Research Article

The causal influence of increasing the statutory retirement age on job satisfaction among older workers in the Netherlands

Patrick Pilipieca School of Business and Economics, Maastricht University, 6211 LM Maastricht, The NetherlandsCorrespondencepatrick.pilipiec@maastrichtuniversity.nl
View further author information

Wim Grootb Maastricht University, School of Business and Economics; Department of Health Services Research; CAPHRI, Maastricht University Medical Center; Faculty of Health, Medicine and Life Sciences, 6211AX Maastricht, The NetherlandsView further author information

Milena Pavlovac Maastricht University, Department of Health Services Research; CAPHRI, Maastricht University Medical Center; Faculty of Health, Medicine and Life Sciences, 6229 GT Maastricht, The NetherlandsView further author information

Research Article

The causal influence of increasing the statutory retirement age on job satisfaction among older workers in the Netherlands

ABSTRACT
Formulae display:?Mathematical formulae have been encoded as MathML and are displayed in this HTML version using MathJax in order to improve their display. Uncheck the box to turn MathJax off. This feature requires Javascript. Click on a formula to zoom.

ABSTRACT

Since 2013, the Netherlands has gradually increased the statutory retirement age. We use a regression discontinuity design to analyse the effect of the increase in retirement age on overall job satisfaction and satisfaction with the organisation’s personnel policy. Date of birth was used as a sharp cut-off to assign workers to the intervention or comparison group. The increase had no effect on overall job satisfaction. An effect was only observed for satisfaction with the personnel policy in one of the groups analysed. Further, there was only a significant effect if the statutory retirement age was increased by seven months, but this relation was no longer significant when controlling for the difference in days between the date of birth and the cut-off.

Keywords:

Job satisfaction Organisation’s personnel policy Pension reform Regression discontinuity Statutory retirement age Work participation

I. Introduction

Many countries are confronted with rapid population ageing (Wheaton and Crimmins 2012; Bernal and Vermeulen 2014; Domonkos 2015; Lüthen 2016). In addition, the number of young people entering the labour market is declining (OECD 2006). If work and retirement patterns remain unchanged, the economic dependency ratio – the unemployed and pensioners as a percentage of the employed – in OECD countries is expected to increase from 38% in 2000, to more than 70% in 2050 (OECD 2006).

As a consequence, population ageing in combination with lower economic growth and a declining influx of younger workers into the labour market, imposes a significant burden to sustaining the public pension programmes (Bernal and Vermeulen 2014; Templer, Armstrong‐Stassen, and Cattaneo 2010). In most EU member states, there is a strong relation between the age of statutory retirement and the age of pension eligibility. In fact, in many of these European countries, pension eligibility starts at the day of statutory retirement. Although workers may voluntarily withdraw from the labour market at any time in their life, they only become eligible to pension benefits upon reaching the retirement age.

To alleviate this burden of an ageing society, many EU member states have taken measures to increase the labour force participation and to delay pension eligibility of older workers by increasing the age for statutory retirement (De Wind, Van der Pas, and Blatter 2016). As a consequence, workers in these countries are required to continue working beyond the age of 65 before they become entitled to pension benefits (Reynolds, Farrow, and Blank 2012). The Netherlands is one of the EU member states that has increased the age of retirement.

In 2012, the Dutch government, trade unions, and employer federations agreed to reform the public pension by increasing the statutory retirement age gradually from 65 in 2012, to 67 in 2023 (Bernal and Vermeulen 2014; Rijksoverheid 2012). Subsequently, by the law, which was signed in July 2012 and, taking effect on 1 January 2013, the statutory retirement age was increased by one month to 65 years and one month. This change was applied to all individuals born after 31 December 1947, but before 1 December 1948 (Staatsblad 2012; Rijksoverheid 2014). Therefore, the implication was that individuals born between 1 January 1948, and 30 November 1948, were compelled to prolong their labour participation by one month to meet the new legal requirements for statutory retirement and to claim 100% of the accumulated basic state pension benefits (Sewdas et al. 2017). After this, the retirement age was further increased by 1 to 3 months each year. The precise incremental increase of the statutory retirement age, which was announced publicly and received a lot of public attention, is outlined per calendar year in Table 1.

Table 1. Incremental increase and relative change of the statutory retirement age per calendar year

CSV Display Table

In addition to increasing the age of retirement and delaying the eligibility to pension benefits, workers were obligated to continue to pay mandatory social contributions and the premiums for voluntary pension plans. While these contributions were also increased, entitlement to pension benefits was delayed and benefits were not increased accordingly. Instead, for many workers, the pension benefits that they were entitled to, were later also reduced.

Similar to other countries (Menon 2017; Wynaendts 2012), there has been fierce public opposition and resistance to increasing the retirement age in the Netherlands (Boschman and Van Alphen 2018; Henkens et al. 2016). This policy intervention was, and continues to be, very unpopular and highly criticised in the Dutch media, by trade unions, politicians, and in society (Boschman and Van Alphen 2018; MAX Vandaag 2016; PlusOnline 2016).

To the best of our knowledge, there is only one study that has investigated the emotional responses of older workers to an increase in the retirement age in the Netherlands. Van Solinge and Henkens (Van Solinge and Henkens 2017) report that 44% of older workers were ‘very or even extremely angry about the increase in the retirement age’. In addition, many older workers felt highly worried about their mental or physical ability to continue working until the new and higher retirement age (Van Solinge and Henkens 2017). Furthermore, it appeared that having access to social resources (i.e. support from partner) and personal resources (i.e. wealth and health) lowered these negative emotions. Overall, Van Solinge and Henkens (Van Solinge and Henkens 2017) found that older workers with demanding jobs had stronger emotions of anger and worry if their access to these social and personal resources was less.

In addition to dissatisfaction with the decision to raise the retirement age and also amplified by the unsuccessful public protests that did not result in reinstating the former pension legislation, older workers may also become dissatisfied with their current job and the conditions of their employment. For example, a frequently used argument was that the policy reform would lead to dissatisfaction among (older) workers, and, therefore, would result in decreased labour force participation and productivity (Henkens et al. 2016; Researchcentrum voor Onderwijs en Arbeidsmarkt (ROA) 2009).

This hypothesised effect may be better understood using the Two-Factor Theory of Motivation by Herzberg, which explains that some work characteristics result in job satisfaction, while other characteristics prevent job dissatisfaction (Herzberg, Mausner, and Snyderman 1959). The former are motivational factors such as meaningfulness of work, opportunities for growth and promotion, recognition, responsibility, and a sense of achievement (Herzberg, Mausner, and Snyderman 1959). The latter are hygiene factors such as company and administrative policies, employment benefits, interpersonal relations, job security, salary, status, and working conditions (Herzberg, Mausner, and Snyderman 1959). We acknowledge that this theory has been criticised because of methodological deficiencies and inconsistency with previous evidence about motivation and satisfaction (House and Wigdor 1967). However, the theory has been very influential in the field of psychology and management science, because it recognises that there is a multitude of factors that can motivate people if satisfied, while other factors may prevent satisfaction if not dealt with appropriately.

This theory is particularly relevant because working longer specifically relates to ongoing work in an organisation. While life satisfaction is a more general proxy of happiness, the national policy that increased the retirement age, and thus results in prolonged labour force participation, may have a very specific impact on the workers’ satisfaction with personnel policy of the organisation. It may be possible that the increase of the retirement age shifted the focus of workers towards hygiene factors, which resulted in job dissatisfaction (Herzberg 1974).

In addition, in line with the notions of fairness and the existence of a psychological contract between the employee, employer, and the government, workers may consider the incremental increase of the statutory retirement age as unfair, unjust, and a breach of their psychological contract. In addition, one study on another pension reform reported that the job satisfaction of workers was strongly affected by a change in the Dutch retirement system, and although this effect was mostly non-monetary, they explained this shock by social comparisons with colleagues (Montizaan, Cörvers, and De Grip 2010). The perceived breach of fairness and the psychological contract may then also result in a decrease in overall job satisfaction and dissatisfaction with the organisation’s personnel policy.

On the contrary, in a study on preferences towards retirement, Brougham and Walsh (Brougham and Walsh 2009) reported that the fulfilment of one’s need for relatedness, identity, and growth are equally important for workers who have early (i.e. voluntarily withdraw from labour participation before the retirement age) and late retirement (i.e. continue voluntary labour participation after the retirement age) intentions. When workers intend to prolong their work participation, they often perceived that continuing their job fulfils their financial need, facilitates their goals for relatedness and identity, and provides opportunities for growth (Brougham and Walsh 2009). Also, in the context of another pension reform, a large Dutch private pension insurer suggested in 2005 that all of its 1.2 million participants had anticipated to their potential losses when changing a pension scheme by increasing their individual pension savings, or by postponing their retirement to increase their pension savings (Montizaan, Cörvers, and De Grip 2010). Another study reported that workers who intended to retire early had similar goals, but perceived that their fulfilment could instead be found in retirement (Brougham and Walsh 2009).

Considering these diverse preferences for retirement documented in the literature, but also the effect that increasing the retirement age in the Netherlands has on the eligibility to pension benefits, it is unclear how the increased retirement age affects job satisfaction of older workers. This paper, therefore, evaluates the impact of an increase in the retirement age on older workers’ overall job satisfaction and satisfaction with the personnel policy of the organisation. To our knowledge, there has been no study on the relationship between the 2012 pension reform in the Netherlands and the overall job satisfaction and satisfaction with the organisation’s personnel policy among older workers.

Job satisfaction

Job satisfaction is an attitude that was defined by Weiss (Weiss 2002) as an ‘evaluative judgment one makes about one’s job or job situation.’ Other authors added that job satisfaction is based on the overall subjective evaluation of beliefs, experiences, and emotions related to the work environment (Davies, Van der Heijden, and Flynn 2017; Roelen, Koopmans, and Groothoff 2008; Alegre, Mas-Machuca, and Berbegal-Mirabent 2016; Cheung and Wu 2013).

Many different factors influence job satisfaction. These can be classified into four sub-groups (Huysse-Gaytandjieva, Groot, and Pavlova 2013). First, socio-demographic features involve the gender, age, health status, education, and marital status of workers. Second, personality attributes, among others, include the factors self-esteem, commitment, locus of control, and the big five personality traits. Third, employment conditions, such as fulltime and part-time work, on-the-job training, and union membership influence job satisfaction. Fourth, work-related contextual features include the local unemployment rates. The former two sub-groups involve personal characteristics of employees, while the latter two sub-groups are work-related factors (Huysse-Gaytandjieva, Groot, and Pavlova 2013).

However, in the literature, there is no agreement on a uniform measurement of job satisfaction. Instead, Linz and Semykina (Linz and Semykina 2012) distinguish two distinctive strategies that are frequently used to operationalise this concept. Economists predominantly gauge the concept as overall job satisfaction which is measured by a single item (Linz and Semykina 2012). Management scientists and psychologists frequently use multiple items to measure various aspects of one’s job, and subsequently calculate job satisfaction as the composite of these dimensions (Linz and Semykina 2012). An example of a well-known and validated instrument that measures different work characteristics that might lead to job satisfaction is the Job Diagnostic Survey (Hackman and Oldham 1975).

