Conference "Identifying ways of raising effective retirement ages"

Conference "Identifying ways of raising effective retirement ages" Organised jointly by the European Commission and the OECD 12 May 2016, Brussels Summary of the deliverables of the OECD project "Identifying ways of raising effective retirement ages" In response to the unprecedented changes triggered by population ageing, extending working lives often appears as a win-win strategy, both improving financial sustainability and enhancing retirement income adequacy. The workshop 1 on Delivering Longer Working Lives and Higher Retirement Ages organised jointly by the OECD and the DG Employment of the European Commission in November 2014 explored solutions to these challenges. It was clearly acknowledged that raising effective retirement ages while achieving more sustainable, adequate and safe pensions requires a high degree of synergy between social and employment policies (see also OECD, 2013 a,b; OECD, 2014 a,b, OECD, 2015 a,b). 2 The success of measures aimed at raising effective retirement ages depend on a wide range of factors. Pension reforms are crucial to determine eligibility to old-age pensions, to provide financial incentives to work longer and to influence social norms about behaviour by both employees and employers about working late in their career. Their impact on labour force participation and employment is the focus of work stream 1 of this Action 3. But increasing statutory retirement ages is not a panacea. The employment opportunities of older workers depend on several factors, and work stream 2 will focus on the determinants of employment at older ages. Labour demand for older workers is in particular greatly influenced by the skills they hold and the age profile of wages and skills, which work stream 3 will analyse. Most countries have begun to put in place a range of measures to buttress pension reform in order to promote longer working lives. Work stream 4 will provide an overview and assessment of the adequacy of these recent reforms in all 28 EU Member States to stimulate employment of older workers. This will be complemented in work stream 5 by a focus on social protection arrangements for the main labour market exit pathways. Finally, the potential impact of alternative social and health policies on the social and economic outcomes of ageing populations will be tested in work stream 6 for a pilot group of three European countries, which could be extended in the future to other EU countries subject to data availability. A joint EU-OECD kick-off meeting will take place in Brussels to launch the Action. 1 Joint EC-OECD workshop "Delivering longer working lives and higher retirement ages", 12-13 November, Brussels http://ec.europa.eu/social/main.jsp?langid=en&catid=88&eventsid=1020&furtherevents=yes 2. OECD (2015a), Ageing and Employment Policies: Poland Working Better with Age; OECD (2015b), Ageing and Employment Policies: Denmark Working Better with Age. OECD (2014a), Ageing and Employment Policies: Netherlands - Working Better with Age. OECD (2014b), Ageing and Employment Policies: France - Working Better with Age. OECD (2013a), OECD Review of the Irish Pension System. OECD (2013b), Ageing and Employment Policies: Norway Working Better With Age. 3. In general for this Action the quality of the results is dependent on the relevant information provided by the EU Member States that are not members of the OECD, and the Commission will ensure all relevant contacts with their authorities. In the event of any difficulties in getting appropriate information from countries, the Commission will be informed and the OECD will not be responsible for any resulting omissions in the results provided to the Commission.

