Expectations of older workers regarding their exit from the labour market and its realization

Pertold, Filip; Federičová, Miroslava Article Expectations of older workers regarding their exit from the labour market and its realization Central European Economic Journal (CEEJ) Provided in Cooperation with: Faculty of Economic Sciences, University of Warsaw Suggested Citation: Pertold, Filip; Federičová, Miroslava (2022) : Expectations of older workers regarding their exit from the labour market and its realization, Central European Economic Journal (CEEJ), ISSN 2543-6821, Sciendo, Warsaw, Vol. 9, Iss. 56, pp. 93-112, https://doi.org/10.2478/ceej-2022-0007 This Version is available at: https://hdl.handle.net/10419/324565 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-nc-nd/4.0/ ISSN: 2543-6821 (online) Journal homepage: http://ceej.wne.uw.edu.pl To cite this article Pertold, F., Federičová, M. (2022). Expectations of older workers regarding their exit from the labour market and its realization. Central European Economic Journal, 9(56), 93-112. DOI: 10.2478/ceej-2022-0007 To link to this article: https://doi.org/10.2478/ceej-2022-0007 Expectations of older workers regarding their exit from the labour market and its realization Filip Pertold, Miroslava Federičová Open Access. © 2022 F. Pertold, M. Federičová, published by Sciendo. This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. Filip Pertold Economics Institute of the Academy of Sciences, Politických vězňů 936/7, 111 21 Praha 1, Czech Republic corresponding author: [email protected] Miroslava Federičová Economics Institute of the Academy of Sciences, Politických vězňů 936/7, 111 21 Praha 1, Czech Republic 1. Introduction One of the greatest challenges of the twenty-first century for the Czech Republic and the whole of Europe is population ageing. The current demographic projections of the Czech Statistical Office (2018) clearly show that the number of people over 65 in the Czech Republic is set to grow in the long run, and by the middle of this century, there will be one million more of them than today.1 This will logically put pressure 1 According to the projections of the Czech Statistical Office (CSO), the number of people over 65 will increase significantly over the next decades. While today there are two million people aged 65 and over, the CSO expects this group to grow to three million within thirty years. At the both on social and health services and on the activation of older workers in the labour market. This study therefore aims to analyse the behaviour of older workers on the labour market, specifically same time, the number of people of working age (15–64) is likely to fall by up to 900,000. Although any projection suffers from a certain degree of uncertainty by definition, the general trend here is clear and is based on the number of births since the 1970s. At the beginning, these are the so-called “Husák’s” children, of whom up to 190,000 were born annually, significantly more than were born in the 1980s and especially in the 1990s. This late twentiethcentury development will then have a significant impact on the demographic situation after 2035, when there will be a significant decline in the labour force and an increase in the number of pensioners. Expectations of older workers regarding their exit from the labour market and its realization Abstract The objective of the paper is to analyse the labour market behaviour of older workers, specifically cross-country differences in expectations regarding the exit from the labour market and subsequent realization. Using longitudinal Survey of Health, Ageing and Retirement in Europe (SHARE) data and econometric analysis, we provide an international comparison of the situation of older workers in the Czech Republic with the other countries of Europe. The data show that although expectations about work activity at the age of 63 are quite similar in the Czech Republic from an international perspective, the work activity realized differs significantly between the Czech Republic and other countries. Our principal finding is that the Czech Republic has a high rate of unexpected retirements compared to all other European countries included in this analysis, even if we control for the socioeconomic background of respondents. The econometric analyses further show that up to about one-third of this difference can be explained by the lower retirement age set by the institutional environment in the Czech Republic, which is anticipated by employees at preretirement age. Conversely, the health status of older workers, and even the different allocation of employees to physically demanding occupations, does not have a significant impact on these cross-country differences in unexpected retirements. Keywords labour supply | expectations | retirement age | SHARE JEL Codes J14, J 26, J32 CEEJ • 9(56) • 2022 • pp. 93-112 • ISSN 2543-6821 • DOI: 10.2478/ceej-2022-0007 95 their expectations of retirement, and provides an international comparison of the Czech Republic with Western European countries. Our ambition is to offer empirical evidence on expectations about future retirement and how these expectations are subsequently fulfilled. We investigate how various factors explain differences across countries in unexpected retirements, specifically in the Czech Republic and eight Western Europe countries: Germany, Sweden, Denmark, Belgium, the Netherlands, France, Spain, and Italy. The results show that the expected age of eligibility for retirement pension is important in explaining cross-country differences. Throughout the study, retirement is understood exclusively as retirement into inactivity.2 Analysing the causes of retirement despite earlier expectations to work can create space and specific advice