EBRI. Retirement Income Opportunities in an Aging America: and Adequacy EMPLOYEE BENEFIT RESEARCH INSTITUTE

Transcription

1 EBRI l I Retirement Income Opportunities in an Aging America: Income Levels and Adequacy EMPLOYEE BENEFIT RESEARCH INSTITUTE

2 Employee Benefit Research Institute The Employee Benefit Research Institute (EBRI) was established in 1978 to contribute to the development of effective and responsible public policy in the employee benefit field. EBRI is a tax exempt trade association with the overall goal of promoting the development of soundly conceived private and public employee benefit plans. A nonprofit organization, EBRI sponsors and conducts research in the employee benefit field. It publishes research results and other information on employee benefits and sponsors lectures, forums and workshops. By acting as an information clearinghouse, EBRI helps to provide private and public sector decisionmakers with useful information. The research sponsored by EBRI complements the work done by others at universities, in government and in private institutions. By helping to avoid duplication of projects and by assisting in the compilation and dissemination of information, EBRI contributes to the orderly expansion of knowledge in the field. EBRI's research is conducted by persons knowledgeable in the employee benefit field. The Employee Benefit Research Institute seeks a broad membership among companies and individuals that provide professional services in the employee benefit field, plan sponsors and other interested individuals and organizations. A broadly based membership helps EBRI sponsor major research and educational activities. Dues paid to EBRI are deductible as ordinary business expenses. EBRI has also established an Education and Research Fund. Contributions and Foundation grants to the Fund are treated as charitable contributions. More information on the Employee Benefit Research Institute can be obtained by writing: Executive Director, EBRI, 1920 N Street, NW, Suite 520, Washington, DC (202)

3 Retirement Income Opportunities in an Aging America: Income Levels and Adequacy Employee Benefit Research Institute Washington, D.C.

4 1982 Employee Benefit Research Institute 1920 N Street, N.W. Suite 520 Washington, D.C (202) All Rights Reserved. The views in this publication are those of the authors and do not necessarily reflect the views of the Employee Benefit Research Institute, its trustees, members or other staff. Library of Congress Cataloging in Publication Data Main entry under title: Retirement income opportunities in an aging America. Includes bibliographical references. Contents: v. 1. Coverage and benefit entitlement--v. 2. Income levels and adequacy-- 1. Old age pensions--united States. 2. Retirement income--united States. I. Employee Benefit Research Institute. HD7106. U5R ' ISBN AACR2 ISBN (set) Printed in the United States of America

5 Table of Contents Page Executive Summary... v Foreword... xi Introduction... 1 Chapter I The Elderly's Income Sources... 3 Major Income Sources; Social Security; Employer Pension Programs; Benefit Structure; Social Security Integration; Retirement, Disability and Death Benefits; Private Pension Programs; State and Local Government Pension Programs; The Civil Service Retirement Program; The Military Retirement Program; Financial Assets and Other Wealth; Employment Earnings; Public Assistance and Other In-Kind Benefit Programs; Medicare; Medicaid; Housing Assistance Programs; Food Stamps; Low-Income Energy Assistance; Supplemental Security Income Program; Other Programs; Current Distribution of Income; Cash Income; In-Kind Benefit Values; Family Asset Values; Employer Pension Benefit Trends Chapter II Income Adequacy under Alternative Standards Alternative Adequacy Standards; Relative Standards; Absolute Income Standards; Retirement Wealth Concepts; The Elderly's Expenditure Patterns; Expenditure Levels; Expenditure Patterns; Inflation; Income Adequacy; Alternative Income Definitions; Income Adequacy under Alternative Standards; Historical Trends; Elderly and Nonelderly Incomes; Characteristics of the Low-Income Elderly; Adequacy of Projected Program Benefits Chapter III Policy Issues Policy and Program Initiatives; Social Security; Replacement Rates; Spouses' Benefits; Retirement Ages; Benefit Indexation; Private Employer Programs; Mandatory Pension System Proposals; Spouses' Benefits; Vesting; Integration; Public Employer Programs; Individual Initiatives; Public Assistance and In-Kind Benefits; Future Research

6 Appendix A Background Tables on Retirement Program Beneficiaries and Benefits Appendix B Methodology for Analyzing the Elderly's Income Appendix C Methodology for Analyzing the Expected Pension and Social Security Benefit Levels Appendix D Background Tables on the Elderly's Income Distribution in

7 Executive Summary Federal retirement programs and policies have played a primary role in determining the elderly's economic status. Five Presidential Commissions have been formed since 1975 to study the question of how to provide adequate income to the elderly. Substantial legislative changes have occurred and others are underway that aim at resolving elderly income problems. For example: --Major Social Security amendments in and 1981: --Employee Retirement Income Security Act of 1974(ERISA): --Multiemployer Pension Plan Amendments Act of 1980(MPPAA): --Amendments to the Age Discrimination in Employment Act of 1978: --Proposed changes to ERISA are pending: --Proposed changes to Social Security are pending. During the last twenty years, the elderly's financial status has improved substantially. Today those who are over age 65 receive income from more sources and have greater financial independence than previous generations of elderly. Expansion of public and private retirement income and benefit programs has increased the proportion of those receiving retirement benefits as well as the average benefit amount. This report concludes that the elderly's income levels and sources will continue to improve during the next twenty years or more. Most significantly, a higher percentage of workers and families will become eligible for employer pensions. When addressing the elderly's financial problems, policymakers should consider the elderly's future income potential, not just their current income levels and sources. The Eiderly's Income Sources Chapter I provides a summary of the elderly's major income sources. In 1979, elderly couples and single persons received over $200 billion from five major income sources. Each source--social Security, employer pension plans, employment earnings, personal assets and public welfare programs--plays a different role in providing income to the elderly: --Social Security is the fundamental program. Over 90 percent of elderly families receive Social Security retirement benefits--this comprises more than one-

8 third of elderly families' total cash income. Thus, it favors lower-income workers by providing them with a larger preretirement income replacement rate than that given to higher-income workers. --Approximately one-third of the elderly receive earnings from part-time or full-time employment. This is the second largest income source for the elderly. --Nearly three-quarters of elderly families received income from family assets in an average of $3,000 per recipient family. --Over one-third of elderly families received employer pension benefits in an average of $4,100 per recipient family. --Most elderly families also receive in-kind benefits such as Medicare and Medicaid. Government programs, such as Supplemental Security Income (SSI), provide a significant source of financial assistance only at the lowest income levels. In 1979, government in-kind benefit and income assistance programs provided over $40 billion in benefits to the elderly. Noncash benefits have become a significant part of the elderly's total economic support: (1) Almost all receive Medicare and/or Medicaid. (2) Many also receive benefits such as food stamps and housing assistance. (3) Significant income tax benefits are also available, e.g., the exclusion of Social Security benefits from federal income tax. Additionally, a majority of the elderly own their homes. Homeownership provides shelter and results in reduced day-to-day living expenses. Home equity is also an important potential income source. During the past thirty years, the elderly's income sources have changed greatly. --In 1950, only 20 percent of elderly households received Social Security benefits and less than 10percent received employer pension benefits. Approximately 40 percent of men age 65 and over worked. --In 1980, over 90 percent of elderly households received Social Security and over one-third received employer pensions. Additionally, the average real Social Security benefit has increased by a factor of three since The average real employer pension benefit also increased. While Social Security and employer retirement program benefits have increased in importance relative to other sources, the role of employment income has declined. Presently, only 20 percent of men age 65 and over are working. Many of these trends are expected to continue. If real wages increase, Social Security benefits will increase. If real Social Security benefits increase, it is likely that even fewer elderly will work. Approximately 34 percent of all elderly families currently receive employer pension plan income, and this is expected to increase significantly. Chapter I indicates that: vi

9 --The percentage of married couples Iwhere at least one person is age 65-69) who are eligible for employer pension benefits will increase from about 50 percent in 1979 to 85 to 90 percent in The percentage of single persons (age 65-69) who are eligible for employer pension benefits should increase from 33 percent in 1979 to approximately 70 percent in Approximately 80 percent of families (where at least one member is age 65 69) will be eligible for benefits from employer pension programs, Keogh plans or IRAs by the year The average size of employer pension benefits is likely to remain constant. Assuming real annual wage increases of 1 percent and moderate pension coverage increases, the average pension benefit for families retiring at age 65 will be approximately $4,000 (in 1981 dollars) in the year Future increases in the percentage of families eligible for employer pension benefits and in the benefit size reflect four primary factors: --The continued maturing of employer retirement programs--an increasing proportion of workers are spending a significant portion of their careers in pension-covered employment. --More liberal private pension plan vesting standards--this increases the proportion of covered workers expecting to receive benefits. It also increases the number of covered years of service for some workers. -- Significant increases in the number of two- worker families--this increases the likelihood that at least one family member will receive a pension. Increases in female labor force participation will also contribute to increases in pension benefits. --A growth in the prevalence of joint and survivor options--as more vested workers elect these options, more individuals will receive pension benefits based on their spouses" work records. Income Adequacy under Alternative Standards Evaluating retirement income adequacy is in part a judgmental exercise. Chapter II describes different standards of retirement income adequacy and how they have been used in: (1) establishing public policy goals; (2) developing individual firm and worker retirement objectives; and (3) identifying problems in current programs and policies. For example, the "poverty level" ($4,508 for couples and $3,528 for single persons in 1979) has been used to identify the low-income elderly. --Approximately 9.4 percent of elderly couples and 26.3 percent of elderly single persons had cash incomes below this level in On a family basis, 18.5 percent of elderly families had cash incomes below the poverty level in vii

10 --Cash incomes considerably understate the elderly's well-being because they exclude in-kind benefit values such as Medicare, Medicaid, food stamps and housing assistance. If in-kind benefit values are added to cash income, the proportion of elderly families below the poverty level declines from 18.5 percent to 12.4 percent. --Family assets also have a significant impact on the elderly's financial wellbeing that is not captured in cash income measures. In addition to savings, many elderly have acquired substantial home equity, which potentially could be converted into income. When these potential income sources as well as in-kind benefit values are added to cash income, the proportion of families below the poverty level is reduced to less than 8 percent. Although other income adequacy standards produce different estimates, most measures identify similar groups of elderly with inadequate incomes: --Single women (primarily those over age 72) represent the largest elderly group with below-poverty-level cash incomes. --Those with below-poverty-level cash incomes probably receive income from fewer sources than the nonpoor elderly. Less than 10 percent of the elderly poor receive employment income and less than 10percent receive employer pension benefits. --The elderly poor also receive less income from each source. The average Social Security benefit for those with below-poverty-level cash incomes is about one-half the Social Security benefit of the nonpoor. The last two conditions result from several factors. Many elderly poor with significant work experience had low preretirement incomes and were probably less able to save than the nonpoor. Many elderly poor have less work experience than the nonpoor. For some, the limited work force experience reflects personal choice; others were unable to work due to poor health or disability prior to retirement. A smaller percentage of future elderly should have below-poverty-level cash incomes because of: (1) increased availability and size of employer pension and IRA benefits; (2) impact of real wage increases on the size of real Social Security benefits; and (3) increased female labor force participation, which should result in more women with pension benefits, as well as larger Social Security benefits. Although these economic and demographic changes are presently occurring, their full effects will not be experienced for many years. Income adequacy estimates provide a useful profile of income available to the elderly. However, no evaluation based on such estimates can be absolute. Chapter II discusses problems associated with evaluating elderly income status and establishing unambiguous adequacy standards. It concludes that income alone does not reflect the entire picture. For example, VIII

11 most elderly have broad options available that have an impact on their income level. One of the most important choices pertains to the available range of retirement ages. In recent years, Social Security and employer pension programs have offered liberal early retirement provisions. Thus, there has been a substantial decline in workers' retirement ages. Some choose early retirement and a lower income in order to enjoy more leisure years. This trend indicates that "adequacy," as defined by the elderly themselves, has improved. Merely examining income levels or replacement rates does not capture such choices. Policy Issues There are specific elderly groups who, on average, receive lower incomes than others. Generally, these include single persons over age 72--nearly 55 percent are widowed females. These elderly typically have little or no work experience; they depend primarily upon Social Security and public assistance. Due to their limited employment history, fewer retirement income sources are available to this group. In the United States, retirement income is provided through several major programs, each with its own primary objective. Tailoring these diverse programs to address problems of the elderly poor is sometimes impractical. Expanded benefits from employment-based programs, such as employer pensions and Social Security, will improve the elderly's overall retirement income level, but may do little to help those with no work experience. In addition, recent or future tax incentives for savings and pension coverage will benefit coming generations, but are unlikely to affect the current elderly. Thus, it may be prudent to address the problems of today's poor with public assistance or Social Security benefits, while providing for future generations with tax incentives, expanded pension coverage or more liberal survivors benefits. This report shows that many low-income elderly have substantial home equity. Policies that facilitate conversion of this equity into cash income may be useful. Those attempting to restrain federal budget growth and to resolve Social Security's financing difficulties, may prefer policies that change the future mix of retirement income programs. Policymakers should understand the interaction of current policies and programs when assessing new policy and program mmatlves. The elderly's economic circumstances have improved and will show continued improvement because of legislative, economic and demographic changes that have already occurred. The full effects of many of these chan_es are not yet evident. For example, as this report shows, the percenix

12 tage of elderly families with employer pension eligibility has increased significantly during the past fifteen years. Under current policies, the percentage of newly retiring elderly with employer pension benefits is expected to double during the next twenty-five years. The average real benefits received by such families are also expected to increase. These trends are not well recognized; many current policies and programs assume that few families will receive pensions. A majority of future retirees will receive employer pensions and proposed policy changes should reflect this trend. An effective determination of what constitutes adequate elderly income cannot be made without understanding: ( 1) the impact that alternative measures of income (e.g., those that consider the value of in-kind benefits or those that do not) have on analyses of the retirement income system's effectiveness; (2) the impact that different income adequacy standards (e.g., the poverty level, the minimum wage or Orshansky's revised index) have on analyses of the retirement income system's effectiveness; (3) the effects of the elderly's expenditure patterns on income adequacy; (4) the impact of personal preferences toward work, leisure and early, normal or late retirement on the elderly's income. This report provides a perspective on these issues and on the economic status of the nation's aged population. The final chapter reviews and analyzes retirement income adequacy issues that relate to previous chapters' findings and identifies future research needs.

13 Foreword Everyone is concerned with his or her income level and its ability to satisfy family needs. However, the question of what is "adequate income," is particularly important for retirees who work minimally or not at all. Retirement income adequacy studies have traditionally focused on cash income. Few previous analyses have compared the elderly's income goals with the goals of other age groups. Attention has seldom been given to the elderly's financial assets, even though such assets enhance income security and provide higher living standards. These and similar economic factors are essential to studies that form the basis of public policy development in the income adequacy area. Retirement income adequacy is increasingly an important public policy issue. It is at the center of debates over proposed changes in Social Security, Supplemental Security Income and similar programs' benefit levels. Additionally, it impacts on the effective distribution of scarce resources. This is vitally important to the overall welfare and economic prosperity of the United States and other nations. Recent headlines highlight the present preoccupation with social welfare and entitlement programs: "'France Takes Aimat Social Security--Mitterand Puts Stopgap Planinto Force" "Reagan's Social Security Proposals Cause Public Outcry in the U.S. "' "'GermanyCuts Social Security Indexing" "Switzerland Moves Towards Mandatory Private Pensions" "Social Security Proposals Bring Down Government in Belgium" "Survival of Private Pensions Threatened by Growth of Public Programs" "Pension Assets Dominate U.S. Capital Markets" The Employee Benefit Research Institute supports the view that public policy decisions must be based on full information and understanding. This is particularly true when working with issues such as income adequacy, which affect the lives of every American. When assessing public policy impact on economic and social conditions, it is essential to consider the future as well as the present. For example, the maturing of retirement programs, changing female labor force participation and similar demographic and economic trends will change the resources and the income levels of future retiree groups. xi

14 This report is intended to provide information that adds perspective on the subject of income adequacy. It offers a review of historical data, statement of present conditions and series of ten to twenty year projections. EBRI is grateful for the extensive help of numerous outside reviewers who provided comments and suggestions. These reviewers, as well as EBRI's sponsors, have not seen the final report before publication; thus EBRI's staff takes full responsibility for the report's content and accuracy. ICF Incorporated prepared the basic analyses under contract with EBRI. Mr. David L, Kennell of ICF Incorporated developed the principal research and writing under the direction of the EBRI research staff; and Ms. Patricia George Moore of EBRI provided the principal editing. DALLASL. SALISBURY Executive Director February 1982 xii

15 Introduction Since the 1960s, there has been ongoing discussion about relevant income measures and income adequacy standards. Income adequacy standards are used to assess the effectiveness of government transfer programs and private income sources in meeting the population's economic needs. Various income measures assist policymakers in determining income adequacy standards. These standards then help determine benefit amounts provided through public welfare and retirement programs. The level of benefits from public programs, in turn, may affect the level of benefits from employerand/or employee-sponsored programs. Income adequacy measures and standards affect employees, employers, taxpayers, program recipients, the federal budget and the economy at large. At a time of rising unemployment, expanding retirement population, high inflation and escalating budget deficits, it is essential that federally set income adequacy standards be carefully developed. If income standards are to carry out their intended goals, they must be based on sound, thorough research. They must carefully consider all present and future, social and economic factors and their interactive effects. This report's purpose is to examine elderly income adequacy by analyzing recently released data on elderly income levels and sources. The report does not focus strictly on retirement income programs; instead, it examines all economic resources available to the elderly. Definitions of Important Terms In this report: The elderly--refers to: ( 1) married couples where at least one spouse is age 65 or over: and (2) single persons who are age 65 or over. The elderly includes retired and nonretired workers, as well as those who have never worked because of personal preferences, disability or poor health. Income adequacy--no consensus exists regarding the definition of an adequate income level for the elderly, This report discusses and illustrates the application of alternative income adequacy standards. The report does not propose an appropriate standard, but it does provide background information and analyses that are necessary for examining this important normative issue. Scope Chapter I examines the wide range of income sources available to the elderly. These include: Social Security, employer pensions, investment 1

16 earnings and personal assets such as homeownership. The economic value of other benefits available to the elderly are also evaluated, e.g., Medicare, Medicaid, tax benefits, Supplemental Security Income (SSI), food stamps, housing and energy assistance. Chapter II examines income adequacy standards used in identifying the low-income elderly. It discusses the difficulty in applying these standards to all retirement income sources. Chapter II illustrates the application of alternative income standards to different retirement income definitions. These comparisons then provide an empirical basis for identifying characteristics of the lowest-income elderly. Finally, Chapter III reviews recent proposals that may affect elderly incomes. It focuses on the effects of such proposals on low-income families. It also considers these proposals' potential effects in the context of present trends which anticipate positive changes in future retirement income levels and sources. This study is part of a broader EBRI research project. A previous report was prepared on retirement income coverage and its expected evolution; 1 and a future report will focus on retirement program funding. _EBRI, Retirement Income Opportunities in an Aging America: Coverage and Benefit Entitlement (Washington, D.C., 1981). 2

17 I. The Elderly's Income Sources During the last thirty years, the income levels and sources for those over age 65 have changed significantly. In 1950, 20 percent of elderly households received Social Security benefits and less than one-tenth received employer pensions. 1 Table I- 1 shows that over 90 percent of elderly households received Social Security benefits and over one-third received employer pensions in The real value of retirement income also increased. For example, after adjustment for inflation, the average 1980 Social Security benefit was three and one-half times larger than it was in While Social Security and employer pension benefits have increased in importance, the role of employment income has diminished. Table I-I 1 shows that approximately 40 percent of men age 65 and over worked and 79 percent of men age 60 to 64 worked in Today, about 20 percent of men over age 65 work and 60 percent of those age 60 to 64 work. TABLE I-1 Income Sources for Elderly Householdsm19791 Total Income Percent from Each Source _ Average Receiving (billions) Income 3 Social Security 92% $ 66 $4,300 Employer Pensions ,100 Savings/Wealth ,000 Employment Earnings ,500 Public Assistance/In-Kind ,200 Programs Total N/A $217 N/A Source: ICF analysis for EBRI of the March 1980 Current Population Survey data. 1Elderly households consist of married couples with a head age 65 and over and single persons age 65 and over. -'Total income provided to all elderly from each source listed. 3Average amount from each source provided to households, which receive income from the particular sources noted. qcf estimates derived from data presented in U.S. Social Security Administration, Demographic and Economic Characteristics of the Aged (1975). 3

18 The increase in female labor force participation also has significant implications for the elderly's income. One-third of all women worked in 1950, but nearly one-half worked in This has had positive effects on family savings and, in turn, on retirement wealth. The number of women with Social Security and employer pensions, and the size of their benefits, have increased. Finally, the adoption of major benefit programs during the last thirty years has increased the elderly's income and has provided retired persons with valuable economic benefits. Medicare and Medicaid, adopted in the sixties, cover a large portion of the eiderly's health care costs. Previously, these costs were primarily borne by the elderly or their families. Food stamps, low-income energy assistance and housing assistance are also recently instituted programs. Such programs have enabled many more elderly to live independently and with higher living standards. Although some elderly live in poverty, the proportion has declined substantially from just ten years ago. Many of these trends will continue. The percentage of households with employer pension income will increase. Female labor force participation will result in more women receiving Social Security and employer pensions. Since the benefits will be based on their own work records, rather than their spouses', family benefits should also be higher. This chapter discusses the levels of income provided to the elderly from five broad sources: (1) Social Security, including retirement benefits under the Railroad Retirement program; (2) employer pensions, including private and public sector defined benefit and defined contribution plans, individual retirement accounts (IRAs) and Keoghs; (3) financial assets and other wealth, such as home equity; (4) earnings from part-time and full-time employment; and (5) in-kind and cash benefits from public assistance programs such as Supplemental Security Income, Worker's Compensation, Medicare, Medicaid, food stamps, housing and energy assistance programs. This chapter describes: (1) the benefit structure characteristics of these five income sources; (2) their relative roles in providing income to the elderly; and (3) recent benefit trends under employer pension programs. Major Income Sources Table I- 1 shows that these five major income sources provided over $200 billion--or an average $13,000 per elderly household in Social Security provided the greatest overall amount--approximately 30 percent of 2u.s. Departmentof Labor,Bureauof LaborStatistics, "Employmentin Perspective: WorkingWomen"(Washington,D.C.: GovernmentPrintingOffice, 1980),p. 1; and AmericanCouncilof Life Insurance,"Womenin the LaborForce:Datatrack7" (Washington,D.C.:ACLI,1980),p. 20.

