Savings Group Says Tax Reform Should Build on Current System

Our nation's employer-based retirement savings system is particularly vulnerable to disruption, because some of the commonly used measures of employer plan coverage are misleading and the true value of the current tax incentives for retirement savings is not widely understood, which can lead to proposals that result in more harm than good.

The American Benefits Council is a public policy organization representing principally Fortune 500 companies and other organizations that assist employers of all sizes in providing benefits to employees. Collectively, the Council's members either sponsor directly, or provide services to, retirement and health plans that cover more than 100 million Americans.

Our employer members believe in and support the continued vitality of voluntary, employer-sponsored retirement savings plans that provide an increasingly important role in the economic security of workers. This statement illustrates how the vast majority of American workers are well served by the system of tax incentives that makes these plans possible. We will:

• Describe the critical role that employers play in providing their employees with a secure vehicle for retirement savings.

• Demonstrate the true value of tax-deferred retirement savings on personal financial security, broader economic growth and future tax revenues.

• Explain how retirement plan coverage and tax policy are intricately intertwined.

• Provide targeted policy recommendations that would improve outcomes without compromising the successes of our current system.

• Offer additional analyses regarding the way in which the tax incentives for savings in employer-sponsored retirement plans are measured and evaluated and offer some insights into how best to interpret retirement plan coverage statistics.

EMPLOYERS PLAY A CRITICAL ROLE IN ENSURING WORKERS' RETIREMENT SECURITY

Voluntary, employer-sponsored defined contribution and defined benefit retirement plans are the foundation of our nation's retirement system. Workplace retirement plans, like those sponsored and administered by the Council's members, successfully assist tens of millions of families in accumulating retirement savings, allowing for a more financially secure retirement and providing sustainable health and financial well-being. The latest available data from the U.S. Department of Labor indicates that more than 130 million active and retired workers (and their spouses) are now covered by nearly 900,000 employer-sponsored retirement plans nationwide.¹ In addition, 46 million households now own some kind of Individual Retirement Account (IRA),² many comprised of assets that have been "rolled over" from employer plans.³ The role of employers in ensuring workers' retirement security, therefore, must not be underestimated.

Employer-sponsored plans provide tangible economic value to American workers saving for retirement. Employers lower administrative costs by achieving economies of scale and simplify participation by offering workers a carefully vetted selection of investment choices, while fiduciary requirements offer security and peace of mind. Simply put, employer retirement plans lower the cost of savings by reducing numerous burdens for individuals.

Payroll deduction has proven to be an effective means of enhancing savings behavior. By pooling resources and offering a collective vehicle for the investment of savings, employers are also able to foster a culture of saving throughout the workplace.

The voluntary, employer-sponsored system is important because it gives companies the flexibility to tailor their plans to diverse and evolving employee populations. Employers have a unique understanding of the retirement needs of their employee population and provide innovative solutions to help employees. For example, companies are increasingly providing their employees with access to education that enhances their understanding of savings principles and helps address the financial stresses that impair overall financial well-being. In the future, successful employee benefits systems will reject "one-size-fits-all" mandates in favor of a flexible approach that allows employees to adjust their goals and behavior according to their own changing needs.

To meet this challenge, employers are continually innovating to improve participation rates and outcomes. Bipartisan legislation has brought many of these innovations into the mainstream. For example, the Pension Protection Act of 2006 (PPA) includes several landmark clarifications to the defined contribution plan rules that encourage voluntary automatic enrollment and automatic contribution escalation, which are improving retirement savings by moderate- and lower-income workers beyond what could otherwise be expected. According to a report by Vanguard, more than 60 percent of new plan entrants in 2013 were enrolled via automatic enrollment and more than half of all contributing participants in 2013 were in plans with automatic enrollment.⁴

It is a testament to the effectiveness of these programs that these innovations have been emulated around the world in both public and private pension systems. Later in this document, we suggest some new ways to build on the employer-sponsored system, but such measures will only be effective if the system remains viable.

Employers have an interest in the continued success of the system because they also enjoy certain advantages of plan sponsorship. Having a strong retirement plan benefits employers by helping them attract and retain talent and by providing their employees with financial security and confidence as they prepare for retirement.

CURRENT TAX INCENTIVES ARE ESSENTIAL TO OUR SUCCESSFUL RETIREMENT SAVINGS SYSTEM

As we stated in our recent strategic plan document, A 2020 Vision: Flexibility and the Future of Employee Benefits, enlightened, bipartisan tax policy has been indispensable to the success of the employer-sponsored benefits system. It has enabled employers to make the commitment to design and fund benefit plans and has encouraged individuals to seek personal and financial well-being for themselves and their family members through participation in those plans.

The U.S. successfully encourages plan sponsorship by employers by providing tax incentives for employers to sponsor plans that deliver meaningful benefits to Americans up and down the income scale. This tax structure also provides powerful motivation for individuals at all income levels to save for their retirement in employer-sponsored retirement plans.

Fundamental Principles of the Current Retirement Tax Incentives System

The fundamental principles of the current retirement tax incentives are straightforward and clear:

• Contributions are excludable or deductible from income -- until withdrawn. Savings in retirement plans are tax-deferred; that is, the contributions and investment earnings accumulate tax-free until the participant receives a benefit payment -- often decades down the line.

• Distributions are taxed as ordinary income. The long-term investment gains within a plan are subject to the higher ordinary income tax rates, rather than capital gains rates.

• Employer contributions are exempt from payroll tax. Because employer contributions to plans are not regarded as wages, neither employees nor employers owe payroll taxes on these amounts. As a result, they do not count toward earnings in the calculation of Social Security benefits.

• Taxes on investment gains are deferred while funds remain inside a retirement plan. This deferral is critical for encouraging savings, as workers know they will not have to divert income year-by-year to pay tax on their retirement savings and can therefore build up a retirement "nest egg" much more rapidly.

• The Saver's Credit supplements exclusion/deduction. The Saver's Credit, which provides a credit to low- and middle-income individuals who contribute to defined contribution plans and IRAs, provides a robust savings tax incentive for lower-income individuals who are below the taxable income thresholds and would therefore be less motivated by the exclusion of contributions.

• Contributions are limited and rules promote fairness. Individual contributions to defined contribution plans and IRAs are subject to coverage, participation and nondiscrimination rules to ensure that they are fairly distributed among plan participants.

The tax incentives for retirement plans also constitute a crucial factor in companies' decisions to establish and keep 401(k) and similar plans. Roughly eight in ten employers recently said that the deferral/exclusion of employee and employer contributions from income taxation are important in their company's decision to sponsor a defined contribution plan.⁶

Tax Incentives as 'Expenditures': A Flawed Comparison

As the Committee considers the economic justification for these incentives, it is critical to remember that retirement savings incentives in the tax code are deferrals of the income tax that would otherwise be paid on contributions, not exclusions, as is the case with the other types of "tax expenditures" with which they are often erroneously compared.

Under current tax expenditure methodology, these retirement savings deferrals of tax are in many cases scored as an immediate federal revenue loss, which in turn distorts any comparison to other tax expenditures. That distortion should not be allowed to influence tax policy discussions.

The most recent estimates provided by the Joint Committee on Taxation (JCT) categorize contributions to qualified employer-based retirement plans, as well as the investment earnings within those plans, as "tax expenditures" totaling $112.3 billion in Fiscal Year 2015, with a total of $699.4 billion for the five-year period from 2014 to 2018.⁷ The U.S. Treasury Department and Office of Management and Budget (OMB) projection published with the President's Fiscal Year 2016 budget, which uses somewhat different economic projections and measurement methods, results in nearly identical numbers of $112.7 billion in Fiscal Year 2015 for employment-based plans and project the totals out for a ten-year period through 2024 at more than $1.5 trillion.⁸

The valuation of these tax incentives, as conducted by Treasury and the JCT, is impossible to interpret correctly and is especially misleading because it is presented side-by-side with other, different exclusions from income, such as mortgage interest or state and local taxes. This is a dramatic misrepresentation of their true nature.

