Designing funded pension arrangements given the level of financial literacy and behavioural biases

Transcription

1 For Official Use Organisation de Coopération et de Développement Économiques Organisation for Economic Co-operation and Development DAF/AS/PEN/WD(2017)3 13 June 2017 DIRECTORATE FOR FINANCIAL AND ENTERPRISE AFFAIRS INSURANCE AND PRIVATE PENSIONS COMMITTEE English - Or. English Working Party on Private Pensions Designing funded pension arrangements given the level of financial literacy and behavioural biases Background and assessment of policies June 2017 This document is circulated for discussion under the agenda of the WPPP meeting For further information, please contact Ms. Stéphanie Payet [Tel: ; stephanie.payet@oecd.org] or Mr. Pablo Antolin [Tel: ; Pablo.antolin@oecd.org] JT This document, as well as any data and map included herein, are without prejudice to the status of or sovereignty over any territory, to the delimitation of international frontiers and boundaries and to the name of any territory, city or area.

2 DAF/AS/PEN/WD(2017)3 FOR OFFICIAL USE Table of Contents Designing funded pension arrangements given the level of financial literacy and behavioural biases: Background and assessment of policies Introduction Why financial literacy levels and behavioural biases matter in the context of DC pension arrangements? Participation decision... 7 Issues posed by poor financial literacy and behavioural biases... 7 Assessment of policies How much to contribute Issues posed by poor financial literacy and behavioural biases Assessment of policies Choice of the pension provider Issues posed by poor financial literacy and behavioural biases Assessment of policies Choice of the investment strategy Issues posed by poor financial literacy and behavioural biases Assessment of policies Choice of the pay-out product Issues posed by poor financial literacy and behavioural biases Assessment of policies Conclusion References

3 FOR OFFICIAL USE DAF/AS/PEN/WD(2017)3 Designing funded pension arrangements given the level of financial literacy and behavioural biases: Background and assessment of policies 1. Introduction 1. The pensions' landscape has changed in recent decades, giving a greater role for funded private defined contribution pension plans. Pension systems are still addressing the challenges posed by population aging, the financial and economic crisis and the economic environment of low growth and low interest rates. These challenges have led to reforms that have increased the diversity of pension arrangements across OECD countries and the importance of funded pension arrangements, especially defined contribution (DC) ones (OECD, 2016a). 2. Defined contribution plans allow a direct link between pension contributions and pension benefits but put most risks related to retirement saving (i.e. investment and longevity) onto individuals. As DC pensions represent a growing share of total retirement income, their design needs to be improved. In addition, OECD (2016a) compares financial education needs across different types of pension arrangements and indicates that the challenges people face in making decisions about their retirement are greater for private pensions (as opposed to public pensions), personal plans (as opposed to occupational plans) and DC plans (as opposed to defined benefit or DB plans). In DC plans, people have to make many important decisions by themselves that will determine how much they will get from their plan, such as how much to save, where to invest, when to retire and how to allocate the assets accumulated when retiring. DC plans therefore offer a greater amount of choice, but imply a greater amount of risk for individuals and require more financial skills than DB plans. 3. However, people have difficulties in planning for retirement, determining their retirement income needs and choosing retirement products. A combination of lack of general financial knowledge and awareness of risk, poor pension-specific knowledge, as well as behavioural and psychological biases undermines people's ability to make appropriate decisions for their retirement. The potential mistakes that people can make along the way therefore put at risk the ability of DC pension plans to deliver pension benefits that people would consider adequate to finance retirement. 4. This document identifies the key decisions that people need to make for their retirement at different stages of their lives in the context of DC pension plans. It is also a first attempt to i) highlight the issues posed by low financial literacy levels and behavioural biases and ii) assess experiments and policies implemented in different OECD and non-oecd jurisdictions to improve the design of DC pension plans through motivating or facilitating the appropriate behaviour by individuals, for each of the key decisions that people need to make along the way. It emphasises the most recent developments in design policies. 5. Policies aiming at improving the design of DC pension plans while addressing the issues posed by low financial literacy levels and behavioural biases can be divided into six broad categories. Default options are widely used to increase participation in and contributions to DC pension arrangements (e.g. automatic enrolment), as well as to help people who are unable or unwilling to choose a pension provider, an investment strategy or a pay-out product. Simplification of information and choice is also a key policy 3

