The relationship between voluntary employer change and work ability among older workers: investigating the honeymoon-hangover effect

Garthe, Nina; Hasselhorn, Hans-Martin Article The relationship between voluntary employer change and work ability among older workers: investigating the honeymoon-hangover effect Journal for Labour Market Research Provided in Cooperation with: Institute for Employment Research (IAB) Suggested Citation: Garthe, Nina; Hasselhorn, Hans-Martin (2021) : The relationship between voluntary employer change and work ability among older workers: investigating the honeymoonhangover effect, Journal for Labour Market Research, ISSN 2510-5027, Springer, Heidelberg, Vol. 55, Iss. 1, pp. 1-12, https://doi.org/10.1186/s12651-021-00294-0 This Version is available at: https://hdl.handle.net/10419/234254 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ Gartheand Hasselhorn J Labour Market Res (2021) 55:12 https://doi.org/10.1186/s12651-021-00294-0 ORIGINAL ARTICLE The relationship betweenvoluntary employer change andwork ability amongolder workers: investigating thehoneymoon-hangover effect Nina Garthe* and Hans Martin Hasselhorn Abstract The aim of the study was to investigate the effect of voluntary employer changes on self-reported work ability among older workers in Germany and whether a honeymoon-hangover effect (HHE) exists here. In research on job satisfaction, three typical periods around a voluntary employer change characterize a HHE: a deterioration in the old job (deterioration), an initial increase in the new job (honeymoon) and a subsequent decline over time (hangover). Whether a HHE exists in respect to work ability following a voluntary employer change remained open. The analyses are based on data from the first three waves of the lidA study (2011, 2014, 2018), a representative cohort study of older employees in Germany born in 1959 or 1965. Data from 2502 workers who participated in all three study waves was analyzed. Fixed-effects regression analyses including lag and lead variables were conducted. A deterioration, honeymoon and hangover period were found. Work ability increased substantially following the voluntary employer change. Our study shows that voluntary employer changes have the potential to maintain work ability at higher working age, but not to increase the work ability in the long-term perspective. However, despite the existence of a hangover period, the positive overall effect of the voluntary change should not be underestimated. Keywords: Turnover, Older workers, Work ability, Honeymoon-hangover effect, Cohort study, Fixed-effects regression © The Author(s) 2021. This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. 1 Introduction At times of worldwide extended working life policies, maintaining work ability at higher working age has received increased public and policy attention (Nilsson etal. 2011). Work ability can be defined as the result of the fit between the individual’s resources and his or her work demands (Tuomi etal. 1997) and may answer the question ‘How good is the worker at present, in the near future, and how able is he or she to do his or her work with respect to work demands, health, and mental resources?’ (Tuomi etal. 1991). Therefore, work ability depends on the actual work situation (Frieling and Kotzab 2014) and is not limited to the individual worker’s resources such as health and functioning (Ebener and Hasselhorn 2019). For over 30 years, work ability has been assessed worldwide with the work ability index (WAI), which has shown to predict various employmentrelated outcomes such as work motivation (Feißel etal. 2018), long-term sick leave, early exit from work and disability (Ebener and Hasselhorn 2019). Today, the WAI is used globally in occupational health practice to assess and maintain the workers´ work ability (Bundesanstalt für Arbeitsschutz und Arbeitsmedizin 2013). Strategies to sustain the fit between older workers and their work are needed as work ability often decreases with age (de Wind etal. 2015; Oakman etal. 2018). One strategy for older workers might be a voluntary change Open Access Journal for Labour Market Research *Correspondence: [email protected] Department of Occupational Health Science, School of Mechanical Engineering and Safety Engineering, University of Wuppertal, Gaußstraße 20, 42119 Wuppertal, Germany 12 Page 2 of 12 N.Garthe , H.M.Hasselhorn of employer, which enables the worker to leave unsuitable workplaces and thereby actively adapt unfavorable working conditions. The change bears the potential to attain suitable working conditions and to increase the ability to work longer at higher working age (Wilke etal. 2019). The large