Although no agreement for a gold standard yet exists (Roelen, Koopmans, and Groothoff 2008), researchers appear to have formed a consensus about the importance of job satisfaction perceived by workers. Specifically, job satisfaction is associated with many labour market outcomes. Prior studies have found that a higher job satisfaction is associated with increased job productivity (Patterson, Warr, and West 2004), increased collective job performance (Whitman, Van Rooy, and Viswesvaran 2010), stronger organisational citizenship behaviour (Foote and Tang 2008), decreased financial turnover (Wright TA, Bonett DG. Job Satisfaction and 2007), and stronger intentions of employees to stay (Meyer et al. 2002). Considering that job satisfaction impacts job performance, Judge et al. (Judge et al. 2001) published a review of seven models that explain the precise relationship between job satisfaction and job performance. While some models received much support, they however find no superior model that best describes this relationship (Judge et al. 2001). Overall, evidence supports that employing satisfied workers is beneficial and that enhancing their job satisfaction is preferred (Linz and Semykina 2012).

Furthermore, job satisfaction was also found to be an essential determinant for retirement decisions (Kosloski, Ekerdt, and DeViney 2001). Specifically, jobs that provide positive social relationships, intrinsic enjoyment, and opportunities for work-related promotion, were negatively related to the employee engaging in retirement planning behaviours (Kosloski, Ekerdt, and DeViney 2001). Therefore, understanding the association between job satisfaction and retirement expectations, and between job satisfaction and the potential decision to continue work beyond retirement, is relevant (Sewdas et al. 2017; Bonsang and Van Soest 2012; Coppola and Wilke 2014). In addition, if older workers are satisfied with their job, they may decide to voluntarily extend their labour force participation beyond the new statutory retirement age, which may further mitigate the financial burden on the public pension system (Ng and Law 2014; MacDonald et al. 2011).

It is, therefore, important and relevant to shed light on the possible influence of public pension reforms that increase the statutory retirement age, on overall job satisfaction (Sewdas et al. 2017).

Pension reform in the Netherlands

Although the legislative change in the Netherlands was long discussed in politics and the media, this reform took effect only six months after the law was signed (Staatsblad 2012), which may have provided insufficient time for workers to adapt to the new situation (cf. 43). Because of this forceful nature of the legislative change, the pension reform, through the mechanisms of fairness and the psychological contract (discussed below), may have impacted the job satisfaction of older workers.

Furthermore, when this bill took effect on 1 January 2013, individuals born between 1 January 1948, and 30 November 1948, were obliged to delay their anticipated retirement day by working one additional month longer, while individuals born before 1 January 1948, remained unaffected and were entitled to retire at age 65, and could thus claim their full public and occupational pension benefits (Staatsblad 2012; Rijksoverheid 2014). Similarly, since further increases of the legal retirement age were based solely on the date of birth, people in subsequent cohorts were obliged to work longer than those born in the preceding cohorts. For example, some cohorts of workers were required to prolong their work relative to the preceding cohort only by one month (e.g. in 2013, 2014 and 2015), while the cohort in 2019 was required to continue working for three additional months, relative to the 2018 cohort (see Table 1).

Workers who were required to continue to work beyond the age of 65 may have perceived working one additional month as a punishment, while those on the other side of this threshold may have perceived the earlier retirement as a positive extrinsic reward. In reference to the discussion above, workers with late retirement intentions may have perceived the prolonged work participation as a reward that allowed them to fulfil their financial need, facilitated their fulfilment for relatedness and identity, and provided opportunities for growth, via job employment (Brougham and Walsh 2009).

Relationship between increasing the statutory retirement age and job satisfaction

The literature suggests that two important mechanisms, namely fairness and a psychological contract, may explain the relationship of increased statutory retirement age with a decrease in overall job satisfaction and with satisfaction with the organisation’s personnel policy.

Van der Heijden and colleagues (Van der Heijden, Nelissen, and Verbon 1997) investigated the influence of perceptions of fairness and altruism on how persons evaluate the public pension system in the Netherlands. Fundamental to this study was that persons base their preferences and opinions on both their current and predicted future financial income (Van der Heijden, Nelissen, and Verbon 1997). While they reported that perceptions of fairness and altruism were primarily present among middle-aged and younger respondents, they also concluded that elderly workers were less concerned with their own interests and more willing to contribute to others (Van der Heijden, Nelissen, and Verbon 1997). If workers have financially contributed to the public pension programme, they may consider it unfair and unjust that, contrary to themselves, the preceding cohort is granted retirement earlier, while the obligation to continue to work is solely based on their date of birth. Furthermore, while at work, these workers contribute by paying income taxes and occupational pension-related premiums longer, but they will eventually receive a similar amount of public and occupational pension as those who retired earlier.

The second relevant mechanism is the psychological contract, defined as ‘the perception of an exchange agreement between oneself and another party’ (Rousseau 1998). This implicit and unwritten contract is based on the exchange relationship between the employee and his employer, and on the perceived reciprocal duties towards each other (Schalk and Roe 2007). In employment relations, the implicit obligations include that employees remain loyal to their employer and perform work in the interest of the organisation, and thereby contribute to achieving the overall strategic corporate goals (Latornell 2007). Similarly, the employer may be expected to provide a secure work environment to workers who perform work according to their job profile and to address the needs of workers to continue working until their retirement. Breaching this psychological contract may result in, among others, contra-productive work behaviors by the employee or may be a reason for voluntary withdrawal (Jensen JM, Opland RA, Ryan AM. Psychological Contracts and 2010). Therefore, perceptions of fairness may also be considered a component of the psychological contract.

In addition, the perceived unfairness and breach of the psychological contract may also result in dissatisfaction with the organisation’s personnel policy. The worker may perceive that, in the event of increasing the statutory retirement age, the personnel policy no longer provides the expected predictability, but instead facilitates the perceived unjust and unfair distinction between workers based on their date of birth (Van der Heijden, Nelissen, and Verbon 1997; Schalk and Roe 2007).

Therefore, this pension reform may have negatively influenced both overall job satisfaction and satisfaction with the organisation’s personnel policy. The opposite could be expected as well. The increased retirement age may also have positive effects on job satisfaction because prolonged work participation results in maintaining a higher income longer, more social contacts, and positive effects on one’s identity.

II. Methods

This study uses observational data from the ongoing prospective open cohort panel – Arbeidsaanbodpanel (AAP) – from the Netherlands Institute for Social Research (Sociaal en Cultureel Planbureau (SCP) 2016). Initiated in 1985, the AAP aims to investigate the work situation of employed and unemployed Dutch inhabitants, and the overall (potential) labour supply of the Dutch population (age 16 to 66) (Van Echtelt et al. 2016). The AAP contains three questionnaires, namely for the employed, unemployed, and students (Van Echtelt et al. 2016). The aim of the AAP is to provide information on a representative sample of the Dutch population; new respondents may, therefore, have been preselected to maintain a representative sample of about 5,000 participants to achieve the desired national external validity of the panel (Vlasblom, Van Echtelt, and De Voogd-Hamelink 2015). Since 1986, data are collected with a two-year interval, and to stimulate participation, participants receive a small financial compensation (7.50 or 10 Euros) upon completion of the questionnaire (Vlasblom, Van Echtelt, and De Voogd-Hamelink 2015; Sociaal en Cultureel Planbureau 2014). For each data wave, the data collection takes place from September to December of that year.

We use cross-sectional data from the 2012 wave (N = 4,837), because the measurement captured the required data not long after the Dutch pension reform was publicly announced, but before the Dutch pension reform of 1 January 2013 took effect. Considering the strong attention in the media and the subsequent protests, we believe that workers were aware of the anticipated pension reform. We use data from respondents who completed the questionnaire Arbeidsaanbodpanel Werkenden, which focusses on employed respondents. This resulted in a sample of 3,290 (68.0% of total) workers. The AAP dataset only includes the year of birth of participants and their calculated age on 1 October 2012. Given the importance of the month of birth, which was required to allocate the respondents to the applicable calendar cohort, we retrieved their corresponding month of birth using microdata from Statistics Netherlands (CBS).

To establish whether participants were salaried employees (i.e. employed by an organisation), the following inclusion and exclusion criteria applied. Eligible participants were in salaried employment (‘How are you currently employed?’) and reported to not currently receive state pension benefits or other pension-related income (‘Do you currently receive one or more of the following income types?’). This resulted in 2,372 (72.1% of the questionnaire employed) observations.

Overall job satisfaction was measured using the self-report item ‘How satisfied are you, everything taken together, with your job?’. The item was rated on a scale from 1 to 4 (1: ‘very satisfied’, 2: ‘satisfied’, 3: ‘dissatisfied’, and 4: ‘very dissatisfied’).

Satisfaction with the organisation’s personnel policy was gauged using the self-report item ‘Which report mark (from 1 to 10) would you give for the personnel policy of your employer?’, and measured specifically the satisfaction with the external hygiene factors (cf. 21). These factors involve work characteristics that Herzberg and colleagues found to be consistently related to job dissatisfaction, such as organisational policies, salary, and work conditions (Herzberg, Mausner, and Snyderman 1959). The item was rated on a scale from 1 (‘most negative’) to 10 (‘most positive’).

The intervention and comparison groups for each calendar cohort were established in line with the exogenous sharp cut-off points as defined by the Law on increasing the age of the General Old Age Pensions Act and the pension target age, which took effect in the Netherlands on 1 January 2013, and were, therefore, based on the date of birth of respondents (see Table 2). This resulted in five intervention and five comparison groups which were rotated depending on their respective analysis.

Table 2. Incremental increase and relative change of the statutory retirement age per calendar year

CSV Display Table

A regression discontinuity design was used to estimate the causal effect of prolonged work participation beyond age 65 on the self-perceived overall job satisfaction and satisfaction with the organisation’s personnel policy (Thistlethwaite and Campbell 1960; Maas et al. 2017). In this quasi-experimental design, the exogenous and sharp cut-off point, which was defined by the Dutch law as the date of birth, regulated to what extent the pension reform applied to the worker (Lee and Lemieux 2010).

In the analyses, we controlled for personal and work-related factors in the second and third model, respectively. The personal factors include gender, education (Lydon and Chevalier 2002; Tåhlin 2007; Fabra and Camisón 2009), and health (Georgellis and Lange 2007). Work-related factors include wage (Lydon and Chevalier 2002; Leppel, Brucker, and Cochran 2012), working hours, occupation, firm size, and being a supervisor (Linz and Semykina 2012). In addition, to control for the distance between the date of birth and the date of cut-off, we calculated this difference in days. Because the day of birth was absent in the dataset, we computed this distance using the first day of each month. Consequently, participants in the comparison group had a negative distance, while participants in the intervention group had a positive distance. A distance of zero was treated as the cut-off in the regression discontinuity plots. We additionally controlled for the first order polynomial of this distance in the fourth model, the second order polynomial in the fifth model, and the third order polynomial in the sixth model.

