United States General Accounting Office GAO Report to the Ranking Minority Member, Committee on Ways and Means, House of Representatives June 1999 SOCIAL SECURITY REFORM Administrative Costs for Individual Accounts Depend on System Design GAO/HEHS-99-131 United States GAO General Accounting Office Washington, D.C. 20548 Health, Education, and Human Services Division B-282845 June 18, 1999 The Honorable Charles B. Rangel Ranking Minority Member Committee on Ways and Means House of Representatives Dear Mr. Rangel: The Social Security program forms the foundation for America’s retirement income system. In 1998, 31 million individuals and their dependents received retirement benefits of $265 billion through this program. In addition, 148 million workers currently contribute to the program in anticipation of future benefits. However, demographic trends, including the aging of the baby boom generation and increased life expectancy, threaten the program’s future solvency and sustainability. In response to this threat, various proposals to reform the program are currently under discussion. Proposed reforms range from traditional changes, such as reducing benefits and raising taxes, to more fundamental changes, such as creating a system of individual accounts for accumulating retirement savings. Under a system of individual accounts, workers would manage their own accounts, and the benefits they received from their accounts would generally be more closely linked to the amount of their contributions and to the gains or losses their investments incurred. Deciding whether and how to implement a system of individual accounts presents several difficult issues. Policymakers will need to consider how to finance the accounts and how they would affect the economy and program solvency, as well as how these accounts would affect the current Social Security benefit structure. In addition, policymakers will need to consider how readily individual accounts can be implemented, administered, and explained to the public.1 The cost of administering individual accounts is among the key factors to consider. The proposed accounts could provide greater individual choice in retirement investments and, according to proponents, would carry the potential for a higher rate of return on contributions than is available 1 In testimony earlier this year, we discussed how these issues can be used as criteria for evaluating reform proposals. See Social Security: Criteria for Evaluating Social Security Reform Proposals (GAO/T-HEHS-99-94, Mar. 25, 1999). Page 1 GAO/HEHS-99-131 Social Security Reform B-282845 under current law.2 However, some experts have asserted that the cost of administering individual accounts is also likely to be higher than the administrative costs of the current Social Security system, and this cost could reduce the amount of savings accumulated in the retirement accounts. Concerned about administrative costs and their effect on account accumulations and benefits, you asked us to determine (1) the factors that influence administrative costs, (2) the estimates that are available for administrative costs associated with individual accounts, and (3) how administrative costs might affect the accumulation of savings in individual accounts and the retirement benefits they provide. It is important to note that this report focuses on only one aspect of individual accounts—the administrative costs associated with them. It does not attempt to discuss how these individual accounts would be financed, how they might affect existing Social Security benefits, or other important issues related to implementing individual accounts. In addition, this report is designed only to illustrate the effects of administrative costs on account accumulations; it does not attempt to predict the effects of any specific proposals or variation in the rate of return on individual account investments. Today we are issuing another report that provides additional information on the key decisions to consider relating to the design and implementation of a system of individual accounts.3 Specific to this report, we conducted our review from October 1998 through May 1999 in accordance with generally accepted government auditing standards. (See appendix I for information on our scope and methodology.) When designing a system of individual accounts, the designers must make Results in Brief critical decisions about who would assume the new administrative and recordkeeping responsibilities, how much choice or discretion individuals would have in selecting and changing their investment options, and how workers would receive their benefits when they retired. The costs of administering a system of individual accounts would vary and would depend on these decisions and the types and level of customer service offered. Customer service features include, for example, how quickly funds are allocated to accounts, how frequently investors are informed of 2 Others, however, believe that returns on contributions are not the only goal of Social Security and that individual accounts are not the only way to increase rates of return. We will address the complex rate of return issue in a forthcoming report. 3 See Social Security Reform: Implementation Issues for Individual Accounts (GAO/HEHS-99-122, June 18, 1999). Page 2 GAO/HEHS-99-131 Social Security Reform B-282845 their account balances, and whether services are handled personally or through automation. While any system has administrative costs, they could be higher for more decentralized systems and for those offering broader investment choices, more customer service options, or both. Because most Social Security reform proposals that include individual accounts do not provide explicit details on how the accounts would be implemented and managed, it is difficult to accurately assess the costs of administering them. Available cost studies have limitations because they do not capture all the likely costs associated with a new system. For example, the cost of government oversight and enforcement activities needed to ensure compliance and the cost of providing public education are generally not included. However, the studies can at least provide a basis for understanding the possible range of administrative costs that individuals might incur under a new system. For example, estimates for a centralized system with limited investment choices and customer service are as low as 0.1 percent of assets per year, while the possible costs for a more decentralized system with broader investment choices or a system with extensive and flexible customer service are as high as 2 or 3 percent of assets annually. Although difficult to predict, administrative costs can have a significant effect on individual account accumulations. Our analysis, which assumed account contributions equal to 2 percent of an individual’s taxable earnings, illustrates this point. In our simulation, for