Stéphanie Payet
OECD Pensions Outlook 2024
Improving Asset-backed Pensions for Better Retirement Outcomes and More Resilient Pension Systems
Abstract
This chapter assesses whether investing in equity markets leads to better retirement income outcomes for members of defined contribution pension plans. It first looks at the current practice and trends in equity investments in defined contribution pension schemes across a wide range of countries. It then assesses the impact of investing in equities on investment performance, on the level of assets accumulated at retirement and on replacement rates.
Defined contribution (DC) pension plans rely on investing the contributions of plan members in capital markets to generate investment income and provide income in retirement.1 However, selecting the most appropriate investment strategy is not straightforward, especially when plan members bear the investment risk individually. Contributions may be invested according to various investment strategies with different risk-return characteristics. In addition, investing these contributions involves solving a trade-off between maximising future retirement income and providing protection from extremely low retirement income (OECD, 2020[1]). Typically, equities produce higher expected returns than bonds over the long term, but they also carry more risk as equity returns are more volatile, especially in the short term (Siegel, 2023[2]). By contrast, conservative investment strategies reduce volatility, but they can result in lower retirement incomes as they generate lower expected returns. Plan members and policy makers, therefore, must find a balance when selecting the most appropriate investment strategy for themselves or for the default option.
This chapter assesses whether investing in equity markets leads to better retirement income outcomes for members of DC pension plans. It starts by looking at the current practices and recent trends in terms of equity investment of DC pension schemes across a wide range of jurisdictions to understand the relevance of that policy question. The chapter then presents three complementary analyses that provide a comprehensive assessment of the impact of equity investment on retirement income outcomes by looking at different indicators. The first analysis looks at case studies of existing DC pension schemes to analyse the link between equity investment and average performance. However, publicly available data only go back 20 years, which is insufficient to assess the impact of investing in equities over an entire career. The second analysis addresses this limitation by using historical returns on selected asset classes since 1900 in 19 OECD countries to calculate the level of assets that successive cohorts of individuals would have accumulated at retirement had a DC pension plan existed in those countries.2 It compares various illustrative investment strategies with different levels and profiles of exposure to equities. The third analysis is forward looking and brings in, on top of investment risk, risks related to discount rates, labour markets and life expectancy. It uses a stochastic model to generate the distribution of replacement rates that the same illustrative investment strategies would provide under different payout options.
The analyses conclude that investing in equity markets would, in general, lead to better retirement income outcomes for members of DC pension plans. Investing in equities is likely to provide a higher average performance, more assets accumulated at retirement and higher replacement rates than investing only in fixed income. Even during the payout phase, people taking regular drawdowns are likely to receive higher pension benefits when investing a significant part of their savings in equities. However, investing in equities comes with caveats, as it requires individuals to invest for long periods to get the full potential from the compounded return effect, it leads to volatile outcomes for individuals and societies, and it makes pension benefits sensitive to equity market downturns happening when people are close to retirement. This trade-off makes it difficult to determine an exact or approximate optimal equity exposure. While conservative investment strategies reduce volatility, they provide only moderate protection to members of DC pension plans and lead to sub-optimal outcomes during the payout phase when taking regular drawdowns. Finally, when considering bequests, combining equity investment with regular drawdowns may bring higher total pension payments (benefits while alive plus bequests) than taking a lifelong annuity, but without full longevity protection. The main policy implication of the analysis is that countries should avoid investment regulations that favour investment strategies that are too conservative, in particular for the default option.
The findings in this chapter are consistent with the existing literature comparing investment strategies in the context of DC pension schemes. For example, two studies find that conservative investment strategies underperform life-cycle investment strategies and strategies with constant equity exposure. Anarkulova, Cederburg and O’Doherty (2023[3]) find that investing in government bills produces lower average balances at retirement and leads to a higher risk of outliving one’s resources during retirement. Khemka, Steffensen and Warren (2021[4]) show that the risk-free strategy produces the lowest balance at retirement at the median and the 10th percentile. Similarly, Berardi and Tebaldi (2023[5]) compare an investment strategy providing a minimum return guarantee and investing 95% in long-term bonds and 5% in equities to three life-cyle investment strategies, and find that the minimum guarantee strategy produces the lowest internal rate of return at the median and the fifth percentile, and the lowest performance adjusted for risk.3 In addition, several studies show that portfolios with a high allocation to equities present a trade-off between risk and return (Dunn and Berg, 2019[6]; Khemka, Steffensen and Warren, 2021[4]; Šebo, Danková and Králik, 2020[7]). By contrast, Anarkulova, Cederburg and O’Doherty (2023[3]) find that, in the United States, the internationally diversified equity portfolio outperforms all the other strategies, producing higher wealth at retirement, higher income replacement rates and higher bequests on average, without reducing outcomes in unfavourable scenarios and with a lower risk of outliving one’s resources during retirement. Finally, Šebo, Danková and Králik (2020[7]) find that life-cycle investment strategies reducing linearly the exposure to equities with age are outperformed by fixed portfolio strategies with the same average equity exposure over 40 years, both in terms of balance at retirement and maximum fall in assets during the accumulation phase. They suggest that life-cycle strategies with a glide path based on age and price movements of equities perform better. For Khemka, Steffensen and Warren (2021[4]), the glide path of life-cycle investment strategies should take into account the level of risk aversion of the individual.
This chapter complements the existing literature. While most studies compare investment strategies using stochastic modelling, Monte Carlo simulations or bootstrapping, this chapter combines stochastic modelling with historical analyses that look at the actual performance of selected pension funds and use past investment returns for selected asset classes to build asset accumulations for successive cohorts of individuals. Additionally, the analysis in this chapter compares investment strategies under two different payout options, lifelong annuities and regular drawdowns. Another interesting feature of this analysis is that it covers more than one country and shows that the main results apply to a range of countries.
This chapter looks in Section 4.1 at the current practices and recent trends regarding equity investments of DC pension schemes using aggregate national data. Section 4.2 presents the case studies of existing DC pension schemes in selected OECD countries. Section 4.3 presents the results of the analysis using historical returns, while Section 4.4 presents the results of the stochastic analysis. Section 4.5 concludes and presents policy implications.4
This section finds that, in many jurisdictions, equity exposure of DC pension schemes has increased over the past 20 years, although the ways they invest in equities and the trends vary across jurisdictions. The section looks at the current practices and recent trends regarding the equity investments of DC pension schemes across a wide range of jurisdictions to understand the relevance of assessing whether investing in equities leads to better outcomes for plan members. It compares the equity investments of DC pension schemes across 23 OECD countries and 15 non-OECD jurisdictions using the OECD Global Pension Statistics database.5 It presents data on asset allocation at end-2022 and uses a sub-sample of jurisdictions to analyse the evolution of equity exposure between end-2001 and end-2022.
The allocation of DC pension schemes into equities varies greatly across jurisdictions. Figure 4.1 presents the aggregate allocation of DC pension schemes into equities (public and private), bills and bonds, cash and deposits, and other investments at the end of 2022. Investment through collective investment schemes (CIS) is only shown when the look-through into the previous asset classes is not available. In some jurisdictions, CIS investment is used to invest indirectly in equities (e.g. Bulgaria). At the end of 2022, total equity exposure represented less than 10% of total investment in Albania, the Czech Republic (hereafter, Czechia), Finland, the Maldives and Nigeria. At the other extreme, total equity exposure represented 60% or more of total investment in Estonia, Hong Kong (China), Lithuania and Poland. On average across all jurisdictions, total equity investment represented 30% of total investment at the end of 2022. Bills and bonds made up the largest share on average, at 47%.
As a percentage of total investment
Note: The category “Collective Investment Schemes (CIS)” is shown only when the look-through into different asset classes is not available. Jurisdictions are ranked in descending order according to the aggregate allocation to equities and CIS.
Source: OECD Global Pension Statistics and Board of Governors of the Federal Reserve for the United States.
Private equity, i.e. unlisted equities and private equity funds, usually represents a small portion of total equity investment. In most jurisdictions with available data, equity investment only comprises listed equities (Figure 4.2). This is due to the fact that many jurisdictions only authorise equites traded on regulated markets (e.g. Armenia, Bulgaria, Costa Rica, Czechia, Greece, Hong Kong (China), Pakistan and Serbia) (OECD, 2024[8]). Figure 4.2 shows that among jurisdictions reporting private equity investment, unlisted equities and private equity funds jointly represented between 2% (in Australia) and 36% (in Finland) of total equity holding at the end of 2022.
As a percentage of the total equity holding
Note: (1) The category “listed equities” includes direct investment in listed equities and equity investments through collective investment schemes.
Source: OECD Global Pensions Statistics.