4 DAF/AS/PEN/WD(2017)3 FOR OFFICIAL USE measure to help people making decisions through a reduced set of options (e.g. investment strategies), better disclosure of comparable information (e.g. cost information) or facilitated comparison of options (e.g. a single internet platform to compare all pay-out offers). More salient information can improve decision making. This information can be conveyed through pension statements, financial education programmes and financial advice. Tax and other types of financial incentives (e.g. matching contributions) are widely used to promote private pension arrangements. Policy makers also introduce product features that may help alleviate some of the fears that people may have when locking in their money in private pension plans (e.g. investment return guarantees, flexible annuity products). Finally, policies reducing costs are important to improve retirement income outcomes. 6. The document starts with a background section highlighting the main reasons why low levels of financial literacy and behavioural biases may have significant implications for retirement income adequacy in the context of DC pension arrangements. The next sections focus on each of the key decisions that people need to make when saving in a DC pension plan: whether to participate in the plan (section 3), how much to contribute (section 4), how to choose the pension provider (section 5), how to invest the contributions (section 6) and how to allocate the accumulated savings when retiring (section 7). Each section is structured in the same way: it first presents the issues posed by poor financial literacy and behavioural biases and then assesses experiments and policies implemented in different OECD and non-oecd jurisdictions to improve the design of DC pension plans through motivating or facilitating the appropriate behaviour by individuals. Finally, section 8 concludes. The Secretariat will share this document with the OECD International Network on Financial Education. 7. This document lists and assesses the different experiments and policies that have been implemented in different jurisdictions. It does not identify the most effective policies to improve the design of DC pension plans. This will be the goal of a future paper that will distil effective practices and policies on how to design DC pension plans given the level of financial literacy and behavioural biases. Delegates are kindly invited to: help the Secretariat complement this document with any relevant policies that have been implemented in their respective country to improve the design of DC arrangements while accounting for low financial literacy levels and behavioural biases provide additional evidence on the cost and effectiveness of policies implemented in their country to improve the design of DC pension plans (for example, whether cost-reducing policies do not lower the quality of the services provided by pension funds). 2. Why financial literacy levels and behavioural biases matter in the context of DC pension arrangements? 8. This section presents how financial literacy levels and behavioural biases interact with retirement decisions in the context of DC pension arrangements. Recent pension reforms have reduced the role of pay-as-you-go (PAYG) public pensions by lowering future pensions (OECD, 2015a). In parallel, funded pension arrangements have grown in 4

5 FOR OFFICIAL USE DAF/AS/PEN/WD(2017)3 importance, as a complement to public PAYG pensions. In addition, within funded pensions, DC pension arrangements are becoming more prominent (OECD, 2016a). 9. DC pension arrangements provide a clear, straightforward link between pension contributions and pension benefits, but put most risks onto individuals, making them more responsible for managing their retirement. In DC pension arrangements, assets accumulated at the end of one's working life (contributions plus investment income earned on those contributions) are used to generate a stream of income, thereby directly determining the amount of retirement income. However, individuals have to bear investment and longevity risks. In addition, they have to make many important decisions that will determine how much they will get from their plan, such as how much to save, where to invest, when to retire and how to allocate the assets accumulated when retiring. 10. DC pension arrangements, unlike with PAYG and funded DB plans, require that individuals understand the features of the different plans offered to them. However, most individuals may not be able or prepared to assume this role, due to low levels of financial literacy and behavioural biases. 11. Low levels of financial literacy are prevalent in many countries. The OECD International Network on Financial Education (INFE) has defined financial literacy as follows: "A combination of awareness, knowledge, skill, attitude and behaviour necessary to make sound financial decisions and ultimately achieve individual financial wellbeing" (OECD, 2016b). OECD (2016b) reports findings from 30 countries and economies that participated in the OECD/INFE international survey of adult financial literacy competencies and collected cross-comparable data. It shows that overall levels of financial literacy, indicated by combining scores on knowledge, behaviour and attitudes are relatively low, with an average score of 13.2 out of a maximum 21. On average, only 56% of adults achieve the minimum target score on financial knowledge, with large differences by gender as 61% of men achieve the minimum target score against 51% for women. The study identifies budgeting, planning ahead, choosing products and using independent advice as weak areas of financial behaviour. The analysis also shows that regarding attitudes, many people have a tendency towards short-termism (50% on average). In addition, studies reviewed in Lusardi and Mitchell (2014) show that financial knowledge is positively correlated with retirement planning, and that those who plan also accumulate more wealth. 12. Behavioural biases are specific ways in which normal human thought systematically departs from being fully rational. Biases can cause people to misjudge important facts or to be inconsistent. Adapted from the list and classification of biases in DellaVigna (2009), the Financial Conduct Authority (FCA) in the United Kingdom categorises ten cognitive biases in retail financial services, according to which component of a decision they affect: preferences, beliefs and decision-making processes (Table 1). 5

