M E A d i s c u ss i o n pA pE rs Earnings test, non-actuarial adjustments and flexible retirement Axel Börsch-supan, Klaus Härtl, duarte n. Leite 06-2017 mea – Amalienstr. 33_d-80799 Munich_phone+49 89 38602-355_Fax +49 89 38602-390_www.mea.mpisoc.mpg.de Earnings test, non-actuarial adjustments and flexible retirement Axel Börsch-supan, Klaus Härtl, duarte n. Leite Abstract: in response to the challenges of population aging, many countries have introduced gradual increases of the statutory eligibility age and shut down pathways to early retirement. There are, however, many incentives left which create early retirement, in particular earnings tests and less than actuarial adjustment rates, both of which are still frequent in Europe. This paper analyzes whether and to what extent abolishing earnings tests will strengthen the sustainability of public pension systems. We employ a life-cycle model of consumption and labor supply where the choices of labor force exit and benefit claiming age are endogenous and potentially separate. Earnings tests force workers to exit the labor market when claiming a pension. After abolishing the earnings test, workers can claim their benefits and can keep on working, potentially increasing labor supply. The difference between exit and claiming age strongly depends on the actuarial neutrality of the pension system and can become very large. We show that abolishing an earnings test as part of a “flexibility reform” creates more labor supply but at the same time reduces the average claiming age when adjustments remain less than actuarial, thereby worsening rather than improving the sustainability of public pension systems. Keywords: population aging, social security, public insurance, pension reform, life-cycle saving, labor supply, retirement age JEL Classification: d91, E17, E21, H55, J11, J22, J26 Zusammenfassung: in reaktion auf die Herausforderungen einer alternden Bevölkerung haben viele Länder das gesetzliche renteneintrittsalter schrittweise angehoben und pfade in die Frührente geschlossen. Es gibt jedoch immer noch viele Anreize zur Frühverrentung in Europa, insbesondere Hinzuverdienstgrenzen und Zu- bzw. Abschläge, die zu gering sind, um anreizkompatibel zu sein. diese studie analysiert, ob und in welchem Ausmaß die Abschaffung von Hinzuverdienstgrenzen die nachhaltigkeit von öffentlichen rentensystemen stärken wird. Wir verwenden ein Lebenszyklusmodell, das Konsum und Arbeitsangebot beschreibt und bei dem die Entscheidungen über das Alter, zu dem man aus dem Arbeitsmarkt ausscheiden möchte, und über das Alter, zu dem man beginnen will, rente zu beziehen, endogen und potentiell verschieden sind. Hinzuverdienstgrenzen zwingen die Arbeitnehmer den Arbeitsmarkt zu verlassen, sobald sie ihre renten beziehen. Wenn diese Hinzuverdienstgrenzen aufgehoben sind, können Arbeitnehmer renten beziehen und gleichzeitig arbeiten, was tendenziell das Arbeitsangebot erhöhen sollte. der unterschied zwischen Arbeitsmarktaustrittsalter und renteneintrittsalter hängt stark von der Anreizneutralität des rentensystems ab und kann sehr groß werden. Wir zeigen, dass die Abschaffung von Hinzuverdienstgrenzen als Teil einer „Flexibilitätsreform“ das Arbeitsangebot erhöht, aber zugleich auch das durchschnittliche renteneintrittsalter verringert, wenn die Zu- und Abschläge niedriger als anreizkompatibel sind. in diesem Fall würde die finanzielle nachhaltigkeit der öffentlichen rentensysteme verschlechtert anstatt verbessert. Acknowledgements: We are grateful to Alexander Ludwig for advice on the computational algorithm. We thank Tabea Bucher-Koenen, Alan Gustman and Johannes Rausch for their helpful comments on earlier versions. The usual disclaimer applies. Earnings test, non-actuarial adjustments and flexible retirement Axel Börsch-Supan a,b,c , Klaus Härtl a,d , Duarte N. Leite a,e a Munich Center for the Economics of Ageing at the Max-Planck-Institute for Social Law and Social Policy b National Bureau of Economic Research (NBER) c Technical University of Munich (TUM) d SAFE, Goethe University Frankfurt e Center for Economics and Finance at the University of Porto (CEF.UP) This version: 31 August 2017 Abstract In response to the challenges of population aging, many countries have introduced gradual increases of the statutory eligibility age and shut down pathways to early retirement. There are, however, many incentives left which create early retirement, in particular earnings tests and less than actuarial adjustment rates, both of which are still frequent in Europe. This paper analyzes whether and to what extent abolishing earnings tests will strengthen the sustainability of public pension systems. We employ a life-cycle model of consumption and labor supply where the choices of labor force exit and benefit claiming age are endogenous and potentially separate. Earnings tests force workers to exit the labor market when claiming a pension. After abolishing the earnings test, workers can claim their benefits and can keep on working, potentially increasing labor supply. The difference between exit and claiming age strongly depends on the actuarial neutrality of the pension system and can become very large. We show that abolishing an earnings test as part of a “flexibility reform” creates more labor supply but at the same time reduces the average claiming age when adjustments remain less than actuarial, thereby worsening rather than improving the sustainability of public pension systems. JEL Classification: D91, E17, E21, H55, J11, J22, J26 Keywords: Population aging, social security, public insurance, pension reform, life-cycle saving, labor supply, retirement age 2 1. Introduction Population aging in most developed countries has posed major challenges to policy makers as they struggle to keep the social security systems sustainable. As a consequence, much attention has been given to labor supply at older ages since working longer helps to decrease a pension system’s dependency ratio. Many countries have introduced gradual increases of the statutory eligibility age and shut down pathways to early retirement. This has, among other factors (Coile et al., 2018), led to a striking reversal around the year 2000 of the long-term trend to ever earlier retirement observed since the 1970s (Figure 1). Figure 1 – Labor force participation among men 60-64 Source: OECD Employment Data (2017) Nevertheless, the percentage of individuals who retire early is still large, regardless of the statutory eligibility age implemented. Except for Sweden, all European countries depicted in Figure 1 feature less than 65% labor force participation among men aged 60-64. In France, Belgium, Italy and Spain the majority of men in this age range have already retired. In Germany, which has experienced the strongest reversal in labor force participation, the actual average retirement age is about 62.1 years, more than 3 years earlier than the current statutory eligibility age which is 65 years and 3 months (Börsch-Supan et al., 2016; OECD, 2015). This observation is in line with the many incentives left which create early retirement, in particular earnings tests and less than actuarial adjustment factors to the benefits claiming age. Their interaction is the subject of this paper. 