Furthermore, reported answers such as ‘unknown’, ‘not applicable’, and ‘I do not want to say’ to confounders were recoded as missing values, where these missing values were subsequently replaced by the sample mode if the variable had the nominal or ordinal type, and to the sample mean if the variable had the continuous type. Only a small fraction of the observations had missing values. In these cases, we used imputed values (see Appendix 1). Also, a dummy variable was added that indicated whether this value was imputed or not. In addition to the confounders, the related constructed dummies were included in the regression analyses as models.

This study investigated and compared six calendar cohorts relative to each other, which resulted in five cut-off points defined by the pension reform bill. The assignment of cohorts to either the intervention or comparison group was rotated. This means that the cohort for which the statutory retirement age was increased, relative to the preceding cohort, was assigned to the intervention group, while the remaining cohort was treated as the comparison group. Therefore, the treatment was prolonged work participation relative to the comparison group. This resulted in five analysis groups (see Table 2).

Our regression discontinuity design followed Lee and Lemieux (Lee and Lemieux 2010). Individuals with the date of birth, denoted as $X$ , that was equal to or above the cut-off point $c$ , received the treatment of the increased statutory retirement age, and thus were expected to prolong their work participation. Contrary, individuals with $X$ below $c$ were granted retirement earlier. Therefore, this exogenously set sharp cut-off point established the discontinuity. For each analysis group, denoted as sub $i$ , where sub $i \in \{1, 2, 3, 4, 5\}$ , as specified in Table 2, we constructed a dummy variable $D$ and congregated individuals below the cut-off point as the comparison group and those who received the treatment as the intervention group. Therefore, $D_{i} \in \{0, 1\}$ , where $D_{i} = 0$ if $X_{i} \geq c_{i}$ and $D_{i} = 1$ if $X_{i} < c_{i}$ . As a result, for Analysis Group 1, the cut-off point $c_{1}$ was 1 November 1949, yielding the dummy $D_{1} = 0$ as the comparison group with individuals born between 1 December 1948, and 31 October 1949, and dummy $D_{1} = 1$ as the intervention group with persons born between 1 November 1949, and 30 September 1950. Subsequently, let $Y$ be the dependent variable, where this variable represents the measure overall job satisfaction and subsequently satisfaction with the organisation’s personnel policy. Therefore, we investigated the causal treatment effect of $D$ on $Y$ .

In this study, we estimated the treatment effect in line with Lee and Lemieux (Lee and Lemieux 2010) as $Y =\propto + β_{0} D + ε$

In this equation, the measures overall job satisfaction and satisfaction with the organisation’s personnel policy of each analysis group $i$ were the result of the sum of the constant $\propto$ , the estimated treatment effect Beta $β$ for $D$ , and the usual error $ε$ . We defined this equation as Model 1.

Subsequently, the regression analyses accounted for two additional models in which the estimated treatment effect in Model 1 was corrected for the confounders, which are denoted as the independent variable $X$ .

In Model 2, the three confounders, categorised as personal factors, were additionally added to Model 1. This resulted in the following equation $Y =\propto + β_{0} D + β_{1} X_{1} + β_{2} X_{2} + β_{3} X_{3} + ε$

In this equation, $X_{1}$ was the level of education, $X_{2}$ represented overall health, and $X_{3}$ was gender.

In Model 3, the five confounders, categorised as work-related factors, were additionally added to Model 2. This resulted in the following equation $Y =\propto + β_{0} D + β_{1} X_{1} + β_{2} X_{2} + β_{3} X_{3} + β_{4} X_{4} + β_{5} X_{5} + β_{6} X_{6} + β_{7} X_{7} + β_{8} X_{8} + ε$

In this equation, $X_{4}$ was satisfaction with the wage, $X_{5}$ represented being a supervisor, $X_{6}$ was the organisation size, $X_{7}$ represented the number of contractual working hours, and $X_{8}$ was the level of the job position according to the SBC 2010 classification.

In addition, per included confounder in Model 2 and Model 3, a dummy variable was generated to register whether the original value of the confounder was a missing value. The transformation of missing values to the sample mode and mean is discussed below. Therefore, these dummy variables were additionally added to Model 2 and Model 3, and the regression analyses were additionally corrected for these dummies.

Furthermore, the distance between the date of birth and the cut-off as the first order polynomial was added to Model 4. Likewise, the second and third order polynomial of this difference were added to Model 5 and Model 6, respectively.

Due to insufficient observations for the cohorts 2012 and 2013, we decided to exclude these cohorts from the analyses. We initially intended to analyse the effect of prolonged work participation of one month to 65 years and one month (i.e. cohort 2013) on job satisfaction, compared to retirement at the age of 65 (i.e. cohort 2012). However, because workers are no longer included in the panel once they retire from work, the dataset included insufficient observations for this analysis. Subsequently, the following analyses were performed.

First, we tested whether, for each analysis group, the personal characteristics and confounders are similar for both the intervention and comparison group. For this, Pearson chi-squares were calculated to compare the personal characteristics and confounders of the nominal and ordinal type, and for continuous variables, we calculated the means and performed the t-test on the difference of the mean between the intervention and comparison group.

Second, for each analysis group, OLS regressions were performed to regress satisfaction with the organisation’s personnel policy on the dummy variable indicating an extension of the retirement age. Model 1 only included this dummy variable as the predictor variable. Model 2 added control variables for personal characteristics and additionally controlled for the presence of missing values. Model 3, in addition, controlled for five confounders categorised as work-related factors, and similar to Model 2, also included dummies for missing values. As indicated above, the first, second, and third order polynomial of the distance between the date of birth and the cut-off were added to Model 4, Model 5, and Model 6, respectively.

Third, for each analysis group, ordered probit regressions were performed to predict the ordinal measure overall job satisfaction. Here, the same explanatory variables as in the OLS regressions were used. The results of the ordered probit regression, which has more stringent underlying assumptions, were compared with the results from the explorative OLS regressions.

Fourth, to increase the size of the intervention group and to generate more statistical power, and for explorative purposes, a new dummy variable was created where the observations from the comparison group of Analysis Group 1 were assigned to the comparison group, and the observations from the intervention group of Analysis Group 1 through 5 were assigned to the intervention group. For this new dummy variable, the analyses outlined under step 1 through 3 were repeated. For step 1, only the difference between the intervention and comparison group for the continuous variables was assessed.

Fifth, because the increase of the statutory retirement age between the intervention and comparison group in all five analysis groups was limited and the difference ranged only from one to three months (see Table 2), we additionally investigated the influence of the treatment between the calendar cohorts over a longer period. Therefore, in five analyses, participants in the calendar cohorts 2014, 2015, 2016, 2017, and 2018 were subsequently treated as the comparison group, while in all analyses the calendar cohort 2019 was defined as the intervention group. This dummy variable substituted the independent variable in the analyses described in step 2 and 3 above. Therefore, we additionally investigated the effect of prolonged work participation between the intervention and comparison groups over a period of three to ten months.

Sixth, for each analysis group, a post hoc power analysis was performed to establish whether the respective group had sufficient statistical power to predict the measures overall job satisfaction and job satisfaction with the organisation’s personnel policy. A high statistical power denotes that the sample size was large enough to detect a significant difference in the effect and avoids that the alternative hypothesis is falsely rejected. Therefore, statistical power helps to understand the findings.

III. Results

The sample size of each analysis group ranged from 50 participants in Analysis Group 1 to 95 observations in Analysis Group 5 (see Table 3).

Table 3. Sample size per intervention and comparison group, by analysis group

CSV Display Table

Sample characteristics

For all analysis groups, the participants in each intervention and comparison group had comparable characteristics (see Table 4). No significant differences were observed regarding gender, level of education, overall health, satisfaction with wage, being a supervisor, level of current job, overall job satisfaction, organisation size, and contractual working hours.

Table 4. Population characteristics for intervention and comparison group, by analysis group

CSV Display Table

Furthermore, for four analysis groups, the calculated age of participants on 1 October 2012, was significantly different between the intervention and comparison group (see Table 4). However, given that the intervention and comparison group of Analysis Group 2 had no variability for the calculated age, a t-test could not be performed to examine the statistical difference between these groups.

In addition, across all calendar cohorts, the participants predominantly rated their perceived overall job satisfaction as satisfied or very satisfied (see Table 4).

Also, across all cohorts, the participants were (reasonably) satisfied with the personnel policy of the organisation. This satisfaction was only significantly different in Analysis Group 5, where on average, participants in the comparison group rated the personnel policy as 7.0, and participants in the intervention group rated the personnel policy as 6.0 (see Table 4).

Analyses by group

The regression discontinuity plots indicate a discontinuity for both overall job satisfaction and satisfaction with the organisation’s personnel policy in all analysis groups (see Appendix 2).

For all analysis groups, the increased statutory retirement age had no statistically significant effect on the ordinal measure overall job satisfaction (see Table 5 and Table 6). This relationship was only positive and significant for the predictor Analysis Group 5 in Model 1 (β = 0.546, p = 0.023) and Model 2 (β = 0.501, p = 0.040), but did not remain significant when the relationship was additionally corrected for in the remaining models. In addition to the ordered probit regression, the OLS regression was performed to explore this relationship because of its less stringent underlying statistical assumptions. Both analyses, however, yielded comparable results.

Table 5. Regression coefficients for overall job satisfaction (1: ‘very satisfied’, 2: ‘satisfied’, 3: ‘dissatisfied’, 4: ‘very dissatisfied’) and satisfaction with organisation’s personnel policy (range from 1: lowest to 10: highest), by analysis group

CSV Display Table

Table 6. Regression coefficients for overall job satisfaction (1: ‘very satisfied’, 2: ‘satisfied’, 3: ‘dissatisfied’, 4: ‘very dissatisfied’), by analysis group

CSV Display Table

Furthermore, the increased statutory retirement age only had a statistically significant effect for the predictor Analysis Group 5 with satisfaction with the organisation’s personnel policy (see Table 5). Although this negative relationship became weaker across the three models (Model 1: β = −0.964, p = 0.003; Model 2: β = −0.908, p = 0.006; Model 3: β = −0.839, p = 0.012), the relationship was no longer significant in the remaining models.

Analyses by combined groups

To increase the size of the intervention group and to increase statistical power, a new predictor dummy variable was generated. Participants in the comparison group of Analysis Group 1 were assigned to the new comparison group, and participants in the intervention group of Analysis Group 1 through 5 were aggregated and assigned to the new intervention group (see Table 2).

The new comparison (n = 19) and intervention (n = 190) group had similar characteristics for organisation size and contractual working hours (see Appendix 3). Furthermore, the participants in these groups had a significantly different calculated age on 1 October 2012.

The regression discontinuity plots indicate a discontinuity for both overall job satisfaction and satisfaction with the organisation’s personnel policy in the combined analysis group (see Appendix 4).

However, the effect size of the new predictor dummy variable and the ordinal measure overall job satisfaction was no longer statistically significant in any model (see Appendix 5 and Appendix 6). Both the ordered probit and OLS regression yielded comparable results.

Likewise, the coefficient of the regression for satisfaction with the organisation’s personnel policy was not statistically significant in any model (see Appendix 5).