Work stream 1. Labour supply implications of recent and on-going pension reforms Recent reforms tightening early-retirement options and raising retirement ages aim at increasing the labour supply of older workers. This work stream is about their effectiveness in raising employment rates of older workers and total employment via possible substitution effects. This part of the Action will build on the close monitoring by the OECD of pension reforms undertaken by EU Member States since the beginning of the world economic crisis in 2007-08. The goal is to extend the analysis to all 28 EU Member States, although the extension to OECD non-member countries will depend on the availability of information related to recent policy action in the pension area. This work stream will present an overview of the relevant measures, their timing and, as far as possible, the numbers and main characteristics of the additional workers that would become available to the labour market following the implementation of reforms, given the size and composition of the relevant age cohorts. The analysis will be based both on the countries demographic structure and on the responsiveness of labour supply to various pension policies reducing the incentives to retire early, including those aimed at extending working lives and benefit cuts. More specifically, the impact of the elimination of early-retirement schemes, of moving towards actuarial neutrality of pension systems and of raising contribution periods and standard retirement ages on the participation rates of older workers will be estimated. The estimated impacts will be based on elasticities of participation rates to retirement ages and financial incentives computed using macroeconomic and microeconomic datasets for EU countries (subject to data availability). They will also be based on the OECD indicator of implicit tax on continued work. These elasticities will be computed for different levels of skills and socio-economic background of older workers. These estimates would provide a good basis for the part of work stream 2 focusing on barriers and opportunities to work longer. While there is a trade-off between country-specific parameter values and the quality of the estimates, the objective would be to group countries based on the labour market institutional settings studied in work stream 4 and for which estimates do not differ significantly. In a second stage, based on labour demand elasticities across different skills and various substitution effects across age groups, and depending on their robustness, this work stream would include the estimation of the impact of reforms on total employment. To the extent possible, the estimated effects will distinguish the short and medium to longer term. The estimation of demand elasticities will investigate the presence of non-linearity; that is, whether they differ according to the level of labour costs following the analysis in work stream 3. More generally, the results will help illustrate possible crowding-out effects across age groups and put in perspective the lump of labour fallacy assumption. Work stream 2. Determinants of older workers employment chances Many pension reforms have tightened the link between contributions paid when working and the pension benefits received when retired. Others have increased the retirement age and the length of the contribution period required to obtain a full pension. The rapid ageing of the population, the effect of the crisis and the need to enhance the financial sustainability of pension systems have driven these reforms. For individuals, these policies have profound effects. In particular, the ability of labour markets to deliver longer and less interrupted working careers is very important for pension adequacy. Longer lives driven by higher life expectancy imply, for example, that an early exit from the labour market may substantially lower the resources the individuals would have over (a longer) retirement. In recent years the employment rates of people older than 50 have steadily increased (which is associated with a small increase in the average effective age of labour market exit), but important employment gaps remain depending on the socio-economic characteristics of population groups; this applies to gaps between workers with high and low education levels, between men and women and between workers with better health compared to those with disabilities or other chronic illnesses. The OECD work on this part of the Action will focus on the employment chances of older workers in order to better identify employment determinants at old ages. Such an analysis would therefore help 2

determine which older people are more likely to be found in employment and under which conditions. Employment chances of older workers are in fact likely to reflect the interactions between several factors: individual attitudes and preferences combine with opportunities, constraints and social structures in complex ways. Economic circumstances (such as education, age and gender) and physical and/or mental health are certainly part of the equation when it comes to the employment of older workers. However, the different ways work places manage age diversity in late careers or accommodate for more flexibility in retirement schedules (while favouring autonomy, job security, learning opportunities, collective interest representation, and other aspects determinant of job quality) are also important. Finally, financial rewards embedded in pension and tax/benefit systems to either continue working or retire, and other contextual factors (such as the macro economic conditions, the features of social security and of pension systems, and national policies concerning the employment of older workers) also matter. Building on work stream 1, the analysis here will first describe older cohorts of workers in EU countries over time in terms of gender, education, sector of activity, occupation