33386 World Bank Pension Reform Primer Retirement Can pension reform reverse the trend to earlier retirement? Retirement is a fairly recent invention in richer At the other end of the spectrum lie 54 countries countries and, in many poorer countries, it where retirement is rare. They are mainly in Af- remains a rarity. To illustrate global patterns of rica, with some in East Asia and Latin America. retirement, we have divided the world into four Activity rates are 90 per cent or more for under groups (Figure 1). The chart shows the proportion 65s and even average 70 per cent for over 65s. of older men that are `economically active': either in paid work or unemployed and looking for work. We have also separated out two intermediate pat- terns. The group of 30 countries marked 2 The 12 countries with the earliest retirement are all includes most of the OECD and former socialist in Europe. They include France, Italy and the countries that are not in the early retirement cate- Netherlands in the west and Hungary, Romania gory. Activity rates are 20 points higher in both and Slovenia in the east. Here, only a little over the 55-59 and 60-64 age ranges than in the earliest half of 55-59 year old men are still active and just retirement countries, group 1. Again, however, one in five men in their early 60s. Virtually no one very few people are active beyond age 65. over 65 is in work. In group 3--made up of 30 mainly low-to-middle Four patterns of retirement 1 income economies--many people retire, unlike the poorest countries in group 4. But over 45 per cent 100 4 very little retirement of people over 65 are still active. 90 (54 countries) 80 Trends in retirement 70 As recently as the 1960s, men in the OECD 3 60 countries worked, on average, for 50 years of their 68 year life-span (Figure 2). Now, they can expect 50 2 to spend just 38 years in paid employment, ac- 40 1 early retirement (12 countries) cording to the OECD. The number of years not 30 working have increased from 18 to 35. 20 activity 10 rates (%) This is partly due to growing participation in fur- ther education and higher unemployment. But 0 source: ILO 50-54 55-59 60-64 65- earlier retirement is the main cause. The OECD age expects this trend will continue, so soon men will spend more years out of than in work. This briefing is part of the World Bank's Pension Reform Primer: a comprehensive, up-to-date resource for people designing and implementing pension reforms around the world. For more information, please contact Social Protection, Human Development Network, World Bank, 1818 H Street NW, Washington, D.C. 20433; telephone +1 202 458 5267; fax +1 202 614 0471; e-mail socialprotection@worldbank.org. All Pension Reform Primer material is available on the internet at www.worldbank.org/pensions 2 Retirement Looking at all countries, 60 have seen significant ciaries and hopeful younger future recipients. The declines in older men's activity rates in the last 15 majority of people can now look forward to many years, according to the ILO. In contrast, there has years of leisure to be enjoyed in good health. But been little change in the age structure of participa- these leisure years have a cost. tion since the 1970s in 30, mainly middle-income nations. In most poorer countries, there has been The price of earlier retirement little decline in older workers' activity in the last The most obvious burden is on the pension sys- 30-40 years. tem, already faced with the cost of an aging population. Each one-year reduction in a coun- Men spend fewer years in work 2 try's average retirement age increases the number of pensioners by more than 5 per cent. And ear- 50 number of years in employment lier retirement also means fewer workers: 2 per cent or more for every year earlier people retire. 40 These twin effects raise the contribution rate to pay for pensions by at least 8½ per cent. And 30 even if early retirees cannot claim a pension, they number of years often rely on other benefits--for unemployment 20 not in employment or disability--putting a similar burden on the pub- lic finances. 10 Aging populations and early retirement have OECD countries 0 broader economic consequences than just for pen- 1960 1970 1980 1990 2000 2010 2020 2030 sion financing. The costs of supporting a larger number of retirees could cut growth rates in As countries become richer, their citizens tend to Europe and the United States by half a percentage retire earlier: compare groups 1 and 4 in Figure 1. point a year, again according to the OECD. This Early retirement in these countries can be seen as a cumulates to a fall in living standards of over 10 great welfare success story, especially to its benefi- per cent after 25 years. Recent policy initiatives to promote employment of older workers Earlier retirement and poor work incentives q Equalizing pension ages. Eight further in pension systems are widespread policy OECD countries will increase pension concerns. Four out of five OECD countries ages for women to equalize them with have recently reformed their