Equitable and Sustainable Pensions
Chapter

Chapter 4. Pension Reform and Equity: The Impact on Poverty of Reducing Pension Benefits

Author(s):
Benedict Clements, Frank Eich, and Sanjeev Gupta
Published Date:
March 2014
Share
  • ShareShare
Show Summary Details
Author(s)
Baoping Shang

Introduction

A primary objective of public pension systems is to provide adequate levels of retirement income to ensure that people are not at risk of poverty in old age. On average, public transfers in Organization for Economic Cooperation and Development (OECD) countries, consisting of earnings-related pensions and means-tested benefits, account for about 60 percent of the total income of the elderly. Income from work and other sources (including private pensions and income from the returns on nonpension savings) each account for about 20 percent of the total (OECD, 2011). Although figures for the OECD as a whole are not readily available, the share of public transfers to the elderly poor is most likely higher. For example, coverage in voluntary funded pension plans among workers in the poorest decile is, on average, about 10–20 percent and much lower than that in the higher income deciles (Antolin, 2008).

Incomes of the elderly are generally lower than that of the working-age population as a whole, and elderly poverty rates are correspondingly higher. However, growing evidence suggests that poverty rates among older people have been declining, reflecting the substantial increase in the generosity of pensions in recent decades (OECD, 2011; Zaidi, 2011). But recent reforms to public pension systems aimed at containing the rise in public pension spending, for example, by reducing pension benefits, are projected to result in substantial declines in income replacement rates (the average pension benefit divided by the average wage) up to 2030.1 The decline will be even larger between 2010 and 2060. Replacement rates in all but one country are projected to decline during that period, and the average decline is nearly 20 percent. These reforms could potentially have adverse impacts on poverty among the elderly if no additional compensating measures are adopted.

The purpose of this chapter is to examine the potential effects of the projected decline in replacement rates on poverty among the elderly. The rest of the chapter is structured as follows: The next section describes the data and patterns of elderly poverty and public pension replacement rates in advanced and emerging European economies. It is followed by a section that provides econometric estimates of the impact of public pension replacement rates on elderly poverty. The subsequent section analyzes the implications on elderly poverty of the projected changes in public pension replacement rates in the coming decades. The final section concludes the chapter with a discussion of policy implications.

Patterns of Elderly Poverty and Pension Replacement Rates

The main data source for this chapter is the Eurostat indicators for poverty and public pension replacement rates. The data are available by gender and age group, and cover 32 advanced and emerging European economies. At-risk-of-poverty data are available for the period 1995–2012, whereas the aggregate replacement rate is only available for 2003–12. In addition, the availability of data varies by country and only a few observations are available for 2012. 2

The at-risk-of-poverty rate is a measure of relative poverty in the elderly population (ages 65 and older). It is defined as the share of persons 65 and older with equivalized disposable income below the at-risk-of-poverty threshold, which is set at 60 percent of the national median equivalized disposable income.3 This type of relative indicator does not measure absolute poverty, but low income in comparison with other residents in the country; thus, it does not necessarily denote a low standard of living. In addition, measuring relative income poverty does not capture poverty’s multidimensional nature. A broader measure would consider the level of indebtedness, the extent of poor health, the number of people living in inadequate housing and poor environmental conditions, and the extent to which people have inadequate access to public services. Such a broad measure, however, is not readily available for a large number of countries for an extended period.

Three types of at-risk-of-poverty rates are available at Eurostat: (1) at-risk-of-poverty rate before pensions and other social transfers, (2) at-risk-of-poverty rate including pensions but not other social transfers, and (3) at-risk-of-poverty rate after pensions and other social transfers. This chapter defines pensions as the sum of the following social benefits: disability pensions, early-retirement benefits resulting from reduced capacity to work, old-age pensions, anticipated old-age pensions, 4 partial pensions, survivors’ pensions, and early-retirement benefits received for labor market reasons. “Other social transfers” refer to nonpension income support, which may or may not be targeted at the elderly.

The poverty rate among the elderly in this sample of countries is more than 85 percent in 2011 when pensions and other social transfers are not included, indicating that work income and income from private savings are low for most of the elderly. Pension income substantially reduces elderly poverty, to about 20 percent; other social transfers further lower elderly poverty to about 15 percent (Figure 4.1). Data also confirm that elderly poverty decreased between 1995 and 2011. Information for 12 countries with complete data between 1995 and 2011 shows that elderly poverty declined by nearly 30 percent during this period.