19 total income. As Table I-1 shows, in 1979, 31 percent of elderly households received employment earnings: this was the elderly's second largest income source overall and for those elderly who worked for pay, employment earnings provided the greatest portion of their household income. Social Security--The Old Age and Survivors Insurance (OASI) program replaces a portion of earnings lost due to a covered worker's retirement or death. It is and probably will continue to be the elderly's largest income source. 3 In 1979, approximately 20 million elderly individuals received nearly $66 billion in OASI benefits (excluding Medicare and Disability Insurance).4 OASI provides two major benefits: (1) the primary insurance amount (PIA); and (2) survivors benefits. The PIA is the principal retirement benefit and is paid to workers retiring at age 65 or over. 5 Benefit amounts are based on each worker's average indexed monthly earnings (AIME) that were subject to Social Security taxes throughout the worker's employment years. Survivors benefits are paid to aged spouses, children under age 18 and dependent parents of insured, deceased workers. Surviving spouses receive benefits in amounts up to 100 percent of the deceased worker's PIA; surviving children receive up to 75 percent; and dependent parents receive up to 82.5 percent. Table 1-2 shows that, in 1980, approximately 22 million of the 30 million OASI benefit recipients were workers and their dependents. The remaining 8 million were survivors. Survivors received about $25 billion or roughly 25 percent of all OASI benefits. Social Security benefits are based largely on a worker's covered earnings. Higher lifetime covered earnings result in larger benefits. The relationship between benefit amounts and lifetime earnings is adjusted by two factors. First, benefits are calculated based on a weighted formula that favors workers with low lifetime covered earnings. In 1980, the PIA benefit formula replaced 90 percent of the first $211 of AIME, 32 percent of the next $1,063 of AIME and 15 percent of the remaining AIME up to the maximum. Based on this formula, workers with low lifetime covered earnings receive a benefit that is proportionally greater relative to their average wages than those with higher earnings. As shown in Table 1-3, in 1979 a single minimum wage earner received a benefit that replaced 57 percent of his or her covered wages; however, those with maximum covered earnings received benefits that replaced only 25 percent of their covered wages. 3Fora detaileddiscussionof coverageundersocialsecurityand otherprogramsdiscussed here, see EBRI, Coverageand BenefitEntitlement. 4ICFanalysisfor EBRIof the March1980CurrentPopulationSurveydata. SThoseretiringearly and receivingbenefitsat age 62 receive80 percentof the PIAthey wouldhave beenentitledtoat age 65.

20 In 1981, no PIA could be less than a specified minimum of $122 monthly.6 Under current law, the amounts of AIME within each bracket are indexed at the rate of aggregate wage increases; but the individual replacement rates within each AIME bracket remain fixed. This formula reflects Social Security's redistributive benefit structure. OASI Beneficiaries TABLE I-2 and Benefit Payments Beneficiaries Benefit Payments Average (millions) (billions) Benefit Retired Workers 19.1 $ 74.3 $4,100 Wives and Husbands ,100 Children ,600 Widowed Mothers ,900 Widows and Widowers ,700 Other ,500 Total 30.3 $105.3 $3,700 Source: u.s. Social Security Administration, Social Security Bulletin, vol. 44, no. 4 (1981), pp TABLE I-3 Age 65 Social Security Pretax Replacement Ratesm1979 Single Worker Worker with Dependent Spouse Maximum Covered Wage Earner 25% 38% Average Covered Wage Earner Full-Time Minimum Wage Earner Source: U.S. Department of Health and Human Services, Advisory Council on Social Security, Social Securi(v Financing arid Bene[_'ts{1979), p The Social Security minimum benefit has been eliminated for all persons retiring after January 1,

21 A second adjustment to the relationship between earnings and benefits is that married workers receive additional protection for their dependents, including spouses and children under 18, even though they pay the same tax rate as single workers. Table I-3 demonstrates that at all levels of lifetime covered earnings, the Social Security benefit formula produces higher replacement rates for workers with dependents than for those without dependents. This feature results in a need for the second adjustment factor in the relationship between earnings and benefits. In 1950, 3.5 million people received Social Security benefits; today over 30 million people receive benefits. Real average benefit values have also increased sharply. Table I-4 shows they doubled between 1950 and 1955 and increased by 75 percent between 1955 and This resulted from increases in initial retirement benefits and real lifetime earnings, as well as indexation of postretirement benefits. OASI benefits increased by 15percent in 1970, 10 percent in 1971 and 20 percent in Additionally, the 1972 benefit change incorporated a provision for adjusting existing beneficiaries' benefits with annual Consumer Price Index (CPI) changes. Employer Pension Programs--In 1979, employer-sponsored pension programs paid almost $60 billion in benefits to approximately 14 million persons. Table I-5 shows that although these programs cover fewer people than Social Security, their averagebenefits are of comparable size or larger. TABLE 1-4 OASI Program Beneficiaries and Benefitsm1950 to Beneficiaries Total Benefits Average Average Benefit Year (millions) (millions) Benefit (1979 dollars) $ 961 $ 276 $ , , , , , , ,796 1,222 2, ,509 2,110 2, ,556 2,984 2,984 Real Source: U.S. Social Security Administration, Social Security Bulletin Annual Statistical Supplement, , pp _These estimates are different from those in Table I-1 because they include beneficiaries of all ages. 7

22 TABLE 1-5 Benefits and Beneficiaries in Employer Pension Programs Percentage of Elderly Benefit Beneficiaries Households Payments Average Employer Plans (millions) Receiving (billions) Benefits Private % $23.6 $2,700 State/Local ,700 Civil Service ,700 Military ,000 Subtotal % $57.2 $4,100 OAS! % $90.6 $3,000 Source: See tables in Appendix A. testimates of beneficiaries, benefit payments and average benefits presented in this table include beneficiaries of all ages. They, therefore, differ from those in Table l- I. Benefit Structure--Employer pension programs provide several kinds of benefits: early and normal retirement, survivors, disability and other ancillary benefits. Although employer pension programs vary greatly in design, they can be categorized into two broad types: defined contribution plans--in which contribution amounts are predetermined while benefit levels are dependent on the total contributions and investment earnings in each employee's account; and defined benefit plans--in which benefit levels are determined in advance and contributions vary with investment earnings and plan experience. In addition, many plans coordinate their benefit or contribution structure with Social Security. The various benefits and types of plans are briefly described below. 7 Defined benefit plan benefit formulas can be divided into two broad types: (1) Unit benefit formulas, which credit a unit of benefit for each year of recognized service with an employer; and (2) Flat benefit formulas, which provide a specified benefit after a minimum period of service. (l) Unit benefit formulas can be based on a pay-related formula or a specific annual dollar amount. --Pay-related formulas are most common for salaried workers; and they usually range from 1 to 2 percent of the defined compensation base. The annual retirement benefit is the sum of amounts credited for each employ- 7For more complete descriptions of the different types of employee benefit programs, see EBRI, "Educational Pamphlet Series" (Washington, D.C.). 8

23 ment year. Pay-related formulas are usually based on a percentage of average earnings for a specified work period, e.g., the average pay during the final one, three, five or ten years of service. These are called final pay plans. Other pay-related plans, known as career average plans, use a compensation base, which is the average of a worker's earnings over his or her entire period of employment. Some plans provide a specific dollar amount to all employees who satisfy a minimum period of credited service--typically ten to fifteen years. Specific dollar amount fi_rmulas are common for hourly or unionized workers. Here, benefits are expressed as a fixed dollar amount per unit of service: e.g., a $15 monthly benefit for each year of service. These plans are often updated for inflationary effects; thus, many retired workers have different fixed dollar amounts applied to different years of service. For example, a worker's monthly benefit may equal the sum of $12 for each year of service before 1975, $15 for each year of service between 1975 and 1978 and $18 for each year of service after (2) Flat benefitj_)rmulas provide specific benefit levels after a minimum service period. They may be based on a percentage of compensation or a specific dollar benefit. Flat benefits, in recently established or amended plans, are usually in the form of a minimum benefit supplemented by a formula that provides an additional benefit for each year of credited service. Defined contribution plans base contributions on predetermined fixed formulas. There are several types of defined contribution plans. A common type is profit sharing. In these plans, annual contributions are based on the sponsor's profitability. Generally, the size of the employer's payment to the plan is derived by a predetermined formula, although it may be decided at the employer's discretion. Allocation of the employer's total contribution is based upon a formula that is usually related to the employee's compensation. A second major type of defined contribution plan is the money purchase pension plan. Annual contributions to money purchase pension plans are usually based on annual compensation. For example, sponsors may contribute 10 percent of total annual compensation to the plan. Benefits will then vary among employees depending on factors such as age, sex and retirement date. A third major category of defined contribution plan is the thrift or savings plan. These plans typically permit employees to voluntarily contribute from 2 to 6 percent of their pay. Employer contributions are usually a fixed percentage of employee contributions--most commonly 50 percent, but sometimes higher. Thrift and savings plans are frequently offered as supplemental protection where employers also offer other defined benefit or defined contribution plans. Stock bonus plans are another form of a defined contribution plan. Stock bonus plans permit employers to contribute shares of company stock to a plan. The shares are then allocated to individual participant accounts, 9

24 usually based on a percentage of annual compensation or a fraction of total participant compensation. Employee Stock Ownership Plans (ESOPs) and Tax Reduction Act Stock Ownership Plans (TRASOPs) are the most common stock bonus plans. Social Security Integration--Internal Revenue Service integration rules permit defined benefit and defined contribution plans to coordinate or "integrate" their benefits with Social Security benefits. The rationale for integration is that employers pay for part of Social Security and should therefore be able to design pension programs that recognize employer financed Social Security benefits. Pension plans are also integrated with Social Security to ensure that: (1) workers with equal years of service but different wage levels will have roughly equivalent replacement rates; (2) each worker's total retirement benefits from Social Security and employer pension plans do not exceed their preretirement pay. Pension integration may seem discriminatory, because it appears to permit employers to provide high-income workers with larger pensions than those provided to low-income workers. However, integration rules restrict the degree to which the Social Security benefit tilt toward low-income workers can be utilized in designing benefit formulas. Integration rules are actually an explicit recognition of the complementary nature of pensions and Social Security. IRS rules limit the extent and manner in which integration is accomplished. There are three basic approaches to integration. Social Security offset formulas are the most frequently used method. This approach directly reduces pension benefits relative to a worker's Social Security primary benefit. The rules limit the rate at which Social Security benefits can be reduced. Most often the plan benefit is reduced 50 cents for each dollar of Social Security primary benefits. The second method bases benefits on earnings in excess of those subject to Social Security taxation. This is the excess integration method. The third method is referred to as the step-rate approach. Step-rate integration formulas provide benefits, which replace a higher percentage of pay above a certain compensation level than below that compensation level. Current integration rules were established in Since then Social Security has been amended several times. The 1977 Amendments made fundamental changes in benefit levels and calculation procedures. Pension integration rules have not been modified to reflect these changes. There is substantial agreement that these rules should be modified. 8 81ntegration regulations are too complex and voluminous to discuss fully in the context of this paper, EBRI is presently developing research on Social Security integration that will investigate current integration rules, their implications and potential policy changes. This research should be completed in mid

25 Retirement, Disability and Death Benefits--In a defined contribution plan, the accumulated account balance is usually provided to the employee or his family in the event of disability or death. Similarly, in a defined benefit plan, the employee's accrued benefit, or a prorated portion of that benefit, is generally provided to an employee or his family at time of death or disability. Early retirement benefits are generally payable to participants who satisfy certain age and service requirements. They may be based on age alone, years of service alone, or an employee's combined age and years of service. In private sector plans, the early retirement benefit is usually the employee's accrued benefit, reduced to reflect his or her longer benefit period resulting from early, rather than normal retirement. Additionally, some pension plans provide supplemental benefits to early retirees until they satisfy Social Security age requirements. In effect, such supplemental benefits are comparable to the employee's future Social Security benefits. When OASI payments begin, the supplemental benefits are terminated. Disability benefits may also be provided. They are usually tied to age and/or service requirements, and they are typically contingent upon satisfying the plan's specified definition of disability. Disability benefits may be a percentage of the participant's compensation when disability began, or they may be a percentage of the participant's accrued or projected pension benefit. Benefits may continue throughout the employee's lifetime or until normal retirement age, when a retirement benefit becomes available. When the employer provides a separate long-term disability program, pension benefit payments are often deferred until normal retirement age. Nearly all employer pension programs provide survivors benefits. If the participant is married when normal retirement benefits begin, benefits must legally and automatically be paid through a qualified joint and survivor annuity unless the participant elects to receive benefits in some other form. Joint and survivor annuities provide the participant and his or her spouse with lifetime benefits. Individual benefit payments are usually reduced to reflect the anticipated additional cost of spreading benefits out over the lifetimes of two persons rather than just one. Early retirement joint and survivor options are handled somewhat differently. Once a pension plan participant becomes eligible for early retirement, he or she must legally be permitted to select coverage under a joint and survivor option. This arrangement allows surviving spouses to receive benefits in cases where an employee becomes eligible for early retirement, but continues working until death. However, this option does not occur automatically. The joint and survivor coverage will only become effective after the employee has chosen this option in writing. Again, benefits may be reduced to reflect the costs of the additional coverage, and survivors benefits 11

26 may also be reduced to reflect the deceased employee's early retirement status at time of death. It should be noted, however, that many private sector plans automatically provide early and normal retirement, survivor benefit protection and do not reduce benefit amounts in conjunction with these added benefit provisions. Private Pension Programs--There are more than 600,000 private pension plans. 9 Each plan was established to respond to local market or collective bargaining conditions, as well as specific employee and employer characteristics. Private retirement programs are designed to respond to individual employee and employer needs and this goal is reflected in their diversity. Real private pension plan benefit payments have increased by a factor of twenty since Table I-6 shows that after inflation adjustment, average real benefits have increased by about 10 percent. A number of recent developments have had a negative effect on this increase in average real benefits: pension coverage growth, faster vesting, increased availability of joint and survivors benefits and Social Security TABLE 1-6 Private Sector Employer Pension Programs to 1979 Real Beneficiaries Total Benefits Average Average Benefit Year (thousands) (millions) Benefit ( 1979dollars) $ 370 $ 822 $2, , ,780 1, , ,750 3,520 1,280 2, ,740 7,360 1,553 2, , ,850 2,087 2, ,700 23,600 2,713 2,700 Sources: Alfred M. Skolnik. "Private Pension Plans, '" Social Security Bulletin, vol. 39, no. 6, June 1976, pp American Council of Life Insurance, Pension Facts (Washington, D.C., 1977), pp, 30-31,36. Private pension values for 1979 were derived from ICF Incorporated, A Private Pension Forecasting Model Profit Sharing Research Foundation, "'Cumulative Growth in Number of Qualified Deferred Profit Sharing Plans and Pensions in the U.S. 1939Through 1980" (Evanston, lh PSRF, 1981), table 3. 12

27 expansion. I From 1950 to 1979, private pension coverage increased dramatically. This recent pension coverage expansion has resulted in a disproportionate number of newly covered workers. These new participants have relatively low average-covered wages; since pension benefits are generally wage related, this group's average benefit is also low. This rapid increase in newly covered workers with relatively low benefits has temporarily restrained the real growth in average pension benefits. Similarly, the liberalization of vesting standards has also lowered average real benefits. Although vesting has increased the number of persons receiving benefit payments, some of the new beneficiaries have only a few years of vested service resulting in relatively small benefits. As previously noted, joint and survivor provisions produce benefits for more retirees (participants and their spouses) over longer periods of time. However, the individual benefit payments are reduced to reflect this additional coverage. Since these provisions have been adopted primarily in recent years, they too have lowered the real benefit growth rate. Finally, Social Security benefit provisions have expanded rapidly in recent years. Since employers must help to pay the costs of expanded Social Security, they may do so with dollars that, otherwise, would be contributed to pension plans--especially where plans are explicitly integrated with Social Security. State and Local Government Pension Programs--State and local governments also sponsor a diverse set of pension programs. A 1976 survey of Public Employee Retirement Systems (PERS) estimated that more than 6,000 of these plans existed, and nearly 80 percent of them had fewer than 100 active members. Plans with more than 1,000 active members comprised only 6 percent of all plans but covered about 95 percent of all active PERS membership. _ State and local employees are more likely to be covered by a defined benefit plan than their private sector counterparts. Eighty-two percent of the workers covered by a PERS were participating in a defined benefit plan according to the Pension Task Force survey. An additional 16 percent were participating in "combination" plans having both defined benefit and defined contribution features. Only about 2 percent participated in a _'pure'" defined contribution plan. 12 State and local plans were also more likely to require employee contributions than private sector plans. The Pension Task Force survey found that 85 _ For more information, see EBRI. Coverage at d Benefit Entitlement. _ U.S. Congress, House. Committee on Education and Labor. Pension Task Force Report on Public Employee Retirement Systems, 95th Cong.. 2nd Sess. (Washington. D.C.: Government Printing Office, 1978). p. 51. I'-lbid., p

28 percent of PERS participants were required to contribute to their plans. Some state and local jurisdictions, however, are moving toward noncontributory plans. Public sector defined benefit plans frequently offer shorter vesting requirements than private plans. Public plan vesting, however, is not strictly comparable to private plan vesting. When private workers vest, their benefit fights are not conditioned on their own contributions remaining in the plan. Public plans generally are different. For example, a worker may vest in a PERS after five years of covered service. However, a terminating worker's rights to the employer's benefit are contingent upon personal contributions remaining in the fund. If the worker withdraws his or her personal contributions, the employer financed share of the "vested" benefit is foregone. In a few select cases (e.g., Los Angeles Police and Fire Plan), employee contributions are also lost if the worker has not vested at the time he or she withdraws. State and local plans are also less likely than private plans to be integrated with Social Security. _3One reason there is less integration among public plans is that approximately 30 percent of state and local PERS participants are not covered by Social Security. Another reason is that many public plans were established before public employees were eligible for Social Security coverage and employers did not modify existing plans when they joined Social Security. However, several state and local jurisdictions have recently modified their PERS and have integrated benefits with Social Security, in order to reduce pension costs by taking advantage of Social Security benefit growth. State and local plans are more likely to provide postretirement cost-ofliving adjustments than private plans. More than one-third of PERS beneficiaries receive automatic increases linked to the CPI. Generally, these limit annual benefit increases to between 1 and 5 percent--most frequently 3 percent. More than 60 percent of the beneficiaries are not provided automatic increases but receive ad hoc adjustments. About half of all PERS, which covers less than 5 percent of state and local workers, have no provisions for benefit indexation._4 The number of beneficiaries in state and local government plans has grown less rapidly than beneficiary levels in other employer retirement systems. Table I-7 shows that although aggregate benefit payments have increased from less than $1 billion in 1955 to more than $10 billion in 1979, 13Marjorie M. Kulash, Integrating Public Employee Pensions with Social SecuriD' Retirement Benefits (Washington, D.C., 1980). _4U.S. Congress, House, Pension Task Force Report on Public Employee Retirement Systems, pp

29 TABLE I-7 State and Local Government Pension Plansm1955 to 1979 Real Beneficiaries Total Benefits Average Average Benefit Year (thousands) (millions) Benefit (1979 dollars) $ 595 $1,393 $4, ,078 1,633 4, ,775 2,003 4, ,291 3,280 2,541 4, ,730 7,025 4,061 5, ,300 10,770 4,683 4,683 Sources: U.S. Social Security Administration, Social Security Bulletin Annual Statistical Supplement pp : American Council of Life Insurance, Pension Facts (Washington, D.C., 1980), table 15. real average benefits have not increased substantially over this period, and they actually declined significantly between 1975 and The Civil Service Retirement Program--The largest federal employer plan is the Civil Service Retirement System (CSRS). CSRS is a defined benefit plan that provides retirement, survivors and disability benefits. All eligible employees must contribute 7 percent of their salary to the plan. Full retirement benefits are provided to retirees who satisfy one of several possible age and service criteria: age 55 with thirty years of service, age 60 with twenty years of service, or age 62 with five years of service. The CSRS benefit formula is based on an employee's average pay during his or her highest three consecutive years' salary. The average salary during these three years forms the employee's compensation base. The employee's pension is then determined by multiplying this base by 1.5 percent for the first five years of service, 1.75 percent for the second five years of service and 2 percent for all additional years of service. Initial benefits cannot exceed 80 percent of the employee's compensation base. Presently, all Civil Service retirement benefits are automatically adjusted with annual changes in the CPI. CSRS disability benefits are generally more liberal than private, state or local government systems; the CSRS disability definition is more lenient and its service eligibility requirement is relatively short (five years of service). As shown in Table I-8, the Civil Service Retirement System has grown rapidly over the past twenty-five years: beneficiaries have increased five- 15

30 TABLE I-8 Benefits under the Federal Civil Service Retirement System to 1979 Beneficiaries Total Benefits Average Average Benefit Year (thousands) (millions) Benefit (1979 dollars) Real $ 380 $1,279 $3, ,581 3, ,385 1,900 4, ,838 2,959 5, ,372 7,056 5,143 6, ,617 12,380 7,656 7,656 Sources: U.S. Social Security Administration, Social Security Bulletin Annual Statistical Supplement, , table 16, pp ; American Council of Life Insurance, Pension Facts (Washington, D.C., 1980), table 13, pp fold and aggregate real benefit payments have increased by a factor of ten. Since 1955, the average real benefit has almost doubled; this reflects the effect of increased salaries and CPI indexation. The Civil Service Retirement System provides higher average benefits than private or state and local government pension plans. However, CSRS-covered employees do not participate in the Social Security program. The Military Retirement Program--The Military Retirement System (MRS) provides higher lifetime benefits than almost any other employer plan. Servicemen are not required to contribute to their pension plan. Retirement benefits are automatically adjusted with CPI changes. However, MRS does not provide vesting of nondisability retirement benefits for military personnel with anything less than twenty years of service. Retirement benefits are calculated by multiplying years of service by 2.5 percent of final basic pay. The annual benefit is 50 percent of final pay for those with twenty years of service and increases to the maximum, 75 percent, for those with thirty or more years. Military retirees who are 62 or older are also entitled to Social Security. Military retirees also receive primary and dependent medical care assistance, commissary privileges, low-cost housing loans and life insurance. Most retired military personnel take civilian jobs which provide income (and may eventually provide retirement benefits) in addition to their military pensions. Members who are physically unable to continue service receive disability benefits. Those who have retired or are eligible to retire can elect to receive survivor benefits. Survivor benefits are equivalent to 55 percent of 16