3 Earnings tests are a specific form of means-testing and impose an upper limit on earnings while receiving a pension. In their most stringent form, earnings tests force workers to exit the labor market when claiming a pension. In contrast, without an earnings test, workers could claim their benefits and continue working, thus potentially increasing labor supply. Earnings tests have been abolished (e.g. in the US, Canada and the UK) quite some time ago. Norway has been the most recent European country to follow their example. A large empirical literature has concentrated its efforts on identifying the effects of the abolishment of earnings tests on labor supply. Studies such as Friedberg (2000), Tran (2002), Gustman and Steinmeier (2008), Michaud (2008), Haider and Loughran (2008), Friedberg and Webb (2009) and Engelhardt and Kumar (2009) for the US, Baker and Benjamin (1999) for Canada, Disney and Smith (2002) for the UK, Shimizutani and Oshio (2013) for Japan, and Brinch et al. (2012) and Hernaes and Jia (2013) for Norway show that the abolishment of an earnings test has led to an increase in labor supply. Table 1, however, shows that many European countries still have earnings tests for individuals who retire before the statutory eligibility age (SEA), e.g. France or Germany, with different limits and different rules (Social Security Administration, 2014). This table also shows how diverse and relatively small adjustment factors are between countries. This poses several challenges on the pension system that we address later in this paper. Adjustment factors link the pension benefit to the age at which individuals begin claiming pension benefits. They reduce benefits by a certain percentage (“adjustment rate”) when an individual claims pension benefits earlier than the statutory eligibility age, and increase benefits when claiming benefits is postponed after the statutory eligibility age. If an individual’s choice of claiming age should be neutral to the sustainability of the pension system, these adjustment factors must be actuarially neutral, i.e., they should equalize the present discounted value of pension benefits across all permissible claiming ages. Depending on age and life expectancy, actuarial adjustment rates are between 6.5 and 8 percent (Börsch-Supan, 2004; Queisser & Whitehouse, 2006; Werding, 2007 and 2012; Gasche, 2012; OECD, 2015). In most European countries, however, they are substantially lower see Table 1. There is a substantial body of evidence showing that smaller than actuarially neutral adjustments exert large incentives to claim benefits earlier than the statutory eligibility age and that this has significantly contributed to the early retirement visible in Figure 1 (e.g. Gruber and Wise, 1999 and 2004; Blondal and Scarpetta, 1999; Börsch-Supan, 2000). 4 Table 1 – Adjustments to retirement age and earnings tests across selected countries Adjustment rate Earnings tests Austria 4.2% Before SEA: when earnings are above a ceiling of 290 € per month, the pension is fully withdrawn; After SEA: no limit Belgium - 1 Before SEA: when annual earnings are above 7,793€ (single) or 11,689 € (dependent child) per year, the pension is reduced by the amount that exceeds the limit. If annual earnings are 25% above the limit, the pension is fully withdrawn for as long as the additional income is higher than the ceiling; After SEA: when earnings are above 22,509 € (single) or 27,379 € (dependent child) per year, the pension is reduced by the amount that exceeds the limit. If annual earnings are 25% above the limit, the pension is fully withdrawn for as long as the additional income is higher than the ceiling. For a retiree older than 65 with at least 42 years of contribution, the ceiling is lifted entirely Denmark - 2 Before SEA: no public pension receipt possible, therefore no conflict between public pension benefits and additional income; After SEA: full basic pension (795€ per month or 9,540€ per year, which is equivalent to around 17% of average earnings) is reduced at a rate of 30% against earned income, if work income exceeds 40,518€ per year (approx. ¾ of average earnings) Finland 4.8% No limit France 5.0% No limit for full pension recipients; workers are eligible for full public pension benefits if they fulfil either both a minimum contributory record (in 2014: 41.25 years for people born in 1953) and the minimum legal pension age (61 years and two months) or the age of 66 years and two months Germany 3.6% Before SEA: for drawing full pension payments the limit is one-seventh of the reference base (i.e. 3,060€ per year or 255€ per month respectively); for drawing a partial pension the ceiling is dependent of the partial pension level, i.e. 1,483€ per month (1/3 partial pension), 1,112€ per month (1/2 partial pension), 741€ per month (2/3 partial pension), multiplied with the individual earnings points in the year before pension claiming; After SEA: no limit Netherlands - 3 Before SEA: no public pension receipt possible, therefore no conflict between public pension benefits and additional income; After SEA: no ceiling on additional earnings for public pension recipients Norway 3.8-4.7% No limit Sweden 4.1-4.7% No limit UK 5.0-6.67% No limit 1 In Belgium, there is no actuarial reduction in the pension calculation in case of early retirement. However, the pension of the early retirees can be incomplete if they have worked less than 45 years. 2 In Denmark, early claiming of pension benefits is not possible. 3In the Netherlands, it is not possible to claim public pensions before the SEA. Early retirement is financed either by private savings or by occupational pensions. Sources: Queisser & Whitehouse (2006) and Börsch-Supan et al. (2016). 5 Abolishing earnings tests and making adjustment factors closer to their actuarially neutral value are therefore often proposed as policies to increase old-age labor supply and thereby strengthen the sustainability of public pension systems. They are elements of a recent policy agenda towards more flexibility in retirement choices to promote active aging and a longer working life (Graf et al., 2011; Huber et al., 2013; Sonnet et al., 2014; Börsch-Supan et al., 2017). This paper employs a life-cycle model of consumption and labor supply to study the interaction between earnings tests and actuarial adjustments during the window of retirement when workers are allowed to choose their retirement age. A key feature of our model is that households may decide separately on their claiming age R and their exit age from the labor force X (𝑅 ⋛ 𝑋), subject to the rules and parameters of the pension system. The main aim of the paper is to study the reaction of these two potentially separate ages R and X to the parameters of the pension system and other determinants driving the retirement decision. Among the parameters of the pension system, we focus on the adjustment factors which link the pension benefit to the chosen claiming age. Other determinants influencing the retirement behavior include declining health, declining productivity, increasing appreciation of leisure versus consumption, and fixed costs of working. Our contribution to the literature combines four aspects. First, we model the three decisions to claim benefits, to choose working hours (intensive margin of labor supply) and to exit the labor force (extensive margin) as separate decisions in a unified life-cycle framework. Second, we offer several alternative mechanisms that create the abrupt jump in the hours’ supply to zero when exiting the labor force, including a minimum hours’ constraint generated by fixed time costs of working. Third, we do not restrict our analysis to the effects of abolishing an earnings test on labor supply but also predict the implications for the financial sustainability of a prototypical public pay-as-you-go pension system. Fourth, we extend this analysis from the actuarially neutral case to the case of distortive adjustment factors which are more typical for the pension systems in Europe. These four aspects are essential to better understand in which institutional setting a reform towards more flexibility will not only boost labor supply but also strengthen the financial sustainability of pension systems. We show that the difference between exit and claiming