Figure 4.1At-Risk-of-Poverty Rate before and after Social Transfers

(Percent)

Sources: Eurostat; and author’s estimates.

The female at-risk-of-poverty rate is slightly higher than that for males. Women are less likely to be in paid employment, tend to have lower pensions, and are more involved in unpaid family care. In addition, when they are in work, they usually earn less than men. In 2011, the female at-risk-of-poverty rate was higher than that for males except in two countries (Ireland and Malta), and the gap is, on average, about 6 percentage points (Figure 4.2). This gap mostly reflects the larger impact of pension benefits for males; private income and other social transfers have very similar impacts on both female and male poverty. Trends for the 12 countries with complete data indicate that the gap has been consistently narrowing—by more than 30 percent between 1995 and 2012.

Figure 4.2At-Risk-of-Poverty Rate by Gender, 2011

(Percent; including pensions and other social transfers)

Sources: Eurostat; and author’s estimates.

Note: AUT = Austria; BEL = Belgium; BGR = Bulgaria; CHE = Switzerland; CYP = Cyprus; CZE = Czech Republic; DEU = Germany; DNK = Denmark; EST = Estonia; GRC = Greece; ESP = Spain; FIN = Finland; FRA = France; GBR = United Kingdom; HUN = Hungary; HRV = Croatia; IRL = Ireland; ISL = Iceland; ITA = Italy; LUX = Luxembourg; LTU = Lithuania; LVA = Latvia; MLT = Malta; NLD = Netherlands; NOR = Norway; POL = Poland; PRT = Portugal; ROM = Romania; SVN = Slovenia; SWE = Sweden; SVK = Slovak Republic.

Elderly poverty also varies significantly by educational attainment. Poverty rates for those with lower secondary education or less are twice the level of those having upper secondary and postsecondary but no tertiary education. They are four times the level of those that have completed the first and second stages of tertiary education (Figure 4.3). This suggests that it may be unrealistic to expect current low-income and low-skilled workers (who are overrepresented among the elderly poor) to save more during their lifetimes to compensate for falling replacement rates.

Figure 4.3At-Risk-of-Poverty Rate by Educational Attainment

(Percent)

Sources: Eurostat; and author’s estimates.

Relative to other age groups, the elderly have higher at-risk-of-poverty rates (Figure 4.4). However, elderly poverty decreased significantly in 2010 and 2011 because pension income has been less sensitive to the economic downturn since the onset of global financial crisis, whereas the poverty of other age groups increased (Figure 4.4, panel 3). It is of interest that the elderly receive smaller social benefits than other age groups as demonstrated by the fact that the poverty rate of the elderly drops less than that of other groups when other social benefits are included (Figure 4.4, difference between panels 2 and 3).

Figure 4.4At-Risk-of-Poverty Rate by Age Group

(Percent)

Sources: Eurostat; and author’s estimates.

The “aggregate replacement rate” is the best measure for approximating the generosity of pension benefits for the entire elderly population. The aggregate replacement rate is defined as the ratio of median individual gross pensions of persons ages 65–74 relative to median individual gross earnings of persons ages 50–59 (excluding other social benefits). The replacement rates for both males and females have been increasing since 2009 (Figure 4.5). This increase, however, may reflect more of a decline in the earnings of the working-age group than an increase in the pension benefits of the elderly. The aggregate replacement rate was very similar for men and women until the onset of the global financial crisis, but a gap has since developed, with men experiencing a larger increase than women.

Figure 4.5Aggregate Replacement Rate by Gender

(Percent)

Sources: Eurostat; and author’s estimates.