for a more appropriate setting of the pension system in order to increase the labour market participation of the older population. The foundation of this analysis is the international longitudinal Survey of Health, Ageing and Retirement in Europe (SHARE), which allow us to follow individuals before and after retirement. At the same time, these data enable us to focus on various factors that could explain the differences in unexpected retirements, such as the subjective health of workers, the different allocation of workers to physically demanding occupations, or simply the shock of becoming eligible for a retirement pension. There is a general consensus in the foreign literature that the expectations of preretirees regarding their entry into retirement are fairly accurate, with respect to both timing and the expected income (Chan & Stevens, 2004). In addition, a number of international studies show that the determinants of expectations about retirement entry are consistent with each individual’s situation, including, for example, health status or type of employment (Benítez-Silva & Dwyer, 2005). The transition from retirement back to work, a phenomenon much more common in the USA than in Europe, has been further explored by Maestas (2010), who has pointed to a strong relationship between expectations to work after retirement age and the transition from retirement to re-employment (unretirement). Overall, however, these studies mainly map the situation in the USA; in European countries, the effect of expectations on the transition to retirement has been virtually unexplored. In the Czech literature, 2 Situations in which the respondent works at least one hour per week and receives a pension are evaluated as work activity. only one study has addressed the topic of retirement and the expectations of older workers. Vidovičová (2013) compares a group of people aged 55–64 who are close to retirement with a group of retired people. The results show that respondents before entering retirement imagine retirement as a joyful event, with this idea reflecting dissatisfaction in their current job. The two groups also differ in their perception and evaluation of the benefits of retired life, indicating that entering retirement is not wholly perceived as a positive step in hindsight. Unfortunately, this study is not longitudinal in nature and therefore it cannot be ruled out whether the measurements and conclusions are influenced by time or cohort effects. Concerning the retirement decision itself, the literature looks at several factors here. Empirical studies show that poor health is an important element in the decision to retire (McGarry, 2004; van den Berg, Elders & Burdorf, 2010; Jones, Rice & Roberts, 2010; Gupta & Larsen, 2010). Poor working conditions, and the associated job dissatisfaction, appear to be another important factor in the decision to leave the labour market (Siegrist et al., 2007; Schnalzenberger et al., 2008; Bockerman & Ilmakunnas, 2020). The study by Riedel, Hofer, & Wögerbauer (2015) also shows that intellectual workers, defined on the basis of the International Standard Classification of Occupations codes 1 and 23 (ISCO), plan their retirement later than workers in other groups. The statutory retirement age also appears to be another important factor for retirement. Studies looking at the effects of the shift in the statutory retirement age in Germany suggest that this shift has led to a delay in retirement (Engels, Geyer & Haan, 2017) as well as a shift in expectations about retirement (Coppola & Wilke, 2014). In this study, we therefore further seek to explain the differences in unexpected retirement between the Czech Republic and Western European countries based on the abovementioned factors. The basic research question of our study is whether and how older workers’ expectations are reflected in their actual behaviour when leaving the labour market. Specifically, we test whether becoming eligible for retirement can explain higher rates of unexpected labour market exits in the Czech Republic versus other European countries. In a regression analysis, we also test whether the observed crosscountry differences can be explained by subjective 3 It is an occupational classification in which codes 1 and 2 belong to a group of legislators and managers and a group of specialists. CEEJ • 9(56) • 2022 • pp. 93-112 • ISSN 2543-6821 • DOI: 10.2478/ceej-2022-0007 96 health status, different segregation of job types across countries, or replacement rates. This analysis does not aim to capture all possible factors that may enter into the different realisations of expectations across Europe. However, it is the first analysis of its kind to look in detail at the factors impacting retirement, particularly for those workers who previously expected to work longer than is actually the case. In the context of the current literature, our research shows that although expectations about work activity at the age of 63 are quite similar in the Czech Republic from an international perspective, the work activity realized differs significantly between the Czech Republic and other countries. Our principal finding is that the Czech Republic has a high rate of unexpected retirements compared to all other European countries included in this analysis. The econometric analysis suggests that up to about one-third of this difference can be explained by the lower retirement age set by the institutional environment in the Czech Republic, which is anticipated by employees at preretirement age. Conversely, the health status of older workers, and even the different allocation of employees to physically demanding occupations, does not have a significant impact on these differences in unexpected retirements. The same can be said for replacement rates, which are higher in the Czech Republic than, for example, in Germany, but only for the low-income group. Our evidence thus shows that the retirement age is crucial for the labour supply of older workers in the labour market. Our findings are important for public policy measures focused on prolonging labour market activity, such as the extension of the retirement age (i.e., the age of eligibility for early retirement), or the incentives of older workers in highly skilled occupations to remain in the labour market after they become eligible for old-age pensions. At the same time, this study offers basic statistical facts on the situation of older workers in the Czech Republic in comparison with Western European countries. 