Unhealthy temporary labour contracts

UNIVERSITEIT LEIDEN

Unhealthy temporary

labour contracts

Examining the effect of the reforms of the Dutch disability insurance system on temporary

contracts for (partially) disabled and unhealthy workers

J.P.F. Mertens – s1535552

Supervisor: Prof.dr. P.W.C. Koning

10-01-2019

TABLE OF CONTENTS ACKNOWLEDGEMENT ...2 ABSTRACT ...3 1 INTRODUCTION ...5 1.1 BACKGROUND ...5 1.2 RESEARCH GOAL ...6 1.3 RESEARCH QUESTION ...6

1.4 ADDED VALUE AND RELEVANCE OF THE RESEARCH ...6

1.5 THESIS OUTLINE ...7

2 THEORETICAL FRAMEWORK ...8

2.1 DISABILITY INSURANCE SYSTEMS ...8

2.2 THE DUTCH DISABILITY INSURANCE SYSTEM ...9

2.3 TEMPORARY CONTRACTS AND THE LABOUR MARKET ... 12

2.4 HYPOTHESES ... 15 3 METHODOLOGY ... 17 3.1 RESEARCH DESIGN ... 17 3.2 DATA COLLECTION... 18 3.3 OPERATIONALIZATION ... 18 3.4 VALIDITY ... 20 4 ANALYSIS ... 22

4.1 LABOUR MARKET PARTICIPATION AND LABOUR CONTRACTS ... 22

4.2 THE EFFECT OF THE REFORMS ... 30

5 CONCLUSION ... 34

5.1 DISCUSSION OF RESULTS ... 34

5.2 LIMITATIONS ... 35

5.3 RECOMMENDATIONS ... 36

BIBLIOGRAPHY ... 37

ACKNOWLEDGEMENT

The process of writing my master thesis started more than a year ago. Before anyone can start writing a master thesis, it is important to have a topic or at least a hunch about a topic. Because my supervisor, Mr. Koning, was specialised in the labour market I started looking for topics in that direction. My first suggestions were about how small employers had to deal with sickness leave of employees by for example insuring themselves. During the year my topic shifted towards the topic of this thesis, the use of temporary contracts for workers with inferior health or disabilities. I think this is an important and actual topic in Dutch politics nowadays, as is illustrated by the recent legislation proposal of the minister of social affairs and employment, concerning the balance between temporary and permanent contracts (Wet arbeidsmarkt in balans). Furthermore, I think the topic of this thesis is a great example of the subject matter of the study of Public administration.

While I originally planned to finish my master before the summer of 2018, I started another master program in February 2018, which led to the decision to postpone my thesis for half a year. I would like to thank Mr. Koning for being my supervisor. Without his help and suggestions about the topic, statistics and details it would have been very hard to finish this thesis. Furthermore, Mr. Koning agreed to continue being my supervisor after I postponed my thesis for half a year, for which I am very grateful.

Joost Mertens

ABSTRACT

For a long time the Netherlands had a reputation for having a very high number of disability benefit recipients. Until 1994, the public expenditure on disability benefits was almost twice as high as in other welfare states in Europe and while the public expenditure on disability benefits decreased, it was still considered as a major economic problem in the Netherlands around the year 2000. That is why several reforms of legislation were introduced in the years 2002, 2004 and 2006. While these reforms had lowered the volume of the disability insurance program, they did create more responsibilities for employers. Employers could risk higher costs if employees became sick for example. That is why it is possible that employers are more hesitant to hire workers with a high risk of becoming sick. Alternatively, employers can hire these high-risk workers on a temporary contract, rather than a permanent contract. To examine this phenomenon, the following research question was stated for this study:

To what extent has the use of temporary contracts for disabled workers or workers with inferior health increased in the Netherlands and if so, can this change in the relative importance in contract types for healthy and unhealthy individuals be explained by disability insurance reforms?

In the literature it became clear that the reforms of the Dutch disability insurance system created or increased several responsibilities for employers. The first reform, the Gatekeeper protocol obliged employers to provide several medical and reintegration documents when an employee got sick. In 2004 another reform was introduced in the disability insurance system: the extension of sickness leave from one to two years. This means that employers are responsible for the payment of sickness leave for two years, after which the disabled/sick person moves to disability benefits. The last reform, the WIA reform, reduced the amount of sick pay in the second year of leave to 70% of the previously earned wage (The WIA reform included some other aspects which are less relevant in this context). This relieved the employers to some extent, because the average was 90% of the wage. Nevertheless, because of these reforms employers have more costs and responsibilities when an employee gets sick.

Based on the literature and the research question, three hypotheses are constructed to guide the analysis. To test them, a longitudinal dataset of Dutch citizens which translates as the “labour supply panel” is analysed over the period 2000-2014.

Hypothesis 1: From the years 2000 until 2014, the relative chance of people with a disability or inferior health to participate on the labour market has decreased. In the analysis it was found that respondents with

inferior health or a disability have a significantly lower chance to participate on the labour market than healthy respondents. Having an inferior health decreases the chance of having a job with 16.7%. It was however not possible to determine if this chance changed throughout the years. Differences in the labour participation over the years between healthy and unhealthy respondents provided some small insights, but those insights rather invalidated than confirmed the hypothesis. It can therefore not be said that from the years 2000 until 2014, the relative chance of people with a disability or inferior health to participate on the labour market has decreased.

Hypothesis 2: From 2000 until 2014, the chance for employees with a disability or inferior health to have a temporary contract has increased when compared with the chance for non-disabled and healthy employees to have a temporary contract. Based on the analysis it can be stated that respondents with inferior health have a

5.1 percentage point more chance of having a temporary contract than healthy respondents on average over the period 2000 until 2014. The results also indicated that the chance for employees with a disability or inferior health to have a temporary contract decreases over time. Although the effect was small, it does contradict the hypothesis. It can therefore not be stated that the chance for employees with a disability or inferior health to have a temporary contract has increased when compared with the chance for and healthy employees.

Hypothesis 3: Due to reforms in the Dutch disability insurance system in 2002, 2004 and 2006, the chance for Dutch employees with a disability to have a temporary contract has increased. Because the results of the

regression analysis for specific years were not significant, it was hard to associate changes in the use of temporary contracts for unhealthy respondents with the reforms of the disability insurance system. The only results that could be linked to the reforms of the disability insurance system were the sudden increases in temporary employment in the years 2004 and 2006. While it is possible that these increases were caused by the reforms, they also diminished after 2008, so they were present in the direct years between and after the reforms. It is therefore not possible to confirm the hypothesis with the available information.