Unlike almost all other exclusions from income taxation, the majority of this tax revenue on contributions to retirement plans is not permanently "lost," but is rather deferred until it is received as benefits, when it is then taxed (along with any investment gains) as ordinary income.

Tax expenditures are typically measured on a "cash flow" basis that considers only the net of taxes foregone on contributions and earnings and those collected on benefit payments for the period.

     ( Taxes foregone    Taxes foregone)   Taxes collected

By failing to account for the future value of revenues outside of the measurement window, this methodology overstates the costs during periods when the employer-sponsored retirement system is approaching maturity or when large demographic cohorts approach retirement age, as is currently the case. This is because government revenue from taxation of withdrawal amounts increases dramatically shortly after the measurement window closes. In fact, a study completed by former staff of the JCT finds that the present value of tax benefits attributable to current-year retirement savings contributions is as much as 54 percent less than estimates of tax expenditures under Treasury's methodology.⁹

Similar distortions are reflected in the "revenue estimating" methodology used to determine the budgetary cost of changes in the retirement plan rules. The revenue scoring performed by the Treasury and the JCT generally produce estimates over five- to ten-year budget periods. Although they are measuring the costs associated with building up an asset base that will be taxed in the future, this measurement (like other revenue estimates) is based on a yearly cash-flow analysis. Under that methodology, the full cost of foregone taxes is realized in the current year, while the taxes an employee will pay when he or she retires and starts taking taxable plan distributions generally occur outside the measurement period This method, which may be appropriate for other types of tax incentives that are never recovered, overstates the true costs of the tax incentives for retirement savings.

As a result, proposals that reduce retirement savings today and raise revenue in the short term will mean the government actually collects less revenue in years outside the budget or scoring window, because retirees will have less taxable retirement income.

These and other issues addressed below are discussed in greater detail in Appendix A: A Better Understanding of the Numbers.

Retirement Plan Incentives Are Not Regressive

Some also suggest that the retirement savings tax incentives are "upside-down" or "regressive," allegedly giving higher-income individuals a disproportionate tax benefit when compared with lower-paid individuals. But the analysis that supports this perception is narrow and misleading, especially when derived from the flawed tax expenditure measurement described above.

While the value of the (temporary) exclusion may be greater for higher-income groups in any given year, this does not provide the full picture of how the overall retirement and tax systems operate for a variety of reasons. It is true that higher-income individuals contribute more to retirement plans and, because the overall value of the initial exclusion is proportional to income tax rates, receive a higher total value of the tax expenditure in any particular year. It is important to keep in mind, however, that the distribution of retirement savings tax expenditures is somewhat less weighted toward the higher-income segments than is the overall value of income taxes paid; stated differently, upper-income tax payers pay a greater share of income taxes than the share of retirement savings incentives they receive.

As described in more detail in Appendix A: Tax Incentives for Retirement Savings: A Better Understanding of the Numbers, the total dollar value of tax expenditures (and, to a significant though lesser extent, the present value of the tax incentive) in relation to income requires context to fully appreciate.

The value of a given year's tax exclusion is largely a function of the amount of an individual's income tax liability. The exclusion can have no value if there is no tax liability and has greater dollar value as more taxes are paid. Aggregate measures of the value of incentives, therefore, depend on the distribution of income taxes.

Because income taxes are progressive, the distribution of tax expenditures for retirement savings are in fact less concentrated in higher-income groups than is the distribution of income taxes paid. An alternative way of looking at this is to consider the distribution of tax expenditures as a percentage of the income for the group. By this measure, they are only somewhat tilted in favor of higher-income groups. Ultimately, the value of a dollar saved by a younger worker (who will accrue tax-free investment earnings over longer periods) or those of moderate incomes (who will move from having taxable income at a 15 percent rate while working to a zero marginal tax rate in retirement) is greater than that of a higher-income individual (who may move from a 33 percent to a 25 percent marginal tax rate).

Furthermore, looking at the income tax exclusion for employer-sponsored plans provides only a partial view of an individual's retirement benefit and ignores the progressivity elsewhere in the system. For instance, we should not look at income tax treatment in isolation but consider the effect of payroll taxes for retirement benefits, which do not apply to employer contributions to retirement plans. This reduces payroll tax liability, producing a progressively greater benefit for lower-income workers.

The largest source of retirement income for American workers is the Social Security system. The Old Age and Survivor benefits from Social Security are progressive in their formula, providing higher income replacement rates and benefits in relation to contributions for low- and moderate-income workers than for higher-income workers. The greater benefits that inure to low- and moderate-income workers in relation to their lifetime payroll taxes effectively balances the complementary distribution of tax benefits from voluntary, employer-sponsored plans, with the net distribution often favoring lower-income individuals.

The Current Tax Incentives Help Smaller Employers Sponsor and Enhance Plans

Moreover, the existing tax structure imposes minimum coverage and eligibility requirements and limits on the value of benefits for higher earners. To qualify for the favorable tax treatment, (1) an employer plan must cover sufficient workers to meet longstanding or established criteria under the Internal Revenue Code and (2) the contributions and benefits for highly-compensated participants must be proportional to those for non-highly-paid individuals.

Part of the misunderstanding of retirement savings tax incentives is the mistaken perception that many high-income individuals are simply shifting savings to tax-qualified plans. This presumption ignores the fact that by establishing these plans, employers are bringing many millions of low- and moderate-wage earners and new retirement savings into a secure and efficient retirement income system.

As we will show, retirement plan coverage and take-up rates among employees at large companies are relatively high and stable. Therefore, gains in retirement plan coverage will need to come primarily from smaller plans. In this regard, coverage gains among smaller employers not only are critical from a social policy perspective, but also serve everyone's interest, because without broad enough coverage, concerns regarding the tax incentives and the private retirement system will increase.

The Council's public policy strategic plan includes a number of recommendations to improve small employer plan sponsorship, such as relaxing the rules to form multiple employer plans. But the first and best step is to maintain the current tax incentives, which not only promote plan formation, but also preserve coverage for the large middle class.

Retirement Tax Incentives Lead to Long-Term Economic Gains

The financial assets held in defined benefit and defined contribution plans and IRAs now total more than $17 trillion, with $3.2 trillion in defined benefit pension plans, $6.8 trillion in defined contribution plans and $7.4 trillion in IRAs.¹⁰ Clearly, the retirement tax incentives continue to be successful in encouraging plan participation and sponsorship. This tax policy is also a vitally important element of the greater economy -- with positive implications for future federal revenue.

These trillions of dollars in assets, which represent ownership of a significant share of the total pool of stocks and bonds, also provide an important and ready source of investment capital for American businesses.¹¹ As of the first quarter of 2014, the financial assets in defined contribution plans and IRA assets alone were collectively larger than the Gross Domestic Product of China.¹² This capital permits greater production of goods and services and enables additional investments that help companies grow, add jobs to their payrolls and raise employee wages.

In addition, analyses have also shown that the tax expenditure for retirement savings produces substantial tax revenue in the future. One simulation, using reasonable assumptions, found that the federal government would produce between $4.42 and $5.89 in revenue for every $1 of foregone federal taxes for contributions to a 401(k) plan as benefits were paid out over 30 years in retirement.¹³

The current tax incentives support plan sponsorship, plan participation (at all income levels), a healthy federal treasury and a healthy economy. In effect, proposals that reduce retirement savings would actually increase the burden on future generations.

RETIREMENT PLAN COVERAGE AND TAX POLICY ARE INTRICATELY INTERTWINED

Evaluation of the current tax incentives for retirement plans are frequently viewed in the context of today's coverage levels. If the tax incentive is operating efficiently, we should see robust rates of plan access and participation across multiple segments of the workforce. This is, in fact, what has been achieved.

