VoxEU Column Labour Markets

Temporary contracts are bad for your cognitive health: Evidence from PIAAC

The negative consequences of dual labour markets have been extensively documented, but so far little attention has been paid to their effects on workers’ on-the-job training and cognitive skills. This column discusses evidence from PIAAC – an exam for adults designed by the OECD in 2013. Temporary contracts are associated with a reduction of 8–16 percentage points in the probability of receiving on-the-job training, and this training gap can explain up to half of the gap in numeracy scores between permanent and temporary workers.

Starting with the seminal work by Saint-Paul (1996), there has been a large literature documenting the negative consequences of dual labour markets in several EU countries.1 Among them, Spain is often cited as the most extreme example, since its labour market is characterised by a large gap between the firing costs of workers with permanent and temporary contracts, and by lax regulation of the use of temporary contracts. Yet, so far not much attention has been paid to the effects of dual labour markets on workers’ on-the-job training (OJT) and the subsequent effect of the latter on cognitive skills.2 A new element to add to the ample evidence of the negative effects of duality on other dimensions of worker’s performance is provided by the Programme for the International Assessment of Adult Competencies (PIAAC), the exam for adults designed by the OECD in 2013, in the spirit of the PISA exams but for the working-age population.

Employment protection legislation and on-the-job training

In a recent paper (Cabrales et al. 2014), we present evidence suggesting that (everything else equal) workers under temporary contracts receive less employer-sponsored training than workers under permanent contracts and that, through this channel, labour market dualism also reduces the cognitive abilities of the former relative to the latter. One plausible mechanism leading to the gap in on-the-job training relies on the large turnover rate among temporary workers, which is induced by the much less stringent employment protection legislation they enjoy relative to permanent workers. Given this differential, whenever collective bargaining prevents neutralising severance pay (i.e. a transfer from employers to workers) through enough wage flexibility, firms will prefer to use temporary contracts in sequence rather than converting them into permanent contracts. As a result, the expected job duration of temporary workers becomes too short, making firms more reluctant to invest in their training. By contrast, the much stronger employment protection enjoyed by permanent workers increases their expected job duration, making firms more eager to invest in them. Thus, in countries with large employment protection gaps and wage rigidity, temporary contracts become dead-end jobs rather than stepping stones, as is often the case in those other countries where employment protection gaps are lower.

Spain is an interesting case study because, since the early 1990s, it has had a large share of temporary work among employees – around 33%. Even after the massive destruction of temporary jobs (almost 2 million) and the mild reduction in the employment protection gap after the labour market reform in 2012, it has fallen only to 24% – still one of the highest rates in the OECD.

In our paper, we provide evidence that differences in employment protection legislation between permanent and temporary contracts is related to differences in on-the-job training within the firm. We also analyse how training can be connected with the cognitive skills (numeracy and literacy) of these workers. Since education is one of the key determinants of the growth rate of total factor productivity, our results suggest a connection between a poor performance in terms of TFP growth and the excessive use of temporary contracts.

Evidence from PIAAC tests

Individuals participating in PIAAC are aged between 15 and 65 years. We look at those who are salaried employees and answered the full questionnaire. The main treatment variable is whether the worker has a temporary contract (more specifically, a fixed-term contract, temporary work with an employment agency, or any kind of training contract) or not. The main outcome variable is the availability of on-the-job training. This binary variable, denoted by DOJT, takes the value 1 if the worker claims to have attended a training session organised or given by their supervisors or co-workers in the past 12 months. According to the PIAAC guidelines, these activities are characterised as planned periods of training, instruction, or practical experience designed to help the respondent do his/her job better or get familiarised with his/her new tasks.