pension sys- men's pension ages, usually at 65. tems with the goal, at least in part, of q Discouraging early retirement. This is promoting employment of older workers. the most popular policy, with 12 OECD There are four key kinds of change. countries taking action. Six will raise the q Increasing pension ages. Nine OECD number of years' contributions needed to countries will raise the age at which pub- qualify for early retirement while five will lic pensions can be claimed. Finland, raise the minimum early retirement age. Italy and the United States plan to raise q Encouraging work after pension age. the age by two years, New Zealand by Five countries aim to promote later than three years and Japan, Korea and Spain normal retirement, usually by introducing by five years. When these changes are or increasing increments for deferring implemented, more than two out of three taking the state pension. OECD countries will have a standard Outside the OECD, a dozen countries, pension age of 65 and in five it will be mainly in eastern Europe and Latin America higher, usually 67. have increased pension ages since 1992. Retirement 3 Some governments, particularly in Europe, explic- We begin with a simple, mandatory retirement itly encouraged earlier retirement to try and savings plan. People save ten per cent of their increase job opportunities for younger workers. earnings each year from age 20. These contribu- But high youth unemployment rates have per- tions then earn a return, net of administrative sisted. Most governments now recognize that it is charges, of five per cent a year. When the individ- a fallacy to think that economies simply have a ual chooses to retire, he or she converts the fund fixed number of jobs that can be divided among to an annuity. By providing an income stream un- their citizens: the so-called `lump-of-labor' fallacy. til pensioners die, this annuity prevents them outliving their resources if they live longer than More than half of OECD governments report low they expect. effective retirement ages and poor work incentives in the pension system among their key social pol- This kind of retirement savings plan is often icy concerns. And four out of five of these richer called a `defined contribution' pension, because the countries have introduced reforms to promote the benefit is determined by the value of contributions employment of older workers (see box). and the returns they earn. Why are people retiring earlier? To explore work incentives, we look at how the Today's older workers benefited from better edu- pension builds up in the savings account. Each cation, enjoy better health and can expect to live extra year's work adds to the pension in three longer than their parents did. And work is now ways. First, an extra year's contributions are paid less physically demanding. Why, then, are so many into the fund. Secondly, the accumulated balance leaving the workforce? in the fund continues to earn investment returns. These two effects work in every year from when One reason is high and persistent unemployment, the plan is started to retirement. The third effect is especially in Europe. Few older workers who lose important when we think about the choice be- their job find new work. And, some governments tween work and retirement. Each year's delay in encouraged earlier retirement in the mistaken be- claiming the pension increases the value of the an- lief that this would reduce unemployment (see nuity that the fund will buy, because the likely above). period of retirement is shorter. Early retirement is a positive choice for many peo- Earnings and pension by age 3 ple, rather than forced on them by ill health or redundancy. Current generations are, on average, richer than their parents, and it is unsurprising that they want to spend some of this wealth on addi- tional leisure time. But a major reason why early retirement has be- earnings come more attractive is the growth in public and, in some countries, private pension benefits. And many of these pension systems are biased against continued paid employment at older ages. defined contribution Valuing pension benefits pension This briefing aims to show how and why pension systems tend to favor early retirement using a basic 50 55 60 65 70 model of pension benefits in different types of age system. 