age strongly depends on the preference for consumption versus leisure and can become very large. This is in line with previous literature and our intuition on how workers tend to react to stimuli that affect their preferences for consumption, leisure and savings. There are, however, also interaction effects that are more complex and not covered by the existing literature. Most importantly, the difference between exit 6 and claiming age strongly depends on the actuarial neutrality of the pension system. We show, for instance, that abolishing an earnings test will reduce the average claiming age when adjustments are less than actuarial, hence worsening rather than improving the sustainability of public pension systems. This effect is not compensated by the increase in labor supply. Our model is in the tradition of several theoretical models that have studied the abolishment of the earnings test in the Anglo-Saxon countries on labor supply (Gustman and Steinmeier, 2008; French, 2005; Benitez-Silva and Heiland, 2007; Michaud, 2008; Fehr and Uhde, 2014; Kudrna and Woodland, 2011). With one exception, these papers do not address the implications for the financial sustainability of the public pay-as-you-go pension systems and they take place in an actuarially neutral pension system. The closest paper to ours is Gustman and Steinmeier (2008) which models the distinction between claiming and exiting the labor force in a very rich dynamic programming model. This model includes heterogeneity of workers’ preferences and job characteristics in order to obtain a good fit to US data. While Gustman and Steinmeier implicitly consider the parameters of the US Social Security system, they do not vary them except for the abolishment of the earnings test. Hence, their analysis refers to a largely actuarial neutral institutional environment which is in stark contrast to the situation in Europe. Our model provides a unified framework which nests and explains a host of empirical findings which have been provided by the literature. Regarding the claiming decision, Gustman and Steinmeier (2008) show that the abolishment of the earnings test increases the share of married men who claim their benefits by about 10% between the early and normal retirement ages. Gruber and Orzag (2003), Song (2004) and Song and Manchester (2007) also find evidence that loosening the earnings test accelerates the claiming of benefits among the eligible population.1 Regarding labor supply, we already mentioned the large number of studies which showed that the abolishment of the earnings tests in the US and the UK had significant positive effects on labor supply. Some of these papers claim that the increase in labor supply stems mostly from the intensive margin, i.e., the decision of working more hours than before, e.g. Disney and Smith (2002), Engelhardt and Kumar (2009), Friedberg (2000), Song (2002), Tran (2002), Hernaes et al (2016) or Hernaes and Jia (2013). Others find that the positive effect is mainly generated on the 1 Michaud (2008) finds no significant evidence on an impact on the claiming age after the elimination of earnings tests. 7 extensive margin, i.e., by an increase in labor force participation, e.g. Baker and Benjamin (1999), Tran (2002), Engelhardt and Kumar (2009), Hernaes et al (2016), Hernaes and Jia (2013), Michaud (2008), Friedberg and Webb (2009).2 Our paper is also related to the literature on means-testing pension benefits against assets (Bütler et al., 2016; Chomik et al., 2015; Fehr and Uhde, 2014; Woodland, 2016). In contrast to our study which investigates earnings tests, these studies concentrate on wealth tests when applying for pension benefits. More specifically, Kudrna and Woodland (2011) and Tran and Woodland (2014) focus on the Australian superannuation scheme and examine the impact of means testing on the incentives of individuals to save and work, on government financial commitments and on the distributional effects on the welfare of individuals. Finally, our paper relates also to the literature on minimum hours constraints (e.g., Gustman and Steinmeier, 2004; Gielen, 2009). We generate such a constraint through fixed time costs of working. The paper is organized as follows. In Section 2, we present our model. Section 3 describes the calibration strategy and our computational solution method. The analysis of our results and sensitivity analyses are presented in Section 4. It includes a prototypical “flexibility reform” which looks promising from a political point of view but is likely to fail in securing the sustainability of a pay-as-you-go pension system. Section 5 concludes the paper. 2. The model Since we want to focus on the households’ labor supply decisions in a complex pension system, we will operate in a partial equilibrium framework in which wages and interest rates are exogenously fixed.3 Our model can be thought of as the household sector plus the pension system in the well-known general equilibrium framework developed by Auerbach and Kotlikoff (1987). Regarding the household model, we need to extend this typical neoclassical set-up by an endogenous retirement decision (extensive margin: labor force participation) in addition to the choice of labor supply (intensive margin: working hours). Regarding the pension system, we need 2 Song (2004), Song and Manchester (2007) and Gruber and Orzag (2003) find no significant effects of the abolishment of the earnings test on employment and/or hours worked in the US. 3 We will thus disregard the general equilibrium effects of abolishing an earnings test on wages and interest rates. Woodland (2016) provides a general equilibrium analysis of means testing and other forms of taxation. While the wage effects are small, the interest rate effects are larger but not in the focus of this labor-supply oriented paper. 8 additional detail by modeling earnings tests and adjustment factors as bonuses for late retirement and penalties for early retirement. Since the main goal of our study is to explain the mechanism and reaction of individuals to earnings tests and different adjustment factors, we abstain from any distributional considerations, e.g. regarding differences by gender, occupational groups or any other differentiable forms between individuals. 2.1 Household consumption and leisure We begin with the traditional set-up of how households choose between consumption and leisure. Households of age j gain utility from consumption 𝑐𝑗 and leisure 𝑙𝑗 according to a CES- type per-period utility function given by 𝒖(𝒄𝒋, 𝒍𝒋) = 𝟏 𝟏−𝜽 [𝒄 𝒋 𝝓𝒋 ∗ 𝒍𝒋 𝟏−𝝓𝒋] 𝟏−𝜽 , (1) where 𝜙𝑗 denotes the utility weight of consumption versus leisure and can be modeled as age- dependent (see Section 3.2). Risk aversion is described by the parameter 𝜃. Households are neoclassical life-cyclers with perfect foresight. They solve a von Neumann- Morgenstern (VNM) expected utility maximization program over the entire life-cycle which lasts for a maximum of J years. The life-time maximization problem of a cohort is therefore given by: 𝒎𝒂𝒙 ∑ 𝜷𝒋−𝟏𝝈𝒋 𝑱 𝒋=𝟏 𝒖(𝒄𝒋, 𝒍𝒋), (2) where β is the pure time discount factor, 𝛽 = 1 1+𝜌 . In addition to pure discounting, households discount future utility with their unconditional survival probability, 𝜎𝑗, expressing the uncertainty about the time of death. We do not include intended bequests in our model and assume that accidental bequests resulting from premature death are taxed away by the government at a confiscatory rate and used for otherwise neutral government consumption. The household’s disposable non-asset income 𝑦𝑗 is 𝒚𝒋 = 𝒉𝒋𝒘𝒋(𝟏 − ) + 𝒑𝒋, (3) which has two components. The first term of the right-hand side reflects labor income (hours worked, ℎ𝑗 = 1 − 𝒍𝑗, multiplied by the net wage, 𝑤𝑗, each age specific) while the second term is pension income. Denoting total assets by 𝑎𝑗, maximization of the household’s intertemporal utility is subject to a dynamic budget constraint given by 9 𝒂𝒋+𝟏 = 𝒂𝒋(𝟏 + 𝒓) + 𝒚𝒋 − 𝒄𝒋. (4) In this traditional set-up, labor supply (working hours) may decline at the intensive margin if the parameters in the utility function change. There is, however, no sudden retirement (withdrawal from the labor force at the extensive margin). Retirement is typically assumed to be exogenously dictated by a mandatory retirement age R at which individuals must stop working and will begin receiving pension benefits. This implies that pj = 0 for 𝑗 ≤ 𝑅 and hj = 0 for j> R in equation (3). 