# Archived - OECD Social Policy Division

Archived information

Archived information is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please contact us to request a format other than those available.

An international perspective

by
Edward Whitehouse

Summary
Affiliations and acknowledgements
Canada's retirement-income provision: An international perspective
1. Pension systems in 12 OECD countries
2. Principles and objectives of pension systems
3. Coverage of the retirement-income system
5. Financial sustainability
6. Economic efficiency
8. Security in the face of risk and uncertainty
9. Conclusions
References

## Summary

This paper's aim is to compare Canada's pension system with those of eleven other OECD countries. It provides an assessment of retirement-income provision in Canada relative to a series of key objectives and principles that all well designed pension systems share.

The first section presents the three tiers under the standard OECD taxonomy – two mandatory, the redistributive components and the insurance components, and one voluntary – and the different types of provision schemes for each.

The paper shows that public programs in Canada have a very broad coverage, in part because the Old-Age Security (OAS) coverage is near universal. Regarding private plans, Canada has one of the highest coverage for occupational plans (RPPs) and close to average for personal plans (RSSPs). Generally, the private coverage depends mostly on age and earnings.

On adequacy for current retirees, Canada's retirement-income system scores very well on measures of income adequacy for today's retirees. Average incomes of older people are high relative to those of the population as a whole. And old-age poverty is relatively rare, compared both with other OECD countries and the position of the working-age population in Canada.

On adequacy of revenues for future retirees, the paper shows that the overall replacement rate for average Canadian earners is relatively high compared to other countries. Canada belongs to a group of country with a replacement rate that exceeds the OECD average for low earners. In Canada, it is possible for low earners to reach the OECD average replacement rate with mandatory provision alone. For average earners, like in most countries, voluntary pension savings are needed to reach the OECD replacement rate.

On financial sustainability, the paper shows that Canada is below the average of the 30 OECD countries for public expenditures on pension as a proportion of the GDP. Although this proportion is expected to grow significantly in the long term, an inward migration rate gives Canada a favourable outlook.

Regarding economic efficiency, the average retirement age in Canada is slightly lower than the average of the 30 OECD countries, though this does not result from a favourable treatment for early retirement in public pension schemes. Regarding administrative efficiency, while the public pension system appears to be administered at low cost, the main issue in Canada is the scale of administrative charges for personal pensions (RRSPs). Regarding security of income in the face of risk and uncertainty, Canada has a well diversified retirement-income provision, though it shows a large proportion of incomes in old age coming from private pensions.

## Affiliations and acknowledgements

Edward Whitehouse leads the pensions team in the Social Policy division of the Organisation for Economic Co-operation and Development (OECD). Other members of the team – Anna D'Addio, Rafal Chomik, Andrew Reilly and Seyda Wentworth – made a significant contribution to the analysis, as did Monika Queisser, head of the Social Policy division.

This report was produced at the request of the Department of Finance Canada as an input to the analysis of the Research Working Group on Retirement-Income Adequacy. The report takes an international perspective; it aims to complement and not to duplicate the work of the national group of experts.

Numerous officials of the Canadian government – from the Department of Finance, Human Resources and Skills Development Canada, Statistics Canada and the Office of the Chief Actuary in the Office of the Superintendent of Financial Institutions – provided helpful comments and information at a series of presentations of a preliminary analysis in Ottawa in September and November 2009.

Nevertheless, the views expressed are those of the author alone, and do not necessarily reflect those of the OECD, nor of any of its member governments, including that of Canada.

## Canada's retirement-income provision: An international perspective

Edward Whitehouse

This report's aim is to compare Canada's pension system with those of other comparable countries. It provides an assessment of retirement-income provision in Canada relative to a series of six key objectives and principles that all well designed pension systems share. Although it points to issues that should be of concern to policy-makers, it does not aim to provide detailed policy recommendations.

The analysis covers 12 countries that are members of the Organisation for Economic Co‑operation and Development (OECD). These comprise the group of seven largest industrialised economies – Canada, France, Germany, Italy, Japan, the United Kingdom and the United States — along with three further countries in Europe – Ireland, the Netherlands and Sweden – and the two Australasian countries – Australia and New Zealand. The pension systems represented here include many that have a similar structure to Canada's and some that were chosen to offer a contrast.

The report focuses on pension systems. Nevertheless, it is important to keep in throughout that pensions are only one route through which individuals provide for incomes in their old age. In many countries, including Canada, people contribute to different sorts of savings schemes that are not labelled as "pensions". People also save in non-financial assets – most notably housing and family businesses – which they intend to draw on in retirement. Again, these are significant avenues for retirement saving in Canada and many of the other countries under study but are not discussed in detail here.

The remainder of the report is structured as follows. The next section provides a brief introduction to retirement-income provision in the 12 countries under study. Section 2 then presents the objectives and principles for pension systems. In sections 3 to 8, national systems are assessed relative to these principles. Section 9 provides some brief conclusions on Canada's retirement income provision from an international perspective. It also points to the issues that are most in need of additional study before an informed decision over the direction of pension reform can be taken.

## 1. Pension systems in 12 OECD countries

The standard OECD taxonomy of pension systems, set out in the Pensions at a Glance reports,1 divides the pension system into three tiers. As illustrated in Figure 1, the first and second tiers are mandatory while third consists of voluntary retirement savings, mainly through private pensions.

The two mandatory tiers are distinguished by their objectives. Redistributive components – the first tier – of pension systems are designed to ensure that pensioners achieve some absolute, minimum standard of living. Insurance components – the second tier – are designed to achieve some target standard of living in retirement compared with that when working.

Because of their redistributive character, first-tier benefits are all publicly provided. However, second-tier schemes can be either publicly or privately provided. Within the first and second tiers, there is also a distinction by the way in which pension benefits are determined.

Figure 1. Different types of retirement-income provision

Using this taxonomy, Table 1 summarises the architecture of the 12 countries under study. Starting with the first tier, the taxonomy distinguishes three different types of provision.

• Resource-tested schemes, such as the guaranteed income supplement (GIS) in Canada, pay a higher benefit to poorer pensioners and lower or zero benefits to richer retirees.

All OECD countries have some sort of safety-net benefit payable in old age. However, most of these schemes play only a limited role in providing old age incomes because most retirees have sufficient entitlements from other parts of the pension system to render them ineligible for these safety nets.

Nevertheless, the OECD pension models do suggest that low-earning, full-career workers would be entitled to resource-tested benefits in three countries: Australia, Canada and the United Kingdom. This is reflected in the proportion of people receiving resource-tested support: 23% in the United Kingdom, 34% in Canada and 78% in Australia. In contrast, coverage of these benefits is just 1% of pensioners in Sweden, 2% in Germany and Japan, 5% in France and Italy and 7% in the United States.2 Since 28% of Irish pensioners receive resource-tested benefits, Ireland is also marked in Table 1.

• With basic-pension schemes, such as old-age security (OAS) in Canada, the benefit is either flat rate (the same amount is paid to every retiree) or it depends only on years of work, but not on past earnings. Additional retirement income from other sources does not change the value of basic pensions.3

Half of the countries in Table 1 have basic pensions. In Canada, the Netherlands and New Zealand, entitlement is effectively based just on residency in the country. In Ireland, Japan and the United Kingdom, entitlements depend on the number of years of contributions (although there are often credits for periods spent out of paid work).

First tier Second tier Universal coverage, redistributive Mandatory, savings Public Public Australia YES DC Canada YES YES DB France YES DB+points Germany Points Ireland YES YES Italy NDC Japan YES DB Netherlands YES DB New Zealand YES Sweden YES NDC DC United Kingdom YES YES YES DB United States DB

Note: DB = defined benefit; DC = defined contribution; NDC = notional accounts.
Source: Country profiles in Part III of OECD (2009).