a man who had average annual earnings every year for 45 years, a change in administrative cost from 0.1 percent to 1 percent reduced accumulations in his account by almost 22 percent. A change from 0.1 percent to 2 percent reduced his account accumulation by almost 40 percent. In more practical terms, he would accumulate $125,430 (in 1998 dollars) in his account under a 0.1-percent annual administrative cost, as opposed to $75,995 under a 2-percent administrative cost. The proportionate effect on accumulations of these changes in administrative costs was approximately the same for all workers in our analysis, regardless of whether they had low, average, or high annual earnings. Further, individuals may incur additional costs if they are required or choose to purchase an annuity, which ensures a steady stream of income throughout retirement. In the current market, the average administrative cost of purchasing an annuity is 5 percent of the amount being converted into the annuity.4 4 This 5-percent administrative cost does not include the additional cost of adverse selection, which is the risk to the annuity provider of having to pay benefits to those who live longer than expected. Page 3 GAO/HEHS-99-131 Social Security Reform B-282845 A number of proposals have been put forth to establish a national system Background of individual accounts; almost all the proposals would affect the Social Security program. Currently, Social Security provides retirement benefits to workers as well as benefits to disabled workers and the families of disabled, retired, and deceased workers. Depending on the reform proposal, individual accounts could replace part of the retirement benefits portion of the Social Security program, or the accounts could be added to the Social Security program. Also, some proposals would mandate worker participation in the system of individual accounts, while others would make such participation voluntary. In general, if the accounts were considered the personal property of individuals, the balances would be considered part of their estates when the account owners died. However, the proposals for individual accounts generally do not clearly delineate how the accounts would be structured and administered. Current Social Security The Social Security Administration (SSA) is responsible for the Program recordkeeping and benefit payment activities of the current Social Security program. The program is financed largely on a “pay-as-you-go” basis, in which the current year’s Federal Insurance Contributions Act (FICA) taxes are used primarily to pay that year’s benefits.5 Employers withhold the employee portion of FICA taxes from employees’ pay and regularly deposit the amount, along with the employer portion, in aggregate, in a designated Federal Reserve Bank or another authorized depository. At the beginning of the next calendar year, employers submit an Internal Revenue Service (IRS) W-2 form to SSA for each worker to report his or her earnings for the previous year. SSA checks this information and posts it to the earnings record it maintains for each individual worker. For tax year 1996, SSA received about 235 million W-2s.6 During this process, SSA and the IRS work together to verify that earnings are recorded in the proper amount in each individual’s record. It is important that the earnings are recorded correctly because these earnings, rather than the FICA amount, form the basis for calculating future benefits. A considerable time lag exists—as much as 7 to 22 months—between the time taxes are deducted from an individual’s earnings and the time these earnings are credited to the individual’s record. A worker must contact SSA to apply for retirement benefits. SSA calculates the retirement benefit and then sends information monthly to the Treasury for issuing a check or making an electronic deposit. 5 FICA taxes are generally composed of equal contributions from both employers and employees. 6 According to SSA, approximately 40 percent of workers receive more than one W-2 annually because they work for more than one employer during the year. Page 4 GAO/HEHS-99-131 Social Security Reform B-282845 Administrative Structure Certain key administrative activities must be performed under any system Options for Individual of individual accounts, much the same as they would under any defined Accounts contribution plan.7 These activities fall into three main categories: collecting contributions and keeping records, investing contributions, and paying benefits: • Collecting contributions and keeping records include enrolling participants, collecting and recording contributions, gathering and updating personal information on each individual (such as current address and investment choices), and correcting errors. • Investing contributions includes transferring the funds to the investment entity, conducting research to support buy and sell decisions, buying and selling investments, and recording gains and losses. • Paying benefits includes processing claims, handling appeals, and, depending on the type of payout option offered, issuing regular checks or processing annuities.8 When designing a system of individual accounts, the designers must make decisions about who would perform each of these administrative activities.9 Depending on who is chosen to assume these new activities—employers, individuals, private sector service providers, or the government—each could be affected in varying degrees. Figure 1 illustrates the three options we identified for the basic administrative structure of a system of individual accounts, each of which builds on an existing retirement system. A discussion of each option follows. 7 A defined contribution plan is a pension plan in which the contributions are allocated to individual accounts by a predetermined formula and benefits vary, depending on the contribution level and the return received from the investment of these contributions. 8 In purchasing an annuity, an individual contracts with an annuity provider, who provides a set monthly payment, usually over the lifetime of the individual, in exchange for an agreed-upon dollar amount. 