4 Retirement Figure 3 shows calculations of the pension value at Each year of contributions obviously increases the different ages, focusing on the final years around pension by adding to the number of years in the likely retirement ages. Earnings are assumed to benefit formula. A second effect comes through rise with age, by three per cent a year. The pen- any change in average lifetime earnings as an extra sion value increases much more rapidly than pay-- year's pay enters the average. around 12 per cent a year--as extra contributions and investment returns accumulate and the delay Figure 4 shows the pattern of pension and earn- in drawing the pension means a shorter retirement ings by age in this benchmark defined benefit plan. to finance. The profile of the defined benefit plan is very dif- Public pension plans ferent from the relation between defined Mandatory retirement savings plans of the sort contribution pensions and age shown in Figure 3. modeled in Figure 3 are an increasingly popular Now, the pension rises much more slowly with form of pension provision. But public pension age, from a higher base: by around 5 per cent a schemes--still the dominant source of retirement year, compared with 12 per cent a year in the de- income in most countries--provide what is termed fined contribution scheme. a `defined benefit'. The pension value is related either to some measure of individual earnings (or Retirement incentives is sometimes flat-rate). Many private plans are also What are the implications for retirement behavior? defined benefit: examples can be found among The financial incentive to work can be measured employer-run schemes in the Netherlands, the by comparing income in work (earnings minus United Kingdom and the United States. contributions) and income out of work (pension). The higher their pension, relative to earnings, the We have modeled a simple, generic plan, which more likely people are to retire. Of course, there pays 1¾ per cent of average lifetime earnings for are other powerful influences on the retirement each year's contribution to the scheme. Public decision such as health, partner's labor market pension plans vary enormously in their generosity, status and the value of non-pension assets. Peo- but this rate of pension accrual is not atypical. ple's preferences for leisure differ and they may simply derive pleasure from their work. Earnings and pension by age 4 A simple measure of the financial returns to working compares in and out of work incomes directly. This is the `replacement rate': the ratio of the pension to earnings. The basic interpretation of this measure is that the higher the replacement rate, the less is the financial reward to working. earnings Comparisons of the pension value and earnings, however, do not capture the dynamic effects of the pension system. Each extra year of work not only defined brings in earnings but changes the pension value. benefit So the total financial return to working must also pension include the present value of the change in pension wealth to earnings less contributions. 50 55 60 65 70 age We have already described how the value of the two different sorts of pension is affected by an extra year's work. In defined benefit schemes, we need to add the loss of a year's benefits for an ex- Retirement 5 tra year's work. (This is captured in the annuity on (lower) pension income will be lower than on calculation for the defined contribution scheme.) (higher) earnings. Figure 5 shows replacement rates--the value of Some countries also give a more favorable tax the pension divided by earnings less contributions treatment either to pensions or to pensioners. plus the change in pension wealth--which adjusts These twin effects mean that net replacement rates for the dynamic impact of the pension system. can be 10-15 percentage points higher than gross, although the outcome varies with income. This picture highlights the inherent differences in the labor market incentives in a defined benefit Pensions and retirement scheme compared with a simple retirement savings Figure 5 makes no assumptions about retirement plan. At age 50, the gross replacement rate is over ages in the two different regimes. But the chart 35 per cent in the defined benefit scheme, com- clearly shows that people are likely to retire at the pared with just 12 per cent in the defined earliest possible opportunity in the defined benefit contribution plan. By age 60, the retirement sav- scheme. In contrast, the defined contribution ings plan would accumulate enough to buy an pension provides a powerful, continuing incentive annuity worth around a quarter of earnings. The to work until advanced ages. defined benefit plan would then offer a replace- ment rate of over 45 per cent. Only at age 68 or `Real-world' pension systems more does the defined contribution pension ex- ceed the defined benefit. Analysis of pension systems must always be careful to distinguish between the inherent characteristics Pension replacement rates 5 of different kinds of system and features which arise from imperfect implementation of these theoretical benchmarks. The simple defined bene- 80 replacement fit plan we explored above indeed abstracts from 70 rate (%) many of the complexities of real-world schemes. 60 And many of these features have important incen- tive effects. 