The findings based on these three hypothesis enable to formulate an answer to the research question. The use of temporary contracts for disabled workers or workers with inferior health has increased, but only in the period between 2000 and 2008. It is possible that this increase was caused by reforms of the Dutch disability insurance system in 2002, 2004 and 2006, although this cannot be said with certainty. Between 2008 and 2014 however the use of temporary contracts for workers with inferior health decreases again.

It is possible that the financial crisis of 2008 influenced the results of this study. For future research it is therefore advisable to recreate this study with other datasets. For policymaking it is advisable to minimize the differences that now exist between temporary and permanent contracts in for example social security. This can be done by stronger regulation of temporary contracts or by transferring some of the responsibilities that now exist for employers to employees with a permanent contract. If employees have to bear more responsibilities, vulnerable groups have to be protected however.

1 INTRODUCTION

1.1

BACKGROUND

In European context, the Netherlands had for a long time a reputation as a country with a high number of people that could not work because of a disability (Versantvoort, van Echtelt, 2012). From 1980 until 1999 the number of disabled people that could participate in the workforce was higher than all (comparable) neighbouring countries like Germany, Great-Britain, Denmark, France, Belgium and Sweden (Jehoel-Gijsbers, 2009) and the Dutch disability program was considered as one of the major economic problems in the Netherlands around the year 2000 (Koning & Lindeboom, 2015). That is why several initiatives were introduced in the Netherlands in the late 1990s and the years 2002, 2004 and 2006. The first reform in 2002, The Gatekeeper Protocol (Wet

verbetering poortwachter), was aimed at improving the responsibilities of the employer and workers as well. The

social benefit administration would only act as a gatekeeper and would be less involved in the reintegration process of sick workers (Koning & Lindeboom, 2015). In 2004, other changes in the disability system were introduced which extended the period that an employee would receive sickness benefits from his employer (Wet

verlenging loondoorbetalingsverplichting bij ziekte). In the last major change in legislation in 2006, the old

Invalidity Insurance Act (Wet op de arbeidsongeschiktheidsverzekering or WAO) was replaced by a new law which translates to the Work and Income (Employment Capacity) Act (Wet werk en inkomen or WIA). This new act included several smaller aspects which would improve work incentives and further limit the eligibility criteria for disability benefits. These new legislation acts will be further explained in the theoretical framework.

While the volume of disability insurance (as well as amount of sick leave) declined after the introduction of the reforms, it is unclear to what situation those people that initially received disability insurance have moved to. It could be that more of those people ended up in unemployment insurance, or that they are now in employment. It is interesting to examine if this has led to more (partially) disabled people participating in the labour market. Furthermore, it is possible that the reform created differences in the type of contracts that (partially) disabled receive. Employment contracts can be roughly divided into two categories: permanent or temporary contracts (also called flexible contracts). Generally, for an employer a permanent contract is financially more interesting than a temporary contract, but permanent contracts can also include more risks than temporary contracts. When the employee gets sick for example, the employer is responsible for paying sickness benefits and with a permanent contract, the employer has to pay sickness benefits for a longer period of time then compared with an employee with a temporary contract. People with health impairments such as disabled people have a higher chance of receiving a temporary contract, because they have a higher risk of being unable to work. Therefore it is interesting to examine if the legislation changes in 2002, 2004 and 2006 changed anything in the distribution of temporary and permanent contracts among people with health impairments.

Alongside with the reforms on disability insurance, there are important trends in the Dutch labour market that catch the eye. In particular, in the past ten to fifteen years the share of employees with a temporary contract has dramatically risen. This trend has been aggravated during the financial crisis (Van Echtelt & de Voogd-Hamelink, 2017). While both the impact of disability insurance reforms and the implications of the shift from permanent

Smits (2015), Muffels (2013) and Koning & Lindeboom (2015) – the interrelationship between labour market flexibilization and the employment of disabled workers has not been touched upon so far. If the Dutch disability legislation changed employer incentives and requirements it could very well be that disabled workers, which are likely to be high-risk workers, will receive different kind of labour contracts. The goal of the disability system reforms was to lower the percentage of people that received disability benefits by increasing the responsibilities of employers. Because the employers received more responsibilities, they also have higher costs when the employee does get sick. An employer can limit this risk by either not hiring people with inferior health and health impairments, or by hiring them under a temporary contract instead of a permanent contract. In this research, the chance of receiving a contract as well as the possible shift of temporary and permanent contracts for (partially) disabled people and people with inferior health will be examined.

1.2

RESEARCH GOAL

The goal of the research is to analyse if there is a difference in the type of labour contracts that (partially) disabled persons or persons with inferior health received over the last couple of years. There is already literature available that shows that the reforms of 2002, 2004 and 2006 would cause less contracts for disabled people (Koning & Hullegie, 2015; Koning & Lindeboom, 2015), but there is no empirical evidence for the distribution of permanent and temporary contracts yet. The research goal therefore is to prove or disprove this assumption with empirical evidence. The research will focus on disabled people and their position on the labour market. Because the recent past situation in the Netherlands will be analysed, the research is mainly retrospective. If there are actually differences in the types of contracts that disabled person receive it will also be analysed what caused the differences.

1.3

RESEARCH QUESTION

The following question will be used to guide the research:

To what extent has the use of temporary contracts for disabled workers or workers with inferior health increased in the Netherlands and if so, can this change in the relative importance in contract types for healthy and unhealthy individuals be explained by disability insurance reforms?

1.4

ADDED VALUE AND RELEVANCE OF THE RESEARCH

In the international literature there is little research on the impact of disability insurance reforms and the use of temporary contracts. There are some studies however that look at the Americans with Disabilities Act (ADA) and these show that ADA has led to less employment of disabled workers (Acemoglu & Angrist, 2001; DeLeire, 2000). The Disability Discrimination Act (DDA) in the United Kingdom has also been studied (Bell and Heitmueller, 2009), but all those studies did not look into changes in type of contracts.

The Netherlands Institute for Social Research (SCP) periodically reports on trends in sick leave, disability insurance and the labour participation in the Netherlands (Beter aan het werk, Beperkt aan het werk, Belemmerd aan het werk). While these reports describe the participation of disabled on the labour market, they do not clarify

what kind of contracts disabled have and therefore the research of this thesis can provide added value. There is also the study of Koning and Lindeboom (2015) that focusses on the effects of the Dutch disability insurance system. Their study already suggested that workers with inferior health are more likely to receive temporary contracts rather than permanent contracts because of the reforms of 2002, 2004 and 2006. This conclusion can be further supported by empirical evidence and that is why the findings in this thesis are of academic relevance and can add value to the existing literature.