2. Data and Methodology This study is based on the SHARE project.4 It is an international multidimensional longitudinal microdata database with more than 140,000 individuals and their partners (approximately 380,000 4 See Börsch-Supan et al. (2013) for methodological details. interviews in total) aged 50+ in 27 European countries and Israel. The research focuses on demographics, family and social ties; education; health and health care; work and retirement; income, consumption, assets; family assistance and financial transfers; housing; activities; expectations; life history; quality of life, and numerous other topics. Importantly, the survey also includes measures of physical and mental health. The result is a unique data set providing information on the state, history, and development of Czech and European society. The first wave of data collection began in 2005 and has been repeated every two years since. Since 2007, i.e., since the second wave, the Czech Republic has participated in all waves of data collection on a panel sample of approximately 5,600 respondents. In this study, we mainly use waves from 2007, 2011, 2013, 2015, and 20175, which allows us to compare health, ageing, and retirement trends in different European countries over a sufficiently long period. The great advantage of these data is the possibility to follow an individual over time and therefore his/her decisions over a certain period of time. In our analysis, we focus on nine European countries, namely, the Czech Republic, Germany, Sweden, Denmark, Spain, Italy, France, Belgium and the Netherlands. The selection of these countries is based on the fact that they are included in all previous waves and therefore provide a chance to fully exploit the longitudinal structure of the data. At the same time, these countries represent various social and economic systems in Europe. The key variables in this study are workers’ expectations about their transition to retirement and the subsequent realization or non-realization of these expectations. In SHARE, we use questions aimed at working respondents, who are asked about their likelihood of working full-time at age 63 for these purposes. The average age of a respondent answering this question is 55. Since this is an expected probability, the response values here range from 0 to 100. Our sample thus consists of those respondents who appear at least in two SHARE waves: in the first wave answering the question about their expectations to work at age 63, and in the second wave those who 5 The analysis does not include data from the third wave in 2009. This is the so-called SHARE LIFE wave, which asks respondents for detailed information about their history, either from their childhood or their work history. At the same time, it does not include all questions from previous waves and therefore does not offer sufficient information for our analysis. CEEJ • 9(56) • 2022 • pp. 93-112 • ISSN 2543-6821 • DOI: 10.2478/ceej-2022-0007 97 have reached at least the age 63. In order to increase the number of observations in our sample, we use any available pair of waves that could provide this information about the respondent.6 In total, around 6,300 individuals were surveyed about their likelihood of working full-time at age 63 in SHARE in selected countries and, in the same time, are observed around age 63.7 To better interpret the results, we then divided this expected probability of working at the age of 63 into responses with high and low, or zero expectations. We then defined high expectations as those responses that reported values for the probability of working greater than 75%. In other words, if a respondent answered that he or she would be likely to work at age 63 with a probability greater than 75%, we label this response as a high expectation of working.8 Due to the longitudinal nature of the data, we are able to further track individuals over time and compare their stated expectations at the age of 50–61 with their actual behaviour at approximately the age of 63. For statistical reasons, we have chosen here an interval of 62–64 years. This allows us to investigate the realization, or possible non-realization, of high expectations to work in the Czech Republic and to compare them subsequently with the situation in other European countries. For economic activity status, we have chosen a definition similar to that used in the Labour Force Survey (LFS). Thus, on the basis of a person’s predominant activity, it is possible to define persons in and out of the labour force. Unemployed people are considered as a part of the labour force.9 6 Using different pair of waves for respondents in our sample, and hence different points in time when they were interviewed, we have to account for possible changes in the conditions and parameters of retirement during the period studied. We do this by including year dummy variables into the regression analysis. 