The effect of psychological incentives and health concerns on labour force participation of old age individuals in OECD countries.
University of Groningen
Faculty of Economics and Business
Course code: EBM877A20
Author: Elpida TZIOUTZIA
Supervisor: Viola ANGELINI
June 8, 2018
Abstract
1. Introduction
Population ageing is one of the main challenges in pension systems and reforms. In particular, high life expectancy and low fertility rates in combination with high early retirement rates weaken the sustainability of societies’ interest. All OECD countries except Turkey and Mexico have an average fertility rate below 1.65 (O’Brien 2010). The life expectancy stands at 76 and 82 for males and females respectively. The ageing population has placed a burden on the government regarding expenditure on social security and health benefits. Consequently, the proportion of the active workforce on which the government relies for tax revenues continues to shrink. Countries around the world meet growing pressure to reform social security arrangements due to these changes in labour force demographics (Manoli and Weber 2016). The situation has led to conflict between the old workers and their younger counterparts who absorb the burden of the former’s subsistence. Older worker labour force participation has, therefore, become a critical policy issue in OECD countries.
In the 1970s early retirement had been proposed and implemented by some countries including Turkey, Hungary, and Poland among others due to the significant decline in employment (O’Brien 2010). This move was aiming to withdraw the older workers from the labour market and place them under their pension contributions to ease the burden of excessive government expenditure. However, the policy was castigated by some as an infringement to the Age Discrimination Acts that are in effect in most OECD countries. For that reason, countries like the United Kingdom, Denmark, United States, and Sweden were initially reluctant to implement the policy. However, early retirement incentives plans had been commonly used in France, Germany, and Italy due to various institutional differences that make their implementation relatively easier. Although the incentives had encouraged many older workers to retire, further research had revealed that they had caused market distortions which reduced labour supply, living standards, and output.
accomplished, many of the OECD countries will sustain a legal retirement age of 65 years (Martin and Whitehouse 2008). Other reforms are related to the offered financial incentives to workers to continue working. For example, many OECD countries have established or increased the rewards or bonuses paid to individuals who retired beyond the statutory retirement age (OECD 2007).
Another element of the pension systems that has been reformed is the calculation of a worker’s earnings that are used for the pension entitlements computation. Finland, Poland, Portugal, Slovakia and Sweden are all in the direction of a lifetime average earnings measure. Many countries are following the same example. Another reform that has also been implemented is the complete alteration of the pension system. Countries with public defined-benefit (DB) system decided to replace it with a defined-contribution (DC) scheme. Hungary, Mexico, Poland, Slovakia and Sweden did the start to initiate compulsory, individual accounts that are privately managed to replace part of the public pension (Whitehouse 2007).
Understanding the motivations behind the early retirement decision is very important. Governments can formulate their pension systems accordingly so they can encourage old age workers to not claim early retirement benefits. Studies so far focus on the pension incentives (Börsch-Supan 2000) or the financial incentives (Blöndal and Scarpetta 1997) for early retirement. A positive correlation exists between these incentives and the decision to retire early, however, the early retirement decision is not based purely on the type of the pension system. A variety of factors can influence an individual’s decision to retire early such as health concerns and psychological incentives. Besides, humans are complex beings and are not driven purely by financial ‘instincts’. This study tries to estimate the effect of these different types of incentives in order to formulate a better understanding of the retirement behaviour.
1.1 Theoretical Basis of Early Retirement Incentives
Theoretically, early retirement incentives are computed through the replacement rate and the change in net pension wealth. The replacement percentage is a ratio of income independent of work and the expected revenue while in employment. For a retirement plan to be effective, the replacement rate must be high enough to guarantee an optimal standard of living even after retirement. If a high replacement rate is available before the statutory retirement age, older workers will have a higher motivation to retire early. However, many governments face a challenge in the calculation of the replacement rate due to the lack of a comprehensive model that takes into account individual labour market characteristics, which might be different from the generic model presented by economic theory.