This thesis can also provide insights that can be of societal relevance. The information that is brought together in this research provides insights in how the labour market for people with a disability operates. Furthermore, the results in this thesis can be used to evaluate specific aspects of the disability reforms of 2002, 2004 and 2006. The labour market implications for people with a disability or with inferior health are most relevant, especially implications regarding type of labour contracts. For future policymaking this can be beneficial, so there is also a societal relevance of this research.

1.5

THESIS OUTLINE

The first chapter has explained the background and the research goal of this thesis. In chapter two, the theoretical framework will be clarified. The reforms of the Dutch disability insurance system will be further explained, as well as the basic functioning of the Dutch labour market, especially with regards to the topics of temporary and permanent contracts. Based on this information, hypotheses and assumptions are defined. Having defined the theoretical framework, the third chapter will explain how the hypotheses will be empirically tested. In chapter four, the actual analysis is described, as well as the results. The last chapter will conclude this thesis and limitations as well as recommendations are discussed.

2 THEORETICAL FRAMEWORK

2.1

DISABILITY INSURANCE SYSTEMS

Disability insurance is inevitably linked to the labour market. When a worker is unable to perform (a part of) her or his work, disability insurance protects the worker from a financial loss due to the fact that the worker is unable to earn wages. Disability insurance, which also includes sick leave, is often a form of public spending and is therefore for a large part funded by national governments. Employers also contribute for a large part to the funding of disability insurance, which can be both in a contribution to the general disability scheme, as in a contribution to the disability benefits of their own employees. Especially the contribution to the disability benefits of their own employees should stimulate employers to provide a safe and healthy working environment, because employers have higher disability insurance costs when their employees have an accident or become ill. For an employer this means however that there is also an incentive to hire only employees with a low risk of moving into disability insurance. This can influence the decision to hire people with a disability or with health problems. This problem has also been observed by Koning and Lindeboom: “The biggest concern is with the high level of sickness and DI risks that are transferred to employers, which probably has made employers more reluctant to hire workers with discernible health conditions.” (2015, p.153). It is therefore likely that the form of the disability insurance system – especially concerning the responsibilities and the contribution of employers – has an influence on the labour market for people with a disability or inferior health.

The labour market can also be influenced by other aspects of disability insurance. The screening discipline with regard to disabilities or illness can have an impact for example. When the screening is relatively tolerant, workers may feign illness or a disability, so that disability insurance could be an attractive alternative (especially for older workers wanting an early retirement). The level of the disability benefits can also influence labour participation decisions, because higher benefits make disability insurance a more attractive option (Gruber, 2000).

When people with a disability or inferior health do want to participate in the labour force, it might not always be possible to apply for the same type of jobs as people that do not have any health issues. In this case, employer behaviour might constrain the possibilities, either because of accommodation costs or because the employer is responsible for the costs (disability benefits for the employee) of an employee being unable to work (Schur, 2003). A worker with a disability can therefore be seen as a risk and this might limit the chances for the disabled worker of having a full-time permanent job. If the employer does decide to hire a worker with inferior health or a disability, an employer could initially hire the person on a temporary basis, so the employer is able to evaluate the “risky” worker (Schur, 2003, p.594). These considerations of the employer are dependent on the responsibilities that the employer has and therefore, reforms of the disability system can alter the considerations of the employer.

There have not been many cases where the disability system of a country underwent such major changes in a short period of time as in the Netherlands, with the aim to promote the employment of disabled workers. One case that can be analysed however is the Americans with Disabilities Act (ADA), which was introduced in 1990 in the United States. The goal of ADA was to remove barriers of employment for disabled people by requiring employers to offer reasonable accommodation and by banning discrimination against disabled workers in wage,

hiring or firing (Acemoglu & Angrist, 2001). Several studies however show that ADA has unintendedly led to less employment of disabled workers (Acemoglu & Angrist, 2001; DeLeire, 2000). The main reason for this was that the costs of hiring disabled workers were raised, because employers were now responsible for the suitable accommodation for disabled workers. Because of the increasing costs of hiring disabled workers, employers were more reluctant to hire them, which resulted in lower employment rates for the disabled. Similar legislation was introduced in the UK in 1995. The Disability Discrimination Act (DDA) was intended as anti-discrimination legislation – just as the ADA –, but the DDA was also not very effective. The DDA at best had no impact on the employment rate of disabled, but possibly worsened it (Bell & Heitmueller, 2009).

In both these cases the reforms raised the costs and responsibilities for employers. They increased costs, causing employers to be more hesitant to hire people with a disability. Instead of helping disabled people, the legislation actually makes it more difficult for them to get a job. Furthermore, as was mentioned before, reforms that are aimed at allocating more responsibilities to employers might cause them to hire “risky” workers under a temporary contract, rather than a permanent contract. It is therefore interesting to learn what the approach in the Netherlands is and how the changes in the disability system affect the job opportunities and type of labour contracts for disabled people.

2.2

THE DUTCH DISABILITY INSURANCE SYSTEM

In this sub-chapter, the general aim and design with regard to the Dutch disability insurance system and the Dutch labour market will be explained. First, the situation before the reforms will be described. Thereafter, the actual legislation reforms will be further explained, as well as the possible effects of these changes.

2.2.1. DUTCH DISABILITY INSURANCE SYSTEM BEFORE REFORMS

From the 1970s up until the early 2000s, the volume of disability benefit recipients in the Netherlands kept increasing. While the total workforce was growing as well, the percentage of disability benefit recipients was very high. This was because the disability insurance system was quite generous and became a popular arrangement for early retirement (Van Sonsbeek & Gradus, 2006). This can be seen in figure 1, which describes the Dutch disability system up until the reforms of the early 2000s. Compared to other European countries, the percentage of GDP spend on disability benefits is substantially higher in the Netherlands (figure 2). From 1980 until 1990, the public expenditure on disability benefit was almost twice as high as in some of the other comparable welfare states in Europe.

Figure 1: Disability benefit recipients in the Netherlands as percentage of the total workforce between 1975-2004

Source: Statistics Netherlands

Figure 2: Public expenditure on disability benefits as % of GDP of several European countries between 1980-2004

Source: OECD

The Dutch disability insurance system differs from most systems in other European countries on two important principles. The first major difference is that the Dutch disability program covers work-related injuries, as well as non-work related injuries and disabilities. Disability insurance recipients are able to access into disability scheme without having any working history requirements, in contrary to many other European countries (Koning & Lindeboom, 2015; Van Sonsbeek & Gradus, 2006). The second difference with other countries is that sick workers already receive wage payments from the period they stop working (or work less) until the moment they receive disability benefits. An important aspect of the Dutch disability insurance system is that disability benefits are accessible if earning capacity is reduced by only 15% (although this number was raised to 35% in 2006). These factors all contributed to the high inflow rates in the disability insurance system.