7 The decline in observations in our sample is due to the fact that some respondents have not yet reached the age of 63 during the survey, and therefore we do not know whether they will decide to retire or continue working at this age. At the same time, there are some respondents answering the question about expectations to work at age 63, but do not appear in the next waves. This fact additionally reduces the number of observations in our sample. 8 The analysis of work expectations is dealt with in the first part of the Results section. For descriptive statistics, see Table 1 and Table A1 in the Appendix. 9 It should be noted here that we have also tried alternative definitions of inactivity, for example, the number of hours worked or, alternatively, whether or not a person receives a pension. However, these alternative 2.1. Basic Control Variables In order to compare how older workers respond to their health status, and specifically how it affects their decision to leave the labour market, we next focus on the declared limitations of older people in their daily approaches did not have a significant impact on the results. Table 1. Descriptive Statistics of the Sample of SHARE Data Czechia Germany All other countries Average expectations of working at the age of 63 (%) 48.17 45.39 43.77 (37.87) (39.97) (38.52) Share of respondents with high expectations of working1 (%) 35.14 34.44 31.96 (46.00) (47.16) (45.27) Expected statutory retirement age 61.2 (2.21) 64 (2.11) 63.4 (2.89) Share of employed 0.730 0.807 0.774 (0.347) (0.330) (0.335) Average age 56.98 56.79 56.98 (3.347) (3.524) (3.561) White-collar occupation20.595 0.724 0.668 (0.486) (0.442) (0.466) Primary education 0.0700 0.00469 0.107 (0.255) (0.0684) (0.309) Lower secondary 0.299 0.0606 0.155 (0.458) (0.239) (0.361) Higher secondary 0.455 0.537 0.368 (0.498) (0.499) (0.482) Tertiary education 0.176 0.398 0.370 (0.381) (0.490) (0.483) Share of respondents with some health limitation in their daily activities 0.315 0.326 0.236 (0.372) (0.368) (0.339) Observations 2,071 2,344 24,793 Note: Standard deviations are given in parentheses. 1 The high expectations of working are those responses that reported values for the probability of working greater than 75%. 2 The types of occupations with ISCO code 1 to 5. CEEJ • 9(56) • 2022 • pp. 93-112 • ISSN 2543-6821 • DOI: 10.2478/ceej-2022-0007 98 activities. The SHARE survey asks older people if they are restricted due to their health condition in their normal daily activities. Respondents could choose between the answers “Severely restricted”, “Restricted but not severely”, and “No restriction”. To facilitate the interpretation of the results, we aggregated the responses into a binary variable that takes the value of 1 if the respondent has at least some limitation (i.e., “Severe limitation” or “Limitations but not severe”) and the value of 0 if “No limitation”. Furthermore, one of the factors discussed is the age at which people become eligible for the old-age pension. Descriptive statistics show that these expectations of receiving an old-age pension differ by almost three years between Czech Republic and Germany. Thanks to the detailed questionnaire of the SHARE project, we are able to further track various characteristics of individuals, such as their highest educational attainment (based on International Standard Classification of Education codes, ISCED) and type of employment (based on ISCO codes). The information on the highest educational attainment allowed us to define workers as highly skilled (i.e., with a university degree) and low skilled (i.e., with a secondary or primary education). For simplicity, we also divided the types of occupations into those requiring higher qualifications (these are ISCO codes 1 to 5) and those requiring lower qualifications (ISCO codes 6 to 9). Descriptive statistics for all the variables mentioned in the Czech Republic, Germany, and all other countries are shown in Table 1. 3. Methodology The actual empirical analysis of the realization or possible non-realization of the expectation to work at the age of 63 is performed using regression functions. Their aim is to statistically estimate the international differences (“country” variable10) in the probability of work activity conditional on previous high expectations to work, controlling for observable characteristics of workers. In this analysis, we then gradually control for other characteristics of individuals. Doing this, we try to explain the differences in the realization of expectations across the European countries by the following three factors: 10 This variable is represented in the model by dummy variables that represent each country separately. education and type of employment, health status, and the expected retirement age. This association between the realization of expectations and the factors that may influence them is described by the following model11: P(Y=1|VO)=F(country, education, occupation, health, retirementage, ε ), where Y is a binary variable equal to one when the respondent retired and 0 when the respondent is the part of labour force at the age of 63, conditional on the previous high expectations of working (VO=1). Thus, in this model, we explain the probability of not realizing high expectations to work at the age of 63. In all regressions, Sweden figures as the reference group. Thus, only binary variables for the other countries enter the baseline model. The total estimate of these variables therefore represents the difference in non-realization of high expectations between the particular country and Sweden. Subsequent regressions then individually ascertain to what extent the mentioned factors can explain this difference. We admit that the presented model suffers from an omitted variable problem, which may cause bias in the presented estimates. However, our goal is to present descriptive