Shifts in the realized retirement age: Europe in times of pension reform and economic crisis

Abstract

Many recent policy reforms aimed to delay retirement. Such a delay seems important because population ageing makes it a necessity for financially sound pension schemes and a sufficiently large workforce. However, pension reforms do not only affect when people retire, they likewise affect what influences the realized retirement age. Moreover, the 2008 economic crisis restructured labour markets, thereby affecting retirement. This article investigates recent shifts in the realized retirement age and influences on it. It does this through multilevel analyses of international data from the Survey of Health, Ageing and Retirement in Europe. It compares individuals who retired in 2005/2006 to those who retired in 2012/2013, thereby identifying changes over time. Findings show that the realized retirement age increased. Simultaneously, inequalities in the realized retirement age decreased within and between countries. Thus, while the economic crisis reshaped social inequalities, pension reforms constrained the options for retirement. Consequently, policymakers need not worry about the economic crisis defeating the goals pursued with pension reforms. However, policymakers and researchers do need to consider new inequalities in life courses and the meaning of retirement. Also, researchers need to consider how social change may have altered social mechanisms when conducting literature reviews on retirement.

Keywords

Economic crisis Europe pension reform realized retirement age

Introduction

The realized retirement age has been the centre of intense debate in Europe (e.g. Börsch-Supan, 2015; Hartlapp and Kemmerling, 2008). This age denotes when people de facto retire from work and start drawing pensions – which can be earlier or later than when pension policies suggest (Bernal and Vermeulen, 2014). Because Europe has the oldest population worldwide, policymakers and researchers argue that the realized retirement age has to change. They argue that pension schemes will only be financially sustainable and the workforce will only be sufficiently large if people start retiring later in life (Kohli and Arza, 2011). Statistics on the old age dependency ratio illustrate this point: in 2015, one European of working age supported 0.26 retirees, while this number is expected to increase to 0.35 by 2030 (United Nations, 2015). Because of these concerns, a rich body of literature on the realized retirement age exists (e.g. Jones et al., 2010; Radl, 2013; Van Bavel and De Winter, 2013). However, recent social change has affected the retirement age. As a result, we may have to update our previous understanding of what pushes people into retirement, and what keeps them in the labour market.

Life-course scholars stress that people experience important life events differently, depending on when and where they take place. They call this fact ‘lives in time and place’ (Settersten, 1999). Retirement is one such important life event. As societies change, the factors that influence retirement change as well. Recently, pension reforms and the 2008 economic crisis have modified the realized retirement age. During the last decades, governments across Europe enacted pension reforms that aimed to prevent early retirement and increase the standard pension age (European Commission, 2015). These reforms increased the realized retirement age (Organisation for Economic Co-operation and Development (OECD), 2013) and strived to make it more uniform. The economic crisis of 2008, however, may have modified these effects. The crisis increased unemployment rates, thereby affecting the opportunities people have in the labour market (Sarfati, 2013). Moreover, it led people to re-evaluate their priorities, lifestyles and choices, thereby creating new behavioural patterns (Coile and Levine, 2011; Elder, 1999; McDaniel et al., 2013). How the twofold social change created by pension reforms and economic crisis affects retirement has not yet been explored. The present article fills this gap of knowledge.

This article explores recent changes in the realized retirement age. It does this by answering two questions: how far did the realized retirement age change? And how far did the factors influencing the realized retirement age change? To answer these questions, I analyse international survey data from across Europe. I compare the situation in 2005 and 2006, a period of time before the 2008 crisis hit, to the situation in 2012 and 2013, after the crisis had unfolded. The comparison reveals within- and between-country differences in changes concerning the realized retirement age. Findings further theories on old age, the labour market, pension reforms and the effects of the 2008 crisis. Additionally, they can help policymakers evaluate the effectiveness of pension policies and identify possible needs for modifications. In the remainder of this article, I first give a brief overview of influences on the realized retirement age. Then, I discuss how recent pension reforms and the 2008 economic crisis modified these effects. I subsequently present the analysis, and conclude by discussing the findings.

Influences on the realized retirement age

The realized retirement age depends on a myriad of factors: characteristics of the older people themselves (micro-level), characteristics of their families and workplaces (meso-level) and characteristics of pension regulations and historical events (macro-level). All these influencing factors have already been extensively studied, and the following paragraphs will therefore only briefly outline them.

Among older individuals’ characteristics, gender, health status and educational level have a particularly strong impact on the realized retirement age (e.g. Jones et al., 2010; Radl, 2013). Women sometimes stop working before retirement to look after their families and household (Erickson, 2005; Trifiletti, 1999). Because of their intermitted workforce participation, they often do not qualify for early retirement schemes and can only retire at the standard pension age. In contrast, men’s more continuous workforce participation often qualifies them for early retirement options (Schils, 2008). As a result, women can have a higher realized retirement age than men (Evandrou and Glaser, 2003; Fasang, 2010; Lund and Villadsen, 2005). Health status is relevant because poor health can hamper workforce participation, thereby leading to an earlier retirement (Jones et al., 2010). Finally, educational level points to the opportunities people have in the labour market. Highly educated individuals are attractive employees in knowledge-based societies, and they, therefore, often have the opportunity to work until old age. These opportunities lead to them retiring late (Gouldner, 2001; Radl, 2013).