The extent to which DC pension schemes invest in equities through foreign or domestic companies varies widely across jurisdictions. Figure 4.3 shows that foreign equities represented more than 50% of total equity exposure at end-2022 in Chile, Croatia, Lithuania, Mexico and North Macedonia. By contrast, DC pension schemes in some countries invest in equities mostly through domestic companies. At the end of 2022, foreign equities represented 10% or less of total equity exposure in Latvia, Poland and Portugal.
As a percentage of total equity investments
Note: The data exclude investments in private equity funds.
Source: OECD Global Pension Statistics.
Quantitative investment limits may constrain investment in equities. Several jurisdictions do not apply quantitative investment limits but request pension entities to invest according to the prudent person principle. This is the case, for example, in Australia, Denmark, Luxembourg, Malawi and the United States. Other jurisdictions have quantitative investment limits, but not for all asset classes. For example, there is no regulatory limit for equity investments in Hong Kong (China), Hungary, Italy, Latvia, the Maldives, Portugal, Slovenia, Spain and Türkiye. Among the jurisdictions with regulatory limits for equity investment, limits vary from 0%, where equity investment is not allowed, to 80% (OECD, 2024[8]).,6 At the end of 2022, equity investment, excluding private equity funds, represented 27% of the portfolio on average among the jurisdictions with quantitative limits for equities, while it was 32% on average among the jurisdictions without quantitative limits. Quantitative investment limits for equity investment may therefore be binding in some jurisdictions, although other factors may be at play.
Over the past 20 years, equity exposure has evolved differently across jurisdictions but it has in general increased in many jurisdictions. Figure 4.4, Figure 4.5, Figure 4.6 and Figure 4.7 show the evolution of equity exposure between end-2001 and end-2022, splitting selected jurisdictions into four groups according to the evolution of equity exposure in the portfolio of DC pension schemes. Figure 4.4 includes jurisdictions where equity exposure has remained relatively stable between end-2001 and end-2022, although fluctuations may have happened in-between. Figure 4.5 includes jurisdictions where equity exposure increased significantly in the first part of the period and then stabilised. Figure 4.6 includes jurisdictions where equity exposure has increased sharply in recent years to reach more than 60% of total investment at the end of 2022. Figure 4.7 includes jurisdictions where equity exposure increased gradually over 2001-22.
Note: 1. Equity exposure also includes investment in CIS as the look-through into different asset classes is not available.
Source: OECD Global Pension Statistics and US Federal Reserve.
Note: 1. Equity exposure also includes investment in CIS as the look-through into different asset classes is not available.
Source: OECD Global Pension Statistics.
Note: 1. Equity exposure also includes investment in CIS as the look-through into different asset classes is not available.
Source: OECD Global Pension Statistics.
Note: 1. Equity exposure also includes investment in CIS as the look-through into different asset classes is not available.
Source: OECD Global Pension Statistics.
Several factors may explain the evolution of equity exposure. Market fluctuations change the value of assets and can affect portfolio composition when the impact varies across asset classes. For example, the 2008 global financial crisis reduced the equity exposure in most jurisdictions as equity values dropped significantly while bond values remained broadly constant. For instance, in Peru, the share of equities in total investment fell from 50% at end-2007 to 32% at end-2008. The equity share was back at 48% at end-2009 thanks to the market recovery. Additionally, in the context of DC schemes with investment choice, the reduction in equity allocations in times of market downturns may also reflect investment switches by plan members to more conservative investment strategies.7
Regulatory changes may also influence the evolution of equity exposure. For example, in Poland, the equity exposure doubled from 41% to 82% between end-2013 and end-2014 as a pension reform banned open pension funds from investing in domestic government bonds, which led to the transfer of all domestic government bonds held by these funds to the social security system. Lithuania introduced life-cycle funds in 2019 leading to a substantial re-allocation of assets from bonds to equities. Between end-2018 and end-2019, equity exposure increased from 43% to 75% of total investment. Estonia relaxed some investment restrictions in 2019 for the second pension pillar. Under the Investment Fund Act 2019, the equity investment limit was raised from 0% to 10% for conservative funds and from 75% to 100% for the other funds. Following this change, pension funds gradually began to increase their equity investments and new funds with full equity exposure were created. Chile introduced the multi-funds system in 2002, leading to an increase in equity exposure in the following years. Before 2002, assets could only be invested in Fund C (with 15% to 40% in equities) and Fund E (with up to 5% in equities). With the introduction of additional funds, a growing proportion of assets has been invested in riskier funds, especially in Fund B (with 25% to 60% in equities) that became the default option for new entrants up to age 35.
The level of maturity of the DC pension system can also explain the evolution of the equity exposure. When jurisdictions introduce a new DC system, equity exposure tends to be small initially and to increase as the system matures, i.e. when DC plan sponsors and members gain comfort in equity investing, often as the domestic capital market deepens and grows. For example, Bulgaria introduced a mandatory DC scheme in 2000. Equity exposure increased gradually from 1% at end-2001 to 31% at end-2007 and then fluctuated around that level up to 2022.
This section studies the link between equity exposure and average investment performance. It provides anecdotal evidence that average performance of DC pension schemes tends to increase with equity exposure, although other aspects of portfolio composition also affect performance, regardless of the equity exposure.8 Equity exposure also increases the volatility of annual investment returns. The section studies existing DC pension funds at the country level for Chile, Czechia and France, and at the pension entity level in Australia (AustralianSuper and QSuper), Denmark (PFA), Sweden (AP7 and Alecta), the United Kingdom (Nest) and the United States (Thrift Savings Plan). Annex 4.A provides a brief description of the pension funds studied. Publicly available data cover the past 7 to 21 years, depending on the fund.9
Members of DC pension plans have access to a wide range of exposures to equities within countries and within pension entities. Table 4.1 presents the average equity exposure, the average performance and the volatility of returns of the studied DC pension funds over the longest available period up to the end of 2022. It shows that equity exposure varies significantly across pension funds.10 In several countries or entities there are pension funds with extreme equity exposures (0% or 100%) used as building blocks for diversified or life-cycle investment strategies.
Notes: See Annex 4.A for a brief description of the different pension funds. “..” means “not available. Returns are measured in local currency. 1. Average equity exposure measured over 2008-2022.
1. Strategic equity exposure as of July 2022.
2. Performance measured over 2012-22.
3. Equity exposure at end-2022.
4. Equity exposure at end-2022 and performance measured for people 30 years before retirement age.
5. Target equity exposure as of January 2023.
Source: OECD calculations based on data collected from the websites of pension supervisors and pension entities, and on data from Europerformance.
In general, performance increases with equity exposure. Table 4.1 shows that when comparing within countries or pension entities, funds with a higher average equity exposure have usually obtained a higher average annual real return. For each country or pension entity, Table 4.1 ranks pension funds in increasing order of equity exposure from top to bottom. In most of the cases, performance increases with equity exposure. For example, in Chile, there is a nearly linear positive relationship between the average exposure to equities and the average annual real return over the past 20 years. For every 10 percentage points increase in average equity exposure, the average annual real return increases by 22 basis points on average.
Equity is not the only asset class affecting performance and some funds have achieved lower average returns than other funds with lower equity exposure. For example, the Indexed Diversified option of AustralianSuper reached a lower performance (4.3%) than the Balanced and Socially Aware options (4.6% and 4.7% respectively) over 2009-22, despite having a higher equity exposure (70% compared to 58%). Beyond differences in equity exposure, the Indexed Diversified option relies on indexing strategies, has a larger exposure to fixed income than the two other funds (25% compared to 9%) and does not diversify through investment in infrastructure, private equity and property, unlike the two other funds.
The destination of investment may also affect the performance of equities. For example, in the United States, the I Fund of the Thrift Savings Plan is fully invested in global equities and achieved a lower performance over 2013-22 than the C Fund and the S Fund, which are fully invested in US equities. Similarly, the Socially Responsible option of QSuper barely invests in Australian equities and reached a lower performance over 2016-22 than the Balanced option and several of the life-cycle funds (with a mix of Australian and international equities), despite having a higher overall equity exposure.
Funds with higher equity exposure tend to experience a greater volatility of investment returns. Table 4.1 presents the standard deviation of real returns for all funds with available information. It shows that higher equity exposure is associated with higher volatility. This is not the case, however, for the Indexed Diversified option of AustralianSuper, which experienced a lower volatility of returns than other funds with lower equity exposure. Its larger exposure to fixed income may have reduced return volatility.
This section complements the previous analysis to assess whether investing in equities provides higher levels of assets accumulated at retirement. It expands the analysis to a full career of 40 years instead of the limited periods of time between 7 and 21 years available for the case studies. More importantly, it looks at the level of assets that individuals accumulate at the time of retirement as it matters more to them than the annual returns they get along the way.