and health status (subject to data availability) using data from the European Labour Force Surveys (LFS) and its special modules. In all its parts, the empirical models will take a close look at age differences. The identification of these characteristics will document to what extent low education attainment, age and bad health shorten working lives, and potentially lead to less than adequate incomes over retirement. This will also help understand the extent to which improvements in educational attainment and health status across cohorts might increase employment prospects. The analysis will examine, in a second step, barriers and opportunities to deliver effectively longer working lives based on econometric models. These estimates will complement the results in work stream 1 based on labour supply and labour demand elasticities. Beyond marital status, education and gender, push factors (such as bad physical and mental health, dissatisfaction with work, insufficient access to training, arduous working conditions in some sectors and occupations) and pull factors would be considered. Again age differences will be considered by introducing 5-year age-group coefficients (such as 50-54, 55-59, 60-64, 65-69 and 70+) and interacting them with education level to assess their effect on the probability of leaving employment/entering retirement which would be a useful basis for work stream 3. In this part of the work, it will be also possible to distinguish dependent employment and self-employment in late careers and examine, for example, whether the propensity to move into retirement is different across work statuses. The analysis will also examine: The effect of economic conditions (such as unemployment rates and GDP growth) and other policies (such as employment protection legislation) using the appropriate indicators. How disabilities related to chronic diseases (and possibly disease prevention programmes) affect health and thereby labour force participation. The availability of sufficient or adequate provision of child and long-term care services and the resulting care responsibilities potentially hampering labour market participation (of older women in particular) will also be addressed in the quantitative analysis using the appropriate variables available in the ageing and retirement surveys (e.g. the number of grand-children, whether the person retired either due to family responsibility or to care for grand-children, etc.). Moreover, as aggregate data suggests that the probability of being re-hired is much lower than the probability of being retained in a company, the analysis (and again subject to data availability) would also attempt to look at factors influencing the probability of being rehired in old-age (such as seniority pay or employment protection legislation). To examine the determinants of employment chances of older workers, various empirical analyses will be conducted. These analyses will be mainly based on the Survey of Health, Ageing and Retirement in Europe (SHARE) data, which implies that in some cases not all EU Member States will be covered. Other surveys [such as the European Labour Force Survey, Eurofound surveys, European Barometer survey and EU statistics on income and living conditions (EU-SILC)] would then be used in addition to the SHARE survey to extend the coverage as far as possible for all 28 EU Member States. Qualitative data would supplement the analysis when quantitative data are missing. The actual 3

geographical coverage of the analysis will eventually depend on data availability. However, policy messages based on a cluster of countries presenting similarities might be offered to allow coverage of countries not included in the analysis. The OECD and the EU would organise a joint international workshop during the timeframe of the Action to generate synergies and facilitate a multilateral exchange of ideas and best practices on how to deliver effectively longer working lives. Work stream 3. Impact of the wage-skill relationship on the labour demand of older workers Beyond the factors analysed in work stream 2, the employment rates of older workers are influenced by a large range of policies. Employment rates of older workers have increased in many OECD member countries, in response to supply-side policies which have raised the legal retirement age and reduced incentives to retire before this age, as analysed in work stream 1 (and in response to other changes in social protection programmes analysed in work stream 5). But now that people are staying longer in the labour market, it is crucial to monitor closely demand-side aspects in order to ensure that rising participation rates of older workers continue to translate into higher employment rates, without increasing unemployment or the ranks of the working poor. Skills that workers hold are a major determinant of both their earnings and their employability. Yet a striking result from the OECD Survey of Adult Skills (PIAAC) is that skills are closely related to age, with the oldest age groups displaying significantly lower levels of proficiency in informationprocessing skills than the youngest (OECD, 2013c). 