Employing young people in South Africa

What role has the Employment Tax Incentive played?

South Africa has some of the highest rates of joblessness in the world and youth unemployment is particularly severe. Roughly two in five young South Africans who want work cannot find it, according to labour force surveys.

The South African government has introduced numerous policies to help young people gain employment and is currently deciding whether or not to extend a tax incentive designed to encourage firms to hire young workers. At stake is whether the policy, called the Employment Tax Incentive (ETI), has successfully boosted youth employment.

UNU-WIDER, the South African National Treasury, and the South African Revenue Service have worked together to make individual and corporate tax forms available to academic researchers. The data-sharing represents a true commitment to a transparent evaluation process according to rigorous academic standards.

Watch this video to learn more about Itumeleng Makgetla's findings on South Africa's Employment Tax Incentive.

Evaluation is an important means of helping policymakers to take the next steps. In the case of the ETI, evidence of effectiveness among smaller firms can be used to assess the incentive’s price tag and decide whether to continue it after 2017, when it is slated to end.

How does the Employment Tax Incentive function?

The policy essentially lowers the relative cost of low-skilled youth labour, incentivizing firms to hire young workers.

The incentive works by reducing the amount of payroll tax that an employer paid per eligible employee. Eligible workers are aged between 18 and 29, earn between ZAR2,000 to ZAR6,000 per month, work in the private sector and were hired after 1 October 2013. 

The government’s expectations were that the incentive would add 178,000 jobs to the economy.

The policy essentially lowers the relative cost of low-skilled youth labour, incentivizing firms to hire young workers. Over time, lower labour costs should encourage expansion and greater employment overall, though evidence of this latter effect may take longer to materialize than that of youth hiring patterns.

In total, around 30,700 firms claimed the incentive for almost 300,000 individuals over fourteen months, from its introduction in January 2014 to February 2015. The total claimed for eligible workers was about R1.5 billion (excluding claims that failed eligibility requirements).

Does the policy deliver expected results?

My study draws on the data made available through the project to explore three questions. First, did firms using the incentive hire more young people than other firms? Second, did employment expand overall in firms using the incentive? And third, did firms fire workers for whom they could not claim the incentive to make way for new workers?

The findings suggest that overall, despite strong take up, the incentive cannot be credited with job growth for young people.

The study finds that firms that used the incentive differed substantively from firms that did not. Firms using the incentive were larger, had more youthful workforces and hired and fired workers at a higher rate than other firms. When we control for those baseline differences using several periods of data before the incentive’s introduction, no significant difference is found in overall employment or employment of young workers in firms that used the incentive compared to similar firms. 

The findings suggest that overall, despite strong take up, the incentive cannot be credited with job growth for young people.

Where can impacts of the policy be seen?

Photo: Beyond access - Youth technology training in South Africa.

Smaller firms did see an increase in the number of young workers that they employed relative to their past performance and comparable firms. This group comprises firms with fewer than 50 employees and roughly 44 per cent of firms of this size in the panel used the incentive. 

Why some firms did not use the incentive is a critical question at the heart of whether the incentive works and how. Additional data are needed to answer this question. Could one make the case that incentive take up was ‘as if random’ or is use of the incentive correlated with something like greater sophistication in a firm’s operations or better management?

By contrast, larger firms with over 50 employees underperformed in job creation relative to their past performance and other large firms. Because these firms employed more employees in the country overall — and have more people ETI — the incentive does not have a net positive effect on youth employment.

Notably, fears that firms would fire older workers to take advantage of the programme were not realized.

Notably, fears that firms would fire older workers to take advantage of the programme were not realized, as separations of older employees did not go up in firms that used the incentive relative to their past trends and other firms.

The challenge of evaluating the incentive was that it was available to all firms simultaneously. The concern was that firms using the incentive may differ from firms that did not in ways that shaped the outcomes we cared about, such as hiring patterns.

Of course, evaluations can control for baseline differences through various conditioning strategies but two problems remain. One, there are likely unobservable factors at play that one can’t identify — let alone control for — driving the choice of incentive take-up and outcomes of interest, like job creation. Two, the tax forms available to analyse the program provide very little information to control for firm characteristics. Even the sector and geographical data were often missing and frequently contradictory. 

I analysed the data using an interrupted time-series with a comparison group. This approach uses a segmented regression to estimate the pre-intervention trend in the dependent variable, any change in the outcome at the time of the intervention, and the post-intervention trend. An interrupted time series with a comparison group enables one to control for differences from the baseline and trend between the treatment and control groups. I constructed the comparison group using nearest neighbour matching on data from before the incentive’s introduction.

This technique resembles a difference-in-difference design but does not require assumptions about the time trend for treatment and control groups. While rarely feasible due to strict data requirements, my study — with 21 pre-intervention and 13 post-intervention periods — was able to make use of the approach.

For my analysis, I created a balanced panel where each unit was a firm-month. 

What’s next?

Mine is one of two independent evaluations that, alongside the Treasury’s analysis provides, a comprehensive assessment of the ETI’s performance. Further analysis of additional data will enhance our understanding of the policy. More periods of implementation have past since my research and should be studied. 

Notably, this panel used in the study excludes 24,503 new firms that emerged after the incentive’s introduction. Of these the 2,416 new firms that used the incentive have more youthful workforces than other new firms. Further work could explore the ETI’s role in causing new business to emerge.

It is also important to consider other methodological approaches. For example, follow up studies of individuals hired on the incentive could illuminate its long-term effects. Qualitative work could also explore how firms used the incentive and why some firms chose not to use it. 

The views expressed in this piece are those of the author(s), and do not necessarily reflect the views of the Institute or the United Nations University, nor the programme/project donors.

WDR 2017 does not disappoint
WDR 2017 does not disappoint
Information asymmetries in extractive industries
In the second part of this blog, Alan R. Roe discusses what is known about the informational failures that pose challenges for governments in projecting revenues from extractive industries. Read the first part here ...
Information asymmetries in extractive industries