A common misconception often put forward in conjunction with the tax expenditure numbers is that the current system has achieved coverage of less than one-half of the workforce, a number reported by the U.S. Department of Labor's Bureau of Labor Statistics (BLS) as part of its annual National Compensation Survey. In 2014, BLS reported that 65 percent of private-sector workers had access to a workplace retirement plan, with 48 percent participating (for a "take-up" rate of 75 percent).¹⁴

As described extensively in Appendix B: Putting Private Pension Coverage Numbers Into Perspective, these figures¹⁵ significantly understate the extent of benefits received from employer plans, because point-in-time measures do not account for the accumulation of benefits for households and over a full working life.

Digging deeper into the BLS figures, we find that coverage gaps can be found primarily in certain segments of the workforce. Most workers in larger firms (with more than 500 employees) and those in the upper half of the income distribution are consistently covered. Part-time, lower-income and younger workers in smaller firms have much lower participation rates:

                                             Access    Participation    Take-up

In a voluntary system, pension coverage is in part a process of supply and demand in the labor markets. Younger and lower-paid individuals have less capacity, less ability and less interest in retirement savings as they seek to pay off student loans, start a family and buy a home. As they transition into higher-paying jobs later in their careers, many will move into jobs with retirement coverage.

If we look at households nearing retirement, we find that nearly two-thirds have accumulated some type of retirement savings account.¹⁶ Those who do not are typically within the lowest income groups, which face many other, more immediate financial needs and will receive lifetime income replacement rates from Social Security similar to those with workplace coverage. A recent study by the Investment Company Institute on the dynamics of private pension coverage concluded that three quarters of workers likely to have the desire to save for retirement had access to an employer-sponsored retirement plan and 93 percent of these workers were participating in the plan.¹⁷ Younger and lower-income workers that are offered the chance to participate in an employer-sponsored plan are far more likely to be saving for retirement than otherwise similarly situated individuals.

The clear differences in these underserved populations demonstrate that the current system is actually working well for the majority of workers. The incremental changes adopted in recent years have made them even more effective, and continued innovation and policy improvements can make the system even stronger. Any public policy reforms, including changes to retirement savings tax policy, should focus exclusively on addressing these coverage gaps rather than disrupting the gains made -- and yet to be made -- by plan participants.

In focusing on these coverage gaps, removing impediments to plan sponsorship should be the first step. The most significant barriers to coverage for lower-wage and part-time workers are the burden and cost of plan administration, often coupled with the complexity of fiduciary responsibility.

The current federal regime of mandatory disclosures, filings and recordkeeping makes plan sponsorship a non-starter for smaller employers -- or larger employers with transitory, seasonal or contingent workers. We believe that simply streamlining the number of retirement plan information disclosure requirements would improve plan formation and access, obviating the need for more dramatic policy changes. The promotion of multiple employer retirement plans would alleviate some of these administrative pressures. This employee population might also be well served by increased portability of retirement plan vehicles, so they wouldn't have to start over each time they change jobs (see recommendation No. 5 below, under "Changes in Retirement Policy Should Build on the Current System").

Many part-time workers have had access to a retirement savings plan at a previous employer, have a spouse accruing retirement benefits at work, will accumulate retirement resources in the future or can save in a tax-advantaged IRA. It is also important to note that this population may well consist of individuals at the very beginning or end of their working lives. While younger workers may be disinterested in retirement planning at the beginning of their careers (or may be saddled with student debt) and older workers may already have accumulated sufficient savings for their retirement, existing programs like the Saver's Credit and the Catch-Up Contribution program can be enhanced to provide additional opportunities for these workers to start (or finish) saving.

For consistently lower-income workers in particular, the most effective and urgent course of action would be strengthening the Social Security system as a foundation of retirement income. Social Security provides retirement income for almost all working Americans, with higher replacement rates for lower-earning households (replacing 70 percent of lifetime earnings for the lowest-income quintile and 47 percent in the second quintile).¹⁸

For most other employees -- the overwhelming majority of all American workers -- employer-sponsored retirement savings are an essential element of their financial future. In a recent study, 90 percent of those with an opinion had favorable impressions of 401(k) plans, with 39 percent saying that they had a "very favorable" impression, and eight in ten households with plans expressed confidence that defined contribution plan accounts could help participants reach their retirement savings goals. Nearly half of defined contribution plan-owning households stated that they probably would not be saving for retirement at all if it were not for their employer-sponsored plans.¹⁹

CHANGES IN RETIREMENT POLICY SHOULD BUILD ON THE EXISTING SYSTEM

Promoting retirement savings must remain one of our nation's top policy priorities. Any changes made should preserve and build upon our existing and successful tax incentive structure so it works even more effectively to facilitate retirement plan coverage and savings for American families.

Harnessing behavioral economics, through features like automatic enrollment and automatic escalation, has proven to be an effective approach for improving outcomes for many employee populations. In many cases, however, additional action from lawmakers will be required.

The Council recently issued a long-term public policy strategic plan, A 2020 Vision: Flexibility and the Future of Employee Benefits, in which we made specific policy recommendations, many of which would empower individuals to save for a secure financial future in retirement. The recommendations include the following:

These recommendations and others incorporated in A 2020 Vision are intended to empower individuals to achieve health and financial well-being in retirement. Our member companies sponsor retirement plans with strong participation levels and we want to ensure that American workers tap into the full value of their retirement plans.

CONCLUSION

As Congress considers retirement issues (in a tax reform context or otherwise), its first priority should be helping individuals and employers generate retirement income sufficient for employees to maintain their standard of living. Too often, retirement legislation is driven by considerations unrelated to retirement policy -- principally to raise revenue to pay for other initiatives. When this happens the result has almost always been unnecessary complexity and costs -- and, worse yet, direct harm to Americans' financial well-being and retirement prospects.

Proposals that purport to raise revenue or offset spending by redirecting, eliminating or eroding the existing retirement savings incentives achieve those additional taxes largely because individuals save less for retirement. Making matters worse, any short-term revenue gain that might be derived from changes in the retirement savings incentives is largely illusory, because when a worker saves less money today, it will mean smaller distributions (and less tax revenue) when the individual retires. That is a lose-lose situation for the retiree and the government.

Eliminating or curtailing the current tax incentives on retirement plans would result in American workers saving less for retirement when we need to encourage them to save more.

A financially secure retirement is the goal for every working American -- but it is also more than that. Creating financially secure retirees should be a goal of U.S. tax policy, because it is an effective, efficient way to promote economic success.

Financially secure retirees are less dependent on federal social safety net programs such as Social Security and Medicare.

Financially secure retirees are able to contribute to the economy as volunteers, entrepreneurs, care providers, tourists and consumers.

Financially secure retirees are better able to withstand the increased expenses of aging and longevity, including higher health care costs.

Just as importantly, financially secure retirees are best capable of enjoying the rewards of their labor.

The employer-sponsored retirement system is the factory that produces financially secure retirees. We stand ready to assist the government in keeping those factory doors open -- to ensure support for Americans as they undertake the substantial challenge of increasing retirement preparedness.

Submitted by:

² Investment Company Institute, 2014 Investment Company Fact Book, Figure 7.15, May 14, 2014.

³ According to the Investment Company Institute (ICI) Fact Book (id, Figure 17.7), "of U.S. households owning traditional IRAs in May 2013, 49 percent (or nearly 18 million) had traditional IRAs that included rollover assets."

⁴ Vanguard, How America Saves 2014, June 13, 2014.

⁵ Investment Company Institute, Americans' Views on Defined Contribution Plan Saving, January 2015.

⁶ American Benefits Institute, Mathew Greenwald & Associates Inc., Attitudes of Employee Benefits Decision Makers Toward Retirement Plan Tax Proposals, 2012.