Like a person’s characteristics, the meso-level institutions of families and workplaces also influence the realized retirement age. Families have an influence because family members often coordinate their activities, including workforce participation. One reason for coordination is to ensure that households generate income and simultaneously cover their demand for homemaking and caregiving (Evandrou and Glaser, 2003). As a result, women are more likely to reduce their workforce participation when children are present and when their spouse works (Evandrou and Glaser, 2003; Trifiletti, 1999). A possible consequence of this behaviour is a later retirement age, as previously explained. A second reason for coordination is a possible desire of spouses to retire at the same time. Because women are often younger than their spouses, this desire can reduce women’s realized retirement age (Syse et al., 2014). Workplaces, on the other hand, have an influence because they affect older people’s opportunities and interest to work. Several studies showed that work autonomy and low health strain make workplaces attractive for older employees (Flynn, 2010; Jones et al., 2010), which can lead to later retirement.

At the macro-level, pension regulations and historical events influence when people retire. Pension regulations are part of modern welfare states, encompassing, for example, the standard and the minimum pension age. They determine at what age people can draw pensions, how much pension benefits they receive and what options for early retirement exist (Kohli and Arza, 2011). Lately, the pension regulations in Northern European countries favoured late realized retirement ages, whereas the regulations in Continental European Countries led to earlier realized retirement ages (OECD, 2013). Historical events such as the end of the communist regime in Eastern Europe or the 2008 economic crisis can likewise influence the realized retirement age. The communist regime in Eastern Europe included a planned economy with an employment guarantee (Turek et al., 2015). When the communist regimes collapsed after 1989, the planned economy came to an end and early retirement became an option (Deighton, 1998; Vanhuysse, 2006). The economic crisis of 2008 increased unemployment rates, which affected the possibilities to work until a late age – and probably also the realized retirement age (e.g. Gustman et al., 2009). The next paragraphs will explain the influences of pension reforms and the 2008 crisis in more detail.

The realized retirement age and pension reforms

Recent pension reforms in Europe have often aimed to increase the realized retirement age. These reforms used two main strategies to reach their goal. First, many reforms increased the standard pension age (Myles, 2002; OECD, 2011). The German government, for instance, decided to increase the standard pension age from 65 to 67 years from 2012 to 2024 (International Social Security Association (ISSA), 2015). Only a few European countries, such as Denmark and France, opposed this trend and facilitated retirement at an earlier age (ISSA, 2015). However, such reforms usually only come into effect years after the laws are passed, and they introduce gradual changes that go on for years, sometimes even decades (ISSA, 2015). As a result, the changes in the standard pension age observed in this study are small (see Table 1). Table 1 displays these changes, identifying only smaller changes in the standard pension ages in Belgium, the Czech Republic, and Italy. A notable exception is Sweden, where in 2005/2006 the standard pension age was 65 years, combined with the right to work until age 67, and in 2012/2013, the standard pension age was within the period from age 61 to age 67 (European Commission, 2015).

Table 1.

Standard and minimum pension age, by gender, country and year.

Country	Standard pension age (men/women)		Minimum pension age (men/women)
Country	2005/2006	2012/2013	2005/2006	2012/2013
Austria	65/60	65/60	60/57	60
Belgium	65/64	65	60	60
Czech Republic	62/60	63/61	59/53	60/54
Denmark	65	65	58	65
France	60	60	55	55
Germany	65	65	60	62/60
Italy	65/60	66/62	55/53	61/60
Netherlands	65	65	65	65
Spain	65	65	60	60
Sweden	65	61 to 67	63	61
Switzerland	65/64	65/64	63/62	63/62
Total	64/63	64/63	60/59	61/60

Source: European Commission (2015).

The pension age is stated for men and women separately where gender differences exist.

Second, many governments tried to increase the realized retirement age by preventing early retirement (see Table 1). One means of doing this was increasing the qualifying periods and conditions for drawing the full amount of pension benefits (Myles, 2002; OECD, 2011). This happened, for example, in Italy (ISSA, 2015). A second means was to diminish – or even abolish – the possibilities of retiring via long-term unemployment benefits or disability pensions. This approach was particularly important in, for instance, Denmark and the Netherlands (OECD, 2005, 2006, 2011). The result of these reforms is that the realized retirement age becomes more uniform and, at the same time, less dependent on an individual’s personal situation. The different kinds of pension reforms suggest two hypotheses: (H1) the realized retirement age increased from 2005/2006 to 2012/2013, and (H2) the effect of a person’s characteristics on their realized retirement age decreased from 2005/2006 to 2012/2013.