4 Lower proficiency in foundation skills (i.e. numeracy and literacy) does not necessarily translate into greater risk of labour market disadvantage since older workers have acquired other skills through longer work experience, which are valuable and that younger workers may not hold. However, their job-specific skills may put them at risk of exclusion from the core labour market if they lose their job. In particular, older jobseekers with weak generic skills (such as IT skills) are likely to encounter considerable difficulties in finding a quality job. Thus, it is important that workers have opportunities and incentives throughout their working lives to keep their skills up to date and learn new skills. In addition, in a number of OECD member countries, wage setting arrangements include a seniority component whereby wage increases are automatically linked to tenure rather than to individual performance. While this can prevent those older workers with the lowest skill levels from falling into in-work poverty, it may discourage employers from retaining older workers, especially if labour demand elasticities are high (work stream 1). To shed further light on these issues, this stream of work will, for the 17 European countries covered by PIAAC 5 : Examine the age-profile of wages and skills proficiency, and the relationship between the two. Investigate in more detail other key determinants of wage age-profiles (i.e. job and worker characteristics), with a particular focus on job tenure so as to search for evidence of a seniority component in wage setting arrangements. Investigate the differences in skill profiles of older people (aged 50-64) by five-year age group who are either out or work or in work, as well as the differences in the skills required in, respectively, their previous or current jobs. Examine whether differences across countries in terms of seniority-based pay practices are associated with differences in unemployment risks or in the incidence of low pay (or in-work poverty) among older workers. 4. OECD (2013c), OECD Skills Outlook 2013. 5. Austria, Belgium (Flanders), Czech Republic, Denmark, Estonia, Finland, France, Germany, Ireland, Italy, Netherlands, Norway, Poland, Slovak Republic, Spain, Sweden and United Kingdom (England and Northern Ireland). 4

While governments have few direct tools to intervene in wage setting arrangements agreed between the social partners, subsidies can also be used to lower the costs of hiring and retaining older workers (such as reductions in employer social contributions for hiring or retaining older workers). The effectiveness of this form of intervention would be assessed. Governments can also seek to indirectly affect wage behaviour in the private sector by changing their own compensation practices in the public sector to align better pay with performance rather than tenure. They can also limit the possibility for employers to set a mandatory retirement age. This would encourage employers to link pay more closely with performance as it would become more difficult to simply rely on mandatory retirement to get rid of older workers whose remuneration exceeds the value of their productivity. Measures should also be taken to ensure that adequate training facilities as well as the right incentives are in place for employers to provide training opportunities to their workforce and for workers to participate in lifelong learning activities. In a concluding section, this work stream will review the main obstacles that firms, and in particular small firms, may encounter in providing sufficient training opportunities to their employees, as well as the main barriers that workers, notably the most vulnerable among them, may face to access training. It will also provide examples of good practices in promoting vocational education and training for middle-aged and older workers with low educational attainment. The intellectual services providers will contribute to the second part of this work stream by reviewing country experiences in providing employers with subsidies to lower the costs of hiring and retaining older workers, as well as in promoting skill development throughout working careers. Work stream 4. Employment policies for older workers This work stream builds on, and extends to all 28 EU Member States, the OECD review of ageing and employment policies to promote employment at an older age. 6 The 2006 synthesis report Live Longer, Work Longer made an important contribution to establishing an age-friendly policy agenda for reform in three broad areas: i) strengthening incentives to carry on working; ii) tackling employment barriers on the side of employers; and iii) improving the employability of workers. Many OECD member countries have since undertaken a number of reforms consistent with that policy agenda, as assessed in a follow-up review conducted in 2011-15. 7 Even though important steps have already been taken by governments, employers, trade unions and non-governmental organisations, there remains considerable scope for further progress to achieve longer working lives and boost labour market prospects for older people, as documented in work streams 1-3. It is especially important to promote and reward job mobility after the age of 50, increase the range and quality of job opportunities open to low-skilled older workers, and reduce gender inequalities among older workers. Therefore, this work stream will provide a monitoring and assessment for all 28 EU Member States of recent reforms and measures to stimulate employment of older workers. This will be complemented by a scoreboard of comparative indicators to monitor the success of these policies in promoting active ageing and longer and better working lives. The following activities would be carried out: The Late-Career Labour Markets scoreboard: This would include quantitative indicators on five topics: employment (employment rates; gender gap in employment; etc.); job quality (incidence of part-time work, of temporary work and of self-employment; age gap in earnings; etc.); labour market mobility (retention rate; hiring rate; job changes in general and after unemployment and after inactivity; effective labour force exit age, etc.); unemployment (unemployment rate; incidence of long-term unemployment; risk of being neither in employment, neither completely in retirement, etc.); and employability (incidence of tertiary education; participation rate in training; age gap in training; etc.) for both the age groups 55-64 and 65 and over. New indicators will be developed to capture transitions from work to retirement (i.e. risk of being neither in employment, neither in retirement). The scoreboard for 6 The OECD conducted a major multi-country review covering 21 OECD member countries during 2003-05 and summarised in the 2006 synthesis report Live Longer, Work Longer, launched after a High-Level Policy Forum in Brussels in October 2005. 7 See more on http://www.oecd.org/employment/emp/ageingandemploymentpolicies.htm. A synthesis of this work will be presented in the report Working Better with Age (forthcoming in 2016). 5