⁷ Joint Committee on Taxation, Estimates of Federal Tax Expenditures for Fiscal Years 2014-2018 (JCX-97-14), August 5, 2014. Includes defined benefit plans, defined contribution plans and plans covering partners and sole proprietors (sometimes referred to as "Keogh plans").

⁸ Office of Management and Budget, Analytical Perspectives, Budget of the United States Government, Fiscal Year 2016, February 2015.

⁹ American Society of Pension Professionals & Actuaries, Retirement Savings and Tax Expenditure Estimates, April 2012.

¹⁰ Investment Company Institute, The U.S. Retirement Market, Fourth Quarter 2014, March 2015. As noted previously, a significant portion of IRA assets is attributable to rollovers from employer plans.

¹¹ At year-end 2013, 51 percent of mutual funds' assets were held in a tax-deferred household account. See Investment Company Institute, 2014 Investment Company Fact Book, Figure A.3.

¹² The World Bank, China, 2015.

¹³ Principal Financial Group, Tax Implication Analysis of Worker Defined Contribution Savings, 2015.

¹⁴ BLS, National Compensation Survey: Employee Benefits in the United States, March 2014.

¹⁵ The Current Population Survey (CPS), produced jointly by the U.S. Census Bureau and the BLS, also produces survey data on workplace retirement plan coverage. The March 2014 CPS indicated that 34.4 percent of all private sector workers report participation in an employer-sponsored retirement plan.

Appendix B to this statement discusses this figure, in comparison to the NCS figures, in more depth. Both the CPS and NCS figures do not distinguish between traditional defined benefit pension plans and defined contribution arrangements such as 401(k) plans.

¹⁶ Board of Governors of the Federal Reserve System, Survey of Consumer Finances, 2014.

¹⁷ Investment Company Institute, Who Gets Retirement Plans and Why, 2013, October 2014.

¹⁸ Congressional Budget Office, The Long-Term Projections for Social Security: Additional Information, Exhibit 10, October 2012.

¹⁹ Investment Company Institute, Americans' Views on Defined Contribution Plan Saving, January 2015.

²⁰ U.S. Centers for Disease Control, Health: United States, 2013, 2013; the Organisation for Economic Cooperation and Development, Statistics on average effective age and official age of retirement in OECD countries, 2012.

A perception advanced by some proponents of reform is that the tax expenditure for retirement savings is (1) excessively expensive, representing the second-largest subsidy provided through the tax system, and (2) disproportionately provided to higher-income groups. This perception often becomes the rationale for proposals to reduce or eliminate the deferral of income taxes on qualified retirement savings or restructure these into various forms of credits. Considering the merits of reform proposals, however, requires a more complete understanding of the nature of the current tax treatment and a more nuanced interpretation of the commonly cited numbers than is typically presented in the rationale for reform proposals.

WHAT ARE THE RETIREMENT SAVINGS TAX INCENTIVES?

Employer-sponsored retirement plans, within specific limits established under the tax code, are afforded favorable tax treatment to encourage employers to establish retirement plans and incent employees to save in those plans. In what are often termed "traditional" plans, allowable contributions (by both employers and workers) are excluded from taxable income in the year in which they are made, and earnings on investment of these funds are not taxed while they remain in the retirement system.

Income tax on the accumulated savings is paid when the funds are distributed to the individual, primarily following retirement. Alternatively, under the so-called "Roth" plans, amounts contributed are included in taxable income but the investment earnings and later distribution of the accumulated funds is not taxable income.¹

This tax treatment is therefore not the exclusion of retirement savings from taxation but rather the deferral of taxes on the funds contributed to the traditional plans and exclusion of taxes on investment earnings in a Roth plan, in comparison to what would have occurred if the savings and investment were outside of the tax-qualified retirement plan.

While the two tax regimes (traditional and Roth) appear to be very different, they are in the final outcome closely equivalent. The tax paid in the initial year for Roth plan contributions results in a lower amount in the account and therefore lesser earnings that typically will be roughly equal to the future tax liability for the "traditional" plan.² Therefore, the incentive (or subsidy) provided is the tax-free earnings of the invested savings that is essentially the same for both arrangements. Since the employee's compensation income is eventually taxed in either of these regimes, both provide a comparable incentive to save within the retirement system because, if the savings were in another form, the accumulated investment earnings would be subject to taxation. By establishing substantially higher allowable amounts in employer-sponsored retirement plans than in Individual Retirement Accounts (IRAs), this tax treatment provides a strong incentive for employers to sponsor plans and for workers to participate in these plans.

Significantly, the tax treatment of retirement savings is fundamentally different from other types of "tax expenditures" with which it is often lumped together in discussions of tax policy or tax reform. In contrast to the exclusion of mortgage interest and amounts paid for state and local taxes from federal income tax, where the value of the payment is excluded in the year in which it is made and never recovered (a true tax expenditure), taxes are simply deferred for traditional plans or the taxes are paid for a Roth plan.³

Because the value of the private retirement system's tax treatment is in the exclusion of investment earnings, for any individual, the tax incentive is primarily a function of the amount of the contributions and the number of years in which they remain in the plan. For a dollar of contributions to a retirement plan, the tax incentive is therefore relatively greater for younger workers who will accrue tax-free investment earnings over longer periods and for those of moderate incomes who will move from having taxable income at a 15 percent rate while working to a zero marginal tax rate in retirement than for a higher-income individual moving from a 33 percent to a 25 percent marginal tax rate.

WHAT ARE THE "TAX EXPENDITURE" NUMBERS FOR RETIREMENT PLANS?

The Joint Committee on Taxation (JCT) and the Department of Treasury issue official estimates of the value of the full range of what are termed "Tax Expenditures." Treasury's estimates are included in the President's budget developed through the Office of Management and Budget (OMB). The term "tax expenditures" is a very specific type of measure that has been developed for a much broader application and is therefore not particularly relevant to the tax incentives provided to retirement plans.

The concept was developed in the 1960s and early 70s in an effort to measure what were perceived as hidden flows of economic value through the federal government. The largest of these are potential tax revenues that were foregone through various provisions in the Tax Code. These were not previously measured and considered in the budgeting process but were seen to be equivalent to direct spending on government programs.

Tax Expenditures are defined as the increased tax revenues that would be received if the laws allowing specific exclusions were to be repealed for the year of measurement. For retirement plans this is the value of taxes that would have been imposed had all of the contributions been included in the taxable income of individuals. It also includes the value of taxes that would be applied to the investment earnings on the accumulated assets held within the plans which, as the assets of pension funds now amount to more than $17 trillion, are about three fourths of the value of the tax expenditure. The value of these foregone taxes is then reduced by the value of taxes collected on distributions from the plans to produce the tax expenditure. This net of tax flows associated with the plans for the specific year is effectively a cash-based accounting approach to the measurement of the value of the tax treatment.

The most recent JCT numbers published in August of 2014 indicate a tax expenditure of $112.3 billion in Fiscal Year 2015 for employment based retirement plans (including Keogh plans) with a total of $699.4 billion for the five year period from 2014 to 2018.⁴ The Treasury/OMB numbers published with the President's FY 2016 budget in February 2015, which use somewhat different economic projections and measurement methods, result in nearly identical numbers of $112.7 billion in FY 2015 for employment based plans and project the totals out for a ten-year period through 2024 at a total of $1,538.4 billion⁵.

WHAT ARE THE CHALLENGES IN INTERPRETING THESE TAX EXPENDITURE NUMBERS?

There are a number of significant issues with both the use of this measure of tax expenditures in relation to the other major categories as well as the application of the accounting measure (cash flow for the year) used in their calculation.

Retirement system incentives are a deferral of tax.

Other major categories of tax expenditure involve forms of economic value that can be fairly accurately measured by periodic cash flows. Although there is some insurance value over the longer term in enrollment in a health plan for an individual because payments in a given year may, partially reflect the value of benefits that will be received over the lifetime, in the aggregate, health care premiums represent the true value of benefits for the period. The economic value of mortgage interest and state taxes are nearly perfectly aligned with the annual cash flows. There is no meaningful future recapture of income and tax revenue from these other exclusions to consider.