The realized retirement age and the 2008 economic crisis

The economic crisis of 2008 deeply affected European societies. Economies slowed down, companies foreclosed, unemployment rates skyrocketed and people changed their outlooks on life (Hemerijck et al., 2009; McDaniel et al., 2013; Zavras et al., 2012). Table 2 outlines the severity of the crisis across countries, using the annual change in the gross domestic product (GDP) and the overall unemployment rate as indicators. The annual change in GDP represents economic growth. Economic growth in 2005/2006 was highest in the Czech Republic and Spain, with 6.7 and 4.0 percent annual change in GDP, respectively. Interestingly, these countries were also the ones with the most dramatic economic slowdowns, amounting to a decline of 6.6 percentage points in the Czech Republic and 4.7 percentage points in Spain between 2005/2006 and 2008–2013. In contrast, Italy had the weakest economic development in 2005/2006 (1.5%) and in 2008–2013 (−0.8%). Also, economic growth slowed down in all countries in our sample, with Germany, France and Belgium being the least affected (−1.8 percentage points difference between 2005/2006 and 2008–2013). The unemployment rate is another indicator for the severity of the economic crisis. Germany had the highest unemployment rate in 2005/2006, amounting to 10.8 percent. Spain had an unemployment rate of only 8.8 in 2005/2006, but it showed the highest increase in the unemployment rate in our sample, leading to a value of 20.3 percent in 2008–2013. In contrast, Switzerland had the lowest unemployment rate at both points of time (4.3 and 4.2%), and it also showed the smallest change in the sample (0.1 percentage point). All these crisis-induced changes may also have an impact on the realized retirement age.

Table 2.

Annual change in GDP and unemployment rate, by country and year.

Country	Annual change in GDP (in %)			Unemployment rate (in %)
Country	2005/2006	2008–2013	Difference	2005/2006	2008–2013	Difference
Austria	2.8	0.3	−2.5	5.5	4.9	−0.6
Belgium	2.3	0.3	−1.8	8.4	7.8	−0.6
Czech Republic	6.7	0.1	−6.6	7.6	6.6	−1.0
Denmark	3.1	−0.3	−3.4	4.4	6.6	+2.2
France	2.0	0.2	−1.8	8.9	9.1	+0.2
Germany	2.2	0.4	−1.8	10.8	6.6	−4.2
Italy	1.5	−0.8	−2.3	7.3	9.1	+1.8
Netherlands	2.9	−0.1	−3.0	4.8	4.6	−0.2
Spain	4.0	−0.7	−4.7	8.8	20.3	+11.5
Sweden	3.8	0.3	−3.5	7.4	8.0	+0.6
Switzerland	3.5	0.7	−2.8	4.3	4.2	−0.1
Total	3.1	0.4	−2.7	7.1	8.0	+0.9

Source: Organisation for Economic Co-operation and Development (OECD) (2016a, 2016b).

GDP: gross domestic product.

The crisis increased unemployment rates in many countries. In times of high unemployment, companies sometimes encourage older workers to retire early instead of laying them off (Ebbinghaus, 2006; Vanhuysse, 2006). Therefore, hypothesis 3 is (H3) the realized retirement age decreased from 2005/2006 to 2012/2013. Rising unemployment rates and experiences of job insecurity lead individuals to lower the self-assessment of their employability and their chances to find a new job (Dorn and Sousa-Poza, 2010; Mäkikangas et al., 2013). Older individuals who are not working may, therefore, give up on any hopes of finding employment again and retire early. Consequently, the fourth hypothesis is (H4) individuals with a spell of non-employment before retirement retired later in 2005/2006 than in 2012/2013.

Analysis

Data

The analysis uses data from the Survey of Health, Ageing and Retirement in Europe (SHARE). SHARE is a European panel study that collects information on the social and financial situation, health and employment status of individuals aged 50+ years (Börsch-Supan et al., 2013). I selected cases from waves 1–5 of the SHARE dataset in three steps. First, I selected individuals who had ever worked and who retired in 2005, 2006, 2012 or 2013. The retirement event must have been observed during the panel participation or described in the life-history interviews. Second, I selected those countries that had a sufficiently large number of retirement events in 2005/2006 and 2012/2013 to allow for substantially meaningful analyses. I combined the data from 2005 with 2006 and that from 2012 with 2013 to obtain a larger sample size. Using a cut-off point of 60 retirement events, I included 11 countries in the analyses: Austria, Belgium, the Czech Republic, Denmark, France, Germany, Italy, the Netherlands, Sweden, Spain and Switzerland. Third, I deleted cases with missing values. Before the case deletion, I had a sample of 3901 and 13 cases with missing values. An analysis of the missing cases suggests that the values are missing at random. Because the number of cases with missing values is low (<1%) and there seems to be no bias, I deleted these cases. Table 3 displays the resulting sample size by year and country.

Table 3.

Sample size, by country and year.

Country	2005/2006	2012/2013
Austria	71	179
Belgium	156	264
Czech Republic	221	192
Denmark	215	186
France	215	177
Germany	171	197
Italy	130	93
Netherlands	195	218
Sweden	190	271
Spain	131	210
Switzerland	85	121
Total	1780	2108

Method

I analysed the data through multilevel modelling. This kind of analysis is useful for observations that are dependent on each other (Gelman and Hill, 2007). The dependency arises because some retirees are nested within households, which – in turn – are nested within countries. Consequently, I built a three-level model, with the levels being individuals, households and countries. The analysis progressed in several steps. First, I calculated an empty model. An empty model contains the constant only, and it is useful for comparative purposes. Second, I conducted a stepwise regression to obtain a full model. In a stepwise regression, explanatory variables are added individually to get a better impression of the analysis process (Harrell, 2001). The resulting full model contained all explanatory variables. It had a random intercept, meaning the average could differ between countries, and fixed effects, meaning the explanatory variables had identical effects in all countries. Third, I tested for random effects. To do this, I calculated the model several times, each time allowing the effect of a different explanatory variable to vary between countries. This approach showed me whether or not a random-intercept, fixed-effects model accurately represented the data structure (Gelman and Hill, 2007). I carried all analyses out twice: once with the data on people who retired in 2005/2006, and once with the data on people who retired in 2012/2013. Splitting the dataset allowed me to compare findings between these years, thereby identifying changes over time. I ran t-tests to determine whether differences between the regression coefficients for both years were statistically significant. For the analyses, continuous variables were centred on their average, and categorical variables were assigned their mode as the reference category. To reach comparability, I chose ‘intermediate’ as the reference category for the educational level in both years.