Only between 1994 and 1996, the amount of disability recipients decreased slightly (figure 1), because of a change in the eligibility criteria in 1993 (Van Sonsbeek & Gradus, 2006). Despite a series of policy changes in the late 1990s, the disability rate in the Netherlands around the year 2000 is among the highest in the world (OECD, 2003). It is likely that the effects of these policy changes took some time to have effect and that therefore the rise of the disability inflow rate continued despite the policy changes.

7% 8% 9% 10% 11% 12% 13% 14% 15% 0.0% 0.5% 1.0% 1.5% 2.0% 2.5% 3.0% 3.5% 4.0% 4.5%

Denmark France Germany Netherlands Sweden United Kingdom OECD - Total

2.2.2. REFORMS IN THE LATE 1990s

In the early 1990s there were already policy changes concerning the disability benefits. Especially in 1993 two changes were made, aimed at reducing the inflow into disability insurance, as well as reducing the level of the benefit in general. As was mentioned before, the changes only had a small effect, so more changes were needed to reduce the inflow. That is why in the 1996 and 1998 reforms of the disability insurance system were introduced. The focus of these two reforms was to change the employer incentives, so that employers would minimize or prevent the amount of workers that end up in the disability insurance system (Koning & Lindeboom, 2015). The first change in 1996 was the privatization of the sickness benefits – which are part of the disability program – so that employers had to insure themselves or had to pay if their employee would get sick. Because employers have to pay 70% of earnings during sickness, the employers have an incentive to minimize the amount of sickness absenteeism through preventive and reintegration measures (De Jong & Lindeboom, 2004).

The other reform concerning the disability insurance system was introduced in 1998. From that moment on, the financing system of disability insurance was experience rated. This means that the level of the premiums that the employers had to pay is dependent on the amount of benefits that are awarded. If a company has a higher risk of employees ending up in the disability insurance program, the company has to pay a higher premium.

The effects of both these reforms are not immediately visible in the amount of disability benefit recipients (see figure 1), but it is likely that the impact of these reforms materialized a few years later (Koning & Lindeboom, 2015). What is evident from both these reforms is that they were aimed at changing the employer incentives. The employers bear more (financial) responsibilities when their employees get sick or end up in the disability insurance system. It is likely however that employers would try to circumvent the risks or costs that came with the increased responsibilities. One way to do this is by using (more) temporary contracts, because the employer is then not individually responsible for the costs if the employee with a temporary contract were to move into the disability insurance system. This is particularly true for high-risk workers, such as workers with inferior health or workers who are partially disabled. This is one of the aspects that will be examined in the analysis.

2.2.3. THE GATEKEEPER PROTOCOL AND EXTENSION OF SICK LEAVE

The Gatekeeper protocol was introduced in 2002 and can be seen as a measure to further increase the responsibilities of the employers. According to Koning and Lindeboom (2015, pp.160) the Gatekeeper protocol has been the most effective policy measure to reduce the amount of people receiving disability benefits. The core of the Gatekeeper protocol is that the employer is responsible for various medical and reintegration documents in the first weeks of absence of the employee. After the initial period, the Gatekeeper – the Employee Insurance Agency (UWV) – will check if the necessary steps are taken and the reports are valid, to make sure the employee is eligible for disability benefits. If this is not the case, the employer is obliged to continue to pay sick pay. Therefore, the Gatekeeper protocol is also an incentive for employers to invest in a quick return of the employee. As is the case with the 1998 policy reform, the employer will also be more hesitant to hire high-risk workers and it is likely that high-risk workers are hired under a temporary contract, rather than a permanent contract.

In 2004 another reform was introduced in the disability insurance system: the extension of sickness leave from one to two years. This means that employers are responsible for the payment of sickness leave for two years, after which the disabled/sick person moves to disability benefits. Koning and Lindeboom (2015, pp.161) consider this reform as part of the Gatekeeper protocol and this extension of the sickness leave is another incentive for the employer to increase efforts to let the employee resume work. At the same time, it can also limit employers in the amount of permanent contracts that they award to high-risk employees, or the amount of contracts they award to high-risk employees at all.

2.2.4. THE WIA REFORM

While the policy changes in the late 1990s, 2002 and 2004 were considered to be effective in limiting the amount of inflow into the disability insurance system, policymakers found further possibilities to improve the system. The element which could still be improved was the work resumption of people with temporary of less-severe disabilities. That is why in 2006 the existing law for invalidity (WAO) was replaced by a new law (WIA), which included three major changes. The first change in the WIA was that it made a difference between workers that were fully and permanent disabled and that therefore could not work again, and between workers that were only partially disabled or sick. The system was split into two schemes: one relatively generous scheme (IVA) for fully and permanent disabled and one less generous scheme (WGA) for partially disabled or fully disabled with a chance of recovery. The WGA therefore also included recurring examinations of the disability in the first three years (Sonsbeek & Gradus, 2013). The second change was the raise of the minimum degree of disability needed for disability benefits. Previously this degree was at 15 percent of the previously earned wage, but in the WIA this was raised to 35 percent. The third measure is related to the extension of the period of sick pay from one to two years, which was introduced in 2004. In the WIA, the amount of sick pay in the second year of leave is set at 70% of the previously earned wage. This relieves the employers to some extent, because the average was 90% of the wage (Sonsbeek & Gradus, 2013, pp.836). At the same time it is an incentive for employees to return to work if possible in the second year of absence, because they receive substantially less wage.

2.3

TEMPORARY CONTRACTS AND THE LABOUR MARKET

2.3.1. THE DUTCH LABOUR MARKET

In the past years, the percentage of workers with a temporary or flexible contract has increased dramatically from 15% in 2004 to 22% in 2014 (Chkalova et al., 2015). In addition, the share of firms that have at least one employee with a temporary contract has increased: from around 30% in the late 1990s, to 57% in 2015 (Van Echtelt & de Voogd-Hamelink, 2017, pp.23). Especially smaller organisations increased their use of temporary contracts in the past years. For smaller organisations the use of temporary contracts is a way to reduce risks that are associated with permanent contracts, for example the risk of an absent employee because of long-term illness. Small organisations can insure themselves against the risk of long-term sickness leave and between 70-80% of them have such an insurance. Such an insurance protects the employer and this should enable the company to use permanent contracts instead of temporary contracts. The employer does have to pay a premium however. The amount of temporary contracts also differs between sectors, because the sector is for example more dependent on seasonal workers.

The factors that mainly contributed to the increase in temporary contracts are globalization and the increased openness and use of information (Chkalova et al., 2015, pp.129). For organisations it is important to be able to adapt to changes that are caused by this globalization and information use. Because it is difficult to fire employees in the Netherlands, the need for labour market flexibility is mainly channelled through temporary employment (Cörvers, as cited in Chkalova et al., 2015, pp.129).