31 retired pay. Such protection requires a small reduction in the normal military retirement benefit. Table I-9 shows that beneficiaries and benefits under the MRS program have increased rapidly during the past twenty-five years. Real average benefits also increased by almost 15 percent. In sum, the benefits from all pension programs have increased significantly in the last twenty-five to thirty years. Many more people now receive benefits and the benefits they receive are larger. This growth has led to greater financial independence and a higher standard of living for many of the elderly. Financial Assets and Other Wealth--Many individuals accumulate financial assets and other forms of wealth throughout their working careers. Most of the elderly receive some cash income from their assets, such as interest earned on savings accounts or stock dividends. Elderly families also hold assets that provide economic value--for example, homeownership. Homeownership, along with other assets such as business equity or life insurance, may provide little or no direct income but could potentially be converted into income. Finally, an increasing proportion of the elderly receive income from individual pension arrangements such as IRAs and Keoghs. Table I-10 shows a breakdown of the types of assets owned by families with a head-of-household who was age in Of this group, 89 percent owned some type of asset. About 80 percent of the single persons and 94 percent of the married couples had assets. Liquid assets, such as funds in checking accounts, savings accounts, stocks and bonds, were the TABLE I-9 Benefits under the Military Retirement Systemm1955 to 1979 Beneficiaries Total Benefits Average Average Benefit Year (thousands) (millions) Benefit (1979 dollars) Real $ 419 $2,341 $7, ,711 6, ,384 2,860 6, ,849 3,686 6, ,073 6,242 5,811 7, ,286 10,279 7,993 7,993 Source: U.S. Department of Defense, Defense Manpower Data Center, Office of Actuary. 17

32 TABLE 1-10 Percentage of Families with Family Head Age 64 to 69 with Assetsm 1975 Median Value of This Asset for Percentage Those with Asset Type of Asset Holding (1979 dollars) Liquid Assets 81% $ 7,300 Life Insurance, Annuities 75 5,1001 Home Equity 69 27,000 Illiquid Assets 24 l 3,500 Any of Above Assets 89 33,100 Sources: Joseph Friedman and Jane Sjogren, "The Assets of the Elderly As They Retire" (Cambridge, Mass.: Abt Associates, Inc., 1980), pp. 15, 36, 46, 49 and 66. Median value of assets in 1979 dollars estimated by ICF Incorporated. _This estimate reflects the cash value rather than the face value of these policies. most widely held asset type. Life insurance was the second most commonly held asset. Home equity was the third most common asset and for most individuals, it was the largest single asset. The majority of elderly families do not have mortgages on their homes, t5 About one-quarter of the elderly have illiquid assets, such as business or real estate equity. Looking at all elderly households headed by individuals age 65 and over, Table 1-16 shows that in 1979, 70 percent received cash income from financial assets. Most of this income was in the form of interest payments from savings accounts, bonds or money market funds. In 1979, about two-thirds of all elderly households received $1,000 or less in cash income from wealth, while 15 percent received more than $6, Of those who did not receive cash income from financial assets, some have other assets such as homes, which provide in-kind income and could also be converted into cash income. Some have business or real estate equity; this too could be converted into cash income. Available data suggest that some elderly families rely heavily upon personal assets for income or in-kind benefits. The fact that over two-thirds own homes suggests further that a large portion of the elderly's assets may not have eroded by recent inflation. _SJoseph Friedman and Jane Sjogren, "'The Assets of the Elderly As They Retire" (Cambridge, Mass.: Abt Associates, Inc., 1980), p _6ICF analysis for EBRI of the March 1980 Current Population Survey data. 18

33 Employment Earnings --Since 1950, employment earnings have declined in importance as an income source for the elderly. This resulted from lower labor force participation of older workers, made possible primarily by Social Security benefit increases, expansion of employer pension programs and introduction of other programs for the elderly. Table I- 11 shows that labor force participation of elderly men has declined sharply since 1950, particularly for those over age 60. From 1950 to 1980, men age 65 and over reduced their labor force participation by one-half. From 1970 to 1980, labor participation rates for male workers age declined by almost 19 percent; and participation rates for those age declined by roughly 8 percent. From t950 to 1970, labor force participation rates for women age increased from 27 percent to 43 percent; but participation rates for women over age 55 have declined since Older workers are more likely to work part-time and in certain industries and occupations. As indicated in Table 1-12, in 1977, the percentage of part-time workers increased after age 60. Of those over age 65, more persons work part-time than full-time. Older workers are more likely to be selfemployed or to work in the services industry; they are also more likely to be in managerial and administrative occupations. Older workers are less likely to work in industries such as mining or construction. TM In 1979, elderly households with work income shown in Table I-1 received approximately $9,500 per household in employment earnings. This is less than the average earnings of nonelderly households, primarily because so many elderly work part-time. However, when combined with the other available income sources, it contributes substantially to the elderly's financial requirements. TABLE 1-11 Labor Force Participation Rates for Mental950 to 1980 Age Group % 88% 90% 83% and over Source: U.S. Department of Labor. Bureau of Labor Statistics. ttebri, Coverage and Bene[it Entitlement, p. 57. _sicf analysis of the May 1979 Current Population Survey data. 19

34 TABLE 1-12 Percentage of Workers Who Work Part-Timem and Over Men 4% 4% 10% 48% Women Total 14% 14% 20% 54% Source: U.S. Department of Labor, Bureau of Labor Statistics, "Work Experience of the Population in 1977," Special Labor Force Report no. 224, p. A-7. Public Assistance and Other In-Kind Benefit Programs Although Social Security, employer pension programs and personal assets have increased rapidly and improved the elderly's financial independence, some elderly groups do not possess these resources. In response to this problem, Congress has enacted a wide range of programs to assist lowincome and disabled elderly individuals. Six major programs have been enacted since In 1980, these programs provided over $40 billion in benefits to the elderly. Medicare provides in-kind benefits to 95 percent of the elderly. Three other programs, Medicaid, food stamps and housing assistance, provide noncash or in-kind benefits. The other two programs, Low-Income Energy Assistance and Supplemental Security Income, provide cash income to low-income elderly. Medicare--As shown in Table 1-13, Medicare is the largest of these programs, in terms of both the number of participants and the level of benefits. Medicare has two parts: Hospitalization Insurance (HI), or Part A, covers inpatient hospital services and posthospital care provided by skilled nursing facilities and home health agencies; and Supplemental Medical Insurance (SMI), or Part B, covers physician care, related supplies and services, outpatient hospital care and home health services. The HI program is financed by payroll taxes. Medicare recipients pay premiums that help to finance SMI. In 1979, HI paid out approximately $18 billion and SMI paid out $7 billion on behalf of the elderly. _9 To become eligible for Medicare, workers must be age 65 and eligible for Social Security or Railroad Retirement benefits; they do not actually have to _"Thomas C. Borzilleri, "In-Kind Benefit Programs and Retirement Income" (Working Paper prepared for the President's Commission on Pension Policy, Washington, D.C., 1980), p

35 TABLE 1-13 Public Assistance and In-Kind Benefit Program for the Elderlym1980 Elderly Individuals Level of Average Participating Benefits Benefits per Programs (millions) (millions) Participant Medicare 24.5 $28,300 $1,200 Medicaid 5.1 4, SSI 1.9 2,400 1,300 Subsidized 1.3 1,600 1,200 Housing Energy Assistance Food Stamps Source: Participant estimates primarily from Thomas C. Borzilleri, -In-Kind Benefit Programs and Retirement Income" (Working Paper prepared for the President's Commission on Pension Policy, Washington, D.C., 1980). retire to receive benefits. Medicare covers early retirees and their spouses once they reach age 65, but not before. Those who are under age 65 and are eligible for Social Security disability for twenty-four consecutive months, as well as insured workers and their dependents with permanent kidney failure, are also eligible for Medicare. There are no income or asset tests associated with the program. Medicaid--Medicaid is the second most important noncash program providing benefits to the elderly. It offers health assistance to low-income individuals and families and to blind and disabled persons. The program is financed jointly by the states and the federal government. Each state may elect whether to participate and administer its own program. Arizona is the only state which does not participate. The federal government requires the provision of certain services from every state's Medicaid program; these provisions include inpatient and outpatient hospital care, physician services, laboratory and X-ray services, skilled nursing care, home health care and physical examinations for children under age 21. States may also provide other coverage, such as clinic services, prescription drugs, dental services and eyeglasses. Many Medicare recipients _irealso eligible for Medicaid. For low-income Medicare recipients, Medicaid pays Part B premiums as well as deductibles and coinsurance. 21

36 Medicaid provides assistance to two elderly groups. One group, categorical beneficiaries, comprise about two-thirds of all elderly Medicaid beneficiaries. They are low-income people receiving Supplemental Security Income. However, not all SSI recipients are beneficiaries. In thirty-five states and the District of Columbia, SSI recipients are automatically eligible for Medicaid. Eligibility standards in the other fourteen states with Medicaid programs are more restrictive than federal standards. Each state determines its own Medicaid eligibility criteria. The second elderly group eligible for Medicaid assistance is the medically needy. These recipients have incomes or assets that exceed SSI eligibility criteria, but they also have very high medical bills. These recipients are permitted to deduct their high medical expenses from their incomes or assets in order to satisfy low-income eligibility standards. Thirty states provide a medically needy program. Housing Assistance Programs--In 1978, there were approximately 1.2 million subsidized housing units that were occupied by elderly persons, z There are presently over a dozen federal housing assistance programs that provide benefits to the elderly. Under public housing, the largest program, elderly families rent housing units at below market prices. In 1978, approximately 500,000 public housing units were occupied by families with a head-of-household who was over age 62. Public housing rents cannot exceed 25 percent of an elderly family's income. Local public housing authorities administer the programs and set income eligibility criteria, zl Section 8 is the second largest housing assistance program; it provides rental subsidies for 400,000 elderly-occupied housing units. 2z As in public housing, rents cannot exceed 25 percent of the occupants' income, and program eligibility is determined locally and is a function of income. Section 202 and 236 programs are the other major housing programs. These provide direct assistance, either through loans or interest subsidies, to developers. This assistance lowers occupants' rent. Program eligibility is determined locally and is limited to lower-income families. Food Stamps--Nearly one million elderly households receive food stamps.23 Eligibility, in 1980, required that a single elderly person's income did not exceed $306 monthly or $3,672 annually. For two-person households with at least one elderly person, income limits were $403 monthly and $4,836 annually. In calculating household income, the elderly are permitted to deduct all shelter expenses such as fuel costs. Household assets, excluding 2Olbid,, p _lbid., p bid. 231bid., p

37 the value of automobiles up to $4,500 and the value of one's home, household goods and life insurance, cannot exceed $1,750 for single elderly persons or $3,000 for couples. Low-Income Energy Assistance Program--The Federal Low-Income Energy Assistance Program is designed to help the low-income population with energy expenses. It has undergone substantial changes since its inception in The states actually design and administer benefit distribution. In some states, benefits are paid directly to households; in others, payments are provided to the utility companies and fuel vendors. Some states require application while others provide automatic benefits to recipients of other welfare programs. Supplemental Security Income Program--SSl was implemented in 1974, is federally administered and provides cash assistance to low-income aged, blind and disabled adults. Those with less than $258 in unearned monthly income or $561 in earned monthly income are eligible for SSI. SSI eligibility criterion requires that individuals cannot have more than $1,500 in assets and couples cannot have over $2,250. In determining eligibility, the value of one's home, life insurance policies with face values below $1,500, up to $6,000 of nonliquid income-producing investments, automobiles valued at up to $4,500 and household goods valued at up to $2,000, are disregarded. Program benefits are indexed to CPI increases; in 1980, the maximum monthly benefit was $238 for individuals and $357 for couples. Thirty-seven states supplement the basic federal benefit. Income from other sources reduces SSI benefits; however, the benefit calculation disregards the first $20 of monthly income (regardless of source), plus the first $65 of earned income and one-half of earnings above $65. Cash and in-kind benefits from other needs-based programs are also disregarded. Cash benefit payments in the SSI program rose from $5.2 billion in 1974 to $7.9 billion in Based on Social Security Administration calculations, the number of elderly SSI recipients has decreased from 2.3 million in 1974 to 1.9 million in _5However, the Social Security Administration classifies SSI recipients according to their initial basis for entitlement. Thus, at the beginning of 1974 there were 87,000 recipients classified as blind and disabled who were also over age 65. By the end of 1979, the number of elderly blind and disabled SS recipients had grown to 386,000--an increase of 299,000. =6 When these elderly SSI recipients are combined -"_U.S. Social Security Administration, Social Security Bulletin. vol. 44, no. 10 (October 1981). table M-I. p '_U. S. Social Security Administration, Social Security Bulletin. vol. 44. no. 4 (April 1981 ), table M-19, p. 55. '61bid.. table M-18, p

38 with those classified as "aged," the total number of elderly SSI recipients rises to 2.3 million in December 1979, roughly the same as the 1974 level. Other Programs--In 1965, the Older Americans Act was passed to respond to a lack of available community social services. A major portion of this program's funds are used for three functions: Area Planning and Social Services ($153 million in 1979); Nutrition Programs ($250 million in 1979); and the Senior Community Services Employment Program ($201 million in 1979). The nutrition program provides low-priced meals to individuals age 60 and older. In 1978, an estimated 2.7 million persons participated in the program at over l, 100 nutrition sites. Participants are asked to pay from 25 cents to 88 cents per meal. The program does not require an income test. 27 Current Distribution of Income As previously noted, over $200 billion is devoted annually to the retirement income needs of the elderly. In 1979, the median cash income of elderly households was $7,900, while the average was approximately $11,000. There is a great deal of variation around this average, however. This section analyzes the variation in income distribution patterns among elderly households in 1979, income sources by income category and each source's relative contribution to total income. 2s In-kind benefit effects and the potential effects the elderly's financial assets could have on income are also discussed. Cash Income--Elderly couples have substantially higher cash incomes than elderly single persons. 29 Table 1-14 shows that in 1979, about 55 percent of all elderly couples had cash incomes greater than $10,000, but less than 22 percent of elderly individuals had this income level. In part, married couples have higher cash incomes because they (1) frequently have two income sources; and (2) typically receive higher Social Security benefits than single persons. Age also affects income distribution. Table 1-15 shows that in 1979, couples and individuals who were under age 72 had higher cash incomes than those over age 72. Of those age 65-72, 63 percent of married couples and 25 percent of single persons had cash incomes above $10,000; this compares with only 44 percent of couples and 19 percent of individuals over 27Borzilleri, "'In-Kind Benefit Programs and Retirement Income," p SAppendixD contains more detailed tables. 29Cash income is defined as employment earnings, employer pension benefits, Social Security and cash payments from other government programs. It excludes the imputed value of homeownership and in-kind benefits. 24

39 TABLE I- 14 Cash Income Distribution of the Eiderlym19791 Single All Elderly Married Couples Persons Households Less than $2, % ( 2.0) 9.1% ( 9.1) 5.8% ( 5.8) $2,500-4, (12.7) 39.4 (48.5) 26.1 (31.9) $5,000-9, (44.6) 29.8 (78.3) 30.8 (62.7) $10,000-14, (68.1) 10.2 (88.5) 16.4 (79.1) $15,000-24, (86.9) 7.4 (95.9) 12.6 (91.7) $25,000 and over 13.2 (100.0) 4.0 (100.0) 8.2 (100.0) Total 100.0% 100.0% 100.0% Source: ICF analysis for EBRI of the March 1980 Current Population Survey data. _Cumulative distribution in parentheses. TABLE 1-15 Cumulative Cash Income Distribution of the Elderly, By Marital Status and Age Single Married Couples Age Persons Age Over Over 72 Less than $2, % 2.1% 9.1% 9.1% Less than $5, Less than $I0, Less than $15, Less than $25, Total 100.0% 100.0% 100.0% 100.0% Source: ICF analysis for EBRI of the March 1980 Current Population Survey data. age 72. Employment income earned by those age explains most of this difference. Much of the income variation is explained by income sources. Highincome elderly families generally receive income from different sources, and from more sources than.low-income families. Social Security income was received by over 90 percent of all households. As shown in Table I- 16, except for those with the lowest and highest incomes, more than 90 percent 25

40 fll f_ fml I.-, _ 26

41 of elderly households within each income classification also received Social Security. Some higher-income households do not receive Social Security because they are still working and receive significant employment earnings, or they have not worked in Social Security-covered employment and, therefore, do not qualify for benefits. However, higher income households frequently receive pensions and income from financial assets. Many lowincome households are not eligible for Social Security because they lack sufficient quarters of covered employment. Table I- 16 also shows that in 1979, 75 percent of elderly households with cash incomes above $25,000 received employment income. Of those with cash incomes of $10,000 to $25,000, approximately half received employment earnings and approximately half did not. Most of the lowest income households did not receive employment income. The effect of employment income on the cumulative distribution of cash income is shown in Table Only one-third of elderly households with employment earnings have cash incomes below $10,000, while threequarters of those without earnings have cash incomes below that level. Table I- 18 shows that married couples are more likely to receive employment earnings than single persons. Additionally, older families are less likely to have full-time workers and employment income. Only 3 percent of married couples and 1 percent of single persons over age 73 work in full-time jobs. Table I- 19 shows the effect of marital status and work status on income distribution. If marital status and work status are held constant, the differ- TABLE 1-17 Cumulative Distribution of Cash Income for Elderly Households, By Earnings Statusm1979 With Without Employment Employment Earnings Earnings Total Less than $2, % 7.8% 5.8% Less than $5, Less than $10, Less than $15, Less than $25, Total 100.0% 100.0% 100.0% Source: ICF analysis for EBRI of the March 1980 Current Population Survey data. 27

42 TABLE 1-18 Working Status of the Elderly, By Age and Marital Statusm 1979 Married Couples SinglePersons WorkStatus and Over and Over Full-Time 14% 6% 3% 7% 4% 1% Part-Time No Work Total 100% 100% 100% 100% 100% 100% Source: ICFanalysisforEBRIof the March1980CurrentPopulationSurveydata. ences in income distribution by age are reduced substantially. However, younger couples and individuals still have higher cash incomes. Equally large differences exist in the relative contributions of different retirement income sources to total income. As shown in Table 1-20, on average, the highest income households receive almost one-half of their cash income from employment. Additionally, Social Security is the largest cash income source for all the elderly, particularly those with cash incomes below $15,000. In-Kind Benefit Values--The cash income estimates presented above do not include the value of all benefits provided to the elderly. Elderly households receive substantial noncash or in-kind benefits from programs such as Medicare and Medicaid. Table I-21 shows that approximately 90 percent of elderly households, in each income bracket, receive Medicare coverage. Of those with incomes below $5,000 in 1979, approximately 30 percent were also enrolled in Medicaid. Sixteen percent of lower-income households receive food stamps and 12to 13 percent receive subsidized housing or other housing assistance benefits. In 1979, more than 95 percent of elderly households with incomes below $5,000 received some in-kind benefits. To analyze the effect of in-kind benefits on the elderly's economic status, a data base was developed, which incorporated in-kind benefit values so that the elderly's total cash and noncash income distribution could be analyzed. 3 The analysis was based upon the March 1980 Current Population Survey (CPS). This Bureau of the Census survey contains information on the characteristics of over 13,000 representative elderly households. After estimating the cash income of each surveyed household, those with specific 30Thismethodologyis describedindetail inappendixb. 28

43 29

44 Lm u c_ m _ c_ j_ _ 30

45 TABLE 1-21 Percentage of Elderly Households Receiving Certain In-Kind BenefitsB1979 Food Housing Medicare Medicaid Stamps Assistance Less than $2,500 88% 30% 16% 12% $2,500-4, $5,000-7, $7,500-9, $10,000-14, $15,000 and over All Elderly 93% 16c/c 6% 5% Source: U.S. Department of Commerce, Bureau of the Census, Characteristics _/'Household_ and Persons Receiving Non-Cash Benefits: 1979, Current Population Reports, series P-23, no. 110, pp. 12, 15, 16 and 18. in-kind benefits were identified. Next, the in-kind benefit's value was estimated and that value was assigned to the couple or individual. There are different methods of valuing in-kind benefits. Some think they should be valued at the cost of providing them. Others think they should be valued at their cash equivalent value (i.e., the cash amount that would leave the recipient equally well off as the in-kind benefit). Most analysts believe that the "cash equivalent value" is a more appropriate method; however, it is difficult to measure because different individuals place different values on different services. For example, some place higher values on food than on housing, and vice versa. This report provides estimates of the cash equivalent value of Medicare, Medicaid, food stamps and housing assistance benefits. The methodology utilized was originally developed by Smeeding and Moon. 31The methodology assumes that the value of each family's benefits is related to family size and income; thus in-kind benefit values are estimated separately for different income and family size groups. The value of a specific in-kind benefit to each family size/income group is equal to the smaller of: (1) the estimated cost to the government for providing the service or good; or (2) the average expenditure incurred in the purchase of that good or service by families of 3_Timothy Smeeding and Marilyn Moon, "'Valuing Government Expenditures: The Case of Medical Care Transfers and Poverty." Review of Income and Wealth, series 26, no. 3 (September 1980), pp

46 equal income and size, based on the Consumer Expenditure Survey (CES) data. For example, the CES data may indicate that, on average, two-person families with $4,000 incomes spend 24 percent of their income on food. If the estimated cost to government of providing such a family with food stamps is less than $960 (24 percent of $4,000), then the government's cost is used as the food stamps' cash equivalent value. Conversely, if the estimated government's cost is over $960, then the average expense of $960 is used. This procedure is utilized separately in calculating the cash equivalent value of food stamps, housing assistance and medical benefits. Based upon this methodology, Table 1-22 demonstrates that including in-kind benefit values changes the elderly's income distribution patterns. Specifically, it reduces the proportion of elderly households with incomes below $5,000 from 32 percent to 26 percent. 32 Family Asset Values--Personal assets are also a significant factor in calculating elderly income. Therefore, this report estimates the potential TABLE 1-22 Cumulative Income Distribution of Elderly Households, Including Cash Income and In-Kind Benefits Cash Cash Income and In-Kind Benefits Income Excluding Including Only Medical Benefits Medical Benefits Less than $2, % 4.9% 3.6% Less than $5, Less than $10, Less than $15, Less than $25, Total 100.0% 100.0% 100.0% Source: ICF analysis for EBRI of the March 1980 Current Population Survey data. See Appendix B for details. _-'lf these in-kind benefits were valued at cost. the cumulative percentage of families with incomes: less than $2,500 would total 1.4 percent; less than $5,000 would total 15.6 percent; less than $10,000 would total 54.7 percent; less than $15,000 would total 75.3 percent; less than $ would total 90.8 percent. 32