body of research on voluntary employer changes mainly focuses on determinants such as job satisfaction, performance, health, leadership quality (see meta-analysis by Rubenstein etal. 2018) and also work ability (e.g. Rongen et al. 2014). The focus on determinants—in contrast to outcomes—was often due to limitations of the data, which did allow researchers to track employees up to the change, but not subsequently. Thus, there is still little research on the consequences of employer changes. The few previous studies confirm that voluntary employer changes have a positive effect on mental health (Liljegren and Ekberg 2009) and job satisfaction (Chadi and Hetschko 2014) and is associated with improved working conditions such as increased job security, reduced working hours (Carless and Arnup 2011), better salary, better possibilities for development and more appropriate work tasks (Grund 2009). To our knowledge, there is no scientific report about the effect on the workers’ work ability. In some studies, it was observed that the positive effects found following the employer change faded over time (Boswell et al. 2005, 2009, Chadi and Hetschko 2014). In theory, this phenomenon is called the honeymoon-hangover effect (HHE) (Boswell et al. 2005). Boswell etal. (2005) first described and examined the honeymoon-hangover effect in voluntary employer changes with respect to job satisfaction. According to the authors, three periods describe the honeymoon-hango- ver effect: (1) In the first period, which may be called deterioration, employees experience a decline in job satisfaction in the old job. This job dissatisfaction precedes turnover. (2) After turnover, job satisfaction strongly increases. This is referred to as the honeymoon period. The increase is explained by organizations creating an overly positive picture of the job for new hires, as well as the employee’s tendency to portray the new organization in a positive light due to high expectations. Unfavorable information about the new job is suppressed. (3) The initially high job satisfaction with the new job declines over time to an individual ‘normal’ level, which is called the hangover period. The longer tenure with the new job brings along increased knowledge of the organization and employees begin to recognize the less attractive aspects of the job; a subsequent disappointment and disillusionment. The honeymoon-hangover effect was confirmed for job satisfaction in three different studies. Boswell etal. (2005) investigated the HHE by measuring job satisfaction in one-year intervals and found the three periods of the HHE in three to four subsequent years. In 2009, Boswell etal. conducted a similar study with four measurement points (day 1, 3months, 6months, 1year). Within this year, they identified the periods honeymoon and hangover for job satisfaction. A third study by Chadi and Hetschko (2014) distinguished between employer changes due to quitting on own initiative, mutual agreement, dismissal and plant closure. They found a strong HHE for employees, who changed on own initiative and a slight HHE for employees, who changed due to mutual agreement. No HHE was found among employees, who changed due to dismissal or plant closure. Furthermore, Clark etal. (2008) found that life satisfaction significantly decreased before and increased after layoffs, but this increase was not longlasting. However, voluntary employer changes were not investigated. Inspired by the studies on the HHE, we aimed to investigate the effect of voluntary employer changes on the work ability of older workers in Germany and whether HHE exist here. However, HHE may not develop in parallel for all possible outcomes (Roe 2008). Oakman et al. (2018), who investigated pathways of work ability over 6 years, state that sustained changes in work are required to change work ability and that these changes in work take time to make an impact. Thus, while an employer change may affect the course of job satisfaction immediately, this is followed by a readaptation after a shorter period of time. We suspect that for work ability, a longer time period is required for re-adaptation as the workers have to adapt to the new work situation, taking into account their health, competencies and values (Tuomi etal. 1997). Now, the German lidA Cohort Study provides the opportunity to investigate the HHE for work ability among older workers in a large representative sample. In line with the theory on the honeymoon-hangover effect, we expect that the work ability of the employer changers is deteriorating in the old job and is therefore lower than the work ability after the change. Therefore, we postulate the following hypothesis: Hypothesis 1 The