Analyses by dummy groups

To further examine the effect of the increased statutory retirement age on a longer period, five new predictor dummy variables were constructed. The calendar cohorts 2014, 2015, 2016, 2017, and 2018 were subsequently treated as the comparison group, while in all analyses, the calendar cohort 2019 was defined as the intervention group.

However, there were no statistically significant relationships across all models for overall job satisfaction and satisfaction with the organisation’s personnel policy (see Appendix 8 and Appendix 9). Indeed, some analyses that included dummy groups yielded significant relationships in some models, but these significant relationships were not stable across models.

Post hoc power analysis

For each initial analysis group (see Table 2), a post hoc power analysis was performed to establish that the group had sufficient statistical power to predict the measures overall job satisfaction and satisfaction with the organisation’s personnel policy (see Table 7). A high statistical power denotes that the sample size is large enough to detect a significant effect and avoids that the alternative hypothesis is falsely rejected.

Table 7. Post hoc power for overall job satisfaction (1: ‘very satisfied’, 2: ‘satisfied’, 3: ‘dissatisfied’, 4: ‘very dissatisfied’) and satisfaction with organisation’s personnel policy (range from 1: lowest to 10: highest), by analysis group

CSV Display Table

For the measure overall job satisfaction, only Analysis Groups 4 and 5 had good statistical power. For the measure satisfaction with the organisation’s personnel policy, good statistical power was only observed for Analysis Group 5, while Analysis Group 4 had sufficient statistical power.

IV. Discussion

The policy reform in the Netherlands to gradually increase the statutory retirement age since 2013 was, and remains to be, highly criticised and has resulted in widespread political and public commotion (Boschman and Van Alphen 2018). Nevertheless, a second law was proposed in 2014 to accelerate the earlier policy reform and, in addition, to increase the statutory retirement age further to age 67 in 2021 (Rijksoverheid 2014; Preesman 2019). Consequently, some political parties initiated campaigns to restore the retirement age to 65 years, and the social, political, and academic support to continue the reform of the pension system remains very low (Boschman and Van Alphen 2018; PlusOnline 2016).

This study resulted from the arguments used in the public debate that such policy reform could negatively impact the job satisfaction of older workers (Henkens et al. 2016; MAX Vandaag 2016). Until now, there was no investigation on the causal effect of the increased statutory retirement age on job satisfaction.

We find no evidence that the increase in retirement age, and, therefore, the delayed right to claim the accrued pension benefits, significantly affected the self-perceived overall job satisfaction of older workers. Results further indicate that, overall, the increased statutory retirement age had no significant effect on the older workers’ satisfaction with the organisation’s personnel policy. If individual predictors are considered, an effect was only observed when we include participants in the comparison and intervention group that were forced to prolong their labour force participation until age 65 and nine months, and age 66, respectively (see Table 2). This finding is remarkable because, contrary to the four other groups analysed, the participants in this intervention group were the youngest and would not be affected by the policy reform until 2019, six years after the policy reform took effect in 2013. This suggests that the level of satisfaction may not be related to the retirement age. However, contrary to the preceding cohorts, participants in the 2019 cohort have to continue working longest, namely three months.

Overall, the analyses yielded no statistically significant effects. Likewise, job satisfaction was conceptualised using two measures, namely the self-perceived overall job satisfaction and satisfaction with the organisation’s personnel policy. The results were consistent for both measures. Also, the fact that in none of the five comparisons we made, and for none of the two outcome measures we used, we found a statistically significant effect, adds to the conclusion that the increase of the retirement age had no effect on job satisfaction.

However, an alternative explanation for the absence of such a statistical difference may be considered. Following Broughham and Walsh (Brougham and Walsh 2009), our sample may contain workers with very diverse preferences for retirement. Workers with late retirement intentions may have perceived the prolonged work participation as a reward that allowed them to fulfil their financial need, facilitated their fulfilment for relatedness and identity, and provided opportunities for growth, via job employment (Brougham and Walsh 2009). On the contrary, workers with early retirement intentions may have perceived the increased retirement age as a punishment because they seek fulfilment of their needs outside work (Brougham and Walsh 2009). The increased retirement age may, therefore, lead to increased job satisfaction in some sample groups, while the policy intervention results in decreased job satisfaction in other sample groups (Brougham and Walsh 2009). Since this study did not account for this preference, we cannot identify these preferences for increased retirement in both sub-groups. Consequently, such positive or negative effect could not be detected in this study.

Furthermore, an alternative explanation for the reported non-significant results may be that some workers, in particular those who were farthest away from retirement, were insufficiently aware about the pension reform and its consequences. The pension reform that resulted in working beyond the original retirement age may, therefore, not impact their job satisfaction. However, because no information about the pension reform was collected, we are unable to verify this hypothesis.

To analyse whether the study had sufficient statistical power, a post hoc power analysis was performed for the predictor Analysis Group 1 through 5 on both measures. The results indicated that the statistical power of Analysis Group 1 through 3 was low. For overall job satisfaction, only good statistical power was observed for Analysis Group 4 and 5. Analysis Group 4 had sufficient, and Analysis Group 5 had good statistical power to detect a real effect on satisfaction with the organisation’s personnel policy. Statistically significant effects for both measures were only observed for Analysis Group 5, but these effects were no longer significant when controlling for the difference in days using a first, second, and third order polynomial. Although the statistical power for Analysis Group 4 was good, no significant effect on overall job satisfaction was observed. Therefore, the limited statistical power for Analysis Group 1 through 3, and specifically Analysis Group 4 for the measure satisfaction with the organisation’s personnel policy, indicates that these sample sizes were insufficiently large to detect a real effect on the measures. This observation is consistent with the regression discontinuity plots, which show a clear discontinuity between the control and intervention groups. Our interpretation is that there may indeed exist a discontinuity, but our sample was too small to detect significant relationships.

V. Conclusion

This study estimated the causal effect of the recently increased statutory retirement age in the Netherlands on both overall job satisfaction and satisfaction with the organisation’s personnel policy among older workers. To conclude, no causal effect of the delayed statutory retirement on the self-perceived overall job satisfaction was found. Furthermore, no evidence was found that the increased statutory retirement age affects the self-perceived satisfaction with the organisation’s personnel policy. Therefore, the findings do not provide support for the societal opinion that the policy reform did impact job satisfaction. However, this study included a relatively small sample and several groups had low statistical power. Moreover, the study did not account for differences in preferences for retirement age or considered an extensive set of control variables that determine job satisfaction. To study the phenomenon in detail, these limitations need to be overcome in future studies.

Table 1. Incremental increase and relative change of the statutory retirement age per calendar year

Year ¹	Individuals born between	Retirement age	Relative change ²
2012	1 February 1947–31 December 1947	65	Not applicable
2013	1 January 1948–30 November 1948	65 + 1 month	1 month
2014	1 December 1948–31 October 1949	65 + 2 months	1 month
2015	1 November 1949–30 September 1950	65 + 3 months	1 month
2016	1 October 1950–31 July 1951	65 + 5 months	2 months
2017	1 August 1951–31 May 1952	65 + 7 months	2 months
2018	1 June 1952–31 March 1953	65 + 9 months	2 months
2019	1 April 1953–31 December 1953	66	3 months
Source: Staatsblad (Staatsblad 2012) and Rijksoverheid (Rijksoverheid 2012). ¹ Cohort defined as the calendar year. ² Change is relative to preceding calendar cohort.

Table 2. Incremental increase and relative change of the statutory retirement age per calendar year

Analysis Group	Group	Cohort ¹	Individuals born between ²	Retirement age
1	Comparison	2014	1 December 1948–31 October 1949	65 + 2 months
	Intervention	2015	1 November 1949–30 September 1950	65 + 3 months
2	Comparison	2015	1 November 1949–30 September 1950	65 + 3 months
	Intervention	2016	1 October 1950–31 July 1951	65 + 5 months
3	Comparison	2016	1 October 1950–31 July 1951	65 + 5 months
	Intervention	2017	1 August 1951–31 May 1952	65 + 7 months
4	Comparison	2017	1 August 1951–31 May 1952	65 + 7 months
	Intervention	2018	1 June 1952–31 March 1953	65 + 9 months
5	Comparison	2018	1 June 1952–31 March 1953	65 + 9 months
	Intervention	2019	1 April 1953–31 December 1953	66
Source: Staatsblad (Staatsblad 2012) and Rijksoverheid (Rijksoverheid 2012). ¹ Cohort defined as the calendar year. ² Dates of birth are defined by the pension reform legislation.

Table 3. Sample size per intervention and comparison group, by analysis group

	Analysis Group 1 (n = 50)		Analysis Group 2 (n = 60)		Analysis Group 3 (n = 64)		Analysis Group 4 (n = 93)		Analysis Group 5 (n = 95)
	C	I	C	I	C	I	C	I	C	I
Sample size	19	31	31	29	29	35	35	58	58	37
C, Comparison Group. I, Intervention Group.

Table 4. Population characteristics for intervention and comparison group, by analysis group