• Thirdly, minimum pensions, which share many features with resource-tested plans, are found in three countries. In these schemes, the value of entitlements is determined by taking account only of pension income. However, unlike resource-tested schemes, they are not affected by income from savings or assets other than the relevant pension. Also, they are contributory, with eligibility typically determined by similar criteria as second-tier benefits. Minimum credits in the earnings-related scheme in the United Kingdom have a similar effect: benefits for workers with very low earnings are calculated as if the worker had earned at a higher level.

Programmes within the "second tier" play the role of "savings" in that they aim to provide retirees with an adequate income relative to their previous earnings, not just a poverty-preventing absolute standard of living. The schemes considered here are, like those in the first tier, mandatory whether public or private. Only Ireland and New Zealand of the 30 OECD countries do not have mandatory, second-tier provision. The first three are variants of earnings-related schemes. They are just based on a different set of parameters, and so it is possible to design a scheme of any type that produces the same results as another type.4

• Defined-benefit plans are provided by the public sector in five of the countries under study. This category includes the Canada and Québec Pensions Plans (CPP/QPP) in Canada. Private (occupational) schemes are "quasi-mandatory" in the Netherlands: they achieve near-universal coverage through industrial-relations agreements. In these schemes, retirement income depends on the number of years of contribution during the length of the working life and on some measure of individual earnings.
• There are points schemes in France (the occupational plans) and Germany. Workers earn pension points based on their individual earnings for each year of contributions. At retirement, the sum of pension points is multiplied by a pension-point value to convert them into a regular pension payment.
• There are notional-accounts schemes in Italy and Sweden. These plans record each worker's contributions in an individual account and apply a rate of return to the accounts. The accounts are "notional" in that both the incoming contributions and the interest charged to them exist only on the books of the managing institution. At retirement, the accumulated notional capital in each account is converted into a stream of pension payments using a formula based on life expectancy. Since this arrangement is designed to mimic the design of defined-contribution schemes, notional accounts are also often called notional defined-contribution plans (NDC).
• Defined-contribution (DC) plans are compulsory in eight OECD countries, including Australia and Sweden of the countries under study. In these schemes, contributions flow into an individual account. The accumulation of contributions and investment returns is usually converted into a pension-income stream at retirement. In addition to the mandatory defined-contribution plan in Sweden, there are quasi-mandatory occupational plans for private-sector workers (with near universal coverage) that are also now of the defined-contribution type.

In addition to these mandatory parts of the pension system, voluntary retirement provision is most significant in Canada, Ireland, the United Kingdom and the United States: as shown below, private pensions provide a substantial part of the retirement-income package. In addition, coverage of private pensions among today's workers is broad in Germany and Japan, in addition to these four countries. (Again, this is discussed below.. In all cases, provision is a mix of the defined-benefit and defined-contribution types, with most of these countries showing a strong trend to defined-contribution plans in recent years.

## 2. Principles and objectives of pension systems

The seemingly simple question – "Which country has the best pension system?" – is one that is dreaded by the author of this report and, probably, by most pension analysts. There is no answer to this question because pension systems have many different objectives and there are trade-offs between these objectives.

The analysis of pension systems in this report is built around a framework of six objectives and principles of retirement-income provision. These have been set out in numerous OECD reports.5

• Coverage of the pension system, by both mandatory and voluntary schemes;
• Financial sustainability and affordability of pensions to taxpayers and contributors;
• Economic efficiency: minimising the distortions of the retirement-income system on individuals' economic behaviour, such as labour supply and savings outside of pension plans;
• Administrative efficiency: keeping the cost of collecting contributions, paying benefits and (where necessary) managing investments as low as possible; and
• Security of benefits in the face of different risks and uncertainties.

This framework illustrates the trade-offs inherent in the design of pension systems and reforms. For example, higher pensions would improve the adequacy of retirement benefits but would also worsen financial sustainability. In other cases, there are synergies between different objectives. Encouraging later retirement improves both economic efficiency and financial stability. Similarly, extending coverage of pensions should also improve adequacy of retirement benefits for today's workers.

The sections that follow cover look at these six objectives in turn. They provide empirical indicators of different countries' pension provision, putting Canada's retirement-income system into an international context.

## 3. Coverage of the retirement-income system

Public programmes for retirement incomes in Canada are very broad in their coverage. For example, the basic scheme – OAS – is near universal because entitlement is based on residency. In contrast, the basic schemes in Ireland, Japan and the United Kingdom can have gaps in coverage, because entitlement is linked to the number of years of contributions. Means-tested benefits – GIS – are received by around a third of older Canadians. Together, these two programmes are effective in providing retirement incomes for groups with interrupted careers – due to caring responsibilities, long-term unemployment etc – or persistent low pay. Moreover, the earnings-related element of Canada's retirement income system -- CPP/QPP – allows for "drop-out" years to boost retirement incomes for people with less-than-full careers. Ongoing OECD work suggests that the different parts of Canada's public retirement-income system, working together, provide strong protection for interrupted work histories without unduly affecting incentives for people to work and save.6

It is also important to look at coverage of private pensions. As shown in the left-hand columns of Table 2, there are mandatory private pensions in Australia and Sweden. Also, occupational plans achieve near-universal coverage of workers in the Netherlands and Sweden through centralised industrial-relations agreements.

In the other nine countries under study – including Canada – private pensions are voluntary. For personal plans, it is clearly up to individuals whether they choose to contribute or not. For occupational plans, "voluntary" means that employers are not obliged to operate a scheme, either by law or as a result of centralised collective bargaining. Also, employees in many cases can choose whether to join the occupational plan or not (although in some countries, employers can and do make participation in the occupational plan a compulsory part of the job contract.

Unfortunately, accurate data on coverage are difficult to obtain and results vary between different national sources. For example, some sources report membership of a plan whereas others give information only on the number of active contributors. Also, there are some problems with comparability, with some sources looking only at employees, some at the economically active (including, for example, the self-employed) and others at the working age population as a whole.

Coverage of occupational plans (RPPs) in Canada is around a third of the workforce. This is somewhat lower than Ireland, Japan, the United Kingdom and the United States, where 40‑50% of workers have an occupational pension. And it is much lower than in Germany, where occupational plans reach nearly two-thirds of the workforce.7

Turning to personal plans, coverage in Canada (of RRSPs) is 36%: the highest for voluntary schemes after Germany. This is a similar degree of coverage as the United States, but well above Ireland and the United Kingdom, among others.

It is important to note that the distinction between occupational and personal plans is often difficult to make. As set out in the notes to Table 2, the mandatory contribution to private pensions in Australia is made to a mix of occupational plans (organised by employers) or to personal schemes (run by financial-services companies). In Canada and the United Kingdom, many employers have established group personal pension schemes for their employees. However, data constraints mean that these are reported as personal plans in Table 2.8

Obtaining data on the overall coverage of private pensions – both personal and occupational -- is further complicated by the fact that many people contribute to both types of plan. The right-hand column of Table 2 shows data on overall private-pension coverage where it is possible to correct for double-counting of such individuals. In Canada, for example, around 18% of people have both occupational and personal plans, with a broadly similar number having only one or the other.