Variables

The explained variable is the realized retirement age. This age denotes when people start receiving old-age pensions. Table 4 shows that the realized retirement age in the sample was 60.4 years in 2005/2006 and 62.8 years in 2012/2013.

Table 4.

Sample descriptive statistics, by country and year.

	Mean		Mode	Percentage
	Realized retirement age	Number of children	Education	Men	Married	Good health	Worked until retirement	Self-employed
2005/2006
Austria	58.9	2.2	Intermed.	51.7	77.5	74.6	73.2	11.3
Belgium	58.9	1.9	Low	59.0	84.0	76.3	76.9	4.5
Czech Republic	58.3	2.1	Low	45.3	76.0	63.3	81.4	7.2
Denmark	62.1	2.3	Intermed.	44.7	77.7	84.7	67.4	8.8
France	59.6	2.4	Low	43.3	74.9	73.5	72.6	13.0
Germany	60.9	1.8	Intermed.	56.1	81.9	67.8	70.8	5.8
Italy	59.5	1.9	Low	59.2	84.6	73.8	76.9	22.3
Netherlands	60.9	2.2	Low	55.9	87.7	76.9	67.7	4.1
Sweden	62.9	2.3	Low	44.7	86.8	85.3	81.6	10.0
Spain	60.6	2.4	Low	64.1	90.1	64.1	86.3	21.4
Switzerland	61.9	1.9	Intermed.	55.3	72.9	89.4	83.5	9.4
Total	60.4	2.2	Low	51.2	81.3	75.1	75.6	10.1
2012/2013
Austria	60.0	2.1	Intermed.	48.0	69.8	77.7	66.5	16.8
Belgium	62.3	2.2	Low	53.4	79.9	79.5	34.5	10.2
Czech Republic	61.0	2.1	Intermed.	43.7	72.9	62.5	56.8	12.0
Denmark	63.5	2.1	Intermed.	49.8	78.5	81.7	45.7	8.1
France	61.4	2.3	Intermed.	53.1	71.2	70.6	39.0	12.4
Germany	63.1	1.8	Intermed.	50.8	84.8	60.4	57.4	6.6
Italy	61.5	1.8	Low	62.4	83.9	63.4	57.0	23.7
Netherlands	64.4	2.2	Low	50.0	83.9	74.8	32.1	16.1
Sweden	64.8	2.4	Intermed.	41.7	77.1	82.3	57.2	7.0
Spain	63.6	2.0	Low	62.9	86.2	68.6	63.8	23.3
Switzerland	64.1	1.9	Intermed.	51.2	81.0	85.1	61.2	14.9
Total	62.8	2.1	Intermed.	50.7	78.9	73.9	50.9	13.0

Intermed.: intermediate.

The analysis includes seven explanatory variables describing individuals. The variables are gender (male/female), educational level (low, representing the International Standard Classification of Education (ISCED) levels 0–2; intermediate, representing ISCED levels 3–4; high, representing ISCED levels 5–6), marital status at retirement (married/not married), number of children, self-reported health at retirement (good/not good), whether people were self-employed before retiring (yes/no) and whether people worked until retirement (yes/no). The latter variable reflects the diversity in people’s employment careers, indicating whether or not people abstained from paid work before retiring due to, for example, unemployment, disability or acting as a homemaker or care-giver. Table 4 shows that the sample contained slightly more men than women and that the average retiree was healthy, married, had two children and worked until retirement as an employee. This table reveals two notable changes over time. In 2005/2006, the average retiree had a low educational level, whereas they had an intermediate educational level in 2012/2013. Moreover, the number of people who worked until retirement dropped from about 75 percent in 2005/2006 to about 50 percent in 2012/2013. Notable geographical differences are that Northern European countries have particularly healthy retirees and a high female workforce participation, whereas Southern European stands out because of its low female workforce participation and its high share of previously self-employed retirees.

Finally, the analysis also includes three explanatory variables at the country level. These variables are the annual change in GDP, the unemployment rate (see Table 2) and the minimum pension age (see Table 1).

Findings

The model fit indices give us a first impression of the findings. Table 5 displays the deviance and the unexplained variance for three of the models calculated: the empty model; the random-intercept, fixed-effects model; and the random-intercept, random-effects model. The latter model includes a random effect for the variable ‘working until retirement’ in 2005/2006 and one for the variable ‘gender’ in 2012/2013 because the model fit only improved with random effects on these variables. The information on the empty model shows that most unexplained variance is at the individual level, followed by the household level and the country level (69%:20%:11% in 2005/2006, 49%:30%:21% in 2012/2013). Table 5, moreover, shows that adding explanatory variables and including a random effect indeed improved the model fit, with the addition of explanatory variables in 2005/2006 have the strongest impact. At the same time, most parts of the unexplained variance decreased during the model building, except for the one at the country level in 2005/2006 when the random effect was added, the one at the individual level in 2005/2006 when the explanatory variables were added, and the one at the household level in 2012/2013 when the explanatory variables were added. These isolated increases in the unexplained variance indicated that some of the elements added had significant effects, while others had not.