In contrast, annual cash flows for employer-sponsored retirement plans are not well aligned with the longer term value of benefit received and are significantly affected by the stage of development of the private system and underlying demographics. Until the private pension system becomes fully mature over multiple generations, annual contributions to pension plans will far exceed benefit payment in any particular year. Annual differences between the flows is accentuated by demographic variations, the most relevant example being the progression of unusually large age cohorts like the "baby boom" generation through the population. With the leading edge of baby boomers just now reaching retirement age this cohort remains largely in the accumulation rather than payout phase of retirement savings.

The underlying demographic trend is exacerbated by two other factors. Coverage within the private pension system only became widespread following the second World War and expanded rapidly up through the early 1970s. The system is just now approaching the point when the higher coverage cohorts are reaching retirement age. This maturing of the system results in higher contribution flows into the system in relation to benefits payments.

In addition, contributions to retirement savings naturally tend to increase as individuals approach retirement age, in part because the value of the tax exclusion on earnings is less over shorter time periods, but also because income and the capacity to save peak in the decade before retirement age. The 401(k)-type savings arrangements, now only 30 years after development, have become the dominant form of retirement savings. The rapid decline of defined benefit coverage and its replacement by these and other defined contribution plans in the 1980s shifted contributions to later years in the life cycle of participants, a process that is accentuated by the introduction of "catch-up" contributions in 2001. When a large group such as the baby boomers passes through prime saving years, as is now happening, contributions are high in comparison to benefit received by the smaller earlier cohorts. Both of these factors result in very high contributions relative to benefits in recent years, a relationship that will continue for at least another decade.

Therefore, because they do not account for the future recapture of the tax exclusion of contributions or consider the stage of development of the private pension system or the significant impact of demographic trends, the tax expenditure measures do not reflect well the economic value of current tax incentives. The treatment in these measures is analogous to utilizing cash flow accounting in the consideration of capital expenditures. While consistent with other budget measures and an approach that makes sense for the other major categories of tax expenditures, measuring retirement savings in this manner is inconsistent with how they would be treated in any other context. This makes any comparisons between the retirement savings incentive and other tax expenditures very misleading.

Tax expenditure estimates do not reflect behavioral responses.

The assumptions implicit in the way in which tax expenditure estimates are made also make them flawed measures of the costs of the incentives to save in tax qualified retirement plans. The estimates are based on a "counter-factual" methodology (cash flows estimated to occur if the relevant tax law is repealed and "ordinary" income tax treatment applied) that incorporates several assumptions that significantly overstate the true value of what tax revenues would be in the absence of the existing legal framework.

Critically, the tax expenditure estimates do not reflect behavioral responses by employers and employees that are certain to occur. The methodology assumes that all of the observed flows into retirement savings would simply be transformed into cash income for the participants in the current year and taxed as ordinary income. While a common simplifying assumption, it is not at all clear that employers would increase cash wages dollar for dollar if they were not making contributions to employer-sponsored pension plans. Were they to keep total compensation equivalent they would be just as likely to increase expenditures for other elements of the benefits package that are tax preferred, which would be more advantageous from a net compensation perspective of the employees. This could result in no net change in revenues for the year. In some instances there would likely not be a full offset of elements of the cost of benefits (such as matching contributions), resulting in higher net earnings for plan sponsors that would be taxed at lower corporate rather than individual rates.

Similarly, these measures assume that behavior, i.e., the propensity to save for retirement, would remain unchanged in the absence of the existing tax treatment. While there is no consensus on the degree to which net savings is influenced by the current ability to exclude contributions to employer-sponsored plans from current income (or, stated differently, the extent to which tax qualified savings substitutes for savings that otherwise would have occurred anyway), the relevant research indicates that there would not be a full substitution of savings especially among lower-income groups.

The tax expenditure measure also assumes that investment gains would be taxed as ordinary income rather than (as noted earlier) effectively excluded from any taxation. In fact, if -- as assumed -- savings behavior remains constant, much of the investment earnings would be transformed into long-term capital gains. For lower-income earners who otherwise might pay a 15 percent marginal income tax, a zero capital gains tax rate would be applied resulting in no change in tax revenues. For higher-income individuals the relevant long-term capital gains rate would be 20 percent rather than the 25 to 39.6 percent assumed in the estimates.

Alternative tax expenditure estimates exhibit other flaws.

In an effort to address some of these limitations of the current year tax expenditure approach to measuring incentives, and to provide an alternative measure that is more consistent with the nature of retirement savings, the Department of Treasury and OMB also publish a "present value" measure of the tax incentive for the current year. This utilizes the same observed level of contributions that are excluded from income tax in the year as the tax expenditures. This approach carries the value of these forward to payment of the associated benefits, applying an assumption regarding foregone taxation of earnings across this period, and then seeks to estimate the value of the taxes that would be received in future years. These flows are then measured in present value by discounting them to the present to account for the time value of money. The most recent numbers in the President's FY 2016 budget estimates the present value of employer-sponsored retirement plans for FY 2014 to be $91.2 billion or about 15 percent less than OMB's $105.3 billion tax expenditure estimate for the same period. A similar analysis by the Congressional Budget Office (CBO) estimates that the present value of retirement savings in 2013 would be about 10 percent less on a present value basis than for its calculation using the tax expenditure method.⁶

This approach is conceptually better aligned with the economic effects of the tax incentive by seeking to capture the full set of flows associated with the retirement savings for a particular year. However, it exhibits different flaws. Most notably these alternative tax expenditure estimates are very sensitive to the assumptions about rates of return on investment and discount rates. OMB and Treasury (like JCT and CBO) do not reveal all of the assumptions from which the present value estimates are derived. Since the government operates on a cash flow deficit and borrows at Treasury rates, it follows that the relevant discount rate is the cost of this borrowing. Less obvious are assumptions regarding the rate of investment returns, allocation of returns between ordinary income and long-term capital gains and individual tax rates in retirement.

The OMB present value measures appear to utilize certain assumptions:

• the rate of return on savings will be equivalent to the discount rate,

• investment returns will be distributed roughly equally between interest, capital gains and dividends, and

• that tax payers will face individual marginal tax rate that are the same in retirement as while working.⁷

The most important of these are assumptions regarding investment returns and applicable income tax rates for retirees. Data on investment patterns of private pension funds indicate significant allocations to equity and other investments that have historically yielded long run returns well above Treasury interest rates. A higher return assumption would increase the foregone revenues during the accumulation period but increase the future tax revenue from retirees. On net, since taxes during the accumulation period would likely be at lower rates applicable to qualified dividends and long-term capital gains while benefit payouts are treated as ordinary income and taxed at higher rates, an increased rate of return would increase future income tax collections more than it would reduce foregone taxation of investment returns despite the differences in timing (the low relevant discount rate, especially today being an important factor) and therefore reduce the present value measure. Individuals also are likely to experience some decrease in income and therefore marginal tax rates as they move into retirement, though changes in marginal tax rates are likely to be relatively small for the income groups with the majority of pension savings and the affected cash flows are far into the future.

If one is to perceive the provision of tax incentives as the federal government financing foregone current revenues with borrowing in return for future tax collections from retirees (that is, effectively leveraging the purchase of a future cash flows at Treasury interest rates) it is perhaps most interesting to note that under some potentially plausible set of assumptions the net present value to the government of retirement tax incentives might even be positive for some years. Using today's cost of borrowing (30-year Treasuries now yield about three percent and a weighted average of outstanding Treasury debt instruments below two percent as reflected in the current yield on the TSP G fund) and historical rates of return on assets invested 50 percent in equities with the remainder in high grade corporate debt and other instruments would generate a considerable spread between the two rates. Historical yields would produce an asset base for future taxes that could exceed the present value of the borrowing costs in part due to these differences but also because of the difference between long-term capital gains and qualified dividend rates and the likely income tax rates paid by retirees. A large baby boom generation making high contributions in their 50s and early 60s and then being taxed at higher rates on their investment gains a decade later could produce revenues that are nearly equivalent to the value of tax exclusion for the contributions.