		Analysis Group 1			Analysis Group 2			Analysis Group 3			Analysis Group 4			Analysis Group 5
		C	I		C	I		C	I		C	I		C	I
		N (%)	N (%)	χ²	N (%)	N (%)	χ²	N (%)	N (%)	χ²	N (%)	N (%)	χ²	N (%)	N (%)	χ²
Gender				0.3347			0.6007			1.2148			0.9209			0.7683
	Male	7 (36.8)	14 (45.2)		14 (45.2)	16 (55.2)		16 (55.2)	24 (68.6)		24 (68.6)	34 (58.6)		34 (58.6)	25 (67.6)
	Female	12 (63.2)	17 (54.8)		17 (54.8)	13 (44.8)		13 (44.8)	11 (31.4)		11 (31.4)	24 (41.4)		24 (41.4)	12 (32.4)
Level of education				0.7610			0.2476			4.1899			1.9481			1.0071
	Low	4 (21.1)	7 (22.6)		7 (22.6)	6 (20.7)		6 (20.7)	14 (40.0)		14 (40.0)	20 (34.5)		20 (34.5)	12 (32.4)
	Intermediary	6 (31.6)	13 (41.9)		13 (41.9)	14 (48.3)		14 (48.3)	9 (25.7)		9 (25.7)	23 (39.7)		23 (39.7)	12 (32.4)
	High	9 (47.4)	11 (35.5)		11 (35.5)	9 (31.0)		9 (31.0)	12 (34.3)		12 (34.3)	15 (25.9)		15 (25.9)	13 (35.1)
Overall health				0.1460			0.0243			2.2237			2.5657			0.7472
	Good	15 (79.0)	23 (74.2)		23 (74.2)	21 (72.4)		21 (72.4)	19 (54.3)		19 (54.3)	41 (70.7)		41 (70.7)	23 (62.2)
	Less than good	4 (21.1)	8 (25.8)		8 (25.8)	8 (27.6)		8 (27.6)	16 (45.7)		16 (45.7)	17 (29.3)		17 (29.3)	14 (37.8)
Satisfaction with wage				0.0901			0.2002			1.6064			1.7749			1.2919
	Relatively high	5 (26.3)	7 (22.6)		7 (22.6)	8 (27.6)		8 (27.6)	15 (42.9)		15 (42.9)	17 (29.3)		17 (29.3)	7 (18.9)
	Relatively low	14 (73.7)	24 (77.4)		24 (77.4)	21 (72.4)		21 (72.4)	20 (57.1)		20 (57.1)	41 (70.7)		41 (70.7)	30 (81.1)
Being a supervisor				0.0901			0.2002			0.6570			1.7918			0.0999
	Yes	5 (26.3)	7 (22.6)		7 (22.6)	8 (27.6)		8 (27.6)	13 (37.1)		13 (37.1)	14 (24.1)		14 (24.1)	10 (27.0)
	No	14 (73.7)	24 (77.4)		24 (77.4)	21 (72.4)		21 (72.4)	22 (62.9)		22 (62.9)	44 (75.9)		44 (75.9)	27 (73.0)
Level of current job (SBC 2010)				0.4259			0.9474			0.3214			0.5734			1.4322
	Low	5 (26.3)	6 (19.4)		6 (19.4)	8 (27.6)		8 (27.6)	8 (22.9)		8 (22.9)	10 (17.2)		10 (17.2)	10 (27.0)
	Intermediary	6 (31.6)	12 (38.7)		12 (38.7)	12 (41.4)		12 (41.4)	14 (40.0)		14 (40.0)	27 (46.6)		27 (46.6)	14 (37.8)
	High	8 (42.1)	13 (41.9)		13 (41.9)	9 (31.0)		9 (31.0)	13 (37.1)		13 (37.1)	21 (36.2)		21 (36.2)	13 (35.1)
Overall job satisfaction				5.6196			2.7970			1.2148			2.2161			5.2042
	Very satisfied	7 (36.8)	9 (29.0)		9 (29.0)	13 (44.8)		13 (44.8)	11 (31.4)		11 (31.4)	27 (46.6)		27 (46.6)	10 (27.0)
	Satisfied	11 (57.9)	16 (51.6)		16 (51.6)	14 (48.3)		14 (48.3)	21 (60.0)		21 (60.0)	26 (44.8)		26 (44.8)	19 (51.4)
	Unsatisfied	0 (0.0)	6 (19.4)		6 (19.4)	2 (6.9)		2 (6.9)	3 (8.6)		3 (8.6)	5 (8.6)		5 (8.6)	8 (21.6)
	Very unsatisfied	1 (5.3)	0 (0.0)		0 (0.0)	0 (0.0)		0 (0.0)	0 (0.0)		0 (0.0)	0 (0.0)		0 (0.0)	0 (0.0)
		Analysis Group 1			Analysis Group 2			Analysis Group 3			Analysis Group 4			Analysis Group 5
		C	I		C	I		C	I		C	I		C	I
		M (SD)	M (SD)	t	M (SD)	M (SD)	t	M (SD)	M (SD)	t	M (SD)	M (SD)	t	M (SD)	M (SD)	t
Age on 1 October 2012		62.8 (0.4)	62.0 (0.0)	12.5984***	62.0 (0.0)	61.0 (0.0)	. ^†	61.0 (0.0)	60.3 (0.5)	8.3806***	60.3 (0.5)	59.4 (0.5)	9.3195***	59.4 (0.5)	58.7 (0.5)	6.8760***
Organisation size		390.5 (901.9)	553.0 (1013.9)	−0.5730	553.0 (1013.9)	308.4 (507.8)	1.1689	308.4 (507.8)	508.9 (896.3)	−1.0700	508.9 (896.3)	447.3 (1394.5)	0.2336	447.3 (1394.5)	438.3 (987.8)	0.0341
Contractual working hours		28.8 (10.7)	29.6 (9.8)	−0.2840	29.6 (9.8)	30.7 (9.5)	−0.4402	30.7 (9.5)	30.7 (10.2)	−0.0048	30.7 (10.2)	30.6 (9.3)	0.0631	30.6 (9.3)	32.1 (9.6)	−0.7652
Satisfaction with the organisation’s personnel policy		6.5 (1.7)	6.0 (2.1)	0.8378	6.0 (2.1)	6.8 (1.5)	−1.6840	6.8 (1.5)	6.6 (1.6)	0.5331	6.6 (1.6)	7.0 (1.3)	−1.3321	7.0 (1.3)	6.0 (1.8)	3.0213**
C, Comparison Group. I, Intervention Group. χ², Pearson Chi-Square. M, Mean. SD, Standard Deviation. t, t statistic on the difference of means (C – I). ^†, Non-variability did not allow the calculation of the statistical difference. * p < 0.05; p < 0.01; * p < 0.001.

Table 5. Regression coefficients for overall job satisfaction (1: ‘very satisfied’, 2: ‘satisfied’, 3: ‘dissatisfied’, 4: ‘very dissatisfied’) and satisfaction with organisation’s personnel policy (range from 1: lowest to 10: highest), by analysis group

		Model 1		Model 2		Model 3		Model 4		Model 5		Model 6
		β	SE	β	SE	β	SE	β	SE	β	SE	β	SE
Overall job satisfaction
	Analysis Group 1 (n = 50)	0.166	0.208	0.144	0.205	0.239	0.238	0.096	0.431	−0.081	0.446	0.349	0.583
	Analysis Group 2 (n = 60)	−0.283	0.171	−0.278	0.173	−0.295	0.192	−0.167	0.397	−0.173	0.401	−0.247	0.484
	Analysis Group 3 (n = 64)	0.151	0.153	0.179	0.155	0.219	0.170	0.149	0.335	0.079	0.347	−0.161	0.436
	Analysis Group 4 (n = 93)	−0.151	0.134	−0.107	0.137	−0.220	0.138	−0.408	0.277	−0.407	0.280	−0.051	0.380
	Analysis Group 5 (n = 95)	0.325*	0.141	0.288*	0.140	0.237	0.138	−0.174	0.278	−0.244	0.306	0.084	0.391
Satisfaction with the organisation’s personnel policy
	Analysis Group 1 (n = 50)	−0.474	0.565	−0.444	0.571	−0.754	0.643	−0.657	1.169	−0.648	1.241	−1.150	1.647
	Analysis Group 2 (n = 60)	0.793	0.471	0.791	0.469	0.520	0.514	1.436	1.053	1.461	1.061	1.700	1.279
	Analysis Group 3 (n = 63)	−0.205	0.384	−0.334	0.402	−0.385	0.413	0.252	0.807	0.645	0.814	0.736	1.030
	Analysis Group 4 (n = 91)	0.403	0.303	0.351	0.311	0.490	0.327	0.709	0.653	0.718	0.661	0.216	0.903
	Analysis Group 5 (n = 94)	−0.964**	0.319	−0.908**	0.323	−0.839*	0.325	0.313	0.653	0.948	0.696	0.208	0.906
Model 1, relationship of independent dummy variable with measure. Model 2, additionally controlled for personal factors: level of education, overall health, gender, and one dummy per confounder to correct for calculated missing values. Model 3, additionally controlled for work-related factors: wage, being a supervisor, organisation size, contractual working hours, level of current job (SBC 2010), and one dummy per confounder to correct for calculated missing values. Model 4, additionally adjusted for difference in days between birth of date and cut-off. Model 5, additionally adjusted for second order polynomial of the difference in days between birth of date and cut-off. Model 6, additionally adjusted for third order polynomial of the difference in days between birth of date and cut-off. β, Beta coefficient. SE, Standard Error. * p < 0.05; p < 0.01; * p < 0.001.

Table 6. Regression coefficients for overall job satisfaction (1: ‘very satisfied’, 2: ‘satisfied’, 3: ‘dissatisfied’, 4: ‘very dissatisfied’), by analysis group

		Model 1		Model 2		Model 3		Model 4		Model 5		Model 6
		β	SE	β	SE	β	SE	β	SE	β	SE	β	SE
Overall job satisfaction
	Analysis Group 1 (n = 50)	0.287	0.330	0.272	0.337	0.447	0.382	0.223	0.668	−0.245	0.727	0.987	0.932
	Analysis Group 2 (n = 60)	−0.487	0.296	−0.494	0.301	−0.548	0.324	−0.344	0.656	−0.359	0.658	−0.493	0.783
	Analysis Group 3 (n = 64)	0.289	0.291	0.356	0.305	0.443	0.331	0.257	0.634	0.097	0.658	−0.394	0.824
	Analysis Group 4 (n = 93)	−0.278	0.245	−0.206	0.250	−0.470	0.280	−0.942	0.560	−0.947	0.562	−0.250	0.755
	Analysis Group 5 (n = 95)	0.546*	0.240	0.501*	0.245	0.454	0.255	−0.360	0.528	−0.494	0.576	0.187	0.754
Model 1, relationship of independent dummy variable with measure. Model 2, additionally controlled for personal factors: level of education, overall health, gender, and one dummy per confounder to correct for calculated missing values. Model 3, additionally controlled for work-related factors: wage, being a supervisor, organisation size, contractual working hours, level of current job (SBC 2010), and one dummy per confounder to correct for calculated missing values. Model 4, additionally adjusted for difference in days between birth of date and cut-off. Model 5, additionally adjusted for second order polynomial of the difference in days between birth of date and cut-off. Model 6, additionally adjusted for third order polynomial of the difference in days between birth of date and cut-off. β, Beta coefficient. SE, Standard Error. * p < 0.05; p < 0.01; * p < 0.001.

Table 7. Post hoc power for overall job satisfaction (1: ‘very satisfied’, 2: ‘satisfied’, 3: ‘dissatisfied’, 4: ‘very dissatisfied’) and satisfaction with organisation’s personnel policy (range from 1: lowest to 10: highest), by analysis group

		Statistical power
Overall job satisfaction
	Analysis Group 1	0.453
	Analysis Group 2	0.427
	Analysis Group 3	0.556
	Analysis Group 4	0.915
	Analysis Group 5	0.967
Satisfaction with the organisation’s personnel policy
	Analysis Group 1	0.481
	Analysis Group 2	0.589
	Analysis Group 3	0.658
	Analysis Group 4	0.755
	Analysis Group 5	0.920

Table 8. Proportion of non-missing and missing (imputed) data (N = 2,372)

Variable	Not missing	Missing
	N (%)	N (%)
Level of education	2,368 (99.8)	4 (0.2)
Overall health	2,372 (100.0)	0 (0.0)
Gender	2,372 (100.0)	0 (0.0)
Satisfaction with wage	2,366 (99.8)	6 (0.3)
Being a supervisor	2,370 (99.9)	2 (0.1)
Organisation size	2,334 (98.4)	38 (1.6)
Contractual working hours	2,257 (95.2)	115 (4.9)
Level of current job (SBC 2010)	2,330 (98.2)	42 (1.8)

Table 9. Means and t-test on the difference of mean between intervention and comparison group

	Group
	C (n = 19)	I (n = 190)
	M (SD)	M (SD)	t
Age on 1 October 2012	62.8 (0.4)	60.1 (1.2)	9.9358***
Organisation size	390.5 (901.9)	453.0 (1060.0)	−0.2478
Contractual working hours	28.8 (10.7)	30.8 (9.6)	−0.8574

C, Comparison Group.

I, Intervention Group.

M, Mean.

SD, Standard Deviation.

t, t statistic on the difference of means (C – I).