The generic model is a program used to analyse data structures in more adaptable forms; thus, explaining how an economy functions (Forrester 2003). It provides solutions to specific deficiencies experienced in the conventional data simulations. Through proper communication, the system establishes an agreement between particular elements making them more substantial as well as reducing their inconsistencies. The model formulates different modes of behaviour which are observed in real economies to provide solutions to the differences found in the economic literature (Forrester 2003). It is utilised in multiple industrial countries including the developed and developing nations, especially in the computation of labour force demographics and retirement schemes. However, the system is not as useful since it does not account for the individual labour market characteristics. For instance, older worker participation in the German labour market continued to increase despite the implementation of the early retirement incentive plans. Similarly, studies have revealed that early retirement in Austria reflects more on the workers’ preferences than policy interventions (OECD 2013). Therefore, there is a need for the development of a more comprehensive model since the incentives might not be effective in withdrawing a significant amount of the labour force from the market.
disabled people. The wealth pension theory is, therefore, more feasible in the computation of early retirement incentives.
Moreover, health status plays an important role to the retirement decision. MacGarry (2002) argues that workers in poor health or individuals whose loving people in their social environment are experiencing negative health conditions have the tendency to retire earlier than those in better health. Siegrist et al (2006) show that both poor quality of work and reduced well-being are independently related to the motive of retirement. As measures of well-being they used indicators associated to self-perceived health, depressive symptoms, quality of life, and bodily symptoms.
Brugiavini et al (2008) describe the cross-section and longitudinal relationship between the health status and participation into the labour market in 2004 and 2006. They have concluded that even though the healthier workers have the tendency to work longer, in many countries there is a great portion of retired individuals who are in a good health. More specifically, using the SHARE 2006 the authors found that a large part of the individuals aged 50-64 does not participate in the labour market even though they are in good health. This result can provide the right motivation for pension systems to revise their approach to solve the early retirement trends.
1.2 Influence of Early Retirement on Labour Force Participation
Studies across the OECD countries reveal an inverse relationship between early retirement incentives and the participation of an ageing labour force (OECD 2011). The implementation of early retirement schemes decreases labour force participation among the older workers. However, several other studies have discovered a spurious relationship between the incentives and workforce participation, especially in the USA and Canada. Social security statistics in the Czech Republic showed a decrease in early retirees by half after the early retirement policy was implemented (Kocourek and Pertold 2011). A study in the Austrian market revealed a 12.5 percent fall in employment among older workers after the government introduced two pension modifications which enhanced the early retirement age (ERA) for men by two years (60 to 62) and females by three years (55 to 58) (Staubli and Zweimüller 2013). It was further revealed that the low labour market participation was due to decreasing motivation among the older workers due to weak prospects of finding employment opportunities.
In conclusion, early retirement incentives have differing impacts on the participation of older workers in OECD countries. A majority of nations have exhibited an inverse relationship between the incentives and labour participation. However, certain cultural and individual-specific factors found in some economies present a challenge in having a linear relationship between the concepts under study. There is a need for a more comprehensive model that captures variations between all demographics.
1.3 Statistical evidence
In all OECD countries, the employment rate falls as age increases. In 2014 the average employment rate in all OECD countries was 67% for workers of age 55 to 59, 44% for workers of age 60 to 64 and 29% for the 65 to 69 age group. It is worth mentioning that the employment rate for workers aged 55 to 64 showed some improvement between 2004 and 2014 from 48% to 56% respectively in most OECD countries. The following figure, Figure 1, illustrates the employment rate of the OECD countries for each age group respectively of the year 2014.
Figure 1: Employment rates of workers aged 55 to 59, 60 to 64 and 65 to 69 in 2014
In particular, the countries which had an employment rate above the OECD 55 to 59 age group were quite plenty. Among them belong Czech Republic, Denmark, Estonia, France, Germany, Israel, Sweden and Switzerland. Furthermore, these countries, except France, also have an employment rate higher than the average OECD 60 to 64 age group. From the same group of countries, only Israel succeeds to have an employment rate above the 29% average employment rate of the OEDC 65 to 69 age group.