6 DAF/AS/PEN/WD(2017)3 FOR OFFICIAL USE Table 1. FCA's classification of behavioural biases in retail financial services Category Bias Description Preferences Beliefs Decision making Present bias Source: Erta et al. (2013). Reference dependence and loss aversion Regret and other emotions Overconfidence Over-extrapolation Projection bias Mental accounting and narrow bracketing Framing, salience and limited attention Decision-making rules of thumb Persuasion and social influence People respond to urges for immediate gratification resulting in overvaluing the present over the future. As such, choices are regretted in the future. Present bias can lead to self-control problems such as procrastination. When evaluating a product or future prospects, people do not think of the choice or product in isolation. Instead they assess it with respect to changes relative to a reference point, thinking in terms of gains and losses from that reference point. Preferences may therefore change when the reference point changes. In addition, psychologically, losses are felt roughly twice as much as gains of the same magnitude. Loss aversion may lead to the endowment effect (valuing a good more just because the individual owns it), a preference for the status quo and distortions in attitudes to risk. People avoid choice or are willing to pay for products just to avoid making a decision that they may come to regret. They may also shy away from ambiguity, uncertainty or stress even if making a choice is likely to result in a positive outcome for them. Their choices can also be distorted by temporary strong emotions (e.g. fear). People can show overconfidence about the likelihood of good events occurring or their own ability and success at different tasks, including the accuracy of their judgements. People often make predictions on the basis of only a few observations, when these observations are not representative. As a result, people also underestimate uncertainty. People expect their current tastes and preferences to continue in the future and underestimate the possibility of change. Mental accounting describes how people treat money or assets differently according to the specific purpose that they have assigned to them, instead of treating all money as the same. Narrow bracketing describes how people often consider the decisions they take in isolation, without integrating these decisions with other decisions that affect their overall wealth and level of risk they take on. People may react differently to essentially the same choice situation because the problem is framed differently. Frames usually work by triggering a particular bias (e.g. loss aversion, reference dependence, regret, a rule of thumb), as certain information is made more salient and limited attention is paid to other factors. Consumers simplify complex decision problems by adopting specific rules of thumb (heuristics). When choosing from a wide range of options, people may choose the most familiar, avoid the most ambiguous or uncertain, choose what draws attention most (e.g. the first option on a list), or avoid choice, including sticking to the status quo. When estimating unknown quantities, people may anchor estimates to some relevant or irrelevant figure and adjust from there. Emotions and norms in social interactions are important: consumers may allow themselves to be persuaded to buy a product just because the sales person is 'likeable' and therefore trustworthy. Emphasising good personality traits or overemphasising bad personality traits may substitute for a reasoned judgement. Consumers may also be influenced by usage patterns without adequately considering whether those apply to their own circumstances. 13. Some of the behavioural biases described in Table 1 particularly affect individuals' retirement planning. Present bias, framing, rules of thumb, persuasion, overconfidence, over-extrapolation, loss aversion, and regret and other emotions are specifically important when making decisions for retirement. First, present bias can be 6

7 FOR OFFICIAL USE DAF/AS/PEN/WD(2017)3 strong as saving for retirement is for the long term and may compete with other shortterm needs. Therefore, the combination of delayed benefits until retirement and small short-term costs (e.g. transaction costs, paperwork) can be a real barrier to action, including participation in retirement savings plans. Second, financial products, and in particular retirement products, are complex. Individuals usually consider that making financial decisions is hard, unpleasant and time-consuming. They often lack motivation to invest time and effort to make informed decisions and, because of the complexity, cannot easily evaluate some products at all. They are therefore more likely to rely on simple rules of thumb and be sensitive to framing and persuasion. Third, effective retirement decisions require sophisticated risk assessments (e.g. longevity and investment risks). Most people lack the skills, practice or intuition to assess risk and uncertainty when making important decisions. Overconfidence and over-extrapolation may therefore lead individuals to underestimate uncertainty and risk. Fourth, many financial decisions are emotional. Emotions, whether positive (like optimism or excitement) or negative (like stress, anxiety, fear and regret), can drive decisions rather than logical cost/benefit analyses. Finally, it can be difficult to improve one's ability to deal with retirement products over time. Decisions related to retirement planning are made infrequently. The consequences of these decisions are often only revealed long after the decision has been made, with little opportunity to learn and correct past decisions. Because of loss aversion and fear of regretting one's decisions later on, people may therefore fail to act. 14. The following sections focus on each of the key decisions that people need to make for their retirement at different stages of their lives: whether to participate in the plan, how much to contribute, how to choose the pension provider, how to invest and how to allocate the accumulated savings when retiring. Poor financial literacy and behavioural biases potentially affect those decisions in a different way. OECD and non-oecd jurisdictions have implemented different policies to address the implications of poor financial literacy and behavioural biases, improving as a result the design of DC pension plans. These policies either simplify the decision-making process or harness the power of behavioural biases to nudge people into acting in their own long-term interest. 3. Participation decision Issues posed by poor financial literacy and behavioural biases 15. Present bias is the main behavioural bias affecting participation in private pension arrangements. The complexity of retirement savings products also makes it difficult for people to plan for retirement and properly assess how well prepared they may be for retirement. 16. Because of present bias, individuals are bad at committing to save for retirement. Procrastination, myopia and inertia lead many individuals to postpone or avoid making the commitment to save for retirement even when they know that this is ultimately in their best interest. In addition, retirement planning competes with other short-term needs, especially at younger ages (e.g. buying a house, pay tuition fees, raising a family). 17. Saving for retirement is a complex financial decision. One potential consequence of this complexity is that individuals put off confronting these decisions. For example, Iyengar, Jiang and Huberman (2004) find that participation rates in 401(k) pension plans in the United States decline as the number of fund option increases. Other things equal, every ten options added was associated with 1.5% to 2% drop in participation rates. Therefore, the complexity of the retirement savings decision, combined with low levels of 7

