April 21, 2003
EXAMINING THE NEW PORTMAN-CARDIN LEGISLATION:Peter R. Orszag
Are Further Pension Tax Subsidies for High-Income Households Affordable or Sound as Pension Policy?
PDF of this report
HTM of fact sheet
PDF of fact sheet If you cannot access the files through the links, right-click on the underlined text, click "Save Link As," download to your directory, and open the document in Adobe Acrobat Reader.
On April 11, Representatives Rob Portman and Ben Cardin introduced legislation to make changes in the tax laws governing pensions and Individual Retirement Accounts (IRAs). As with earlier Portman-Cardin proposals, this legislation includes a mix of promising and problematic provisions. Most of the provisions that involve the largest revenue losses, however, represent problematic policy. At a time when substantial budget deficits loom as far as the eye can see, these provisions would provide additional tax subsidies to high-income individuals who would likely save without these new tax breaks and who already tend to be much better prepared for retirement than other Americans with less income and wealth.
The revenue losses that these provisions would engender would exacerbate the already dire fiscal outlook. Although an official revenue estimate is not yet available for the legislation, Representative Portman has indicated that the bill as a whole would cost more than $100 billion over the next 10 years. Such costs would presumably be on top of the $350 billion to $550 billion in reconciliation tax cuts allowed by the recently adopted budget resolution. In the face of large budget deficits, there is serious doubt that the nation can afford the costly and not-well targeted subsidies the legislation includes.
Among the provisions in the new Portman-Cardin legislation that involve the most substantial revenue reductions are: the acceleration of scheduled increases in the amounts that can be contributed to 401(k)s and IRAs; an increase from $160,000 to $220,000 in the income limit for contributions to Roth IRAs by married couples; expansions in income limits for contributions to traditional IRAs by married couples; making permanent all of the pension and IRA provisions from the 2001 tax act; and a weakening of the "minimum distribution" rules intended to ensure that tax-advantaged retirement accounts are used primarily to finance retirement needs, rather than for other purposes such as estate planning by wealthy individuals. These proposals would provide substantial additional tax subsidies to upper-income households who least need additional help in preparing for retirement, while providing little or no benefit to the majority of families struggling to save for retirement.
The legislation includes some beneficial changes. It expands and makes permanent the "saver’s credit" created by the 2001 tax legislation (this credit currently is scheduled to sunset at the end of 2006), although the expansion of this credit is less substantial than may initially appear to be the case because the credit is not refundable. The legislation also reforms current rules in the Supplemental Security Income (SSI) program under which poor individuals who become disabled can be disqualified from receiving SSI benefits unless they liquidate their retirement accounts and spend the proceeds, leaving them with little or no assets for their old age. These and other provisions are described briefly in the below.
This analysis focuses on the most expensive and regressive components of the Portman-Cardin legislation — the acceleration and extension of increased contribution limits and income limits for tax-advantaged retirement accounts, along with the liberalization of the "minimum distribution" rules. The analysis finds that these proposals represent unsound pension policy and would not be fiscally prudent in the current budget climate.
Expanded Tax Subsidies for Retirement Saving Among High-Income Households
Before examining the specific provisions, it is important to recognize that higher-income households are much more likely to have pensions and to save adequately for retirement than other workers. Moreover, the pension contributions they make are less likely to represent new saving and more likely to represent asset shifting (that is, the shifting of existing savings from a taxable account to a tax-preferred account) than the pension accumulations of lower-income earners. Conversely, lower-income households are less likely to be saving adequately for retirement and are less likely to have pensions than higher earners are. In addition, the pension contributions they do make are more likely to represent new saving.
For these and other reasons, current tax benefits related to pensions accrue disproportionately to high-income households but do not appear to have a large effect on the overall amount that such households save, since these households tend to shift savings from one account to another to take advantage of pension tax breaks. By contrast, lower- and middle-income households gain much less from the current pension system, but the pension benefits they do receive appear both to increase the amount they save and to help them accumulate more adequate assets for retirement. This strongly suggests that pension reform should encourage expanded pension coverage and participation among low- and middle-income households, since that would be more likely to boost national saving and build wealth for households who often are saving very little. Unfortunately, most of the pension tax measures approved in recent years have moved in the opposite direction and concentrated on expanding tax-preferred savings opportunities further for high-income households.