2.2 The retirement decision Modern pension systems deviate from this rigid set-up. First, most pension systems have a window of retirement defined by an earliest and a latest eligibility age 𝑅𝐸 ≤ 𝑅 ≤ 𝑅𝐿 which bracket the statutory or “normal” eligibility age 𝑅. Workers have the choice to retire within this window which we need to model. Second, “flexibility reforms” permit combinations of work and pension benefit receipt both before and after the statutory retirement age. In this case, “retire” refers to two separate decisions, namely to stop working at age X and to begin receiving pension benefits at age R. Both decisions are influenced by common determinants such as institutional parameters and the individual’s preferences. Earnings tests and mandatory retirement may enforce 𝑅 = 𝑋. In other institutional settings, however, R may be earlier or later than X. If workers have saved sufficiently, they may want to stop working before they receive pension benefits (X 5%), workers shift their claiming age to very late ages around 67-70 in order to benefit from higher pension payments. The optimal claiming age depends on 𝜔𝑅 relative to the discount rate including mortality risk. 21 Table 4 – Claiming age R as a function of actuarial adjustment – no earnings test Actuarial adjustment rate (𝜔) Initial level of utility weight of consumption (𝝓) 0% 3.6% 5% 6.3% 7% Low (0.55) 60 60 65 69 69 Middle (0.60) 60 60 64 67 69 High (0.65) 60 60 65 68 70 Source: own calculations. Regarding exit age decisions, households work until the utility from consumption is dominated by the utility of leisure and labor costs. As Table 5 shows, labor force exit age X is decoupled from the benefit claiming age R in Table 4. Exit ages are mostly higher than claiming age, which means that workers request their pension benefits but keep working for some more years before exiting the labor market. The duration of labor supply beyond claiming age strongly depends on the preference for consumption, expressed by the parameter 𝜙. The higher the preference for consumption, the longer workers remain in the labor force. Moreover, it depends on total income (wages plus pensions). The higher the preference for consumption, the longer individuals work and receive wages along with pensions. The difference between claiming and exit ages diminishes with increasing adjustment rates which results from labor costs and decreasing consumption preferences with age. Otherwise, the impact of the adjustment rates on exit ages has no general pattern since exit ages are only indirectly affected by them. This is in stark contrast to claiming ages which are directly dependent on the adjustment rates. 22 Table 5 – Exit age X as a function of actuarial adjustment – no earnings test Actuarial adjustment rate (𝜔) Initial level of utility weight of consumption (𝝓) 0% 3.6% 5% 6.3% 7% Low (0.55) 69 69 70 67 67 Middle (0.60) 73 72 73 75 76 High (0.65) 85 79 79 85 85 Source: own calculations. Figure 5 and the subsequent tables have shown retirement and labor force exit ages. They represent the extensive margins of labor supply. We now turn to the intensive margin (hours worked if participating, Figure 6). The product of both margins (total labor volume) is a key variable for a PAYG system since it directly determines a country’s overall wage bill from which proportional contributions are paid. Figure 6 – Labor supply for different adjustment rates with and without earnings test Source: own calculations. Figure 6 shows that abolishing an earnings test has positive impacts on total labor volume in terms of both margins. Labor force exit ages are shifted to older ages and hours are increased relative to the scenarios with an earnings test imposed. It would be premature, however, to conclude that this higher labor volume helps to stabilize PAYG pension systems by lowering contribution rates. As we have seen in Figure 5, claiming age decreases when the adjustments rates are smaller than actuarially neutral. This increases pension expenditures and thus 23 contribution rates. The combined effect is show in Figure 7. It shows that contribution rates are higher after the abolishment of an earnings test for low adjustment rates and decrease only for adjustment rates that are actuarially neutral (around 6.3%) or higher. Figure 7 – Contribution rates with and without earnings test Source: own calculations. 4.2. Flexibility reform and partial retirement Some recent “flexibility reforms” and “partial retirement” proposals entail a slightly modified scenario of how benefits are calculated when individuals keep working after having claimed pension benefits. Examples are the abolition of the earnings test in 2000 in the US (Social Security Administration, 2008) and the proposal for a “Flexi-Rente” announced by the German government (Bundesgesetzblatt, 2016). These reforms abolish the strict earnings test but depart from the scenario presented in the previous subsection by collecting pension points after claiming pensions which is not permitted in traditional systems without an earnings test. This modification allows individuals to increase their pension benefits after claiming. On the one hand, this creates even larger incentives to work longer since besides receiving wages individuals also receive higher pension benefits in the end – a double incentive. On the other hand, however, net wages are lower due to contributions to the pension system. While the double incentive has an intuitive appeal to many policy makers, this subsection shows that these proposals have the same negative impact on claiming ages and sustainability of the pension system when they are applied to the existing public pension system with less than actuarially neutral adjustment factors. Figure 8 parallels Figure 5 and provides an overview of the results based on the same benchmark parameters as in the preceding subsection. 24 Figure 8 – Retirement decisions under a flexibility reform Source: own calculations. The flexibility-reform in Figure 8 generates slightly later claiming ages but earlier labor force exit ages compared with the abolishment of the earnings test depicted in Figure 5. At the same time, individuals also work less intensively under a flexibility reform compared to the traditional scenario without an earnings test (Figure 9). Figure 9 – Labor supply under a flexibility reform Source: own calculations. The total effect on the pension system is shown in Figure 10 in terms of the contribution rate. It is in between the two cases shown in the preceding subsection (earnings test imposed and traditional 58 60 62 64 66 68 70 72 74 76 78 0.0% 3.6% 5.0% 6.3% 7.0% R et ir e m e n t ag e s Adjustment rates No earnings test: exit Flexi - Rente: exit With earnings test No earnings test: claim Flexi - Rente: claim 0,15 0,20 0,25 0,30 0,35 0,40 0,45 0,50 55 57 59 61 63 65 67 69 71 73 75 H o u rs w o rk e d Age 3.6% with Earnings Test 3.6% no Earnings Test 3.6% Flexibility reform 7% with Earnings Test 7% no Earnings Test 7% Flexibility reform 25 scenario without an earnings test). The contribution rate is slightly lower after a