8 DAF/AS/PEN/WD(2017)3 FOR OFFICIAL USE financial literacy to appropriately assess the different options, discourages employees from timely enrolment in private pension plans, even when they prefer participation to non-participation. 18. It is also of concern that many people have a misperception of their retirement preparedness, as they would fail to take action in case their future retirement income falls short of their expectations. Munnell, Hou and Sanzenbacher (2017) show that 57% of households in the United States have realistic expectations about their retirement preparedness, with 33% recognising that they are at risk of being unable to maintain their standard of living in retirement. Other households have a misperception of their retirement readiness, with 24% of households reporting that they are inadequately prepared while the index calculated by the authors says they are not at risk and 19% of households being less worried than they should be about their retirement preparedness. The key drivers of being in the "not worried enough" group are having a DC plan and having high income. Households with a DC plan may indeed suffer from "wealth illusion", not realising how much income can be derived from their DC balances. In addition, high-income households may not properly assess how much wealth accumulation is required to maintain their standard of living. Assessment of policies 19. Policies aiming at increasing participation in DC pension plans fall into three broad categories: policies related to the enrolment process, policies providing financial incentives and policies increasing people's access to knowledge. Policies related to the enrolment process include making participation compulsory, enrolling workers automatically in pension plans, simplifying the choice for individuals and prompting an active decision about participating in a private pension arrangement. Financial incentives can help re-aligning people's wish to save for retirement with the action to join a private pension plan. Historically, tax incentives have been the main type of financial incentives to promote private pensions. Matching contributions and flat-rate subsidies are more recent types of financial incentives. Finally, increasing people's knowledge about the pension system and pension reforms, through better communication and financial education, can help people realise about the importance of saving for retirement and encourage them to take action. Policies related to the enrolment process 20. Making enrolment into private pensions compulsory is ultimately the most effective policy in reaching high and uniformly distributed coverage rates. As shown in Figure 1, coverage rates are higher in countries having mandatory or quasi-mandatory pension plans, usually at or above 70% of the working-age population. OECD (2012) also shows that coverage is more evenly distributed across different socio-economic characteristics in such systems, as compared to voluntary systems. Compulsory enrolment addresses the behavioural biases identified above (procrastination, myopia and inertia), ensuring that individuals save for retirement and start saving early in their career. 8

9 FOR OFFICIAL USE DAF/AS/PEN/WD(2017)3 Figure 1. Coverage of private pension systems As a percentage of the woking-age population Mandatory/Quasi-mandatory Auto-enrolment Voluntary Note: Coverage rates are provided with respect to the total working-age population (i.e. individual aged 15 to 64 years old) for all countries except Germany, Ireland, Sweden and the United Kingdom for which coverage rates are provided with respect to total employment. Source: OECD (2015a) and The Pensions Regulator (2016) for the United Kingdom. 21. One of the limits of compulsory enrolment is that it may not be necessary for all individuals depending on the design of the overall pension system. Forcing low income workers to contribute may lead them to become more indebted or divert funds from other necessary expenses, such as children education or housing. When these workers can already expect high replacement rates from the public pension system, making private pension contributions compulsory for them may not be justified. Recognising this problem, in Australia, only workers 18 years old or over and earning at least AUD 450 a month are entitled to mandatory employer contributions to the superannuation system. 22. Another limit to compulsory enrolment is when the informal sector is large. Workers outside the formal economy are not paying social security contributions, let alone pension contributions. In Mexico for example, nearly 60% of all workers are informal. In that country, informal workers are not only common among self-employed workers and unpaid workers, but also among salaried workers (according to the National Survey of Occupation and Employment, 45.5% of salaried workers were informal in the third quarter of 2014). This explains why the mandatory private pension system only covers around 60% of the working-age population (Figure 1). 23. As an alternative to compulsory enrolment, automatic enrolment has gained popularity in the last decade. In some countries, compulsory enrolment would be difficult to implement politically, because mandatory contributions to private pensions would be perceived as another tax. Automatic enrolment involves signing up people automatically to private pensions but giving them the option to opt out within specified timeframes and conditions. The policy exploits individual behavioural traits such as inertia and procrastination to make people engage in retirement and pension saving, while preserving individual choice and responsibility for the decision about whether to participate in a private pension arrangement. It is used in the context of occupational pension plans and 9