Relationship Between Public Pension Replacement Rates and Elderly Poverty

Few studies have attempted to quantify the impact on poverty among the elderly of reducing the generosity of pension benefits, given the difficulty of estimating the corresponding effects on other sources of income, either through changes in working and saving behavior or through changes in other social transfers. The literature, in general, uses two approaches, each with its advantages and disadvantages. The first approach uses household survey data to simulate the impact of changes in pension income on elderly poverty. It makes explicit, often simplifying, assumptions about other types of income. For example, asset income and employment income are assumed to grow at the same rate as GDP (Jackson, Howe, and Naka-shima, 2010). This approach is transparent, but how well it represents reality is unclear. The second approach empirically estimates a reduced-form relationship between pension income and the elderly poverty rate using historical data that already take into account changes in other sources of incomes. It then simulates the impact of pension reform on poverty, assuming the relationship holds in the future (Zaidi, Grech, and Fuchs, 2006). One shortcoming of this approach, like most reduced-form models, is its lack of clearly specified channels. One of the key challenges for this approach is to address the potential endogeneity of pension replacement rates, that is, pension replacement rates are determined as part of the social protection system and thus are likely correlated with the characteristics of the rest of the social security system.

This chapter builds on the second approach to estimate the relationship between pension replacement rates and elderly poverty. The dependent variable is the at-risk-of-poverty rate after pensions and other social transfers for the 65 and older age group, and the key independent variable is the aggregate replacement rate. The analysis takes advantage of the panel nature of the Eurostat data and uses several panel data models to address the endogeneity problem, including a fixed-effects model, a fixed-effects model with a first-order autoregressive (AR(1)) error term, and a difference generalized method of moments (GMM) model, and an ordinary least squares (OLS) model is used as a reference (see Appendix 4A for more details).

The aggregate replacement rate can affect poverty through several channels. First, a lower replacement rate directly reduces the income of the elderly. Second, the elderly with reduced income as a result of lower replacement rates may qualify for a higher amount of other social benefits, which would offset some of the effects of lower replacement rates. Third, some of the elderly may respond to lower replacement rates by increasing labor supply and thus increasing labor income. And fourth, in anticipation of lower replacement rates, people may choose to increase private savings, including participation in private pensions and accumulating other assets, to raise future retirement income. Because the at-risk-of-poverty rates after pensions and social transfers are based on income from all sources, including private savings and pensions, work income, public pensions, and other social transfers, the reduced-form model estimation implicitly takes all four channels into account.

Table 4.1 reports the econometric estimates for two age groups, ages 65 and above and ages 25–54, under all four specifications. As expected, the aggregate replacement rate has no effect on poverty of those ages 25–54. The coefficients under all specifications, except OLS, are small and insignificant. This outcome provides evidence that estimates from the simple cross-section OLS specification are biased.

Table 4.1Econometric Estimates of the Relationship between Public Pension Replacement Rates and Elderly Poverty
Fixed effects
model with an
OLS modelFixed effects modelAR(1) error termDifference GMM
Ages 65 and older
Log aggregate replacement rate−1.43***−.48***−0.40***−0.46***
Male−0.43***−0.45***
Log GDP per capita−0.16***3.07***1.69***1.70***
Sample size466466404356
Adjusted R squared0.470.820.21
Ages 25–54
Log aggregate replacement rate0.23***0.030.050.04
Male−0.05**−0.05***
Log GDP per capita−0.09***−0.64***−0.19−0.40***
Sample size456456394334
Adjusted R squared0.280.870.27
Sources: Eurostat; and author’s estimates.Note: AR(1) = first-order autoregression; GMM = generalized method of moments; OLS = ordinary least squares.*, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively.
Sources: Eurostat; and author’s estimates.Note: AR(1) = first-order autoregression; GMM = generalized method of moments; OLS = ordinary least squares.*, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively.

As expected, the risk of poverty in old age is negatively related to the pension’s income replacement rate. A simple OLS model appears to overestimate the elasticity between the aggregate replacement rate and elderly poverty with a coefficient of -1.43, indicating that potential bias is introduced by the likely correlation between the replacement rate and country fixed effects, which may include the generosity and progressivity of other social transfers.

The elasticity estimate from panel data models is about -0.4, which implies that a 10 percent reduction in the aggregate replacement rate would increase the elderly poverty rate by about 4 percent. This elasticity is based on the at-risk-of-poverty threshold of 60 percent of the national median equivalized disposable income; it would change under alternative thresholds as the concentrations of elderly population differ. In addition, elderly women have a higher poverty rate than elderly men, even after controlling for aggregate replacement rate and country effects. An increase in per capita GDP is associated with higher elderly poverty, likely capturing the fact that average income (for the working-age population) grows faster than pension income during periods of economic growth.