evidence about cross-country differences in unplanned retirements. We take advantage of SHARE data to control for variables that are normally not available on a national level. Variables representing the education and occupation type control in our model for the differences across countries in the proportion of the workforce with lower skills and education. Thus, these are mainly those employed in manual jobs, which are more physically demanding than highly skilled jobs, and thus may play a larger role in the final decision not to work despite previously high expectations. Using the health variable, we then control for cross-country differences in the health status of the population in question, specifically for the subjective perceptions of individuals, the older employed, regarding their health constraints in daily activities. The last factor in the regression is the reported expected age at which respondents will be eligible for an unreduced old-age pension. We therefore take into account cross-country differences in the age at which people become eligible for an old-age pension. 11 The model is described in greater detail in Appendix A1. CEEJ • 9(56) • 2022 • pp. 93-112 • ISSN 2543-6821 • DOI: 10.2478/ceej-2022-0007 99 3.1. The Current Situation of the Economically Active on the Labour Market In this section, we describe in more detail the work activity of the elderly population in the Czech Republic and its development over time. At the same time, we look at the dynamics of the labour market exits for retirement in general, and also in particular, for individual groups of the population on the basis of their health status and the type of employment. Here we compare the situation in the Czech Republic with the situation in the countries of Western Europe, which in this section, for simplicity, is represented by Germany. It is the country closest to the Czech Republic in Western Europe. The ageing of the Czech population is reflected in the shrinking size of the labour force. One of the possible solutions to this development is to increase the labour market participation rate of those population groups with the lowest employment rates. These are mainly women and older workers. By this measure, the situation in the Czech Republic appears to be favourable, or at least moving in the right direction. This is demonstrated by Figure 1, which shows the share of workers in the 55–64 age group based on the LFS. The figure clearly shows the increase in the labour force in this age group in recent years, which is mainly due to the increase in the employment rate of women. It has risen from 22% in 2000 to over 57% in 2018, largely converging with the rates common in Western Europe, and more specifically in Germany, where the female employment rate in the 55–64 age group is 67%. This development is probably due to the gradual increase in the retirement age, especially for women, over the last 10 years. Although not as steep, but still positive, employment trends can also be observed for the male population in the Czech Republic, where it has increased by 22 percentage points over the last 20 years, from 52% to 74%. Compared to Germany, employment of men in the 55–64 age group is therefore at about the same level as in the Czech Republic, which was not the case before. 3.2. Dynamics of Labour Market Exit In this section, we look at the evolution of men’s and women’s move from the labour market to retirement in the Czech Republic and compare this evolution with neighbouring Germany. A more detailed insight into the employment rate and specifically the dynamics of labour market exit for women and men is shown in Figure 2. The horizontal axis shows the age of individuals, and the vertical axis shows the share of employed people for a given age, which is taken as a continuous variable for simplicity’s sake. The results are shown in the figure for the Czech Republic in 2007 and 2017, and for comparison we also report the data for Germany in 2017. Figure 2 (right), showing the employment rate for women by age, clearly indicates that these have increased over time for all age groups up to age 62. After the age of 62, the employment rate for women Figure 1. Employment Rate in the 55–64 Age Group as a Percent of the Population in the Czech Republic Source: Eurostat (LFS, Employment and activity by sex and age) CEEJ • 9(56) • 2022 • pp. 93-112 • ISSN 2543-6821 • DOI: 10.2478/ceej-2022-0007 100 in the Czech Republic is almost zero between 2007 and 2017. Although there has been a significant shift in female employment in this age group in the Czech Republic over the last decade, the Czech Republic is still lagging behind compared to Germany, where around 40% of women aged 63 are still working. Moreover, it is not until around 67 years of age that Germany reaches near-zero female employment. As far as male employment is concerned, an increase in employment can also be observed here between 2007 and 2017 for the 59–63 age group (Figure 2, right). Thus, the employment of men aged 63 has increased from about 15% in 2007 to just over 20% in 2017. The dynamics of men leaving the labour market in neighbouring Germany is also similar to the Czech Republic. In Germany, however, the employment rate of men after the age of 62 declines more slowly than in the Czech Republic, reaching around 43% at the age of 63. While people in the Czech Republic become eligible for retirement earlier on average than in other countries, this may also have a different impact on their decision to retire, despite their earlier expectations to work. In Figure 3 below, we therefore show how respondents’ employment changes with respect to the distance in age from the expected12 retirement pension age in Germany and the Czech Republic. Indeed, on the horizontal