47 effect such assets could have on income distribution, assuming the assets were converted into indexed annuities. 33 The March 1980 CPS data was used to identify elderly homeowners as well as those with dividend and interest income. 34 Based on mortality and interest assumptions, these three asset types were converted into annuities with an annual indexed increase of 5 percent. Additionally, married couples were assumed to receive joint and survivor annuities, which when one spouse died, provided the survivor with a benefit that equaled one-half the annual annuity payment that had been paid to the couple. Based on CPS data, the actual ages of husbands, wives and single persons were used to calculate these annuities. Table 1-23 shows that when the value of the elderly's in-kin d benefits and assets are both included in the income definition, the percentage of elderly households with incomes below $5,000 decreases from 32 percent to 17 percent. However, annuitizing asset values has an even greater effect on households with cash incomes above $5,000. TABLE 1-23 Cumulative Income Distribution of Elderly Households, Including Cash Income, In-Kind Benefits and Annuitized Wealth Income Cash Income, In-Kind Benefits and Cash Annuitized Wealth Income Excluding Including Only Medical Benefits Medical Benefits Less than $2, % 2.9% 2.3% Less than $5, Less than $10, Less than $15, Less than $25, Total 100.0% 100.0% 100.0% Source: ICF analysis for EBRI of the March 1980 Current Population Survey data. See Appendix B for details. 33Thisanalysis attempts to provide a perspective of the cash income potential of personal assets. In evaluating this information, it should be understood that some elderly are unable or unwilling to convert personal assets into cash. 34Thismethodology is also described in Appendix B. 33

48 Employer Pension Benefit Trends As previously discussed, employer pension benefits are an increasingly important elderly income factor. Employer pensions include benefits from private employer retirement plans (including Keoghs), state and local government pension plans, the Civil Service Retirement System and military pensions. In 1979, married couples and single persons with pension benefits had higher total incomes than those without pensions. Table 1-24 shows that approximately 44 percent of elderly married couples and 26 percent of elderly single persons received employer pension benefits. To examine recent employer pension trends, analysis was developed on the historical rates of pension receipt for two groups: those age and those over age 75. Differences in the rate of pension receipt between these two groups reflect whether a larger or smaller percentage of newer retirees are receiving pension benefits. As shown in Table 1-24, the percentage of couples and single persons age receiving pensions is higher than the percentage of those over age 75 with pensions. Forty-six percent of younger married couples received pensions, while only 39 percent of older couples received employer pensions. This difference is primarily attributable to increased benefit receipt from private pension plans. The other pension sponsors did not exhibit as significant an increase. The increase in benefit receipt is largely due to rapid growth of private plan coverage over the last twenty-five years and the liberalization of vesting provisions. TABLE 1-24 Percentage of Elderly Receiving Employer Pension Benefitsu1979 Married Couples Age Single Persons Age Type of 75 and 75 and Employer Pension Over Total Over Total Private 35% 27% 32% 17% 14% 16% State and Local Government Federal Military Any Employer Pension 46% 39% 44% 28% 23% 26% Source: ICF analysis for EBRI of the March 1980 Current Population Survey data. 34

49 Important questions are: (1) Will this increasing pension receipt trend continue? (2) What effect will it have on the size of pension benefits? To analyze recent trends more carefully, a microsimulation model was used to estimate: (1) the potential future rate of pension benefit receipt; and (2) the expected level of pension benefits under alternative workforce, economic and pension coverage assumptions.3s Estimates were developed for a representative sample of families; the.sample contained individuals age in The model simulated each individual's work and wage histories, their pension coverage and retirement benefits. Appendix C contains a description of the model and the assumptions used in this analysis. For the age group, the model simulated the percentage of elderly receiving benefits and the size of their initial benefits under three pension coverage cases. In Case 1, employer pension coverage in each industry was assumed to remain at a constant level of 66 percent from 1979 to In Case 2, pension coverage was assumed to increase from 66 percent in 1979 to 75 percent in In Case 3, pension coverage was assumed to increase from 66 percent in 1979 to 79 percent in These assumptions are not intended as forecasts. Instead, they are illustrative of a potential range of estimates for future pension coverage growth. Table 1-25 shows that the percentage of couples age 65-69, who will be eligible for employer pension benefits, will probably increase from the current 53 percent level to 85 to 90 percent by For single persons, the results are similar--employer pension benefit receipt should increase from 33 percent to 66 to 73 percent by These results are illustrated in Figure 1-1. Although the proportion of families eligible for employer pension benefits may double in the next twenty to twenty-five years, this is not likely to affect significantly the average benefit level for those receiving pensions. This is because many beneficiaries will be widows with survivors benefits, or workers whose wages are at or below the average wages that were earned by today's retirees. Additionally, many will receive benefits based on periods of service that are shorter than those of the 1979 retired population. Nonetheless, Table 1-26 shows that during the next twenty to twenty-five years, the role of employer pensions is expected to increase substantially. As discussed above, this results primarily from the increased proportion of families with employer pensions. _SThismodelwas originallydevelopedby ICFIncorporatedfor the Office of Pensionand WelfareBenefitProgramsofthe Departmentof Laborand the President'sCommissionon PensionPolicy.It has recentlybeenextensivelyrevisedforthe AmericanCouncilof Life Insurance.It is describedin AppendixC. 35

50 TABLE 1-25 Percentage of Families with Individuals Age 65 to 69 Eligible to Receive Employer Pension Benefits _ Case 1 Case 2 Case 3 Married Couples 29% 53% 85% 88% 90% Single Persons All Families 22% 42% 77% 81% 83% Sources: 1967 values from U.S. Social Security Administration, Demographic and Economic Characteristics of the Aged (1975), p ICF analysis of March 1980 and May 1979 Current Population Survey data for 1979 values. ICF estimates for See Appendix C for methodology and assumptions. _Values for 1979 are higher than the percentage of elderly currently receiving benefits because some individuals age have a vested right to employer pension benefits but have not accepted them. Estimates for 2004 are for those who will be eligible to receive benefits, not necessarily for those receiving them. TABLE 1-26 Projected Employer Pension Benefits for Families with Individuals Age 65 to 69m1979 and 2004 _ Percentage of Average Pension Average Pension Families Eligible Benefit Per Benefit for All to Receive Recipient Aged Families Pension Benefits Family Married Couples 1979 $2,500 53% $4, , ,400 Single Persons 1979 $1,200 33% $3, , ,800 All Families 1979 $1,800 42% $4, ,1 O0 81 3,900 Source: 1979 values from ICF analysis of March 1980 Current Population Survey data. ICF estimates for 2004 under coverage Case 2. See Appendix C for methodology and assumptions. _Estimatcs are in 1979 dollars. Estimates are not directly comparable because estimates for 1979 reflect actual benefit payments while estimates for 2004 reflect amounts families will be eligible for, if they accept benefits at age

51 37

52 In this analysis, an annual average real wage increase of 1 percent was assumed. Real pension benefit formulas were not assumed to increase. If real wages should increase by more than 1 percent, or if real benefits increase, employer pension benefits may be larger than those estimated in the analysis. In addition, these estimates do not include the potential effects of tax deductible employee contributions or the effects of increased IRA and Keogh maximum contribution limits resulting from the 1981 Economic Recovery Tax Act (ERTA). These recent developments are expected to increase the size of employer pension benefits. On the other hand, these estimates assume that workers do not cash out vested benefits upon job change and that they will receive unreduced pension benefits. If workers do cash out vested benefits or if they retire early, before eligibility for unreduced benefits, the estimates presented here may be overstated. Increases in the percentage of families who will be eligible for employer pension benefits reflect primarily four key factors: (1) The continued maturing of employer retirement programs. An increasing proportion of workers receive the opportunity to spend a significant portion of their careers in covered employment. (2) Liberalized vesting standards in private sector plans: this increases the proportion of covered workers expected to receive benefits, as well as some workers" length of covered service. (3) Significant increases in the number of two-worker families. This increases the likelihood that at least one, and often two, family members will receive pensions. Increases in female labor force participation also lead to an increase in the number of single persons (primarily widows) with pension benefits based on their own workforce experience, (4) An expanded prevalence of joint and survivor options. As more workers become vested and elect these options, more individuals become eligible for pension benefits based on their spouses" work records. These trends are significant. They suggest that a careful evaluation of retirement income patterns must consider the current rate of pension benefit receipt, as well as the expected future rate. Pension benefit receipt, especially private plan benefit receipt, has grown rapidly in the last twenty years and should continue to do so in the future. 36 _'An earlier EBRI study reviewed the rapid increases in private plan coverage over the last thirty years. Based on these coverage increases, alone, pension receipt is expected to increase for at least the next ten to fifteen years. For a more detailed discussion, see EBRI, Coverage and Benefit Entitlement. 38

53 II. Income Adequacy under Alternative Standards Recent public discussion reflects that there are differences of opinion regarding the definition of an appropriate standard of income adequacy for the elderly. In part, these differences exist because retirement income standards have different functions. One function involves the establishment of overall public policy goals, e.g., the poverty level, the minimum income level established by SSI and the President's Commission on Pension Policy's proposed replacement rate goals. Income adequacy standards offer a frame of reference for evaluating and designing welfare and benefit programs. Some standards, such as replacement rates, are also used to help individuals plan for retirement and select appropriate retirement programs. Some employer pension programs aim at replacing a specified percentage of full-time workers" preretirement income and others aim at providing a specified annual income. Finally, income standards are helpful in identifying problems with present retirement income programs and policies. Although it may not be possible to achieve a consensus regarding appropriate retirement income levels, alternative standards can help in making useful comparisons among relevant groups. The poverty standard, for example, may be too high or too low; nonetheless, it enables evaluation of the relative condition among different population groups. Retirement income standards are used here primarily to identify major problem areas. This report attempts to assess the elderly's well-being by examining: (1) the elderly's ability to consume essential goods and services; and (2) the elderly's degree of financial independence resulting from their wide range of available economic resources. As previously noted, a problem in choosing appropriate elderly income adequacy goals is that all presently established adequacy goals are built around cash income measures. Using cash income alone to evaluate the elderly's well-being excludes: (1) in-kind benefit values and (2) asset values. _ Cash income measures ignore that most elderly own their homes, the inherent value of homeownership and the potential income of converting such assets into an annuity. This report attempts to provide some perspective on these problems by estimating the elderly's in-kind benefits and the potential family income that could be derived from assets. _Theelderly's cash income usuallyincludesemploymentearnings, incomefrom Social Security.employerpensionbenefitsand othercash transfers. 39

54 Cash income measures do not adequately capture two important changes of the past twenty years: (1) a reduction in risk the elderly J_tce--The risk of destitution is reduced. Medicare, Medicaid, food stamps, housing assistance, Low-Income Energy Assistance and SSI all contribute to the elderly's improve welfare. These programs ensure that the poorest families will receive a minimum level of health care, food, shelter, fuel and income. They also insure against risk of catastrophic events by providing a floor of income and services. Additionally, some of the risk of inflation has diminished. For example, in 1979, elderly families with total incomes of less than $8,000 had more than 75 percent of their income indexed to the CPI? Although inflation may erode the value of income for the low-income elderly, full indexation of Social Security and SSI reduces this problem substantially. (2) increase in choices of retirement age--spurred by liberal early retirement provisions in employer pension programs, as well as in the Social Security program, there has been a significant decline in workers' retirement ages. Such increases in retirement age options have also contributed to the elderly's improved welfare. Finally, average household cash income measures may be misleading unless analyzed along with other considerations. For example, many elderly prefer to live alone, but if they live away from relatives, their overall household income is less than it would be if they lived with relatives. Although these changes do not expand cash incomes, they have had an important effect on the elderly's overall improved circumstances. Alternative Adequacy Standards Three primary standards are used in evaluating the elderly's income adequacy: --Relative standards--for example, the replacement rate is commonly used to measure the relationship between standards of living before and after retirement. --Absolute standards--a variety of standards measure the relationship between a consuming unit's income and a specified income standard. -- We,hh st_m_h_rds--measure the amount of wealth required to sustain individuals during retirement or old age. Each of these standards has conceptual and practical problems. Relative standards, such as the wage replacement concept, may be inappropriate for 2u.s. Congress. Congressional Budget Office, -Low-lncome Energy Assistance: Issues and Options" (June 1981), p O

55 evaluating the income of those without substantial lifetime work experience. Absolute income standards, such as the poverty index, are simple in approach, but they are based upon judgments regarding adequate expenditure levels for food, shelter, health care and similar necessities. These standards both fail to reflect variations in wealth. Wealth standards are useful because they incorporate all available income sources. However, defining an adequate wealth level at retirement requires a reference point based either on: (l) preretirement wealth, or (2) specification of an adequate level of wealth, which presents problems similar to those affecting relative and absolute income standards. Finally, all these standards share a common problem--they are based upon value judgments. Relative Standards--The replacement rate concept compares an individual's or family's retirement income to its preretirement income. Replacement rates are often used as a retirement planning guide, and some have proposed using replacement rates as an adequacy standard. Table II- 1 shows the sensitivity of the income replacement level to: (1) specific retirement income sources included in the fate's numerator, and (2) preretirement TABLE II-1 Pretax Replacement Rates for Single Persons with a Final Salary of $15,000 Measure of Retirement Benefits Measure of Pre- (in numerator) retirement Income OASI and OASI, Pension (in denominator) Employer Pension and Other Income Salary Only Final Five-Year Average 70.9% 87.0% Final Year All Income Final Year Salary and Employee Benefits Final Year Salary, Employee and Other Benefits Income Source: Towers, Perrin, Forster & Crosby, Business Roundtable Study of Retirement Levels. Costs and Issues (August 1978). Estimates for a single worker retiring on January 1, 1979 at age 65 with thirty years service, under a pension plan that provides a benefit of: 1.67 percent, x high five-year average pay, x years of service, less 1.67percent, x applicable Social Security benefit, x years of service. 41

56 income sources included in the denominator. When the numerator and denominator are defined consistently, replacement rates reflect the relationship between retirement income and preretirement living standards. When examining replacement rates, one must determine: --How the family unit is defined--for establishing overall elderly income goals, the family is the most frequently used measure. For program (e.g., Social Security, employer pension plan) objectives, it is usually more convenient to use the individual worker as a basis for replacement rates. --How retirement income is defined--when assessing the elderly's income, most analysts agree that Social Security, employer pensions, income from financial assets, cash transfers and earnings should be included in the retirement income definition. Most agree that retirement income should be measured after taxes. Not all agree, however, that in-kind benefits or family wealth should be included. --How preretirement income is defined Preretirement income commonly includes employment earnings, income from assets, other transfers and in-kind benefits. Some argue that work-related expenses and savings should be excluded. If one excludes taxes, work-related expenses and savings from preretirement income, the postretirement income needed to replace final earnings may decline substantially. --The preretirement income period--to reflect the income level, income is usually measured over a recent period. Otherwise, replacement rates will not reflect current living standards. Most applications of the replacement rate concept use a period that is more stable than the final year prior to retirement. 3 One drawback of using final year's income is that it may be particularly low, reflecting a period of low workforce activity due to poor health. -- Whether the numerator and denominator are defined consistently--whatever measures are used, if the numerator is an after-tax measure, the denominator should also be an after-tax measure. If preretirement income includes total family income, the denominator should also include family income. The most significant problem associated with the replacement rate concept is that little justification or rationale has been provided for an "adequate" replacement rate. Arguments for using replacement rates often presume that any reduction in living standards after retirement is "undesirable." Replacement rates also fail to reflect standard-of-living preferences over each individual's lifetime. Especially for middle- and higher-income groups, replacement rates result from individual choice, regarding work, savings and a tradeoff between wages and pensions. It is unclear whether 3u.s. Social Security Administration, "'The Concept of Replacement Ratios under Social Security," Actuarial Note no. 96, Joseph Ar Applebaum and Joseph F. Faber (July 1979), p

57 variance in replacement rates resulting from individual choice should be an issue of public concern. There is justification for public concern when individual planning or program design is upset by unanticipated economic events or government decisions. For example, many individuals retire with adequate incomes, but over time their incomes are eroded by unanticipated inflation. Additionally, many plan their retirement around current Social Security law. Sudden changes in this law could upset such retirement planning. Three other problems may make relative standards unworkable public policy goals: --They are sensitive to the timing of retirement income evaluations. For many, retirement income buying power declines during later retirement years. Replacement rates may show that some incomes are adequate in the first retirement year. but inadequate in subsequent years. --A specified replacement rate for all families, or a set of replacement rates, fail to reflect long-term real growth in workers" earnings. As real earnings grow, fixed replacement rates will provide greater real income. This may not be affordable or desirable. --For families with low preretirement incomes, a 100 percent replacement rate may be insufficient to provide adequate income. Absolute Income Standards--The elderly's income may also be evaluated relative to estimates of minimum expenditures required to maintain a household. The goal of ensuring minimum family incomes is commonly accepted. However, the problem of determining such a minimum has not been satisfactorily resolved. None of the standards discussed here are based upon a minimum subsistence level. Rather, they are based on judgments about minimal "adequate" income levels. Four standards are used to evaluate income adequacy and to establish benefit eligibility under income assistance programs. Table II-2 shows: (1)the poverty level; (2) the revised Orshansky index: (3) the Bureau of Labor Statistics (BLS) lower, intermediate and higher standard of living budgets for elderly couples; and (4) the 1979 minimum wage standard. The poverty level is lower than the other standards. The four major standards are described below: --Federal poverty levels--mollie Orshansky at the Social Security Administration developed the various levels under this standard. The government uses this as its official poverty stanqlard, and the Census Bureau uses this standard to estimate the number of persons in poverty. The indices approximate the money amount needed by families to achieve a minimum adequacy level. The poverty level presumes that family members form a single economic unit, and their combined incomes determine the family's economic status. The standard has 43

58 TABLE II-2 Alternative Adequacy Standards for the Elderly Couples Single Persons _ Poverty Level $ 4,508 $3,528 Revised Orshansky Index 6,085 4,868 BLS Low Budget 6, BLS Intermediate Budget 8, BLS High Budget 12, Minimum Wage 2 6,032 6,032 Sources: u.s. Department of Commerce, Bureau of the Census, Consumer Income: Characteristics of the Population Below the Poverty Level: 1978, Current Population Reports, series P-60, no. 124, p. 208; U.S. Department of Labor, Bureau of Labor Statistics, "'Three Budgets for a Retired Couple, Autumn 1978" (August 20, 1979); U.S. Department of Health and Human Services, Social Security Administration, Office of Research and Statistics, "Poverty Lines: Alternative SSA Poverty Matrix" (April 1980). _The Bureau of Labor Statistics' budgets are computed for couples only. 2Assumes forty hours per week for fifty-two weeks at $2.90 per hour. 124 indices, which vary by family size, number of children under age 18, sex and age of family head and farm/nonfarm residence. 4 The poverty definition is based on a hypothetical economy food plan, which is the least costly of four nutritionally sound food plans that were designed by the Department of Agriculture. From the Department of Agriculture's 1955 survey of food consumption, analysts estimated that families with three or more persons spend about a third of their after-tax income on food. Thus, the poverty level is set at three times the cost of providing the economy food plan to this size family. For smaller families and persons living alone, the factor is slightly higher than three in order to compensate for relatively larger fixed expenses of smaller households. The poverty level is updated annually to reflect price changes of food budget items. --Revised Orshanskv index--mollie Orshansky developed this in It is a revision of the official poverty level described above. This index reflects the more recent findings on poor families' expenditure patterns. These revisions are also based on the family unit concept, and they vary by the same family characteristics used in the official index. However, as shown in Table Ii-2, the Orshansky indices are percent higher than the official poverty level indices. Two factors contribute to this 4The elderly's standard is slightly lower than that of persons under age

59 variation: First, the Orshansky index uses a more recently developed food plan, which is 22 percent higher than that used in the official index. Second, as the result of a 1965food consumption survey, the original ratio of total expenditures to food expenditures was changed from 3 to 1 to 3.4 to BLS elderly couples budgets-- BLS prepare s annual estimates of the resource s required to purchase the typical market basket of goods consumed by elderly couples at low, intermediate and high budgets. The items included are determined from studies of food, housing and medical care needs as well as the elderly's consumption patterns as reported in the Consumer Expenditure Survey. Consequently, these standards are not estimates of family subsistence requirements. Rather, they reflect the costs to elderly couples of maintaining three alternative living standards. These three budget levels are defined in terms of purchased goods and services, rather than family income. It is assumed that: ( 1) The lower budget family lives in rental housing without air conditioning, relies primarily on public transportation (supplemented perhaps by an older car), performs more services for itself and uses free recreation facilities. (2) The intermediate and higher budget families enjoy progressively greater degrees of homeownership, more household appliances, more paid-for services and are more likely to have a new car. These budgets" assumed consumption patterns are shown in Table II-3. --Minimum wage--under this standard, the adequacy level is equal to the annual earnings of a full-time, full-year worker who is paid the minimum wage. This index offers the advantage of incorporating a direct relationship with the national policy, whictr defines the minimum acceptable hourly wage level. Adjustments are not made for variations in family characteristics. Consequently, this standard may be less sensitive than others to the unique circumstances of retired families. TABLE II-3 Expenditure Patterns Assumed in BLS Budgets for a Retired Couple BLS Budget Lower Intermediate Higher Food 32% 31% 27% Housing Transportation Clothing/Personal Care Health Care Other Items Total Consumption 100% 100% 100% Source: U.S. Department of Labor, Bureau of Labor Statistics, Interview Survey, Consumer Expenditure Survey Series, 1972and 1973report table 4B. 45

