what extent it supports job creation or higher incomes outside the programme.

South Africa has an exceptionally high unemployment rate (32% or 41%, depending on the definition), particularly concentrated among the youth. Economic growth has been stagnant for the last 15 years, and increasing pressure has been placed on the national budget. While the core objective of a public employment programme is to provide direct employment and improve service provision, in this context it is also important to understand how it might stimulate economic activity.

At the Southern Africa Labour and Development Research Unit at the University of Cape Town, we were commissioned by the Presidency to draw on our knowledge of programme evaluation and South Africa’s social assistance policy to try to answer this question. We evaluated spending associated with the largest component of the programme – the Basic Education Employment Initiative.

We found that the programme likely does support broader economic activity, and these effects partly persist after the end of the programme. Participants buy goods which are produced to some extent in local value chains, and which employ local labour, rather than being imported. The programme spending does not just “disappear”, but recirculates in the South African economy.

The study

The Basic Education Employment Initiative has employed about 245 000 young people per phase to assist schools across the country. The duration of employment has varied with each phase. More recently it has been eight months. Participants are employed in full-time positions and are paid the monthly national minimum wage, which is approximately R4 000.

The programme completed its fourth phase in 2023. Since it was launched in December 2020 it has employed over 850 000 young people, becoming the largest youth employment programme in South Africa’s history.

In our study, we focused on phases 2 and 3 of the programme, from November 2021 to August 2022.

First, we looked at how the programme affected participant spending patterns. We then estimated what kind of economic activity this spending supported.

Our initial evidence came from a WhatsApp survey of 31 250 participants we ran with Harambee Youth Employment Accelerator, a non-profit which supported the programme in partnership with the Department of Basic Education. Harambee holds contact details of most participants for phases 2 and 3, with permission that the records may be used for programme evaluation.

The survey response rate was unfortunately low. But it showed participants spent their cash mostly on groceries (about 50%), transport and rent.

Most of their income went to necessities, much of it from local stores.

However, our main evidence comes from information provided by a leading grocery retailer. The retailer gave us limited access to fully anonymised sales records from its loyalty rewards programme.

In partnership with Omnisient, a privacy-preserving data collaboration platform, we were able to see who in the data was a participant in the programme and who wasn’t, while retaining individual anonymity. We explain in the paper how this was done without revealing or sharing any personally-identifying information. The data collaboration partnership went through a rigorous legal process and received University of Cape Town Research Ethics clearance.

Using this data, we found that average participant spending at the retailer increased from R327 per month before the start of the programme to R437 during the programme.

When compared to a control sample of other customers who shopped at the same locations and kinds of stores as the participants, using a statistical analysis method called difference-in-differences, we found that participant spending sharply increased by 15% during the programme.

Even after the programme ended, participants’ spending remained 4% higher than the baseline.

This might be due to participant savings during the programme, or participants being better placed to find work after the programme ends.

But this aggregate spending increase hides quite a lot of variety, per Table 1. In the largest spending categories, participant spending increased by 16% (groceries; refrigerated and frozen perishables) and 20% (toiletries), but in some smaller categories the percentage increase was much higher (off a low base).

For example, spending on home and small appliances increased by 51%, and kitchenware by 40%. In general, percentage spending increases were lower for food items. This was unsurprising as these necessities already took up a large part of participants’ budget before the programme.

This means the spending increase of 15% at the retailer is likely an underestimate of how much the programme increased participant spending overall, because food items make up over 80% of expenditure at the retailer and are therefore over-represented.

Another reason the 15% increase is probably an underestimate is because we can only see each individual’s shopping, and not the rest of their household. But some participants were probably shopping on behalf of their family before the programme, and during the programme someone else took over shopping responsibilities, using income from the Basic Education Employment Initiative.

Income effects

What can we then say about who receives income from this increased expenditure? This part of the paper is exploratory and speculative, because we cannot directly see how spending from the programme flows through the economy, and how firms respond to this increased revenue.

Instead, we have to use back-of-the-envelope calculations to scale up the expenditure, use input-output data from Stats SA to guide assumptions about which industries produce which kinds of goods, and use other firm data to see how firms’ wage bills and profits usually respond to sales increases. In our paper we explain the methods, assumptions and limitations in detail.

With these caveats in mind, the implied direct effect of the programme on the retailer’s sales is about R8-million per month. Directly, this likely increased the wage bill for workers at the retailer by about R1-million per month.

Indirectly, the increase in the retailer’s sales would have increased demand from their suppliers, and in turn their suppliers’ suppliers, which we estimate increased employment and wages outside the retailer by another R1.7-million per month.

What about participant spending outside the retail firm? By scaling up the retailer-specific results, we estimate that overall the programme generates about R38-million per month in additional value added in the national economy, which translates to R19-million in additional employment and wages per month, R13-million of which went towards local community employment.

What next

The main beneficiaries of the Basic Education Employment Initiative programmes are the young people who are directly employed by it, and the students in the schools. But the money does not get “thrown away” – one person’s spending is another person’s income.

And the participants do buy goods which are produced locally, using local workers. When evaluating the costs and benefits of the programme, and similar programmes such as social grants, these “extra” economic benefits need to be part of the calculation.

Written by Joshua Budlender, PhD candidate in Economics, UMass Amherst and Ihsaan Bassier, Researcher in Economics, London School of Economics and Political Science

This article is republished from The Conversation under a Creative Commons license. Read the original article.