This section calculates the level of assets that successive cohorts in 19 OECD countries would have accumulated at retirement, had a DC pension plan existed. It looks at different illustrative investment strategies and uses historical performance for different asset classes to compare outcomes across investment strategies with different levels and profiles of equity exposure. Box 4.1 describes the methodology.
The calculations assume that each cohort is represented by a typical individual. That individual joins the labour market at age 25 and receives wages that grow in line with historical inflation and a fixed productivity rate of 1.25%.1 Under the baseline assumptions, contributions are paid continuously over 40 years based on a constant contribution rate of 5%.2 The individual pays an annual 1% asset-based fee and retires at age 65. The assets accumulated at retirement are expressed as a multiple of the sum of contributions paid.
The analysis uses historical data on returns and inflation between 1900 and 2021 for 19 OECD countries. Assuming a career of 40 years (from age 25 to 65), this means that the level of assets accumulated at retirement is calculated for 83 successive cohorts retiring between 1939 and 2021 for each country.
The analysis compares the level of assets accumulated at retirement that the following illustrative investment strategies would have produced:
A fixed portfolio invested in domestic fixed income - 88% in government bonds and 12% in cash and deposits (“domestic fixed income”)3,4
A fixed portfolio invested 70% in domestic equities and 30% in international equities (“diversified equities”)5
A fixed portfolio invested 60% in equities and 40% in fixed income, with an equal mix of domestic and international securities for equities and government bonds (“diversified 60/40 portfolio”)6
A life-cycle investment strategy with a proportion invested in equities (70% domestic and 30% international) starting at 80% and declining linearly during the last 10 years before retirement to 20%, and with equities substituted by domestic fixed income (“life cycle 10y”)
A life-cycle investment strategy with a proportion invested in equities (70% domestic and 30% international) starting at 80% and declining linearly during the last 20 years before retirement to 20%, and with equities substituted by domestic fixed income (“life cycle 20y”).
Notes:
1. The analysis does not assume productivity improvement across cohorts. Each cohort has the same starting wage at age 25.
2. The assumed contribution rate reflects the fact that, in many OECD countries, DC pension plans are voluntary and complement public pay-as-you-go pension plans. In countries where DC pension plans are mandatory and represent the main source to finance retirement, contribution rates tend to be higher, between 10% and 20%.
3. These proportions are based on the average asset allocation of pension assets at the end of 2022 across OECD countries.
4. Treasury bill returns are used as a proxy for returns of cash and deposits.
5. In the case of countries from the European Union (EU), the domestic equity portfolio becomes an EU equity portfolio from 1990 to take into account the full liberalisation of capital movements between EU countries enacted that year.
6. The fixed-income part uses the same proportions between government bonds and cash and deposits as the domestic fixed income investment strategy.
The analysis shows that most cohorts in all countries would have accumulated significantly more assets at retirement had they invested at least part of their contributions in equities instead of only in domestic fixed income. For the minority of cohorts for which investing in domestic fixed income would have turned out to produce higher levels of assets accumulated at retirement, the additional level of assets achieved compared to investing in equities would have been modest. The analysis also shows that people need to save for retirement for long periods to get the full potential of investing in equities because the compounded return accumulates over time. However, higher investment in equities also leads to more volatile outcomes across cohorts.
Portfolios with some equity exposure would have led to higher assets accumulated at retirement on average across cohorts than a portfolio with only domestic fixed income. Figure 4.8 shows the average level of assets accumulated at retirement across cohorts by investment strategy for the different countries. The portfolio with domestic fixed income would systematically have produced lower accumulated assets on average compared to the other investment strategies. For the portfolio with domestic fixed income, the average level of accumulated assets varies between 1.9 and 3.5 times the sum of contributions depending on the country. By contrast, for the portfolio with diversified equities, the average level of accumulated assets varies between 4.5 and 11.2 times the sum of contributions. The other investment strategies usually fall in-between these two extremes.11
As a multiple of total contributions paid
Most cohorts would have been better off investing in equity markets. Figure 4.9 shows the proportion of cohorts that would have accumulated more at retirement by investing at least part of their contributions in equities rather than investing them only in domestic fixed income. In all countries, at least 80% of the 83 cohorts analysed would have been better off investing in equity markets. This proportion reaches 100% for the four portfolios with equity exposure in Australia, Austria, Belgium, Finland, New Zealand, Portugal and the United Kingdom. Looking for example at Denmark, Figure 4.10 shows that only four cohorts among all those retiring between 1939 and 2021 would have accumulated more at retirement with the portfolio with domestic fixed income than with the portfolio with diversified equities.
As a multiple of total contributions paid
Assets accumulated at retirement would have been significantly higher when investing in equities rather than only in domestic fixed income. Figure 4.11 presents the average difference in accumulated assets between the portfolio with domestic fixed income and the other four investment strategies. Panel A compares the portfolio with domestic fixed income with the portfolio with diversified equities. The blue bars show the average difference in accumulated assets for the cohorts that would have been better off with the portfolio with diversified equities, while the black bars show the average difference in accumulated assets for the cohorts that would have been better off with the portfolio with domestic fixed income. The other panels do the same, comparing the portfolio with domestic fixed income with the diversified 60/40 portfolio (light-grey bars), the life-cycle investment strategy reducing the equity exposure 10 years before retirement (light-blue bars) and the life-cycle investment strategy reducing the equity exposure 20 years before retirement (grey bars), respectively. Among cohorts better off with investment strategies with some equity exposure, the additional level of assets they would have accumulated at retirement compared to the portfolio with domestic fixed income tends to be substantial. For example, in Sweden, for cohorts better off with the portfolio with diversified equities (74 cohorts), the additional level of assets they would have accumulated at retirement compared with the portfolio with domestic fixed income reaches 6.1 times the sum of contributions on average. The average difference in accumulated assets corresponds to 2.8 times the sum of contributions for the diversified 60/40 portfolio, 3.7 times for the life-cycle investment strategy reducing the equity exposure 10 years before retirement, and 3.0 times for the life-cycle investment strategy reducing the equity exposure 20 years before retirement.
Note: FI means fixed income.
By contrast, for the few cohorts better off with the portfolio with domestic fixed income, the additional accumulated assets at retirement would have been modest compared to what would have been accumulated with the other investment strategies. Still in Sweden, the average additional level of accumulated assets achieved with the portfolio with domestic fixed income corresponds to 0.4 time the sum of contributions when comparing with the portfolio with diversified equities, 0.3 time when comparing with the diversified 60/40 portfolio, 0.5 time when comparing with the life-cycle investment strategy reducing the equity exposure 10 years before retirement, and 0.4 time when comparing with the life-cycle investment strategy reducing the equity exposure 20 years before retirement. Therefore, cohorts that would have invested in equities would have been neutral or only moderately worse off in the minority of cases when the portfolio with domestic fixed income would have produced higher assets accumulated at retirement. In the other cases, they would have been significantly better off.
However, given that portfolios with equity exposure produce more volatile outcomes, the likelihood for a cohort to accumulate a level of assets at retirement different than the other cohorts is greater when investing in equities than when investing in fixed income only. Figure 4.12 shows that the standard deviation of accumulated assets is higher for the portfolios with equity exposure in all countries. The longer and the higher the exposure to equities is, the higher the variability of accumulated assets tends to be. Hence, in most countries, the portfolio with diversified equities is the one producing the most volatile outcomes, leading to greater differences in accumulated assets across cohorts. By contrast, accumulated assets are much more comparable across cohorts with a portfolio with domestic fixed income only. However, this comes at the cost of lower levels of assets accumulated for most cohorts.
As a multiple of total contributions paid
Life-cycle investment strategies can partially address the issue of volatility across cohorts when equity market downturns happen at the end of the accumulation phase. On average across cohorts, the level of assets accumulated at retirement with the life-cycle investment strategies would have been lower in all countries than with the portfolio with diversified equities (Figure 4.8). This is because reducing the equity exposure prior to retirement lowers the overall rate of return. However, accumulated assets would have fallen less from one cohort to the next with the life-cycle investment strategies had equity markets dropped towards the end of the accumulation phase. For example, Figure 4.13 shows that in Belgium, accumulated assets with the portfolio with diversified equities would have fallen from 12.3 times the sum of contributions for the cohort retiring in 1999 to 5.6 times for the cohort retiring just 3 years later in 2002. With the life-cycle investment strategy reducing the equity exposure 10 years before retirement, the accumulated assets would have been stable at around 7 times the sum of contributions across cohorts retiring between 1999 and 2002.