Accelerating and Extending Higher IRA and Pension Limits from the 2001 Tax Legislation
The Economic Growth and Tax Relief Reconciliation Act of 2001(the 2001 tax-cut law) included a series of major changes in the tax laws governing pensions and IRAs. The main provisions make various changes that allow larger contributions by, and on behalf of, high-income individuals, such as business owners and executives. The 2001 tax-cut law gradually phases in most of these changes and then sunsets all of them by the end of 2010.
In 2001, for example, workers were allowed to make a maximum of $10,500 in tax-favored contributions to a 401(k) account. The 2001 tax-cut law raises the maximum gradually to $15,000 by 2006 (and to $20,000 for those aged 50 or over). Similarly, the 2001 tax legislation also more than doubles the amount that a taxpayer and spouse can contribute each year to an IRA. Under prior law, a taxpayer and spouse could each contribute $2,000; the 2001 legislation gradually raises the maximum contribution to $5,000 apiece by 2008 (and to $6,000 apiece for those aged 50 or over).
Accelerating the Increases in Contribution Limits
The Portman-Cardin legislation would accelerate these increases, making the full increases in the 401(k) and IRA contribution limits effective in 2003. The proposed acceleration of the 401(k) and IRA contribution limits is problematic both as pension policy and as economic policy:
- As noted above, the primary effect of increasing the contribution limits is likely to be that high-income households shift other saving they already are undertaking from taxable accounts to the tax-preferred accounts. By shifting funds, such households would be able to capture the additional tax subsidies without raising their overall level of saving.
- Moreover, increasing the contribution limits would have little effect on middle- and upper-middle-income families and individuals. The vast majority of Americans do not make the maximum contributions to their 401(k)s or IRAs today and therefore would benefit little, if at all, from accelerating the increases in the maximum contribution levels. A study conducted in 2000 by an economist in the Office of Tax Analysis at the Treasury Department found that only four percent of all taxpayers who were eligible for traditional IRAs in 1995 made the maximum allowable contribution, which was $2,000 at that time. Likewise, the proposed 401(k) changes would affect only a very small percentage of the population. The General Accounting Office found that an increase in the contribution limit for 401(k)s would directly benefit fewer than three percent of participants; this indicates that increasing the limit on employee 401(k) contributions to $15,000 immediately would disproportionately benefit those on the higher rungs of the compensation scale. Other recent studies have reached similar conclusions, finding that the fraction of individuals constrained by the IRA or 401(k) limits in place prior to enactment of the 2001 tax-cut legislation was very small.
- Finally, given current economic conditions, the acceleration would represent unsound economic policy if it succeeded in boosting saving. Whatever their actual effects, increases in contribution limits are typically advertised as inducing additional saving, not additional spending. Yet to boost the economy in the near term, inducing additional spending now — rather than more saving — is the appropriate policy approach. (Since firms have excess capacity and could produce more if there were more demand for their goods and services, it is additional consumption that would spur the economy in the near term.) The new Portman-Cardin legislation thus is a peculiar bill to be advocating in the current sluggish economic environment. If it were successful in achieving its ostensible goal of immediately raising retirement saving, it would be counterproductive now from an economic perspective.
Making the Increases Permanent
Like most of the rest of the 2001 tax legislation, the increases in contribution limits that are contained in that legislation sunset in 2010. The new Portman-Cardin bill would not only accelerate these increases in the contribution limits but also would make these increases permanent.
Making these provisions permanent at this time is another step that does not represent sound policy, especially since little information is yet available on whether these provisions are achieving their intended effects. In view of the staggering deficits projected when the baby-boom generation retires in large numbers, Congress should wait for information on the impact of the 2001 changes on pension coverage before rushing to lock these provisions into permanent law.
· These provisions represent an untested "trickle down" approach to pension coverage: they were promoted on the grounds that if the tax laws governing retirement savings were made more generous for higher-income owners and executives, the increased generosity would encourage more small businesses to offer pension plans, which would result in pension coverage being extended to more rank-and-file workers. This approach lacked solid empirical backing when the legislation was passed, and little information has emerged since enactment of the legislation to indicate that these provisions are promoting retirement saving among middle and lower earners. In fact, the early evidence on the effects of the legislation is not particularly encouraging. This suggests caution and taking the prudent course of waiting to see how the approach works in practice over the next few years before permanently locking in these changes at significant cost.