work ability in the old job is deteriorating before a voluntary employer change (deterioration). Page 3 of 12 12 The relationship betweenvoluntary employer change andwork ability amongolder workers:… Further, we expect an increase of the work ability following the employer change due to the new, probably more appropriate working conditions and the positive expectations in the new job: Hypothesis 2 The work ability at the new job is initially higher than the work ability with the old job (honeymoon). Lastly, we suppose that the higher work ability declines over time up to a level which is similar to the employees who have stayed with their employer, as the voluntary employer changers adapt to the new work situation and negative aspects of the job are also recognized. Hypothesis 3 The initial high work ability in the new job will decline over time (hangover). 2 Methods 2.1 Data andsample The data used for this study derive from the German lidA Cohort Study on Work, Age, Health and Work participation, a representative cohort study of older employees in Germany. Initially, employed people subject to social security contributions (no self-employed or sworn civil servants), born in either 1959 or 1965, are interviewed every three to four years in their homes (computerassisted personal interviewing, CAPI). The analyses are based on data from the first three waves of the study, 2011 (n = 6585), 2014 (n = 4244) and 2018 (n = 3586). In 2018, the participants were 53 and 59years old. A more detailed description of the lidA Cohort Study and its design has been given elsewhere (Hasselhorn etal. 2014; Rauch etal. 2015). For this study, participants were excluded if they were not employed full time, part time or marginally in any of the waves, if they were self-employed and if they had an involuntary change of employer between any of the waves. This balanced sample allows examination of intraindividual changes in work ability over the course of the three study waves. In all, 2502 workers were included who participated in all three study waves. 2.2 Measures 2.2.1 Groups ofvoluntary employer change In wave two (2014) and three (2018), the participants were asked whether they changed employer on their own initiative. If the participants had multiple changes between two waves, only the last change was measured. In the analyses we distinguish four groups: (1) participants, who had no change, either between 2011 and 2014 nor between 2014 and 2018, thus, they stayed in the same job for all three waves (Job A, Job A, Job A; AAA), (2) participants, who had no change between 2011 and 2014 and changed between 2014 and 2018, thus, they had a new job since wave 3 (Job A, Job A, Job B; AAB), (3) participants, who changed between 2011 and 2014 and had no change between 2014 and 2018, thus, they had a new job since wave 2 (Job A, Job B, Job B; ABB) and (4) participants, who changed between 2011 and 2014 and between 2014 and 2018, thus, they had new jobs in wave 2 and also in wave 3 (Job A, Job B, Job C; ABC). 2.2.2 Work ability In each wave, work ability was measured by the second dimension (WAI2) of the Work Ability Index (WAI). The WAI is an established questionnaire to assess work ability in occupational health research. Short measures such as the WAI2, which assesses the work ability in relation to the demands of the job, were recommended for large surveys and shown to be suitable short measures for work ability in occupational health research and employee surveys (Ebener and Hasselhorn 2019). The WAI2 consists of three questions: In two questions the participants were asked to rate their actual work ability with respect to mental and physical demands at work, respectively (very poor/rather poor/moderate/rather good/very good). The answers were weighted depending on the third question, which measures whether the participant was mainly mentally active in the main job, mainly physically active or both equally. The weighting of the WAI2 is described by Hasselhorn and Ebener (2016). The resulting sum score ranges from 2 (no work ability) to 10 (high work ability). 2.2.3 Socio‑demographics Socio-demographic information includes gender (male/ female), year of birth (1959/1965), vocational education (low/medium/high) (based on Jöckel etal. 1998) as time-independent variables and having a partner (yes/no) assessed in each wave. 2.2.4 Work factors The work factors include working hours (full time/part time/marginal employment), mental and physical work (mainly mental/mainly physical/both) and the income level (up to 1500 Euro/1500 to 3000 Euro/3000 Euro and more), each assessed in each wave. 2.2.5 Health Mental and physical health were assessed with the Short Form Health Survey (SF-12) (Ware etal. 1995, Nübling etal. 2006). The component scores range from 0 to 100 with high scores indicating better health. Both SF-12 12 Page 4 of 12 N.Garthe , H.M.Hasselhorn scales were found to have acceptable psychometric properties and validity (Ware etal. 1996). 