Controlling for workers’ motivation

We control for those observable factors in the sample that could potentially affect both the above-mentioned outcome and the treatment variable. Controls can be divided into two types. First, we use basic individual characteristics, such as age, gender, education level, marital status, whether there are children in the household, immigrant status, and the level of education of their parents. Second, we control for occupation (two digits) and industry (one digit) dummy variables. The PIAAC data also include information that is usually absent in this type of analysis and whose absence can seriously bias the results – the worker’s motivation. To capture this information, we exploit a question on the reasons for learning new skills. In particular, we use a dummy variable, Motivation, which takes the value 1 when the individual reports being identified ‘largely’ or ‘to a very great extent’ with learning new skills to improve her performance in the workplace. Notice that not controlling for individual motivation may provide a non-causal correlation between temporary contracts and skills acquisition, as less motivated individuals may also be more likely to have a temporary contract.

Empirical findings

  • Depending on the specification, we find that a temporary contract is associated with a reduction of between 8 and 16 percentage points in the probability of receiving on-the-job training (the unconditional probability of receiving training in our sample is 44%).3

These results are consistent with the view that entrepreneurs do not find it profitable to train workers who will be fired with high probability at the end of their contracts, as the still-high differential in severance payments between permanent and temporary contracts is for them an insurmountable obstacle (the annual temporary-to-permanent conversion rate in Spain barely exceeds 7% at present, and never went above 15% in the past).

We then turn to the association between cognitive skills and having received training in the company, again for various specifications. In the case of numeracy scores, the relationship is positive and quite robust.4

  • The difference in PIAAC test results between workers with permanent and temporary contracts was 14 points (261 and 247, respectively), and our results indicate that between 4 and 7 points of this gap is explained directly by the lower occupational training received by temporary workers.

These results remain fairly similar when we carry out several robustness checks. For example, this is the case when we use propensity score matching techniques to minimise the differences between the treatment and control groups, besides having a temporary contract, or when we use a restricted control group corresponding to those employees who had the possibility to attend employer-sponsored training but ended up not doing so for exceptional and unexpected reasons.

Finally, we provide evidence for other EU countries.5

  • In relatively dual labour markets (Italy, France) workers receive less training than in countries with more unified labour markets (Denmark, UK).

Furthermore, differences in cognitive abilities between permanent and temporary workers are strong in dual countries while they are insignificant in less dual countries.

Concluding remarks

The evidence presented above is not, strictly speaking, a ‘smoking gun’ of a harmful effect of temporary contracts. Due to the nature of our data (a cross section), we cannot credibly establish unambiguous causality for the statistical relation that we find. However, the evidence is quite suggestive. We hope that future analysis will allow for a clearer causal inference, but given all the evidence (not just ours) accumulated so far on the detrimental effects of dualism, we should keep the suspect under house arrest and heavy police surveillance. If it really is a social danger, we should at least try not to do more damage.


Arulampalam, W, A Booth, and M Bryan (2004), “Training in Europe”, Journal of the European Economic Association 2: 346–360.

Blanchard, O J and A Landier (2002), “The Perverse Effects of Partial Labor Market Reform: Fixed Duration Contracts in France”, Economic Journal 112: 214–244.

Bentolila, S, P Cahuc, J Dolado, and T Le Barbanchon (2012), “Two-Tier Labour Markets in the Great Recession: France versus Spain”, Economic Journal 122: F155–F187.

Cabrales, A, J Dolado, and R Mora (2014), “Dual Labour Markets and (Lack of) On-the-Job Training: PIAAC Evidence from Spain and Other EU Countries”, CEPR Discussion Paper 10246, November.

OECD (2014), Employment Outlook, Chapter 4: “Non Regular Employment, Job Security and the Labour Market Divide”.

Saint-Paul, G (1996), Dual labor markets: A macroeconomic perspective, MIT Press.


[1] See, e.g., Blanchard and Landier (2002), Cahuc and Postel-Vinay (2002), and Bentolila et al. (2012).

[2] See, e.g., Arulampalam et al. (2004), and more recently Chapter 4 in OECD (2014).

[3] For details, see Table 2 in Cabrales et al. (2014).

[4] For details, see Table 6 in Cabrales et al. (2014).

[5] For details, see Table 13 in Cabrales et al. (2014).

3,045 Reads