Impacts of the Projected Decline in Replacement Rate on Elderly Poverty

According to the 2012 Ageing Report from the European Commission, sizable decreases in pension generosity are projected for the coming decades in many European countries as a result of pension reforms in recent years that were mostly directed to strengthening the financial sustainability of pension systems by decreasing coverage and benefits (European Commission, 2012).

Of the 23 European Union member states for which data are available, an average 8.8 percent decline in the replacement rate is projected between 2010 and 2030 (Figure 4.6). Only 3 of the 23 countries are projected to have an increase in the public pension benefit ratio (Belgium, Cyprus, and Italy). In six countries (Estonia, Malta, Poland, Romania, the Slovak Republic, and Sweden), the decline is projected to be more than 15 percent. A rather substantial decline in the public pension benefit ratio is projected for the period 2010–60, amounting to an average reduction in the replacement rate of nearly 20 percent. A reduction of 20 percent or more is projected in seven countries (Estonia, Greece, France, Poland, Romania, the Slovak Republic, and Sweden). Only Cyprus is projected to have a slightly increasing public benefit ratio over the projection horizon.

Figure 4.6Projected Changes in Pension Replacement Rates

(Percent)

Sources: Eurostat; European Commission; and author’s estimates.

Note: AUT = Austria; BEL = Belgium; BGR = Bulgaria; CHE = Switzerland; CYP = Cyprus; CZE = Czech Republic; DEU = Germany; DNK = Denmark; EST = Estonia; GRC = Greece; ESP = Spain; FIN = Finland; FRA = France; GBR = United Kingdom; HUN = Hungary; HRV = Croatia; IRL = Ireland; ISL = Iceland; ITA = Italy; LUX = Luxembourg; LTU = Lithuania; LVA = Latvia; MLT = Malta; NLD = Netherlands; NOR = Norway; POL = Poland; PRT = Portugal; ROM = Romania; SVN = Slovenia; SWE = Sweden; SVK = Slovak Republic.

In light of the large projected decline in replacement rates in recent pension reforms, it is thus important to assess what effect these reforms could have on pension adequacy and elderly poverty. An elasticity of -0.8 between the public pension replacement rate and elderly poverty indicates that elderly poverty would, on average, increase by nearly 0.5 percentage point (3 percent) between 2010 and 2030. There are considerable variations, however, by country. Elderly poverty is projected to increase in all but three countries, with the increase exceeding 1 percentage point in seven countries (Bulgaria, Estonia, Malta, Poland, Romania, Sweden, and Slovenia) (Figure 4.7).

Figure 4.7Poverty Impact of Changes in Pension Replacement Rates

Sources: Eurostat; European Commission; and author’s estimates.

Note: AUT = Austria; BEL = Belgium; BGR = Bulgaria; CHE = Switzerland; CYP = Cyprus; CZE = Czech Republic; DEU = Germany; DNK = Denmark; EEM = European Economic and Monetary Union; ESP = Spain; FIN = Finland; FRA = France; GBR = United Kingdom; HUN = Hungary; HRV = Croatia; ISL = Iceland; ITA = Italy; LUX = Luxembourg; LTU = Lithuania; LVA = Latvia; MLT = Malta; NLD = Netherlands; NOR = Norway; POL = Poland; PRT = Portugal; ROM = Romania; SVK = Slovak Republic; SWE = Sweden.

The increase in elderly poverty is substantially larger between 2010 and 2060. Elderly poverty is projected to increase in all countries but Cyprus, and on average, rise by 1.1 percentage points, or roughly 7 percent. Again, variations are substantial across countries, and projected increases range from 0.08 percentage point in the Czech Republic to nearly 3½ percentage points in Poland. 5 The percentage point increase in poverty is highest in Bulgaria, Estonia, Greece, Poland, and Sweden, with the increase in elderly poverty exceeding 2 percentage points. The poverty increase in France and the Slovak Republic is moderate, despite the large reduction in their replacement rates, because of their low initial poverty rates in 2011. Similarly, the large poverty impact for Estonia, despite the moderate reduction in its replacement rate, is due to its high initial poverty rate of more than 20 percent in 2011.