2.2.6 Statistical analyses The analysis consists of two steps, the description and the regression analysis. In the description, the four groups of voluntary employer change described above were compared in terms of socio-demographics, work factors, health and work ability across the three study waves. For work ability, the group means were displayed along with confidence intervals for each wave. This allows depicting the course of work ability for each group over time. The regression analysis is conducted as a fixed effects regression analyses including lag and lead variables for employer changes in order to investigate the individual changes of work ability before and after a voluntary employer change. To examine the individual effect of voluntary employer changes, only participants who reported a change in one or more of the waves were included in the regression analyses. With the fixed effects transformation, the individual mean value for work ability over the three waves is subtracted from each single work ability score for each participant. Through this transformation, the individual relationship between the values of each participant remains the same, but potential level differences between the participants are eliminated. Therefore, the fixed effects regression analyses allow to investigate individual work ability changes and unobserved individual heterogeneity, i.e. level differences between study participants, is removed from the work ability data. The honeymoon-hangover effect is examined by including lag or lead dummy variables for voluntary employer changes, respectively. The three study waves allow to integrate two lag variables, which indicate whether a participant has a new job since one or two waves (examining the honeymoon and hangover period, models 1 and 2). Similarly, two lead variables could be integrated, which indicate whether a participant will have a new job in one or two waves (examining the deterioration period, models 3 and 4). In Models 2 and 4 control variables are added. Sensitivity analyses were conducted by performing separate analyses for men and women and for participants born in 1959 and 1965. All statistical analyses were performed using SPSS version 26.0. 3 Results 3.1 Description Table 1 shows the socio-demographics, work factors, health and work ability in the total sample and across the change groups. The majority of the participants stayed with their employer over the seven years (89.6%; AAA ). 9.2% of the participants changed once (AAB, ABB) and 1.2% changed twice (ABC). In some aspects, these three change groups differed from the group of stayers. Women and the younger cohort born in 1965 were overrepresented in all change groups. Participants with medium educational level were overrepresented in groups AAB and ABB and participants with low educational level were overrepresented in group ABC. In all change groups, participants more often had a partner. Participants who changed once were less likely to work full-time in each of the waves. Only in wave one, those marginally employed were overrepresented among the three change groups. In wave three, participants working full-time were overrepresented in the ABC group. Furthermore, participants, who work mainly physically, were overrepresented among all change groups. After changing, these participants more often did both, physical and mental work. More participants with a low income changed employer voluntarily. Mental health increased after a voluntary change. In relation to the stayers, changers reported worse mental health before a change and better physical health afterwards. Figure1 displays the course of work ability for the four groups of voluntary employer change over time. The figure indicates different patterns for each groups’ work ability: the work ability of the stayers (AAA) slightly deteriorated over time and the work ability of the group, who changed twice (ABC), considerably improved after each change. The groups, who changed once (AAB, ABB), had reverse patterns. The work ability of group ABB initially improved following the change and then deteriorated considerably while staying with the new employer. The work ability of group AAB slightly deteriorated while staying with the old employer and improved following the change. Overall, these patterns indicate a honeymoon-hangover effect, as the work ability of the group AAB deteriorated before the change (deterioration), the work ability of all change groups (AAB, ABB, ABC) improved after the change (honeymoon) and the work ability of the group ABB deteriorated while staying with the new employer (hangover). This supports hypotheses 1, 2 and 3. 