According to the OECD data about early retirement trends, many countries attempt to restrict the access to early retirement. For instance, the government of Austria has increased to 40 years the required insurance period which allows individuals to claim early retirement. It used to be 38 years in 2013. In line with this, Austria also increased the minimum early retirement age from 60 to 62 years for the male population and from 55 to 57 for the female population. Belgium is following the same example. By the year of 2016 the government of Belgium aimed to increase the early retirement age from 60.5 years in 2013 to 62 years. Moreover, the contribution period was also programmed to rise from 38 years to 40 years. Denmark is aiming to increase the early retirement age to 64 years by 2023. The Netherlands is slowly eliminating the early retirement options for workers who are occupied in manual jobs. In the case of Spain, the Spanish government decided to increase the early retirement age in accordance with the change in the legal retirement age with the former one to reach 63 years by 2027.
Many OECD countries are also trying to offer financial incentives to workers in order to work longer and not claim early retirement benefits. In Australia, employers have greater economic motivations to hire or keep older workers and this may result in a potential reduction of public expenditures on Age Pension. Austria decided to move in a different direction. The Austrian government will increase each year’s penalties of early retirement assertion from 4.2% to 5.1% for workers who were born in 1955 or after. In Canada the rewards for postponing retirement after the age of 65 were increased. In addition, Canada offers the possibility to individuals to combine work and pension benefit which will be provided by the mandatory public scheme. In Spain individuals can work and withdraw retirement money at the same time. Sweden raised in 2014 the earned income tax credit for labourers over 65 resulting in stronger financial incentives to work more.
age who entered the labour market in a young age in the form of full pension benefits without any penalty. Naturally, this kind of measurement increases pension entitlements but at the same time promote the workers’ behaviour to withdraw from the labour market earlier. In France, the legal minimum retirement age is 62 years. Nevertheless, labourers who started at the labour market when they were younger than 18 years’ old and have already completed 41.5 years of work, they can claim full pension benefits at the age of 60. In Germany workers who have completed 45 years of contributions can retire at the age of 63. It should be mentioned that after 2015 this age will be increased by two months each year until it becomes 65.
3. Data
3.1 The SHARE data on retirement
In this study, I use the sixth (6th) wave of the SHARE data. “The Survey of Health, Ageing and Retirement in Europe (SHARE) is a multidisciplinary and cross-national panel database of micro data on health, socio-economic status and social and family networks of more than 120,000 individuals aged 50 or older (more than 297,000 interviews). SHARE covers 27 European countries and Israel” (SHARE 2018). The sixth wave includes 17 European countries and Israel; in total 18 countries. The European countries are in alphabetical order: Austria, Belgium, Croatia, Czech Republic, Denmark, Estonia, France, Germany, Greece, Italy, Luxembourg, Poland, Portugal, Slovenia, Spain, Sweden, and Switzerland. A computer-assisted personal interviewing (CAPI) was used for the SHARE data collection. The interviewers conducted direct interviews with the respondents using a laptop computer which was equipped with the CAPI instrument. The sixth wave conducted and published in 2015. The target population was people born in 1964 or earlier, and people who are a spouse/partner of a person born in 1964 or earlier. The dataset consists of 68,231 observations. More than the half of the respondents (39,530 individuals) in the dataset assessed their current job situation as “retired”. 1 In this research retired people are defined as those who self-report themselves as being retired. It
1 The other choices of the assessment of the current job situation were: “employed or self-employed”, “unemployed”, “permanently sick or disabled”, “homemaker”, “other”.
should be mentioned that self-reported unemployment, housekeeping and disability status were not included in the definition of retired person. For the creation of the sample all respondents who stated their current job situation as permanently sick or disabled, homemaker, and other were eliminated. This elimination resulted in 58,915 observations. From these observations, only the individuals, whose age is smaller or equal than the legal retirement age of their corresponding country, were kept and this resulted in 23,271 observations. Next, the sample was finalised by including only respondents whose age is larger or equal to 50 years and smaller or equal to 70 years. The last elimination deleted 705 observations resulting in 22,566 observations. Having eliminated the observations in that order, the sample includes retirees who are by definition ‘early retirees’. Table 1 illustrates the labour market status of the respondents in 2015. Note that the current job situation responses are only 67,561 observations. Therefore, the percentages are calculated based on this size of sample.