50 defined benefit 40 Final and best earnings systems 30 First, we assumed that the defined benefit pension formula is based on average real earnings. But, in 20 practice, most systems use a subset of lifetime 10 defined contribution earnings. For example, two-thirds of developing countries and 40 per cent of OECD countries use 0 a measure of final earnings, ranging from the last 50 55 60 65 70 month's pay to the last ten years'. And best years' age earnings periods are used in around a fifth of countries. Only a third of OECD countries' pen- sion systems and less than 15 per cent of The impact of taxes developing countries' are based on average earn- These replacement rates are gross, so they ignore ings. But there has been a clear trend in countries the effect of taxes other than the pension contri- with best and final earnings formulae to lengthen bution. Personal income tax systems are the number of years that count. And some coun- `progressive', which means that the proportion of tries have moved to an average earnings base. income paid in tax increases as income increases. Net replacement rates, taking account of income The effect on incentives of formulae based on fi- tax, are higher than gross, because the tax levied nal or best years (compared with the benchmark, average earnings scheme) is complicated. It de- 6 Retirement pends on the pattern of earnings with age, and rate is substantially higher than the adjustments there is no consensus in empirical studies. Final- applied in nearly all lower-income countries and pay schemes discourage people from working once more than half of the OECD countries. real earnings have passed their peak. Partial re- tirement--reducing working hours or taking a Secondly, the adjustments do not seem to have different, part-time job--is inhibited. much of an effect on retirement behavior. Most people retire early if given the opportunity, even Maximum pensions when the pension is reduced. And few people The effect of other features of public, defined seem take advantage of pension increases for later benefit schemes is less ambiguous. Thirty-eight retirement. About 2 per cent of people of pension countries pay no more pension after 30 or fewer age in the United Kingdom and 8 per cent of 65- years' contributions and in another 19 countries 69 year olds in the United States defer their the limit is between 30 and 35 years. This provides pension. a strong incentive to retire once the ceiling has been reached. And 50 countries vary the pension A final issue is the fiscal consequences of these accrual rate so that the first years' contributions adjustments. If they are actuarially fair, then peo- (usually between 10 and 20 years) earn more pen- ple retiring later will receive the same net present sion than subsequent contributions. So the value of pension benefits. So there is no saving in increment to pension for older workers with a full pension costs from delayed retirement. There may contribution record is relatively small. Again, this be some fiscal gain from any extra taxes paid on encourages earlier retirement than a scheme that earnings and on the higher pension in payment. pays the same accrual rate for all contributions. But other policies to remove the incentive to retire early bring this gain plus lower pensions costs, Adjusting early and late pensions which will typically be more valuable. Some public, defined benefit schemes adjust pen- sion benefits with the age at which the pension is Combining work and pensions drawn. Nearly half of the OECD countries and 18 Some countries require people to give up work lower-income countries have these adjustments. completely to be able to draw a public pension, although a pension claim is compatible with a Most common are increments awarded when peo- limited amount of paid work in the majority of ple retire later than the normal pensionable age countries. Usually, some pension is withdrawn and defer claiming their pension. Many countries once a relatively low earnings threshold is reached. reduce pensions drawn earlier than the standard These earnings tests can impose very high effective age. The size of the adjustments varies signifi- tax rates on working, ensuring that very few peo- cantly: between 1 and 12 per cent for each year the ple work beyond a certain age. pension is drawn early or late. In OECD countries they average 6½ per cent a year compared with Some countries have either eliminated or substan- less than 3½ per cent a year in outside the OECD. tially diluted pension earnings tests as a way of encouraging older people to work. As with the These adjustments naturally improve work incen- actuarial adjustments, these reforms do not seem tives, but there are two caveats. to have had much effect, usually because they ap- ply at ages well after the majority has retired. But First, the actuarially fair adjustment is generally there is a risk these schemes can, in a way, become larger than the rate applied. The neutral adjust- too effective. People continue working with the ment depends critically on life expectancy (and the added bonus of a pension on top of their earnings. assumed discount rate). So the actuarially fair level varies with age--from around 7 per cent at age 50 Allowing people to combine work and pensions is to 12 per cent at 70--and sex--it is around 2 per- costly when effective pensionable ages are low. centage points higher for women. And this neutral But if pension ages are high enough to avoid abuse