- Indeed, making the pension provisions permanent at this time would cause a further deterioration in a budget outlook that has already worsened dramatically since the 2001 tax legislation was enacted. The Congressional Budget Office has reported that for the 2002-2011 period, the budget has deteriorated from a projected surplus under then-current policies of $5.6 trillion in January 2001 to a projected deficit under current policies of $378 billion today. The difference between the unified budget projections made in January 2001 and the current projections is $6 trillion over this ten-year period, or an average of $600 billion per year.
- Furthermore, the official $378 billion deficit projection for the 2002-2111 period is based on the unrealistic assumptions that all of the 2001 tax cut will be allowed to expire by the end of 2010 and none of it will be extended, and that the Alternative Minimum Tax will be allowed to mushroom and to affect 38 million tax filers a year by 2011 (as compared to roughly three million today).
Under more realistic assumptions, deficits from 2004 through 2013 will total well over $1 trillion, and deficits outside Social Security will amount to approximately $4 trillion over this period. Moreover, even these less rosy figures do not include funds for any new tax cuts or spending increases, such as the tax-cut "growth" package that Congress is expected to approve in May, a Medicare prescription drug benefit, or the costs of the war in Iraq and subsequent reconstruction. When these costs are taken into account, the projected deficits spiral to even higher levels.
Given this budgetary situation, policymakers should impose a high burden of proof on new legislation with significant long-term budget costs, particularly in cases where it is unclear whether the legislation will achieve its ostensible goals. The troubling fiscal outlook adds weight to the view that it would be more prudent to wait before acting to make the retirement provisions of the 2001 law permanent. Retirement planning does require some certainty about the long-term rules applying to pensions, but at this point, the benefits of waiting to evaluate the effects of last year’s retirement provisions before making them permanent substantially outweigh the costs.
New Tax Subsidies for Higher-income Households
The Portman-Cardin legislation not only accelerates and extends the higher IRA and retirement savings contribution limits included in the 2001 tax legislation. It also expands tax subsidies for higher-income households beyond those included in the 2001 act.
For example, it increases the income limit for full contributions to Roth IRAs by joint filers from $150,000 to $190,000 and the income limit for any contributions to Roth IRAs from $160,000 to $220,000. It also eliminates all income limits on tax-deductible IRA contributions by a high-income worker who is not covered by an employer-provided pension plan even if his or her spouse is covered by such a pension plan. In addition, it substantially weakens the "minimum distribution" rules, which are intended to ensure that the tax subsidies provided for pension saving are used to finance needs during retirement, not as estate planning devices for affluent individuals.
Higher Income Limit for Roth IRAs and Traditional IRAs
Under current law, eligibility for contributions to Roth IRAs is phased out between $150,000 and $160,000 in adjusted gross income for married couples filing joint returns. Eligibility for tax-deductible contributions to traditional IRAs for joint filers who are covered by employer-provided plans is currently phased out between $60,000 and $70,000 in adjusted gross income. That phase-out range is scheduled to increase to $80,000 to $100,000 by 2007.
If neither member of a couple is covered by an employer-provided plan, no income limit applies to tax-deductible contributions to traditional IRAs. If one member of the couple is covered by an employer-provided plan but the other one is not, the uncovered member of the couple can make tax-deductible contributions to a traditional IRA, subject to the same limits as Roth IRA contributions (i.e., the phase-out begins at $150,000).
The Portman-Cardin legislation would increase all of these limits for joint filers, as shown in the table below. The increases would be substantial: In 2004, for example, couples earning $190,000 would be entitled to make $6,000 in contributions to Roth IRAs ($3,000 per spouse), compared to zero under current law. (Since current law phases in an increase in the traditional IRA limit and since Portman-Cardin also phases in its expansion in those limits, the table below shows the results under current law and the Portman-Cardin proposal for 2010, at which point all the provisions would be fully in effect.)
Phase-Out Range for IRA Contributions
for Married Couples Filing Joint Returns, 2010
Both members of couple covered by employer-provided plar
One member covered by employer-provided plan
* Effective 2004
** Effective 2007
The press materials accompanying the Portman-Cardin legislation present these changes as ensuring that the income limits for joint filers are twice those for single filers in order to address a "marriage penalty" in IRA eligibility. Yet these changes would have no effect on the vast majority of married families. The changes would have no effect on joint filers with income below $150,000:
- The expanded eligibility for Roth IRAs included in Portman-Cardin affects only married couples with incomes between $150,000 and $220,000, most of whom already benefit from tax-preferred employer-based retirement plans and who tend to accumulate adequate retirement saving.