Table 1: Self-reported labour market status in 2015 Retired Employed
or self employed
Unemployed Permanently sick or disabled
Homemaker Other Total
Men 28.6% 11.65% 1.39% 1.35% 0.11% 0.55% 43.7%
Women 29.9% 13.2% 1.46% 1.70% 8.73% 1.35% 56.3%
Source: SHARE dataset, 6th wave, 2015
Table 2: Statutory retirement age for basic pension, 2014-2015
OECD countries Men Women
Austria 65.0 60.0 Belgium 65.0 65.0 Czech Republic 62.7 61.3 Denmark 65.0 65.0 Estonia 63.0 61.0 France 61.2 61.2 Germany 65.3 65.3 Greece 65.0 65.0 Israel 67.0 62.0 Italy 66.3 62.3 Luxembourg 65.0 65.0 Poland 65.0 60.0 Portugal 66.0 66.0 Slovenia 65.0 65.0 Spain 65.2 65.2 Sweden 65.0 65.0 Switzerland 65.0 64.0 Croatia* 65.0 61.3
Source: OECD (2015), Pensions at a Glance 2015, Chapter 11
Table 3: Retirement year of the early retiree sample Percentiles Smallest 1% 5% 10% 25% 50% 75% 90% 95% 99% 1970 1946 1984 1948 1988 1948 1994 1949 2001 Largest 2008 2015 2012 2015 2013 2015 2015 2015 Obs Sum of Wtg. Mean Std. Dev. Variance Skewness Kurtosis 28,457 28,457 2000.233 9.716483 94.41004 -.901173 4.387327
Source: SHARE dataset, 6th wave, 2015
3.2 Explanatory variables
The retirement decision can be rather complicated as a lot of factors influence a person’s decision to retire; from institutional perspective to personal factors. For my analysis I include the health-related variables in combination with the socio-economic and socio-demographic characteristics of the individuals.
the data from ISCED-1997 were used and thus groups of educational level where “low” (primary and lower secondary education), “medium” (upper secondary and post-secondary non-tertiary education) and “high” (first and second stage of tertiary education) were specified. This international classification is recommended for international comparisons of education systems. Concerning the socio-economic characteristics of the respondents, the total household net income and the household’s net financial assets, both provided by the SHARE dataset, were added as explanatory variables. The value of the household’s net financial assets is the result we get if we subtract from the sum of the household’s bank accounts, bonds, stocks, mutual funds, and savings for long-term investment the household’s financial liabilities.
Physical health variables. Two different indicators of physical health were used; health in general (self-assessed health) and chronic diseases. Health in general was measured by a single question asking the people to evaluate their health as “excellent”, “very good”, “good”, “fair”, and “poor”. A dummy was generated in order to capture the different levels of self-reported health; 1 indicates the people who reported from fair to excellent health and 0 corresponds to those with poor health. By definition, “fair” means neither very bad nor very good; it is more than a neutral description. It was better to be considered as a good indicator of health because the poor status is related to serious health conditions. The chronic diseases indicator contains a number of diseases from heart attack and high cholesterol to emotional disorders and dementia. The participant has to select the disease or the diseases the doctor has diagnosed her for. A more detailed list of the chronic diseases indicator can be found in Table A1 of the Appendix.
The dependent variable is the probability of someone to be retired and taken into account how the sample was formulated, the dependent variable is automatically also the probability of someone to be early retired. Having tested the correlation of all the mentioned explanatory variables with each other and with the dependent variable, it was derived that most of the variables do not have strong correlation neither with the dependent variable nor with each other. It should be stressed though that correlation does not imply causation.