Mandatory/quasi-mandatory Voluntary Total Occupational Personal Occupational Australia 85.0 18.8 9.7 Canada 34.0 36.0 52.6 France 15.0 Germany 64.0 44.0 Ireland 42.9 14.9 Italy 10.6 5.1 Japan 45.0 Netherlands >90.0 New Zealand 13.0/32.6 5.5/10.7 na/43.3 Sweden >90.0 >90.0 United Kingdom 47.1 18.9 59.1 United States 46.0 34.7 57.7

Note: Empty cells indicate that there is no legal basis for that scheme type in a particular country or that coverage is negligible (less than 1%). The entry ">90.0" indicates that coverage is near universal. The column for total coverage is only filled where there are adequate data to deal with double-counting of people with both occupational and personal plans.

Australia: the mandatory superannuation-guarantee scheme allows individuals to choose between an employer-wide scheme, industry-wide funds, a financial-services firm or to invest the funds themselves: a mix between occupational and personal provision. Germany: coverage of occupational pensions is a percentage of employees covered by the public pension. New Zealand: the second figure in each cell shows people covered by KiwiSaver (either through their employer – occupational – or a financial-services firm – personal). The first figure shows coverage of traditional occupational and personal pensions (excluding people contributing to personal pensions aged over 65 for tax reasons).

Source: OECD (2009), pp. 140-141, Statistics Canada.

It is also interesting to see how coverage of private pensions varies between different groups of individuals. The two most important factors in determining whether people have a private pension are age and earnings, as illustrated in Figure 2.

The pattern of private-pension coverage by both age and earnings is remarkably similar between Canada, Ireland, the United Kingdom and the United States. Coverage increases with age (left-hand panel of Figure 2) from 20-30% of people in their early 20s to 65-70% of people in the 45-54 age-bracket in these four countries. Apart from the United States, coverage declines for the oldest workers (aged 55-64). The most likely explanation for this decline is the early retirement of people with private pensions, particularly with occupational schemes.9

Coverage of private pensions increases strongly with earnings. In Canada, just 10% of people in the lowest two deciles of the earnings distribution have private pensions, compared with over 85% of people in the highest two deciles. Low earners are somewhat more likely to have occupational or personal pensions in the United Kingdom and the United States than in Canada.

The pattern of coverage of voluntary private pensions by both age and earnings is significantly different (from the other four countries in Figure 2) in Australia and Germany. In Australia, this most likely reflects the fact that mandatory private pensions are also in place. Few people want voluntary plans in addition.

Figure 2. Coverage of voluntary private pensions by age and earnings in six OECD countries

A. Coverage by age group B. Coverage by earnings decile

Source: OECD (2009), Figure 4.6; Statistics Canada

Germany has much higher coverage of younger and low-income workers than the other countries analysed in Figure 2: around twice as many 20-24 year olds and workers with earnings in the lowest tenth (decile) of the earnings distribution. The reasons for Germany's success in bringing younger and low-income workers into private pensions are many and complex. But they most likely include the specific design of fiscal incentives to join a private pension and the greater need for low earners to save for old age than in countries – such as Canada, Ireland, the United Kingdom and the United States – that have much more redistributive pension systems (see below).

## 4. Adequacy of retirement incomes

What constitutes an "adequate" income in old age is a contentious issue. The scope of the debate encompasses both the analysis of older peoples' incomes and fundamental questions in the design of retirement-income systems.

#### Adequacy of incomes of today's retirees

The analysis begins with a broad definition of adequacy: comparing average incomes of older people (aged over 65) with average incomes of the population as a whole. Incomes are measured from the data in national household surveys that were carried out in the mid‑2000s. They are calculated, following standard methods:

• Income comprises earnings, self-employment income, income from capital (dividends, interest etc.), rental income and public transfers (such as retirement, family, unemployment, housing and disability benefits).
• The measure used is "disposable" income, i.e., net of personal income taxes and social security contributions.
• Incomes are aggregated on a household basis and household members assumed to share resources.
• Total household income, divided among household member(s), is "equivalised" to adjust for differences in household size. For example, the OECD's adjustment effectively says that two people can live as cheaply as approximately 1.4 people living alone.

Figure 3 shows that incomes of older people – aged over 65 – were, on average in OECD countries, 82.4% of population incomes in the mid 2000s. Canada's figure of 90.8% is well above the OECD average, with only France and Germany of the 12 countries having higher relative incomes for older people.

Figure 3. Relative incomes of older people (aged over 65)
Equivalent household disposable income, mid-2000s

Source: OECD (2009), Figure 2.1; OECD (2008), Figure 2.4.

In contrast, Australia, Ireland, New Zealand and the United Kingdom have a much greater gap between incomes of older people and the population as a whole. Indeed, the over 65s have incomes of less than 70% of the population in the first three of these countries.

The second empirical results focus on older people with low incomes, presenting data on old-age poverty in OECD countries. This embodies a narrower definition of "adequacy": it compares older people's incomes with a poverty line rather than comparing their living standards, on average, with the population as a whole.

For the purposes of international comparison, the OECD treats poverty as a relative (rather than an absolute) concept. It is relative in two senses of the word. First poverty is measured against a yardstick dependent on median household incomes. Secondly, the poverty thresholds are country-specific, so poverty is measured against prevailing norms for living standards in a particular country at a particular time.

This means that a person classified poor in a prosperous OECD country will have a higher income than many of the non-poor in other countries that are less prosperous overall. The general approach of measuring poverty relative to a proportion of median income, adopted by the OECD for its cross-country analysis, is also widely used in national poverty statistics as well as by other international organisations, such as the European Union in its social-reporting system.

Figure 4. Old-age income poverty rates, mid-2000s
Percentage of over 65s with incomes of less than half
median equivalised population incomes

Source: OECD (2009), Figure 2.5; OECD (2008), Figure 5.3.

Recent analysis sets the threshold for poverty at 50% of median, equivalised household disposable income.10 People with incomes below this level are counted as "income poor".

In the mid-2000s, 13.3% of older people (aged over 65) were income poor on average in OECD countries (Figure 4). The old-age poverty rate was just 4.4% in Canada, the fifth lowest among the 30 OECD countries. Australia and Ireland are among the four OECD countries with the most widespread old-age poverty with the highest rates among the 12 under study. The poverty rate is also much above average in Japan and the United States.

One of the main drivers of the differences in old-age poverty rates shown in Figure 4 is the level at which old-age safety-net benefits are set. In Australia, for example, the full age pension in 2005 was A$12 700 a year, lower than the poverty threshold of A$ 14 770 for a single person.11 The difference between the two is larger in Ireland: € 8 870 for the basic pension and € 10 775 for the poverty threshold. In both countries, there are many people clustered around the income level of these programmes, which are 86% of the poverty threshold in Australia and 82% in Ireland.

However, Canada illustrates that there is not a deterministic relationship between the level of basic and resource-tested benefits and old-age poverty. For 2005, the safety-net income level in Canada was C$12 500 a year, just 83% of the OECD poverty line of C$ 15 000.

There are substantial differences in overall poverty rates between countries. The important issue for pension policy is whether older people are more likely to be income poor than the population as a whole. The 12 countries studied here divide into three groups, as illustrated in Figure 5:

• In Australia, Ireland, Japan and the United States, overall poverty rates are relatively high, but older people are still more likely to live in income poverty than the population as a whole.
• In France, Germany, Italy, Sweden and the United Kingdom, old-age and population poverty rates are similar: with the exception of Germany, old-age poverty is a slightly greater than the overall poverty rate, but not by much.
• In Canada, the Netherlands and New Zealand, old-age poverty rates are significantly below those for the population as a whole.

To summarise the data presented in Figures 3-6, Canada's retirement-income system scores very well on measures of income adequacy for today's retirees. Average incomes of older people are high relative to those of the population as a whole. And old-age poverty is relatively rare, compared both with other OECD countries and the position of the working-age population in Canada.