60 Except for the minimum wage index, these standards are based upon analyses of money amounts necessary to purchase a broadly defined market basket of goods and are somewhat arbitrary measures of adequate incomes. For example, the BLS budgets' assumed market baskets of goods are based on unsubstantiated judgments about food, medical care and housing requirements. Similarly, the poverty level implies that families with a female head-of-household require less food--an assumption that is not always valid. Additionally, the minimum wage is more politically than scientifically determined. Problems in defining meaningful income adequacy standards are illustrated by the large differences in these income standard amounts. Absolute standards offer the advantage of providing a minimum income floor, which can be used in assessing all household incomes regardless of family members' prior work histories. These standards also help in determining the number of people without sufficient income to attain some minimum standard of adequacy. However, although some of these standards vary somewhat with unique family circumstances, they do not adjust for variations in family health status. Moreover, unlike wage replacement rates, these measures do not indicate whether individuals' and families' living standards decline after retirement. Retirement Wealth Concepts--Chapter I discussed the income-producing assets and other wealth that many elderly families have. When comparing assets with annual income, it is helpful to convert annual income to a present value measure. Thus, retirement wealth would include an estimate of the present value of expected pension and Social Security benefits as well as home equity, savings and financial investments. Using such estimates, alternative adequacy standards can be developed without regard to the form of economic value available from each source. Wealth measures differ from the other adequacy concepts because, in evaluating retirement living standards, they explicitly include financial assets. These measures are particularly useful because living standards are influenced substantially by asset (e.g., home equity and durable goods) availability. They also permit assessment of retirement income adequacy over an individual's entire retirement period. However, wealth adequacy measures present the same analytical problems as income adequacy measures. To identify adequate retirement wealth levels, analysts must rely on definitions of preretirement living standards or minimum living standards. Additionally, these measures are very sensitive to assumptions about inflation, mortality and investment yields. Finally, it is difficult to incorporate employment earnings in this measure. 46

61 The Eiderly's Expenditure Patterns Some absolute income adequacy standards are based on the elderly's observed expenditure patterns. Replacement rate standards may also assume that preretirement living standards and expenditure patterns should be preserved during retirement. However, effective analysis must examine these patterns closely, because actual expenditure and savings patterns are substantially influenced by income levels, household size and family members' ages. Expenditure Levels--The elderly's expenditure levels and patterns differ from those of other groups due to their lower incomes, smaller family size, reduced labor force participation and overall tendency to spend less. The Consumer Expenditure Survey data suggest, for example, that households headed by 65 to 71-year-old persons spend about half as much on goods and services as households headed by persons age This finding is also true among strictly elderly groups. Table II-4 shows that families with heads-of-household who are over age 72, spend about 75 percent of the amount on goods and services as that spent by families with heads-of-household who are age These differences are partly explained by household size and composition; since elderly households are more likely to have only one member, they spend less. As previously noted, income is another important factor influencing family expenditures. After controlling for income, the elderly's household TABLE II-4 Family Expenditures By Age and Household Compositionm1972 to 1973 Age of Single Single Married Household Head Men Women Couples Average Less than 62 $8,255 $5,622 $11,578 $10, ,813 4,278 7,295 6, ,229 3,707 6,503 5, or more 2,881 2,907 5,207 4,100 Source: Sheldon Danziger, Jacques van der Gaag, Eugene Smolensky and Michael K. Taussig, "'A Consumption Measure of the Economic Status of the Aged," unpublished paper (August 1980), p. I 1. _This is the most recent cross-sectional data available on household expenditure patterns. 47

62 expenditures are a lower percentage of after-tax income than household expenditures of younger families. In other words, for a given level of disposable income, elderly families spend less and save more than younger families. This is true even when household size and work status are controlled 6 This finding, which indicates savings rates increase with age, is not consistent with the "life-cycle theory of consumption." This theory suggests that families save during working years and spend their savings during retirement. Generally, low-income elderly persons do spend more than their annual reported income--this can occur primarily because this group liquidates savings and other asset wealth or purchases items with credit. However, for higher-income elderly families, the savings rate becomes positive and continues to increase with income. When income and family characteristics are held constant, household saving rates among those with heads-of-household who are age 65 or more actually increase as family members' ages increase. That is, those over age 72 are saving more and consuming less than those age 65 to 71. This is consistent with Internal Revenue Service data showing that size of family estates grows with decedents' ages, There are several potential explanations for the elderly's increased savings pattern. For example: (1) Many elderly continue to save in order to leave a bequest for their family. (2) Retired persons have more time to shop for bargains and to perform household services than those who are working. (3) A decline in health that frequently accompanies old age reduces some elderly persons' mobility. Therefore, many do not desire, or are unable, to maintain previous spending patterns. (4) Today's elderly may be saving in response to growing uncertainty about their financial future. As individuals grow older and become less able to work, they may feel increasingly concerned that their total assets and retirement income are insufficient to sustain them throughout their remaining years. Increased savings levels can act as an insurance policy against unexpected long life. Expenditure Patterns--Table II-5 shows expenditure distributions for various goods and services purchased by families. The table provides this information for families of all ages, as well as elderly families, in As indicated, elderly families tend to spend more than average on necessities (e.g., food, housing and medical care) and less on nonessential items. 7 6Sheldon Danziger, Jacques van der Gaag, Eugene Smolensky and Michael K. Taussig, "A Consumption Measure of the Economic Status of the Aged," unpublished paper (August 1980), p. 9. 7Due to changes in relative prices, the market basket of goods and services consumed today differs from that reported in the 1973 CES. 48

63 TABLE II-5 Relative Importance of Expenditures for Selected Groups of Goods and Servicesm Type of Expenditure All Households Elderly Households Food 19.2% 22.7% Housing Medical Care Apparel Transportation Reading, Recreation Miscellaneous Total 100.0% 100.0% Source: Calculated from Thomas C. Borzilleri, "The Need for a Separate Consumer Price Index for Older Persons." The Gerontologist. vol. 18, no. 3 (1978), pp Expenditures as a percentage of total consumption outlays. These differences are largely explained by the elderly's overall lower incomes. Among all demographic groups, the proportion of income spent on necessities usually falls as income rises, and consumption of nonessential goods and services increases. Since elderly households have lower incomes, they spend larger than average proportions of income on essential goods and services. Family size and composition also have a substantial impact on the elderly's consumption patterns. For example, elderly married couples tend to spend larger portions of income on food, fuel and medical care than elderly persons who live alone. Elderly couples, however, spend smaller portions of income on shelter; this suggests that substantial savings are realized by sharing living facilities, s Although household size and income influence food expenditures, the elderly tend to maintain the same food consumption patterns as they grow older. The elderly's lower transportation expenses are largely explained by their reduced labor force participation. Because they work less, there are subsequent reductions in work-related transportation costs. For similar reasons, transportation costs generally continue to decline as age increases. 8Approximately 70 percent of elderly families own their own homes. A higher percentage of married couples (approximately 85 percent) than single persons (approximately 60 percent) own their homes. 49

64 The elderly's higher medical expenses are largely explained by a general decline in health that frequently occurs with the aging process. Recent expanded life expectancy has been accompanied by increased chronic ailments among older Americans. Medicare and Medicaid, however, cover most of the elderly's medical expenses. Inflation--Inflation is a particular problem for the elderly because older persons frequently have less ability to adjust workforce participation or living expenses to offset inflationary effects. Many nonelderly respond to cost-of-living increases by shifting consumption patterns, working longer hours, taking a second job or reentering the labor force. However, the elderly generally do not have this flexibility due to health problems or limited labor market opportunities. Despite this, the elderly do have some inflation protection because a substantial portion of their income is indexed and nontaxable. For the low-income elderly, Social Security and SSI income are both fully indexed to CPI changes. Recent estimates indicate that, on average, over 75 percent of the low-income elderly's income is fully indexed. 9 Additionally, as shown in Table II-4, average annual family expenditures tend to decline as family members age. Some elderly have had more problems with inflation than younger persons, because they spend greater proportions of income on goods and services which have experienced the largest price increases (e.g., food, fuel and medical care). Borzilleri estimates that prices of goods purchased by the elderly in 1973 increased at an average annual rate that was 4 percent greater than the prices of goods purchased by the overall population in Concern has developed, however, over using the CP1 as the elderly's cost-of-living index. For example, mortgage interest rates carry substantial weight in the CPI. Thus, the unusually high interest rates of the past three years may have caused the CPI to overstate inflation's overall impact on the elderly. Income Adequacy The number of elderly families with incomes below the different standards varies substantially depending upon particular adequacy standards and income sources used. However, most standards lead to similar conclusions '_Thisestimate is for households in 1979with incomes below $8.000 and persons age 65 or older. U.S. Congress, Congressional Budget Office, "'Low-Income Energy Assistance: Issues and Options" (June 1981), p. 16. "_Thomas C. Borzilleri, "The Need [br a Separate Consumer Price Index for Older Persons," The Gerontologist, vol. 18. no. 3 (1978), p O

65 regarding: (1) historical trends in the number of low-income retirees; and (2) the distribution of low-income elderly persons, according to sex, marital status and age. Under any standard, the elderly's financial conditions have improved considerably over the past twenty years. Under the standards examined in this report, persons over age 72 have the lowest incomes. Throughout the following analyses, the official poverty level is used to analyze the elderly's incomes. This is not intended to imply that this is the most realistic or desirable standard. The poverty level was selected for five primary reasons: First, it is the U.S. government's official poverty standard. Second, it is the most widely used income distribution standard. Third, it is adjusted for family size and other family characteristics while many of the other standards are not. Fourth, unlike the BLS budgets, an official poverty level has been established for both elderly and nonelderly families. Fifth, unlike replacement rates, the poverty level can measure income levels of workers as well as nonworkers. Alternative lm'ome Definitions--When estimating the number of persons in poverty, the Census Bureau defines income as all family members' combined cash income, including employment earnings, earnings from financial assets and cash transfer payments (e. g., employer pensions, Social Security and cash benefits from other government programs). This definition is not adjusted for income tax payments or in-kind benefit values. As discussed in Chapter I, including taxes and in-kind benefit effects in the income definition changes substantially the elderly's financial status. Federal, state, local and Social Security taxes reduce disposable incomes. Consequently, they tend to increase the observed number of families who cannot maintain a given standard's suggested level of income adequacy. Taxes have a more substantial effect on the nonelderly, however, than on the elderly. Since a smaller proportion of the elderly work, they have lower taxable earnings. In addition, their Social Security benefits are not taxed. Elderly persons also receive increased tax exemptions. In-kind benefit programs provide goods and services that increase the elderly's living standards. Family assets also significantly affect the elderly's welfare. These values are not captured by the Census cash income definition. Some of the elderly regularly draw upon assets to purchase goods and services. Many have accumulated substantial home equity that is potentially convertible into income. Even where retirees are unwilling or unable to convert their home equity into cash income, homeowners without mortgages generally have lower-than-average annual housing costs. Thus, they do not require as much income to satisfy living standards as that implied by the different income adequacy definitions. The March 1980 Current Population Survey was used to estimate poverty rates among the elderly under alternative income definitions. As described in 51

66 Chapter I and Appendix B, the estimated value of in-kind benefits and the annuitized value of home equity and other financial assets were included in the income definition. Social Security, federal, state and local taxes were also included, l_ Table II-6 shows the percentage of elderly families with incomes below the official poverty level in 1979 under seven alternative income definitions. Under the Census cash income definition, approximately 18.5 percent of all elderly families were in poverty. The poverty rate is highest for those over age 75. About two-thirds of these families are single women who generally have lower incomes than their male counterparts. If food stamps and public housing assistance values are added to income, the elderly poverty rate drops slightly from 18.5 to 16.2 percent. However, if the annuitized value of wealth is included, the rate declines from 16 percent to about 10 percent--the largest decline is for those over age 74. After considering tax effects, the poverty rate increases slightly for those age and those age 75 and over. The percentage of elderly families with below-poverty-level incomes declines further when Medicaid and Medicare values are included. The rate declines to approximately 8 percent when taxes, food stamps, rent subsidies and potential wealth values are included in the income definition. Medicare and Medicaid benefits have a substantial impact because most elderly persons are covered by one or both of these programs. Food stamps and housing assistance were received by fewer elderly families, but over 70 percent of the elderly have financial assets. Tax deductions have a minimal impact on poverty counts, primarily because the federal tax code provides older persons with additional exemptions and a tax credit. Thus, low-income elderly families have little or no tax liabilities. Under each income definition, the poverty rate for single persons is more than twice as high as the rate for married couples. Table II-7's estimates indicate that in using the cash income definition, elderly married couples and single persons have poverty rates of 9.4 percent and 26.3 percent, respectively. If the potential income from assests, in-kind benefits and taxes are included, incidence of poverty drops to 5.4 percent for married couples and 14.4 percent for single elderly persons. If Medicare and Medicaid values are included, the poverty rate falls to 3.7 percent for married couples and to 11.5 percent for single persons. Income Adequacy under Alternative Standards--There is substantial variation in the number of elderly who fall below the different income adequacy standards considered here. Table II-8 presents percentage estit_seeappendixbfor a detaileddescriptionof this methodology. 52

67 TABLE II-6 Poverty Rates among Elderly Families under Alternative Income Definitionsm1979 _ Family Head's Age 75 and Income Definition Over Total 1. Total Cash Income 15.4% 16.4% 22.8% 18.5% 2. Medicare and Medicaid Values Not Included in Income Definition Cash Income, Food Stamps 13.5% 13.9% 19.8% 16.2% and Housing Assistance Cash Income, In-Kind Benefits and Annuitized Wealth Values Cash Income, In-Kind Benefits and Annuitized Wealth Values, Less Taxes 3. Medicare and Medicaid Values Included in Income Definition Cash Income, Food Stamps 10.9% 10.6% 14.8% 12.4% and Housing Assistance Cash Income, In-Kind Benefits and Annuitized Wealth Values Cash Income, In-Kind Benefits and Annuitized Wealth Values, Less Taxes Source: ICF analysis for EBRI of March 1980 Current Population Survey data. See Appendix B for details. 1Data include single persons and married couples. 53

68 TABLE II-7 Poverty Rates among Elderly Married Couples and Single Persons under Alternative Income Definitionsu 1979 Married Single All Elderly IncomeDefinition Couples Persons Families 1. Total Cash Income 9.4% 26.3% 18.5% 2. Medicare and Medicaid Values Not Included in Income Definition Cash Income, Food Stamps 8.3% 22.9% 16.2% and Housing Assistance Cash Income, In-Kind Benefits and Annuitized Wealth Values Cash Income, In-Kind Benefits and Annuitized Wealth Values, Less Taxes 3. Medicare and Medicaid Values Included in Income Definition Cash Income, Food Stamps 5.6% 18.3% 12.4% and Housing Assistance Cash Income, In-Kind Benefits and Annuitized Wealth Values Cash Income, In-Kind Benefits and Annuitized Wealth Values, Less Taxes Source: ICFanalysisforEBRIof March1980CurrentPopulationSurveydata.See Appendix Bfor details. 54

69 TABLE ll-8 The Percentage of Elderly Families with Incomes below Alternative Standards _ Married Single All Elderly Alternative Standards Couples Persons Families Official Poverty Level 3.7% 11.5% 7.9% Revised Orshansky Poverty Indices BLS Elderly Couples' 9.4 N/A N/A Low Budget Minimum Wage Index Source: ICF analysis for EBRI of March 1980 Current Population Survey data. tlncome includes the after-tax value of cash income, the value of in-kind benefits and the annuitized value of wealth. See Appendix B for details. mates of elderly families with annual income below four standards in In this table, income is defined as the after-tax value of cash income, plus in-kind benefit values and the potential annuitized value of assets. These data show that about 1.3 million, or 7.9 percent, of all elderly families had total cash incomes that were less than the official poverty level. The number of elderly families with incomes below standards set by the revised Orshansky index is nearly double the number below the official poverty standard percent.12 A larger proportion of elderly families, 24.6 percent, had annual cash incomes below the minimum wage index. Elderly married couples are generally better off than single persons. The percentage of married couples with incomes below the various standards ranges from 3.7 percent (under the official poverty level) to 9.7 percent (under the minimum wage index). The number of elderly couples with incomes below the standards is roughly equivalent under the revised Orshansky poverty indices, minimum wage index and BLS lower-level budget for aged couples. Variations are greater among single-person households, because the poverty level is adjusted for family size while the minimum wage index is not. Regardless, it is possible that adjustments for family size do not reflect actual adequacy differences. _2As noted earlier, the revised Orshansky index is percent higher than the official povc_ylcvct. 55

70 TABLE II-9 Elderly Married Couples with Incomes below Alternative Standards, By Agem19791 Age of Family Head 73 and All Elderly Alternative Standards Over Married Couples Official Poverty Level 3.7% 3.7% 3.7% Revised Orshansky 9. l Poverty Indices BLS Elderly Couples' Low Budget Minimum Wage Index Source: ICF analysis for EBR! of March 1980 Current Population Survey data. 1Income includes the after-tax value of cash income, in-kind benefit values and the annuitized value of wealth. See Appendix B for details. The incidence of elderly couples' incomes below standard levels increases slightly with age.t3 As shown in Table II-9, the proportion of elderly couples with incomes below standard levels is not substantially different between families with heads-of-household over age 73 and families with household heads age Less than 11 percent of couples in either group had incomes below any of the standard levels. Table II- 10 shows that 11.4 percent of single elderly women have belowpoverty-level incomes, compared with 12.0 percent of single elderly men. However, using the revised Orshansky index, the proportion of single elderly men and women with below-standard-level incomes increases to 18.9 and 19.7 percent, respectively. Under both standards, poverty rates decrease with age, although not as dramatically as the poverty rate decline observed for married couples. Table II-10's results contradict a common finding that poverty increases with age. The primary reason for this is: For a given asset level, annual annuity payments rise with the annuitant's age. Consequently, annuitized wealth values tend to have more significant effects on the poverty rates among the oldest persons. This result should be interpreted carefully, however. Evidence cited earlier suggests that many of the elderly continue to save after retirement, thus increasing the value of their assets as they grow older. Also, t3as indicated in the tables, the analyses use an income definition which includes the after-tax value of cash income, in-kind benefit values and the annuitized value of wealth. 56

71 TABLE II-10 The Percentage of Elderly Single Persons with Incomes below Alternative Standardsm1979 _ Income Below Official Income Below Poverty Level Orshansky Revision Single Men 12.0% 18.9% Single Women Single Persons Age: or more Average 11.5 % 19.6% Source: ICF analysis for EBRI of March 1980 Current Population Survey data. _Income includes the after-tax value of cash income, the value of in-kind benefits and the annuitized value of wealth. See Appendix B for details. life expectancy grows shorter with age. Both of these factors increase the annuity value of assets with age. If the elderly did convert their assets to annuities at retirement, this phenomenon would be less pronounced. This report's analysis partially accounted for this problem by assuming that the annuity would be annually indexed at a 5 percent rate. Historical Trends--Since 1959, the number of persons with belowpoverty-level cash incomes has declined substantially. Table II-11 shows the number and percentage of all persons and elderly persons, with belowpoverty-level cash incomes. When only cash income is considered, the number of persons in poverty has fallen from 39.5 million in 1959 to 24.5 million in Since the nation's population has grown by about 21 percent in this same period, this improvement in poverty levels is particularly encouraging. The number of elderly with cash incomes below the poverty level declined from 5.5 million in 1959 to 3.2 million in The decline in the elderly's poverty rates occurred at a different time than the rate decline for the general population. The largest decrease in the overall poverty rate occurred between 1959 and it fell from 22.4 to 12.8 percent. In this same period, the elderly's poverty rate fell less rapidly and in the sixties, it was about twice as high as the national rate. The elderly's poverty rate declined most rapidly from 1967 to it fell from 29.5 to 16.3 percent. The national and the elderly's poverty rates have decreased only slightly since 1975; currently they are about 11 and 14 percent, respectively. 57

72 TABLE li-11 The Number and Percentage of Persons with Cash Incomes below the Poverty Levelm1959 to 1978 All Ages Elderly Year (millions) (percent) (millions) (percent) (22.4) 5.5 (35.2) (14.7) 5.1 (28.5) (14.2) 5.4 (29.5) (12.8) 4.6 (25.0) (12.1) 4.8 (25.3) (12.6) 4.7 (24.5) (12.5) 4.3 (21.6) (11.9) 3.7 (18.6) (11.1) 3.4 (16.3) (11.6) ) (12.3) ) (11.8) ) (11.6) ) (11.4) ) Source: U.S. Department of Commerce, Bureau of the Census, Characteristics of the Population Below the Pover O' Level: Current Population Reports, series P-60, no. 12. Income assistance programs--particularly newly developed programs targeted at families with children--were largely responsible for this rapid decline in poverty rates. Social Security benefit increases in the seventies greatly affected the elderly's incomes--especially the 1972 Amendments to the Social Security Act, which indexed benefits to CPI changes. Additionally, as discussed in Chapter I, the percentage of families with employer pensions has increased dramatically in recent years. These changes, along with increases in real lifetime earnings and female labor force participation, explain the sharp reductions in the elderly poverty rate. As noted early in this chapter, historical trends that are based on cash incomes and, which reflect income adequacy, probably understate the elderly's present economic status for five primary reasons: (1) Cash income measures exclude in-kind benefit values. (2) The risk of destitution that elderly families face has been reduced. 58

73 (3) Inflationary effects have been considerably reduced for low-income elderly, because Social Security and SSI are indexed to the CPI. (4) Workers have more options regarding retirement age--many can choose between early, normal or late retirement. (5) Many elderly families can now choose whether they live alone or with relatives. Living independently may improve the elderly's overall living standards; however, it may also reduce their total available household cash income. During the last thirty years, the percentage of the elderly living alone has increased. This is often measured by the household "headship rate" (i.e., the ratio of the number of households headed by an individual 65 or over to the number of persons 65 or over). Table II- 12 demonstrates that this rate for elderly persons increased from 56 to 63 percent from 1950 to In addition, the percentage of single women living with relatives has declined from 40 percent in 1951 to about 20 percent in This trend can be largely attributed to rising incomes, t4 Table II-13 reflects the relationship between the head-of-household's income, marital status and living arrangements. As income increases, the likelihood of living alone increases. Elderly and Nonelderly Incomes--Using only cash income, poverty among the elderly is greater than it is among other age groups. Table II-14 reflects the number and percentage of families in various age groups with total after-tax cash incomes that were less than the official poverty index in TABLE The Household Headship Rate for Elderly Persons to Headship Rate Male Female All Elderly % 34% 56% Source: Sheldon Danziger, Saul Schwartz and Eugene Smolensky, "Living Arrangements in a Simulation Model of Retirement Policy" (1981), p. 5. IThe headship rate is defined as the ratio of: the number of households headed by an individual 65 or over to the number of persons 65 or over. _4SheldonDanziger, Saul Schwartz and Eugene Smolensky, "Living Arrangements in a Simulation Model of Retirement Policy" (1981), p