4. Model and methodology
As it has been mentioned in the data section, for the purposes of the analysis the 2015 SHARE data were used. In this wave each person was observed only once. Therefore, a cross-sectional approach is recommended. I consider the conditional probability of being retired, and due to the nature of the sample also the probability of being early retired, 𝑃(𝑦𝑖 = 1|𝑥𝑖), as a function of the respondents’ socio-demographic, socio-economic, physical health, and mental health characteristics. A probit estimation model is used to estimate the following model:
𝑃2015(𝑦𝑖 = 1|𝑥𝑖) = 𝛷(𝛼 + 𝛽𝑥𝑖),
where Φ is the cumulative distribution function. One of the advantages of the probit model is that the probit curve approaches the axes more quickly than the curve whereas the logistic function has slightly flatter tails.
In order to capture the heterogeneity among countries, I separated the countries into three groups; north countries, east countries, and south countries. In the north group of countries, Austria, Belgium, Denmark, Germany, Luxembourg, Sweden, and Switzerland were included. The east group of countries contains Croatia, Czech Republic, Estonia, Poland, and Slovenia. France, Greece, Italy, Portugal and Spain were included in the south group of countries and finally, Israel stands for its own category due to the cultural differences with any of the other group countries. Additionally, dummies for each country, a dummy for gender and age dummies for each age available in the sample were also created.
married and living together with spouse, registered partnership, married and living separated from spouse, never married, divorced, and widowed. Due to the number of the possible choices, a dummy variable to describe the marital status of the individuals was generated. The respondents who reported themselves as married regardless if they are living or not together with their spouse and those who have a registered partnership, were considered as one group of people; that of having a partner. It should be mentioned that this group contains 16,960 observations. The remaining observations were considered as people with no partner.
Together with the probit regression, the fitted values were also computed. The lasts ones were used for the calculation of the probability of ‘being retired’. This last calculation assigns the probability of each observation of the sample to be retired and thus, early retired and allows for comparisons among countries, gender and among any other characteristics like age, income and so on.
The descriptive statistics of the variables that are used in the regression is given by Table A4 of the Appendix.
5. Empirical results
Table 5 shows the results of the estimated model of the probability ‘being retired’. I provide the output of the both the probit regression and the average marginal effects regression. Table A3 and Table A4 of the Appendix illustrate the age dummies and country dummies of the regressions respectively.
The two variables related to physical health, self-assessed health and chronic diseases are both statistically significant. If chronic conditions increase by one unit, ceteris paribus, the probability of a person to be retired, and given the settings of the sample also early retired, is higher. However, the coefficient of the variable related to health has a negative sign. In particular, a one unit change in health, ceteris paribus, decreases the probability of someone to be retired by 0.1069673. Similar studies show also a negative association between health and decision for early retirement (Borsch-Supan 2000). Less healthy individuals decide to retire early.
The socio-economic variable, namely the household’s net income, appears to have a strong influence on the probability of a person to be retired. More specifically, a one unit increase at the household’s net income, ceteris paribus, will decrease the probability of the income earner to be retired and thus, early retired. From this outcome it can be claimed that the wealthier the household becomes the less likely is for the income earner to retire early. This can be partly explained from the worker’s non-willingness to substitute her high income over the relatively lower pension benefits. Furthermore, a higher income may imply an advancement in career so the individual would not prefer to drop her career and switch to the retirement social status. The variable related to the household’s net financial assets was not included in the sample due to its large amount of missing values which account for almost 14% of the sample.
Concerning the dummy variable related to the gender, it is highly significant and negatively associated with the probability of a person to be (early) retired. According to the regression output, a male person has a 0.025 lower probability to be retired than a female person. The variable related to the marital status of the respondents does not show statistically significant coefficient and therefore, no correlation can be claimed between the probability of a person to be retired and having or not a partner.