axis, time is normalized by the value of the expected age of receipt of the old-age pension, so the value -2 is two years 12 We have no information about when they are actually eligible for a pension. We use the expected retirement age as a proxy for this variable in our analysis. until the respondent expects to receive the old-age pension.13 This confirms that people in Germany and the Czech Republic behave similarly on this scale, i.e., they retire when their current age is the same as the age at which they expected to receive a pension. The main difference between Germany and the Czech Republic therefore remains the age at which people expect to start receiving a retirement pension. In order to compare how older workers respond to their health status, and specifically how it influences their decision to leave the labour market, we further focus on the declared limitations of older people in their daily activities. Looking to the dynamics of labour market exit of those with limitation in daily activities (Figure 4), they follow the overall dynamics of labour exit documented in Figure 2. The drop for women in the Czech Republic is therefore again most likely due to the overall earlier exit of women from the labour market in the Czech Republic compared to women in Germany. However, according to the World Health Organization (WHO), healthy life expectancy is two years higher in Germany than in the Czech 13 A similar analysis, this time for the Czech Republic only, was carried out on the basis of data from the statistical yearbook of the Czech Social Security Administration (CSSA). These show the share of age groups of men in newly granted old-age pensions for 2018. In a simplified way, we could interpret this share as the probability of entering early or normal retirement by age. Again, this confirms that the observed sharp decline in male employment at the age of 63, and hence their exit from the labour market, can therefore be explained by their newly accrued pension entitlement (see Figure A1 in the Appendix). Figure 2. Share of Employed People for a Given Age in Years 2007 and 2017 in the Czech Republic Note. For comparison purposes, there also are data for Germany in 2017. Source: SHARE waves 2, 4, 5, 6, and 7 (own calculations) CEEJ • 9(56) • 2022 • pp. 93-112 • ISSN 2543-6821 • DOI: 10.2478/ceej-2022-0007 107 P01_AG08291, P30_ AG12815, R21_AG025169, Y1-AG-4553-01, IAG_BSR06-11, OGHA_04-064, HHSN271201300071C) and from various national funding sources is gratefully acknowledged (see www. share-project.org). This work was supported by the Ministry of Education, Youth and Sports of the Czech Republic through the project SHARE-CZ+ (CZ.02.1.0 1/0.0/0.0/16_013/0001740). Funding This research was supported by a grant from the Ministry of Education, Youth and Sports SHARE-CZ+ (CZ.02.1.01/0.0/0.0/16_013/0001740). References Benítez-Silva, H., & Dwyer, D. S. (2005). The rationality of retirement expectations and the role of new information. Review of Economics and Statistics, 87(3), 587-92. Bockerman, P., & Ilmakunnas, P. (2020). Do good working conditions make you work longer? Analyzing retirement decisions using linked survey and register data. Journal of the Economics of Aging, 17. https://doi. org/10.1016/j.jeoa.2019.02.001 Börsch-Supan, A. (2020a). Survey of health, ageing and retirement in Europe (SHARE). Wave 1. Version 7.1.0. [Data set]. SHARE-ERIC. https://doi.org/10.6103/ SHARE.w1.710 Börsch-Supan, A. (2020b). Survey of health, ageing and retirement in Europe (SHARE). Wave 2. Version 7.1.0. [Data set]. SHARE-ERIC. https://doi.org/10.6103/ SHARE.w2.710 Börsch-Supan, A. (2020c). Survey of health, ageing and retirement in Europe (SHARE). Wave 4. Version 7.1.0. [Data set]. SHARE-ERIC. https://doi.org/10.6103/ SHARE.w4.710 Börsch-Supan, A. (2020d). Survey of health, ageing and retirement in Europe (SHARE). Wave 5. Version 7.1.0. [Data set]. SHARE-ERIC. https://doi.org/10.6103/ SHARE.w5.710 Börsch-Supan, A. (2020e). Survey of health, ageing and retirement in Europe (SHARE). Wave 6. Version 7.1.0. [Data set]. SHARE-ERIC. https://doi.org/10.6103/ SHARE.w6.710 Börsch-Supan, A. (2020f). Survey of health, ageing and retirement in Europe (SHARE). Wave 7. Version 7.1.0. [Data set]. SHARE-ERIC. https://doi.org/10.6103/ SHARE.w7.710 Börsch-Supan, A., Brandt, M., Hunkler, C., Kneip, T., Korbmacher, J., Malter, F., Schaan, B., Stuck, S., & Zuber, S. (2013). Data resource profile: The survey of health, ageing and retirement in Europe (SHARE). International Journal of Epidemiology, 42(4), 992–1001. https://doi.org/10.1093/ije/dyt088 Chan, S., & Stevens, A. (2004). Do changes in pension incentives affect retirement? A longitudinal study of subjective retirement expectations. Journal of Public Economics, 88(7–8), 1307–33. Coppola, M., & Wilke, C. B. (2014). At what age do you expect to retire? Retirement expectations and increases in the statutory retirement age. Fiscal Studies, 35(2), 165–88. Czech Statistical Office. (2018). Projekce obyvatelstva ČR do roku 2100. [Projection of the population of the Czech Republic until 2100]. Praha: Český statistický úřad. https://www.czso.cz/csu/czso/ projekce-obyvatelstva-ceske-republiky-2018-2100 Engels, B., Geyer, J., & Haan, P. (2017). Pension incentives and early retirement. Labour Economics, 47, 216–31. https://doi.org/10.1016/j.labeco.2017.05.006 Gupta, N. D., & Larsen, M. (2010). The impact of health on individual retirement plans: Self-reported versus diagnostic measures. Health Economics, 19(7), 792–813. https://doi.org/10.1002/hec.1523 Jones, A. M., Rice, N., & Roberts, J. (2010). Sick of work or too sick to work? Evidence on selfreported health shocks and early retirement from the BHPS. Economic Modelling, 27(4), 866–80. https://doi. org/10.1016/j.econmod.2009.10.001 