10 DAF/AS/PEN/WD(2017)3 FOR OFFICIAL USE changes the default enrolment from "not participating" to "participating" in the employersponsored pension plan. It is also associated with default options for the contribution rate and the investment strategy to simplify the choice of individuals. The policy has been introduced at the national level in Italy (2007), New Zealand (2007), Turkey (2017) and the United Kingdom (2012), it is developed at the state level in the United States (2012) and it is encouraged by regulation in Canada (2012) and the United States (2006). Table 2 provides a description of the different schemes. Table 2. Description of automatic enrolment schemes Country Canada Italy New Zealand Turkey United Kingdom United States Description A Pooled Registered Pension Plan (PRPP) is a kind of multi-employer DC pension plan in which unrelated employers and self-employed persons are eligible to participate. Where an employer elects to offer a PRPP, enrolment of employees would be automatic unless an employee chooses to opt out. The PRPP framework, introduced in 2012 at the federal level, will be fully in place across Canada pending provincial enabling legislation. So far, six provinces have introduced PRPP legislation. Automatic enrolment was introduced in January For all private sector employees, it involved the payment into pension funds of the future flow of the severance pay contributions (Trattamento di fine rapporto, TFR), set at 6.91% of salary. Individual workers were given a period of six months in order to decide whether to opt out of this arrangement, keeping their rights regarding the TFR as in the past. The same mechanism applies since then to all first-time private sector employees. KiwiSaver was introduced on 1 July Employers must enrol new employees (i.e. those starting a new job) into the scheme and individuals have 8 weeks to opt out. The minimum contribution is 3%, which is deducted from employee earnings, and an employer contribution of 3% of salary is added. The government also contributes 50 cents for every dollar of member contribution annually up to NZD Existing employees not subject to the automatic enrolment rule can also join (opt in) the KiwiSaver plan on a voluntary basis. The kick start payment of NZD has been removed for contracts opened after 21 May Since 1 January 2017, employers have to choose a private DC pension plan and automatically enrol employees younger than 45 into it. Employees may choose to opt out of the system within the first two months following their automatic enrolment. Employers are not required to contribute, while employees must contribute at least 3% of their gross income. The government matches 25% of an employee's contributions and makes an additional one-time contribution of TRY 1,000 for those who do not opt out within the first two months. The duty on employers is being staged in between January 2017 and January 2019, starting with the largest employers and the public sector. Automatic enrolment has been introduced in 2012 for all those workers who are not already covered by a private pension arrangement. Employers are required to automatically enrol their eligible jobholders (worker aged at least 22 but under State Pension Age, who earn more than GBP 10,000 a year) into a qualifying workplace pension. Workers can opt out within one month, and if so, will be automatically re-enrolled by their employer on a three-year cycle. Employer and employee contributions are being phased in from October 2012 to a minimum total contribution of 8% of qualifying earnings by October The duty on employers is being staged in between October 2012 and February 2018, starting with the largest employers. - Pension Protection Act: Automatic enrolment in 401(k) pension plans was introduced in 1998 for newly hired employees. Since 2000, the automatic enrolment was extended to current workers who were not enrolled in a pension scheme. In 2006, the adoption of the Pension Protection Act greatly encouraged automatic enrolment by giving employers incentives to automatically enrol their employees into a retirement savings plan. - Automatic IRA: Since 2012, several states have created state-facilitated retirement savings plans following the "Automatic IRAs" model, which has been proposed by the Obama Administration in 2009 but has never been enacted. 24. As detailed in OECD (2014), available evidence in the United States, New Zealand and Italy shows that automatic enrolment has a positive impact on participation in retirement savings plans. However, its impact on participation may be reduced when other incentives compete and interact negatively with its introduction. Several studies in the United States demonstrate that automatic enrolment is associated with significant 10