Because the elasticity estimate is based on current social protection systems, the projections assume the same social protection system design in the future, including, for example, the progressivity of public pensions and other social benefits. The projections, however, are not intended to predict elderly poverty in the future given that changes in the social protection system and economic structure could help offset the effects on poverty. Rather, the purpose of the projections is to provide a sense of how poverty among the elderly could evolve if future social protection systems remain similar to those of the past. The following section provides a discussion of options to achieve this.

Discussion and Conclusion

Most of the recent pension reforms are projected to lead to a reduction in public pension benefits. The analysis in this chapter suggests that measures are needed to mitigate the adverse impact of pension reform on the well-being of the elderly, for example, through the design of social security systems or labor market policies.

The most direct way to limit the impact of pension reforms on old-age poverty is to lower pension benefits only for those with higher income—that is, to increase the progressivity of public pension benefits. However, this policy should be weighed against any possible adverse effects on the labor market. Similarly, other social transfers could be scaled up to provide additional assistance to the elderly. As discussed earlier, other social transfers currently are shown to have a relatively small impact on elderly poverty.

Although the efforts to better target these benefits to the poor could have a substantial effect on elderly poverty while helping contain the fiscal costs of these benefits, it may be challenging to increase voluntary pensions and other private savings during working lives to offset the reduction in public pension benefits for the low skilled and less educated because of their low earnings and low participation in voluntary pension plans. Automatic enrollment in voluntary pension plans, however, may help increase coverage (Benartzi and Thaler, 2013; Beshears and others, 2008).

The elderly poor can also respond to lower public pension benefits by increasing their labor supply. A number of barriers to old-age employment will need to be overcome for this to be successful:

  • Employers often believe that older workers are less productive and less able to adapt to change. Legislation against age discrimination has been effective in many countries, and public information campaigns could also help correct this misperception (OECD, 2011).

  • In some countries, older workers cost too much and retirement provides a convenient way of adjusting the size of the workforce, and strict employment-protection legislation can make it costly to hire older workers (Daniel and Heywood, 2007; OECD, 2004). Labor market reforms and reforms to the seniority-based wage structure could help limit these impacts (IMF, 2013).

  • Employment opportunities for older workers are often limited. In addition, older workers are less likely than their younger counterparts to take part in training (OECD, 2011). Encouraging the elderly to work may require many of the effective Active Labor Market Programs and training programs to be expanded to include the elderly population (IMF, 2011).

  • Available employment opportunities may also be unattractive to the elderly because of poor working conditions or unsuitable and inflexible working-time arrangements; improving working conditions for elderly workers could help mitigate this effect (OECD, 2011).

Several options could help reduce the higher old-age poverty among women. Policies that help promote labor force participation of women, for example, strengthening child care benefits, could enhance their ability to save during working lives and entitle them to higher replacement rates at retirement (IMF, 2011). Increasing the replacement rates for survivors’ pensions could help reduce old-age poverty among women because life expectancy is higher for women and they often live into widowhood while receiving survivors’ pensions (Hoff, 2008; Anzick and Weaver, 2001).

Appendix 4A. Methodology to Estimate the Relationship Between Public Pension Replacement Rate and Elderly Poverty

This chapter uses a panel data econometric model to estimate the relationship between pension income replacement rates and elderly poverty. The dependent variable is at-risk-of-poverty after pensions and other social transfers for the 65 and older age group, and the key independent variable is the aggregate replacement rate. The econometric model takes the following form:

in which i indicates country; j indicates gender; and t indicates year. The variable X includes gender and real per capita GDP (in log terms), R denotes the aggregate replacement rate (in log terms) and captures the generosity of pension benefits, and Poverty denotes the at-risk-of-poverty rate (in log terms). Year fixed effects control for common shocks across countries in the sample. Country denotes the unobserved country-level effects, and ɛ is an error component. Log transformation is applied to both at-risk-of-poverty rates and the aggregate replacement rate to reduce the impact of outliers on the estimates of the regression coefficients. Coefficient β1 is thus an elasticity measure that represents the percentage change in the elderly poverty rate due to a 1 percent change in the pension income replacement rate.

Because the aggregate replacement rate is not exogenously determined and is likely to be correlated with unobserved country characteristics, addressing endogeneity is a key challenge. This analysis takes advantage of the panel nature of the Eurostat data on the at-risk-of-poverty rate and aggregate replacement rate and compares several model specifications:

  • A simple OLS model. Because replacement rate is likely correlated with country fixed effects, a simple OLS estimate would be biased. This chapter uses the OLS estimate as a reference point.