flexibility reform than after simply abolishing an earnings test if the adjustment rates are lower than actuarially neutral. This poses incentives for earlier claiming in order to benefit from higher net wages after claiming. In turn, the higher total income (wages and pensions) changes the labor/leisure trade-off and makes workers leave the labor force earlier than in the straight no-earnings test scenario. The pension system faces two opposite effects. On the one hand, after claiming, workers can still contribute the pension system benefits through more contributions until later in life which increases receipts to the system. On the other hand, the pension claims of retirees will be higher which increases expenditures. If adjustment rates are actuarially neutral, there is no effect on the contribution rate. If they are higher than actuarially neutral, contributions rates are slightly higher than in the traditional scenario without an earnings test since the costs of the additional pension benefits outweigh the additional contributions. Figure 10 – Contribution rates under a flexibility reform Source: own calculations. In summary, keeping contributions active when working even after claiming benefits as part of a flexibility reform is less harmful in terms of the financial sustainability to the simple abolishment of an earnings test when adjustment rates are less than actuarially neutral – but the effect on the financial sustainability is still negative. As was the case for the simple abolishment of an earnings test, a flexibility reform is not a substitute for making adjustment rates actuarially neutral. 14.0% 14.5% 15.0% 15.5% 16.0% 16.5% 17.0% 17.5% 18.0% 0.0% 3.6% 5.0% 6.3% 7.0% C o n tr ib u ti o n r at e s Adjustment rates No earnings test With earnings test Reform 26 4.3 Sensitivity to alternative retirement mechanisms This subsection investigates whether these harmful effects occur also under different assumptions about the aging process of which we have no good evidence. We simulate systematically all non-benchmark assumptions presented in Subsection 3.2 and show that they yield similar behavior of individuals when faced with the abolishment of an earnings test. Since the flexibility reform presented in the preceding subsection is an intermediate case, we only show the difference between the more extreme cases of Subsection 4.1. We separate the sensitivity analyses into two sets of scenarios. The first set changes the assumptions on the evolution of productivity during the life-cycle under different fixed costs profiles, representing higher or lower effects of declining health on the disutility of work (Table 6). The second set examines different age-dependent preferences on consumption and leisure under different fixed costs profiles (Table 7). We start by comparing vertically the different productivity scenarios (see light grey boxes in Table 6, results displayed in Figure 11). We then compare scenarios horizontally (dark grey boxes in Table 6, results displayed in Figure 12). Table 6 – Scenarios with different productivity profiles and fixed cost levels Productivity Time costs of working Zero Moderate High Flat Appendix Figure 11 Appendix Moderate see Figure 12 Benchmark in Section 4.1 Figure 12 Steep Appendix Figure 11 Appendix Finally, we compare vertically the different preference scenarios (see light grey boxes in Table 7, results displayed in Figure 13). The more extreme scenarios are relegated to an appendix. Table 7 – Scenarios with different consumption preferences and fixed cost levels Consumption preferences Time costs of working Zero Moderate High Flat Appendix Figure 13 Appendix Moderate Figure 12 Benchmark in Section 4.1 Figure 12 Steep Appendix Figure 13 Appendix The first set of sensitivity analyses addresses the different productivity profiles (flat and steep decline). As referred to in Subsection 3.2, different strands of literature show that productivity may present different shapes with age. Börsch-Supan and Weiss (2016) show that 27 productivity tends to not decrease with age. This profile is presented as the flat productivity profile. The other two experiments refer to the estimated benchmark productivity profile and to an even steeper productivity profile (“steep decrease”) where the decline on productivity is stronger after age 50 (Figure 3 in Subsection 3.2). Figure 11 shows that the abolishment of an earnings test with less than actuarial neutral adjustment rates leads to large differences in claiming ages between scenarios with and without an earnings test. Under a flat or steep declining productivity profile and for low adjustment rates, the claiming age is always higher in the earnings test scenario than in the no earnings test scenario (except for adjustment rates of 0% since the incentives for very low values lead to the earliest possible claiming age). Figure 11 – Claiming and exit ages with and without earnings test (different productivity profiles) Source: own calculations. These outcomes are in accordance with the benchmark scenario and support the claim that individuals tend to claim earlier under the no earnings test scenario when adjustment factors are not actuarially neutral. Only for very high adjustment rates, this pattern turns around and claiming ages in the no earnings test scenario are equal or higher than in the earnings test case. As can be seen in Figure 11, this happens in the steep scenario with adjustment rates of 5%. In the flat- productivity simulation, this intersection occurs only for very high adjustment rates. For the flat productivity profile, individuals exit the labor force very late in life because of the constant, high wages at older ages that incentivizes individuals to work longer. 28 When analyzing the effect of different fixed cost profiles on the outcomes of the model (Figure 2), we again obtain the same patterns as we did in the benchmark case. For fixed costs equal to zero (“zero costs”), the differences in claiming ages are large: individuals claim much earlier in the scenario without an earnings test. The intersection occurs, again, only for very high adjustment rates. In the case of high fixed costs (“high costs”), retirement and claiming ages are identical. Figure 12 – Claiming and exit ages with and without earnings test (different fixed costs profiles) Source: own calculations. Two main types of labor force exit behavior are observed when no earnings tests are imposed. While fixed costs of zero lead to very late exit ages, high fixed costs lead to early exit ages. For the highest fixed costs considered (31% of total available time), exit ages become substantially lower (age 61/64) because decreasing health conditions represented by the fixed costs of work force workers to exit the labor market much earlier. In some cases, workers even exit labor force before claiming their pensions. These individuals finance their retirement consumption temporarily with the savings which they have accumulated during their working lives. Retirement ages in the earnings test scenario are quite sensitive to the fixed costs of working. This is not the case for claiming ages when no earnings tests are in place. Claiming ages remain relatively stable whereas the corresponding exit ages decline substantially. This can lead to some very particular cases where the claiming age after flexibilization is higher than the retirement age before flexibilization. However, this is a very specific situation which results from 29 the higher sensitivity of individuals under an earnings test to fixed costs. This happens because individuals do not have as much degrees of freedom as they had when no earnings test was imposed. We conclude from this analysis that individuals released from the restriction of an earning test first