74 TABLE II-13 Percentage of Elderly Couples and Single Persons Who Live Alonem1975 Married Couples Single Men Single Women Income of Head White Nonwhite White Nonwhite White Nonwhite Less than $1,000 77% N/A 54% N/A 32% 15% $ 1,000-2, % 57 26% $2,000-3, $3,000-4, $4,000-5, N/A 68 N/A $5,000-6, N/A 79 N/A 69 N/A More than $6, N/A 74 N/A All 83% 56% 69% 42% 63% 45% Source: Sheldon Danziger, Saul Schwartz and Eugene Smolensky. "Living Arrangements in a Simulation Model of Retirement Policy" CI981), p. 15. TABLE II-14 The Number and Percentage of Families in Poverty and Labor Force Participation Rates of Families Families in Poverty t Labor Force Age of Number Participation Family Head (thousands) Percentage Rates , % 94.8% , ,491 l ,321 15,6 42,1 73 or more 1, ,9 Total 11, % 79.9% Source: ICF analysis for EBRI of March 1980Current Population Survey data. _Theseare families whose after-tax cash income is less than the official poverty level. -'This is the proportion of families who had earnings from employment in

75 1979. As the head-of-household's age increases beyond 35, the poverty rate also rises. Among families with household heads age 35 to 54, the poverty rate is about 9 percent. The incidence of poverty rises to approximately 12 percent among those age 55 to 64 but remains below the overall national poverty rate of 13 percent. The poverty rate rises to almost 16 percent for families with household heads age 65 to 72, and nearly 22 percent for those with household heads who are over age 72. A primary reason for this increase is the reduced labor force participation that accompanies the acceptance of retirement benefits. Table II-14 demonstrates that the percentage of families with employment earnings falls with age, from almost 95 percent for those under 55, to about 82 percent for those age 55 to 64. Labor force participation rates drop dramatically for persons 65 to 72 and for those over age 72--to 42.1 and 18.9 percent, respectively. Characteristics of the Low-Income Elderly--To better understand the low-income elderly's particular problems, this group's characteristics were examined in greater detail. Table II-15 shows that three-quarters of poverty-level families consist of single persons. Fifty-four percent are widowed women.15 This table also demonstrates that 57 percent of these families are headed by individuals 73 or over. Table II-16 reveals that only 5 percent of families in poverty receive employer pensions, 9 percent receive employment earnings and 36 percent receive wealth income. However, if home equity is included, almost 60 percent own some type of assets. The benefit level received by poor families from each income source is also much lower. Table II-17 shows that of those with Social Security TABLE II-15 Distribution of Elderly Families in Poverty Age of Family Head Family Status and Over Total Married Couples l2% 12% 24% Single Men Single Women Total 43% 57% 100% Source: ICF analysis for EBRI of March 1980 Current Population Survey data. tslcf analysis of the March 1980 Current Population Survey data. 61

76 TABLE II-16 Percentage of Elderly Families Receiving Different Sources of Income Families in Poverty All Other By Age of Head Elderly or More Total Families Social Security 82% 88% 86% 93% Employer Pensions Employment Earnings Public Assistance Wealth Income Other Government Source: ICF analysis for EBR1 of March 1980Current Population Survey data. TABLE II-17 Average Income Received By Families from Each Income Source t Poverty Families Nonpoverty Families Social Security $2,400 $ 4,700 Employer Pensions 1,000 4,100 Employment Earnings 1,000 10,000 Public Assistance 1,200 1,900 Wealth Income 300 3,300 Other Government 1,000 2,200 Total $2,800 $12,700 Source: ICF analysis for EBRI of March 1980Current Population Survey data. _Averageincome for only those families who receive income from the specified source. 62

77 benefits, poor families received average benefits that were about half the amount going to nonpoor families. Poor families are more likely to be one-person households, and thus they receive smaller household Social Security benefits. Of those with employer pension benefits, poor families receive much smaller benefits than nonpoor families. Poor families with pensions may have had lower wages and/or fewer years of service in covered jobs. Additionally, many poor have low benefits because they receive survivor benefits, which are frequently lower than workers' primary benefits. Poor families are also likely to have either low income or no income from employment. Table II-16 shows that less than 10 percent of poor families work while 36 percent of nonpoor families work. As indicated in Table II-17, poor elderly workers receive an average of $1,000 in employment income, while nonpoor working families receive an average of $10,000. Table II- 18 reflects analysis of families with below-poverty-level incomes and the sources of such incomes. It reveals that if Social Security and in-kind benefits are not included, the poverty rate jumps from 12 to 27 percent. _6A number of other findings are also reflected in this table. First, as discussed above, the poverty rate increases with age. Second, there is substantial reliance on Social Security from all age groups and this increases with age. Third, wages are important for all elderly but particularly for those under age 73. Many of the differences in work behavior between the poor and nonpoor are explained by illness and disability. Of households with persons who are age 62 to 64, _7Table II- 19 indicates that the most striking difference between the poor and all other households is reflected in the percentage of workers and the percentage of disabled persons. While over one-half of all 62 to 64-year-old workers were employed over 1,500 hours annually, only 25 percent of poor workers were employed over 1,000 hours. _8 Disability accounts for much of this difference. Table I1-20 reveals that of those age 55 to 71, over 60 percent of poor persons have some degree of disability, but less than 30 percent of the nonpoor are disabled. Low retirement income levels may result from the shortened working career of those becoming disabled. These disabilities _6of course, the poverty rate would not jump this much if there were no Social Security benefits. Many individuals would continue to work, some would receive SSI benefits, and many of those in employer pension plans that are integrated with Social Security would receive larger pension benefits. tvthisis an appropriate age group to consider because many persons in this group are getting ready to retire. _8F.Thomas Juster. "Current and Prospective Financial Status of the Elderly Population, _' unpublished paper (1981), p

78 _J CJ,_ 64

79 TABLE II-19 The Status of Persons Age 62 to 64m1979 Persons Persons Not in Poverty in Poverty Working 14% 42% Laid Off, Unemployed 2 4 Retired Disabled 29 3 Housewife 26! 7 Other 0 1 Total 100% 100% Sources: F. Thomas Juster, "Current and Prospective Financial Status of the Elderly Population," unpublished paper (1981), p. 24. Data derived from the Panel Study of Income Dynamics. TABLE II-20 Extent of Disability among the Elderly By Poverty Status and Agem1979 Elderly Poor Elderly Nonpoor Extent of Disability No Disability 35% 38% 39% 87% 77% 71% "'Somewhat" or _'A Little" "'A Lot" Total 100% 100% 100% 100% 100% 100% Sources: F. Thomas Juster, "'Current and Prospective Financial Status of the Elderly Population," unpublished paper ( 1981 ), p. 32. Data derived from the Panel Study of Income Dynamics. 65

80 may require an unanticipated use of family assets. They may also reduce benefits from employer pension programs. Adequacy of Projected Program Benefits Chapter I discussed the expected trend in employer pension benefit receipt. This chapter has examined the close relationship between benefit receipt and income adequacy. The unpredictability of human behavior makes it impossible to accurately forecast the future percentage of families with adequate incomes. For example, if women continue to enter the labor force and if employer pension benefit trends continue, it is likely that workers will retire earlier than they do today. Alternatively, if Social Security benefits are reduced, it is possible that without compensating changes people will work longer. Using the microsimulation model discussed in Chapter I, the expected levels of Social Security and employer pension benefits were estimated under three pension coverage assumptions. It was assumed there would be 1 percent real annual wage growth, that current labor force participation patterns would continue and that current Social Security retirement ages would remain constant. Under these assumptions, the future percentage of married couples and single persons with pension benefits is likely to increase substantially. This will raise the aggregate level of net retirement income available to the elderly. As this report has noted, single elderly women are one group with particularly low incomes. Focusing specifically on this group, Table II-21 shows that in 1979, 27 percent of all single women age 65 to 69 received employer pension benefits. Although the average benefit per recipient is not expected to increase, almost two-thirds of these women are expected to become eligible for pension benefits by TABLE II-21 Employer Pension Benefits for Single Women Age 65 to Percentage Receiving Benefits 27% 65% Average Benefit Per Recipient $3,000 $2,600 Source: 1979valuesfromICFanalysisfor EBRIof March1980CurrentPopulationSurvey data. ICFestimatesfor 2004.See AppendixC for details. 66

81 III. Policy Issues Previous chapters have shown that there are specific elderly groups who, on average, have systematically lower incomes than other groups. These primarily include single persons over age 72--especially widowed females. Such individuals typically have little or no work history and depend primarily upon Social Security and public assistance for income. Due to their limited employment experience, some retirement income vehicles may be more effective than others in addressing this group's income problems. U.S. retirement income is provided through different programs, each with its own major objective. Pensions are primarily intended as a form of deferred compensation. The worker's benefit right and benefit size are both a function of employment experience. Alternatively, public assistance programs such as SSI and food stamps attempt to establish a floor of income that guarantees some minimum benefit level. The Social Security system incorporates the deferred compensation concept and a benefit structure that favors low-wage workers. Like pension plans, Social Security eligibility and benefit levels are based on work experience. But workers with a career of low eamings covered by Social Security receive higher benefits in relation to the payroll taxes the2_pay than workers with higher levels of covered earnings. 1 Tailoring existing programs to specifically address the elderly poor's problems requires great care. Liberalization of employment-based retirement income programs could improve the general population's retirement income level, but may do little for those with no work experience. Tax incentives that encourage savings and expanded pension coverage may benefit future retirees, but they are unlikely to help the present elderly generation. Thus, it may be prudent to address: (1) the problems of today's poor with increased public assistance or Social Security benefits; and (2) to provide for future generations with tax incentives, expanded pension coverage or more liberal survivors benefits. Policymakers who seek to reduce the federal budget or to resolve Social Security's financing problems, may lean towards policies that change the mix of retirement income programs. For example, some believe that Social Security should not be a vehicle for income redistribution. They have supported elimination of the minimum benefit and they have recommended removal of Social Security's other redistributive components. Others think _As noted earlier, the Social Security minimum benefit has been eliminated for all persons retiring after January 1, This does not, however, eliminate the redistributive nature of the Social Security benefit formula. 67

82 that Social Security benefits should be reduced in general. They propose reductions in the proportion of preretirement earnings replaced by Social Security benefits for all future retirees. Additionally, many are considering alternative procedures for postretirement benefit indexation, normal retirement age changes and early retirement benefit reductions. Reductions in various programs' in-kind benefits are also contemplated. To assure effective policymaking, an understanding of the interaction between cul-rent policies and programs is essential. For example, many current policies and programs are based on the assumption that few families will receive employer pension benefits. However, this report has shown that employer pension eligibility has increased in recent years. More importantly, the percentage of newly retiring elderly families with employer pensions should double in the next twenty-five years._ Table 1-26 shows that average real benefits may remain relatively stable for future recipients. However, the higher benefit recipiency rates will result in significant future growth of average pension benefits for all aged families. Such trends have not previously been well recognized. When evaluating potential retirement benefit policy proposals, several specific questions should be addressed: --How do the proposals afj_'ctthe current, low-income elderly? Many proposals, such as incentives for greater personal savings, may improve income adequacy for future retirees but have slight or no impact on today's low-income elderly. --How do these proposals affect Jim, re retirees? Although higher benefits from Social Security or other government programs provide immediate help to current retirees, a need for such assistance may diminish in future years as employer pensions increase. Pension receipt and average benefit levels are expected to increase. --How effectively do the proposals assist the lowest income groups? Many proposals, such as mandatory pension coverage, help broadly defined elderly groups but may not be effectively targeted to assist the lowest income groups. For example, over 65 percent of the benefits that would be generated by the President's Commission on Pension Policy's mandatory pension proposal would go to households already expecting pension benefits. --How will proposals it!fiueme human behavior? Workers may decide to modify their savings behavior as a result of policy changes that increase private or public benefits. For example, workers who find that their Social Security benefits will be lower than anticipated may respond by saving more, working longer hours or negotiating greater private pension benefits. Thus, the overall effect of such policy initiatives on income adequacy may be minimal or negative. '-See Appendix C for a detailed description of the assumptions used in making these estimates. 68

83 This chapter examines some of the policy issues prompted by such questions. It briefly: (1) reviews selected proposals that are aimed at changing retirement income levels; and (2) highlights these proposals' potential impact on income levels as well as current retirement program mix and funding practices. Policy and Program Initiatives Many proposed policies would reduce emphasis on Social Security (by increasing retirement ages, reducing benefit indexation or lowering replacement rates) and increase emphasis on employer pensions or family savings. This report's analysis suggests that Social Security changes should be designed to: (1) specifically address problems of the truly needy, primarily older single persons: and (2) recognize continued expected increases in pension receipt and benefit levels. This approach may require a careful rebalancing of the entire Social Security program to reflect the role of means-tested public assistance programs and employer pension plans. Social Securit3'--ln recent years, policymakers have suggested Social Security modifications ranging from benefit increases to benefit reductions for some or all recipients. A key issue is whether Social Security should be used specifically to assist the poor or whether other public assistance programs are more appropriate. Proponents of increasing Social Security benefits argue that this is an effective and efficient means of increasing the elderly poor's living standards, because almost 90 percent of these individuals already receive Social Security benefits. Others argue that income transfers should be funded out of general tax revenues rather than out of the Social Security retirement fund. They note that Social Security benefit increases may go to both high-income and low-income recipients. Thus, public assistance programs may be better at targeting aid to the needy elderly. Major proposals affecting Social Security benefits are summarized below. Replacement Rates--Social Security benefits are designed to replace a proportion of workers' preretirement earnings. Under Social Security's redistributive benefit structure, the earning's replacement rate declines as lifetime preretirement earnings increase. Changing this replacement rate formula is one approach for increasing the elderly's income. A general increase in these rates would raise benefits for all covered retirees and, hence, reduce the elderly's overall poverty rate. However, this 69

84 type of policy change may not efficiently increase the elderly poor's incomes. A general benefit increase would provide higher incomes to most elderly while having little effect on those in poverty. Because many elderly poor persons have had little or no covered work experience, they are entitled either to minimal benefits or no benefits. Additionally, unless benefit increases are specifically granted to those who are already retired, replacement rate changes would not help the present low-income elderly. Lowering replacement rates is one way to reduce Social Security's costs. The Social Security benefit formula has three replacement rate thresholds; each threshold is indexed to wage level increases. In 1982, the Social Security income replacement rate is 90 percent of the first $230 of average indexed monthly earnings (AIME), plus 32 percent of the next $1,158 of AIME and 15 percent of AIME above $1,388 up to the maximum. The Reagan Administration has proposed to reduce indexation of these thresholds at a rate that is 50 percent of wage increases for 1982 through After 1987, the thresholds would again be indexed with wage increases. Table IlL 1 shows this proposal's effects on benefits. For those retiring at age 65 in 1987, monthly benefits would be 4 to 9 percent lower--depending upon the person's preretirement income history--than under current policy. If other policies and behavior remain constant, this change would almost certainly temporarily increase the number of elderly poor. However, some workers would probably decide to retire later. In the long-run, workers may respond by increasing savings or working years. Because real wage in- TABLE III-1 Illustrative Monthly Benefits for Workers Retiring at Age 65 under Present and Proposed Law' Percentage Present Law Proposed Law Reduction Full-Time Minimum $ $ % Wage Earner Average Covered Wage Earner Maximum Covered Wage Earner Source: U.S. Department of Health and Human Services, -HHS Fact Sheet," May 12, _Worker is assumed to reach age 65 and retire on January 1,

85 creases cause real Social Security benefits to increase, the long-run effect on the elderly's real income may be small. Spouses' Benefits--Widows and divorced women are disproportionately represented among the elderly poor. Many argue that this is partially due to this group's "inequitable" treatment under Social Security and that the survivor and dependent benefit formulas should be changed. Divorced persons with less than ten years of marriage are not entitled to dependent or survivor benefits based on their former spouses' work experience. Consequently, divorced women with less than ten years of marriage and limited personal work experience may qualify for little or no Social Security benefits. Many argue that this is unfair because "marriage is a partnership where both parties' labors (including the work of homemakers) determine the couple's standard of living." Thus, some feel that divorced individuals should share the retirement wealth acquired during marriage; i.e., when calculating divorced persons' benefits, both spouses should divide equally the Social Security credits earned during marriage. Although these proposals present substantial administrative problems, many single elderly poor women may receive higher benefits. Such a proposal would take benefits away from retired men, however, and may actually increase the total number of elderly poor. Liberalizing the Social Security's survivor benefit formula is another potential change under consideration. Currently, where two couples each have the same total preretirement household income, the surviving spouse of a one-worker family receives higher benefits than the surviving spouse of a two-worker family. To adjust for this inequity, survivor benefits could be based upon the combined preretirement incomes of the husband and wife. This change would increase benefits for many widows who have substantial work experience, but may provide little help to other elderly poor. Retirement Ages--Eligible covered workers may retire with benefits as early as age 62. However, since early retirees receive actuarially reduced benefits, they generally have lower overall income levels than those who retire at later ages. Despite this, early retirement has become an increasingly popular option over the past two decades. Table III-2 shows that the proportion of early retirees has increased from 16.3 percent in 1961 to 61.1 percent in Some believe that this has placed a strain on the Social Security system, because the number of beneficiaries has increased more rapidly than the number of workers supporting the system through payroll taxes. Proposals have been made to reverse this trend by increasing incentives for later retirement. Under a recent Reagan Administration proposal, those retiring at age 62 would receive 55 percent--rather than the current 80 percent--of the benefits they would receive if they retired at age 65. Table 71

86 TABLE III-2 Social Security Beneficiaries with Reduction for Early Retirement to 1979 Number Percent of Year (millions) Worker Beneficiaries % Source: U.S. Social Security Administration, Social Security Bulletin, March 1979, table Q-4. III-3 shows that if this proposal were implemented in 1982, the average wage earner would receive a $247 monthly benefit instead of the presently available $373 benefit. For full-time minimum wage earners, the benefit would fall from $248 to $164. Others have proposed changing Social Security's benefit eligibility ages. For example, the President's Commission on Pension Policy recommended a twelve-year gradual increase of the normal retirement age, from age 65 to age 68, over a twelve-year period beginning in This would be accompanied by an increase in the early retirement benefit eligibility age from age 62 to age 65. Longer labor force participation will reduce benefit payout periods for retirees and simultaneously increase Social Security revenues, because workers will pay Social Security taxes for longer periods. Additionally, some pension plans which are integrated with Social Security may increase their pension benefits to offset the Social Security reductions. Those who do not wish to change their expected retirement date, will be encouraged to increase their personal savings rather than to rely on the government for support. 72

87 TABLE III-3 Illustrative Monthly Benefits for Workers Retiring at Age 62 under Present and Proposed Law t Present Law Proposed Law Full-Time Minimum Wage Earner $ $ Average Covered Wage Earner Maximum Covered Wage Earner Source: U.S. Department of Health and Human Services. "'HHS Fact Sheet." May 12, _Assumes retirement on January Worker is assumed to reach age 62 on that date. A three-year shift in retirement age would mandatorily and uniformly affect all retirees. The reduced early retirement benefit, however, would affect only early retirees and would provide workers with a choice of accepting the lowered benefit or receiving the normal benefit at age 65. Both proposals could cause an increase in the number of elderly poor--particularly among those relying on Social Security as their major income source. Even if these proposals succeed in discouraging early retirement, some persons will be unable to work until a later age, because their age or health prevents them from finding suitable employment. However, these proposals' adverse impact would be at least partially offset by food stamps and other assistance programs. To accommodate individual worker preferences, some have proposed establishing a range of retirement eligibility ages with benefit levels actuarially adjusted to account for later retirement. This would provide a selfadjusting system where individual retirement decisions help to keep benefits and costs in balance. However, greater uncertainty may develop regarding total future program costs. On the other hand, a uniform retirement eligibility age reduces uncertainty about program costs, but it may not satisfy individual worker preferences or our nation's social and economic needs. Benefit lndexation--a considerable amount of controversy exists regarding Social Security benefit indexation. Benefits are currently inflationadjusted based on annual CPI changes. The CPI, itself, as well as the use of the CPI for inflation adjustments in elderly persons' incomes, have both come into question. Some believe that the elderly should have a separate price index because: (1) they have substantially different consumption patterns than the general population: and (2) annual expenses vary differently as individuals become 73

88 older. A separate index for the elderly probably would have resulted in larger benefit increases than those resulting from CPI adjustments during the 1970_ 1977 period. This is because the elderly consume greater than average proportions of the particular items which experienced the largest price increases during that period. Alternatively, recent interest rate increases, particularly mortgage rate increases, carry substantial weight in the CPI. This may have caused the CPI to overstate inflation's impact on the elderly because older families are generally not major purchasers of homes? Others have proposed indexing benefits to changes in average annual earnings. They argue that Social Security benefits are currently adjusted with price changes even though the wages of workers who support the system have increased at a lower rate. An earnings-based index, would more evenly distribute the burden of inflation among beneficiaries and taxpayers. It would lead to more rapid benefit growth during periods when wages increase more than prices, however. Another proposal would require adjusting benefits to changes in either the CPI or the wage index--whichever is lower. During most of the last thirty years, wages grew at a faster annual rate than the CPI. However, in recent years, prices have risen faster than wages. Some analysts argue that if the lower of the two indices is used, Social Security's financial stability will be less vulnerable to unanticipated wage or price shifts. The effect of such indexing proposals could provide the elderly with higher or lower benefits, depending on the future behavior of prices relative to wages. Private Employer Programs--Rather than expanding our dependence on Social Security, many suggest that the public should be provided with incentives that encourage personal savings and/or additional employer pension program development. This section discusses proposed policy changes that would affect employer retirement income programs and elderly income levels. Mandato_ Pension System Proposals--Some groups have advocated a mandatory pension system. Such a system would require all employers to offer a pension plan. The President's Commission on Pension Policy proposed a Minimum Universal Pension System (MUPS) with a 3 percent-ofpayroll employer contribution requirement. MUPS would require: (1) coverage of all employees over age 25 with one year of service and 1,000 hours of annual employment; and (2) immediate vesting of employer contributions. Table III-4 shows the effects of the 3 percent MUPS contribution on average retirement benefits in the initial year of retirement. MUPS would 3A changein the treatmentof housingcostsin the CPI, which willtake effect for Social SecurityindexationinJanuary1985.willmakethisindexlesssensitivetofluctuationsinthe interestrate. 74