Retirement 7 of the system, flexibility avoids discouraging later Retirement depends on many other factors, such retirement. as individual health or the level of unemployment. But financial incentives in the tax and pension `Parametric' reforms system have important effects on the retirement Retirement is often perceived as an `institutional' decision. phenomenon in countries where public or private pension systems have broad coverage. Yet most Pension replacement rates 6 people in these countries retire before the standard pensionable age, sometimes well before. In most 100 effective OECD countries, for example, the normal pension 90 tax rate on Netherlands France working (%) age is now at least 65. But under half of 60-64 80 Italy Belgium year olds are economically active in 48 of the 175 70 Spain countries summarized in Figure 1. 60 Sweden Germany Japan 50 Similarly, the trend to earlier retirement is com- 40 UK mon across countries over the past three decades, US 30 especially the OECD and former socialist coun- 20 tries. Yet only a few countries reduced normal Canada 10 later retirement earlier retirement pension ages in this period. A few have even in- 0 creased pension age while effective retirement ages have fallen. Both of these facts imply that it would be too op- timistic to assume that any increase in pension age would automatically deliver later retirement. Further reading Disney, R.F. and Whitehouse, E.R. (1999), `Pen- Incentives and retirement behavior sion plans and retirement incentives', Pension We have shown how the incentive to retire varies Reform Primer Series, Social Protection Dis- hugely between different pension systems. But it cussion Paper no. 9924, World Bank. is, of course, also important to demonstrate that Gruber, J. and Wise, D.A. (1999), Social Security people's retirement behavior responds to these Programs and Retirement around the World, Univer- incentives. Figure 6 summarizes an international sity of Chicago Press for National Bureau of study of retirement. The study of 11 countries, co- Economic Research. (Survey chapter--no. ordinated by the National Bureau of Economic 6134--and country chapters--nos 6097, 6135, Research, looked at the effective tax rate on con- 6136, 6137, 6153, 6155, 6156, 6169, 6214, tinuing to work (which is related to the 6308--also available as NBER Working Pa- replacement rate). The Figure plots this measure pers. Internet: www.nber.org) of incentives against the degree of early retirement: Blöndal, S. and Scarpetta, S. (1998), `The retire- the total number of years people retire before the ment decision in OECD countries', Ageing standard pension age. Countries to the right of the Working Paper no. 1.4, OECD, Paris. (Coun- chart have earlier retirement than countries on the try studies available as Economics Department left. Working Papers nos 203-207. Internet: www.oecd.org) There is a clear relationship between early retire- OECD (1999), `Workforce ageing in OECD ment and incentives in the pension system, a result countries', Employment Outlook, pp. 123-151. confirmed by other international studies and by (Also available as Ageing Working Paper no. more detailed econometric analysis of individual 4.1) countries' systems. 8 Retirement The box on recent policy reforms is based on: Conclusions and recommendations Kalisch, D.W. and Aman, T. (1998), `Retirement income systems: the reform process across q defined benefit pension schemes OECD countries', Ageing Working Paper no. encourage early retirement in a number 3.4, OECD, Paris. of different ways Schwarz, A.M. and Demirguç-Kunt, A. (1999), q higher pension accruals for early years' `Taking stock of pension reforms around the contributions and maximum pensions world', Pension Reform Primer series, Social mean the increase in pension from Protection Discussion Paper no. 9917, World working at older ages is low or zero Bank. q some systems still levy contributions even when no extra pension is earned q early retirement is often allowed with no reduction or a less than actuarially fair reduction in benefits q schemes that use final or best earnings to calculate the pension encourage retirement once earnings have peaked q they also discourage partial retirement, where people cut their hours of work or take a part-time job q late retirement is discouraged, either by earnings tests which prevent work or by less than fair pension increments for deferring the pension claim q retirement savings plans based on defined contributions are neutral over the choice of retirement age q many of the problems of defined benefit plans can be mitigated with fair actuarial reductions in pension benefits for earlier retirement q but defined benefit plans still tend to encourage earlier retirement relative to defined contribution schemes paymentor age periodic n.1. onretirement specified etc.)better-for´m re pe´nsion made above REFORM ofimperfections, PENSION PRIMER byremoval&i.1. v.t. faults equipprimer make orerrors orabandonment (institution, personn. 1. with elementary procedure information book to