Table 5.

How well the different models fit the data.

		Empty model: constant only	Full model: random intercept, fixed effects	Full model: random intercept, random effect
2005/2006	Unexplained variance
	Country level	1.9	1.2	1.5
	Household level	3.5	1.2	1.0
	Individual level	11.9	12.7	12.6
	Deviance	9945.6	9764.8	9738.6
2012/2013	Unexplained variance
	Country level	2.2	1.2	0.9
	Household level	3.1	3.5	3.5
	Individual level	5.0	4.3	4.1
	Deviance	10,413.5	10,325.9	10,292.8

The random intercept, random effect model contains a random effect for the low educational level.

Table 6 provides more detailed information on the findings, showing the results of the regression analysis in the random-intercept, random-effects models. The constant presented in this table shows that the average realized retirement age increased from 60 years in 2005/2006 to more than 62 years in 2012/2013. At the same time, the number of countries that differ in their average realized retirement age decreased from four to two. In other words, the average realized retirement age became more homogeneous across countries. Also, the effect of the minimum pension age in a country on the individual’s realized retirement age increased from 2005/2006 to 2012/2013, which means that a country’s pension regulations became more influential. Among the variables describing the individual, we see fewer variables with significant effects in 2012/2013 than in 2005/2006 because health status and the low educational level lost their significance over time. Those individual-level variables with significant effects in both years had a smaller effect in 2012/2013 than in 2005/2006. These changes indicate that the realized retirement age became more homogeneous within countries, depending less on an individual’s situation. The variable ‘work until retirement’ had a random effect in 2005/2006, which means that the influence of this variable differed across countries. In most countries, the realized retirement age of individuals with a period of inactivity before retirement was 2.2 years above that of individuals who worked until retirement – with the difference being 2 months in the Czech Republic and 6 months in the Netherlands. The random part of this effect no longer occurred in 2012/2013, and the size of the fixed effect of this variable decreased from 2005/2006 to 2012/2013. These changes indicate that the pathway to retirement became less important for the realized retirement age, and that its influence became uniform across countries. Gender had a random effect in 2012/2013, and the fixed part of this variable had a significant effect in 2005/2006 only. These changes mean that the difference in the realized retirement age between men and women disappeared in some countries. In 2005/2006, women retired 1 year earlier than men. In 2012/2013, Dutch women retired 1.5 years earlier than men, Austrian and Italian women retired about a year earlier than men, and Danish women retired about a year later than men. In 2005/2006, the people who retired the earliest were continuously employed, ill women with intermediate education. They retired at age 58. The people who retired the latest in 2005/2006 were highly educated, healthy, self-employed men with a period of inactivity before retirement. They retired at age 64.4. In 2012/2013, continuously employed individuals with an intermediate educational level retired the earliest, at age 62.4. In contrast, highly educated, self-employed individuals with a period of inactivity before retirement retired the latest in 2012/2013, at age 64.5. Thus, the latest realized retirement age remained fairly stable, whereas the earliest realized retirement age increased by 4.4 years from 2005/2006 to 2012/2013.

Table 6.

Influences on the realized retirement age, by year.

Variables		2005/2006, B (SE)	2012/2013, B (SE)
Constant		60.0 (0.4)***	62.4 (0.3)***
	Austria	−1.3 (0.6)*	−1.9 (0.4)***
	Belgium	−1.6 (0.8)*	−0.3 (0.6)
	Czech Republic	2.2 (0.6)***	1.6 (0.4)***
	Denmark	−0.2 (0.7)	0.0 (0.7)
	France	0.4 (0.6)	1.1 (0.6)
	Germany	−1.9 (0.6)***	−0.5 (0.4)
	Italy	1.2 (0.7)	−0.1 (0.6)
	Netherlands	0.4 (1.0)	0.5 (0.5)
	Spain	0.4 (0.8)	0.1 (0.4)
	Sweden	0.6 (0.7)	0.3 (0.6)
	Switzerland	−0.2 (0.8)	−0.8 (0.6)
Gender: female		−1.0 (0.2)***	−0.4 (0.3)
	Austria		−0.8 (0.4)*
	Belgium		0.4 (0.4)
	Czech Republic		0.5 (0.4)
	Denmark		1.1 (0.4)*
	France		0.5 (0.4)
	Germany		0.2 (0.4)
	Italy		−1.1 (0.4)**
	Netherlands		−1.5(0.4)**
	Spain		0.3 (0.4)
	Sweden		0.3 (0.5)
	Switzerland		0.2 (0.5)
Educational level	Low	0.5 (0.2)*	0.0 (0.2)
Educational level	High	0.8 (0.2)**	0.3 (0.2)*
Number of children		0.1 (0.1)	0.0 (0.0)
Marital status: not married		0.3 (0.2)	0.0 (0.2)
Health status: not good		−1.0 (0.2)***	−0.2 (0.1)
Self–employed: yes		1.4 (0.3)***	1.1 (0.2)***
Worked until retirement: no		2.2 (0.5)***	0.7 (0.1)***
	Austria	−0.1 (0.8)
	Belgium	1.0 (0.6)
	Czech Republic	−2.0 (0.7)**
	Denmark	1.0 (0.6)
	France	−0.2 (0.8)
	Germany	1.0 (0.7)
	Italy	0.9 (0.6)
	Netherlands	−2.5 (0.7)***
	Spain	0.5 (0.7)
	Sweden	0.2 (0.9)
	Switzerland	0.1 (0.7)
Change in GDP		−0.2 (0.2)	0.3 (0.7)
Unemployment rate		−0.1 (0.1)	0.0 (0.0)
Minimum pension age		0.2 (0.1)*	0.4 (0.1)***

B: regression coefficient; SE: standard error; GDP: gross domestic product.

p < 0.05; **p < 0.01;***p < 0.001; reference categories not shown.