89 TABLE III-4 Potential Effect of MUPS on Average Initial Retirement Benefits at Age 65 for Workers Currently Age 25 to 29t Average Benefits at Age 65 Current Policy MUPS Married Couples Expecting to: Receive Pension Benefits $15,700 $16,800 Not Receive Pension Benefits 8,500 10,600 Single Persons Expecting to: Receive Pension Benefits $11,200 $12,700 Not Receive Pension Benefits 5,200 7,200 Source: President's Commission on Pension Policy and U.S. Department of Labor, Office of Pension and Welfare Benefit Programs, "Background Analysis of the Potential Effects of a Minimum Universal Pension System," ICF Incorporated (1981), p. 38. _These estimates are in 1980 dollars and assume retirement at age 65 and an average annual real wage growth of I percent. Under MUPS, all private and public employers must offer a retirement plan that permits participation of all employees who are age 25, with one year of service and 1,000 hours of service annually. Additionally, MUPS would require that all participants vest within five years. significantly increase household retirement benefits for those who would not otherwise receive private pensions. Single persons who, without MUPS, would not receive pensions are especially affected. MUPS is expected to increase their benefits by about 40 percent. Although MUPS would improve retirement benefits for some future retirees, it would not help current retirees. MUPS would also have little effect on those with below-poverty-level incomes in the future. Of those age in 1979, only 1 percent of married couples and 3 percent of single persons are expected to have below-poverty-level retirement incomes, assuming they retire at age 65 under current policies. Roughly the same percentage of these couples and individuals would fall below the poverty line if MUPS were implemented. MUPS would, however, improve the incomes of those in the _'near-poor" group. 4 MUPS is not the most efficient way to target benefits to low-income elderly. Households that already expect to receive pension benefits would 4The -near-poor'" include those individuals or families whose income falls between the povertyline and 125percentof the povertyline. 75

90 receive even larger amounts under MUPS. Approximately 65 percent of all MUPS-generated retirement benefits would go to these households, s Although MUPS would help some low-income elderly, it may reduce these persons" living standards during their working years and/or add to inflation through unit labor cost increases. To offset these costs, some workers may lose their jobs and others, particularly low-wage workers, may receive lower future wage increases. Finally, this system's administrative costs, especially with its full and immediate vesting, may be substantial. Spouses' Benefits--Employer pension programs currently provide many widowed spouses with retirement income. One way to increase the percentage of widows and widowers with pension benefits is to require all pension plans to provide survivor benefits to covered workers' spouses. Some propose establishment of an automatic 50 percent joint and survivor option for survivors of workers who die in the ten-year period prior to normal retirement age. Additionally, some feel that in the event of a divorce, pension benefits earned during the marriage should be divisible. Mandatory survivor benefits would help reduce the future number of low-income elderly widows. Although expensive to administer, the divisibility of pension benefits for divorced couples could also reduce the future number of elderly poor. On the other hand, pension plans may have to reduce benefits to all recipients to cover the added costs of mandatory survivor benefits. These changes would do little to assist divorced or widowed persons with retirement incomes now below poverty. Vesting--Proposals have been made to liberalize ERISA's minimum vesting provisions so that all workers would vest after five years of pension plan participation. Shorter vesting would help many future retirees but might not significantly help the low-income elderly. Liberalized vesting could also result in high administrative costs, particularly immediate vesting or vesting after one year of service. Integration--Many defined benefit and defined contribution pension plans coordinate, or integrate, their benefits with Social Security. As noted in Chapter I, the reasons for integration are: --Employers pay for part of Social Security and should, therefore, be able to design pension programs that recognize employer-financed Social Security benefits. --Workers with equal years of service but different wage levels should have roughly equivalent replacement rates. SPresident's Commission on Pension Policy and U.S. Department of Labor, Office of Pension and Welfare Benefit Programs, "'Background Analysis of the Potential Effects of a Minimum Universal Pension System," ICF Incorporated ( 1981). 76

91 --Each worker's total retirement benefits from Social Security and employer pension plans should not exceed his or her preretirement pay. Current tax law prohibits pension plans from providing benefits that discriminate in favor of highly selected groups, e.g., officers, shareholders and high-income workers. Although low-wage workers receive proportionately more generous Social Security benefits due to the system's redistributive benefit structure, some believe that total retirement income remains inadequate at the bottom of the income spectrum. Some also believe that, despite nondiscriminatory tax law provisions, current pension benefit rules do discriminate against lower-income workers. Low-wage employees can work many years under an integrated pension plan and receive low benefits relative to high-wage employees. 6 Due to coordination between the pension and Social Security systems, integrated plans may play an important role in offsetting any future Social Security benefit reductions. For example, legislation has recently been enacted which eliminates Social Security's $122 minimum benefit for future retirees. This will result in lower Social Security benefits for some future beneficiaries. Such individuals who are also covered by an integrated pension plan, under the offset method, will receive higher pension benefits that replace part of the Social Security benefit reduction. Others think that the current integration rules are necessary to assure that lower-income workers' benefits do not exceed their preretirement income. However, some argue that current integration rules already produce relatively generous benefits for lower-income workers because they do not adequately reflect: (1) differences in tax treatment between Social Security and pension benefits; and (2) unlike Social Security benefits, pension benefits are not fully inflation-adjusted. Public Employer Programs--Federal policy changes affecting public pension plans have also been considered. All federal and many state and local employees are not currently covered by Social Security. Unlike private plans, state, local and federal pension plans are not required to satisfy ERISA's participation, vesting and funding requirements. Based upon the findings of the Universal Social Securit3' Coverage Study, the National Commission on Social Security and the President's Commission on Pension Policy have recommended mandatory Social Security coverage for all private and public workers. This would help eliminate windfall Social Security benefits that some public workers receive because they have spent a short period working in Social Security-covered employment. 6Note,however,that the combinedpensionandsocialsecuritybenefitswillstillbe greater for low-wageworkerscomparedtohigh-wageworkers,relativeto finalearnings. 77

92 Many have suggested the development of ERISA-type legislation for state and local employees to assure these workers of protection similar to that of private sector pension participants. It is difficult to predict the impact of such legislation on benefits. Many public plans provide liberal vesting and early retirement policies. However, ERISA-type funding requirements could lead to benefit reductions because some public plans have very large unfunded liabilities. The cost of amortizing these unfunded liabilities could jeopardize current benefits for systems that rely heavily on present pension revenues to meet present benefit payments. However, to reduce future pension liabilities, some public plan sponsors are now changing retirement ages, benefit levels and other requirements. Modifications to the Federal Civil Service Retirement plan have also been proposed. Proponents of such changes believe that federal workers who retire early should receive actuarially reduced benefits similar to Social Security's early retirement provisions. Individual Initiatives--The 1981 Economic Recovery Tax Act expands workers' incentives to save for retirement. Under this law, workers who are covered by employer pension plans, as well as those who are not covered by employer plans, may establish an individual retirement account and deduct up to $2,000 ($2,250 for spousal IRAs) in annual contributions from their taxable income. This liberalizes existing IRA legislation which: (1) permitted that only workers who were not currently covered by employer plans could set up IRAs; and (2) restricted the maximum annual deductible deposit to the lesser of $1,500 ($1,750 for spousal IRAs) or 15percent of earnings. Similarly, ERTA permits self-employed workers to contribute up to $15,000 annually to a Keogh plan (formerly the limit was $7,500). 7 Finally, ERTA's income tax rate reductions were designed to stimulate general savings that, in turn, may increase individuals' retirement savings. These incentives' direct impact on the number of future elderly poor may be minor. Low-wage workers will generally be unable to save as much as high-wage workers. Additionally, higher-income families will receive greater tax savings than lower-income families because of contributions to IRAs and Keoghs. ERTA may result in improved overall economic conditions which could, in turn, favorably affect workers' abilities to save for retirement. However, whether ERTA will impro,_e economic conditions remains unclear. Public Assistance and In-Kind Benefits--Many elderly are unable to work, have few assets and receive minimal retirement income. These indi- 7Keoghplans (sometimesreferred to as HR 10 plans}may be establishedby the selfemployedin unincorporatedbusinessesor individualswithoutsideself-employmentincome. EquivalentKeoghcoveragemust be providedfor all full-timeemployees(with at leastthreeyears of employment)of the self-employed. 78

93 viduals frequently have limited lifetime work experience, and thus policies aimed at modifying work-based programs (e.g., pension plans and Social Security) may offer little help to this group. In the short-term, public assistance programs may be the most efficient means for providing adequate incomes to the elderly poor. As discussed in Chapter I, many new welfare programs have been developed and others have expanded to provide economic assistance to the elderly poor. However, in an effort to reduce federal spending, Congress has recently cut benefits under Medicaid, food stamp and housing assistance programs. These cuts are likely to affect some elderly poor families; however, the states have considerable latitude in deciding how Medicaid and housing assistance benefits will be distributed. Therefore, effects on the elderly poor may be substantially reduced if the states elect to place high priority on elderly aid. Congress has imposed a cap on the dollar amount of state Medicaid expenditures that is reimbursable by the federal government. This may prompt states to tighten Medicaid eligibility criteria and perhaps to eliminate altogether Medicaid's coverage of the medically needy. However, states could structure eligibility rules so the elderly, or low-income elderly, receive fewer benefit reductions than other groups. Proposals are being considered that would reduce benefit levels and the number of eligible households under the food stamp program. Recently, Congress enacted an additional income test, which eliminates from eligibility households with total gross cash incomes that are more than 130 percent of the official poverty level. Benefit levels and the maximum allowable deductions for dependent care, medical care and work-related expenses will be only partially adjusted for inflation in However, Congress has expressed some interest in fully indexing these items for elderly families. Housing assistance programs are reducing the number of available units and rent subsidies. Participants' rental payments are permitted to rise from 25 percent to 30 percent of a family's income, thus increasing participants' share of the rent payments. However, state and local governments may elect to maintain current levels of service for the elderly. Future Research Proposed changes to one retirement income source may have serious implications for other retirement income components. To be effective, retirement income studies should consider the interrelated effects of proposed changes on all retirement income components. The retirement system's primary goal is to provide adequate income for retirees with the efficient use of available resources. The principal policy 79

94 issue is to define and encourage an appropriate mix of private and public programs for achieving this goal. In assessing policy initiatives, population demographics, program structures, income and wealth variations and the potential ramifications of various proposals should be carefully evaluated, s Before efficient retirement income adequacy policies can be developed, additional research is needed. For example: (1) How do workers accumulate asset wealth and use it in retirement'? What is wealth accumulation's role in retirement planning'? Should wealth be considered when determining poverty estimates'? If so, how should wealth be counted'? What portion of retirement consumption is paid for by liquidating family assets'? To the extent that this occurs, does it suggest that retirement income programs are inadequate; or, is this an appropriate vehicle for financing some retirement expenditures'? (2) The lowest income elderly group is largely comprised of single women. One question is whether or not these women have low incomes because of their lack of work experience. Have joint and survivor options and life insurance coverage been effectively utilized in retirement planning? (3) How would proposed Social Security changes affect the elderly'? Is it likely that employer pension benefits will increase to offset any Social Security reductions? (4) The majority of the elderly own their homes. Many elderly homeowners are financially limited but are unable or unwilling to liquidate their home equity. Additional information is needed to determine the potential methods, such as reverse equity mortgages, for using the value of home equity to satisfy retirement income needs. (5) The full implications of raising Social Security's retirement age have not been thoroughly examined. Although life expectancy is increasing, it is not clear that older workers" physical abilities are comparably improving. Before Social Security's retirement age is modified, this proposal's implications for individuals, employers, the overall economy and other retirement system components must be completely analyzed and understood. What proportion of workers who retire at age 62 and age 65 have no income source other than Social Security'? How will this change in the future'? What proportion of the age population is unable to work, disabled or in poor health? As a result of retirement age eligibility changes, what adjustments should be made to Medicare, Disability Income and other in-kind benefit programs'? What is an appropriate method for addressing the problems of those who cannot physically work, or otherwise adjust their retirement age to a later date'? (6) How do retirees" living expenses change over time? Very little is presently known about how expenditure patterns evolve in old age. This is important for understanding indexation requirements, as well as the design of income, health and other benefit programs. _EBRI, Coverage attd Benefit Entitlement, p

95 APPENDIX A Background Tables on Retirement Program Beneficiaries and Benefits TABLE A-1 Beneficiaries and Benefits of the Federal Old-Age and Survivors Insurance Programm1950 to 1979 Beneficiaries Total Benefits Average Average Benefit Year (thousands) (millions) Benefit (1979 dollars) ,477 $ 961 $ 276 $ ,379 1, , ,026 2, , ,981 3, , ,886 3, , ,961 4, , ,128 5, , ,979 7, , ,162 8, , ,244 9, , ,157 10, , ,468 11, , ,778 13, , ,583 14, , ,236 14, , , , , ,797 18, , ,565 19, , ,225 22,642 1,019 2, ,827 24,209 1,061 2, ,564 28,796 1,222 2, ,361 33,413 1,371 2, ,205 37,122 1,473 2, ,310 45,741 1,739 2, ,941 51,618 1,916 2, ,732 58,509 2,110 2, ,397 65,699 2,314 2, ,217 73,113 2,502 2, ,718 80,352 2,704 3, ,348 90,556 2,984 2,984 Source: U.S. Social Security Administration. Social Security Bulletin Annual Statistical Sul_plement pp

96 TABLE A-2 Beneficiaries and Benefits of Insured and Noninsured Private Pension Plans to 1979 Beneficiaries Total Benefits Average Average Benefit Year (thousands) (millions) Benefit (1979 dollars) $ 370 $ 822 $2, , , , , , ,090 1, , ,240 1, , ,400 1, , ,590 1, , ,780 1, , ,910 1,970 1,031 2, ,100 2,330 1,110 2, ,280 2,590 1,136 2, ,490 2,990 1,201 2, ,750 3,520 1,280 2, ,180 4,190 1,318 2, ,460 4,790 1,384 3, ,920 5,530 1,411 2, ,180 6,450 1,543 3, ,740 7,360 1,551 2, ,180 8,590 1,657 2, ,550 10,000 1,801 3, ,080 11,220 1,843 3, ,390 12,930 2,023 2, ,115 14,850 2,087 2, ,612 16,760 2,202 2, ,993 18,768 2,348 2, ,363 21,146 2,529 2, ,713 23,587 2,707 2,707 Sources: Alfred M. Skolnik, "'Private Pension Plans, ," Social Security Bulletin, vol. 39, no. 6 ( 1976); American Council of Life Insurance, Pension Facts (Washington, D.C., 1977), pp , 36_ All private pension values for years after 1975 were taken from [CF Incorporated, A Private Pension Forecasting Model (1979). 82

97 TABLE A-3 Beneficiaries and Benefits of State and Local Government Pension Plans to 1979 Beneficiaries Total Benefits Average Average Benefit Year (thousands) (millions) Benefit (1979 dollars) $ 595 $ 1,393 $4, ,078 1,633 4, ,775 2,003 4, ,291 3,280 2,541 4, ,379 3,620 2,625 4, ,463 4,335 2,963 5, ,550 5,075 3,274 5, ,635 5,835 3,569 5, ,730 7,025 4,061 5, ,840 7,700 4,188 5, , 173 8,455 3,891 4, ,240 9,550 4,263 4, ,300 10,770 4,683 4,683 Sources: U.S. Social Security Administration, Social Security Bulletin Annual Statistical Supplemenl pp : American Council of Life Insurance, Pensi_m Facts (Washington. D.C., 1980). table

98 TABLE A-4 Beneficiaries and Benefits of the Federal Civil Service Retirement System to 1979 Beneficiaries Total Benefits Average Average Benefit Year (thousands) (millions) Benefit (1979 dollars) $ 380 $1,279 $3, ,581 3, ,385 1,900 4, ,838 2,959 5, ,231 3,146 5, ,002 3,668 6, ,853 4,068 6, ,144 4,701 6, ,056 5,143 6, ,037 5,612 7, ,517 6,396 7, ,821 6,914 7, ,380 7,656 7,656 Sources: U.S. Social Security Administration, Social Security Bulletin Annual Statistical Supplement, , table 16; American Council of Life Insurance, Pension Facts (Washington, D.C., 1980), table

99 TABLE A-5 Beneficiaries and Benefits of Military Pension Plansm1955 to 1979 Beneficiaries Total Benefits Average Average Benefit Year (thousands) (millions) Benefit (1979 dollars) $ 419 $2,341 $7, ,71 l 6, ,384 2,860 6, ,849 3,686 6, ,386 4,075 7, ,885 4,365 7, ,390 4,631 7, ,012 5, 128 5,067 7, ,073 6,242 5,817 7, , ,445 8, ,199 8,216 6,852 8, , ,378 8, , ,993 7,993 Sources: 1979 DOD Statistical R_7_ort_[' the Military Retirement System, supplement. Unpublished data. U.S. Department of Defense, Defense Manpower Data Center, Office of Actuary. 85

100 Table A-6 Beneficiaries and Benefits of the Supplemental Security Income Programm1974 to 1979 Beneficiaries Total Benefits Average Average Benefit (thousands) (millions) Benefit (1979 dollars) Aged ,286 $2,414 $1,056 $1, ,307 2,517 1,091 1, ,148 2,420 1,127 1, ,051 2,364 1,153 1, ,968 2,342 1,190 1, ,872 2,421 1,293 1,293 Blind ,680 2, ,716 2, ,763 2, ,844 2, ,922 2, ,104 2,104 Disabled ,636 2,557 1,563 2, ,933 3,072 1,589 2, ,012 3,346 1,663 2, ,109 3,628 1,720 2, ,172 3,882 1,787 1, ,201 4,286 1,947 1,947 Total ,996 5,097 1,276 1, ,314 5,716 1,325 1, ,236 5,900 1,393 1, ,238 6,134 1,447 1, ,372 1,511 1, ,150 6,869 1,655 1,655 Source: u.s. Social Security Administration, SocitH Se_'urity Bulletin Annual Statistical Supplement, , pp

101 APPENDIX Methodology the Elderly's B for Analyzing Income Overview Because of data limitations, past studies of the elderly's financial wellbeing have tended to examine income, assets and in-kind benefits separately. This has made it difficult to determine how these three resource categories interact to provide economic security to individuals. Most studies have been done on a pretax basis without analyzing disposable (after-tax) income. The March 1980 Current Population Survey (CPS) contains recently available data on the asset income, cash income and in-kind benefits that the elderly receive. The CPS, a representative U.S. population sample, includes data on over 13,000 elderly families. The March 1980 CPS was used to analyze cash income, asset wealth and in-kind benefit distribution to the elderly by estimating: (1) the level of cash income each elderly family received in 1979; (2) the individuals who received in-kind benefits: (3) the cash equivalent value for each of four in-kind benefit types: (4) the potential income of converting asset wealth (including home equity) into an indexed annuity; and (5) the disposable (after-tax) income of each family. This appendix describes how this unique data set was constructed. (1) The number of individuals with income and the income amount received from any of the sources listed below were determined from the March 1980 CPS: employment earnings --Business or farm earnings --Unemployment compensation Worker's compensation --Social Security --SSI (federal and state) --Interest from financial assets --Stock dividends --Rental income --Educational assistance --Public assistance/welfare 87

102 --Child support --Alimony --Regular financial assistance from friends --Private pensions --Civil Service pensions --U.S. military pensions --State/local government pensions --U.S. Railroad Retirement --Annuities/insurance policies --IRA/Keogh plans --Veteran survivors pensions --Black Lung benefits --Payments from estates/trusts --State temporary sickness plans --Other, miscellaneous. (2) Using the March 1980 CPS, the numbers of participants in four major in-kind benefit programs--food stamps, public housing assistance, Medicare and Medicaid--were estimated.t (3) These individuals' in-kind benefit values were then estimated. Some think in-kind benefit values are equal to the government's cost of providing them. A major problem with this approach is that individuals may value in-kind benefits at less than government's cost. For example, a family receiving $160 in food stamps may be equally satisfied with $140 in cash. One way to avoid overstating the impact of in-kind benefits is to value them at their cash equivalent value to the recipient. The cash equivalent value of these benefits was developed with a methodology utilized previously by Smeeding and Moon.2 First, the in-kind benefit's cost to government was added to each family's disposable income. To estimate the government cost of food stamp benefits, the food stamps' face value, as recorded on the CPS, was used. For Medicare and housing _TheCPS data underreport Medicaid eligibility because many are unaware of their eligibility. To compensate for underreporting, each state's Medicaid eligibility rules were used to determine which CPS-surveyed individuals were entitled to Medicaid. Individuals determined to be eligible for Medicaid through this process were added to those who had indicated their Medicaid enrollment on the CPS to get an estimate of the total Medicaid population. ZTimothy Smeeding and Marilyn Moon, "Valuing Government Expenditures: The Case of Medical Care Transfers and Poverty," Review of Income and Wealth, September 1980.For other research on the cash equivalent valuation of in-kind benefits see Eugene Smolensky, et. al., "'Adding In-Kind Transfers to the Personal Income and Outlay Account: Implications tbr the Size Distribution of Income," ed. Thomas Juster, NBER Studies inincome and Wealth, no. 41 (Cambridge, Mass.: Ballinger, 1977); and Timothy Smeeding, "The Anti-Poverty Effectiveness of In-Kind Transfers," Journal of Human Resources, no. 52, Summer

103 assistance, a methodology developed by Thomas Borzilleri was used? Medicaid's value was assumed to be equal to the average 1979 government Medicaid expenditure per eligible individual. Second, the benefit value was constrained to assure that it did not exceed the average amount spent on that good by families of equal income and size, as reported in the Consumer Expenditure Survey (CES). For example, CES data may indicate that two-person families with $4,000 incomes spent approximately 24 percent of their income on food. If the government cost of their food stamps was less than $960 (24 percent of $4,000), the government cost would be used as the cash equivalent value of their food stamps. However, if the government cost was over $960, then the average expense of $960 would be used. This procedure is followed separately for food stamps, public housing and medical benefits. 4 (4) The potential effect on income distribution of converting asset wealth into income was analyzed. Data on the flow of income from interest payments and dividends were used to estimate the underlying value of financial asset wealth. Home equity values were then added to this estimate to determine estimates of total wealth. This total value of wealth was then converted into an indexed annuity. The annual payment of this annuity was then substituted for the value of interest and dividend income. As discussed above, the March 1980 Current Population Survey was used to determine which individuals received dividend income and/or interest income and which individuals owned or were buying homes. Asset values were estimated using 1979 individual interest and dividend data as well as 1979 investment return data. Data on 1976 home equity values were inflated to determine present home equity values. It was then assumed that elderly persons with financial assets or home equity converted such assets into indexed annuities.-_ Based on mortality and interest assumptions, the three asset types were converted into annuities with an annual indexed increase of 5 percent. Married couples were assumed to receive joint and survivor annuities, which, when one spouse died, provided the survivor with one-half the annual annuity that had been paid to the couple. The indexed annuity payment's annual value was then substituted _Thomas C. Borzilleri, "'In-Kind Benefit Programs and Retirement Income" (Working Paper prepared for the President's Commission on Pension Policy, Washington, D.C ). 4Although most analysts think the cash equivalent approach is appropriate, it is difficult to measure because different individuals place different values on services such as medical care and goods such as food and housing. 5This analysis attempts to provide a perspective of the cash income potential of personal assets. In evaluating this information, it should be understood that some elderly are unable or unwilling to convert personal assets into cash. 89