Maestas, N. (2010). Back to work: Expectations and realizations of work after retirement. Journal of Human Resources, 45(3), 718–48. McGarry, K. (2004). Health and retirement: Do changes in health affect retirement expectations? Journal of Human Resources, 39(3), 624–48. OECD. (2019). Pensions at a glance 2019: OECD and G20 indicators. Paris: OECD Publishing. https:// www.oecd-ilibrary.org/social-issues-migration- health/pensions-at-a-glance-2019_b6d3dcfc-en CEEJ • 9(56) • 2022 • pp. 93-112 • ISSN 2543-6821 • DOI: 10.2478/ceej-2022-0007 108 Pertold, F., & Šatava, J. (2018). Obezita v České republice: mezinárodní srovnání s využitím dat z projektu SHARE [Obesity in the Czech Republic: An international comparison using data from the SHARE project]. IDEA study 9, 2018. https://idea.cerge-ei. cz/files/IDEA_Studie_9_2018_Obezita_v_Cesku/ mobile/index.html#p=1) Riedel, M., Hofer, H., & Wögerbauer, B. (2015). Determinants for the transition from work into retirement in Europe. IZA Journal of European Labor Studies, 4(4). Schnalzenberger, M., Schneeweis, N., Winter- Ebmer, R., & Zweimüller, M. (2008). Job Quality and Retirement Decisions. In Börsch-Supan, A., Brugiavini, A., Jürges, H., Kapteyn, A., Mackenbach, J., Siegrist, J., and Weber, G. (Eds.). First results from the Survey of Health, Ageing and Retirement in Europe (2004–2007) (pp. 215–21). Mannheim: Mannheim Research Institute for the Economics of Aging. Siegrist, J., Wahrendorf, M., von dem Knesebeck, O., Jürges, H., & Börsch-Supan, A. (2007). Quality of work, well-being, and intended early retirement of older employees—baseline results from the SHARE Study. European Journal of Public Health, 17(1), 62–8. van den Berg, T., Elders, L.A., & Burdorf, A. (2010). Influence of health and work on early retirement. Journal of Occupational and Environmental Medicine, 52(6), 576–83. Vidovičová, Lucie. (2013). Představy o starobním důchodu a jeho časování: Vliv legislativy, práce a rodiny [Ideas about old-age pension and its timing: The influence of legislation, work and family]. Fórum Sociální Politiky, 7(2), 2–9. CEEJ • 9(56) • 2022 • pp. 93-112 • ISSN 2543-6821 • DOI: 10.2478/ceej-2022-0007 109 Appendix A1: Model Model: The Realization of Expectations to Work Below, we present in greater detail the model of the non-realization of the expectation to work at the age of 63 and its comparison across European countries. Only respondents who indicated that they expect to be working at age 63 enter the model. We use the least squares method to estimate the model. Differences in unrealized expectations of working at the age 63 across European countries are described by the baseline model, 𝑌𝑌𝑌𝑌 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 =𝛼𝛼𝛼𝛼+𝛽𝛽𝛽𝛽𝑖𝑖𝑖𝑖𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖+𝛽𝛽𝛽𝛽𝑖𝑖𝑖𝑖𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖+𝜀𝜀𝜀𝜀𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖, (P1) (P1) where Y is a binary variable equal to one when the respondent retired conditional on his previous high expectations of working for respondent i in country c and at time t; Country is a vector of binary variables equal to one for each country (except for two countries—Sweden and Denmark—which form our reference group), and Year is a vector of binary variables for each year in which the data were collected (i.e., 2007, 2011, 2013, 2015, and 2017). By adding binary variables for each wave of SHARE testing, we control for possible changes in the conditions and parameters of retirement during the period studied. Subsequently, we try to explain the estimated differences from model (P1), as found in the existing literature, based on three factors: education and type of employment, health status, and expected retirement age. We first add education and type of employment to the model (P1), and thus estimate the model, 𝑌𝑌𝑌𝑌 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 =𝛼𝛼𝛼𝛼+𝛽𝛽𝛽𝛽𝑖𝑖𝑖𝑖𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖+𝛽𝛽𝛽𝛽𝑖𝑖𝑖𝑖𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖+𝛾𝛾𝛾𝛾𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 1𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶𝐸𝐸𝐸𝐸𝑌𝑌𝑌𝑌𝐶𝐶𝐶𝐶𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 +𝛾𝛾𝛾𝛾𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 2𝑂𝑂𝑂𝑂𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶𝑂𝑂𝑂𝑂𝑌𝑌𝑌𝑌𝐶𝐶𝐶𝐶𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 +𝜀𝜀𝜀𝜀𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖, (P2) (P2) where Education is the highest level of education attained by the respondent and Occupation is the type of the respondent’s last job (based on ISCO code). In model (P2), we further control for the health status of the respondent, 𝑌𝑌𝑌𝑌 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 =𝛼𝛼𝛼𝛼+𝛽𝛽𝛽𝛽𝑖𝑖𝑖𝑖𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖+𝛽𝛽𝛽𝛽𝑖𝑖𝑖𝑖𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖+𝛾𝛾𝛾𝛾𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 1𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶𝐸𝐸𝐸𝐸𝑌𝑌𝑌𝑌𝐶𝐶𝐶𝐶𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 +𝛾𝛾𝛾𝛾𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 2𝑂𝑂𝑂𝑂𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶𝑂𝑂𝑂𝑂𝑌𝑌𝑌𝑌𝐶𝐶𝐶𝐶𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 + +𝛾𝛾𝛾𝛾𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 3𝐻𝐻𝐻𝐻𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝐻𝐻𝐻𝐻𝐶𝐶𝐶𝐶ℎ𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 +𝜀𝜀𝜀𝜀𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 , (P3) (P3) where Health is a binary variable equal to one for those respondents who experience some health limitation in their daily activities. In the last model, we then control for the expected retirement age of the respondent (i.e., the age of eligibility for an old-age