11 FOR OFFICIAL USE DAF/AS/PEN/WD(2017)3 increases in 401(k) plan participation. 1 In terms of participation, the KiwiSaver scheme in New Zealand is a real success with nearly 2.7 million people enrolled as at December 2016, of which 41% have been enrolled automatically. In addition, KiwiSaver members' distribution by income is similar to the one of the eligible population, meaning that lowincome individuals are not disproportionately left out of the scheme (Inland Revenue, 2015). Finally, while the increase in pension fund coverage in Italy following the TFR reform was significant, with 1.4 million additional workers enrolled in a pension fund between 2006 and 2007, only around 5% of new members in that period were automatically enrolled. As shown in Figure 1, private pensions covered less than 20% of the working-age population at the end of 2013, six years after the TFR reform. Rinaldi (2011) identifies structural and implementation factors to explain the relative failure of the reform. Fornero and Monticone (2011) show that the lack of knowledge of basic financial concepts among Italians may also have reduced the impact of automatic enrolment. They find that financial literacy increases the probability of participating in a pension fund, as well as the probability of transferring the TFR contributions to a pension fund. 25. Despite the success of automatic enrolment at the company level in the United States, membership in 401(k) plans has remained broadly constant at the national level after the adoption of the Pension Protection Act in 2006, leading several states to take further actions. According to data from the US Department of Labor, 401(k) coverage increased from 27.5% of the working-age population in 2005 to 29.1% in 2006 and stagnated at 29-30% thereafter. To further increase coverage, the US Administration proposed in 2009 the "Automatic IRAs" programme, aiming at mandating employers not offering an occupational pension plan to automatically enrol their employees in an Individual Retirement Arrangement (IRA), with contributions deducted from the salary and an opt-out option for the employee. Despite this, no legislation has been enacted at the federal level yet. Several states have stepped in and created state-facilitated retirement savings plans following the "Automatic IRAs" model (John & Antonelli, 2017; Munnell, Belbase, & Sanzenbacher, 2016). According to the Georgetown Center for Retirement Initiatives, since 2012, 40 states have considered to establish state-facilitated retirement savings programmes. Seven states have already passed legislation (California, Connecticut, Illinois, Maryland, New Jersey, Oregon and Washington). These plans use private sector providers to manage investments and provide other services. By combining state facilitation with private providers, they aim at allowing small employers to offer simple, low-cost retirement savings plans to their employees Automatic enrolment also has a positive impact on participation in the United Kingdom and in Turkey. Cribb and Emmerson (2016) show that automatic enrolment in the United Kingdom led to an increase of 37 percentage points in the probability of participating in a workplace pension scheme for eligible private sector employees. According to The Pensions Regulator (2016), as at March 2016, 66% of all employees were active members of a pension scheme (Figure 1), compared with just 47% in Only 4.5 months after the introduction of the new system in Turkey (as at 19 May 2017), 1 See Madrian and Shea (2001), Choi et al. (2001), Choi et al. (2004), Beshears et al. (2009) and Butrica and Karamcheva (2015). 2 It is not clear what will happen to these plans already stablished in seven states after the Senate, in May 2017, voted to reverse the policy. States may still be able to set up plans as long as they comply with ERISA rules (therefore excluding the use of IRAs). 11

12 DAF/AS/PEN/WD(2017)3 FOR OFFICIAL USE nearly 2.9 million employees had already been automatically enrolled. As a comparison, the individual voluntary system was introduced in 2003 and covered 5.6 million individuals at the end of Other forms of enrolments have been tested voluntarily by employers in the United States. As an alternative to traditional opt-in programmes (employees are not enrolled in the 401(k) plan unless they make an affirmative election) and to automatic enrolment (employees are automatically enrolled and can opt out), Choi, Laibson and Madrian (2009) and Beshears et al. (2013) studied the effects of Quick Enrollment. This programme gives workers the option of enrolling in the 401(k) plan provided by their employer by opting into a pre-set default contribution rate and asset allocation. The goal of this policy is to reduce complexity. Instead of evaluating all possible contribution rate and asset allocation options, employees just face a binary choice between participating based on the default options provided by the programme and nonparticipating. The authors find that Quick Enrollment resulted in substantial 401(k) participation increases, although typically smaller than automatic enrolment: it increased the participation among new hires by 16 percentage points after three months of tenure (relative to a standard enrolment mechanism in which employees must actively select both a contribution rate and an asset allocation); and it prompted 10% to 20% of previously hired employees who were not participating in their 401(k) plan to enrol in the plan. They also find that the participation increases produced by Quick Enrollment are durable and that employees who join the pension plan in this way often remain at the default contribution rate and asset allocation for years. 28. Automatic enrolment and Quick Enrollment involve default options, but defaults may not be suitable when they apply to a large number of people with heterogeneous needs and preferences. Studies show that most participants enrolled under those arrangements stick to both the default contribution rate and the default asset allocation for long periods. For example, KiwiSaver members tend to retain the default contribution rate they have been assigned to when joining the system. Inland Revenue statistics show that as of 30 June 2011, 80% of people who joined KiwiSaver after April 2009 contributed 2% (the default between April 2009 and April 2013), while 62% of those who joined before that date were still contributing 4% (the default before April 2009). However, even well-chosen defaults may not be optimal when they apply to individuals with highly heterogeneous situations. For example, in a company whose workforce includes both financially constraint young parents and older employees who need to save for their retirement, a single contribution rate to the occupational pension plan may not be appropriate. Default contribution rates and investment strategies could vary according to some observable demographic characteristics such as age, but unobserved employee heterogeneity may limit the usefulness of such employee-specific defaults and the practical implementation may be difficult (in particular from a legal perspective as not all employees would be treated equally). 29. An alternative to default mechanisms is to force people to take active decisions about their participation in a pension plan. Carroll et al. (2009) compare two kinds of 401(k) enrolment: standard enrolment (i.e. the default is not to participate) and active decisions (i.e. there is no default but rather a compulsory choice between participating or not). On the one hand, the active decision mechanism encourages individuals to think about an important decision and avoid procrastinating. On the other hand, an active decision mechanism requires individuals to deal with a potentially time-consuming and complex issue and then explicitly express their decision at a time which may be inconvenient. The authors describe a natural experiment at one firm and find that 12

