Wage subsidies and youth employment in South Africa: Evidence from a randomised control trial

• Youth unemployment in South Africa is high, differs substantially by race group and is increasing. In 2012, close to two-thirds of young Africans were broadly unemployed. Over the four years prior to this the unemployment rate had increased by almost ten percentage points. • A wage subsidy is one type of intervention which aims to reduce youth unemployment by providing a subsidy to firms which covers part of the cost of employing young people. The outline of a youth employment incentive was provided by the National Treasury in 2011. • Evidence from other countries suggests that the success of a wage subsidy can be context specific and depends on the nature of the intervention and the structure of the labour market amongst other things. Thus, in order to understand how a wage subsidy may affect youth unemployment it is useful to know how South African young people and firms may react to a wage subsidy. A randomised control trial (RCT) is one way to investigate this. • In an RCT the participants in the study are randomly divided into two groups – one which received the intervention, in this case a voucher for a wage subsidy which a firm who employs the individual could claim for six months (called the treatment group), and the second group which does not receive anything (called the control group). Since allocation to the groups is random and both groups share similar characteristics, any observed changes on average should be the result of the wage subsidy voucher. We can thus attach a causal interpretation to our results. • The key finding of the paper is that those who were allocated a wage subsidy voucher were more likely to be in wage employment both one year and two years after allocation. The impact of the voucher thus persisted even after it was no longer valid. The magnitude of these effects was relatively large – those in the voucher group were 7.4 percentage points (approximately 25 percent) more likely to be in wage employment one year after allocation and of similar magnitude two years later. This impact was not driven by changes in the sample composition. • This suggests that those young people who entered jobs earlier than they would have because of the voucher were more likely to stay in jobs. This confirms the important dynamic impacts of youth employment. It also suggests that government interventions which successfully create youth employment are important and can virtuous longer-term effects. • Relatively few firms actually claimed the voucher. Interviews with firms and young people suggest that this was for a number of reasons: the young people did not even get a chance to show the voucher to someone who makes hiring decision; the administrative burden associated with claiming the money, although not onerous, could not be overcome (for example, larger firms did not have a process for accepting subsidy money, human resource functions were centralised and HR had little incentive to engage in the process of claiming the voucher); or managers or firm owners questioned the legitimacy of the voucher. This suggests that a national wage subsidy policy would need to be widely advertised and information and support provided to firms who would like to claim the subsidy. • However, the impact of the voucher among those individuals who were employed in firms who claimed or enquired about the subsidy was much larger than the broader estimated effect. In this paper we are unable to ascertain whether these jobs were new or not. • Even after controlling for firm take up and enquiry there is still a difference in the probability of wage employment between the group with a voucher and the group without. This indicates that part of the impact of the voucher is through supply side responses of those allocated the voucher. • The results indicate that the observed impact of the voucher is not driven by changes in search, increases in search intensity or movement either to look for jobs or to take up employment. • Rather it seems that part of the impact may be driven by people turning down job offers. Those in the control group, especially those in households with other employed members, were more likely to turn down job offers than those in the treatment group. This suggests that there is some queuing in the South African youth labour market as young people who can wait for better paid jobs do. • We can only speculate about why those in the treatment group did not engage in this behaviour. One explanation is that the voucher changed their perceptions of potential success in the job market and thus they were more willing, or able, to go to these jobs, since they thought the voucher advantaged them or they were able to borrow money from their households to travel and incur the initial costs associated with accepting a job. It may also be that households which contained voucher holders were more likely to encourage the holder to take up the job since it was perceived as part of a special programme, or it may be that more information about jobs was passed onto the households of voucher holders with employees since they were linked into firms and people in their network may have known about the voucher. • These results confirm that the structure of the household is important for success in the labour market. Research on South African labour markets shows that networks are the main channel through which information about jobs is transmitted. Households with working members are thus advantaged since members receive more information about jobs. These types of households can also provide intra-household cash transfers to help pay for transport costs or other costs associated with taking up a job. • However, there can also be relatively negative implications for young people in households with other earners – they can afford to turn down jobs as they wait for potentially better paying or better matched jobs. The consequences of this may not necessarily be negative if these types of jobs eventually arrive but if they do not then these young people have sacrificed both earnings and work experience as they wait.

C93, D22, H25, J22, J23, J58

wage subsidies, youth employment, South Africa, impact evaluation, randomised control trials, employment incentives