As a multiple of total contributions paid
Investing in equity markets produces better results when people save for retirement for long periods as the compounded return accumulates over time. Many individuals may not be able to save for retirement during a full career. When participation in a pension plan is voluntary some individuals may delay saving into their 40s or 50s. Changing the length of the contribution period has different implications for the different investment strategies. In all the countries, average levels of accumulated assets across cohorts increase with the length of the contribution period for all investment strategies, but in greater proportions for portfolios with equity exposure. For example, Figure 4.14 shows that in the case of the United Kingdom, after only 20 years of contributions (starting to save from age 45 instead of 25), the different investment strategies would have produced relatively comparable outcomes, with average accumulated assets across cohorts ranging from 1.4 time the sum of contributions with the portfolio with domestic fixed income to 2.3 times with the portfolio with diversified equities. Differences are much larger after 40 years of contributions, with average levels of accumulated assets across cohorts ranging from 2.9 times the sum of contributions with the portfolio with domestic fixed income to 7.6 times with the portfolio with diversified equities. As the length of the contribution period increases, the effect of compounded return becomes stronger, in particular for portfolios with higher average returns.
As a multiple of total contributions paid
Shorter contribution periods reduce the proportion of cohorts that would have been better off investing in equity markets. However, most cohorts would have remained better off even with 20 years of contributions. Figure 4.15 compares the proportion of cohorts that would have accumulated more assets at retirement with the portfolio with diversified equities than with the portfolio with domestic fixed income after 20 or 40 years of contributions. For all the countries, the proportion of cohorts better off investing in equities increases with the length of the contribution period. With 20 years of contributions, the proportion of cohorts better off with diversified equities is still higher than 69% across countries, showing that investing in equities would still have been worthwhile for most people. Results are similar when comparing the diversified 60/40 portfolio and the life-cycle strategies with the portfolio with domestic fixed income.
The analysis now brings a forward-looking perspective and considers risks related to capital markets, labour markets and life expectancy to complement the historical analyses. It uses a stochastic model to simulate different realisations of the world and for each of them generate the replacement rates from a generic DC pension plan under different investment strategies and payout options. This allows to compare the distribution of replacement rates across investment strategies.12 Additionally, the analysis compares replacement rates and total pension payments (including bequests) when individuals take a lifelong annuity or take regular drawdowns in order to assess the impact of continuing investing in capital markets during the payout phase. Annex 4.B describes the stochastic model.
The analysis shows that, when assuming that individuals take a lifelong annuity, there is a probability of around 90% that investing in equities during the accumulation phase would result in a higher replacement rate than investing only in fixed income. However, high equity exposure up to the end of the accumulation phase makes the assets accumulated sensitive to equity market downturns close to retirement. When assuming that individuals take regular drawdowns, investing in equities also leads to higher replacement rates. However, higher equity exposure leads to more volatile replacement rates during retirement. When comparing the two payout options, the analysis shows that combining regular drawdowns with equity investment is likely to lead to higher total pension payments (i.e. adding up benefits while alive and bequests) than taking a lifelong annuity.
4.4.1. Replacement rates with a lifelong annuity
This sub-section compares the replacement rates that individuals taking a lifelong annuity would obtain at retirement under different investment strategies. The lifelong annuity pays fixed nominal benefits during the entire remaining lifespan of the individual in retirement. Under this payout option, individuals only invest their retirement savings during the accumulation phase.
Investment strategies with some equity exposure outperform the fixed-income portfolio when looking at the distributions of replacement rates. Figure 4.16 presents the five illustrative investment strategies in a risk-return diagram, using the median replacement rate as the return measure and the replacement rate at the fifth percentile as the risk measure.13 The higher the median replacement rate, the better. By contrast, the lower the replacement rate at the fifth percentile, the higher the risk because it indicates that individuals would receive a lower replacement rate in case of extreme unfavourable scenarios.14 The median replacement rate varies from 17% for the fixed-income portfolio to 39% for the equity portfolio when individuals take a lifelong annuity at retirement. Moreover, the fixed-income portfolio also produces the lowest replacement rate at the fifth percentile, meaning that in extreme negative circumstances that may only have a 5% probability of happening, the investment strategies with some equity exposure would still produce a higher replacement rate (12%-13%) than the fixed-income portfolio (9%). The fixed-income portfolio is therefore outperformed by the other strategies as it is likely to provide a lower return (lower median replacement rate) with a higher risk (lower replacement rate at the fifth percentile).
As a percentage of the last wage
Note: LC10 = Life cycle 10 years; LC20 = Life cycle 20 years.
In most cases, individuals annuitising at retirement would be better-off investing only in equities during the whole accumulation period. Table 4.2 compares the probabilities to obtain a higher replacement rate with each investment strategy compared to the others. For example, the second column shows that the equity portfolio would produce a higher replacement rate with a probability of 87% when compared to the fixed-income portfolio, 81% when compared to the 60/40 portfolio, 76% when compared to the life-cycle strategy reducing the equity exposure 10 years before retirement, and 78% when compared to the life-cycle strategy reducing the equity exposure 20 years before retirement. By contrast, the first column shows that the fixed-income portfolio would produce a higher replacement rate than the other strategies with a probability of no more than 13%.
The life-cycle strategy that reduces the equity allocation 10 years before retirement instead of 20 years is more likely to produce higher replacement rates for individuals taking a lifelong annuity. The median replacement rate is slightly higher when the reduction in the equity exposure happens later in the accumulation phase (31% versus 28%, Figure 4.16). Moreover, there is a 76% probability of obtaining a higher replacement rate with the life-cycle strategy reducing the equity allocation 10 years before retirement rather than 20 years. Reducing the equity exposure earlier rather than later reduces the expected return because fixed-income securities become prominent in the portfolio earlier in the accumulation phase and produce lower returns on average.
Investment strategies with some equity exposure are expected to improve replacement rates by a larger margin than the fixed-income portfolio for individuals annuitising at retirement. Figure 4.17 shows the median difference in replacement rates when the fixed-income portfolio produces a higher replacement rate than the alternative strategies (blue bars), and the median difference in replacement rates when the alternative strategies produce a higher replacement rate than the fixed-income portfolio (grey bars). Among the simulations where the fixed-income portfolio produces a higher replacement rate (between 9% and 13% of the cases according to Table 4.2), the fixed-income portfolio only improves replacement rates by up to 5 percentage points, showing that the extent of the loss when having some equity exposure is relatively modest. By contrast, when focusing on the simulations where the investment strategies with some equity exposure produce a higher replacement rate than the fixed-income strategies (between 87% and 91% of the cases according to Table 4.2), having some equity exposure improves replacement rates by 12 to 26 percentage points when compared to the fixed-income portfolio.
In percentage points
Falls in equity markets towards the end of the accumulation phase affect the replacement rates of the equity portfolio the most. Table 4.3 presents the median replacement rate produced by each investment strategy when considering all simulations, and when only considering the simulations where equity markets fall by at least 10% in the year just before retirement and in the fifth year before retirement. The median replacement rate from a lifelong annuity produced by the equity portfolio falls from 39% to around 30% when equity markets fall towards the end of the accumulation phase. Given that life-cycle strategies have a reduced allocation to equities in the final years of the accumulation phase, the effect of a market fall before retirement is more modest. For example, the replacement rate produced by the life-cycle strategy reducing the equity exposure 20 years before retirement remains at 28% when the equity market fall happens the year before retirement and declines to 26% when the equity market fall happens in the fifth year before retirement. The fixed-income portfolio is not sensitive to equity market movements.
Note: P(Eq better) = Probability that the equities portfolio produces a higher replacement rate.
Life-cycle investment strategies are nearly on a par with the equity portfolio when there is a fall in equity markets just before retirement. Table 4.3 shows that the median replacement rate produced by the life-cycle strategy reducing the equity exposure 10 years before retirement (30%) is only one percentage point below the one produced by the equity portfolio (31%) when equity markets fall in the year just before retirement. Moreover, the probability to be better-off with the equity portfolio than with the life-cycle strategy falls from 76% when considering all the simulations to 52% when equity markets fall in the year just before retirement. When comparing with the life-cycle strategy reducing the equity exposure 20 years before retirement, the probability to be better-off with the equity portfolio declines from 78% to 59%.
However, the attractiveness of life-cycle strategies diminishes when the equity market fall happens further away from the year of retirement. For example, the probability to be better-off with the equity portfolio than with the life-cycle strategy reducing the equity exposure 10 years before retirement increases from 52% when the equity market fall happens just before retirement, to 58% when the equity market fall happens in the fifth year before retirement. This is because there is some time to recover when the equity market fall happens five years before retirement. Having large equity exposure may allow to benefit from a potential market recovery. The lower equity exposure at the end of the accumulation period with the life-cycle strategy does not allow for the same potential rebound in asset values when the individual is annuitising at retirement.