The different education levels give us an interesting and expected result. Low educated and medium educated workers have a higher probability to be retired and thus, early retired than the highly educated individuals. This result lays on the fact that low and medium educated individuals usually work under relatively high demanding conditions and are occupied in ‘heavy’ jobs in terms of manual labour. Fischer and Sousa-Poza (2006) show also a positive linkage between beyond primary school education and early retirement decision. Note that the high educated dummy variable was omitted from the regression because of collinearity. All the dummy variables need a base category. Stata drops one dummy arbitrarily, otherwise they would be perfectly collinear. Based on the high significance level of the low educated and medium educated, it can be assumed that high educated is also statistically significant. Assumptions about the sign of its coefficient can be derived from the sign of the coefficient of the household’s net income due to the fact that highly educated people have greater income.
aged 50-year-old has a 0.73 lower probability to be retired and consequently early retired. The older someone gets the higher the probability for them to be retired. For example, a 55-year-old person has a 0.50 lower probability to be retired. By the age of 60, a person has a 0.28 lower probability to be (early) retired.
As it was expected, the country dummies are also statistically significant with positive coefficients except Sweden which carries a negative coefficient but at the same time is not statistically significant. In particular, for an Austrian citizen the probability for her to be retired is 0.30 higher than a citizen of another country. Slovenia appears the higher probability of an individual to be retired and thus, early retired and that probability is 0.36. Denmark, Estonia, Germany, Israel, and Spain have a small probability, under 0.10, for a person to belong to the retired population compared to other countries respectively. This is a quite interesting outcome due to the fact that these countries do not have anything in common, maybe except Denmark and Germany as north countries. The rest of the countries show a high probability from 0.18 to 0.23 except Italy whose probability of a person to be retired is 0.13.
Table 5: Probability of being early retired across Europe and Israel
(1) (2)
VARIABLES Probit Marginal effects
(0.000233) (4.89e-05) With partner 0.0242 0.00508 (0.0278) (0.00583) Constant 0.212 (0.269) Observations Log-likelihood Pseudo R² 21,833 -8192.2362 0.3497 21,833
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Using the calculated probability of each observation of the sample to be retired, the following comparisons were made. Table 6 illustrates the different mean probabilities among genders to be retired in each country under the strict assumption that all the other characteristics are constant. It can be observed that in most countries the mean probability of a female worker to be retired and given the settings of the sample, early retired, is lower than the corresponding probability of male workers. Denmark, France, Germany, and Sweden present the opposite result; a female worker has a higher probability to be early retired than a male worker.
Table 6: Mean probability of ‘being retired’ by country and gender
Italy 920 0.3000813 552 0.1243932 Luxembourg 317 0.4069488 292 0.3929898 Poland 300 0.2961679 219 0.1192381 Portugal 272 0.4364957 297 0.4001015 Slovenia 768 0.5753068 1,004 0.5640694 Spain 750 0.2372253 667 0.209229 Sweden 492 0.1539432 620 0.1626217 Switzerland 432 0.1159312 499 0.0899847
Source: SHARE dataset, 6th wave, 2015
Next, I compare the mean probability of ‘being retired’ across the different group of countries (north, east, south countries, and Israel) including also the mean age corresponding to the mentioned probability. Table 7 demonstrates the outcome of the comparison.
Table 7: Mean probability and mean age of ‘being retired’ across the different group of countries Group of countries Number of observations Mean probability of ‘being retired’ Mean age of ‘being retired’ North countries 9,341 0.2256544 58.57927 South countries 5,898 0.256401 58.44168 East countries 6,082 0.3320548 57.98553 Israel 512 0.2346615 60.33789
Source: SHARE dataset, 6th wave, 2015
Republic, Estonia, Poland, and Slovenia) show a higher mean probability of early retirement in combination with a lower mean age. In particular, a worker with a mean age 57 year has a 0.33 mean probability to be early retired with all the characteristics being constant. In Israel the situation is different. The mean age of someone to be early retired with mean probability 0.23 is 60 years, ceteris paribus.
6. Conclusions
Over the years, OECD countries face an unsustainable burden on pension financing due to high early retirement rates. Many reforms on the pension systems have been implemented during the last years in order for the early retirement rates to be decreased. Some of the prime objectives of the pension reforms are related to the pension benefits, work incentives, and sometimes complete replacement of the pension system.