prefer to reduce the number of years in the labor force and change their claiming age only when necessary. We now turn to Figure 13 and the third alternative to model the aging process, namely different trajectories of how the relative preference for leisure versus consumption increases with age, modelled by a declining 𝜙𝑗 in the utility function (1). Fixed costs and productivity are set to the benchmark values, i.e. fixed costs are 21% of total available time and productivity is assumed to decline moderately after age 50. We compare the two extreme cases presented in Figure 4 of Subsection 3.2. The first case entails constant preferences during the life-cycle while the second case comprises a decrease that is steeper than in the benchmark case. As it was the case in our earlier sensitivity analyses, claiming ages decrease when abolishing an earnings test and the adjustment rates are not actuarially neutral. These results do not change qualitatively when different slopes of increasing leisure preferences are assumed. In the case of a steep decrease, the intersection takes place between adjustment rates of 5% and 6.3% (Figure 13). When preferences are flat, the intersection occurs for an adjustment rate of 6.3%. Exit ages are also affected by the abolishment of an earnings test, being considerably later than before, as it would be expected. Figure 13 shows that this effect is slightly larger when preferences are constant over the life course than with steeply increasing preferences for leisure, again in line with intuition. 30 Figure 13 – Claiming and exit age decisions with and without earnings test (different consumption preferences) Source: own calculations. In summary, this subsection has shown that the patterns of outcomes and the main messages are preserved when more extreme assumptions about the process of aging and the mechanisms for retirement are assumed. 4.4. Sensitivity to other model parameters As explained in Subsection 3.2, our calibration roughly targeted the actual average age of labor market exit in the large Continental European countries. In this section, we investigate how our results change if different preference and rate price parameters are chosen. We focus on the outcomes of retirement decisions for adjustment rates of 0%, 3.6% and 6.3%. The first part of the analyses assumes that households are less risk averse, namely 𝜃 = 1 rather than 𝜃 = 2. Individuals are less concerned about large income shocks and therefore save less. Results are shown in Table 8. Retirement ages under an earnings test are usually lower for all consumption preferences and adjustment rates. For values close to the actuarially neutral adjustment rate, the claiming age is equal or slightly higher than the retirement age under earnings tests. The predictions regarding the impact of the abolishment of an earnings test on individuals’ behavior remain unchanged to the benchmark case. 31 Table 8 – Sensitivity analysis for 𝜽=1 and r=4% and ρ=2% Earnings test No earnings test (Claim) No earnings test (Exit) Actuarial adjustment rate (𝜔) Actuarial adjustment rate (𝜔) Actuarial adjustment rate (𝜔) Utility weight of consumption (𝝓) 0% 3.6% 7.0% 0% 3.6% 7.0% 0% 3.6% 7.0% Low (0.55) 60 60 69 60 66 69 66 62 59 Middle (0.60) 60 60 69 60 60 69 69 68 62 High (0.65) 60 61 69 60 60 72 71 71 70 Source: own calculations. Table 9 shows the sensitivity to our assumptions regarding interest rates. A lower interest rate makes individuals work longer in order to compensate for the loss of interest income from their savings. They will therefore retire much later even when an earnings test is implemented. With a low non-actuarially neutral adjustment rate, an earnings test yields very early retirement. Abolishing the earnings test makes most individuals claim their pension earlier, but at the same time they leave the labor force later than in our main scenario in Subsection 4.1. These results might be interpreted as an indication of the younger generations’ behavior if current low interest rates were to persist in the longer run. Table 9 – Sensitivity analysis for 𝜽=2 and r =3% and ρ=2% Earnings test No earnings test (Claim) No earnings test (Exit) Actuarial adjustment rate (𝜔) Actuarial adjustment rate (𝜔) Actuarial adjustment rate (𝜔) Utility weight of consumption (𝝓) 0% 3.6% 7.0% 0% 3.6% 7.0% 0% 3.6% 7.0% Low (0.55) 60 64 70 60 62 72 76 75 85 Middle (0.60) 60 66 71 60 62 72 85 85 85 High (0.65) 60 67 72 60 63 72 85 85 85 Source: own calculations. For a higher interest rate, we can observe the same results (Table 10). The patterns observed in the benchmark case now occur already for lower adjustment rates. For instance, for initial values of consumption preferences of 0.6, abolishing an earnings test no longer leads to earlier claiming ages for an adjustment rate of 3.6%. This means that the adjustment rate which equates the claiming ages in the two types of pension systems is now lower than in the benchmark case. 32 The exit age is also slightly lower than in the benchmark case. This is due to higher returns on savings that can make up for lower number of years at work force. Table 10 – Sensitivity analysis for 𝜽=2 and r =5% and ρ=2% Earnings test No earnings test (Claim) No earnings test (Exit) Actuarial adjustment rate (𝜔) Actuarial adjustment rate (𝜔) Actuarial adjustment rate (𝜔) Utility weight of consumption (𝝓) 0% 3.6% 7.0% 0% 3.6% 7.0% 0% 3.6% 7.0% Low (0.55) 60 60 67 60 60 67 66 65 63 Middle (0.60) 60 60 67 60 60 68 69 69 66 High (0.65) 60 62 67 60 60 68 60 60 67 Source: own calculations. In summary, the central conclusions drawn in the benchmark case also hold if different preference and cost parameters are assumed. For lower than actuarially neutral adjustment rates, workers tend to claim their pension earlier rather than later when an earnings test is abolished, worsening the financial sustainability rather than improving it. Exit ages, however, will occur later in life, increasing total labor volume. 5. Conclusions Increasing dependency ratios in aging societies pose a threat to the financial sustainability of pension systems. Policy makers have faced this challenge with several reforms. Among them, increasing the statutory eligibility age is effective but very unpopular. Alternative reforms, particularly if sold to the public under the label of increasing flexibility, are more popular. It is essential to understand how these measures affect labor supply and retirement behavior. These behavioral effects are complex since flexibilization drives a wedge between claiming pension benefits and exiting the labor force. In the best case, more labor supply creates additional resources to finance the pension system; in the worst case, however, such “flexibility reforms” do harm to the sustainability of pension systems because the added flexibility allows individuals to claim pension benefits earlier. It is thus crucial to take behavioral reactions into account. In order to shed some light on this topic, we built a life-cycle model of saving and labor supply under a PAYG pension system that allowed us to study the incentive effects of a pension system on three distinctive decisions of workers: when to claim benefits, how many hours to work (intensive margin) and when to exit the labor force (extensive margin). Several key parameters 33 shape these decisions, mainly institutional parameters such as the adjustment factors to the actual claiming age and the existence of an earnings test, and preference parameters such as the consumption/leisure trade-off which may change with age. Workers tend to exit early from the labor force when an earnings test is enforced. Lifting this restriction appears to be a good way to keep older workers in the labor force and make pension systems more sustainable. The