4.4.2. Replacement rates with regular drawdowns and comparison with a lifelong annuity
This sub-section compares the replacement rates that individuals taking regular drawdowns would obtain at retirement under different investment strategies. The benefits paid under the regular drawdowns are determined by dividing the level of assets in the pension account at the end of each year by the remaining life expectancy. This payout option illustrates what happens when individuals keep investing their retirement savings during retirement. This sub-section also compares replacement rates between the two payout options, a lifelong annuity and regular drawdowns.
High equity exposure during the payout phase leads to better outcomes when taking regular drawdowns. Figure 4.18 presents the five investment strategies in the risk-return diagram and shows that the fixed-income portfolio and the life-cycle strategies are outperformed by the other two strategies. Having no equity exposure or keeping equity exposure at 20% during the payout phase as in the life-cycle strategies leads to lower median replacement rates and lower replacement rates at the fifth percentile. Given an average life expectancy of 20 years at 65, individuals still investing their assets during retirement would be better-off with higher equity exposures to get higher returns and higher benefits. The probability of obtaining a higher replacement rate with the investment strategies having some equity exposure than with the fixed-income portfolio ranges from 91% (when compared to the equity portfolio) to 94% (when compared to the balanced portfolio).
As a percentage of the last wage
Note: LC10 = Life cycle 10 years; LC20 = Life cycle 20 years.
However, higher equity exposure in retirement also implies a higher volatility of benefits received. Calculating the standard deviation of yearly replacement rates during retirement and taking the median across all simulations results in a volatility of replacement rates ranging from 10% with the fixed-income portfolio to 116% with the equity portfolio (Table 4.4). This means that benefit payments would vary significantly from one year to the next with the equity portfolio. This makes it more difficult for individuals to plan their expenditures as they would be uncertain about how much they would receive each year.
Taking regular drawdowns while keeping a high equity exposure during the payout phase tends to lead to higher replacement rates than taking a lifelong annuity. Figure 4.19 compares the median replacement rate and the replacement rate at the fifth percentile when the individual takes a lifelong annuity and when the individual takes regular drawdowns. When the regular drawdowns are associated with the equity portfolio, replacement rates are higher than with a lifelong annuity. When the regular drawdowns are associated with an investment strategy with a low equity exposure (0% or 20%), it is the other way around. Replacement rates are similar between the lifelong annuity and the regular drawdowns when savings are invested according to the 60/40 portfolio. Additionally, the probability for individuals to be better-off with the lifelong annuity than with the regular drawdowns (higher average replacement rate) is 87% with the fixed-income portfolio and 75% with the two life-cycle strategies. This probability falls to 48% with the 60/40 portfolio and 40% with the equity portfolio. The regular drawdowns can only produce higher benefits than the lifelong annuity when the average return of the portfolio between the time of retirement and the time when the individual passes away is higher than the discount rate at the time of retirement used to determine annuity payments. Portfolios with low equity exposures during the payout phase may fail to achieve a sufficient investment return.
As a percentage of the last wage
Note: Median = Median replacement rate; 5th pctl = Replacement rate at the fifth percentile.
As a multiple of total contributions paid
Individuals taking regular drawdowns may have assets left at the time they pass away and leave a bequest, unlike with a traditional lifelong annuity.15 The formula used to determine the benefits from the regular drawdowns avoids the risk for individuals of outliving their resources during retirement because the remaining assets are divided each year by a factor always greater than one (remaining life expectancy). However, it also implies that there will be assets left in the account when the individual passes away, and these assets can be passed on to heirs.16 Table 4.5 shows that individuals taking regular drawdowns would leave a significant bequest. The median bequest amount increases with the equity exposure in the portfolio during retirement.
When combining the sum of pension benefits received by individuals while alive and the bequests, total pension payments tend to be higher when taking regular drawdowns than when taking a lifelong annuity. Figure 4.20 compares total pension payments at the median and the fifth percentile when the individual takes a lifelong annuity and when the individual takes regular drawdowns. Unlike with replacement rates, total pension payments are larger with the regular drawdowns and the difference with the lifelong annuity is larger for investment strategies with higher equity exposure. Overall, the probability for individuals of being better-off with the regular drawdowns rather than with the lifelong annuity, when combining pension benefits and bequests, is equal to 57% with the fixed-income portfolio, 70% with the life-cycle strategies, 84% with the 60/40 portfolio, and 85% with the equity portfolio.
As a multiple of total contributions paid
Note: Median = Median total pension payments; 5th pctl = Total pension payments at the fifth percentile.
This chapter argues that investing in equity markets leads to better retirement income outcomes for members of DC pension plans. It assesses whether this is the case in a comprehensive manner by using complementary analyses. The analysis first looks at the current practice and trends of equity investment in DC schemes across a wide range of countries to help understand the relevance of the above policy question. The analysis then assesses whether investing in equities provides higher average investment performance by looking at actual case studies, higher levels of assets accumulated at retirement using historical returns, and higher replacement rates when incorporating capital, labour market and longevity risks using stochastic modelling.
4.5.1. Main findings
Equity investment represents a significant share of the portfolio of DC pension schemes and has been rising steadily over the past 20 years in many jurisdictions. At the end of 2022, the total equity exposure of DC schemes, including public and private equities, represented more than 40% of total investment in 13 out of 38 jurisdictions analysed. By contrast, total equity exposure represented less than 20% of total investment in 7 jurisdictions. Average equity exposures tend to be lower in jurisdictions capping equity investment.17 Moreover, there has been a general upward trend in the equity exposure of DC pension schemes in many jurisdictions, with an increase of more than 20 percentage points between 2001 and 2022 in 5 OECD countries. Therefore, the policy question to address is whether investing in equities may lead to better retirement income outcomes.
Investing in equity markets would lead to better retirement income outcomes for members of DC pension plans. The analysis shows that higher equity investment usually brings higher average performance when comparing DC pension funds that have different levels of equity exposure within countries or within pension entities. Still, diversification in terms of asset classes and geographical coverage plays a role in investment performance for a given equity exposure. Beyond investment returns, what matters to members of DC pension plans is the level of assets they will accumulate by the time of retirement and the pension benefits they will receive during retirement. Based on historical returns on selected asset classes, the analysis shows that a vast majority of cohorts (at least 80%) across 19 different OECD countries would have accumulated more assets at retirement had they invested at least part of their retirement savings for 40 years in a mix of domestic and foreign equities instead of only investing in domestic fixed income securities (government bonds plus cash and deposits). The analysis from the stochastic model considering uncertain risk variables provides consistent results, with a probability of 87% to 94% of getting a higher replacement rate when investing in equities compared to only investing in fixed income, depending on the investment strategy and payout option.
However, investing in equities comes with three main caveats:
It leads to better retirement income outcomes when people save for retirement for long periods, either by starting to save early or by delaying retirement. The analysis shows that cohorts that would have saved for 20 years instead of 40 would have accumulated less at retirement with all investment strategies, but the difference is much larger for portfolios fully invested in diversified equities than for other investment strategies. As the length of the contribution period increases, the effect of compounded return becomes more powerful, in particular for portfolios achieving higher average returns. Still, even with only 20 years of contributions, at least 69% of cohorts across countries would have been better-off if they had invested in equities rather than in domestic fixed income.
It leads to volatile outcomes for individuals and societies. The analysis from the case studies shows that DC pension schemes with higher equity exposures have usually achieved more volatile annual investment returns over the past 7 to 21 years. Additionally, the volatility of replacement rates increases with equity investment, and, during the payout phase, yearly benefits are also more volatile when retirement savings remain invested in equities rather than in fixed income. Individuals, therefore, face more uncertainty regarding their level of pension benefits when investing in equities, making financial planning in retirement more difficult. Uncertainty and volatility also apply at the level of societies, as successive cohorts of individuals are more likely to accumulate different levels of assets at retirement when savings are invested in equities instead of domestic fixed income. Volatile equity returns imply that two individuals born just one year apart could end up with large differences in accumulated assets at retirement despite having the exact same earnings history, contribution rate and investment strategy. Even though such differences could be seen as unfair and hard to understand, they should be nuanced by the fact that both individuals would in general be better-off with a portfolio invested in equities than with any other investment strategy.