* p < 0.05; ** p < 0.01; *** p < 0.001.

Table 10. Regression coefficients for overall job satisfaction (1: ‘very satisfied’, 2: ‘satisfied’, 3: ‘dissatisfied’, 4: ‘very dissatisfied’) and satisfaction with organisation’s personnel policy (range from 1: lowest to 10: highest)

	Model 1		Model 2		Model 3		Model 4		Model 5		Model 6
	β	SE	β	SE	β	SE	β	SE	β	SE	β	SE
Overall job satisfaction (n = 209)	0.021	0.161	−0.015	0.162	−0.011	0.164	−0.023	0.195	0.305	0.259	−0.020	0.333
Satisfaction with the organisation’s personnel policy (n = 207)	0.061	0.399	0.051	0.406	0.042	0.411	−0.037	0.489	−1.141	0.644	−0.214	0.827
Model 1, relationship of independent dummy variable with measure. Model 2, additionally controlled for personal factors: level of education, overall health, gender, and one dummy per confounder to correct for calculated missing values. Model 3, additionally controlled for work-related factors: wage, being a supervisor, organisation size, contractual working hours, level of current job (SBC 2010), and one dummy per confounder to correct for calculated missing values. Model 4, additionally adjusted for difference in days between birth of date and cut-off. Model 5, additionally adjusted for second order polynomial of the difference in days between birth of date and cut-off. Model 6, additionally adjusted for third order polynomial of the difference in days between birth of date and cut-off. β, Beta coefficient. SE, Standard Error. * p < 0.05; p < 0.01; * p < 0.001.

Table 11. Regression coefficients for overall job satisfaction (1: ‘very satisfied’, 2: ‘satisfied’, 3: ‘dissatisfied’, 4: ‘very dissatisfied’)

	Model 1		Model 2		Model 3		Model 4		Model 5		Model 6
	β	SE	β	SE	β	SE	β	SE	β	SE	β	SE
Overall job satisfaction (n = 209)	0.033	0.272	−0.030	0.278	−0.023	0.286	−0.030	0.338	0.526	0.449	−0.065	0.582
Model 1, relationship of independent dummy variable with measure. Model 2, additionally controlled for personal factors: level of education, overall health, gender, and one dummy per confounder to correct for calculated missing values. Model 3, additionally controlled for work-related factors: wage, being a supervisor, organisation size, contractual working hours, level of current job (SBC 2010), and one dummy per confounder to correct for calculated missing values. Model 4, additionally adjusted for difference in days between birth of date and cut-off. Model 5, additionally adjusted for second order polynomial of the difference in days between birth of date and cut-off. Model 6, additionally adjusted for third order polynomial of the difference in days between birth of date and cut-off. β, Beta coefficient. SE, Standard Error. * p < 0.05; p < 0.01; * p < 0.001.

Table 12. Regression coefficients for overall job satisfaction (1: ‘very satisfied’, 2: ‘satisfied’, 3: ‘dissatisfied’, 4: ‘very dissatisfied’) and satisfaction with organisation’s personnel policy (range from 1: lowest to 10: highest), by dummy group

		Model 1		Model 2		Model 3		Model 4		Model 5		Model 6
		β	SE	β	SE	β	SE	β	SE	β	SE	β	SE
Overall job satisfaction
	Dummy Group 1: 2014 and 2019 (n = 56)	0.209	0.202	0.133	0.215	0.136	0.221	−5.041*	2.261	−5.188*	2.328	7.468	14.710
	Dummy Group 2: 2015 and 2019 (n = 68)	0.043	0.171	0.007	0.172	0.057	0.180	−0.104	1.389	−0.780	1.437	−4.127	6.804
	Dummy Group 3: 2016 and 2019 (n = 66)	0.325	0.166	0.308	0.163	0.353*	0.174	−0.424	1.100	−0.598	1.105	−0.336	4.377
	Dummy Group 4: 2017 and 2019 (n = 72)	0.175	0.155	0.169	0.150	0.050	0.156	−0.205	0.577	−0.359	0.605	2.300	1.688
	Dummy Group 5: 2018 and 2019 (n = 95)	0.325*	0.141	0.288*	0.140	0.237	0.138	−0.174	0.278	−0.244	0.306	0.084	0.391
Satisfaction with the organisation’s personnel policy
	Dummy Group 1: 2014 and 2019 (n = 56)	−0.447	0.494	−0.356	0.532	−0.314	0.549	11.905*	5.661	10.387	5.702	0.601	36.334
	Dummy Group 2: 2015 and 2019 (n = 68)	0.027	0.473	0.069	0.472	0.030	0.501	4.772	3.810	6.957	3.906	9.878	18.534
	Dummy Group 3: 2016 and 2019 (n = 66)	−0.766	0.413	−0.830*	0.414	−0.818	0.437	5.415*	2.628	5.989*	2.606	1.424	10.301
	Dummy Group 4: 2017 and 2019 (n = 71)	−0.561	0.402	−0.522	0.403	−0.368	0.438	1.627	1.709	2.612	1.761	−2.360	4.970
	Dummy Group 5: 2018 and 2019 (n = 94)	−0.964**	0.319	−0.908**	0.323	−0.839*	0.325	0.313	0.653	0.948	0.696	0.208	0.906
Model 1, relationship of independent dummy variable with measure. Model 2, additionally controlled for personal factors: level of education, overall health, gender, and one dummy per confounder to correct for calculated missing values. Model 3, additionally controlled for work-related factors: wage, being a supervisor, organisation size, contractual working hours, level of current job (SBC 2010), and one dummy per confounder to correct for calculated missing values. Model 4, additionally adjusted for difference in days between birth of date and cut-off. Model 5, additionally adjusted for second order polynomial of the difference in days between birth of date and cut-off. Model 6, additionally adjusted for third order polynomial of the difference in days between birth of date and cut-off. β, Beta coefficient. SE, Standard Error. * p < 0.05; p < 0.01; * p < 0.001.

Table 13. Regression coefficients for overall job satisfaction (1: ‘very satisfied’, 2: ‘satisfied’, 3: ‘dissatisfied’, 4: ‘very dissatisfied’), by dummy group

		Model 1		Model 2		Model 3		Model 4		Model 5		Model 6
		β	SE	β	SE	β	SE	β	SE	β	SE	β	SE
Overall job satisfaction
	Dummy Group 1: 2014 and 2019 (n = 56)	0.347	0.320	0.222	0.340	0.255	0.354	−10.576*	4.212	−12.242*	4.727	35.524	28.078
	Dummy Group 2: 2015 and 2019 (n = 68)	0.069	0.272	0.015	0.286	0.100	0.300	−0.192	2.269	−1.465	2.396	−8.032	11.337
	Dummy Group 3: 2016 and 2019 (n = 66)	0.553	0.285	0.561	0.294	0.680*	0.316	−0.886	1.907	−1.198	1.930	−0.743	7.730
	Dummy Group 4: 2017 and 2019 (n = 72)	0.303	0.267	0.322	0.274	0.158	0.301	−0.005	1.177	−0.331	1.248	6.537	3.584
	Dummy Group 5: 2018 and 2019 (n = 95)	0.546*	0.240	0.501*	0.245	0.454	0.255	−0.360	0.528	−0.494	0.576	0.187	0.754
Model 1, relationship of independent dummy variable with measure. Model 2, additionally controlled for personal factors: level of education, overall health, gender, and one dummy per confounder to correct for calculated missing values. Model 3, additionally controlled for work-related factors: wage, being a supervisor, organisation size, contractual working hours, level of current job (SBC 2010), and one dummy per confounder to correct for calculated missing values. Model 4, additionally adjusted for difference in days between birth of date and cut-off. Model 5, additionally adjusted for second order polynomial of the difference in days between birth of date and cut-off. Model 6, additionally adjusted for third order polynomial of the difference in days between birth of date and cut-off. β, Beta coefficient. SE, Standard Error. * p < 0.05; p < 0.01; * p < 0.001.

Disclosure statement

The authors declare that they have no competing interests.

Availability of data and material

The data that support the findings of this study are available from Microdataservices at Netherlands Statistics.

Authors’ contributions

PP conducted this study. All authors contributed to the text of the manuscript and read and approved the final manuscript.

Code availability

The code used to analyze the data is available from the corresponding author.

Additional information

Funding

This study was funded by the Department of Health Services Research of Maastricht University.

References

Alegre, I., M. Mas-Machuca, and J. Berbegal-Mirabent. 2016. “Antecedents of Employee Job Satisfaction: Do They Matter?” Journal of Business Research 69 (4): 1390–1395. doi:10.1016/j.jbusres.2015.10.113. [Crossref], [Web of Science ®], [Google Scholar]

Bernal, N., and F. Vermeulen. 2014. “The Impact of an Increase in the Legal Retirement Age on the Effective Retirement Age.” De Economist 162 (2): 115–145. doi:10.1007/s10645-014-9226-8. [Crossref], [Web of Science ®], [Google Scholar]

Bonsang, E., and A. Van Soest. 2012. “Satisfaction with Job and Income among Older Individuals across European Countries.” Social Indicators Research 105 (2): 227–254. doi:10.1007/s11205-011-9879-5. [Crossref], [Web of Science ®], [Google Scholar]

Boschman, O., and F. Van Alphen 2018. “Dutch Parliament: Government Must Slow down Retirement Age Increases.” https://www.ipe.com/countries/netherlands/dutch-parliament-government-must-slow-down-retirement-age-increases/10028306.article [Google Scholar]

Brougham, R. R., and D. A. Walsh. 2009. “Early and Late Retirement Exits.” The International Journal of Aging and Human Development 69 (4): 267–286. doi:10.2190/AG.69.4.b. [Crossref], [PubMed], [Web of Science ®], [Google Scholar]

Cheung, F., and A. M. S. Wu. 2013. “Older Workers’ Successful Aging and Intention to Stay.” Journal of Managerial Psychology 28 (6): 645–660. doi:10.1108/JPM-09-2011-0062. [Crossref], [Web of Science ®], [Google Scholar]

Coppola, M., and C. B. Wilke. 2014. “At What Age Do You Expect to Retire? Retirement Expectations and Increases in the Statutory Retirement Age.” Fiscal Studies 35 (2): 165–188. doi:10.1111/j.1475-5890.2014.12027.x. [Crossref], [Web of Science ®], [Google Scholar]

Davies, E. M. M., B. I. J. M. Van der Heijden, and M. Flynn. 2017. “Job Satisfaction, Retirement Attitude and Intended Retirement Age: A Conditional Process Analysis across Workers’ Level of Household Income.” Frontiers in Psychology 8: 891. doi:10.3389/fpsyg.2017.00891. [Crossref], [PubMed], [Web of Science ®], [Google Scholar]

De Wind, A., S. Van der Pas, and B. M. Blatter. 2016. “Van Der Beek AJ. A Life Course Perspective on Working beyond Retirement—results from A Longitudinal Study in the Netherlands.” BMC Public Health 16 (1): 499. doi:10.1186/s12889-016-3174-y. [Crossref], [PubMed], [Google Scholar]