Figure 1 illustrates how the combination of within- and between-country differences developed over time. This figure shows the realized retirement age for the average individual in our sample by gender, country and year. In this figure, the realized retirement age for men and women in 2005/2006 and 2012/2013 is indicated through markers (dots and squares). Countries are indicated through vertical lines and labels at the bottom of the graph. Figure 1 clearly shows that the realized retirement age increased over time and that gender differences disappeared in many countries. The lowest realized retirement age in 2005/2006 was that of German women (57.3 years); in 2012/2013, it was that of Austrian women (59.4 years). The highest realized retirement age in 2005/2006 was that of Czech men (61.5 years), and in 2012/2013, it was that of Danish women (65.3 years). Denmark in 2012/2013 is the only occasion where the realized retirement age of women exceeds that of men. The smallest changes in the realized retirement age occurred in France and Austria, and the biggest ones in Denmark, Germany and Belgium.

Figure 1.

Realized retirement age by gender, country and year.

Discussion and conclusion

This article studies recent changes in the realized retirement age. These changes are important because population ageing makes the realized retirement age a key factor for financially sound pension schemes and a sufficiently large workforce. This article suggests that the realized retirement age shifted because of pension reforms and the 2008 economic crisis. Findings show that fewer people worked until retirement in 2012/2013 than in 2005/2006. Also, as the realized retirement age increased from 2005/2006 to 2012/2013, the individuals’ characteristics became less influential, and countries became more similar.

Findings failed to reject some, but not all hypotheses formulated. They failed to reject both hypotheses about the impact of pension reforms on the realized retirement age. These hypotheses were that pension reforms increase the realized retirement age, and that they make the realized retirement age less dependent on an individual’s characteristics. However, findings indicate that we have to reject both hypotheses about the impact of the 2008 economic crisis on the realized retirement age. These hypotheses were that the crisis lowered the average realized retirement age and the realized retirement age of individuals with a period of non-employment before retirement. This pattern of findings can have one of two reasons. First, it can mean that pension reforms affected the realized retirement, whereas the 2008 economic crisis had no such effect. Second, it can mean that both pension reforms and the 2008 crisis affected the realized retirement age, but that the effects of pension reforms outweighed those of the crisis. Findings showed that the average realized retirement age was increasingly affected by pension regulations, but not by the country’s economic situation. However, this study cannot definitively determine the cause of all changes in the average realized retirement age and in the effects of individual characteristics because both pension reforms and the 2008 crisis occurred at the same time. It can only observe that changes occurred while both events took place. Future studies are needed to separate the effects of the crisis and pension reform. Qualitative studies that explore shifts in the motivations and experiences of older workers might be particularly useful for this purpose.

The findings have important practical and theoretical implications. The foremost theoretical implication is that policymakers need not worry about the economic crisis defeating the goals pursued with pension reforms. Despite the crisis, the realized retirement age is increasing and becoming less dependent on personal characteristics. However, these changes are, among other things, due to changes in early retirement, which no longer results from poor health or a low educational level. Individuals in poor health often have problems in physically carrying out paid work, and particularly individuals with a low educational level have problems finding paid work in old age. Thus, recent developments force people to stay in the labour market who have reasons to retire early. Policymakers may want to consider whether or not these shifts in social inequalities in retirement are desirable.

One theoretical implication concerns social inequalities and the meaning of retirement. The characteristics of the people who retired in 2005/2006 differ markedly from those of people who retired in 2012/2013. Most notably, in 2012/2013, fewer people worked until they retired. This shift in characteristics is probably due to the economic crisis. However, the analysis showed that social inequalities in the realized retirement age decreased at the same time. In other words, people became more unequal, but these inequalities became less relevant for the realized retirement age. While the retirement age now looks more homogeneous, it actually had a more diverse meaning. Most notably, it now means ‘stopping the activity of working for pay’ for fewer people, and instead means ‘receiving pension benefits after a period of not working’ for more people. It would be worthwhile if future studies further investigated these changes in the meaning and in the pathways to retirement. Such studies seem fruitful undertakings for research on labour markets, ageing, life-courses, social inequalities and culture.

A second theoretical implication concerns the specificity of research findings. This study showed that influences on the realized retirement age differ across time and place. In other words, what was true at one point in time may no longer be true even a few years later. Also, what is true in one country may not be true in another country – despite the country convergence. These findings are in line with the idea that lives are rooted in time and place, which life-course studies propose (Settersten, 1999). As a consequence, researchers writing literature reviews need to pay special attention to the date and place where studies were carried out. If they suspect that social change might have occurred since a study was carried out or that countries differ markedly, then they should be very circumspect when referencing the findings of this study.