3.2 Regression analyses The regression analysis provides further insight into the honeymoon-hangover effect, while controlling for socio-demographic information and work factors. The fixed-effects transformation of the outcome work ability allows prediction of the individual changes of work ability, as positive values do not indicate high work ability, but higher work ability in relation to another observation of the same individual. In all models, only Page 5 of 12 12 The relationship betweenvoluntary employer change andwork ability amongolder workers:… Table 1 Sample socio-demographics, work factors, health and work ability in the total sample and across change groups Groups of voluntary employer change (jobs in 2011, 2014, 2018) Total sample Job A, Job A, Job A (AAA) Job A, Job A, Job B (AAB) Job A, Job B, Job B (ABB) Job A, Job B, Job C (ABC) n (%) N = 2502 (100.0) n = 2242 (89.6) n = 139 (5.5) n = 92 (3.7) n = 29 (1.2) % Mean (SD) % Mean (SD) % Mean (SD) % Mean (SD) % Mean (SD) Socio-demographic factors Gender n = 2502 χ2(3) = 11.98, p = 0.007 Male 45.3 46.5 33.8 38.0 34.5 Female 54.7 53.5 66.2 62.0 65.5 Year of birth n = 2502 χ2(3) = 19.79, p = 0.000 1959 45.6 47.1 34.5 30.4 31.0 1965 54.4 52.9 65.5 69.6 69.0 Vocational education n = 2488 χ2(6) = 5.31, p = 0.504 High 23.1 23.6 19.0 16.3 20.7 Medium 57.2 56.9 61.3 62.0 51.7 Low 19.7 19.5 19.7 21.7 27.6 Partner (2011) n = 2493 χ2(3) = 3.89, p = 0.273 Yes 88.2 87.8 92.8 89.1 a No 11.8 12.2 7.2 10.9 a Partner (2014) n = 2499 χ2(3) = 6.07, p = 0.108 Yes 88.1 87.6 93.5 91.3 a No 11.9 12.4 6.5 8.7 a Partner (2018) n = 2480 χ2(3) = 3.36, p = 0.339 Yes 86.9 86.5 91.4 89.1 89.7 No 13.1 13.5 8.6 10.9 10.3 Work factors Working hours (2011) n = 2502 χ2(6) = 43.07, p = 0.000 Full time 67.5 69.0 52.5 53.3 69.0 Part time 27.5 26.8 33.8 38.0 20.7 Marginal employment 5.0 4.2 13.7 8.7 10.3 Working hours (2014) n = 2502 χ2(6) = 39.62, p = 0.000 Full time 66.9 68.3 51.8 55.4 62.1 Part time 29.3 28.2 36.7 a a Marginal employment 3.9 3.4 11.5 a a Working hours (2018) n = 2502 χ2(6) = 26.05, p = 0.000 Full time 67.5 68.9 51.1 58.7 72.4 12 Page 6 of 12 N.Garthe , H.M.Hasselhorn Table 1 (continued) Groups of voluntary employer change (jobs in 2011, 2014, 2018) Total sample Job A, Job A, Job A (AAA) Job A, Job A, Job B (AAB) Job A, Job B, Job B (ABB) Job A, Job B, Job C (ABC) n (%) N = 2502 (100.0) n = 2242 (89.6) n = 139 (5.5) n = 92 (3.7) n = 29 (1.2) % Mean (SD) % Mean (SD) % Mean (SD) % Mean (SD) % Mean (SD) Part time 29.4 28.0 a a a Marginal employment 3.0 3.1 a a a Mental and physical work (2011) n = 2502 χ2(6) = 17.36, p = 0.008 Mainly mental 50.0 50.8 41.7 42.4 48.3 Mainly physical 9.9 9.1 15.1 18.5 17.2 Both 40.1 40.1 43.2 39.1 34.5 Mental and physical work (2014) n = 2502 χ2(6) = 16.48, p = 0.011 Mainly mental 51.0 52.3 40.3 40.2 34.5 Mainly physical 9.9 9.5 14.4 10.9 13.8 Both 39.2 38.2 45.3 48.9 51.7 Mental and physical work (2018) n = 2502 χ2(6) = 13.98, p = 0.030 Mainly mental 51.0 52.1 41.0 44.6 a Mainly physical 7.4 7.3 7.9 9.8 a Both 41.6 40.5 51.1 45.7 62.1 Income level (2011) n = 2404 χ2(6) = 50.43, p = 0.000 Up to 1500 Euro 40.1 37.8 63.3 57.8 50.0 1500 to 3000 Euro 47.0 49.1 28.1 27.8 32.1 3000 Euro and more 12.9 13.1 8.6 14.4 17.9 Income level (2014) n = 2459 χ2(6) = 56.32, p = 0.000 Up to 1500 Euro 34.7 32.4 59.0 51.6 44.8 1500 to 3000 Euro 49.7 51.7 30.6 31.9 34.5 3000 Euro and more 15.7 15.9 10.4 16.5 20.7 Income level (2018) n = 2438 χ2(6) = 46.85, p = 0.000 Up to 1500 Euro 27.1 25.1 48.9 40.7 37.9 1500 to 3000 Euro 52.4 53.8 38.3 42.9 44.8 3000 Euro and more 20.5 21.1 12.8 16.5 17.2 Health n = 2495 F(3, 2491) = 0.766, p = 0.513, partial η2 = 0.00 Mental healthb (2011) 52.1 (9.5) 52.2 (9.4) 52.6 (10.0) 50.7 (10.7) 51.9 (9.1) Page 7 of 12 12 The relationship betweenvoluntary employer change andwork ability amongolder workers:… Table 1 (continued) Groups of voluntary employer change (jobs in 2011, 2014, 2018) Total sample Job A, Job A, Job A (AAA) Job A, Job A, Job B (AAB) Job A, Job B, Job B (ABB) Job A, Job B, Job C (ABC) n (%) N = 2502 (100.0) n = 2242 (89.6) n = 139 (5.5) n = 92 (3.7) n = 29 (1.2) % Mean (SD) % Mean (SD) % Mean (SD) % Mean (SD) % Mean (SD) Mental healthb (2014) n = 2490 F(3, 2486) = 3.170, p = 0.023, partial η2 = 0.00 50.6 (10.0) 50.7 (9.9) 48.5 (11.4) 51.7 (9.4) 53.4 (7.7) Mental healthb (2018) n = 2495 F(3, 2491) = 1.777, p = 0.149, partial η2 = 0.00 51.7 (9.7) 51.5 (9.7) 53.2 (9.9) 51.9 (9.3) 53.9 (7.5) Physical healthb (2011) n = 2495 F(3, 2491) = 0.632, p = 0.595, partial η2 = 0.00 50.3 (8.8) 50.3 (8.8) 50.8 (8.8) 51.3 (8.0) 51.0 (7.3) Physical healthb (2014) n = 2490 F(3, 2486) = 3.253, p = 0.021, partial η2 = 0.00 50.0 (8.8) 49.8 (8.9) 51.3 (8.5) 50.9 (8.5) 53.7 (7.7) Physical healthb (2018) n = 2495 F(3, 2491) = 2.928, p = 0.033, partial η2 = 0.00 48.2 (9.2) 48.0 (9.2) 49.6 (9.0) 49.9 (8.4) 50.5 (7.2) Work ability Work abilityc (2011) n = 2493 F(3, 2489) = 1.359, p = 0.253, partial η2 = 0.00 8.1 (1.5) 8.1 (1.5) 8.3 (1.4) 8.1 (1.6) 7.7 (1.6) Work abilityc (2014) n = 2494 F(3, 2490) = 5.022, p = 0.002, partial η2 = 0.01 8.0 (1.4) 7.9 (1.4) 7.9 (1.7) 8.5 (1.4) 8.3 (1.1) Work abilityc (2018) n = 2497 F(3, 2493) = 7.727, p = 0.000, partial η2 = 0.01 7.8 (1.5) 7.7 (1.5) 8.2 (1.6) 7.9 (1.4) 8.6 (1.2) SD standard deviation For each socio-demographic and work factor Chi-square tests and for health and work ability one-way ANOVAs are shown for the groups comparisons between the four groups AAA, AAB, ABB and ABC a This data cannot be shown for reasons of data protection b Range from 0 to 100. High scores indicate good health c Range from 2 to 10. High scores indicate high work ability 12 Page 8 of 12 N.Garthe , H.M.Hasselhorn observations of participants reporting a change (groups AAB, ABB; ABC) were included(Table2). Models 1 and 2 include the two lag variables, which allow to investigate the effect of being in a new job since one or two waves on work ability. The two lag variables are dummy variables indicating if a participant was in a new job since one or two waves, respectively. The models show that the work ability in the new job was significantly higher one wave after the change. The lag variable ‘new job since 2 waves’ showed no significant effect, which indicates that the work ability was not significantly higher two waves after the change. This supports hypotheses 2 and 3 and the existence of a honeymoon and hangover period. Models 3 and 4 include the two lead variables, which allow to examine if and how the individual work ability before changing the employer differed from work ability following the change. The two lead variables are dummy variables indicating if a participant will be in a new job in one or two waves, respectively. The models show that work ability in the wave prior to the change was significantly lower than after the change, which is indicated by the negative regression coefficient. The lead variable ‘new job in 2 waves’ showed no significant effect. This result supports hypothesis 1 and the deterioration period, as the work ability one wave before the change was significantly lower than after the change (significant negative regression coefficient), but not two waves before (no significant regression coefficient). The inclusion of control variables in models 2 and 4 did not affect these results. In the sensitivity analyses (not shown), similar patterns as described above were found for men and women, participants born in 1965 and 1959 and when including the observations of all participants (groups AAA, AAB, ABB and ABC) in the regression analyses. However, significant regression coefficients were only found for women and younger workers, which may be due to the low number of cases among male participants and participants born in 1959 reporting a voluntary employer change. In further sensitivity analyses covering not only those who changed job but all participants (n = 2479) and observations (n = 7437), the findings and significances remained largely stable. However, the corrected within R2 was considerably lower which may be explained by a dilution effect due to the—logically—reduced within-panel variation of the lag and lead variables. 4 Discussion In summary, our data indicate that voluntary employer changers are a specific group of employees among older workers. In this study younger and female employees were overrepresented among changers. The higher proportion of female changers may be due to the fact that women are more likely to work in jobs with more opportunities and needs to change employer. For example, in part-time and marginal employment, in jobs with a lower income level and in occupations which more easily allow 7 7.5 8 8.5 9 9.5 00.5 11.5 22 .5 AAAAABABBABC 2011 20142018 Fig. 1 Work ability of voluntary employer changers between 2011 and 2018 with confidence intervals (2011: n = 2493, 2014: n = 2494, 2018: n = 2497). Notes: Possible range from 2 (no work ability) to 10 (maximal work ability). AAA: Job A, Job A, Job A (n = 2234–2237); AAB: Job A in 2011, Job A in 2014, Job B in 2018 (n = 138–139); ABB: Job A in 2011, Job B in 2014, Job B in 2018 (n = 92); ABC: Job A in 2011, Job B in 2014, Job C in 2018 (n = 29)