Figure 5. Old-age and population income poverty rates, mid-2000s
Percentage of people with incomes of less than half the median equivalised income

Source: OECD (2009), Figure 2.5; OECD (2008), Figure 5.3.

#### Adequacy of incomes of future retirees

The incomes of today's retirees are determined by the labour-market experience of these generations when they were working and, in particular, by the parameters, rules and structure of the pension system that were in place in the past. Most OECD countries have significantly reformed their retirement-income provision in the past two decades. In many cases, these changes mean that public pension entitlements for today's workers will be substantially lower than their parents or grandparents. For example, 10 OECD countries – a group that includes France, Germany, Italy, Japan and Sweden – will reduce benefits for average earners by an average of around 20%.12

Figure 6 therefore shows calculations of the pension entitlements for a worker entering the labour market in 2006, assuming that the parameters and rules of the pension system in that year remain unchanged. The calculations show the pension for a full-career, single worker using common assumptions for economic variables, such as price and wage inflation and investment returns on defined-contribution plans.13

Figure 6 shows the results for an average (mean) earner in the form of the pension replacement rate: that is, the income in retirement as a percentage of earnings when working. The standard OECD results show the replacement rate from all mandatory pension schemes, illustrated in the chart by the two darkest shaded bars. Countries have been ranked in the chart by the level of the mandatory replacement rate: from 88% in the Netherlands at the top to 31% in the United Kingdom at the bottom.

The replacement rate from the mandatory schemes in Canada (GIS, OAS and CPP/QPP, all of which are publicly provided) is 45%. This is rather less than the OECD average of 59%, but higher than seven of the 12 countries studied here, including Germany, Japan, the United Kingdom and the United States. Mandatory and quasi-mandatory private pensions in Australia and the Netherlands are expected to provide the majority of retirement benefits for average earners and they play a substantial role in Sweden.

Figure 6. Projected gross pension replacement rates for average earners
(by type of retirement-income scheme)

Source: OECD (2009), pp. 118-119.

Figure 7 deepens the analysis, by looking at replacement rates from the mandatory (public and private) parts of retirement-income systems for people with different levels of earnings. The 12 countries under study have been divided into four groups, depending on how closely benefits are tied to individual earnings when working.14 In each panel of Figure 7, the OECD average for the replacement rate is shown. As discussed above, this is 59% for average earners. But it is much higher for low earners – with pay of half the economy-wide average – at 72%. Replacement rates, on average, fall with earnings, to 54% for a worker with 150% of average earnings and 50% at double average earnings.

Starting in panel A at the top left are countries where there is a very strong pension-earnings link. In Italy, for example, the replacement rate is the same across the earnings range for people with the same number of years of contributions. In the Netherlands, the public, basic scheme means that replacement rates decline with earnings, but the overall impact of this scheme is dominated by the occupational plans.

Figure 7. Projected gross pension replacement rates from mandatory schemes by individual earnings

A. Italy, Netherlands, Sweden B. France, Germany
C. Australia, Japan, United States D. Canada, Ireland, New Zealand, United Kingdom

Source: OECD (2009), pp. 128-129.

The pattern in Sweden is complex. Replacement rates fall with earnings up to 85% of the economy-wide average due to the effect of the minimum (guarantee) pension. The replacement rate then increases for people above average earnings because at these earnings levels, the contribution rate to the defined-contribution occupational plan is 30%, compared with 4.5% below the threshold. Compared with the OECD average, Italy and Sweden pay very similar replacement rates to the OECD average for low earners, but much higher benefits to higher-income workers.

France and Germany, in panel B, have a slightly weaker link between pension and earnings. In Germany, this is because of the ceiling on pensionable earnings, which is significantly below the average for OECD countries. In France, the impact of the minimum pension on low earners and the ceiling on public pensions can be seen clearly. The pattern of the replacement rate by earnings is similar to the OECD average, but 5-10 percentage points lower. Germany's replacement rate is much lower, especially for low earners.

Moving to panel C, the link between individual earnings and pensions becomes weaker still. In the United States, this is because the public pension pays a replacement rate for full-career workers of 90% on the first slice of earnings, then 32% and finally 15%, up to the ceiling on pensionable earnings. Japan has a two-tier public scheme: the basic pension ensures that replacement rates are higher for lower earners. Australia's means-tested public scheme has a similar effect.

Canada belongs in the final group of countries, in panel D, which have strongly progressive mandatory retirement-income systems. For low earners, the replacement rate exceeds the OECD average, but then the gap between Canada and the OECD average grows larger as earnings increase. Indeed, the pattern in Canada is very similar to the two OECD countries – Ireland and New Zealand – that have just public, basic schemes and no mandatory second-tier provision.

#### Voluntary private pensions

In the countries with low replacement rates from the mandatory parts of the pension system, coverage of voluntary private pensions is widespread. This is to be expected: people will naturally wish to top up low public pensions with voluntary private provision, either through employers (occupational schemes) or individually (personal plans).

The lightest bars in Figure 6, therefore, show replacement rates for an average earner including voluntary private pensions. For consistency, these are also calculated under the assumption of a full career of contributions, although, as discussed below, this is much less likely to be achieved than in the case of mandatory (public or private) plans. In each case, a defined-contribution plan has been assumed. For Canada, Ireland and the United States, the contribution rate assumed in the calculations is based on data on average contribution rates in occupational plans: around 8‑10% (including both employee and employer contributions). For Germany, New Zealand and the United Kingdom, the calculations are based on parameters of the pension system. For Germany, a 4% contribution rate is the norm because this is the rate that attracts fiscal incentives. A 4% contribution rate is the default in New Zealand's KiwiSaver programme. In the United Kingdom, a contribution of 8% of earnings is around the average for people in occupational defined-contribution plans and will be the default contribution rate in the new automatic-enrolment, "personal accounts" that will be introduced in 2012. (More discussion of these new programmes will come later.)

Figure 6 shows that a full career of contributions at these assumed contribution rates would deliver a total replacement rate – including public and voluntary, private pensions – of around 75% in Canada, Ireland, the United Kingdom and the United States. This is significantly higher than the OECD average replacement rate from mandatory schemes of just under 60%. The smaller, 4% contribution rates, assumed for Germany and New Zealand are naturally expected to deliver rather lower overall replacement rates: around 60% and 55% respectively.

#### Pension gaps

A full-career of contributions is used as the baseline in the OECD's analysis of replacement rates in Figure 6. While this is a useful benchmark for mandatory schemes, it is clearly not for voluntary, private pensions. However, a "realistic" assumption is hard to come by. We know from the snapshot picture that around half of Canadians have a private pension at any one time. But this could result from half of the workforce being covered for their whole careers, or from the whole workforce being covered for half of their careers. The true picture will lie somewhere between the two. But the OECD has not been able to analyse long periods of panel data, following the same individuals over time, to generate a more realistic assumption.

The OECD has therefore turned the analysis around, and asked instead how much and for how long and how much would people need to contribute to reach a particular replacement rate. The most difficult part of this approach is the question of what the appropriate target replacement rate should be. The replacement-rate level will depend on many factors that affect resource needs during retirement relative to those when working. Important elements include the costs of raising children, the costs of work, housing provision and health and long-term care needs in old age. The last two, in particular, vary both between countries and between individuals in a particular country. For example, some countries have comprehensive public provision of long-term care, while in others, all individuals bear these costs or only the poorest individuals are effectively insured by the state. People who own their own homes and have paid of their mortgages in retirement will need a lower replacement rate than people who rent housing throughout their lives.

One approach is simply to ask people what replacement rate they would like. But such questions need to be phrased carefully: a higher target replacement rate for the pension system would appeal to most people. And this needs to be balanced against the fact that higher benefits can only be obtained at the cost of higher contributions or taxes.