It makes pension benefits sensitive to equity market downturns occuring when people are close to retirement. For example, the analysis from the stochastic model shows that the median replacement rate of the equity portfolio declines from 39% for all simulations to 31% for simulations with a fall in equity markets of at least 10% in the year just before retirement, when assuming that individuals buy a life annuity at retirement. Life-cycle investment strategies can mitigate that risk. By keeping a high exposure to equities during the first part of the accumulation phase and reducing it gradually as the retirement date approaches, life-cycle strategies use the compounded return effect to grow savings and reduce the risk of large losses when accumulated assets are at their peak. The analysis from the stochastic model shows that, while the equity portfolio outperforms life-cycle strategies in terms of replacement rates in more than 75% of the cases, the life-cycle strategies produce higher replacement rates than the equity portfolio with a probability between 40% and 50% when there is a fall in equity markets the year just before retirement. The attractiveness of the life-cycle strategies compared to the equity portfolio diminishes when the equity market fall happens further away from the year of retirement, as there is more time for a market rebound before retirement.
Conservative investment strategies provide only moderate protection to members of DC pension plans. A common approach to address the caveats associated with equity investment is to offer investment strategies that provide more certainty to individuals. The analysis shows that retirement income outcomes are less volatile with a fixed-income portfolio than with investment strategies with some equity exposure. However, most people are likely to be worse-off with portfolios only invested in fixed income and to forego significant pension benefits. Additionally, even when a fixed-income portfolio may be the best option, the difference with other investment strategies may only be moderate, as portfolios with equity exposure may only produce slightly lower accumulated assets and replacement rates in case of negative events.
Finally, there are important trade-offs to consider when investing in equities during the payout phase. The level of equity investments in the payout phase affects the comparison between regular drawdowns and lifelong annuities. The analysis from the stochastic model shows that, when equity exposure does not exceed 20% during the payout phase, people would be better-off taking a lifelong annuity than taking regular drawdowns in at least 75% of the cases. By contrast, there is a 60% probability of getting higher replacement rates when staying invested fully in equities during the payout phase and taking regular drawdowns instead of buying a lifelong annuity at retirement. Therefore, if individuals value flexibility and take regular drawdowns, large investments in equities would increase expected benefits. However, this comes at the cost of higher volatility of benefits and the risk of outliving one’s resources, which are risks that lifelong annuities address. Moreover, if individuals take regular drawdowns and withdraw too little during the payout phase or pass away early, they may leave a bequest to their heirs. When considering the sum of all pension payments (benefits while alive plus bequests), the probability of receiving more with regular drawdowns and equity investment than with a lifelong annuity increases to 85%.
4.5.2. Policy implications
Pension regulators should avoid setting frameworks that lead to default investment strategies that are too conservative as equity investment tends to bring better retirement income outcomes. People investing their retirement savings in the default option tend to stay in the default even though it may not match their level of risk tolerance (OECD, 2018[9]). While high exposures to government bonds and low equity investment in the default option may not penalise plan members in the short term in countries where government bonds provide high returns, lack of investment diversification may increase the concentration risk and economic development may reduce future expected returns from government bonds, thereby affecting accumulated assets and pension benefits in the long run.
Countries where DC schemes invest mostly in fixed income should assess the appropriateness of their investment regulations. Investment limits for equity investment may be binding for DC schemes in some countries. Countries should ensure that their investment regulations are not constraining equity investment in a way that could reduce risk-adjusted returns. Similarly, attention should be given to investment limits affecting foreign investment and alternative investments, such as infrastructure and real estate, as they play a role in diversification.
Pension regulators should allow providers to offer life-cycle investment strategies to alleviate the risk of large falls in the level of assets accumulated when people lack the time to benefit from a market recovery. The regulatory framework should allow for innovation in designing the glide path to adapt to the needs of individuals. For example, the reduction in equity exposure may start in the last 10 years before retirement when people are planning to take an annuity, but if they take regular drawdowns and remain invested during the payout phase, the reduction in equity exposure may be smoother and continue into the payout phase as the investment horizon is longer.
Finally, the ideal level and profile of equity exposure is country-specific, and relevant stakeholders in each country should consider following a precise methodology to determine what would be the most appropriate equity exposure for default investment strategies. Default options are important for people unwilling or unable to select their own investment strategy. In some countries, policy makers may want to define a single default investment strategy for the whole population, while in others, policy makers may allow pension providers to define their own default option within a harmonised framework. Selecting an appropriate default investment strategy requires pension providers and policy makers to solve a trade-off between maximising the level of retirement income for plan members and minimising the risk that some plan members may get a retirement income that is deemed too low (OECD, 2020[1]). To solve this trade-off and select a default investment strategy, countries could follow the OECD framework described in (OECD, 2020[1]), which involves five steps: i) pre-selecting the investment strategies to be assessed; ii) assessing these strategies using a stochastic model to reflect the uncertainty of possible outcomes; iii) calculating indicators reflecting their potential riskiness and performance; iv) defining thresholds for risk indicators that reflect the importance given to the downside risk relative to the upside potential; and v) selecting the investment strategy meeting the thresholds for the risk indicators and maximising the performance indicators. While using this framework, pension providers and policy makers should consider the role of DC schemes in the overall pension system, the population’s level of risk aversion and the characteristics of the target population for the default option, especially the length of the contribution period, the contribution level, and the payout options they tend to select.
References
[3] Anarkulova, A., S. Cederburg and M. O’Doherty (2023), “Beyond the Status Quo: A Critical Assessment of Lifecycle Investment Advice”, SSRN Electronic Journal, https://doi.org/10.2139/ssrn.4590406.
[10] Antolin, P. and S. Payet (2011), “Assessing the labour, financial and demographic risks to retirement income from defined-contribution pensions”, OECD Journal: Financial Market Trends, Vol. 2010/2.
[5] Berardi, A. and C. Tebaldi (2023), “Saving for retirement in Europe: the long-term risk-return tradeoff”, Journal of Pension Economics and Finance, Vol. 23/2, pp. 272-293, https://doi.org/10.1017/s1474747223000136.
[6] Dunn, R. and M. Berg (2019), “Lifecycle Design – To and Through Retirement”, 27th Colloquium on Pensions and Retirement, https://www.ricewarner.com/wp-content/uploads/2019/12/Rpt-Lifecycle-Design-To-and-Through-Retirement-CEPAR-December-2019-002.pdf.
[11] HMD (2023), Human Mortality Database. Max Planck Institute for Demographic Research (Germany), University of California, Berkeley (USA), and French Institute for Demographic Studies (France), http://www.mortality.org (accessed on 17 October 2023).
[4] Khemka, G., M. Steffensen and G. Warren (2021), “How sub-optimal are age-based life-cycle investment products?”, International Review of Financial Analysis, Vol. 73, p. 101619, https://doi.org/10.1016/j.irfa.2020.101619.
[8] OECD (2024), 2024 Annual Survey of Investment Regulation of Pension Providers, https://www.oecd.org/content/dam/oecd/en/topics/policy-sub-issues/asset-backed-pensions/2024-Survey-of-Investment-Regulation-of-Pension-Providers.pdf.
[1] OECD (2020), “Selecting default investment strategies”, in OECD Pensions Outlook 2020, OECD Publishing, Paris, https://doi.org/10.1787/1c7381db-en.
[12] OECD (2020), “Switching investments in defined contribution retirement savings arrangements”, in OECD Pensions Outlook 2020, OECD Publishing, Paris, https://doi.org/10.1787/ed47ae07-en.
[9] OECD (2018), “Improving retirement incomes considering behavioural biases and limited financial knowledge”, in OECD Pensions Outlook 2018, OECD Publishing, Paris, https://doi.org/10.1787/pens_outlook-2018-8-en.
[7] Šebo, J., D. Danková and I. Králik (2020), “In Search of the Optimal Saving Strategy for Pan-European Pension Products”, Financial Assets and Investing, Vol. 11/2, pp. 54-72, https://doi.org/10.5817/fai2020-2-4.
[2] Siegel, J. (2023), Stocks for the Long Run: The Definitive Guide to Financial Market Returns & Long-Term Investment Strategies, Sixth Edition, McGraw Hill.
This annex provides a brief description of the pension funds studied in Section 4.2.
Every pension fund administrator in Chile offers five different types of funds, from Fund A that can invest 40% to 80% in equities, to Fund E that can invest no more than 5% in equities. Each pension fund administrator must allocate a growing part of investment to variable income securities from Fund E to Fund A. Men older than 55 and women older than 50 can select any Fund except Fund A. Fund B (25% to 60% in equities) is the default option for men and women up to age 35, Fund C is the default option for men aged 36 to 55 and women aged 36 to 50, and Fund D is the default option for men aged 56 and older and women aged 51 and older.