pension), based on the Retirement Age variable, and estimate the regression equation, 𝑌𝑌𝑌𝑌 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 =𝛼𝛼𝛼𝛼+𝛽𝛽𝛽𝛽𝑖𝑖𝑖𝑖𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖+𝛽𝛽𝛽𝛽𝑖𝑖𝑖𝑖𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖+𝛾𝛾𝛾𝛾𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 1𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶𝐸𝐸𝐸𝐸𝑌𝑌𝑌𝑌𝐶𝐶𝐶𝐶𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 +𝛾𝛾𝛾𝛾𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 2𝑂𝑂𝑂𝑂𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶𝑂𝑂𝑂𝑂𝑌𝑌𝑌𝑌𝐶𝐶𝐶𝐶𝐸𝐸𝐸𝐸𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 + +𝛾𝛾𝛾𝛾𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 3𝐻𝐻𝐻𝐻𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝐻𝐻𝐻𝐻𝐶𝐶𝐶𝐶ℎ𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 +𝜀𝜀𝜀𝜀𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖, (P3) . (P3) The results of these models are shown in Table 2. CEEJ • 9(56) • 2022 • pp. 93-112 • ISSN 2543-6821 • DOI: 10.2478/ceej-2022-0007 110 Appendix A2 Figure A1. Share of Age Groups of Men in Newly Granted Old-age Pensions, Year 2018 (Czech Republic) Note: Figure values are expressed for regular and early retirement Source: Own calculations based on the statistical yearbook (CSSA) Table A1. Descriptive Statistics, All Countries Germany Czechia Sweden Denmark Italy Spain France Belgium The Netherlands Average expectations of working at the age of 63 (%) 38.96 45.26 51.93 41.89 41.13 55.44 17.90 22.03 19.61 (41.67) (40.45) (41.00) (41.32) (40.61) (38.69) (31.51) (34.56) (33.85) Share of respondents with high expectations of working (%) 30.46 34.38 44.13 34.08 31 43.65 11.28 15.26 15.22 (46.05) (47.53) (49.68) (47.42) (46.29) (49.64) (31.65) (35.98) (35.96) Share of employed 0.438 0.185 0.667 0.474 0.367 0.504 0.160 0.225 0.280 (0.496) (0.389) (0.471) (0.500) (0.482) (0.500) (0.367) (0.418) (0.450) White-collar occupation 0.734 0.582 0.824 0.797 0.582 0.528 0.701 0.707 0.783 (0.442) (0.494) (0.381) (0.402) (0.494) (0.500) (0.458) (0.455) (0.413) Primary education 00.0770 0.0721 0.0379 0.242 0.354 0.212 0.100 0.0676 (0) (0.267) (0.259) (0.191) (0.428) (0.479) (0.409) (0.301) (0.251) Lower secondary education 0.0761 0.320 0.165 0.0702 0.243 0.268 0.0793 0.219 0.321 (0.265) (0.467) (0.371) (0.256) (0.429) (0.443) (0.270) (0.414) (0.468) Higher secondary education 0.468 0.427 0.296 0.383 0.323 0.172 0.405 0.317 0.266 (0.499) (0.495) (0.457) (0.486) (0.468) (0.377) (0.491) (0.465) (0.442) Tertiary education 0.456 0.177 0.467 0.509 0.192 0.206 0.304 0.364 0.345 (0.498) (0.382) (0.499) (0.500) (0.394) (0.405) (0.460) (0.481) (0.476) Expected retirement age 63.7 (2.08) 61.2 (2.19) 64.2 (2.00) 65.4 (1.26) 62 (3.31) 64 (2.67) 60 (2.43) 62.3 (2.77) 65.05 (1.67) Good health 0.423 0.454 0.559 0.532 0.438 0.420 0.492 0.543 0.754 (0.494) (0.498) (0.497) (0.499) (0.497) (0.494) (0.500) (0.498) (0.431) Observations 788 701 818 898 600 559 807 1055 414 CEEJ • 9(56) • 2022 • pp. 93-112 • ISSN 2543-6821 • DOI: 10.2478/ceej-2022-0007 111 Table A2. Differences in Unplanned Retirement Across Countries Compared to Sweden and Possible Factors That Explain These Differences (Probit Estimation of Main Model). Probit (1) (2) (3) (4) Czech Republic 1.143*** 1.054*** 1.035*** 0.658*** (0.082) (0.084) (0.085) (0.092) Germany 0.302*** 0.341*** 0.316*** 0.259*** (0.084) (0.086) (0.087) (0.088) Denmark 0.036 0.050 0.038 0.119 (0.083) (0.084) (0.084) (0.085) Italy 0.510*** 0.377*** 0.348*** 0.151 (0.096) (0.098) (0.099) (0.102) Spain 0.294*** 0.107 0.071 0.068 (0.092) (0.098) (0.098) (0.100) France 0.746*** 0.691*** 0.673*** 0.347*** (0.106) (0.108) (0.108) (0.115) Belgium 0.478*** 0.505*** 0.495*** 0.414*** (0.094) (0.095) (0.095) (0.097) The Netherlands 0.298** 0.288** 0.335** 0.407*** (0.146) (0.146) (0.147) (0.147) Lower secondary education -0.166*-0.184*-0.195* (0.100) (0.100) (0.102) Higher secondary education -0.235** -0.252*** -0.266*** (0.095) (0.096) (0.097) Tertiary education -0.549*** -0.559*** -0.523*** (0.099) (0.099) (0.101) White-collar occupation -0.103*-0.102*-0.113* (0.058) (0.058) (0.059) Good health -0.214*** -0.204*** (0.048) (0.048) Expected retirement age -0.154*** (0.014) _cons -0.817*** -0.381*** -0.252** 9.653*** (0.057) (0.102) (0.107) (0.935) N3255 3255 3255 3255 R2 Note: Standard errors in parentheses. *p < 0.1;**p < 0.05; ***p < 0.01. CEEJ • 9(56) • 2022 • pp. 93-112 • ISSN 2543-6821 • DOI: 10.2478/ceej-2022-0007 112 Table A3. Differences in Unplanned Retirement Across Countries Compared to Sweden and Possible Factors That Explain These Differences (LPM) (1) (2) (3) (4) Czech Republic 0.421*** 0.383*** 0.374*** 0.245*** (0.028) (0.028) (0.028) (0.030) Germany 0.096*** 0.107*** 0.097*** 0.077*** (0.028) (0.028) (0.028) (0.028) Denmark 0.010 0.013 0.008 0.030 (0.027) (0.026) (0.026) (0.026) Italy 0.173*** 0.123*** 0.112*** 0.053 (0.033) (0.033) (0.033) (0.033) Spain 0.094*** 0.026 0.014 0.013 (0.031) (0.032) (0.032) (0.032) France 0.265*** 0.240*** 0.233*** 0.127*** (0.037) (0.037) (0.037) (0.037) Belgium 0.160*** 0.166*** 0.161*** 0.134*** (0.032) (0.032) (0.032) (0.031) The Netherlands 0.095*0.087*0.102** 0.117** (0.050) (0.049) (0.049) (0.048) Lower secondary education -0.060*-0.066*-0.060* (0.035) (0.035) (0.034) Higher secondary Education -0.086*** -0.092*** -0.085*** (0.033) (0.033) (0.032) Tertiary education -0.189*** -0.191*** -0.166*** (0.034) (0.034) (0.033) White-collar occupation -0.038*-0.038*-0.041** (0.020) (0.020) (0.020) Good health -0.071*** -0.067*** (0.016) (0.016) Expected retirement age -0.049*** (0.004) _cons 0.207*** 0.367*** 0.412*** 3.537*** (0.018) (0.035) (0.036) (0.280) N3255 3255 3255 3255 R20.086 0.107 0.112 0.145 Note: Standard errors in parentheses. *p < 0.1; **p < 0.05; ***p < 0.01.