each EU Member State will compare the country s late-career labour market situation with those of other EU Member States as well as with the EU and OECD averages. 28 country notes (max. 15 pages per country note) summarising the implementation of policies to promote longer working lives over the last decade, including a Late-Career Labour Markets scoreboard and an assessment of progress towards raising effective labour force exit (or retirement) ages for both genders. A policy database on good practices to promote longer working lives (both as wage and salary earners and as self-employed), their use across EU Member States, the involvement of the social partners and the potential for improvement. This will include the use and effectiveness of wage subsidies (including reductions in employer social security contributions) for hiring and retaining older workers. The intellectual services providers will contribute to the second part of this work stream by preparing country notes and providing examples of good practices to promote longer working lives. Work stream 5. Social protection arrangements for longer working lives The contribution of social protection arrangements to longer working lives is essential. However, many knowledge gaps still exit. The OECD work in this part of the Action would be structured as follows. It will first analyse the main pathways of exiting the labour market, thus complementing the analysis in work streams 1 and 2. The analysis will be based on data from the Labour Force Survey in EU countries distinguishing by gender and by age groups. It will then provide a description of the arrangements representing either barriers or opportunities to longer working lives, in particular: (1) early retirement options for older workers in general, those applying for certain professions, for mothers, and for the unemployed; (2) the possibilities to retire gradually and/or partially such as parttime pensions; (3) the options to delay retirement; (4) the incentives or penalties to combine work and pensions; (5) disability benefits; (6) means-tested income schemes for the elderly; and (7) access to affordable child and elderly care. In doing so, the Action will review the financial conditions under which these options are granted and the extent to which these measures create an incentive to leave at an early or later stage. In particular, it will quantify the impact of extending the working life on pension benefits and the extent to which the pension designs make the systems close or far from actuarial neutrality. 8 In doing so the work will use estimates of changes in pension wealth (i.e. the present value of the cumulated flow of pension benefits taking account of discounting, mortality, indexation of pension payments) at different ages from the OECD pension model. Ideally, work stream 4 will assess whether the public support is well targeted or generates windfall effects by reaching people who would keep working in the absence of the measures. The analysis will also assess whether (1) the conditions under which people can reduce working time at the end of the career are not too strict; and (2) whether gradual retirement or partial retirement options have adverse effects on pension entitlements. It will also discuss the alternatives that could be used to avoid early retirement (such as better working conditions and career management, higher accrual rates during periods of unhealthy work). This work stream will discuss the nexus between the tightening of early retirement possibilities, the search for greater flexibility in retirement ages and the design of financial incentives. It will elaborate on the extent to which the related policy interventions might generate some inconsistencies in the achievement of the various objectives or whether there is some complementarity between the various 8 Actuarial neutrality is a central concept to work incentives around retirement ages. There are two main interrelated but different definitions, capturing changes in pension benefits at the margin. According to the first (se e.g. Duval, 2003), the pension system is neutral if the cost in terms of foregone pensions and contributions paid for working an additional year is exactly offset by an increase in future benefits. According to the second (see e.g. Queisser and Whitehouse, 2006), the system is actuarially neutral if the present value of accrued pension benefits for working an additional year is the same as in the year before (meaning that benefits increase only by the additional entitlement earned in that year). The main difference between the two definitions is that contributions paid or benefits earned during the additional year are not considered in the second one. 6