- Similarly, the proposal that would eliminate income limits on tax-deductible contributions to traditional IRAs for workers who are not covered by employer-provided plans, but whose spouses are, would only affect couples with more than $150,000 in income. For those with lower incomes, the current rules already allow tax-deductible contributions to traditional IRAs.
According to results from the Tax Policy Center micro-simulation model, 90 percent of joint filers in 2003 have incomes below $150,000. For 90 percent of joint filers, the proposed changes thus are not relevant. The top 10 percent of joint filers who would benefit from the changes tend to have high incomes and extensive wealth. For these high-income married couples, the additional new tax subsidies are likely to result primarily in asset shifting and more tax sheltering, rather than new saving.
Other Provisions of the Bill With Significant Revenue Effects
Several provisions of the Portman-Cardin legislation that also seem likely to have a significant revenue effect are potentially beneficial, although some of these proposals have problematic aspects, as well.
- The legislation expands and makes permanent the "saver’s credit" (which currently is scheduled to sunset at the end of 2006) created by the 2001 tax legislation. Expanding this credit and making it permanent is a promising proposal, since the credit could play an important role in making the pension system more progressive and encouraging saving among middle and lower earners. The Portman-Cardin approach to expanding the credit, however, has a fundamental weakness: The legislation does not make the credit refundable, so the credit will continue to be of little or no benefit to millions of workers with modest incomes.
- The legislation encourages annuitization — that is, the transformation of an accumulated balance in a 401(k) or IRA into a payment per month that lasts as long as the worker or spouse is alive — by allowing up to $2,000 per year of annuitized income to be tax-free. The legislation would phase this tax-free preference out for couples with incomes above $150,000. On the one hand, the objective of this proposal — to encourage broader annuitization — is sound, since annuitization is crucial to ensuring that retirees will not outlive their savings. On the other hand, this proposal would exempt from taxation income that has already enjoyed a substantial tax preference: The initial contributions to the 401(k) or IRA will have been tax-deductible, and the accumulation of funds within the account will not have been taxed. Providing yet another tax advantage to these funds – by partially exempting withdrawals from taxation if the account is annuitized – is not necessarily sound tax policy. Moreover, it is unclear whether this approach is the best mechanism for encouraging annuitization. Roughly 66 percent of elderly tax filers face a marginal tax of 15 percent or lower. Thus, for two-thirds of the elderly, the $2,000 exclusion would save them at most $300 a year in taxes. It is unclear how many people will be encouraged to annuitize as a result of those modest savings. Finally, $150,000 in income is a high threshold for this provision. About 95 percent of elderly couples — a substantial number of whom have considerable wealth — have current incomes below $150,000. An elderly couple aged 65 could transform roughly $2 million in assets into an annuity paying about $150,000 per year and qualify for this tax break.
- Finally, the legislation reforms current rules in the Supplemental Security Income (SSI) program under which poor individuals who become disabled can be disqualified from receiving SSI assistance unless they liquidate their retirement accounts and spend most of the proceeds, leaving them with little or no assets for their old age. Under the legislation, SSI beneficiaries would be allowed to maintain up to $75,000 in retirement-account assets without being disqualified for SSI. (Beneficiaries will be expected to begin drawing down these funds upon reaching retirement age, with the withdrawals being counted as income in determining SSI eligibility and benefit levels.) The provision will help to address a current inequity in the SSI program, under which assets in defined benefit pension plans do not count against the program’s stringent resource limit, but assets in defined contribution plans do count.
The Portman-Cardin legislation includes a number of other pension changes, as well. Several of them may have significant revenue costs. Others do not have much effect on revenue, but do have important implications for the pension system.
Loosened "Minimum Distribution" Rules
The legislation would also loosen the minimum distribution rules for defined contribution plans, such as 401(k)s. These rules are intended to ensure that the substantial tax benefits provided for pensions and IRA contributions are actually used to finance retirement needs. As Mark Warshawsky, currently a Deputy Assistant Secretary at the Treasury Department, has previously written: "The public policy purpose of the minimum distribution requirements…is to ensure that tax-qualified retirement plans serve primarily as vehicles for providing income during the retirement of the plan participant and his or her spouse….The government also intends that the use of retirement plans for tax avoidance schemes and the accumulation of large estates should be minimized."