Governments face a crucial challenge to understand the early retirement behaviour due to its inverse relationship with the labour force participation of old age individuals. In 2014, the average OECD employment rate of people aged between 55 to 59 years was 67% whereas for workers of age 60 to 64 and 65 to 69 was 44% and 29%, respectively.
The need to understand the retirement behaviour it is of high importance. It has been shown that pension and financial incentives influence someone’s decision to retire early. Although the financial approach of the problem is essential it does not however provide a complete understanding of the retirement behaviour. Other factors should be included such as physical and mental health concerns as well as psychological factors.
On the health level, this study reveals that the less healthy someone is, the more likely is for them to decide to retire early. In addition, the presence of chronic diseases is positively associated with the probability of someone to be early retired. Even though SHARE 2015 provide data about the EURO depression scale and the level of life satisfaction of the respondents, the probit estimation showed not statistically significant results and therefore, there was no evident correlation between these variables and the probability of someone to be early retired.
them to have withdrawn early from the labour market. In addition, workers who are up to medium educated have a high probability to be early retired while those with high income have a lower probability.
Finally, the comparison of the mean probabilities of each gender to be early retired among the different countries revealed that in most countries female workers have a lower probability to be early retired than male workers. Furthermore, the comparison of the group countries showed that east countries like Croatia, Czech Republic, Estonia, Poland, and Slovenia, have a higher probability of a citizen to be early retired around the age of 57.
7. Discussion
The sixth wave of the SHARE data that is used for the purposes of this study is the latest wave available from SHARE. Therefore, the findings of the study can be considered as a representative outcome of the current retirement behaviour in OECD countries.
In addition, the use of only one wave of data allowed for a cross-sectional approach. Under this type of analysis, the comparison of the differences among the participants was achieved. However, no conclusion can be derived regarding the retirement behaviour in OECD countries in comparison with previous years. Future researchers who would like to show how the retirement population in OECD countries is evolving, could include more waves and examine the data under a time series analysis approach.
Furthermore, in the spring of 2019 the seventh wave of SHARE data will be released. Consequently, the results of a similar study using the data of the 7th wave can be compared to the results of this study that uses the 6th wave and possible changes could be observed in the landscape of early retirement in OECD countries.
8. References
Börsch-Supan, A. (2000) ‘Incentive effects of social security on labor force participation: evidence in Germany and across Europe’, Journal of Public Economics 78, pp. 25–49 Blöndal, S., and Scarpetta S. (1997), ‘Early Retirement in OECD Countries: The Role of Social Security Systems’, OECD Economic Studies No. 29, 1997/II
Brugiavini, A., Pasini, G. and Peracchi, F. (2008). First Results from the Survey of Health, Ageing and Retirement in Europe (2004-2007). Mannheim: Mannheim Research Institute for the Economics of Aging (MEA), pp. 204-212
Forrester, J. 2003, Economic theory for the new millennium. In: International System Dynamics Conference.
Kocourek, D. & Pertold, F. 2011, The impact of early retirement incentives on labor market participation: Evidence from a parametric change in the Czech Republic. Czech Journal of Economics and Finance, vol. 61, no. 5, pp.467-483.
Martin, J.P. and E. Whitehouse (2008), “Reforming RetirementIncome Systems: Lessons from the Recent Experiences of OECD Countries”, OECD Social, Employment and Migration Working Paper 66
MacGarry, K. (2002). Health and Retirement: Do Changes in Health Affect Retirement Expectations? NBER working paper No. 9317. National Bureau of Economic Research
Mortel, T., (2008). Faking it: social desirability response bias in selfreport research. Australian Journal of Advanced Nursing, vol.25, no. 4, pp. 40-48
O’Brien, M. 2010, Exploring older male worker labor force participation across OECD countries in the context of ageing populations: A reserve army of labor? [ebook] Available at: http://ro.uow.edu.au/commwkpapers/218. [Accessed 25 May 2018].