measures. In particular, the interrelation between the increase in retirement ages, its speed, and financial incentives will be analysed to feed the normative discussions. A discussion of the possible impact of a provision increasing retirement ages with life expectancy on people working in particular arduous and hazardous jobs will also be presented, beyond the effects discussed in work stream 1. The information collected by the European Social Policy Network (ESPN) for the 2016 report on social protection arrangements for people in arduous jobs is an important basis for the work carried out in this part of work stream 5. Automatic adjustment mechanisms, triggered by changes in some economic or demographic variables, allow smoother changes in pensionable ages and avoid the political risks of ad hoc pension reforms. However, increasing the pensionable age by linking it to life expectancy does not imply that people will effectively work more. Moreover, these mechanisms may make some of the workers who are unable to extend their working lives increasingly vulnerable to retirement-income inadequacy, unless other systems are in place to prevent this from happening. The report will discuss the main challenges to overcome in terms of social protection tools in order to deliver longer working lives. The report will be used as the basis to prepare a joint workshop with the EU aimed in particular at those countries that need to make their social protection system more effective in delivering longer working lives to older workers. The workshop should involve various stakeholders, including representatives from governments, trade unions, employers and civil society. Work stream 6. Modelling optimal reform strategies The OECD is developing a policy foresight model, based on dynamic microsimulation techniques. This model, called the Global Future Elderly Model, is a dynamic micro-simulation model building on a model originally developed by the University of Southern California (USC). Many features make it different from other dynamic micro-simulation tools. First, it is very powerful because it allows assessing the implications of health developments on health expenditures and on population projections (reasons for which it was initially created). Since it takes into account the distribution and incidence of chronic/non-chronic diseases, the pace of population ageing and the impact of those diseases (e.g. diabetes and obesity) on life expectancy can be analysed. The model is flexible and could accommodate other important behavioural processes (such as retirement and labour market participation). The OECD s work would focus precisely on extending the model on pension and labour market aspects. With this model, the OECD aims to test the potential impact of alternative social and health policies, and various policy combinations, on the social and economic outcomes of ageing populations in a limited number of OECD member countries presenting different characteristics in terms of the ageing process. The purpose is to see how current and projected populations would respond to policy changes according to their educational attainment, employment and earnings trajectories, health behaviours, health conditions and disabilities. This model is meant to help countries develop costeffective policies to produce better and more equal outcomes for different cohorts/generations. It will provide an opportunity to benefit from a consistent international platform to test health and social policies before they are implemented. In addition, it will live on to become a policy advice tool for tackling the social and health policy challenges that we will face in the future. The OECD work in this part of the Action will sharpen the focus of some of the analyses carried out in the overall project for 3 EU Member States (Belgium, Italy and Sweden). This is a pilot group of countries which can be potentially extended in the future to other EU Member States subject to data availability as explained below. The model will be used: (1) to forecast the life of three cohorts of individuals from the Baby Boomers, Generation X and Millennials starting at ages 51-52; and (2) to project outcomes of current policies and potential alternative policies for representative populations, by sex, age group, generation and socio-economic status and other characteristics including educational attainment, labour-force participation (employed and unemployed) and earnings, life expectancy and deaths. The work that is currently being undertaken at the OECD with the policy foresight model aims primarily to better understand the implications of changes in the health status of the elderly in terms of 7

health expenditures and health policies. The work the OECD will do with the policy foresight model in this Action would be rather to test the effect of some pension and labour market policies scenarios on key economic variables including employment and pension levels by distinguishing how different cohorts are affected according to their socio-economic characteristics and life expectancies. In this respect, the model would analyse the effect of increasing pension age on public pensions, taking into account differences in expected life expectancy and health improvements across socioeconomic characteristics. The analysis in work stream 6 will complement the work carried out in the preceding work streams by making projections on the future pension outcomes of current young cohorts in three countries based on individual micro-data under various reform scenarios. This stream of work would highlight the degree of disparities across groups within and across the three aforementioned EU Member States, and the effects of some labour market and pension policies in terms of equity and affordability. In order to create this model, the OECD will lead a joint effort involving the USC Schaeffer Center and the University of Rome Tor Vergata. A first phase (the pilot) will focus on creating a workable full model on a limited scale that is capable of projecting health and economic outcomes and testing alternative social and health policy options to improve outcomes and reduce inequalities. The primary goal of the pilot is to have an operational model ready for policy analysis in Belgium, Italy and Sweden. Additional countries might be added in a second stage when the model will be stable enough to deliver reliable results. Further extensions may occur using other platforms such as the MIDAS- LIAM2 model, a new open source development tool for discrete-time dynamic microsimulation models. The data will be based on ageing/retirement and population health surveys (SHARE) and on the OECD pension models to map pension entitlements rules over time. 8