key result of this paper is that this aim is achieved if and only if adjustment factors are actuarially neutral. This result also holds when individuals who keep working after claiming benefits continue to contribute to the pension system with associated benefit increases. This is the case for some flexibility reforms which have actually been enacted, e.g. in Germany and the US. If pension benefits are not actuarially neutrally linked to the claiming age, abolishing an earnings test is indeed harmful to the financial sustainability of a pension system. These conclusions are derived from a theoretical model. Like any model, one may criticize its underlying assumptions, specifications and parameter choices. We have performed a series of sensitivity analyses to ascertain that results are robust under different parameter choices and alternative specifications of the retirement mechanism. The results are also corroborated by the few empirical studies on recent flexibility reforms (Graf et al., 2011; Huber et al., 2013; Börsch- Supan et al. 2015a, 2015b, 2016). They also explain the results from the much larger empirical literature on the abolishment of the earnings tests in the US, Canada and the UK. Our predictions can intuitively be understood as follows. If there is no earnings test, the decisions of when to claim a pension and when to exit the labor force ages are detached. The decision to claim a pension is essentially driven by the adjustment factors which balance additional contributions to be paid by individuals with additional pension benefits later on. Low adjustment factors create incentives for workers to claim early. This incentive is strengthened once the earnings test has been abolished because individuals can now combine their pension benefit with additional wage income. Early claiming, however, means additional years of benefits that have to be financed by the pension system, threatening its sustainability. The gap between claiming and labor force exit age shrinks with higher adjustment factors. If they are actuarially neutral, having or not an earnings test will not influence the decision on claiming age but will provide higher labor supply until later in life. Only in this case, abolishing an earnings test meets the aim of policy makers to strengthen the financial sustainability of a pension system. 34 Abolishing an earnings test must therefore carefully be contemplated by policy makers. Understanding the interplay of the benefit computation with the claiming age right is essential to avoid worsening the financial sustainability of pension systems. Flexibility per se is no substitute to fixing a pension system that creates early retirement incentives due to adjustment factors that are lower than actuarially neutral. Permitting workers to keep contributing to the system in order to earn additional earnings points leads to less harm done to the pension system than simply abolishing the earnings test. The key condition for actually improving the financial sustainability, however, is again an actuarially neutral linkage between benefits and claiming age. 35 References Altig, D., et al. (2001). Simulating Fundamental Tax Reform in the United States. American Economic Review, 91(3), 574-595. Auerbach, A. J., & Kotlikoff, L. J. (1987). Dynamic Fiscal Policy. New York, USA: Cambridge University Press. Bansal, R., & Yaron, A. (2004). Risks for the long run: A potential resolution of asset pricing puzzles. Journal of Finance, 4(4), 1481-1509. Baker, M., & Benjamin, D. (1999). How do retirement tests affect the labour supply of older men? Journal of Public Economics, 71, 27–5. Blondal, S. and S. Scarpetta (1999). Early Retirement in OECD Countries: The Role of Social Security Systems, OECD Economic Studies, No. 29. Börsch-Supan, A. (2004). Faire Abschläge in der gesetzlichen Rentenversicherung. Sozialer Fortschritt, 53(10), 258-261. Börsch-Supan, A., & Stahl, K. (1991). Life Cycle Savings and Consumption Constraints. Journal of Population Economics, 4(3), 233-255. Börsch-Supan, A. (2000): Incentive Effects of Social Security on Labor Force Participation: Evidence in Germany and Across Europe. Journal of Public Economics, Vol. 78, 25-49. Börsch-Supan, A., & Weiss, M. (2016). Productivity and age: Evidence from work teams at the assembly line. Journal of the Economics of Ageing, 7, 30-42. Börsch-Supan, A., Bucher-Koenen, T., Kluth, S., Haupt, M., & Goll, N. (2015a). Vor- und Nachteile höherer Flexibilität als Instrument zur Erhöhung der Beschäftigung Älterer. MEA Discussion Paper, 06-2015. Börsch-Supan, A., Bucher-Koenen, T., Kluth, S., Haupt, M., & Goll, N. (2015b). Internationale Evidenz zu flexiblen Übergängen in den Ruhestand. MEA Discussion Paper, 05-2015. Börsch-Supan, A., Bucher-Koenen, T., Kutlu-Koc, V., & Goll, N. (2016). Dangerous Flexibility- Retirement Reforms Reconsidered. MEA Discussion Paper 03-2017, MEA - Max Planck Institute for Social Law and Social Policy. Börsch-Supan, A., Härtl, K., & Leite, D. N. (2016). Social security and public insurance. Handbook of the Economics of Population Aging, 1B, 781-863. Brinch, C., Hernæs, E., & Jia, Z. (2012). Labor supply on the eve of retirement. Disparate effects of immediate and postponed rewards to working. Statistics Norway working paper number 698 Browning, M., Hansen, L. P., & Heckmann, J. J. (1999). Micro data and general equilibrium models. Handbook of Macroeconomics, 1(8). 543-633. Bundesgesetzblatt (2016). Gesetz zur Flexibilisierung des Übergangs vom Erwerbsleben in den Ruhestand und zur Stärkung von Prävention und Rehabilitation im Erwerbsleben. Bundesgesetzblatt, Teil I (59), Bonn. Bütler, M., Peijnenburg, K., & Staubli, S. (2016). How much do means-tested benefits reduce the demand for annuities?. Journal of Pension Economics and Finance, 1-31. Cecchetti, S. G., Lam P. S., & Mark, N. C. (2000). Asset pricing with distorted beliefs: Are equity returns too good to be true? American Economic Review, 90(4), 787-805. Chomik, R., Piggott, J., Woodland, A. D., Kudrna, G., & Kumru, C. (2015). Means Testing Social Security: Modeling and Policy Analysis. University of Michigan Retirement Research Center (MRRC) Working Paper, WP 2016-337. 36 Coile, C., Milligan, K., & Wise D. A. (2018). Social Security Programs and retirement Around the World: Working Longer (tentative title), forthcoming, University of Chicago Press: Chicago. Cogan, J. F. (1981). Fixed costs and labor supply. Econometrica, 49(4), 945-963. Deutsche Rentenversicherung Bund. (2015a). Die richtige Altersrente für Sie. Deutsche Rentenversicheung Bund. Berlin. Deutsche Rentenversicherung Bund. (2015b). Rentenversicherung in Zeitreihen. DRV-Schriften, (22). Deutsche Rentenversicherung Bund. Berlin. Eurostat (2016), http://ec.europe.eu/eurostat/statistiscs-explained/index.php/ mortality_and_life_expectancy_statistics, demo_mlifetable, retrieved 18 June 2017. Fehr, H., & Uhde, J. (2014). Means-testing and economic efficiency in pension design. Economic Modelling, 44, 57-67. Frederick, S., Loewenstein, G., & O´Donoghue, T. (2002). Time discounting and time preference: A critical review. Journal of Economic Literature, 40, 351-401. French, E. (2005). The effects of health, wealth and wages on labour supply and retirement behaviour. Review of Economic Studies, 72(2), 395-427. Fullerton, D. & Rogers, D.L. (1993). Who bears the Lifetime Tax Burden? Washington, DC: Brookings Insistution Press. Gasche, M. (2012). Was sind die richtigen Rentenabschläge? – Neue Perspektiven. Jahrbuch für Wirtschaftswissenschaften, 63(2), 187-235. Gielen, A. C. (2009). Working hours flexibility and older workers’ labor supply. Oxford Economic Papers, 61(2), 240-274. Göbel, C., & Zwick, T. (2009). Age and productivity – evidence from linked employer-employee data. ZEW - Centre