13 FOR OFFICIAL USE DAF/AS/PEN/WD(2017)3 compelling new hires to make active decisions about 401(k) participation raises participation by 28 percentage points relative to a standard opt-in enrolment procedure after three months of tenure. In addition, contribution rates in the active decision cohort reach levels that would take 30 months to achieve under standard enrolment, i.e. employees immediately choose a contribution rate that is similar on average to what they would take up to 30 months to attain under standard enrolment. Providing financial incentives 30. Historically, tax incentives (tax exemptions, tax deductions, tax credits or rate reliefs) have been the main type of financial incentives provided by governments to promote private pensions. DAF/AS/PEN/WD(2017)2 shows that tax incentives, especially tax deductions, encourage participation in private pension plans. The impact of tax deductions on participation increases with income, as high-income individuals usually face higher marginal tax rates. Individuals also respond to changes in tax incentives (e.g. tax reforms, changes in contribution limits) by adjusting their participation in private pension arrangements. 31. The evidence on the impact of rate reliefs on private pension participation is less conclusive. For example, Feng (2014) fails to demonstrate a positive impact of a reduced flat tax rate on contributions on participation in salary sacrifice arrangements in Australia. A salary sacrifice arrangement is where employees agree contractually and voluntarily to give up part of the remuneration that they would otherwise receive as salary or wages, in return for their employer providing contributions to a superannuation fund of the same amount. Salary sacrifice contributions attract a favourable tax treatment, being taxed at a fixed rate of 15% (same as for mandatory employer contributions), rather than at the marginal income tax rate. Using data from wave 10 of the Household Income and Labour Dynamic in Australia survey, the author determines the effect of higher tax incentives on the participation in salary sacrifice arrangements at two marginal tax rate jump points of the personal income tax system, from 15% to 30% and from 30% to 38%. 3 The results indicate that tax incentives lead to a small but insignificant increase in participation in salary sacrifice arrangements. The author explains that this result is likely due to the complexity of the incentive schemes and competing demands for long-term savings. In addition, the name of the arrangement, "salary sacrifice", is framed as a loss which may discourage participation. 32. Tax incentives can be complemented by other types of financial incentives to encourage participation in private pension, such as matching contributions (from the state or from the employer) and state flat-rate subsidies. These incentives are provided to eligible individuals who actually participate or make voluntary contributions to the private pension system. Both matching contributions and state subsidies are paid in the pension account, thus increasing the assets accumulated to finance retirement. 33. After tax incentives, matching contributions are the most common type of financial incentive used by OECD and EU countries to promote private pensions (OECD, 2015b). Usually, the matching contribution is conditional on the individual contributing and corresponds to a certain proportion of the individual's own contribution, up to a maximum amount. The generosity of the match rate varies greatly across countries, from 3 The higher the marginal tax rate, the higher the drop in the tax rate on contributions and therefore the higher the incentive to contribute. 13