2.3.2. THE DUTCH LABOUR MARKET IN EUROPEAN CONTEXT

Flexibility on the labour market is not only caused by the use of temporary contracts. In countries with more stringent employee protection it is more difficult to fire employees, so it is likely that in those countries temporary employment has become more popular in the past years. There are however also countries where the employee protection is not that strict and it is relatively easy to fire employees with a permanent contract. In those countries there can be flexibility on the labour market as well, although the percentage of temporary contracts might be relatively low.

In the Netherlands, employee protection for permanent contracts is quite stringent when compared with other European countries. In 2013, only Belgium and Germany had a more stringent employee protection legislation (EPL), although in Belgium the level of EPL is mainly caused by high collective dismissal protection. The United Kingdom and Ireland are the countries with the lowest employee protection for permanent contracts from the EU-15. In the Netherlands, the terms on which an individual can be fired are strict; the level of the discharge fee is relatively less stringent. (Chkalova et al., 2015)

While the Netherlands is one on the highest scoring countries in 2013 on employee protection for permanent contracts, they are one of the lowest when it comes to employee protection for temporary contracts (Chkalova et al., 2015). That is probably why the Netherlands also was the country with the highest rise in use of temporary contracts in the period 2000-2017 (Euwals, de Graag-Zijl, van Vuuren, 2016), as can be seen in figure 3. The connection between employee protection and permanent or temporary contracts is also visible in other countries. The UK for example has very limited employee protection for both temporary and permanent contracts, so for employers it does not really matter what type of contracts they award. That is why the share of temporary contracts in the UK is also relatively low.

Figure 3: Change in the share of employees with a temporary contract between 2000 and 2017 of EU15 countries

Source: Eurostat

2.3.3. ARGUMENTS FOR TEMPORARY CONTRACTS

Both employers and employees may have preferences when it comes to choosing between permanent or temporary contracts. Two main benefits of temporary contracts that were already mentioned are that a temporary contract can minimize risks (for example inferior health risks) and that it can increase flexibility. Permanent contracts however are more suitable for example when it is important that employees develop and become more efficient in the organisation. A survey among Dutch employers showed that for 73% of the employers, the quality of the employee was an important reason to choose for a permanent contract (Ecorys, 2014, pp.25). Quality in this case is specified as the preservation of knowledge and skills. When employers choose for a temporary contract this is mainly because of volatility, risk (aversion) and the recruitment procedure (Ecorys, 2014). Volatility is important in seasonal work and a temporary contract can be used as part of the recruitment procedure when the employer uses this as a trial period. For this research, the argument of risk aversion is the most important factor to examine, because this is important when it comes to employees with a disability or inferior health.

The risks and costs for an employer are illustrated in the following example. Assume an employee with a gross income of €36.500. If the employee gets sick or has a minor disability and is absent the daily costs can be as high as €410 a day. This is caused by continued wage payment, the costs of replacement and the costs of sickness counselling. In the first two years of sickness leave, the employer is responsible for the continued payment of the employee where the employer has to pay at least 70% of the wage of the employee. Moreover, the employee cannot be fired during this period on the account of being sick. If the employee would have had a temporary contract, the employer is still responsible for the same continued wage payments as with a permanent contract, but with a temporary contract of for example one year, the employer only has to pay one year of wage. A temporary contract can therefore lower the financial risk of the employer

-8.0% -6.0% -4.0% -2.0% 0.0% 2.0% 4.0% 6.0% 8.0% 10.0%

ES GR DE* UK FI EU15* BE FR AT SE* PT IE LU* DK IT NL * 2005 data

Not only employers have incentives for temporary contracts, employees can have their reasons as well. They often have less choice than employers; 61% of the employees say that their motivation for a temporary contract is involuntary and that they are actually looking for a permanent job. Only 39% of the employees mention that their decision is driven by their own availability or personal preferences (Ecorys, 2014).

As was mentioned before, a labour market with temporary contracts can provide flexibility. It is hard to assess what the optimal level of flexibility is and a flexible labour market can have advantages as well as disadvantages. Flexibility comes with adaptability, which can be an advantage especially for employers, because adaptability enables employers to adjust their labour force quickly to the needs of their organisation or the status of the economy. For employees, a high level of flexibility can have disadvantages. It was already mentioned that a high percentage of Dutch employees see their temporary contract as unavoidable and that they are actually looking for a permanent contract. Moreover, in the Netherlands especially vulnerable groups experience the disadvantages of a flexible labour market (Euwals, de Graag-Zijl, van Vuuren, 2016).

In addition to the changes in legislation in the Dutch disability insurance market, there are also other factors that influence the chance of receiving a temporary or permanent contract. In the Netherlands, there is a higher probability of having a temporary contract if a person is female, young or lower educated (EU Labour Force Survey 2008, as cited in Muffels, 2013, pp. 82). While these factors are already known, it is interesting to examine if these factors influence the disabled group that is the focus of this research.

2.4

HYPOTHESES

Based on the literature some assumptions and hypotheses can be made that will help answer the research question. In this part, the hypotheses that will be tested in the analysis will be described.

Hypothesis 1: From the years 2000 until 2014, the relative chance of people with a disability or inferior health to participate on the labour market has decreased.

The first hypothesis is stated to examine if people with inferior health or people with a disability have any labour contracts at all. As has been mentioned earlier, the increased responsibilities for employers that were part of the reforms of the Dutch disability insurance system could discourage employers from hiring “risky” workers. While this thesis focusses more on the type of labour contracts that people with a disability or inferior health receive, it is also important to examine if they receive any contracts at all. Therefore, the period 2000-2014 is taken – the period in which the reforms took place – to examine if the relative chance of people with a disability or inferior health to have a job has decreased. The relative chance is mentioned because there are several factors, like economic or political climate, which influence the labour market. Therefore, the chance of having a job for people with a disability or inferior health is compared with the chance for people with good health. Generally, changes in the labour market will not differentiate between people with good or bad health, so if there are differences between these groups then they can be linked to the reforms.

Hypothesis 2: From 2000 until 2014, the chance for employees with a disability or inferior health to have a temporary contract has increased when compared with the chance for non-disabled and healthy employees to have a temporary contract.

This hypothesis is based on the assumption that the overall chance of an employee having a temporary contract has increased. This assumption is based on the literature, but to examine the hypothesis this effect will also be tested in the empirical analysis. This is necessary because the goal of the research is to examine if an increase in temporary contracts is higher for employees with a disability than for non-disabled employees.