Figure 8 shows the result of a survey in the United Kingdom. At the median income level (£200-249 at the time), around a third of 45-54 year olds said they wanted a retirement income the same as they had at the time, with 40% saying that they could get by with less. Apart from the very lowest income groups, the proportion of people saying that they would like around the same income in retirement is generally around one third. However, the proportion wanting a higher pension substantially declines with income and, equivalently, people with higher incomes at age 45-54 said they would need less in retirement.

Figure 8. Desired retirement income compared with current income, 45-54 year olds, 2002

Source: Pensions Commission (2004), Table 4.4 and Figure 4.10 based on Mayhew (2003).

However, the data reproduced underneath the chart show that desired median replacement rates were still very high: around 100% for people with the median income or a little below and well above 100% for the lowest two income categories. Desired replacement rates are around 70% for middle to high earners and nearly 60% for the highest income group. These are well above the replacement rates that recent retirees have achieved in the United Kingdom and well above the likely outcome for these prime-age workers.

The OECD's analysis has therefore taken the benchmark replacement rate for overall pensions (including voluntary private provision) as the average of its 30 member countries' replacement rates from mandatory schemes. As shown in Figure 7, this benchmark is a grossreplacement rate of just under 60% for average earners, just over 70% for low earners (with half average earnings) and 50% for people with pay of double the economy-wide average.

Figure 9 reprises the analysis of Figure 6, focusing on the replacement rate from mandatory parts of the retirement-income system and comparing national calculations with the OECD average. In three of the 12 countries under study, the replacement rate from mandatory schemes exceeds the OECD average. But in the other nine, including Canada, there is a "pension gap": a measure of the voluntary pension savings needed to finance retirement.

Figure 9. The pension gap for average earners
(Gross replacement rate for average earners from mandatory schemes and difference from OECD average)

Source: OECD (2009), Figure 4.1.

#### Filling the pension gap

How much will people have to contribute to voluntary, private pensions to lift overall replacement rates from the national, mandatory level to the average for OECD countries. For simplicity and comparability, the calculations assume that people with voluntary pensions have a defined-contribution plan, where the value of the benefit depends on contributions and investment returns.15 The modelling makes the same general assumptions as the calculations of replacement rates above: in particular it assumes an annual real return of 3.5% on pension savings, net of administrative charges.

Figure 10 shows the percentage of earnings that an average earner would need to pay into a private pension plan to plug the retirement-savings gap in the respective country. The darkest bars show the contribution rate needed with a full history, that is contributions in each year from age 20 to the normal pension age in the country.

The chart shows the impact of different national pension ages. The required contribution rate in Germany, the United Kingdom and the United States is lower because pension ages will increase to 67 or 68. Similarly, the low pension age in France reduces the number of contribution years and increase the retirement duration. There are also differences in life expectancy. In Japan (which has the longest life expectancy), 65 year olds are expected to live 22.2 years longer, compared with an OECD average of 19.7 years.

#### Effect of incomplete contribution histories

With a full contribution history, the proportion of earnings that would need to be paid into retirement savings plans to fill the pension gap is not generally large: around 6% in Ireland and the United Kingdom and around 7% in Japan. In many cases – Australia, Canada, Germany and the United States – the required contribution rate is 3.5-4.5%.

However, workers are unlikely to have full contribution histories. The lighter bars show two scenarios: one with 10 missing contribution years, the other with 20. It is assumed that these missing years occur at the start of the career: that is, people delay joining a private pension until they are 30 or 40 years old. For the countries shown, the average of the required contribution rate increases from 4.7% to 6.5% with ten missing years, and to 9.8% with 20 years missing.

#### Policy implications: voluntary private schemes

It is important to bear in mind that in some countries – particularly Australia, Canada, Ireland and New Zealand – it is possible for low earners to reach the OECD average replacement rate with mandatory provision alone. But even for average earners, the contribution rate needed to reach this benchmark replacement rate is not large, provided people contribute for most of their careers. However, the evidence in Figure 2 above implies that many people only begin contributing to voluntary private pensions around age 30.

Figure 10. Filling the pension gap: average earners with different contribution histories
(Contribution rate to voluntary pension private plan needed to achieve OECD average gross replacement rate)

Note: assumes a 3.5% real rate of return per year, net of administrative charges.
Source: OECD (2009), Figure 4.2.

#### Retirement-income adequacy over time: indexation policy

The final issue affecting retirement-income adequacy of future retirees is the policy for uprating pension benefits. Indexation policy can affect retirement incomes in two different ways: through the adjustment of pensions in payment and through changes in key parameters of the pension system.

In Canada, CPP/QPP benefits are primarily affected through the first mechanism: adjustment of pensions in payment. If inflation is 2.5%, then an unindexed benefit would be worth 21% less over the lifetime than a benefit indexed to prices. If real wages are growing by 2% a year, then an earnings-indexed pension would be worth 23% more than a price-indexed benefit (see Table below).16

 Indexation Unindexed Prices Earnings (1%) Earnings (2%) Lifetime value of benefits (% difference from prices) -20.9 0.0 10.6 23.0

The second mechanism – adjustment of parameters of the pension system – strongly affects entitlements to the basic and means-tested schemes (OAS/GIS). This is because indexation matters both before retirement and afterward: CPP/QPP benefits are effectively indexed to earnings before retirement and prices only during retirement.

Canada's current policy is to index the basic pension level and the target income for the means-tested benefit in line with prices. This means that these benefits will fall substantially behind general living standards in Canada if real earnings grow over time. During the 45-or-so years between a person starting work and beginning to draw a benefit, growth of real wages at 1% a year would mean that the benefits would fall by a third relative to average earnings. If real wage growth were 2% a year, the decline would be 60%. In effect, these benefits would "wither on the vine".

The OECD pension models implicitly assume as a baseline that these important value parameters are indexed to earnings. Otherwise, their value would be nugatory for example individuals entering the labour market today. In addition to Canada, this assumption affects the minimum (guarantee) pension in Sweden. The basic pension in the United Kingdom used to be indexed to prices, but once the new reform is in place, will be wage-indexed.

#### Indexation in practice

Figure 11 explores indexation practices over a long period, as much as 1960 to 2006 where data are available. In some cases, the data show the value of first-tier benefits. This is the basic pension in Ireland, the Netherlands, New Zealand and the United Kingdom and the means‑tested scheme in Australia. For Canada, the variable is the sum of the basic and means-tested components (OAS plus GIS).

In the other cases, the charts aim to capture the impact of changes in the adjustment of earnings-related pensions in payment. Increases in benefits over time are converted into an index with a value of 100 in 2006, the final year for which data were collected.

The first row of Figure 11 shows the adjustment of benefits relative to the index of consumer prices. A common pattern is a period of strong real increases in benefits followed by a period of stability. The period of real increases in pensions came to an end in the United States in the mid-1970s (when indexation of benefits to prices was formally adopted). The turning point was the late 1970s in France and Germany, the early 1980s in Italy and New Zealand and the mid- 1980s in Canada. Australia, Japan and the United Kingdom have seen continual real increase in pensions, but at a much slower rate. But, like the other countries, there is a pronounced shift to a slower rate of real growth since the 1970s.

Figure 11. Indexation policies in practice, 1960-2006

Real value of pension relative to prices, index 2006=100
Real value of pension relative to average earnings, index 2996=100

Source: Whitehouse (2009a).