Pension management companies in Czechia can offer a range of participating funds with different investment strategies and risk profiles, including the mandatory conservative participating fund. They also manage transformed funds, where participants do not have investment choice but have an annual non-negative return guarantee. In 2013, participating funds were introduced and transformed funds were closed to new entrants. At the end of 2022, 63% of members of pension funds were still in transformed funds.
Asset managers provide some of the occupational pension plans available in France. These are the former PERCO plans, now called mandatory or collective company PER plans. Every asset manager offers a choice to plan members between four categories of funds: cash, bonds, equities and diversified. The diversified funds mix the other three categories of funds. These three funds are also the building blocks of life-cycle investment strategies.
AustralianSuper is the largest superannuation fund in Australia. It is owned by the Australian Council of Trade Unions (ACTU) and the Australian Industry Group (Ai Group), an employer organisation. The entity offers six pre-defined diversified investment options with different risk profiles: Stable, Conservative Balanced, Balanced, Socially Aware, Indexed Diversified and High Growth. The Balanced option is the default.
QSuper is part of Australian Retirement Trust and is the superannuation fund for employees of the Queensland Government and their family. It offers a range of pre-defined diversified investment options (Moderate, Balanced, Socially Responsible and Aggressive), as well as a default option following a life-cycle investment strategy (Lifetime). This life-cycle investment strategy is composed of eight funds (Outlook; Aspire 1 and 2; Focus 1, 2 and 3; Sustain 1 and 2). Members’ assets are invested in one fund at a time based on their age and their account balance. Assets are transferred to more conservative funds as people reach age 40, 50 and 58 and when the account balance exceeds certain thresholds defined for each age bracket.
PFA is the largest commercial pension company in Denmark and among the largest in Europe. It is one of the providers of occupational pension plans negotiated by collective agreements. PFA offers four different life-cycle investment strategies with different risk profiles. The four profiles (A to D) vary in their allocation between the high-risk fund and the low-risk fund according to the number of years until retirement. Profile A is the one with the lowest allocation to the high-risk fund, while Profile D is the one with the highest allocation to the high-risk fund.
Alecta is one of the providers of occupational pension plans in Sweden and among the largest in Europe. It is the default provider for salaried employees in the private sector. All those born from 1979 onwards are in a DC pension plan. The default option in the DC plan, called Alecta Optimal Pension, is a life-cycle investment strategy based on three funds. Up to age 63, members’ assets are invested in a fund with a target asset allocation of 60% in equities, 30% in fixed income securities and 10% in alternative investments (e.g. real estate and infrastructure). Between 63 and 65, the assets are invested in a fund with a target asset allocation of 50% in equities, 40% in fixed income securities and 10% in alternative investments. From age 65 onwards, the assets are invested in a fund with a target asset allocation of 40% in equities, 50% in fixed income securities and 10% in alternative investments.
AP7 is the public provider in the Swedish premium pension system. It provides the default investment option. This default option is a life-cycle investment strategy based on two funds, an equity fund and a fixed income fund. The equity fund is invested in global listed and unlisted securities, 99% outside Sweden. The fixed income fund is invested in Swedish fixed income securities. For members in the default option, 100% of the assets are allocated to the equity fund up to the age of 55. From the age of 56, the allocation is annually and linearly rebalanced towards the fixed income fund until reaching an allocation of two-thirds in the fixed income fund and one-third in the equity fund at the age of 75.
Nest is the public provider in the UK’s automatic enrolment system. It is the provider of last resort for employers of all sizes. It offers several investment options as well as a range of one-year target date funds, called Retirement Funds, as a default option. The glide path of the Retirement Funds includes three phases during the accumulation phase. The first phase, called the Foundation phase, lasts 5 years and is for people with more than 40 years until retirement. It has moderate exposure to risky assets to promote confidence in savings and minimise the impact of investment shocks that might make younger members stop contributing. The second phase, called the Growth phase, lasts 30 years and aims at outperforming inflation by 3 percentage points a year after charges. The third phase, called the Consolidation phase, is for people within 10 years to retirement. It reduces gradually the allocation to risky assets.
The Thrift Savings Plan (TSP) is a retirement savings and investment plan for US federal government employees and uniformed services members. It offers five individual funds with fixed asset allocations that plan members can mix according to their risk profile. The Government securities investment fund (G Fund) is fully invested in short-term US Treasury securities. The Fixed income index investment fund (F Fund) is invested in government securities, securitised assets and credit instruments. The Common stock index investment fund (C Fund) is fully invested in US equities. The Small cap stock index investment fund (S Fund) is fully invested in small cap US equities. The International stock index investment fund (I Fund) is fully invested in global equities. TSP also offers 10 target date funds, called L Funds, providing a pre-determined diversified mix of the 5 individual funds that evolves as the retirement date approaches.
The baseline assumptions for the stochastic model are similar to those used in Section 4.3 with the analysis based on historical returns. The model assumes an individual joining a DC pension plan at age 25 and contributing 5% of wages each year in employment until retirement at age 65. A 1% annual assets-based fee is charged during the accumulation phase, as well as during the payout phase when savings remain invested.18 The amount of assets accumulated at retirement is transformed into a stream of income according to the selected payout option. The analysis calculates the replacement rate as the level of retirement income divided by the wage just before retirement. When the level of retirement income varies over time, the average replacement rate is calculated over the retirement period.
The analysis considers the same five illustrative investment strategies as in Section 4.3. The only difference is that the DC plan is assumed to invest only in international securities to avoid referring to a particular country. There are three portfolios with constant asset allocation invested only in fixed income, only in equities, and in a mix of equities (60%) and fixed income (40%). There are two life-cycle strategies reducing the equity exposure linearly from 80% to 20% either 10 years or 20 years before retirement. The relative share of government bonds (88%) and cash and deposits (12%) is assumed to remain constant in the fixed-income part of the different investment strategies.
The analysis considers two different payout options, a lifelong annuity and regular drawdowns. The lifelong annuity pays fixed nominal benefits during the entire remaining lifespan of the individual in retirement. The benefits are equal to the level of assets at retirement divided by the annuity factor. The annuity factor is calculated using survival rates between age 65 and 110, and the government bond yield at the time of retirement as the discount rate. The benefits paid under the regular drawdowns are determined by dividing the level of assets in the pension account at the end of each year by the remaining life expectancy. Given the formula, the individual is ensured to receive benefits during his/her entire remaining lifespan in retirement, although the level of benefits varies and is not guaranteed. In addition, there may be assets left in the pension account to leave as a bequest when the individual passes away. For the regular drawdowns, savings remain invested during the payout phase and the illustrative investment strategies keep the asset mix at the time of retirement constant during the payout phase.
The stochastic model derives uncertainty about financial, labour market and demographic risks by generating 10 000 Monte Carlo simulations. Each Monte Carlo simulation represents one possible realisation of the world for the asset returns, inflation rates, government bond yields, unemployment spells, real wage-growth profiles and remaining lifespan in retirement. This provides the distribution of replacement rates generated by the different investment strategies.
Financial risks, i.e. investment returns, inflation rates and yields on government bonds, affect different stages of the retirement savings journey. Indeed, contributions are taken from wages that grow in line with inflation. These contributions are invested in portfolios mixing three different asset classes that generate different rates of return. Finally, when taking an annuity at retirement, benefits are determined based on a discount rate, which is assumed to be the yield on government bonds at the time of retirement.
The model generates 10 000 simulations of investment returns, yields and inflation rates by drawing random numbers from different distributions. The model assumes that the returns on government bonds and on cash and deposits, the inflation rates and the yields on government bonds are drawn from a normal distribution with mean and standard deviation provided by historical values between 1900 and 2021.19 The equity returns are drawn assuming a geometric Brownian motion for the underlying equity price, with drift and volatility provided by historical equity returns between 1900 and 2021. Each simulation covers the accumulation phase and the payout phase. Annex Table 4.B.1 presents the moments of the distributions. The model does not intend to reflect financial markets perfectly, but rather to provide a framework to compare the retirement outcomes produced by different investment strategies.
Note: Treasury bills are used as a proxy for cash and deposits.
The model also assumes correlations between two sets of variables. The correlation coefficients ensure that the value of the different risk variables in each simulation are likely to materialise together and form a plausible realisation of the world. The model assumes that yields on government bonds, returns on government bond, returns on cash and deposits, and inflation are correlated. Using the same historical data as in Annex Table 4.B.1, Annex Table 4.B.2 presents the Pearson correlation coefficients between the four variables. In addition, the model assumes that the risk of unemployment is correlated with the performance of equity markets. The risk of suffering unemployment indeed tends to be lower when the economy is booming, and to increase when the economy slows down or enters into recession, generally with a lag. Moreover, improvements in the economy or higher economic growth may push up returns on equity investment. Therefore, when the economy is doing well, returns on equity investments rise and the risk of suffering spells of unemployment falls, always with a lag, reinforcing the positive feedback cycle regarding the accumulation of income for retirement. The opposite occurs when the economy tanks. To take these patterns into account, the model adds a shock to unemployment rates linked to the performance of equity markets, with a lag. A correlation coefficient of -0.24 is assumed (Antolin and Payet, 2011[10]).