Hypothesis 3: Due to reforms in the Dutch disability insurance system in 2002, 2004 and 2006, the chance for Dutch employees with a disability to have a temporary contract has increased.

The function of the last hypothesis is to test if a possible change in the amount in temporary contracts for employees with a disability or inferior health is caused by the changes of legislation in the Dutch disability insurance system. The literature suggests that this is the case, because the employer has more responsibilities in the new system and it is likely that employers will therefore be more reluctant to offer permanent contracts to high-risk potential employees.

With these three hypotheses it is possible to eventually answer the research question. There are however more assumptions and expectations that are examined in this study, such as potential effects for disabled employees with certain characteristics. These characteristics can for example be gender, age and severity of the disability. The literature claims that gender and age do matter when it comes to the change of having a temporary contract, but this is not specifically for employees with a disability. It is therefore interesting to examine if these factors also make a significant difference when it comes to disabled employees.

3 METHODOLOGY

3.1

RESEARCH DESIGN

The purpose of this study is to examine whether Dutch employees with a disability have a higher chance of having a temporary contract rather than a permanent contract and if this chance has changed because of legislation. In order to test this possible change, longitudinal data about employees is analysed in a quantitative research design. It is necessary to use longitudinal data, because the situation before and after the change in legislation has to be analysed. Regression analysis is used to explain the relationship between the dependent and independent variables.

The most important dependent variable in this research is the probability of having a temporary or permanent contract. This will be further specified in the operationalization. The most important independent variable is the variable that specifies whether an employee has a disability or not. Several control variables are added like degree of disability, sex and age. Then regressions will be run for the several waves to examine the relationship between the independent and dependent variables throughout the years. The used regression analyses are based on a linear regression. In equation [1] a linear regression is described where Y represents the dependent variable or the outcome of the regression. The intercept is β0 and the independent variable is described in β1Xit, where β1

represents the effect of the independent variable on the dependent variable.

𝑌

= 𝛽

+ 𝛽

𝑋

+ 𝜀

[1]

It most cases, the regression will include more than one independent variable. In that case the formula is expanded, which can be seen in equation [2]. When the example from hypothesis one is used, Xnt can for example

represent a dichotomous variable that describes the sex or age of the respondent. Βn describes the effect of this

particular variable on Y.

𝑌

= 𝛽

+ 𝛽

𝑋

+ 𝛽

𝑋

+ 𝜀

[2]

Because in some cases the dependent variable has a binary outcome (either 0 or 1), linear regression is not possible. Therefore, a probit regression will be used to find a possible relation between the variables. Another method that will be used is the Heckman selection model. This method is necessary because in some of the probit regressions selection bias can occur. With the Heckman selection model it is possible to estimate two models at the same time. The analysis will be conducted with the statistical software program Stata, version 14. The script that is used to produce the results from the analysis is provided in appendix A.

The analysis will focus on the period 2000 – 2014. This range is selected because this research focusses on the legislation in the years 2002, 2004 and 2006. More information about the data is provided in the next sub-chapter.

3.2

DATA COLLECTION

The data that is needed for the regression analysis should contain at least the variables that are mentioned in the hypotheses. Therefore, a longitudinal dataset of Dutch citizens, preferably with a large sample, is needed. The dataset that best fits the criteria is the “Arbeidsaanbodpanel (AAP)”, which translates as the “labour supply panel”. This questionnaire is sent out every other year to a sample of the Dutch population between the age of 16 and 66, the potential workforce, and is initiated by the “Organisatie voor Strategisch marktonderzoek” from 1985 until 2010. From 2010 onwards, the Netherlands Institute for Social Research (SCP) is responsible for the questionnaire. The sample size is between 4500 and 5000 and the focus of the questionnaire is on employment, but the questionnaire contains many other questions which can be important for the control variables. More extensive information about the labour supply panel can be found at the website of the SCP.1

As was already mentioned, this research will focus on the period 2000 - 2014. While there is data from before 2000, the questions that are used in this research are not present in the waves before 2000. Because the period is 2000 – 2014 and the questionnaire is published every other year, eight waves are examined in this research.

3.3

OPERATIONALIZATION

Labour participation

Before a distinction is made between permanent and temporary contracts, the labour participation is assessed. The labour participation is important to take into account because labour participation can influence the other variables. Moreover, when employers are worried about hiring high-risk workers, they can lower their risk by giving them a temporary contract, but they can also not hire them at all. This illustrates that it is not only important to look at the type of contracts for workers with inferior health or disabilities, but to examine their labour participation as well.

For labour participation no distinction is made between full-time and part-time workers. There are multiple options for type of labour participation, but for the simplicity of this study the possible options will be converted to a binary variable for having a job and not having a job.

Table 1: Operationalization of concept of labour participation

Variable Question in questionnaire Included answers

Has a job _{How are you employed at this moment?} Salaried employee, self-employed Does not have a job _{How are you employed at this moment?} No paid work, day training

Temporary or permanent contract

Another variable which has to be specified is the dependent variable that is the focus of this study: the probability of having a temporary or permanent contract. According to the Dutch government, a temporary contract is a contract with an expiration date (Rijksoverheid, n.d.). If an employee wants to continue working at the company

after the expiration date, a new contract has to be signed. A permanent contract on the other hand, is a contract without a set expiration date. In principle, no distinction will be made for part-time and full-time jobs, because temporary and permanent contracts can be used in both part-time and full-time jobs.

What is important to keep in mind in this operationalization is that the variable should have a binary outcome: either a temporary contract or a permanent contract. In the questionnaire, more than two options are available when participants have to mention their type of employment contract such as ‘temporary contract with expectation of permanent contract’. For the focus of this study the possible options will be converted to a binary variable for just a temporary or permanent contract.

Table 2: Operationalization of concept of type of contract

Variable Question in questionnaire Included answers

Temporary contract What type of employment contract do you have?

Temporary contract, temporary contract with expectation of permanent contract

Permanent contract What type of employment contract do you

have? Permanent contract

Inferior health and disability

The most important independent variable is the variable that specifies inferior health or a disability of an employee. For this variable, many options can be taken into account like the degree of disability, whether the respondent receives disability benefits and health as rated by the employee itself. The most obvious way to operationalize the concept of a disabled employee is to ascertain if a person received disability benefits. This information is present in the data and can provide a correlation between being disabled and the type of contract. The downside however is that only a small sample of people in the data receive disability benefits. This is also the case with sickness benefits.

It is therefore useful to use an additional operationalization of the concept of being disabled. In the data there are also questions about the respondents’ health. The question that can be useful for this research is: ‘How would you rate your health on average?’ While this question might be a slightly less adequate representation of the concept of the disabled employee, it does provide a much larger sample. An overview of the operationalization of the concept of disabled employee can be found in table 3.