To ensure that retirement plan assets are used primarily to finance retirement needs, workers generally must begin to draw down their accumulated pensions by age 70½, or when they retire, whichever is later. This rule ensures that pension accumulations are used at least in part during retirement. In the absence of such a rule, high-income individuals could use the tax benefits associated with pensions and IRAs as tax shelters, making contributions to tax-preferred pension and IRA accounts that they never intend to use for retirement needs. Instead, in the absence of some form of minimum distribution rule, tax-preferred pension and IRA accounts could be used to accumulate substantial estates (rather than to provide income during retirement). In that case, the tax preferences associated with pensions and IRAs would not be serving their basic public policy purpose of bolstering retirement security. As Professor Jay Soled of Rutgers University and Bruce Wolk of the University of California at Davis have written, "There seems little justification for a system that, on one hand, allows the highly compensated to amass significant tax-favored wealth on the theory that it was needed for retirement, but, on the other hand, permits them to perpetuate their own financial dynasties as this wealth moves across multiple generations, retaining its tax-favored status."
Pension experts agree that the minimum distribution rules are complicated. Efforts to simplify them are already underway, however, including important simplifications contained in recent IRS regulations. If further steps are required, an alternative approach of exempting a moderate level of assets from the minimum distribution rules would ensure that the rules do not apply to the vast majority of retirees.
For example, the rules could be modified so that each person could exempt up to $50,000 of pension and retirement account assets from the minimum distribution requirements. Data from the 2001 Survey of Consumer Finances suggest that more than 70 percent of households aged 55-64 own defined contribution and IRA assets of less than $50,000. If the minimum distribution rules did not apply to assets of less than $50,000, these rules would thus cease to affect approximately two-thirds or more of retirees. The impact of the rules also would be greatly reduced on retirees who have pension and retirement assets of modestly more than $50,000. This approach could eliminate the need for most retirees to be concerned about the minimum distribution rules and would do so without creating powerful incentives to use retirement tax preferences primarily as estate-building mechanisms.
Moreover, the approach taken in Portman-Cardin — delaying retired from 70½ to 75 the age at which mandatory distributions must begin if the worker is already — is problematic for a number of reasons.
- Such a delay in the age at which distributions from pension plans must begin being made would provide a significant tax benefit to high-income individuals who have sufficient assets or income to enable them to delay withdrawals from their pensions and IRAs past age 70½. The vast majority of American workers retire before age 70½ and need to begin withdrawing funds from their pensions before then. Thus, for the vast majority of workers, the minimum distribution rules simply are not relevant, either because these workers lack retirement assets or because they will have begun taking regular distributions from their pensions well before the age by which distributions must begin.
- Raising the required age consequently would primarily affect high-income households that have sufficient other income and assets to delay withdrawals from their tax-preferred pension accounts, and would significantly expand the potential for such households to use their tax-preferred retirement accounts as estate planning devices.
- Raising the required age for minimum distributions also could discourage work among high-income elderly individuals. Currently, an affluent individual aged 72, for example, needs to continue working if he or she is intent on not withdrawing any funds from a 401(k), since the rules requiring distributions to start at age 70½ do not apply if the individual remains employed. The Portman-Cardin bill would enable such individuals to retire without having to make any withdrawals from their 401(k)s until age 75.
The Portman-Cardin pension legislation would make numerous changes in tax provisions governing retirement saving. Some of these changes are beneficial, but the bulk of the revenue loss from the legislation results from provisions that are problematic. This analysis focuses on those provisions.
According to Rep. Portman, the Portman-Cardin bill will reduce revenue by more than $100 billion over the next 10 years. As this analysis indicates, many of the most expensive provisions involve increased or expanded opportunities for high-income households to shelter savings from taxation. Such tax sheltering does not generate increases in private saving; instead, it shifts assets from taxable accounts to tax-preferred accounts.
Pension reform should instead be focused on expanding tax incentives for lower- and moderate-income earners to save for retirement. Contributions to tax-preferred retirement accounts by such workers are more likely to represent new saving, rather than asset shifting. They also are much more likely to reduce the risk of living in poverty during retirement.