DISTRIBUTION OF THE VALUE OF TAX INCENTIVES FOR RETIREMENT SAVINGS

The other major area where there are significant challenges in interpreting the value of tax incentives for retirement savings is in the measurement of how they are spread across the income distribution. CBO periodically updates an analysis of the value of tax expenditures by income, with the latest version published in 2013.⁹ This analysis is based on the principles of tax expenditures that evaluate the net cash flows for this tax treatment with all of the attendant issues outlined above.

These estimates do not separate the various components of the tax expenditures to distinguish employer-sponsored plans from IRAs and other elements of the pension system so are not exactly comparable to the more specific numbers on tax expenditures. They do provide an indication of the overall distribution of tax expenditures. CBO's analysis concludes that about two-thirds of the total $140 billion in tax expenditures for "net pension contributions and earnings" in 2013 accrued to the top one fifth (quintile) of the income distribution. The next highest quintile received something close to a proportional share with 18 percent with the bottom three quintiles ranging from two percent for the lowest to nine percent for the middle quintile of income.

This distribution is somewhat more tilted in favor of the highest income quintile than the overall distribution of tax expenditures for the period, with 51 percent attributed to the top quintile and eight percent to the bottom quintile. The share received by the highest income group, however, is considerably less than the 81 percent resulting from deductions for mortgage interest, state and local taxes and charitable contributions. It is more evenly distributed than any of the other major categories of deductions or exclusions from income except the exclusion for employer-sponsored health insurance and capital gains transferred on death. The most evenly distributed category is employer-sponsored health insurance for which 34 percent is received by the highest income quintile and 8 percent for the lowest fifth of the income distribution with the middle three quintiles ranging from 14 percent to 26 percent.

                  CBO Estimated Distribution of Tax Expenditures:

    CBO Estimated Distribution of Tax Expenditures by Income Quintile: 2013

          Distribution of Tax Incentives for Employer-sponsored Plans

The numbers provided by these various methods of evaluating the distribution of retirement savings incentives provide some useful insights into the dynamics and outcomes of the current tax treatment. There are, however, a number of important considerations in their interpretation. These include:

Distribution in Relation to Overall Income Tax Liability

Although higher-income households are much more likely to participate in employer-sponsored retirement savings plans and able to make larger contributions, the progressive structure of the income tax system inevitably provides a value of the deferral of income tax and exclusion of investment earnings that is far greater for household facing a 39.6 percent marginal income tax rate than one at the zero or 15 percent marginal rates faced by the lower half of the income distribution. One simple way of putting this into perspective is to compare the value of total income taxes paid (taking into consideration the deductions) with the distribution of tax expenditures shown above. The CBO periodically estimates the distribution of federal taxes in relation to the same income quintiles by which tax expenditures are measured. The most recent version published in 2013 provides the distribution for 2010.¹² The CBO estimates for total income taxes paid in 2010 and the value of tax expenditures for pension contributions and earnings in 2013 (these are the most recent years for which CBO has published the relevant tables) are shown below. This indicates that the upper income quintiles pay a significantly greater portion of income taxes than the share of tax expenditures for retirement savings that are attributed to them. This shows that the highest income group receives a far lower share of the total value of tax expenditures that its share of individual income taxes while the other 80 percent of the income distribution receives a significantly greater share in relation the proportion of income taxes paid.

          Share of Income Taxes in Relation to Pension Tax Expenditures,

Also, the distribution of tax incentives estimates outlined above use point-in-time measures of income. As is well documented, incomes vary considerably for individuals over time, generally taking the classic "hump shape" life cycle distribution in which income gradually increases through the ages of 55 to 60 and then declines rapidly, largely because this is the point at which many people stop working. Higher-income groups are usually found to experience greater income differences over their working lives. The capacity and propensity to participate in retirement savings plans and make contributions is strongly correlated with both age and income, with rates for both increasing rapidly after the age of 35 as individuals transition from savings for the purchase of a home and the costs of raising children to saving for retirement -- and also sort themselves into jobs with private pension coverage. This leads to point-in-time measures of income that are not necessarily indicative of the relative position in the income distribution that an individual occupies over their lifetime. Therefore the concentration of tax benefits in the top quintile is influenced by the timing of contributions at peak income periods in the life cycle. A recent study of "permanent" (income averaged over the life cycle) and "transitory" (point-in-time) income found that the relationship between the tax benefits and income are somewhat less for the life cycle than a point-in-time measure.¹³ The extent of the difference, however, is likely muted because the study uses a present-value measure of tax benefits in which (as discussed above) the investment earnings rate is equivalent to the assumed discount rate which will understate the value for younger workers.

Of likely greater impact on the income related measures is the introduction of "catch-up" contribution rules that permit individuals over the age of 50 to make far larger tax preferred contributions to retirement plans. Another recent study found a significant increase in contributions to 401(k)-type plans as individuals passed the age of 50 and entered into their highest earning period.¹⁴ The combination of the demographic bulge of the baby boom generation now passing through peak earning years and the recent introduction of catch-up contributions significantly skews any measure of the distribution of tax benefits toward the highest income segments.

Interactions with Social Security Taxes and Benefits

The retirement savings incentives also should be viewed in the context of broader retirement policy, including especially Social Security. The retirement income system in the United States, to a greater extent than that of any other major developed country, is distributed relatively evenly among the three elements of (1) the Social Security System, (2) employer-sponsored pension benefits and (3) individual savings. In addition, after decades of decline, as a result of a wide variety of factors the proportion of older Americans who remain working has increased in recent years. The Social Security system utilizes a progressive benefit and financing structure that calculates benefits based on an average of lifetime earnings (adjusted to reflect the overall growth of earnings over time) and replaces income at rates that range from 90 percent for the first increments of earnings to 15 percent for the highest increments. This has the effect of providing much higher-income replacement rates for the lower half of the income distribution and no income replacement for the top 10 percent of income (that which is above the earnings cap for the payroll tax). This provides something close to 70 percent lifetime average income replacement rate for the bottom half of the distribution, representing a very positive net present value of benefits in relation to contributions. In contrast, higher-income groups will receive a negative return on their "investment" in Social Security and must accrue significant sources of private retirement income to achieve a similar replacement rate.

Tax incentives for retirement savings are therefore properly viewed only in a more holistic manner, by taking into account payroll taxes paid by employers and employees and Social Security benefits. A recent research project financed by the Social Security Administration analyzed the combined effects of Social Security and defined contribution retirement plans. The initial findings of this study presented in August 2014 found that "the size of the progressively distributed OASDI (Social Security) benefits swamp the size of the regressively distributed benefits from DC savings plans, making the combined system progressive. . . ."¹⁵

The Coverage Effects of Pension Tax Qualification Requirements

Perhaps the most important (and also the most difficult to measure) effect of the provision of tax incentives for retirement savings is the extent to which the capacity of higher-income groups to gain access to the much higher limits that apply to employer-sponsored plans has the effect of inducing retirement savings among the lower paid, an effect that is sometimes referred to as "crowding-in." Allowable annual tax deferred contributions to an employer-sponsored plan can be more than six times greater than what is permitted for Individual Retirement Accounts (IRA). To access these higher contribution limits for highly-compensated employees, employers must satisfy strict coverage, participation and other so-called "non-discrimination" rules that require broad coverage in a plan and limit differences in the share of income that can be contributed by higher paid and other participants in the plan. Efforts by employers to enable their higher-income workers to make greater tax preferred contributions by eliciting participation by low- and moderate-income workers draws many workers into the private pension system who would not otherwise be saving for retirement. Employers use various measures to encourage broad-based participation, including direct employer contributions, matching contributions, and automatic enrollment and escalation in the plan design that have become very prevalent in employer-sponsored plans. Studies of participation in retirement savings have consistently found that individuals working for an employer sponsoring a pension plan are more than twice as likely to be saving for retirement as workers with otherwise similar characteristics whose employer does not sponsor a plan. This is especially the case for younger workers, who benefit the most from the long periods for which they will accrue investment earnings tax-free, yet who show up in the distributional tables as low- and moderate-income earners.