Besides its merits, this study also has some limitations. First and foremost, this study cannot separate social change caused by pension reforms from the one caused by the 2008 crisis. Because both events occurred simultaneously, this study can only report the overall effect. Future studies which separate the effects of both events would be valuable. Second, this study does not consider paid work after retirement. It is possible that the economic crisis did not only affect when people retired but also whether people worked or plan to work after retirement (Walsh et al., 2015). A separate research project focusing on post-retirement work and the economic crisis would be interesting. Third, this study uses only two indicators for the economic crisis: the change in GDP and the unemployment rate. Even though these indicators are commonly used, they are neither universally valid crisis indicators, nor do they represent all possible ways in which the 2008 crisis may affect the realized retirement age. For example, some countries saw changes in their unemployment rates and in the change in GDP in the wake of the crisis, whereas other countries saw changes in only one of these indicators. Also, some publications argue that crisis effects may unfold because of, for example, drops in the perceived job security (Lubcke and Erlinghagen, 2014). A separate study that investigates through which mechanisms the 2008 crisis affects retirement would be worthwhile.

Summing up, this study sheds light on shifts in the realized retirement age. It demonstrates that the realized retirement age increased and became more uniform when pension reforms and economic crisis coincided. Thus, the observed shift has both a quantitative and a qualitative dimension. Policymakers can consider the findings to be an indication that reforms aiming to delay retirement are still realized. However, policymakers and researchers also need to consider shifts in social inequalities and in the meaning of retirement. Moreover, researchers need to scan for possible effects of social change when conducting literature reviews.

Footnotes

Funding

The author(s) received no financial support for the research, authorship and/or publication of this article.

References

Bernal

Vermeulen

(2014) ‘The Impact of an Increase in the Legal Retirement Age on the Effective Retirement age’, De Economist 162(2): 115–45.

Börsch-Supan

(2015) ‘Challenges for European Welfare States’, International Tax and Public Finance 22(4): 534–48.

Börsch-Supan

Brandt

Hunkler

Kneip

Korbmacher

Malter

et al . (2013) ‘Data Resource Profile: The Survey of Health, Ageing and Retirement in Europe (SHARE)’, International Journal of Epidemiology 42(4): 992–1001.

Coile

C.C.

Levine

P.B.

(2011) ‘The Market Crash and Mass Layoffs: How the Current Economic Crisis May Affect Retirement’, The B.E. Journal of Economic Analysis & Policy 11(1): 22.

Deighton

(1998) ‘The Remaking of Europe’, in Howard

Louis

(eds) The Oxford History of the Twentieth Century, pp. 190–202. New York: Oxford University Press.

Dorn

Sousa-Poza

(2010) ‘‘Voluntary’ and ‘Involuntary’ Early Retirement: An International Analysis’, Applied Economics 42(4): 427–38.

Ebbinghaus

(2006) Reforming Early Retirement in Europe, Japan and the USA. Oxford: Oxford University Press.

Elder

Jr (1999) Children of the Great Depression: Social Change in Life Experience. Boulder, CO: Westview Press.

Erickson

R.J.

(2005) ‘Why Emotion Work Matters: Sex, Gender, and the Division of Household Labor’, Journal of Marriage and Family 67(2): 337–51.

10.

European Commission (2015) ‘Mutual Information System on Social Protection: Comparative Tables Database’, Available at www.missoc.org/MISSOC/INFORMATIONBASE/COMPARATIVETABLES/MISSOCDATABASE/comparativeTableSearch.jsp (accessed 7 July 2016).

11.

Evandrou

Glaser

(2003) ‘Combining Work and Family Life: The Pension Penalty of Caring’, Ageing & Society 23(5): 583–601.

12.

Fasang

(2010) ‘Retirement: Institutional Pathways and Individual Trajectories in Britain and Germany’, Sociological Research Online 15.

13.

Flynn

(2010) ‘Who Would Delay Retirement? Typologies of Older Workers’, Personnel Review 39(3): 308–24.

14.

Gelman

Hill

(2007) Data Analysis Using Regression and Multilevel/Hierarchical Models. Cambridge: Cambridge University Press.

15.

Gouldner

A.W.

(2001) ‘The Future of Intellectuals and the Rise of the New Class’, in Grusky

D.B.

(ed.) Social Stratification: Class, Race and Gender in Sociological Perspective, pp. 817–30. Boulder, CO: Westview Press.

16.

Gustman

A.L.

Steinmeier

T.L.

Tabatabai

(October 2009) ‘What the Stockmarket Decline Means for the Financial Security and Retirement Choices of the Near-retirement Population’, NBER Working Paper 15435. Cambridge: National Bureau of Economic Research

17.

Harrell

F.E.

(2001) Regression Modeling Strategies. New York: Springer.

18.

Hartlapp

Kemmerling

(2008) ‘When a Solution Becomes the Problem: The Causes of Policy Reversal on Early Exit from the Labour Force’, Journal of European Social Policy 18(4): 366–79.

19.

Hemerijck

Knapen

Van Doorne

(eds) (2009) Aftershocks: Economic Crisis and Institutional Choice. Amsterdam: Amsterdam University Press.

20.