  • A fixed-effects model. A fixed-effects model can effectively control for unobserved time-invariant country-specific effects. However, it does not address unobserved time-variant country-specific effects. In addition, it is not efficient when serial correlations are present.

  • A fixed-effects model with an AR(1) error term. This model can improve the efficiency of the fixed-effects model when the error term follows an AR(1) process.

  • A difference GMM model. A difference GMM approach is appropriate when elderly poverty depends on its own past realizations; aggregate replacement rates are not strictly exogenous, meaning correlated with past and possibly current realizations of the error; and heteroskedasticity and autocorrelation occurs within individuals but not across them.

References

    AntolinPablo2008Coverage of Funded Pension Plans,OECD Working Papers on Insurance and Private Pensions No. 19 (Paris: OECD Publishing).

    AnzickMichael and DavidWeaver2001Reducing Poverty among Elderly Women,ORES Working Paper No. 87 (Washington: Office of Research, Evaluation, and Statistics, U.S. Social Security Administration).

    BenartziShlomo and RichardThaler2013Behavioral Economics and the Retirement Savings Crisis,Science Vol. 339 No. 6124 pp. 115253.

    BeshearsJ.J.J.ChoiD.Laibson and B.C.Madrian2008The Importance of Default Options for Retirement Saving Outcomes: Evidence from the United States,in Lessons from Pension Reform in the Americased. by S.J.Kay and T.Sinha (Oxford: Oxford University Press) pp. 5987.

    DanielK. and J.Heywood2007The Determinants of Hiring Older Workers: UK Evidence,Labor Economics Vol. 14 pp. 3551.

    European Commission2012The 2012 Ageing Report: Economic and Budgetary Projections for the 27 EU Member States (Brussels: Directorate-General for Economic and Financial Affairs, European Commission).

    HoffAndreas2008Tackling Poverty and Social Exclusion of Older People: Lessons from EuropeWorking Paper No. 308 (Oxford: Oxford Institute of Aging, Oxford University).

    International Monetary Fund (IMF)2011Fiscal Policy and Employment in Advanced and Emerging Economies” (Washington: International Monetary Fund).

    International Monetary Fund (IMF)2013Jobs and Growth: Analytical and Operational Considerations for the Fund” (Washington: International Monetary Fund).

    JacksonRichardNeilHowe and KeisukeNakashima2010The Global Aging Preparedness Index (Washington: Center for Strategic and international Studies).

    Organization for Economic Cooperation and Development (OECD)2004OECD Employment Outlook” (Paris: OECD Publishing).

    Organization for Economic Cooperation and Development (OECD)2011Pensions at a Glance 2011: Retirement-Income Systems in OECD and G20 Countries (Paris: OECD Publishing).

    ZaidiAsghar2011Poverty and Income of Older People in OECD Countriesunpublished.

    ZaidiAsgharAaronGeorge Grech and MichaelFuchs2006Pension Policy in EU25 and Its Possible Impact on Elderly Poverty,CASEpaper No. 116 (London: Centre for Analysis of Social Exclusion).

Projections for the 23 European Union countries for which data are available indicate that public pension replacement rates will decline in all but three countries, averaging a decline of about 9 percent between 2010 and 2030. The 23 European Union countries are Austria, Belgium, Bulgaria, Cyprus, the Czech Republic, Denmark, Estonia, Germany, Greece, Spain, Finland, France, Hungary, Italy, Luxembourg, Lithuania, Malta, Norway, Poland, Romania, the Slovak Republic, Slovenia, and Sweden.

Turkey is dropped from the analysis because data are only available for a few years.

Equivalized disposable income is the total income of a household, after taxes and other deductions, that is available for spending or saving, divided by the number of household members converted into equalized adults; household members are equalized or made equivalent by weighting each according to their age, using the so-called modified OECD equivalence scale. The 60 percent threshold is most often referred to by Eurostat, but alternative poverty thresholds are also available at Eurostat.

Anticipated old-age pensions are periodic payments intended to maintain the income of beneficiaries who retire before the legal or standard age as established in the pension scheme.

Poland has adopted measures to gradually increase its retirement ages, to 67 in 2020 for men and in 2040 for women.

    Other Resources Citing This Publication