Other Benefits to Lower-Income Workers

A related issue is the even more complex challenge of determining where the full costs of financing employer-sponsored retirement savings reside. Although in an accounting sense the contributions to private pension funds are roughly evenly divided among employers and employees, economists generally assume that the worker pays the full costs of any employer contributions through reduced cash wages. While comporting with standard economic theory, the evidence supporting this assumption is difficult to produce because it is impossible to determine what the wages for any individual might have been had their employer not provided pension coverage. The efforts to evaluate this "cost incidence" therefore rely on efforts to match the characteristics of workers and observed differences in the composition of compensation between those receiving employer-sponsored benefits and otherwise similar individuals who are not covered under a plan. A recent study of this issue found that there is a significant difference in these wage offsets depending on income with an additional employer contribution for high-income male workers offset by a 90 cent reduction in other compensation while a dollar of DC plan contribution for male workers in low-income households offsetting by only a 29 cent reduction.¹⁶ The disparity was even greater for women at 99 cents for high-income and 11 cents for low-income. This suggests that the tax qualification rules lead employers to subsidize the contributions of low-income workers to a degree that significantly compensates for the differential value of the tax exclusion. It also suggests that any changes in the retirement savings incentives could disproportionately harm lower-income workers.

CONCLUSIONS

The tax treatment of employer-sponsored retirement plans is an important but often poorly understood incentive that has been very effective in fostering a critical element of the retirement income system in the United States. Because they are in fact a deferral of taxes on savings rather than a full exclusion (as is the case for other types of tax expenditures) the long-term value of the tax treatment is in the exclusion of taxes on investment earnings. This makes the value of the tax incentive for any given dollar of retirement savings greater for younger workers who typically require the strongest incentives to save.

The way in which the costs of these tax incentives are evaluated and presented, done in the same way as other exclusions from taxable income (such as the mortgage interest deduction or the exclusion of state and local taxes), provides a flawed and misleading measure. Using cash accounting measures significantly overstates the long-term costs. Likewise, revenue estimates using a ten-year budget window overstate the budget impact of any reduction of these incentives. When viewed and evaluated as a long-term investment, tax preferred retirement savings can be a very cost-effective form of public finance.

Although commonly used measures appear to indicate that these tax expenditures are heavily concentrated in higher-income segments, this does not reflect the reality of their distribution. This result is significantly a function of the distribution of income tax payments which is even more concentrated by income that tax expenditures for retirement savings.

A number of other important issues are relevant to the consideration of the distribution of retirement tax incentives. The concentration of retirement savings in older individuals when they are at peak earning years skews any measurement of their value in relation to lifetime income. When considered in relation to payroll taxes and benefits received from Social Security, the net effect is likely to be redistribution in favor of lower-income groups. Of equal importance is to consider the outcomes achieved by employer plans in terms of how they impact retirement savings in relation to what would occur in their absence. The tax qualification rules bring in many moderate income and younger workers who would not otherwise be saving for retirement.

² For a more in depth discussion of this equivalence with some illustrative examples see "The Benefits and Revenue Costs of Tax Deferral," Investment Company Institute, Washington DC, 2012.

³ Lower taxes may result from this deferral if the individual moves to a lower tax bracket in retirement however this is likely to be relatively small in relation to the value of the exclusion from taxation in the initial year both because changes in marginal tax rates are typically much less than the rates themselves (those participating in retirement plans tend to have similar pre and post retirement incomes) and because it will occur many years in the future and thus have less value in the present.

⁴ Joint Committee on Taxation, "Estimates of Federal Tax Expenditures for Fiscal Years 2014-2018,"JCX-97-14, August 5, 2014.

⁵ Office of Management and Budget, "Analytical Perspectives, Budget of the United States Government. Fiscal Year 2016," February 2015

⁶ Congressional Budget Office, "The Distribution of Major Tax Expenditures in the Individual Income Tax System," p 24, May 2013

⁷ This can be inferred from the assumptions CBO has published in the report cited above and the similar level of the present value estimates in relation to the tax expenditure approach.

⁸ American Society of Pension Professionals and Actuaries, "Retirement Savings and Tax Expenditure Estimates" Judy Xanthopoulos, Ph.D. & Mary M. Schmitt, Esq.), April 2012.

⁹ Congressional Budget Office, "The Distribution of Major Tax Expenditures in the Individual Income Tax System," May 2013. JCT Publishes some distributional analysis but does not specifically delineate retirement savings in its analysis.

¹⁰ The Urban Institute and Tax Policy Center, "Tax Subsidies for Asset Development -- An Overview and Distributional Analysis" (Harris, Benjamin H., C. Eugene Steuerle, Signe-Mary Mckernan, Caleb Quakenbush and Caroline Ratcliffe), August 2014

¹¹ CBO also estimates the present value of tax expenditures however did not publish this in the same detail as the TPC although reaching a similar conclusion that it reduces the value of tax incentives by about 10 percent and distributes them more to the middle three quintiles.

¹² Congressional Budget Office, "The Distribution of Household Income and Federal Taxes, 2010," December 2013

¹³Contemporary Economic Policy, "Taxes, Income and Retirement Savings: Differences by Permanent and Transitory Income" (Heim, Bradley T. and Ithai Z. Lurie), V 32 Issue 3, 2014

¹⁴ Center for Retirement Research, "Do Tax Incentives Increase 401K Retirement Saving? Evidence from the Adoption of Catch Up Contributions" (Rutledge, Mattew S., April Yanyuan Wu and Francis M. Vitigliano), November 2014

¹⁵ The Urban Institute, "How Progressive Are the Combined Benefits of Social Security and Tax Benefits for Retirement Savings?" (Smith, Karen E. and Eric J. Toder), August 2014

¹⁶ The Urban Institute Retirement Policy Discussion Paper, "Do Low-Income Workers Benefit from 401(k) Plans," (Toder, Eric and Karen E. Smith), September 2011

WHAT DO THE MOST CURRENT CPS NUMBERS SHOW?

The Current Population Survey is a large household survey that asks a rotating sample of 57,000 households each month a set of basic questions about their employment status and earnings. The CPS is the source of the highly visible monthly employment (and unemployment) numbers that are released on the first Friday of every month. The survey sample is very large and was established to ensure the precision and validity of the employment statistics. The questionnaire is carefully constructed and maintained in a constant form over the years to ensure the comparability of the numbers across long time periods. In addition to the core questionnaire each month the survey includes a short set of supplementary questions that are directed toward the short-term measurement and development of long-term data to track trends in a set of policy issues. These supplements are highly valued because of the very large sample size and, most importantly, the capacity to link the supplements to the core question on employment status and other socioeconomic and demographic characteristics.

However, the CPS questionnaire is also an inevitable compromise between precision and scope. There is only limited space available for each supplement and the large sample size makes each supplement very expensive to undertake, so they are generally only a few questions on each issue. For example, the March CPS supplement also includes questions on health insurance coverage that are the source of the highly visible numbers on coverage in this area. The CPS supplement asks two pension coverage questions: (1) does the respondent work for an employer that offers pension coverage for any of its employees, and (2) were you included in the pension plan. The wording of the question may significantly undercount pension plan coverage because the individual who participates in an employer-sponsored 401(k) plan or other similar plan may not recognize the plan may not recognize the plan as a pension plan.