Domonkos, S. 2015. “Promoting A Higher Retirement Age: A Prospect-theoretical Approach.” International Journal of Social Welfare 24 (2): 133–144. doi:10.1111/ijsw.12101. [Crossref], [Web of Science ®], [Google Scholar]

Fabra, M. E., and C. Camisón. 2009. “Direct and Indirect Effects of Education on Job Satisfaction: A Structural Equation Model for the Spanish Case.” Econ Educ Rev 28 (5): 600–610. [Crossref], [Web of Science ®], [Google Scholar]

Foote, D. A., and T. L. P. Tang. 2008. “Job Satisfaction and Organizational Citizenship Behavior (OCB): Does Team Commitment Make a Difference in Self‐directed Teams?” Management Decision 46 (6): 933–947. doi:10.1108/00251740810882680. [Crossref], [Web of Science ®], [Google Scholar]

Georgellis, Y., and T. Lange. 2007. “Participation in Continuous, On-the-job Training and the Impact on Job Satisfaction: Longitudinal Evidence from the German Labour Market.” The International Journal of Human Resource Management 18 (6): 969–985. doi:10.1080/09585190701321112. [Taylor & Francis Online], [Web of Science ®], [Google Scholar]

Hackman, J. R., and G. R. Oldham. 1975. “Development of the Job Diagnostic Survey.” J Appl Psychol 60 (2): 159–170. [Crossref], [Web of Science ®], [Google Scholar]

Henkens, K., H. Van Solinge, M. Damman, and E. Dingemans. 2016. “Langer Doorwerken Valt Nog Niet Mee [Working Longer Is Not Easy].” Demos 32 (2): 1–4. [Google Scholar]

Herzberg, F. 1974. “Motivation-hygiene Profiles: Pinpointing What Ails the Organization.” Organizational Dynamics 3 (2): 18–29. doi:10.1016/0090-2616(74)90007-2. [Crossref], [Web of Science ®], [Google Scholar]

Herzberg, F., B. Mausner, and B. Snyderman. 1959. The Motivation to Work. New York: John Wiley. [Google Scholar]

House, R. J., and L. A. Wigdor. 1967. “Herzberg’s Dual-Factor Theory of Job Satisfaction and Motivation: A Review of the Evidence and A Criticism.” Personnel Psychology 20 (4): 369–390. doi:10.1111/j.1744-6570.1967.tb02440.x. [Crossref], [Web of Science ®], [Google Scholar]

Huysse-Gaytandjieva, A., W. Groot, and M. Pavlova. 2013. “A New Perspective on Job Lock.” Social Indicators Research 112 (3): 587–610. doi:10.1007/s11205-012-0072-2. [Crossref], [PubMed], [Web of Science ®], [Google Scholar]

Jensen JM, Opland RA, Ryan AM. 2010. “Psychological Contracts and Counterproductive Work Behaviors: Employee Responses to Transactional and Relational Breach.” Journal of Bus Psychology 25 (4): 555–568. [Crossref], [Web of Science ®], [Google Scholar]

Judge, T. A., C. J. Thoresen, J. E. Bono, and G. K. Patton. 2001. “The Job Satisfaction–Job Performance Relationship: A Qualitative and Quantitative Review.” Psychological Bulletin 127 (3): 376–407. doi:10.1037/0033-2909.127.3.376. [Crossref], [PubMed], [Web of Science ®], [Google Scholar]

Kosloski, K., D. Ekerdt, and S. DeViney. 2001. “The Role of Job-Related Rewards in Retirement Planning.” The Journals of Gerontology: Series B 56 (3): P160–P9. doi:10.1093/geronb/56.3.P160. [Crossref], [Google Scholar]

Latornell, J. 2007. “The Relationship between the ‘Psychological Contract’ and ‘Employment Relations’.” Journal of Industrial Relations 49 (2): 277–286. doi:10.1177/0022185607074922. [Crossref], [Web of Science ®], [Google Scholar]

Lee, D. S., and T. Lemieux. 2010. “Regression Discontinuity Designs in Economics.” Journal of Economic Literature 48 (2): 281–355. doi:10.1257/jel.48.2.281. [Crossref], [Web of Science ®], [Google Scholar]

Leppel, K., E. Brucker, and J. Cochran. 2012. “The Importance of Job Training to Job Satisfaction of Older Workers.” Journal of Aging & Social Policy 24 (1): 62–76. doi:10.1080/08959420.2012.629136. [Taylor & Francis Online], [Web of Science ®], [Google Scholar]

Linz, S. J., and A. Semykina. 2012. “What Makes Workers Happy? Anticipated Rewards and Job Satisfaction.” Industrial Relations: A Journal of Economy and Society 51 (4): 811–844. doi:10.1111/j.1468-232X.2012.00702.x. [Crossref], [Web of Science ®], [Google Scholar]

Lüthen, H. 2016. “Rates of Return and Early Retirement Disincentives: Evidence from a German Pension Reform.” German Economic Review 17 (2): 206–233. doi:10.1111/geer.12070. [Crossref], [Web of Science ®], [Google Scholar]

Lydon, R., and A. Chevalier. 2002. Estimates of the Effect of Wages on Job Satisfaction. London: Centre for Economic Performance. [Google Scholar]

Maas, I. L., S. Nolte, O. B. Walter, T. Berger, M. Hautzinger, F. Hohagen, W. Lutz, et al. 2017. “The Regression Discontinuity Design Showed to Be a Valid Alternative to a Randomized Controlled Trial for Estimating Treatment Effects”. Journal of Clinical Epidemiology 82: 94–102. doi:10.1016/j.jclinepi.2016.11.008. [Crossref], [PubMed], [Web of Science ®], [Google Scholar]

MacDonald, B.-J., K. D. Moore, H. Chen, and R. L. Brown. 2011. “The Canadian National Retirement Risk Index: Employing Statistics Canada’s LifePaths to Measure the Financial Security of Future Canadian Seniors.” Canadian Public Policy 37: S73–S94. [Crossref], [PubMed], [Google Scholar]

MAX Vandaag. 2016. “60-plusser Boos over Verhoging Pensioenleeftijd [People over the Age of 60 Angry about Raising Retirement Age].” https://www.maxvandaag.nl/sessies/themas/geld-werk-recht/60-plusser-boos-over-verhoging-pensioenleeftijd/ [Google Scholar]

Menon, R. 2017. “Towards a Higher Retirement Age.” https://www.ipe.com/analysis/analysis/towards-a-higher-retirement-age/10020430.article [Google Scholar]

Meyer, J. P., D. J. Stanley, L. Herscovitch, and L. Topolnytsky. 2002. “Affective, Continuance, and Normative Commitment to the Organization: A Meta-analysis of Antecedents, Correlates, and Consequences.” Journal of Vocational Behavior 61 (1): 20–52. doi:10.1006/jvbe.2001.1842. [Crossref], [Web of Science ®], [Google Scholar]

Montizaan, R., F. Cörvers, and A. De Grip. 2010. “The Effects of Pension Rights and Retirement Age on Training Participation: Evidence from a Natural Experiment.” Labour Economics 17 (1): 240–247. doi:10.1016/j.labeco.2009.10.004. [Crossref], [Web of Science ®], [Google Scholar]

Ng, E. S. W., and A. Law. 2014. “Keeping Up! Older Workers’ Adaptation in the Workplace after Age 55.” Can J Aging 33 (1): 1–14. [Crossref], [PubMed], [Web of Science ®], [Google Scholar]

OECD. 2006. Live Longer, Work Longer. Paris: OECD Publishing. [Crossref], [Google Scholar]

Patterson, M., P. Warr, and M. West. 2004. “Organizational Climate and Company Productivity: The Role of Employee Affect and Employee Level.” Journal of Occupational and Organizational Psychology 77 (2): 193–216. doi:10.1348/096317904774202144. [Crossref], [Web of Science ®], [Google Scholar]

PlusOnline. 2016. “Ontevredenheid over Pensioenakkoord [Dissatisfaction with Pension Agreement].” https://www.plusonline.nl/pensioen/ontevredenheid-over-pensioenakkoord [Google Scholar]

Preesman, L. 2019. “Dutch Government to Press Ahead with Plans for Pensions Reform.” https://www.ipe.com/countries/netherlands/dutch-government-to-press-ahead-with-plans-for-pensions-reform/10029317.article [Google Scholar]

Researchcentrum voor Onderwijs en Arbeidsmarkt (ROA). 2009. De Arbeidsmarkt Naar Opleiding En Beroep Tot 2014 [The Labor Market by Education and Occupation until 2014]. Maastricht: Researchcentrum voor Onderwijs en Arbeidsmarkt. [Google Scholar]

Reynolds, F., A. Farrow, and A. Blank. 2012. ““Otherwise It Would Be Nothing but Cruises”: Exploring the Subjective Benefits of Working beyond 65.” International Journal of Ageing and Later Life 7 (1): 79–106. doi:10.3384/ijal.1652-8670.127179. [Crossref], [Google Scholar]

Rijksoverheid. 2012. “AOW-leeftijd Stapsgewijs Omhoog Naar 66 Jaar in 2019 En 67 Jaar in 2023 [Age in General Old Age Pensions Act Increases Incrementally to 66 Years in 2019 and 67 Years in 2023].” https://www.rijksoverheid.nl/actueel/nieuws/2012/05/25/verhoging-aow [Google Scholar]

Rijksoverheid. 2014. “Wetsvoorstel Versnelde Verhoging AOW-leeftijd Naar De Tweede Kamer [Law Proposal Accelerated Increase in State Pension Age].” https://www.rijksoverheid.nl/actueel/nieuws/2014/11/17/wetsvoorstel-versnelde-verhoging-aow-leeftijd-naar-de-tweede-kamer [Google Scholar]

Roelen, C. A. M., P. C. Koopmans, and J. W. Groothoff. 2008. “Which Work Factors Determine Job Satisfaction?” Work 30 (4): 433–439. [PubMed], [Google Scholar]

Rousseau, D. M. 1998. “The ‘Problem’ of the Psychological Contract Considered.” J Organ Behav 19 (S1): 665–671. [Crossref], [Web of Science ®], [Google Scholar]

Schalk, R., and R. E. Roe. 2007. “Towards a Dynamic Model of the Psychological Contract.” Journal for the Theory of Social Behaviour 37 (2): 167–182. doi:10.1111/j.1468-5914.2007.00330.x. [Crossref], [Web of Science ®], [Google Scholar]

Sewdas, R., A. De Wind, V. D. Z. LGL, W. E. Van der Borg, R. Steenbeek, A. J. Van der Beek, C. R. L. Boot, et al. 2017. “Why Older Workers Work beyond the Retirement Age: A Qualitative Study.” BMC Public Health 17 (1): 672. doi:10.1186/s12889-017-4675-z. [Crossref], [PubMed], [Google Scholar]

Sociaal en Cultureel Planbureau. 2014. Toelichting Arbeidsaanbodpanel 1985-2014 [Explanation Arbeidsaanbodpanel 1985-2014]. The Hague: Sociaal en Cultureel Planbureau. [Google Scholar]

Sociaal en Cultureel Planbureau (SCP). 2016. “Arbeidsaanbodpanel 1985 T/m 2014.” DANS, editor. [Google Scholar]