The second row of Figure 11 shows the adjustment to pensions relative to an index of economy-wide average earnings. In a few countries – France, Germany, Sweden and the United Kingdom – pensions have fallen substantially behind wages as their growth in real terms slowed or came to a halt. In Australia and Japan, pensions fell behind wage growth in the 1980s and 1990s, but by a much smaller margin. In Canada and the United States, despite real (price-adjusted) benefits begin fairly constant from the mid-1980s and mid-1970s respectively, pensions have not fallen behind average earnings. This is because real earnings growth has tended to be slower than in Europe.

The implication of this analysis is that Canada can maintain adequacy of public retirement benefits over time if real earnings continue to grow at only a slow rate. However, it is important to reconsider increasing the real value of basic and means-tested benefits if real earnings are growing, to avoid pensioners' living standards lagging behind those of the rest of the population and a risk of old-age poverty returning.

## 5. Financial sustainability

Canada currently spends around4.5% of national income on pensioners (Figure 12). This is significantly below the average for the 30 OECD countries of 7.4%. Of the 12 countries under study here, only Australia and Ireland have lower public pension expenditure as a percentage of gross domestic product (GDP). In contrast, France spends 13% of GDP on pensions, and Italy, 14%.

Figure 12. Public expenditure on pensions, per cent of GDP
(Data for 2005-2007 and projections to 2060)

Source: OECD Social Expenditure database: see OECD (2009), pp. 138-139. European Commission (2009), Table 11; Office of the Chief Actuary, Office of the Superintendent of Financial Institutions Canada.

The arrows in Figure 12 show projections of the change in public pension expenditure over the next 50 years from a recent study by the European Union. The 27 EU member states currently spend nearly 9% of GDP on pensions on average, well above the average figure for the OECD countries. By 2060, this is expected to increase to nearly 13% of GDP.

The increase is, of course, driven primarily by population ageing. Indeed, the forecasts suggest that, holding everything else constant, demographic change would approximately double public pension spending to an average of 18%. The main reason that a smaller increase is expected is the series of pension reforms that have taken place over the past two decades, many of which will mean lower public pensions for future generations of retirees than are paid today and/or an increase in pension eligibility ages.17 Thus, Italy's public pension spending is expected to fall slightly as a share of national income by 2060 and the increase in spending in France is relatively modest.

National financial projections for Canada show an increase in pensions spending from around 4.5% of GDP now to 6.2% by 2060. This is a sizeable increase, but the growth in spending is at a slower rate than most of the other countries in Figure 2 for which comparable forecasts are available. One of the main drivers is the expectation of continued inward migration of around 0.5% of population per year, which gives Canada a more favourable demographic outlook than many European countries. The analysis suggests that Canada does not face major challenges of financial sustainability with its public pension schemes.

## 6. Economic efficiency

Retirement-income systems can affect individuals' economic behaviour in many ways. The most significant impacts of pension schemes are on work and savings decisions.

The average age at which Canadians leave the labour market was 63.3 years for men and 61.9 for women, both of which are slightly lower than the average for the 30 OECD countries (Figure 13).18

Effective ages of labour-market exit for men exceed the normal pension age of 65 in Ireland, Japan, New Zealand and Sweden. At the other end of the spectrum, both men and women leave the labour market on average before age 60 in France.

Figure 14 looks in more detail at labour-force participation rates by age and sex over time, comparing Canada with the average for all 30 OECD countries. Starting with men aged 55-59, the highest lines on the chart, participation rates in Canada fell from the late 1980s to the late 1990s: at their nadir, they were five percentage points below the OECD average. Since then, there has been a rebound in Canada, with participation rates for men in this age range now a little above the OECD average.

Participation rates for women at all ages have grown strongly over time. For women aged 55-59, the participation rate grew from less than 45% in the late 1970s to more than 55% in the latest year. But in Canada, the growth was much more rapid: from well under 40% in the late 1970s to nearly 64% recently.

Figure 13. Average effective age of labour-market exit, by sex, 2002-07

Men Women

Source: Chomik and Whitehouse (2009), which updates analysis in OECD (2006).

In the older range – age 60-64 – Canada's participation rates for both men and women have been consistently below the OECD average. The series for both Canada and the OECD for men shows a slow but steady decline until the early 1990s, with a period of stability followed by a rebound in the period since 2000. For women, the period of stability extends from 1975 again to around 2000, followed by a period of increase.

Low participation rates of older workers – in France and Italy in Figure 13, for example – are primarily a result of favourable treatment for early retirement in public pension schemes. This is not a problem in Canada. The basic and means-tested benefits (OAS and GIS) are not available until age 65. The reduction in benefits for early retirement in the earnings-related schemes is 6% a year. This is rather higher than the average for OECD countries that have early-retirement provisions in public schemes, of just over 5% a year. Moreover, there are plans to increase this adjustment to 7.2%. This is close to the actuarially neutral level – where early retirement is neither subsidised nor excessively penalised – according to OECD calculations.19

Nevertheless, pension eligibility ages may be an issue. In Australia, Ireland and the United Kingdom, public pensions cannot be drawn before the normal pension age, which is 65, except for women in the United Kingdom (which will increase from 60 to 65 over the next 10 years. Furthermore, Australia, Germany and the United States will increase normal pension age to 67 and the United Kingdom, to 68, in the coming years.20 While France, Italy and Japan allow public pensions to be drawn as early as 65, early pension ages are 61 in Sweden, 62 in Finland and the United States and 63 in Germany.

Figure 14. Labour-force participation rates by age and sex, Canada and OECD average, 1975-2007

Source: OECD labour-market database.

Canada's public pension system appears to be administered at low cost. The World Bank has collected information on administrative costs of a range of countries' public pensions; the database includes 11 OECD countries. Relative to national income, Canada spends just one quarter of the average for these countries (which include 5 of the 12 studied in this report). Costs are similar to the Netherlands, a little lower than the United States and substantially lower than Australia. Only New Zealand's public pension system is cheaper to run than Canada's. In part, this reflects the nature of the pension systems: New Zealand and the Netherlands run only residency-tested, basic pensions, which have the lightest load of information collection and processing. Australia's public pension is means-tested, which has a much higher informational requirement. Given the fairly large proportion of retirees on means-tested benefits in Canada, administrative expenses are perhaps lower than might be expected.

The main issue in Canada is the scale of administrative charges for personal pensions (RRSPs). Information provided to the author by Canadian officials suggests that many RRSPs have charges of 2% of assets per year, or even more. These higher-cost options tend to be actively managed, individual RRSPs. Nevertheless, there are lower cost options. For example, investing through indexed rather than actively managed funds involves typically only around half the costs and exchange-traded funds are cheaper still. Also, many people have group RRSPs where, due to economies of scale, costs also tend to be lower than individual plans. However, it is difficult to gauge from national sources the proportion of people covered by different types of plan and so the average and distribution of charges that individuals are paying for their RRSPs.

Moreover, 2% sounds rather a small number. Thus, both individuals and policy-maker. often overlook the impact such charges have on retirement incomes. The impact is more clearly captured by the "charge ratio": the proportion of the overall retirement accumulation paid in charges. (This is equivalent to the proportion of every contribution that goes in charges.

The effect is illustrated in Figure 15. The left-hand chart shows the charge levied as a percentage of assets on the horizontal axis and the charge ratio on the vertical. A levy of 1% of assets implies that 21.5% of the total retirement accumulation (or, equivalently, 21.5% of contributions) are paid in fees. With a levy of 2% of assets, the charge ratio is 37.3%.

(Percentage of contributions/retirement accumulations paid in charges)

by level of charge on assets by investment return

Note: baseline assumptions are a 45-year investment term, 2% annual growth in individual real earnings, constant rate of contributions relative to individual earnings. In addition, the left-hand-side chart assumes a 3.5% real return on investments in gross terms and, the right-hand-side chart, a charge of 1% of assets.
Source: OECD calculations. See Whitehouse (2000a,b and 2001) for details of the calculations.