14 DAF/AS/PEN/WD(2017)3 FOR OFFICIAL USE 3% in Austria (paid by the state) to 325% in Mexico (paid by the state for civil servants). A match rate of 50% can be found in Australia (paid by the state), Iceland (paid by the employer) and New Zealand (paid by the state). Matching contributions usually apply to all workers, but can also be targeted to specific groups, such as young workers (as in Chile) or low-income individuals (as in Australia). 34. Empirical evidence from a large body of the literature in the United States suggests that matching contributions increase participation in private pension plans. Choi (2015) and Madrian (2013) review the rich literature in the United States regarding the effect of matching contributions on participation and conclude that matching contributions increase participation. Madrian (2013) however qualifies the quantitative impact as small. This statement relies on a study by Engelhardt and Kumar (2007). Indeed, using a nationally representative random sample of individuals aged 51 to 61 and their spouses from the Health and Retirement Survey, they find that increasing the match rate by 25 percentage points (for example, from 25 cents per dollar of employee contribution to 50 cents) raises 401(k) participation by 5 percentage points. Choi (2015) reports that an increase of 10 percentage points in the match rate increases the participation rate in a range from 2.5 percentage points to 6.3 percentage points, depending on the study. 35. Flat-rate subsidies paid by the state into pension accounts can be found in Chile, Germany, Lithuania and Mexico. They are designed to attract low-income individuals, as the fixed subsidy represents a higher share of their income. In Germany, financial institutions can offer so-called Riester pension plans since The state encourages people to contribute to a Riester plan via two types of incentives: tax exemptions and flatrate subsidies. Flat-rate subsidies are particularly valuable for low-income individuals, while tax deductions attract more high-income individuals. There are three types of flatrate subsidies: a basic subsidy of EUR 154 per year and per person, a child subsidy of EUR 300 per year and per child and a young worker subsidy of EUR 200 granted once at the age of 25. In order to receive the maximum subsidy, the sum of the member's contributions and the subsidy must be at least equal to 4% of his/her previous year's annual income. Only very low-income households can get the full subsidy without investing 4% of their income if they contribute at least EUR 60 annually. In 2012, 44% of eligible households were covered by the Riester scheme (Börsch-Supan et al., 2016). 36. Evidence related to the introduction of the Riester pension scheme in Germany shows that subsidies are effective in attracting families with children and low-income earners. Indeed, Pfarr and Schneider (2013) find that child subsidies have an impact on the take up of Riester plans. In addition, Börsch-Supan, Coppola and Reil-Held (2012) and OECD (2012) show that despite the fact that coverage rates of Riester pensions increase with income, Riester pensions are more equally distributed by income than occupational pensions and unsubsidised private pension plans. For example, in 2009, only around 19% of households in the lowest income bracket had taken up Riester pension plans, while almost a half of those in the two upper income brackets participated in the scheme. However, the distribution of participation is even more skewed towards highincome individuals for occupational pension plans and other private pension plans, which cover only around 5% of households in the lowest income bracket. Therefore, among low-income households, Riester pensions are much more common than any other form of private pension provision. As a result, Riester pensions are more equally distributed by income than occupational pensions or unsubsidised private pension schemes. 14

15 FOR OFFICIAL USE DAF/AS/PEN/WD(2017)3 Increasing access to information, knowledge and skills 37. Policy makers increasingly recognise the role of financial education in supporting individuals to plan for their retirement. OECD (2016a) summarises the different financial education tools used in different jurisdictions to address various financial education needs in relation to retirement planning. These can be split into three broad categories: Providing information on retirement options and increasing awareness of retirement issues: this includes general information through websites, awareness campaigns covering retirement issues, comparison tools presenting plan features in a standardised way, personalised pension statements, access to personal information online, as well as calculators and simulators; Instruction: this takes the form of seminars and workshops about retirement planning, helping participants acquire financial knowledge and skills relevant for retirement, explaining the risks that individuals may be exposed to and suggesting how to manage them, and helping individuals estimate their retirement income needs, with the ultimate goal to help them taking decisions about their pensions; Advice: this ranges from factual information to fully personalised advice. 38. Evidence on the effectiveness of financial education for retirement is currently limited but suggests that workplace financial education can be effective in increasing enrolment in occupational pension plans (Atkinson, Messy, Rabinovich, & Yoong, 2015). Workplace financial education initiatives that are efficient in increasing participation in private pension arrangements include seminars and workshops and online courses. For instance, Duflo and Saez (2003) studied a university that encouraged a random sample of its employees to attend to an annual event providing information on benefits, including an occupational pension plan. They found that 5 to 11 months after the event, plan participation was higher in treated departments (i.e. those where a random sample of employees received an invitation letter promising a reward for attending the event) than in control departments. Collins and Urban (2016) show that online financial education courses offered to employees increases self-reported IRA participation by six percentage points. 39. In addition, providing more information on the employer's pension plan and how to join it can also increase participation. Clark, Maki and Morrill (2014) find that young employees (18 to 24 years old) who received a flyer containing information about their employer's 401(k) plan and the value of contributions compounding over a career, were more likely to begin contributing to the plan compared to workers of a similar age that did not receive the flyer. Lusardi, Keller and Keller (2009) study the impact of helping employees form and implement a savings plan through the provision of a planning aid that (a) encourages individuals to set aside a specific time for enrolling in their savings plan, (b) outlines the steps involved in enrolling in the plan (e.g., choosing a contribution rate and an asset allocation), (c) gives an approximation of the time each step will take, and (d) provides tips on what to do if individuals get stuck. This planning aid increased enrolment in the studied occupational pension plan by 12 to 21 percentage points for newly hired employees. 40. The take up of financial advice may be increased by facilitating the payment for advice. For example, the United Kingdom has introduced a pension advice allowance to enable people to withdraw money from their DC pension plan to access pension and retirement advice. Since April 2017, pension providers are able to offer the allowance to their members. Individual members and beneficiaries can, at any age, withdraw GBP 500 tax free, no more than once in a tax year, and up to a maximum of three times in total. 15