Staatsblad. 2012. “Wet Verhoging AOW- En Pensioenrichtleeftijd [Law on Increasing the Age of the General Old Age Pensions Act and the Pension Target Age].” https://zoek.officielebekendmakingen.nl/stb-2012-328.html [Google Scholar]

Tåhlin, M. 2007. “Skills and Wages in European Labour Markets: Structure and Change.” In Employment Regimes and the Quality of Work, edited by D. Gallie, 35–76. Oxford: Oxford University Press. [Crossref], [Google Scholar]

Templer, A., M. Armstrong‐Stassen, and J. Cattaneo. 2010. “Antecedents of Older Workers’ Motives for Continuing to Work.” Career Development International 15 (5): 479–500. doi:10.1108/13620431011075349. [Crossref], [Web of Science ®], [Google Scholar]

Thistlethwaite, D. L., and D. T. Campbell. 1960. “Regression-discontinuity Analysis: An Alternative to the Ex Post Facto Experiment.” J Educ Psychol 51 (6): 309–317. [Crossref], [Web of Science ®], [Google Scholar]

Van der Heijden, E. C. M., J. H. M. Nelissen, and H. A. A. Verbon. 1997. “Altruism and Fairness in a Public Pension System.” Journal of Economic Behavior & Organization 32 (4): 505–518. doi:10.1016/S0167-2681(97)00008-5. [Crossref], [Web of Science ®], [Google Scholar]

Van Echtelt, P., S. Croezen, J. D. Vlasblom, M. De Voogd-Hamelink, and L. Mattijssen. 2016. Aanbod Van Arbeid 2016: Werken, Zorgen En Leren Op Een Flexibele Arbeidsmarkt [Supply of Labor 2016: Working, Caring, and Learning on a Flexible Labor Market]. The Hague: Sociaal en Cultureel Planbureau. [Google Scholar]

Van Solinge, H., and K. Henkens. 2017. “Older Workers’ Emotional Reactions to Rising Retirement Age: The Case of the Netherlands.” Work, Aging and Retirement 3 (3): 273–283. doi:10.1093/workar/wax010. [Crossref], [Web of Science ®], [Google Scholar]

Vlasblom, J. D., P. Van Echtelt, and M. De Voogd-Hamelink. 2015. Aanbod Van Arbeid 2014: Arbeidsdeelname, Flexibilisering En Duurzame Inzetbaarheid [Supply of Labor 2014: Labor Participation, Flexibilization, and Sustainable Employability]. The Hague: Sociaal en Cultureel Planbureau. [Google Scholar]

Weiss, H. M. 2002. “Deconstructing Job Satisfaction: Separating Evaluations, Beliefs and Affective Experiences.” Human Resource Management Review 12 (2): 173–194. doi:10.1016/S1053-4822(02)00045-1. [Crossref], [Google Scholar]

Wheaton, F., and E. M. Crimmins. 2012. “The Demography of Aging and Retirement.” In The Oxford Handbook of Retirement, edited by M. Wang, 22–41. New York: Oxford University Press. [Google Scholar]

Whitman, D. S., D. L. Van Rooy, and C. Viswesvaran. 2010. “Satisfaction, Citizenship Behaviors, and Performance in Work Units: A Meta-Analysis of Collective Construct Relations.” Personnel Psychology 63 (1): 41–81. doi:10.1111/j.1744-6570.2009.01162.x. [Crossref], [Web of Science ®], [Google Scholar]

Wright TA, Bonett DG. 2007. “Job Satisfaction and Psychological Well-Being as Nonadditive Predictors of Workplace Turnover.” J Manage 33 (2): 141–160. [Web of Science ®], [Google Scholar]

Wynaendts, A. 2012. “The Aegon Readiness Survey 2012: The Changing Face of Retirement.” https://www.aegon.com/contentassets/d9914f4e7f274bb3850c4c582f234162/aegon-retirement-readiness-survey-2012.pdf [Google Scholar]

Appendices

Appendix 1

Table 8. Proportion of non-missing and missing (imputed) data (N = 2,372)

CSV Display Table

Appendix 2

Display full size

Figure 2. Regression discontinuity plot for overall job satisfaction in Analysis Group 2

Display full size

Figure 3. Regression discontinuity plot for overall job satisfaction in Analysis Group 3

Display full size

Figure 4. Regression discontinuity plot for overall job satisfaction in Analysis Group 4

Display full size

Figure 5. Regression discontinuity plot for overall job satisfaction in Analysis Group 5

Display full size

Figure 6. Regression discontinuity plot for satisfaction with personnel policy in Analysis Group 1

Display full size

Figure 7. Regression discontinuity plot for satisfaction with personnel policy in Analysis Group 2

Display full size

Figure 8. Regression discontinuity plot for satisfaction with personnel policy in Analysis Group 3

Display full size

Figure 9. Regression discontinuity plot for satisfaction with personnel policy in Analysis Group 4

Display full size

Figure 10. Regression discontinuity plot for satisfaction with personnel policy in Analysis Group 5

Display full size

Appendix 3

Table 9. Means and t-test on the difference of mean between intervention and comparison group

CSV Display Table

Appendix 4

Figure 11. Regression discontinuity plot for overall job satisfaction in combined Analysis Group

Display full size

Figure 12. Regression discontinuity plot for satisfaction with personnel policy in combined Analysis Group

Display full size

Appendix 5

Table 10. Regression coefficients for overall job satisfaction (1: ‘very satisfied’, 2: ‘satisfied’, 3: ‘dissatisfied’, 4: ‘very dissatisfied’) and satisfaction with organisation’s personnel policy (range from 1: lowest to 10: highest)

CSV Display Table

Appendix 6

Table 11. Regression coefficients for overall job satisfaction (1: ‘very satisfied’, 2: ‘satisfied’, 3: ‘dissatisfied’, 4: ‘very dissatisfied’)

CSV Display Table

Appendix 7

Figure 13. Regression discontinuity plot for overall job satisfaction in Dummy Group 1

Display full size

Figure 14. Regression discontinuity plot for overall job satisfaction in Dummy Group 2

Display full size

Figure 15. Regression discontinuity plot for overall job satisfaction in Dummy Group 3

Display full size

Figure 16. Regression discontinuity plot for overall job satisfaction in Dummy Group 4

Display full size

Figure 17. Regression discontinuity plot for overall job satisfaction in Dummy Group 5

Display full size

Figure 18. Regression discontinuity plot for satisfaction with personnel policy in Dummy Group 1

Display full size

Figure 19. Regression discontinuity plot for satisfaction with personnel policy in Dummy Group 2

Display full size

Figure 20. Regression discontinuity plot for satisfaction with personnel policy in Dummy Group 3

Display full size

Figure 21. Regression discontinuity plot for satisfaction with personnel policy in Dummy Group 4

Display full size

Figure 22. Regression discontinuity plot for satisfaction with personnel policy in Dummy Group 5

Display full size

Appendix 8

Table 12. Regression coefficients for overall job satisfaction (1: ‘very satisfied’, 2: ‘satisfied’, 3: ‘dissatisfied’, 4: ‘very dissatisfied’) and satisfaction with organisation’s personnel policy (range from 1: lowest to 10: highest), by dummy group

CSV Display Table

Appendix 9

Table 13. Regression coefficients for overall job satisfaction (1: ‘very satisfied’, 2: ‘satisfied’, 3: ‘dissatisfied’, 4: ‘very dissatisfied’), by dummy group

CSV Display Table

Applied Economics

The causal influence of increasing the statutory retirement age on job satisfaction among older workers in the Netherlands

Research Article

The causal influence of increasing the statutory retirement age on job satisfaction among older workers in the Netherlands

ABSTRACTFormulae display:?Mathematical formulae have been encoded as MathML and are displayed in this HTML version using MathJax in order to improve their display. Uncheck the box to turn MathJax off. This feature requires Javascript. Click on a formula to zoom.

I. Introduction

Published online:

Table 1. Incremental increase and relative change of the statutory retirement age per calendar year

Job satisfaction

Pension reform in the Netherlands

Relationship between increasing the statutory retirement age and job satisfaction

II. Methods

Published online:

Table 2. Incremental increase and relative change of the statutory retirement age per calendar year

III. Results

Published online:

Table 3. Sample size per intervention and comparison group, by analysis group

Sample characteristics

Published online:

Table 4. Population characteristics for intervention and comparison group, by analysis group

Analyses by group

Published online:

Table 5. Regression coefficients for overall job satisfaction (1: ‘very satisfied’, 2: ‘satisfied’, 3: ‘dissatisfied’, 4: ‘very dissatisfied’) and satisfaction with organisation’s personnel policy (range from 1: lowest to 10: highest), by analysis group

Published online:

Table 6. Regression coefficients for overall job satisfaction (1: ‘very satisfied’, 2: ‘satisfied’, 3: ‘dissatisfied’, 4: ‘very dissatisfied’), by analysis group

Analyses by combined groups

Analyses by dummy groups

Post hoc power analysis

Published online:

Table 7. Post hoc power for overall job satisfaction (1: ‘very satisfied’, 2: ‘satisfied’, 3: ‘dissatisfied’, 4: ‘very dissatisfied’) and satisfaction with organisation’s personnel policy (range from 1: lowest to 10: highest), by analysis group

IV. Discussion

V. Conclusion

Disclosure statement

Availability of data and material

Authors’ contributions

Code availability

Additional information

Funding

References

Published online:

Table 8. Proportion of non-missing and missing (imputed) data (N = 2,372)

Published online:

Published online:

Published online:

Published online:

Published online:

Published online:

Published online:

Published online:

Published online:

Published online:

Published online:

Table 9. Means and t-test on the difference of mean between intervention and comparison group

Published online:

Published online:

Published online:

Table 10. Regression coefficients for overall job satisfaction (1: ‘very satisfied’, 2: ‘satisfied’, 3: ‘dissatisfied’, 4: ‘very dissatisfied’) and satisfaction with organisation’s personnel policy (range from 1: lowest to 10: highest)

Published online:

Table 11. Regression coefficients for overall job satisfaction (1: ‘very satisfied’, 2: ‘satisfied’, 3: ‘dissatisfied’, 4: ‘very dissatisfied’)

Published online:

Published online:

Published online:

Published online:

Published online:

Published online:

Published online:

Published online:

Published online:

Published online:

Published online:

Table 12. Regression coefficients for overall job satisfaction (1: ‘very satisfied’, 2: ‘satisfied’, 3: ‘dissatisfied’, 4: ‘very dissatisfied’) and satisfaction with organisation’s personnel policy (range from 1: lowest to 10: highest), by dummy group

Published online:

Table 13. Regression coefficients for overall job satisfaction (1: ‘very satisfied’, 2: ‘satisfied’, 3: ‘dissatisfied’, 4: ‘very dissatisfied’), by dummy group

Alternative formats

Related research

Information for

Open access

Opportunities

Help and information

Keep up to date

ABSTRACT
Formulae display:?Mathematical formulae have been encoded as MathML and are displayed in this HTML version using MathJax in order to improve their display. Uncheck the box to turn MathJax off. This feature requires Javascript. Click on a formula to zoom.