Table 3: Operationalization of concept of health

Variable Question in questionnaire Included answers

Health scale How would you rate your health on average? Very good, Good, Average, Bad, Very Bad

Health binary (good health) How would you rate your health on average? Very good, Good

Health binary (inferior health) How would you rate your health on average? Average, Bad, Very Bad

Received disability benefits Did you receive disability benefits? Yes

Does not receive disability

Control variables

The control variable of education level can have an impact on the dependent variable, as the literature suggests. For this operationalization, three levels of education are used: lower, middle and higher education. They are based on the variable in the questionnaire that indicates the completed education level. They are arranged according to the Dutch Standard Classification of Education (Standaard Onderwijsindeling), which consists of seven levels. For the operationalization of this variable the levels are rearranged to three levels.

Table 4: Operationalization of control variables

Variable Question in questionnaire Included answers

Gender _- Male, Female

Age _- #

Marital status _{What is your marital status?} Married, living with partner, divorced, widow/widower, single

Education _{What is your highest finished education?}

Primary education, lower secondary education, higher secondary education, higher professional education, university education.

Year _- 2000 - 2014

For this research, a period from 2000 until 2014 is examined. The Dutch Standard Classification of Education changed during this period, but the seven levels remained almost entirely the same. This will therefore not have an impact on the variable that represents education level. The operationalization can be found in table 4.

3.4

VALIDITY

3.4.1. INTERNAL VALIDITY

Some aspects have to be acknowledged about the internal validity in this research. Because the data is obtained through a questionnaire, there is only data about stated information instead of observed information. The variable that represents overall health is an example of a variable that is completely dependent on the subjective judgement of the respondent. With other variables like education level, the stated level will likely be a better representation of the observed level.

Another aspect which should be taken into account is the fact that the questions can differ between the different waves of the questionnaire. These differences are limited however and should have minimal or no impact on the answers of the respondents. Related to this aspect is the fact that the order of the questions can change between different waves.

3.4.2. EXTERNAL VALIDITY

The external validity represents how well the results of this research can be generalized to other similar situations. In this research this sheds a light on how well the results of the regression analysis on the data from the survey can be used as a measure for the entire Dutch labour force. Because the sample is sufficiently large – between

4500 and 5000 respondents per wave – it should be a good representation of the Dutch labour force. The sample did become a little bit skewed throughout the years however, because the respondents were a slightly older and higher educated than the average of the Dutch labour force. That is why from the year 2000, sampling weights were introduced to have a more even distribution. In 2012, the sampling weights were altered slightly and because this influences two waves examined in this study, this may have a small effect on the validity.

4 ANALYSIS

In this analysis the hypotheses will be tested with the available data. The first part of the analysis will focus on the first three hypotheses and will explain if people with inferior health and disabilities have labour contracts and if they do, what type of contracts they receive. The second part of the analysis focusses on the reforms of 2002, 2004 and 2006 in the Dutch disability insurance system and is aimed at finding a correlation between the reforms and possible changes in the share and type of contracts among people with a disability or inferior health.

4.1

LABOUR MARKET PARTICIPATION AND LABOUR CONTRACTS

4.1.1. HEALTH AND LABOUR MARKET PARTICIPATION

Before a possible change in type of contracts will be examined, it is important to look at labour contracts in general. This aspect is described in hypothesis 1:

From the years 2000 until 2014, the relative chance of people with a disability or inferior health to participate on the labour market has decreased.

In table 5 the distribution of labour participation among several characteristics is described. The first characteristic is gender which shows that the percentage of males that have a job is more than 10% higher than the percentage of females that have a job. This is caused by the fact that in some cultures historically, women were expected to stay at home to care for the family, which is in this case not perceived as labour market participation. Although the labour participation of women has been increasing for many years now, there are still many women that are inactive on the labour market. When the characteristic of age group is taken to measure labour market participation, the results are as expected. In the age group of 16-24 years the labour participation is relatively low, because young people are often still studying or might live with their parents. In the following age groups the labour participation is much higher. The age group of 55-64 and especially the age group of 65-66 have a low average labour participation, which can be explained by the fact that a large share of people in these age groups have already retired from work. The level of education also has an impact on the labour participation of respondents. It can be seen in table 5 that respondents with higher education generally have a higher chance of having a job. This can be explained by the fact that people with a higher education have a large pool of jobs to choose from, while lower educated people will generally have a limited choice in their job opportunities.

Table 5: Distribution of labour participation among gender, age group and education level

Characteristics Labour market participation

No job Has a job

Gender Male 18.95 % 81.05 % Female 30.87 % 69.13 % Age group 16 - 24 years 32.61 % 67.39 % 25 - 34 years 13.89 % 86.11 % 35 - 44 years 13.76 % 86.24 % 45 - 54 years 16.27 % 83.73 % 55 - 64 years 44.94 % 55.06 % 65 - 66 years 90.79 % 9.21 % Education Primary education 49.52 % 50.48 % Lower secondary education 35.87 % 64.13 % Upper secondary education 22.63 % 77.37 % Higher professional education 16.35 % 83.35 % University education 13.39 % 86.61 %

In this thesis, the focus is on people with a disability and inferior health. In table 6 several health and disability indicators are included with their distribution among labour participation. The first indicator represents health on a five-step scale. In general it can be concluded that respondents that classified themselves in lower health scales also have a lower chance of having a job. The second health indicator, which is on a binary scale, is based on the first indicator, and shows similar results. The last indicator is less dependent on the respondent’s opinion of health than the first indicator, because the respondent has to indicate whether he or she receives disability benefits. This indicator shows that most respondents who receive disability benefits do not have a job.

Table 6: Distribution of labour participation among several health indicators

Indicators health/disability Labour market participation

No job Has a job

Health scale Very good 20.57 % 79.43 % Good 18.97 % 81.03 % Average 36.60 % 65.40 % Bad 64.01 % 35.99 % Very bad 79.46 % 20.54 % Health binary Good 19.39 % 80.61 % Average/Bad 40.31 % 59.69 % Disability benefits

Receives disability benefits 70.92 % 29.08 % Does not receive disability benefits 19.40 % 80.60 %

To actually test hypothesis 1, regression analyses are conducted. The first regression uses the health scale as an indicator and the results can be seen in table 7. Several control variables are taken into account, but the primary focus is on the independent variable Health scale. It shows that the variable has a negative correlation with the dependent variable Labour market participation, which indicates that lower levels of health have a lower chance of having a job. This result is significant and it confirms the observations from table 6. All the other independent variables are significant as well, which shows that there is a correlation between these independent variables and labour market participation. Some of the observations made in table 5 are confirmed by the results of the independent variables age, gender and education.