OECD 2011, "Trends in retirement and working at older ages." Pensions at a glance 2011. [ebook] Paris: OECD Publishing, pp.39-47. Available at: https://www.oecd- ilibrary.org/finance-and-investment/pensions-at-a-glance-2011/trends-in-retirement-and-in-working-at-older-ages_pension_glance-2011-6-en. [Accessed 26 May 2018].
OECD 2013, OECD economic surveys: Austria 2013. Paris: OECD Publishing.
OECD (2007), Pensions at a Glance, Public Policies across OECD Countries, Paris
Rabate, S. 2017, Can I stay or should I go? Mandatory retirement and labor force participation of older workers. [ebook] pp.1-36. Available at: https://halshs.archives-ouvertes.fr/halshs-01521150. [Accessed 25 May 2018].
Staubli, S. & Zweimüller, J. 2013, Does raising the early retirement age increase employment of older workers? Journal of Public Economics, vol. 108, pp.17-32.
Siegrist, J., Wahrendorf, M., Knesebeck, O., Jurges, H. and Borch-Supan, A. (2006). Quality of work, well-being, and intended early retirement of older employees—baseline results from the SHARE Study. European Journal of Public Health. Vol. 17, No. 1, pp. 62–68
SHARE-project.org, (2018). SHARE-Survey of Health, Ageing and Retirement in Europe
official website. [online] Available at: http://www.share-project.org/
9. Appendix
Table A1: List of the chronic diseases
1. A heart attack including myocardial infarction or coronary thrombosis or any other heart problem including congestive heart failure
2. High blood pressure or hypertension 3. High blood cholesterol
4. A stroke or cerebral vascular disease 5. Diabetes or high blood sugar
6. Chronic lung disease such as chronic bronchitis or emphysema
7. Cancer or malignant tumour, including leukaemia or lymphoma, but excluding minor skin cancers
8. Stomach or duodenal ulcer, peptic ulcer 9. Parkinson disease
10. Cataracts 11. Hip fracture 12. Other fractures
13. Alzheimer's disease, dementia, organic brain syndrome, senility or any other serious memory impairment
14. Other affective or emotional disorders, including anxiety, nervous or psychiatric problems 15. Rheumatoid Arthritis
Table A2: Description of the mental health variables (EURO depression scale) Variable examined Question asked
Depression In the last month, have you been sad or depressed? Hopes for the future What are your hopes for the future?
Felt would rather be dead In the last month, have you felt that you would rather be dead? Feels guilty
If yes,
Do you tend to blame yourself or feel guilty about anything?
So, for what do you blame yourself?
Trouble sleeping Have you had trouble sleeping recently? Less os same interest in things
Keeps up interest
In the last month, what is your interest in things?
So, do you keep up your interests? Irritability Have you been irritable recently? Appetite
Eating more or less
What has your appetite been like in the last month?
So, have you been eating more or less than usual?
Fatigue In the last month, have you had too little energy to do the things you wanted to do?
Concentration on entertainment How is your concentration? For example, can you concentrate on a television programme, film or radio programme?
Table A3: Descriptive statistics Variable Number of
observations
Mean Std. Dev. Min Max
Age 62 21,833 .070398 .2558225 0 1 Age 63 21,833 .0662758 .2487693 0 1 Age 64 21,833 .058581 .2348443 0 1 Age 65 21,833 .0586268 .2349304 0 1 Age 66 21,833 .0069619 .0831492 0 1 Age 67 21,833 .0013741 .0370438 0 1
Table A4: Probability of being early retired across Europe and Israel, age dummies
(1) (2)
VARIABLES Probit Marginal effects
Constant 0.212 (0.269) Observations Log-likelihood Pseudo R² 21,833 -8192.2362 0.3497 21,833
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
Table A5: Probability of being early retired across Europe and Israel, country dummies
(1) (2)
VARIABLES Probit Marginal effects
Switzerland (omitted) - - Constant 0.212 (0.269) Observations Log-likelihood Pseudo R² 21,833 -8192.2362 0.3497 21,833