These calculations are made at the OECD's baseline assumption of 3.5% investment return. The right-hand side chart in Figure 15 shows how the charge ratio varies with the investment return. Assuming a levy of 1% of assets, the charge ratio increases from 21.5% at the baseline assumption to 23.4% with a 5% return and 25.6% with a real return of 7% a year. Similarly, lower returns mean a lower charge ratio. There are also variations in the charge ratio with other variables. For example, with the investment term: charges are lower for people with a shorter contribution history. But they are larger for people with gaps in their contribution record after joining a personal pension scheme.

By way of international comparison, the fees in Canada are probably at a similar level to so-called retail funds in Australia and personal pensions in the United Kingdom. However, in both these countries there has been substantial downward pressure on fees. In Australia, industry funds have low charges (0.5-1% of assets) and these funds have expanded into the retail market, driving charges down. In the United Kingdom, stakeholder pensions, introduced in the late 1990s, had a ceiling on charges of 1% of assets. The target for the new personal accounts scheme is for charges of 0.5% of assets in the short term, declining to 0.3% in the long term assets in the scheme build up.

The impact of charges on retirement incomes cannot be stressed enough: moving from a levy of 2% of assets per year to 0.5% would increases net benefits by more than 40%.

## 8. Security in the face of risk and uncertainty

The financial and economic crisis has never been far from the headlines over the last couple of years. And the impact on the investments of private pension plans has had wide coverage. Although the issue of investment risk is at the forefront of people's minds, this is only one of the many different risks and uncertainties that affect pension systems and individual workers and retirees: political, economic, social and demographic, as well as financial.

The best approach for an individual faced with such uncertainty – and, by extension, for a government seeking to do the best thing for its citizens – is to use a mixture of ways of providing retirement incomes. The OECD has long advocated diversified retirement-income provision, arguing that "diversity has many virtues" (OECD, 1998). The report on Maintaining Prosperity in an Ageing Society went on to say that "each of the elements of the system has its own strengthens and weaknesses and a flexible balance among them not only diversifies risk but also offers a better balance of burden-sharing between generations".

Canada already has well diversified retirement-income provision. Figure 16, shows, for example, the sources of income of today's older people. These are divided into three categories: public transfers, work (earnings and self-employment) and capital. The last of these is mainly private pensions, but includes the income from non-pension financial assets. Canada, along with Australia, the Netherlands, the United Kingdom and the United States, shows a large proportion of incomes in old age coming from private pensions.

The OECD has carried out a detailed analysis of the impact of the crisis on pension systems and the measures that can be taken to mitigate the effects of this crisis and try and avoid the worst impacts of any future crisis.21 Of the many measures discussed, one that deserves mention here is the issue of "lifecycle investing": moving savings in defined-contribution plans, such as RRSPs, towards less risky assets as people near retirement. While such investment options are often available in Canada, the United States and other OECD countries, they are often rarely chosen by individuals, have high charges and can involve only modest moves in asset allocation over the lifecycle. However, encouraging people to invest in such funds, perhaps through default options, would avoid the problem recently encountered by many vulnerable people in the years just prior to their planned retirement.

Figure 16. Sources of incomes of older people
(percentage of household disposable income, mid-2000s)

Note: Income from work includes both earnings (employment income) and income from self-employment. Capital income includes private pensions as well as income from the investment returns on non-pension savings.
Source: OECD (2009), Figure 2.9. See also OECD (2008) for a detailed discussion of the OECD's income-distribution database, from which these data are drawn.

## 9. Conclusions

Viewed from an international perspective, Canada has a high-performing pension system.

• Old-age poverty is comparatively rare and retirees' incomes compare favourably with those of the population as a whole.
• The basic pension – old-age security, OAS – and the means-tested scheme – guaranteed-income supplement, GIS – ensure universal coverage and form a very effective safety-net for the old-age incomes of people who spend periods out of the labour market (caring for children, for example) or have low earnings. The provision for "drop-out" years in calculating CPP/QPP benefits is also effective and appropriate.
• Long-term projections show that public retirement-income provision is financially sustainable. Population ageing will naturally increase public pension spending, but the rate of growth is lower and the starting point better than many OECD countries. Moreover, the earnings-related public schemes (CPP/QPP) have built up substantial reserves to meet these future liabilities.
• Public pension schemes, unlike many OECD countries, do not encourage people to retire early. And the incentives to continue working after age 60 in the earnings-related schemes are being strengthened.
• Retirement-income provision in Canada is well diversified: between public and private provision and pay-as-you-go and pre-funding as financing mechanisms. Diversification offers security against the range of many different risks and uncertainties affecting pension systems and individuals' retirement incomes.

Nevertheless, this report has also set out a number of areas of concern about retirement-income provision for people of working age today.

• Coverage of private pensions, particularly among low-to-middle earners and, to a lesser extent, younger workers, is less-than-complete. While the lowest earners will be able to get by on public pensions, projected replacement rates for middle earners from public benefits are below the OECD average. The analysis here suggests that most workers with a full contribution history will fill this pension gap through voluntary retirement savings. Nevertheless, there are concerns that interrupted contribution histories will leave a retirement-savings gap.
• Contribution rates for people with personal plans (RRSPs) are often relatively small. For example, calculations carried out by Human Resources and Skills Development Canada show that balances in RRSPs for people late in their careers (and so nearing retirement) are significantly smaller than those of people with occupational plans (RPPs).
• Although the public pension scheme provides incentives to remain in work, labour-market participation rates for people in the years up to the normal retirement age of 65 are relatively low.
• Administrative charges for personal pensions (RRSPs) are high for people with individual plans, especially those invested through actively managed funds. Such charges can take a substantial proportion of people's retirement savings.

The aim of this report is not to set out a prescription for Canada's retirement-income system. Nevertheless, the diagnosis set out suggests a number of avenues for research and consideration by policy-makers.

• Encouraging coverage of and contributions to private pensions. New Zealand has adopted and the United Kingdom will soon introduce "automatic-enrolment" into private pensions. This could help expand coverage among groups – such as low earners and the young – who are less likely to have private pensions now. Other countries – such as Germany, Ireland, New Zealand and the United Kingdom – have changed the tax incentives for private pensions to simplify them and focus them on lower earners (or they plan to do so).
• Consider indexing the target minimum retirement income provided by the basic and means-tested schemes (OAS and GIS) to growth in earnings to ensure that future generations of retirees' living standards do not fall behind those of the population as a whole.
• Increasing pension age. There is no pressing financial or fiscal need to increase pension ages in the foreseeable future. However, the reduced benefit expenditures from a higher pension age could be used to augment the value of benefits, for example.
• Bearing down on administrative charges for personal pensions (RRSPs). Countries have adopted many different policies with some successes in reducing charges. These can range from standard-setting, ensuring comparability of charges of different providers to institutional changes (such as a centralised contribution clearing house to reduce costs) and statutory ceilings on levies.
• Improving the investment of people's defined-contribution pension savings, by moving automatically towards less risky assets as people near retirement. This would provide some protection for old-age incomes in the face of volatile stockmarkets and any future financial crisis.

## References

Antolín, P. and E.R. Whitehouse (2009), "Filling the Pension Gap: Coverage and Value of Voluntary Retirement Savings", Social, Employment and Migration Working Paper No. 69, OECD, Paris.

Chomik, R. and E.R. Whitehouse (2009), "Trends in Pension Eligibility Ages and Life Expectancy, 1950-2050", Social, Employment and Migration Working Paper, OECD, Paris, forthcoming.