Table 7: Regression analysis of health (scale) on labour participation

Independent variables Labour market participation

B SE Health scale -0,084*** 0,004 Age 0,059*** 0,002 Age^2 -0,001*** 0,000 Gender -0,163*** 0,007 Marital status -0,009*** 0,003 Education 0,052*** 0,003 Year 0,006*** 0,001 Note: *p < .1; **p < .05; ***p < .01

To actually define a gap between healthy and less healthy respondents, it is necessary to use the binary indicator for health, which is based on the five-step health scale. The results of this regression are reported in table 8. Because the binary indicator for health is based on the health scale variable, the results are almost the same as with the previous regression. The correlation between the independent variable Health binary and the dependent variable Labour market participation is -0,163. Because a binary variable is used, it is now possible to conclude that having inferior health decreases the chance of having a job with 16.3 percentage point. The correlation is significant as well as the correlations for the other variables.

Table 8: Regression analysis of health (binary) on labour participation

Independent variables Labour market participation

B SE Health binary -0,163*** 0,009 Age 0,058*** 0,002 Age^2 -0,000*** 0,000 Gender -0,164*** 0,007 Marital status -0,009*** 0,003 Education 0,053*** 0,003 Year 0,006*** 0,000 Note: *p < .1; **p < .05; ***p < .01

The results that are presented in table 8 provides insights in the effect of health on labour market participation, but hypothesis 1 states an effect over time. As is already suggested in table 8, the variable Year has a very small

positive effect on labour market participation, but this variable does not specify the respondent’s health over time. For this an interaction variable is created, which will be included in the next part of this chapter.

To analyse the changes in the labour participation of respondents with inferior health it is necessary to compare the labour market for people with inferior health with the labour market for people with good health. This is because the labour market itself is dependent on factors like the economical and political climate. In table 9 the difference in labour participation between good and bad health respondents is calculated to examine how the labour market participation of people with inferior health has changed compared with the labour market participation of people with good health. A high difference – reported in the last column of table 8 – therefore means there is a large gap on the labour market between people with good health and people with inferior health. In this thesis the size of the gap is less important, what is more important is how this gap changes over time. Overall it can be concluded that the gap became smaller, although there is a small growth again in the years 2012 and 2014. However, overall it can be concluded that the average gap of the years 2000 and 2002 is higher than the average gap of the years 2012 and 2014.

Table 9: Differences in labour participation for respondents with good and inferior health

Year Good health Inferior health Difference

2000 81.89 % 58.06 % 23.83 % 2002 82.41 % 60.54 % 21.87 % 2004 77.90 % 55.99 % 21.91 % 2006 78.30 % 57.26 % 21.04 % 2008 80.44 % 62.31 % 18.13 % 2010 79.67 % 62.65 % 17.02 % 2012 82.99 % 62.36 % 20.63 % 2014 82.26 % 59.43 % 22.83 %

Another way to test if the labour participation chances of people with a disability or inferior health change through time is to use an additional variable that is composed of the time variable and the variable that indicates health. In Table 10 this variable is included, but it can be seen that the result is not significant on at least the 10% level. Therefore, it is not possible to base conclusions on this variable from this regression analysis. What can be concluded however is that that respondents with inferior health have a 16.7 percentage point less chance to participate on the labour market than healthy respondents on average over the period 2000-2014 This result is almost similar to the result that was found in table 8.

Table 10: Regression analysis of health trough time on labour participation

Independent variables Labour market participation

B SE

Health binary -0,167*** 0,016

Health trough time 0.000 0,001

Age 0,058*** 0,002 Age^2 -0,001*** 0,000 Gender -0,164*** 0,007 Marital status -0,009*** 0,003 Education 0,053*** 0,003 Year 0,006*** 0,001 Note: *p < .1; **p < .05; ***p < .01

Because it is not possible to draw conclusions from the labour participation chances of people with a disability or inferior health through time with the regression analysis, it is not possible to provide clear evidence for hypothesis 1. Table 9 provides some small insights, but those insights rather invalidate than confirm the hypothesis. It can therefore not be said that from the years 2000 until 2014, the relative chance of people with a disability or inferior health to participate on the labour market has decreased.

4.1.2. HEALTH AND TYPE OF LABOUR CONTRACTS

In the previous sub-chapter the effects for labour market participation were described. While this is an important aspect, the focus of this thesis is on the type of contracts that respondents have. This is illustrated in the second hypothesis:

From 2000 until 2014, the chance for employees with a disability or inferior health to have a temporary contract has increased when compared with the chance for non-disabled and healthy employees to have a temporary contract.

The distribution of labour contracts for various characteristics is described in table 11. What is important to note is that this only includes respondents that have an actual labour contract.

Table 11: Distribution of type of labour contracts among gender, age group and education level

Characteristics Type of labour contract

Permanent contract Temporary contract Gender Male 88.17 % 11.83 % Female 83.62 % 16.38 % Age group 16 - 24 years 43.60 % 56.40 % 25 - 34 years 82.59 % 17.74 % 35 - 44 years 91.19 % 8.81 % 45 - 54 years 93.48 % 6.52 % 55 - 64 years 94.92 % 5.08 % 65 - 66 years 77.92 % 22.08 % Education Primary education 69.63 % 30.37 % Lower secondary education 86.49 % 13.51 % Upper secondary education 84.46 % 15.54 % Higher professional education 89.44 % 10.56 % University education 87.13 % 12.87 %

When gender is taken into account, it can be seen that males have a slightly higher average of permanent contracts than females. This can again be explained by the fact that in some cultures historically, women were expected to take care for the family. Women will therefore be more likely to have a job which can be combined with taking care for the family than men. If age groups are examined it can be seen that young people have the highest average of temporary contracts, which can be explained by the fact that they often have to combine a job with their education. Young people will therefore be more likely to seek temporary jobs as well as part-time jobs. In the age from 25 until 64, the percentage of permanent contracts is much higher than the percentage of temporary contracts. In the highest age group, the percentage of permanent contracts is significantly lower. This is because it is difficult for older people to find a job if they do not have one. One aspect of old people that is an important theme in this thesis is that they are more likely to have health issues and that could be one of the reasons that employers are hesitant to hire older workers. Moreover, older people might lack certain skills or knowledge that younger employees do have. Also, in this age group people are about to retire, so a temporary contract might cover them until their retirement. The last characteristic, education, is distributed quite evenly except for the group with primary education. The percentage of permanent contracts for the group with primary education is much lower than all the other groups. In table 5 it already was clear that respondents with only primary education had the lowest percentage of labour participants, so it is probably hard for people with only primary education to actually find a job, especially a permanent job.