International Social Security Association (ISSA) (2015) ‘Social Security Country Profiles’, Available at https://www.issa.int/en_GB/country-profiles (accessed 17 August 2015).

21.

Jones

A.M.

Rice

Roberts

(2010) ‘Sick of Work or Too Sick to Work? Evidence on Self-reported Health Shocks and Early Retirement from the BHPS’, Economic Modelling 27(4): 866–80.

22.

Kohli

Arza

(2011) ‘The Political Economy of Pension Reform in Europe’, in Binstock

R.H.

George

L.K.

(eds) Handbook of Aging and the Social Sciences, pp. 251–64. London: Elsevier.

23.

Lubcke

Erlinghagen

(2014) ‘Self-Perceived Job Insecurity across Europe over Time: Does Changing Context Matter?’, Journal of European Social Policy 24(4): 319–36.

24.

Lund

Villadsen

(2005) ‘Who Retires Early and Why? Determinants of Early Retirement Pension among Danish Employees 57–62 years’, European Journal of Ageing 2(4): 275–80.

25.

McDaniel

S.A.

Gaszo

(2013) ‘Generationing Relations in Challenging Times: Americans and Canadians in Mid-life in the Great Recession’, Current Sociology 61(3): 301–21.

26.

Mäkikangas

De Cuyper

Mauno

Kinnunen

(2013) ‘A Longitudinal Person-Centered View on Perceived Employability: The Role of Job Insecurity’, European Journal of Work and Organizational Psychology 22(4): 490–503.

27.

Myles

(2002) ‘A New Social Contract for the Elderly?’, in Esping-Andersen

Gallie

Hemerijck

Myles

(eds) Why We Need a New Welfare State, pp. 130–72. Oxford: Oxford University Press.

28.

Organisation for Economic Co-operation and Development (OECD) (2005) Ageing and Employment Policies: Netherlands. Paris: OECD Publishing.

29.

Organisation for Economic Co-operation and Development (OECD) (2006) Ageing and Employment Policies: Live Longer, Work Longer. Paris: OECD Publishing.

30.

Organisation for Economic Co-operation and Development (OECD) (2011) Pensions at a Glance 2011. Paris: OECD Publishing.

31.

Organisation for Economic Co-operation and Development (OECD) ( 2013) ‘Average Effective Age of Retirement in 1970-2012 in OECD Countries’, Available at http://www.oecd.org/els/emp/Summary_1970%20values.xls(accessed 17 August 2015).

32.

Organisation for Economic Co-operation and Development (OECD) (2016a) ‘Gross Domestic Product’, Available at http://stats.oecd.org/index.aspx?DatasetCode=SNA_TABLE1 (accessed 7 July 2016).

33.

Organisation for Economic Co-operation and Development (OECD) (2016b) ‘Labour Force Statistics – By Sex and Age: Indicators’, Available at http://stats.oecd.org/Index.aspx?ysisDataSetCode=LFS_SEXAGE_I_R (accessed 7 July 2016).

34.

Radl

(2013) ‘Labour Market Exit and Social Stratification in Western Europe: The Effects of Social Class and Gender on the Timing of Retirement’, European Sociological Review 29(3): 654–68.

35.

Sarfati

(2013) ‘Coping with the Unemployment Crisis in Europe’, International Labor Review 152(1): 145–56.

36.

Schils

(2008) ‘Early Retirement in Germany, the Netherlands, and the United Kingdom: A Longitudinal Analysis of Individual Factors and Institutional Regimes’, European Sociological Review 24(3): 315–29.

37.

Settersten

(1999) Lives in Time and Place: The Problems and Promises of Developmental Science. Amityville, NY: Baywood.

38.

Syse

Solem

P.E.

Ugreninov

Mykletun

Furunes

(2014) ‘Do Spouses Coordinate Their Work Exits? A Combined Survey and Register Analysis from Norway’, Research on Aging 36(5): 625–50.

39.

Trifiletti

(1999) ‘Southern European Welfare Regimes and the Worsening Position of Women’, Journal of European Social Policy 9(1): 49–64.

40.

Turek

Perek-Bialas

Stypinska

(2015) ‘Socioeconomic Status in Poland: A Question of Cumulative Advantages and Disadvantages’, in Komp

Johansson

(eds) Population Ageing from Lifecourse Perspective: Critical and International Approaches, pp. 85–106. Bristol: Policy Press.

41.

United Nations (2015) ‘Elderly Dependency Ratio (per cent)’, Available at http://data.un.org/Data.aspx?d=PopDiv&f=variableID%3A44 (accessed 17 August 2015).

42.

Van Bavel

De Winter

. (2013) ‘Becoming a Grandparent and Early Retirement in Europe’, European Sociological Review 29(6): 1295–308.

43.

Vanhuysse

(2006) Divide and Pacify: Strategic Social Policies and Political Protests in Post-Communist Democracies. Budapest: Central European University Press.

44.

Walsh

Carney

G.M.

Leime

A.M.

(eds) (2015) Ageing through Austerity: Critical Perspectives from Ireland. Bristol: Policy Press.

45.

Zavras

Tsiantou

Pavi

Mylona

Kyriopoulos

(2012) ‘Impact of Economic Crisis and Other Demographic and Socio-Economic Factors on Self-Rated Health in Greece’, The European Journal of Public Health 23(2): 206–10.