When goals get in the way: Imperfect information in job search in South Africa


Published 10.04.20

Reducing transport costs and increasing exposure to the labour market leads to job seekers improving targeting in the job search

Youth unemployment remains high throughout the developing world. Persistent youth unemployment can pose several short- and long-term economic and social challenges. It can compromise lifetime earnings, reduce incentives to invest in education, and potentially lead to increased criminal activity. Given rising demographic pressure throughout the developing world, understanding why youth unemployment is so high and designing effective labour market policies that remove constraints to employment, remain pressing policy concerns.  

In South Africa, a staggering 54% of the population aged 15-24 report being unemployed (ILOSTAT 2017). Around 82% of the unemployed claim to have searched for jobs for over a year, and approximately 30% for over five years (Labour Market Dynamics in South Africa 2015). At the same time, firms complain about being unable to find workers and that when they do, workers often quit after a short period of time on the job (Banerjee et al. 2008). 

One plausible explanation is the existence of a skills gap: available jobs require more skills than most workers can offer. However, there is also evidence that even skilled workers have a hard time finding jobs (Pritchett 2001, Mckenzie 2017). It is therefore unlikely that this skill mismatch in the labour market explains the full story. 

Distance to jobs and employment prospects

An alternative explanation for this puzzle is that there is a mismatch between the kind of jobs that workers are looking for and those that are available. Workers might quit existing jobs because they over-estimate their chances of getting into more desirable occupations. Such inaccurate beliefs can persist when poor urban infrastructure, high transport costs, and residential segregation, combine to place low-income groups away from the city centre where most high-paying jobs are located. While they aspire to one day get a high-paying job in finance or government in the city centre, transport costs are high so they search too little while hanging on to their goals. In the meantime, they turn down jobs that are at their skill level and closer to home. 

Biased beliefs

In order to assess the presence of biased beliefs, we surveyed a sample of 1,082 job seekers in a large township in the outskirts of Johannesburg (Banerjee and Sequeira 2020). Our sample, aged between 18 and 32, had completed secondary (83%) or tertiary education and were actively searching for jobs. The township is 20 kilometres away from the city centre, where most high-paying jobs are located. While in our setting this spatial divide is a clear legacy of Apartheid, there is growing evidence that poor urban planning and limited transport infrastructure are leading to similar spatial mismatches across large cities in the developing world  (Abebe et al. 2018, Franklin 2017). 

Our findings show a striking optimistic bias: job seekers expect to earn 1.7 times more than the actual median salary earned by someone in their region with a similar skill level and age profile. We also find that this bias is driven primarily by an overestimation of the probability of getting into high-paying occupations, as opposed to job seekers not knowing average wage levels for a given job category. We find that 75% of job seekers reported looking for a professional job in government or finance, even though the actual probability of an individual with a similar age and skill profile getting such a job is only 11%. 

Job seekers are also over-confident. They believe that their own starting salary is likely to be, on average, twice as large as the salary of job seekers in their township who have a similar skill profile. Consistent with targeting high-paying jobs and being over-confident, job seekers overestimate the probability of getting any type of job. 60% of them believe that it is likely or very likely that they will find a job within two months, despite having been searching for an average of 15 months.

But how can these biased beliefs persist? 

The transport costs to the city centre are paramount in explaining these biased beliefs. 91% of job seekers in our sample believe that professional jobs that match their skills and interest are located in the city centre, and that the two most likely strategies to get these jobs are to get referrals and drop off CVs in person at firms. Since they do not have the social connections, and cannot afford to go often to the city centre, they do not search enough for the jobs that they want. This means they learn little over time about the probability of getting into a high-paying job.

Increasing access to the labour market

To assess the impact of changing expectations, we designed an experiment varying exposure to a wider labour market (Banerjee and Sequeira 2020). For a randomly selected treatment group, we provided job search subsidies through smartcards for job seekers to search for jobs in the city centre. By contrast, the control group received no financial support for the job search but they received a smartcard that enabled us to observe their travel patterns. 

Providing job search subsidies led beneficiaries to search more intensively relative to the control group, by traveling more frequently and covering a wider geographic area in the job search. This increased exposure to the labour market changed job seekers' expectations about the net returns to searching for, and accepting, jobs in the city centre. 12 months after the intervention, job seekers that received the subsidy adjusted downwards their salary expectations (5%) and reservation wage, i.e. the lowest wage they would be willing to accept for a particular type of job (8%). More importantly, they revised downward the probability of getting a high-paying job as shown in Figure 1. 

Figure 1 Cumulative perceived probability of getting a job in finance or government

As a result, they were 11 percentage points more likely to accept a job in their own township, despite having reported a preference for a job in the city centre initially. Job seekers also reported a significant increase (23%) in how much they valued an existing job. 

It is also possible that the job search subsidies enabled commuters to learn about the cost of commuting to the city centre on a regular basis. In fact, accounting for transport costs rendered jobs in the township at least as attractive as jobs in the city centre as shown in Figure 2, particularly for those with only secondary schooling.  

Figure 2 Actual monthly earnings, net of commuting costs

Exposure to the labour market may make job seekers more realistic and reconcile them to a job in the township. Confirming that the main mechanism is learning through the job search, those who search the most also adjust their beliefs the most and are more likely to accept a job in the township. 

Policy implications: The importance of addressing imperfect information 

Our findings highlight the importance of imperfect information in shaping both the job search and the employment outcomes of young job seekers. Job seekers living in the periphery of large labour markets can significantly overestimate their employment prospects in the city centre due to high transport costs that limit search activity. Increasing exposure to the wider labour market partially de-biases job seekers as they adjust their expectations on the likelihood of getting a professional job. 

These findings suggest that policies to support job seekers in low-income settings categorised by imperfect information, might need to include de-biasing interventions. Alongside investments in increasing the accessibility of jobs, this would optimally guide job seekers in the job search and improve targeting. 


Abebe, G, S Caria, M Fachamps, P Falco, S Franklin and S Quinn (2018), “Anonymity or distance? Experimental evidence on obstacles to youth employment”, Mimeo, Stanford University.

Banerjee, A, S Galiani, J Levinsohn, Z McLaren and I Woolard (2008), “Why has unemployment risen in the new South Africa,” Economics of Transition, 16(4): 715–740.

Banerjee, A and S Sequeira  (2020), “ Spatial mismatches and imperfect information in the job search”, manuscript.

Franklin, S (2017), “Location, search costs and youth unemployment: Experimental evidence from transport subsidies”, The Economic Journal 128(614): 2353- 2379.

ILOSTAT (2017), Unemployment Statistics, International Labor Organization.

Labour Market Dynamics in South Africa (2015), LMDS, South Africa

Mckenzie, D (2017), “How effective are active labour market policies in developing countries?”, World Bank Policy Research, Working Paper 8011.

Pritchett, L (2001), “Where has all the education gone?”, The World Bank Economic Review, 15: 367-391