104 for the interest or dividend income each individual reported on the 1980 CPS. These procedures are described in greater detail below. Individual taxes were estimated as described below. First, family members were put into tax-filing units. Federal, state, local and Social Security taxes were then calculated. Estimates were made of the number of individuals who itemized deductions and their deduction levels. Included in these estimates were probabilities of reporting capital gains and the level of such gains. These estimated tax payment values were then subtracted from gross income to determine after-tax incomes for elderly households. Housing Assistance The March 1980 Current Population Survey identifies those who live in subsidized housing. Thomas Borzilleri has estimated that in 1978, these households received an average subsidy of $1, This amount was adjusted to reflect the CPI changes from 1978 to The adjusted amount was then used to estimate the government cost of elderly housing assistance ($1,621}. This amount was assigned to each family unit which reported receipt of housing assistance benefits in the March 1980 CPS. These benefits were converted to their cash equivalent value by constraining the subsidy amount. It was constrained so it did not exceed average rental expense for equal income/equal size families as reported in the Consumer Expenditure Survey. 7 Food Stamps The March 1980 CPS also records the food stamp face value received by households. This face value amount was used as the government cost estimate of food stamp benefits. Benefits were converted to their cash equivalent values by taking the lesser of: (1) the food stamps' face value; or (2) the average food-at-home expense reported in the CES for families of comparable size and disposable income level.8 Medical Assistance In the March 1980 CPS, individuals were asked whether they were covered by Medicaid. As previously noted, the data understate the number 6Borzilleri, "In-Kind Benefit Programs and Retirement Income," p Rental expenses from the CES were adjusted to 1979 levels to account for inflation. SFood-at-home expenses from the CES were adjusted to 1979 levels to account for inflation. 90

105 of covered persons because many are unaware of their Medicaid eligibility. To adjust for this, Medicaid state eligibility rules were used in assigning Medicaid coverage status. Family members were assumed to be eligible for Medicaid if: (1) in the March 1980 CPS, they reported they were covered by Medicaid; or (2) they were receiving SSI, Aid for Dependent Children (AFDC) or other cash assistance (in most states, these families are automatically eligible for Medicaid); or (3) their family income was less than their state's Medicaid income eligibility limits. Due to variation in different states' Medicaid benefit levels, separate benefit amounts were calculated for each state. The average Medicaid payment per eligible person in each of four age groups was then calculated. Each state's eligible population was estimated using the three-step procedure described above. Medicaid payment averages based on recipients' age groups were used as the Medicaid payment estimates. These average amounts were then used as the government cost of the eligible population's Medicaid coverage. Thomas Borzilleri has estimated the average value of Medicare coverage using insurance cost estimates for comparable private insurance coverage. He determined Medicare's annual value to be $1,075 for single persons and $1,613 for married couples. 9 Government's cost was assumed to be equal to this value for each household that reported Medicare eligibility on the March 1980 CPS. Because Medicaid and Medicare coverage overlap, a family's total medical assistance benefit was not permitted to exceed its Medicaid benefit. The total medical benefit cash equivalent value was estimated by constraining benefit amounts to the minimum of: (1) government's estimated cost of the benefits; or (2) average medical expenses reported in the CES by families of equal income and size.l Housing Wealth Estimates The following steps were taken to estimate housing wealth values: I 1) The March 1980 CPS was used to determine which individuals or married couples owned or were buying a home. (2) The March 1980 CPS did not contain data on home equity. Therefore. the 1976 Survey of Income and Education (SIE) was used to determine 1976 average '_The Borzilleri 1978 and 1979 figures were adjusted for inflation to produce the numbers shown here. _ Medical expenses from the CES were adjusted to 1979 levels to account for inflation. 91

106 home equity values fl_r 392 age/region/farm status family income groups. These 392 groups are a product of: --Seven age groups Less than or more --Census North North -South -West region East Central --Farm/nonfarm slatus --Seven annual family income groups ( 1976 income in 1979 dollars) -Less than $3,000 $3,000-5,999 $6,000-8,999 -$9,000-14,999 -$15,000 24,999 -$25,000-39,999 $40,000 or more Home equity values.for each of these 392 groups were divided into 13 groups (1976 values in 1979 dollars): -Less than $4,999 -$5,000-9,999 -$10,000-14,999 -$15,000-19,999 -$20,000-24,999 -$25,000-29,999 -$30,000 39,999 -$40,000-49,999 -$50,000 59,999 -$60,000-69,999 -$70,000-79,999 -$80, ,000 -More than $100,000 (3) The proportion of individuals whose home equity fell into each of the 13 classes listed above was calculated for each of the 392 groups. (4) These proportions were used as probabilities to randomly assign CPS respondents with similar age/regitm/tarm/income characteristics to one of the thirteen home equity value groups. The mean equity vatue in the home equity category was assigned to CPS respondents. {5) Home equity values were converted into income as described below. 92

107 Interest Wealth Value Estimates To estimate the underlying value of wealth on which interest income was received, three steps were followed: (1) The March 1980 CPS was used to determine which families had interest payments. (2) To estimate savings levels that generated interest income, the value of these interest payments was divided by.0722 (the 1979 average rate of return on savings and loan deposits)._t (3) This interest "'wealth" was then converted into income as described below. Dividend Wealth Value Estimates To estimate the underlying value of wealth on which dividend income was received, three steps were followed: I l) The March 1980 CPS was used to determine households with dividend income. {2) To estimate value of dividend wealth, the value of dividend payments was multiplied by 7.0 tthe average 1979 price/earnings ratio). _2 13) This dividend wealth was then converted into income as described below. Annuitizing Wealth The following procedure was used to convert housing and other financial asset values into an income stream: ( 1) The value of housing equity was added to dividend and interest wealth for each married couple or single person on the March 1980 CPS. Wealth values were calculated l\_r families as described above. (2) The cost of providing an indexed annuity of one dollar annually to married couples with a husband age X and a wife age Y or a single man or woman age X was calculated. X and Y were determined for each couple or single person based upon the March 1980 CPS data. The indexed annuity's cost was calculated using the UP-84 mortality assumptions and a 6 percent interest _Data Resources. Inc., "'U.S. Long-Term Review: Summer 1980"" (Lexington. Mass ). _21hid. This factor may understate the value of wealth because some companies may not pay dividends. If the elderly are more likely to invest in stocks with higher price earnings ratios than the average individual, however, this factor may overstate the value of wealth. 93

108 assumption. For married couples, the annuity was assumed to be payable in the following fl_rm: one dollar annually to the husband for life and one dollar to the wile l\_rlife. For single persons, the annuity was payable as a life annuity. For both married couples and single persons, the annuity was indexed at 5 percent annually. (31 The lump-sum value of wealth calculated in the first step was divided by the annuity factor calculated in the second step to determine the annual annuity payment. {4) This annual annuity payment was added to each individual's record. Calculation of State, Federal and Social Security Taxes To calculate after-tax incomes, a tax calculation model was developed to simulate the federal income tax-filing process for each CPS surveyed household. _ The model determines household members with sufficient income to file returns and those with dependency status. All individuals included in a given tax return are referred to as the tax-filing unit. Adjusted gross income (AGI) for tax-filing units is calculated by summing each member's taxable income. This information is taken from CPS data on cash income and includes taxable portions of estimated wealth annuities. The model also imputes capital gains and adds them to the AGI figure. Taxable income is then calculated by subtracting the appropriate amount per exemption as well as allowable deductions from adjusted gross income. Federal tax liabilities are calculated directly from the appropriate tax table. Earned income and elderly tax credits are calculated for eligible individuals and are then subtracted from their federal taxes. State taxes are assumed to be a fixed proportion of federal taxes. Finally, Social Security payroll taxes are computed for all employed persons. --F_Jrming a Filing Unit. CPS records are used to identify households that are required to file tax returns. Married heads-of-household in each census family unip4 are grouped with their spouses into one filing unit and are considered eligible for joint return filing. Heads-of-household who are not married and have dependents are assumed to file head-of-household returns. All other single family members are assumed to file separately. Tax returns are simulated for all persons whose taxable income is high enough for him or her to be required by IRS to file a return. Members of one filing unit may be financially dependent t_the tax calculation methodology described here is similar to those used by the Congressional Budget Office and the Department of Health and Human Services to calculate taxes with Census data. t4censusfamily units include: the primary family (household head and his family members), secondary families (families living with the primary family in the household) and subfamilies (single or divorced daughter of family head with a child, etc.). 94

109 upon others, and thus they may be claimed as dependents on a head-ofhousehold's tax return. For example, it is assumed that anyone totally without taxable income is dependent upon another household member for support. Other individuals who have to report taxable income may still be supported by another household member. In this case, the model assigns members of some filing units dependency status in other tax-filing units within the household. However, these people still file a return as long as they have a positive tax liability. The algorithm for assigning individuals to a tax unit as dependents is described below. -All family heads and their spouses are counted as dependents on their own tax return. -All children under age 15 within a family are counted as dependents. For persons over age l 5 whose income is less than the average income of all family members, dependent status is assigned in the family head's tax return. -If the person is a secondary family head (a family living with a household head's own family) without a positive tax liability, the head, spouse and all family dependents of such a family are counted as dependents on the household bead's return. -Any person not counted as a dependent in a secondary family is counted as a dependent of the household head if the person's income is less than the average household income (proxy for IRS support test). -Persons living in group quarters are assumed to file separate returns and are not counted as dependents in any other return. -- Computation (q'adjusted Gross Income. Adjusted gross income is calculated by summing taxable incomes and imputing capital gains to the filing unit. These income sources include wages and salaries, farm or nonfarm self-employment, rent and private or government pensions. Taxable portions of estimated wealth annuities are also included in AGI. A portion of the filing unit's unemployment insurance is also included. _s Capital gains are imputed based upon 1978 Statistics of lncome (SOl) data, which summarizes federal tax return information. These data estimate the probability that an individual in a given income interval will report capital gains, as well as the size of such capital gains income relative to adjusted gross income (shown in Table B- 1). These data are used to randomly assign capital gains income to tax-filing units with the Monte Carlo technique._6 lsthe unemployment insurance (UI) amount that is added to AGI is computed as follows: If the sum of AGI and unemployment compensation is less than the limit for a particular return type (e.g., $20,000 for single, $25,000 for joint and $5,000 for other returns), UI is left out of AGI. If UI is not less than the return limit, the smaller of either (1) UI; or (2) one-half of AGI, plus the amount of UI over the limit, is added to AGI. _6The Monte Carlo technique is used to randomly assign capital gains to filing units as follows: ( 1)A random number (between zero and one) is generated for each CPS filing unit. (2) If this number is less than or equal to the probability that this filing unit will report capital gains, this type of income is imputed to its tax return. (3) If the random number is above this probability, no capital gains are imputed. The capital gains amount is calculated by multiplying a unit's AGI (before adding in unemployment compensation) by the average ratio of capital gains to AGI (as reported in the Statistics of Income). 95

110 TABLE The Probabilities and Proportions Used to Impute Capital Gains Reported in the 1978 Statistics of Income Adjusted Gross Probability Capital Gains As a Proportion Income Range of Reporting of AGI for All Those Who ( 1978dollars) Capital Gains Reported Capital Gains $0-$2, $2,000-$4, $4,000-$6, $6,000-$8, $8,000-$10, $10,000-$ $12,000-$ $14,000-$ $16,000-$ $18,000-$ $20,000- $ $25,000-$ $30,000-$50, $50,000-$100, $100,000-$ 200, More than $200, B-I Source: U.S. Internal Revenue Service, Statistics of Income, 1978, table Taxable Im'ome. Taxable income is computed by subtracting all allowable deductions and exemptions from the AGI. Filing units have the option of itemizing or taking standard deductions. The model uses the Monte Carlo technique to randomly determine which units itemize, based upon the 1978 SOl probabilities. Within each of sixteen AGI income groups, itemized deductions are assumed to be a fixed proportion of AGI income. These proportions are estimated from 1978 SOl data that show the size of itemized deductions as a proportion of AGI. These probabilities and the proportion of itemized income are shown in Table B-2. --Federal Taxes. Different sets of tax rates are used for those filing single, joint or head-of-household tax returns. Appropriate tax rates are used to calculate taxes to be paid on taxable income. Earned and elderly income tax credits are also computed. --State Taxes. All persons with positive federal tax liabilities are assumed to pay state taxes. These tax amounts are assumed to pay state taxes. These tax 96

111 TABLE B-2 The Probability of Itemization and the Proportion of Income Which Is Itemized A_usted Gross Proportion of Income Range Probability of Income Itemized by (1978 dollars) Itemizing Income Those Who Did Itemize $0-$2, $2,000-$4, $4,000-$6, $6,000-$8, $8,000-$10, $10,000-$12, $12,000-$14, $14,000-$16, $16,000-$18, $18,000-$20, $20,000-$25, $25,000-$30, $30,000-$50, $50,000-$100, $100,000-$200, Morethan $200, Source: U.S. Internal Revenue Service, Statistics of Income. 1978, table 1.4. amounts are assumed to be I7.8 percent of federal tax, based upon analysis of the ratio of federal income tax revenues to state income tax revenues tor individuals. Social Security Payroll Tax. Each person's Social Security tax is calculated directly from his or her earnings using the 1979 Social Security payroll tax formula. 97

112 APPENDIX C Methodology for Analyzing the Expected Pension and Social Security Benefit Levels Overview Estimating the distribution of retirement benefits and replacement rates requires: (1) the selection of a representative population group; (2) estimates of this population's family and work histories; and (3) estimates of their periods of Social Security and pension plan coverage. This information can then be used to calculate retirement benefits. An ICF model was used to perform these tasks for this report. It utilized the most recent available data on pension plan characteristics, coverage and labor force participation.l The model is specifically designed to estimate both Social Security and employer pension benefits for individuals and families under alternative policies. Its central focus is the analysis of alternative retirement and pension income policies. Currently, it is the only model which uses: (1) data from the Pension Supplement to the May 1979 Current Population Survey (CPS); and (2) detailed pension plan characteristics and provisions based on a representative sample of public and private sector plans. The model has four basic components: (1) a population data base consists of individual records from a modified version of the May 1979Current Population Survey, including imputations for nonresponses: (2) a workhistory simulation model uses dynamic aging simulation techniques to generate individual work, wage and family histories selected from the population data base: (3) a retirement benefit simulation model assigns each individual who is covered byan employer retirement program a representativepension plan sponsor, and then estimates the participants' retirement benefits; _Thismodel was originallydeveloped for the Office of Pension and Welfare Benefit Programsat the U.S. Departmentof Laborand the President'sCommissionon Pension Policy.Ithasrecentlybeenextensivelyrevisedforthe AmericanCouncilof LifeInsurance. 99

113 (4) a retirement income analysis model estimates replacement rates under alternative definitions for both individuals and families. These components and the key assumptions used are discussed below. Population Data Base Actual individual records from the May 1979 Current Population Survey, Special Pension Supplement, provide the population data base used in the wage history and retirement benefit simulation models. The Special Pension Supplement is a statistically representative sample of workers in the U.S. It includes approximately 30,000 households and contains information on the family structure and labor force characteristics of each individual surveyed. It also contains data on each individual's pension coverage, participation and vesting status. In a limited number of cases, statistical procedures are used to impute responses to labor force and pension questions that individuals failed to answer, e This data base is used as the basic population information utilized in the simulation models. For example, when analyzing the expected benefit receipt of workers age in 1979, the records of all workers age who were in the survey were selected and simulated. Work History Simulation Model To estimate future pension and Social Security benefits, a dynamic simulation model was developed. It estimates labor force experience of a large number of representative households drawn from the Special Pension Supplement. The work history model, as shown in Figure C- 1, uses various probabilities to determine year-to-year changes in family composition and labor force characteristics. These probabilities vary with individuals' demographic characteristics. They are taken from recent Bureau of the Census, Bureau of Labor Statistics and Social Security Administration data. The data sources used in these submodels and the factors by which the probabilities vary are also shown in Figure C-1.3 The model's outputs provide individual data records containing sufficient marital status and labor force information 2For further discussion on the problems that were encountered with the pension supplement and the statistical procedures that are used to correct them, see President's Commission on Pension Policy and Department of Labor, Office of Pension and Welfare Benefit Programs, "Background Analysis of the Potential Effects of a Minimum Universal Pension System (MUPS)," Appendix B (Washington, D.C., 1981). 3More detailed information on these probabilities can be obtained from ICF Incorporated upon request. 100

114 FIGURE C-1 Overview of Work History Simulation Model --Select INITIAL a cohort or all cohorts DATA BASE --Uses the enhanced May 1979 Current Population Survey data base or other data bases 4 I --Social 1. DEATH SecurityAND data DISABILITY and assumptions l 2. MARITAL STATUS TRANSITION --Using NCHS and Social Security data 3. HOURS WORKED ANNUALLY --By age, sex, marital status, education --From longitudinal analysis of CPS data 4. JOB CHANGE --By age, hours worked, years of service --From longitudinal analysis of CPS data 5. INDUSTRY OF EMPLOYMENT --For job changers, labor force entrants --From longitudinal analysis of CPS data I --From longitudinal 6. WAGE analysis RATES of CPS data I REPEAT PROCESS SIMULATION FOR EACH YEAR 101

115 to estimate both pension and Social Security coverage. The records also permit linking of data on different family members. The first file is then used as input to the Retirement Benefit Simulation Model. Retirement Benefit Simulation Model This model simulates work histories and data on the characteristics of a representative group of retirement plan sponsors. These simulations are used to estimate pension and Social Security retirement benefits. As shown in Figure C-2, this model uses five major steps to estimate retirement benefits. The first step in the retirement benefit simulation model estimates retirement plan coverage by using the individual work history data files. Coverage is assigned according to the following procedure: (1) First year pension coverage (1979) is determined directly from the modified May 1979 CPS data base, using respondents' indications of whether their employers sponsored plans. (2) As workers change jobs or enter the labor force during the simulation, pension coverage is estimated as a function of the individual's industry of employment, hours worked, age and indexed wage rate. These probabilities are based upon coverage rates reported for job changers and labor force entrants in the modified May 1979 CPS. (3) Pension plans are created for some workers who do not change jobs, using assumptions about industry coverage levels for the population group studied. These assumptions about plan creation and industry coverage rates are specified exogenously. They are shown in Table C-I. (4) If a job changer in a multiemployer plan remains in the same industry, he or she remains covered by the same plan. Otherwise, the individual is treated like a nonmultiemployer worker. In the second step, a pension plan sponsor, selected from a representative sample of employer retirement plan sponsors, is randomly assigned to each covered worker. This representative plan sponsor data base is stratified by industry and the multi/single employer status of the worker. It provides plan provision data for both primary and supplemental defined benefit and defined contribution plans offered by public and private sector sponsors. The data base includes all relevant plan provisions for participation, vesting, hours credited, retirement, death, disability, joint and survivors benefits and Social Security integration. Within an industry and multi/single employer category, the probability of being assigned to a given plan sponsor is proportional to the number of persons actually covered by that plan sponsor. The third step estimates Social Security and employer sponsored retirement benefits. If an individual is covered by Social Security, a Social 102

116 FIGURE C-2 Retirement Benefit Simulation Model WORK HISTORY MODEL J WORK HISTORIES FROM 1. PLAN COVERAGE DETERMINATION FOR EACH JOB I I 2. PLAN ASSIGNMENT FOR COVERED JOBS --From representative plan sponsor data base, including supplemental plans --By industry, employer category 3. SOCIAL SECURITY AND EMPLOYER RETIREMENT PLAN BENEFIT ESTIMATION --Participation standards --Vesting standards --Benefit formula 4. IRA ADOPTION DETERMINATION FOR NONCOVERED WORKERS --Probability --Estimated of adoption annual contribution 5. BENEFIT ACCUMULATION FROM ALL JOBS 103

117 104,fi

118 Security benefit is estimated based upon his or her work history data using the average indexed monthly earnings (AIME) formula. The model then estimates employer-sponsored retirement plan benefits by: (1) examining each worker's covered jobs to determine whether the worker's benefits are vested; and (2) using the plan's actual benefit calculation formula to determine the individual's pension benefit amount. When specified in the pension plan provisions, the model: (1) integrates benefits with Social Security; (2) calculates supplemental plan benefits; and (3) estimates survivors benefits. 4 This process is repeated for each job that an individual had. For every job a benefit record is produced. This record includes the first year the individual was covered, became a participant or became vested. This record also includes information on benefit amounts to be paid and the earliest age at which reduced and unreduced benefits are payable. In a fourth step, the model estimates whether noncovered workers will choose _.oestablish an individual retirement account (IRA). The probability of adopting an IRA is based on the worker's age and family earnings level. Where IRAs are determined to be established, an assumption about the annual contribution level is made. This is a function of two variables: (1) the worker's family earnings level; and (2) whether the IRA is an individual or a spousal IRA. 5 Benefits from employer-sponsored plans and IRAs are summed for all jobs to provide estimates of each individual's total retirement income. Retirement Income Analysis To simplify the analysis of retirement benefits under alternative policies, a retirement income analysis model was developed that is based on individual and household benefit records. This model estimates average individual and family retirement benefits and replacement rates according to preretirement and retirement income class. It permits examination of the relationship between benefit size and other selected characteristics, e.g., years of service, education, employment, wage rates, firm size, etc. 4Becausethismodelwasdevelopedpriortothe passageofthe 1981EconomicRecoveryTax Act,the provisionsfor tax deductibleemployeecontributionsare notincluded. SBecausethisanalysiswas conductedpriortothe passageof the 1981EconomicRecovery Tax Act, the new IRAlimitsare not includedin theseassumptions. 105