D'Addio, A.C. (2009), "Pension Entitlements of Women with Children: the Role of Credits within Pension Systems in OECD Countries", Social, Employment and Migration Working Paper, OECD, Paris, forthcoming.

D'Addio, A.C., J. Seisdedos and E.R. Whitehouse (2009), "Investment Risk and Pensions: Measuring Uncertainty in Returns", Social, Employment and Migration Working Paper No. 70, OECD, Paris.

Disney, R.F. and E.R. Whitehouse (1999), "Pension Plans and Retirement Incentives", Pension Reform Primer series, Social Protection Discussion Paper no. 9924, World Bank, Washington, D.C.

European Commission (2009), The 2009 Ageing Report: Economic and Budgetary Projections for the EU 27 Member States (2008-2060), European Economy No. 2/2009, Brussels.

Holzmann, R. and R.P. Hinz (2005), Old-Age Income Support in the 21st Century: An International Perspective on Pension Systems and Reform, World Bank, Washington, D.C.

Martin, J.P. and E.R. Whitehouse (2008), "Reforming Retirement-Income Systems: Lessons from the Recent Experiences of OECD Countries", Social, Employment and Migration Working Paper No. 66, OECD, Paris.

Mayhew, V. (2003), Pensions 2002: Public Attitudes to Pensions and Planning for Retirement, Research Report No. 193, Department for Work and Pensions, United Kingdom.

OECD (1998), Maintaining Prosperity in an Ageing Society, Paris.

OECD (2001), Ageing and Income: Financial Resources and Retirement in Nine OECD Countries, Paris.

OECD (2005), Pensions at a Glance: Public Policies across OECD Countries, Paris.

OECD (2006), Live Longer, Work Longer, Paris.

OECD (2007), Pensions at a Glance: Public Policies across OECD Countries, Paris.

OECD (2008), Growing Unequal? Income Distribution and Poverty in OECD Countries, Paris.

OECD (2009), Pensions at a Glance: Retirement-Income Systems in OECD Countries, Paris.

Palacios, R.J. and E.R. Whitehouse (2006), "Civil-Service Pension Schemes around the World", Pension Reform Primer Series, Social Protection Discussion Paper No. 06/02, World Bank, Washington, D.C.

Pearson, M.A. and E.R. Whitehouse (2009), "Social Pensions in High-Income Countries", in R. Holzmann, D. Robalino and N. Takayama (eds), Closing the Coverage Gap: The Role of Social Pensions, World Bank, Washington, D.C.

Pensions Commission, United Kingdom (2004), Pensions: Challenges and Choices, The First Report of the Pensions Commission, The Stationery Office, London.

Queisser, M. and E.R. Whitehouse (2006), "Neutral or Fair. Actuarial Concepts and Pension-System Design", OECD Social, Employment and Migration Working Paper No. 40, OECD, Paris.

Queisser, M., P. Whiteford and E.R. Whitehouse (2007), "The Public-Private Pension Mix in OECD Countries", Industrial Relations Journal, Vol. 38, No. 6, pp. 542–568.

Whitehouse, E.R. (2000a), "Administrative Charges for Funded Pensions: Measurement Concepts, International Comparison and Assessment", Journal of Applied Social Science Studies, Vol. 120, No. 3, pp. 311-361.

Whitehouse (2000b), "Paying for Pensions: An International Comparison of Administrative Charges in Funded Retirement-Income Systems", Occasional Paper No. 13, Financial Services Authority, London.

Whitehouse, E.R. (2001), "Administrative Charges for Funded Pensions: Comparison and Assessment of 13 Countries", in Private Pension Systems: Administrative Costs and Reforms, Private Pensions Series, Vol. 3, OECD, Paris.

Whitehouse, E.R. (2006), "New Indicators of 30 OECD Countries' Pension Systems", Journal of Pension Economics and Finance, Vol. 5, No. 3, pp. 275-298.

Whitehouse, E.R. (2007), "Life-Expectancy Risk and Pensions: Who Bears the Burden?" Social, Employment and Migration Working Paper No. 60, OECD, Paris.

Whitehouse, E.R. (2009a), "Pensions, Purchasing-Power Risk, Inflation and Indexation", Social, Employment and Migration Working Paper No. 77, OECD, Paris.

Whitehouse, E.R. (2009b), "Pensions during the Crisis: Impact on Retirement-Income Systems and Policy Responses", Geneva Papers on Risk and Insurance, Vol. 34, pp. 536–547.

Whitehouse, E.R., A.C. D'Addio, R. Chomik and A. Reilly (2009), "Two Decades of Pension Reform: What has been Achieved and What Remains to be Done?" Geneva Papers on Risk and Insurance, Vol. 34, pp. 515-535.

1 (OECD, 2005, 2007, 2009).

2 Source: OECD (2009), Figure 2.7. See also the extensive discussion in Pearson and Whitehouse (2009).

3 Formally, the basic pension (OAS) in Canada is resource-tested in that it is subject to a claw-back once income in old-age reaches a certain level. However, the threshold is set at a very high level and the claw-back rate is only 15%. Effectively, this is a basic pension as defined in the main text.

4 See Queisser and Whitehouse (2006).

5 OECD (1998, 2001, 2009), for example. See also Whitehouse et al. (2009). The World Bank has also set out a similar list of aims: see Holzmann and Hinz (2005).

6 This work involves modelling pension entitlements of people with interrupted careers. D'Addio (2009) looks at the position of women who spend time out of paid work caring for children. This paper forms part of a major OECD project on "Women and pensions", which is supported by Human Resources and Skills Development Canada. Another forthcoming OECD paper will set out the results of calculations of retirement incomes for people with long periods of unemployment.

7 However, these schemes are designed to provide much lower benefits than most occupational plans in Canada, Ireland, the United Kingdom and the United States: see Figure 6 and the surrounding discussion, and section 8 below.

8 National estimates of the proportion of RRSPs members in group schemes provided by employers vary between 20% and 40% of the total.

9 This can be explained by an "income effect": workers with large accumulations in private pensions wish to retire. It can also result from the rules of defined-benefit occupational plans, many of which have incentives to retire early. See Disney and Whitehouse (1999) for a detailed discussion.

10 OECD (2008, 2009).

11 Benefit levels for 2005 are taken from the database underlying the calculations in Whitehouse (2009). The "Country profiles" in Part III of OECD (2009) give 2006 values. The level of poverty thresholds for 2005 is taken from OECD (2008), Table 5.A1.1.

12 See Whitehouse et al. (2009), Martin and Whitehouse (2008) and OECD (2007, 2009) for a detailed description and analysis of these reforms and their impact on future retirement incomes.

13 See OECD (2005) for a detailed discussion of the methodology and OECD (2009) for the details of the calculations presented here.

14 The grouping is based on a measure of the progressivity of pension benefit formulae, the results of which are presented in OECD (2009), pp. 126-127. See Whitehouse (2006) for a detailed definition and discussion.

15 A detailed, step-by-step illustration of the calculations is set out in OECD (2007), pp. 83 84.

16 The calculations, carried out with the OECD pension models, are based on OECD average unisex mortality rates for a pension paid from age 65.

17 See the discussion in Whitehouse et al. (2009) and Table 12 of European Commission (2009).

18 Note that these average rates of labour-market exit were calculated over the period 2002-07 to mitigate the impact of cyclical effects on the results. The ages are calculated using a standard methodology from OECD data on labour-market participation (see OECD, 2006 for details).

19 Queisser and Whitehouse (2006). See Figure 2 and Table 5 for detailed parameters for OECD countries and calculation.

20 See Chomik and Whitehouse (2009).

21 See OECD (2009), Part I.1 and Whitehouse (2009b).