Many of the principal provisions of the Portman-Cardin legislation consequently move in the wrong direction. Given the troubling fiscal outlook facing the nation, the case for the bill’s expanded tax subsidies for higher-income households thus is weak. The Portman-Cardin bill is an example of the type of attractive-sounding measure that the nation now can ill afford and that policymakers must begin summoning the courage to resist if there is to be much hope of bringing our fiscal house in order.
 Peter R. Orszag is the Joseph A. Pechman Senior Fellow in Economic Studies at the Brookings Institution. The views expressed here do not necessarily represent those of the staff, officers, or directors of the Brookings Institution. The author thanks William Gale, Robert Greenstein, and Mark Iwry for the joint work upon which much of this note is based.
 Representative Portman indicated the legislation would cost $112 billion over 10 years. See National Journal’s Congress Daily, April 11, 2003.
 For the empirical evidence on these points and for a discussion of proposals for progressive pension reform, see William G. Gale and Peter R. Orszag, "Private Pensions: Issues and Options," in H. Aaron, J. Lindsay, and P. Nivola, eds., Agenda for the Nation (Brookings: forthcoming), and Peter Orszag and Robert Greenstein, "Toward Progressive Pensions: A Summary of the U.S. Pension System and Proposals for Reform," in M. Sherraden and L. Morris, eds., Inclusion in the American Dream: Assets, Poverty, and Public Policy (Oxford University Press: forthcoming).
 For further discussion of this point in the context of the Administration’s proposal for expanded tax-free savings accounts, see Leonard Burman, William Gale, and Peter Orszag, "The Administration’s Saving Proposals: A Preliminary Analysis," Tax Notes, March 3, 2003.
 Robert Carroll, "IRAs and the Tax Reform Act of 1997," Office of Tax Analysis, Department of the Treasury, January 2000.
 General Accounting Office, "Private Pensions: Issues of Coverage and Increasing Contribution Limits for Defined Contribution Plans," GAO-01-846, September 2001. The GAO also found that 85 percent of those who would benefit from an increase in the 401(k) contribution limit earn more than $75,000. (These figures reflect the effects of other changes included in EGTRRA that have already taken effect, such as the elimination of the previous percentage cap on the amount of combined employer-employee contributions that can be made to defined contribution plans.)
 See, for example, Craig Copeland, "IRA Assets and Characteristics of IRA Owners," EBRI Notes, December 2002, and David Richardson and David Joulfaian, "Who Takes Advantage of Tax-Deferred Saving Programs? Evidence from Federal Income Tax Data," Office of Tax Analysis, US Treasury Department, 2001.
 To the extent that Portman-Cardin results purely in such asset shifting, it does not have this adverse short-term macroeconomic effect since it does not raise saving in the short run. This component of the legislation is thus misguided short-term macroeconomic policy, misguided pension policy, or some combination thereof. Despite the claims of its proponents, the legislation is unlikely to raise private saving significantly, so its principal shortcoming is that it fails as pension policy.
 See Peter R. Orszag, "The Retirement Savings Components Of Last Year’s Tax Bill: Why It Is Premature To Make Them Permanent," Center on Budget and Policy Priorities, September 18, 2002.
 See, for example, Bridget O’Brian, "Older Investors Miss the Chance to Catch Up in Retirement Plans," Wall Street Journal, September 13, 2002.
 The AMT figure for 2011 assumes that the 2001 tax cut is extended.
 Mark J. Warshawsky, "Further Reform of Minimum Distribution Requirements for Retirement Plans," Tax Notes, April 9, 2001.
 The rules for distributions from traditional IRAs are slightly different. Distributions from IRAs are required to begin by age 70 ½ regardless of whether the owner is retired. No minimum distribution rules apply to Roth IRAs until the death of the owner.
 Jay A. Soled and Bruce A. Wolk, "The Minimum Distribution Rules and Their Critical Role in Controlling the Floodgates of Qualified Plan Wealth," Brigham Young University Law Review, 2000, page 616.
 See Internal Revenue Bulletin (2001-11, page 865) on March 12, 2001.
 The typical retirement age – that is, the age at which half of men are no longer in the labor force -- is approximately 63. See Gary Burtless and Joseph Quinn, "Retirement Trends and Policies to Encourage Work Among Older Americans," Brookings Economic Papers, January 1, 2000.