In addition to financial risks, the model also considers the uncertainty surrounding labour market outcomes. Labour market outcomes, in particular employment and wages, determine contribution levels, and thus the level of assets accumulated at retirement. Contributions to DC pension plans may be discontinued during periods of unemployment and depend on individuals’ wages.
In the stochastic model, contributions are calculated assuming a fixed contribution rate of 5% of wages. Wages for all simulations start at USD 22 000 at age 25. This corresponds to the OECD average annual wage in 2021 divided by two to approximate the wage level at the start of the career.20 Wages then grow in line with stochastic inflation and real wage growth. No contributions are paid in years of unemployment. For each simulation, the model determines whether the individual would suffer any unemployment spell and if so, in which years. The model also builds three different real wage-growth profiles and allocates each simulation to one of the profiles.
Unemployment spells
The model proceeds in two steps to determine stochastically the years of unemployment in the simulations. The first step considers cohort-level unemployment, while the second one considers economy-wide unemployment.
In each cohort, only a certain proportion of individuals will suffer spells of unemployment during their career. On average, only around 40% of individuals in any given cohort suffer spells of unemployment (Antolin and Payet, 2011[10]). Therefore, the model generates only 40% of the simulations with at least one unemployment spell. For the other 60%, the individual will have a full career.
For the simulations where the individual suffers unemployment, the model determines in which years the spells will occur during the career. The economy-wide unemployment rate is the best proxy of the probability of suffering unemployment in any year. The model considers that the unemployment rate varies with age, with younger individuals experiencing higher rates of unemployment than other age groups. Moreover, the unemployment rate shows a large degree of persistence, in particular in European countries. This means that someone unemployed in a given year will have a higher probability of being unemployed the following year.
OECD data show that the unemployment rate in the OECD area declines with age up to approximately 40 years old and remains relatively constant around 5% thereafter (Annex Table 4.B.1). For each simulation where the individual suffers unemployment, the model calculates the unemployment rate by drawing a base rate from a normal distribution with mean 5.04% and standard deviation 0.73%. This corresponds to the observed mean and standard deviation over 2000-22 of the unemployment rates for the age group 40 to 64 in the OECD area. For those aged 25 to 40, the model adds another component that declines linearly to 0 and starts with a draw from a normal distribution with mean 3.71% and standard deviation 0.51%. This corresponds to the observed mean and standard deviation over 2000-22 of the differences in the unemployment rates between the age groups 25-29 and 40-64 in the OECD area. The model then determines whether there is an unemployment spell in each year by drawing from a binomial distribution with probability equal to the resulting yearly unemployment rates.21
In per cent
Source: OECD Labour Force Statistics by sex and age.
In addition, the model accounts for the persistence of unemployment. Once the model has determined the unemployment spells following the methodology above, additional spells may be added to reflect the fact that someone unemployed in a year has a higher risk of staying unemployed the next year. Someone unemployed in year N is assumed to have a 75% probability of being unemployed in year N+1 if the unemployment rate as determined before has increased between N and N+1. The probability drops to 50% if the unemployment rate as determined before has decreased between N and N+1 (Antolin and Payet, 2011[10]).
The results are consistent with expectations. Among all the simulations, 64% have no unemployment spells. For those with unemployment spells, the median number of spells is 4 and the average is 5.
Real wage-growth profiles
Labour market risk also originates from the uncertainty surrounding the trajectory of real wages during one’s career. Real-wage gains during a career vary across individuals according to their socio-economic situation (e.g. occupation, educational level and income). Labour market studies document that there are three main career paths for real wages. In general, real wages experience the largest gains during the early part of a person’s career, with lower gains, even sometimes negative gains, in the latter part. This pattern results in real-wage paths that for some people reach a plateau at the end of their career, while for others, real wages plateau earlier, around ages 45 to 55, and fall thereafter. A minority experience flat real wages throughout their working lives. Annex Table 4.B.2 presents the three profiles of real-wage growth used in the model. The model assumes that 42% of the individuals experience real wages that keep rising during the entire career (high income/education), 55% experience real wages that fall during the second half of the career (middle income/education), while 3% experience flat real wages (low income/education) (Antolin and Payet, 2011[10]).
Nominal wages grow with inflation and account for the impact of unemployment. The model assumes that individuals suffering spells of unemployment re-enter the labour market at the nominal wage level they had when last working (i.e. they do not get inflation or real wage growth). However, if inflation or real wage growth declined in between, the last wage is adjusted downward.
The demographic risk in this model stems from the uncertainty around the number of years the individual may live in retirement (idiosyncratic longevity risk). The model assumes that the individual lives with 100% certainty until age 66, so the retirement period lasts at least one year. At older ages, the possibility of passing away is derived by drawing a random number from a binomial distribution with probability equal to the mortality rate at each age. The mortality rate used for each age from 66 to 110 correspond to the average mortality rate in 2019 across 33 OECD countries (HMD, 2023[11]). The mortality rate is equal to 1 at age 110. The first occurrence of death is used to determine the remaining lifespan in retirement.
← 1. In this chapter, the term “DC pension plan” covers both occupational DC plans and personal plans. In these plans, members can usually select their investment strategy and bear the investment risk.
← 2. The 19 OECD countries covered by the analysis are Australia, Austria, Belgium, Canada, Denmark, Finland, France, Ireland, Italy, Japan, the Netherlands, Norway, New Zealand, Portugal, Spain, Sweden, Switzerland, the United Kingdom and the United States.
← 3. The performance adjusted for risk or “risk-reward ratio” corresponds to the ratio between the median internal rate of return and the internal rate of return at the fifth percentile.
← 5. Data for the United States come from the Federal Reserve and only cover DC private pension funds (i.e. pension funds for public-sector employees are not included). Data for Australia include a small proportion of defined benefit assets.
← 6. In most jurisdictions, the limits vary for different types of pension arrangements.
← 7. Frequent switching tends to result in worse net investment performance. Many jurisdictions impose implicit and explicit barriers to switching between investment strategies and pension providers (OECD, 2020[12]).
← 8. An analysis of the correlation between equity exposure and average performance using aggregate cross-country data from the OECD Global Pension Statistics database between 2011 and 2022 for 28 jurisdictions does not find a statistically significant correlation coefficient due to country disparities and data limitations, including the relatively short period for analysis (2011-22).
← 9. All the data except for France were collected from the websites of pension supervisors (Chile and Czechia) and pension entities. For France, the data source is Europerformance.
← 10. In Czechia, although people have access to participating funds with high equity allocations (over 90% for some funds), most people are in conservative funds so the average exposure of members in participating funds was only 7.7% over 2014-22.
← 11. In Austria, expressing returns for international securities (equities and government bonds) from USD into national currency leads to very large returns during the years of the First and Second World Wars due to extreme exchange rates. This affects the diversified 60/40 portfolio more than the other investment strategies as foreign investment makes up half of that portfolio.
← 12. The analysis considers the same five illustrative investment strategies as in the previous section. The only difference is that the DC plan is assumed to invest only in international securities to avoid referring to a particular country.
← 13. There is a 5% probability to get a replacement rate lower than the one at the fifth percentile.
← 14. Figure 4.16 shows the y axis in reverse order so that investment strategies producing a lower replacement rate at the fifth percentile are higher in the graph (higher risk).
← 15. Some annuity products continue to pay benefits to beneficiaries after the policy holder passes away (e.g. guaranteed period, survivor option). Such annuities pay lower benefits upfront to the policy holder compared to traditional annuities.
← 16. Only individuals passing away at the age of 110, the maximum age assumed in the model, have no assets left when they pass away.
← 17. See the Annual Survey of Investment Regulation of Pension Providers - OECD for equity investment limits in OECD and selected non-OECD jurisdictions.
← 18. The analysis assumes no fees are charged during the payout phase when the individual takes a lifelong annuity. The annuity provider gets a remuneration by investing the assets in a wide range of securities while using yields on government bonds to price the annuity.
← 19. Although past performance may not be indicative of future results, historical data is the only available information to project expected returns together with their variability.
← 20. In the United States, the median yearly earnings of 25 to 34 years old full-time workers with high-school degree are approximately half the average wage of the population.
← 21. The unemployment rate also accounts for a shock linked to equity returns, as explained before.