for European Economic Research Discussion Paper 09-020 Graf, N., Hofer, H., & Winter-Ebmer, R. (2011). Labour supply effects of a subsidized old-age part-time scheme in Austria. Zeitschrift für ArbeitsmarktForschung, 44 (3), 217-229. Gruber, J. & Wise, D. A. (1999, eds.), Social Security and Retirement around the world, University of Chicago Press, Chicago, London. Gruber, J. & D. Wise (2004, eds): Social Security Programs and Retirement around the World: Micro-Estimation. University of Chicago Press, Chicago, London. Gruber, J., & Orszag, P. (2003). Does the Social Security Earnings Test Affect Labor Supply and Benefits Receipt? National Tax Journal, 56(4), 755 - 773. Gustman, A. L., & Steinmeier, T. L. (2004). Minimum Hours Constraints, Job Requirements and Retirement’. NBER Working Papers, No. 10876, National Bureau of Economic Research, Inc. Gustman, A. L., & Steinmeier, T. L. (2008). Projecting behavioral responses to the next generation of retirement policies. Research in Labor Economics, 28, 141 – 195 Huber, M., Lechner, M., & Wunsch, C. (2013). The Effect of Firms' Partial Retirement Policies on the Labour Market Outcomes of their Employees’. CESifo Working Paper, No. 4343, Munich. Ludwig, A. (2006). The Gauss-Seidel-Quasi-Newton Method: A Hybrid Algorithm for Solving Dynamic Economic Models. Journal of Economic Dynamics and Control, 31(5), 1610- 1632. 37 Maestas, N. (2010). Back to work: Expectations and Realizations of Work after Retirement. Journal of Human Resources, 45(3), 718-748. OECD. (2015). Pensions at a Glance. OECD, Paris. OECD. (2017). Labour Market Statistics: Labour force statistics by sex and age: indicators, OECD Employment and Labour Market Statistics (database). Retrieved from: http://dx.doi.org/10.1787/data-00310-en (Accessed on 19 April 2017). Queisser, M., & Whitehouse, E. (2006). Neutral or fair? Actuarial concepts and pension system design’. OECD Social, Employment and Migration Working Papers, No. 40, Paris. Reil-Held, A. (2000). Einkommen und Sterblichkeit in Deutschland: Leben Reiche länger?’ Schriftenreihe des Sonderforschungsbereichs 504 »Rationalitätskonzepte, Entscheidungsverhalten und ökonomische Modellierung«, No. 00-14. Universität Mannheim, Mannheim. Rogerson, R.., & Wallenius, J. (2013). Nonconvexities, Retirement and the Elasticity of Labor Supply’. American Economic Review, 103(4), 1445-1462. Social Security Administration. (2014). Social Security Programs Throughout the World: Europe. SSA Publication No. 13-11801, Washington D.C. Social Security Administration. (2008). https://www.ssa.gov/OACT/COLA/rteahistory.html, accessed 08-24-2017. Sonnet, A., Olsen, H., & Manfredi, T. (2014). Towards More Inclusive Ageing and Employment Policies: The Lessons from France, The Netherlands, Norway and Switzerland’. De Economist, 162, 315-339. Tran, B. (2005). The Effect of the Repeal of the Retirement Earnings Test on the Labor Supply of Older Workers. Manuscript, University of Maryland. Werding, M. (2007). Versicherungsmathematisch korrekte Rentenabschlage für die gesetzliche Rentenversicherung. ifo Schnelldienst, 16, 19-32. Werding, M. (2012). Rentenbemessung und Renteneintrittsalter: Korrekte Abschläge bei vorzeitigem Rentenzugang. Mimeo, Lehrstuhl für Sozialpolitik und öffentliche Finanzen, Ruhr-Universität Bochum. Woodland, A. (2016). Taxation, Pensions, and Demographic Change. Handbook of the Economics of Population Aging, 1B, 713-780. 38 Appendix This appendix shows the results for the extreme scenarios not described in Subsection 4.3. Scenarios with different productivity profiles and fixed cost levels Productivity Time costs of working Zero Moderate High Flat Scenario A-1 see Figure 6 Scenario A-2 Moderate see Figure 7 Benchmark in Section 4.1 see Figure 7 Steep Scenario A-3 see Figure 6 Scenario A-4 Scenarios with different consumption preferences and fixed cost levels Consumption preferences Time costs of working Zero Moderate High Flat Scenario A-5 see Figure 8 Scenario A-6 Moderate see Figure 7 Benchmark in Section 4.1 see Figure 7 Steep Scenario A-7 see Figure 8 Scenario A-8 39 Scenario A-1 (productivity: flat; fixed costs: zero) phi=0.55 With Earnings Test No Earnings Test Flexibility Reform 𝜔 = 0% 72 60 85 60 85 𝜔 = 3.6% 72 60 85 63 85 𝜔 = 6.3% 72 70 85 71 85 phi=0.60 With Earnings Test No Earnings Test Flexibility Reform 𝜔 = 0% 72 60 85 60 85 𝜔 = 3.6% 72 60 85 63 85 𝜔 = 6.3% 72 70 85 71 85 phi=0.65 With Earnings Test No Earnings Test Flexibility Reform 𝜔 = 0% 72 60 85 60 85 𝜔 = 3.6% 72 60 85 63 85 𝜔 = 6.3% 72 70 85 70 85 40 Scenario A-2 (productivity: flat; fixed costs: high) phi=0.55 With Earnings Test No Earnings Test Flexibility Reform 𝜔 = 0% 60 60 64 60 61 𝜔 = 3.6% 60 62 60 60 60 𝜔 = 6.3% 68 68 60 68 60 phi=0.60 With Earnings Test No Earnings Test Flexibility Reform 𝜔 = 0% 60 60 67 60 63 𝜔 = 3.6% 60 60 66 60 63 𝜔 = 6.3% 67 67 63 67 63 phi=0.65 With Earnings Test No Earnings Test Flexibility Reform 𝜔 = 0% 60 60 69 60 67 𝜔 = 3.6% 61 60 69 60 66 𝜔 = 6.3% 68 68 66 68 66 41 Scenario A-3 (productivity: steep; fixed costs: zero) phi=0.55 With Earnings Test No Earnings Test Flexibility Reform 𝜔 = 0% 62 60 76 60 76 𝜔 = 3.6% 69 60 76 61 75 𝜔 = 6.3% 71 69 77 68 76 phi=0.60 With Earnings Test No Earnings Test Flexibility Reform 𝜔 = 0% 63 60 77 60 76 𝜔 = 3.6% 69 60 76 61 76 𝜔 = 6.3% 71 69 77 68 77 phi=0.65 With Earnings Test No Earnings Test Flexibility Reform 𝜔 = 0% 64 60 77 60 77 𝜔 = 3.6% 70 60 77 61 76 𝜔 = 6.3% 71 69 77 68 77 42 Scenario A-4 (productivity: steep; fixed costs: high) phi=0.55 With Earnings Test No Earnings Test Flexibility Reform 𝜔 = 0% 60 60 59 60 58 𝜔 = 3.6% 60 60 58 60 58 𝜔 = 6.3% 68 68 58 68 58 phi=0.60 With Earnings Test No Earnings Test Flexibility Reform 𝜔 = 0% 60 60 63 60 61 𝜔 = 3.6% 60 63 61 60 61 𝜔 = 6.3% 68 68 61 68 61 phi=0.65 With Earnings Test No Earnings Test Flexibility Reform 𝜔 = 0% 60 60 66 60 63 𝜔 = 3.6% 60 60 65 60 63 𝜔 = 6.3% 68 68 63 68 63 43 Scenario A-5 (fixed costs: zero; consumption weight: flat) phi=0.55 With Earnings Test No Earnings Test Flexibility Reform 𝜔 = 0% 65 60 85 60 85 𝜔 = 3.6% 72 60 85 62 85 𝜔 = 6.3% 72 69 85 70 85 phi=0.60 With Earnings Test No Earnings Test Flexibility Reform 𝜔 = 0% 66 60 85 60 85 𝜔 = 3.6% 72 60 85 62 85 𝜔 = 6.3% 72 69 85 70 85 phi=0.65 With Earnings Test No Earnings Test Flexibility Reform 𝜔 = 0% 67 60 85 60 85 𝜔 = 3.6% 72 60 85 62 85 𝜔 = 6.3% 72 69 85 70 85 44 Scenario A-6 (fixed costs: high; consumption weight: flat) phi=0.55 With Earnings Test No Earnings Test Flexibility Reform 𝜔 = 0% 60 60 63 60 60 𝜔 = 3.6% 60 61 59 60 60 𝜔 = 6.3% 68 68 59 68 60 phi=0.60 With Earnings Test No Earnings Test Flexibility Reform 𝜔 = 0% 60 60 65 60 63 𝜔 = 3.6% 60 60 64 60 63 𝜔 = 6.3% 68 68 62 68 62 phi=0.65 With Earnings Test No Earnings Test Flexibility Reform 𝜔 = 0% 60 60 68 60 66 𝜔 = 3.6% 61 60 67 60 65 𝜔 = 6.3% 68 68 65 68 65 45 Scenario A-7 (fixed costs: zero; consumption weight: steep) phi=0.55 With Earnings Test No Earnings Test Flexibility Reform 𝜔 = 0% 64 60 85 60 85 𝜔 = 3.6% 71 60 85 62 85 𝜔 = 6.3% 72 69 85 70 85 phi=0.60 With Earnings Test No Earnings Test Flexibility Reform 𝜔 = 0% 65 60 85 60 85 𝜔 = 3.6% 72 60 85 62 85 𝜔 = 6.3% 72 69 85 70 85 phi=0.65 With Earnings Test No Earnings Test Flexibility Reform 𝜔 = 0% 66 60 85 60 85 𝜔 = 3.6% 72 60 85 62 85 𝜔 = 6.3% 72 69 85 70 85 46 Scenario A-8 (fixed costs: high; consumption weight: steep) phi=0.55 With Earnings Test No Earnings Test Flexibility Reform 𝜔 = 0% 60 60 58 60 58 𝜔 = 3.6% 60 60 58 60 58 𝜔 = 6.3% 68 68 58 68 58 phi=0.60 With Earnings Test No Earnings Test Flexibility Reform 𝜔 = 0% 60 60 64 60 62 𝜔 = 3.6% 60 63 61 60 61 𝜔 = 6.3% 68 68 61 68 61 phi=0.65 With Earnings Test No Earnings Test Flexibility Reform 𝜔 = 0% 60 60 66 60 64 𝜔 = 3.6% 60 60 65 60 64 𝜔 = 6.3% 68 68 63 67 64 DP 06-2017 Titel DP_06_2017