The latest version of the CPS for which tabulations are available (there is a considerable lead time before the data is made available) is the March 2013 Supplement. The Employee Benefit Research Institute (EBRI) CPS tabulations indicate that among all workers, 48.6 percent reported that they had worked in a job for which pension coverage was available for at least some workers (though not necessarily themselves) and 39.4 percent reported that they had participated in the plan. This includes individuals of all ages as well as self-employed and agricultural workers. It also includes employees of both the public and private sector, which have widely differing patterns of pension coverage in which the vast majority of public sector workers, due to a variety of factors, work for an employer that provides pension coverage. Restricting the measurement to only private sector workers yields an overall pension participation rate of 34.4 percent which is typically the lowest pension coverage number that is cited.

The estimate of only about one-third coverage of private sector workers, however, is potentially misleading for a variety of reasons. In addition to the potential lack of recognition of the 401(k) plan as a pension plan, two more basic reasons are that it includes workers of all ages, some of whom may be too young to be required to even be eligible to participate in an employer pension plan and others who are old enough to receive benefits through Social Security or another employer pension plan. It also includes part-time, part-year or other types of workers with relatively tenuous employment relationships that are not conducive to employer-sponsored pension plan coverage.

More relevant coverage measures are therefore those that look only at individuals working full-time that are below the age of 65 and above an age when they are likely to have completed their education and begun to work in a sustained basis usually establishing a lower bound of 21 or 25 years of age. EBRI's tabulations indicate a pension "offer" rate for private-sector workers, age 21 to 64, of 48.5 percent, and a participation rate of 39.1 percent. A similar analysis from the Center for Retirement Research (CRR) at Boston College finds that for 25-to-64-year olds, the "offer" rate for full-time private sector workers was 59 percent with 51 percent of these "prime age" workers reporting participation in an employer-sponsored plan.

Interpretation of these pension coverage numbers requires consideration of several key questions, including how accurate and representative are the widely cited CPS numbers, how do they fit into any longer term trend and does the roughly 50 percent participation rate represent a glass that is half-empty or one that is half-full.

ARE THE CPS COVERAGE NUMBERS INDICATIVE OF CURRENT CIRCUMSTANCES?

There are several alternative measures of pension coverage currently available. These include other household-based surveys and data on individuals as well as measures that collect information from employers. There are three main alternative sources of household survey data on pension coverage, the Survey of Income and Program Participation (SIPP), which is a panel survey that asks questions similar to the CPS every two to three years; the Panel Study on Income Dynamics (PSID) and the Federal Reserve Board's Triennial Survey of Consumer Finances (SCF). Each of these uses a different survey method and asks slightly different questions. But there is enough similarity in the approaches to provide some rough comparability. The previously referenced CRR analysis found considerable variation in measured pension participation rates among full time private sector workers, age 25 to 64, among the four household surveys, ranging from a low of about 40 percent in the PSID to a high of nearly 55 percent in the SIPP. The study found that the differences among the surveys was relatively consistent over the past 20 years, indicating that the variations are the result of systematic differences in the way coverage is measured.

This would seem to indicate that there is some significant error in the reporting of pension coverage by individuals as previously mentioned. While some of the variation is due to somewhat non-comparable definitions of workers and differences in the designs of the survey panels, these differences are also consistent with other research over this time period that indicates that there is not a high level of awareness among many individuals, especially younger and lower income, of the nature of their coverage under employer sponsored pension plans. It is therefore likely that these household surveys provide a material undercount of actual pension coverage.

An alternative approach that sheds some light on the degree of underreporting of pension coverage is to measure coverage through employers. The largest and most representative source that measures coverage through a survey of potential plan sponsors indicates that pension coverage is likely to be higher than is reported in the household surveys. The Bureau of Labor Statistics conducts the National Compensation Survey (NCS) of employers, which also uses a very large and carefully calibrated sample to measure the cost and incidence of wages and benefits with a very high level of precision. The March 2014 NCS indicated that the firms in its' survey reported that 65 percent of all private sector workers and 74 percent of full-time private sector workers had access to a retirement plan with 48 percent of all workers and 58 percent of full time workers participating in the plan. While this is due in part to the exclusion of some potentially low coverage groups (i.e. self-employed and agricultural workers) from the BLS sample, as EBRI notes in its analysis, excluding the groups not counted by BLS and considering only full-year workers raises the CPS estimate to 42.8 percent, which is still well below the BLS number.

The perception that household surveys provide estimates of pension coverage that are too low is reinforced by various efforts to reconcile administrative and survey data on pension coverage. A 2011 study sought to link income tax records indicating contributions to a defined contribution pension (the vast majority of which now involve 401(k) elective deferrals) with household survey data to determine the accuracy of the household surveys in measuring pension coverage. This study concluded that while there appear to be errors in the reporting of both plan sponsorship and participation in the SIPP (which interestingly provide the upper end of coverage estimates from surveys) of three percentage points for employers offering a plan and five percentage points for the proportion offered the opportunity that are participating in the plan.

If the household survey data is aggregated to estimate total numbers of participants for earlier years, these similarly provide lower participation levels than the aggregates reported on the Form 5500 annual reports compiled by the Department of Labor, although these pose the challenge of adjusting for the reporting of dual coverage for individuals who have been covered under multiple plans of the same employer or more than one employer during the reporting period. Although DOL no longer calculates these coverage estimates, in the past they yielded higher estimates of overall participants in pension plans that are greater than the aggregates suggested by household surveys, indicating an especially low awareness of defined benefit coverage when compared to the household survey data. This is consistent with an extensive body of research over the past two decades that demonstrates significant gaps in individual participants' knowledge and awareness of their pension coverage.

The other important issue in the interpretation of survey data such as the CPS is that it represents what is known as "cross-sectional" or effectively a "slice-in-time" measure of pension coverage. The household and employer surveys measure coverage at the point of the survey response, or about what the individual has experienced in the preceding year. This is, at best, a very rough indication of what kind of pension benefits an individual will accrue over a lifetime. In the extreme case, if all individuals spend forty years in the labor force (from age 25 to age 65) and half are covered at any time with an equal probability of coverage for any one year, then all workers would reach a retirement age of 65 with 20 years of accrued benefits.

In addition, with the very rapid increase in female labor force participation beginning in the 1960's, a large proportion of individuals are members of a household in which both spouses have accrued significant work history. This is increasingly the case as the first generation of women with high labor force participation rates are reaching normal retirement ages. There are unfortunately only limited sources of information on accrued benefits at the household level. The best known and most comprehensive source of data to look at households' accrued pension status is the Federal Reserve's Survey of Consumer Finances (SCF), which interviews 5,000 households every three years to obtain very detailed information about earnings, assets and other aspects of their financial status. This survey has been the source of much of the documentation of stagnant or declining real earnings among all but the top 10 percent of the income distribution in the United States over the past 30 years. The latest SCF numbers from 2013 that were released in September 2014 indicate that the proportion of families that report having some type of retirement account has ranged from 53.0 percent in 2007 to 49.2 percent in 2013. The survey also found that 59.3 percent of household nearing retirement age (with the head of the household between age 45 and 64) has ranged from about 56 percent to as high as 62 percent. As with other survey data these numbers are likely to reflect a lower bound due to the limitations in awareness of retirement plan coverage observed on other surveys.

WHAT IS A REASONABLE ESTIMATE OF PRIVATE PENSION COVERAGE?

Considering all of the recent data that is available on private pension coverage, it seems reasonable to conclude that about 60 percent of private sector workers will be employed at some point during the year in a job that affords them an opportunity to save for retirement through an employer-sponsored plan and about one half of the workforce will be covered by the plan. As these workers move through their careers, about two-thirds of households will approach retirement age with some accumulated value in accounts or accrued benefits from these plans.