9 In GAO/HEHS-99-122, we discuss in more detail the issues to consider when making each of these key decisions. Page 5 GAO/HEHS-99-131 Social Security Reform B-282845 Figure 1: Options for Account Administration and Recordkeeping Page 6 GAO/HEHS-99-131 Social Security Reform B-282845 The first option is based on a centralized recordkeeping system. A federal agency, such as SSA, could assume administrative and recordkeeping responsibilities. This way the structure would build on the current payroll reporting and tax collection system. Alternatively, a centralized clearinghouse could assume recordkeeping responsibilities, similar to how the federal government’s Thrift Savings Plan (TSP) uses the National Finance Center.10 The TSP is a tax-deferred defined contribution retirement plan for federal employees that contains features typically found in private sector 401(k) plans. The Federal Retirement Thrift Investment Board, an independent federal agency, manages the TSP. The Board holds the funds in trust, since they are owned by the participants, not the government, and thus are to be managed independent of political and social considerations. Federal employees may contribute each pay period either a percentage of their basic pay or a fixed dollar amount. All contributions, however, must be made through payroll deductions.11 Currently, employees can allocate their contributions into three TSP funds: the Government Securities Investment Fund (G Fund), the Common Stock Index Investment Fund (C Fund), and the Fixed Income Index Investment Fund (F Fund).12 Twice a year, employees can change their contribution amounts and how future contributions are invested. Employees can also transfer their account balances between funds monthly. Employees may take from their accounts loans that they repay through payroll deductions, using the interest rate for the G Fund at the time of the loan. Upon leaving the government or retiring, an employee’s account balance is paid through one of three options: (1) lifetime annuity, (2) lump sum payment, or (3) timed withdrawal in the form of monthly payments. Throughout this report, we discuss the government-managed and independently managed options for a centralized system together because they are similar in how information and money would flow from the employer through one central entity. Under either centralized system, the 10 The National Finance Center provides recordkeeping and payroll services for the Department of Agriculture and other federal agencies. For the TSP, the Center provides detailed recordkeeping and software development and maintains an office to provide service to participants. 11 The government automatically contributes 1 percent of basic pay for participants in the TSP who are covered by the Federal Employees Retirement System, regardless of whether the employees make personal contributions. For participants who choose to make personal contributions, the government matches the first 3 percent of their contributions at 100 percent and the next 2 percent of contributions at 50 percent. 12 The TSP will add two additional funds in May 2000—the Small Capitalization Index Investment Fund (S Fund) and the International Stock Index Investment Fund (I Fund). Page 7 GAO/HEHS-99-131 Social Security Reform B-282845 investments could be managed centrally by the recordkeeper or contracted out. A second option would build on the current decentralized system of 401(k) plans. A 401(k) plan is an employer-sponsored defined contribution plan that allows individuals to contribute, before taxes, a portion of their salary to a qualified retirement account. Unlike the previous option, under which all funds and information would flow through a centralized nationwide structure, the employer would collect individuals’ contributions and forward them directly to the investment manager. In the current system, the employer determines which services and investment vehicles the plan will provide, and, depending on the plan, individuals may choose how the assets are allocated among investment options. Under an employer-sponsored option, employers who do not currently offer 401(k) plans would bear the responsibility for creating an infrastructure to deposit contributions and provide employees with links to and choices among investment managers. Alternatively, some proposals suggest that a new system could build on the 401(k) system and permit individually managed or government-sponsored accounts for individuals who do not have access to a plan like a 401(k) plan through their employers. Finally, the third option would build on the decentralized individually managed system of individual retirement accounts (IRA). This structure does not involve employers in recordkeeping. Individuals send their personal funds to a financial service provider who places the money in an individual tax-deferred account. Individuals deal directly with their providers for services and payment of benefits. Under a decentralized individually managed option, workers would bear the responsibility for selecting an investment manager, or managers, and depositing their contributions. While any system of individual accounts has administrative costs, their Administrative Costs level will vary depending on the complexity of the system’s design. Design Increase With System complexity can be linked to four factors: (1) the administrative structure Design Complexity selected, including who maintains the records; (2) the flexibility in selecting and changing investments; (3) the level of customer service provided; and (4) the variety and type of options offered for paying benefits. In general, the more complex the system design and the more flexibility offered to individual investors, the higher the administrative costs. As a result, implementing a system of individual accounts involves Page 8 GAO/HEHS-99-131 Social Security Reform B-282845 making decisions about trade-offs between costs and flexibility. Table 1 summarizes the factors that can affect cost. Table 1: Factors That Affect Administrative Costs May decrease costs May increase costs Centrally managed recordkeeping Decentralized recordkeeping Limited investment choices Wide range of investment choices Limited customer services Varied and readily available customer services Mandated, centralized payout option Varied or decentralized payout options The administrative structure selected for a system of individual accounts—beginning with who is responsible for recordkeeping—will affect the costs of administering the system. A centralized management structure, whether run by SSA or a new centralized clearinghouse, could keep administrative costs down by taking advantage of economies of scale.13 For example, using one centralized system to record data and keep records on individual accounts for 148 million workers could minimize the costs per individual. Also, investing large sums of pooled contributions could lower transaction costs per account. However, centralizing these administrative activities would be likely to increase government administrative responsibilities and, under some proposals, increase government involvement in investment choices. Alternatively, a decentralized management structure could decrease direct government control over investment choices but could add to administrative costs. For example, if recordkeeping and management were distributed among a number of private companies, administrative costs per participant would likely rise. Moreover, depending on the investment options available, decentralizing recordkeeping and investments would be likely to increase the need for government regulation and oversight and the costs accompanying such activities. In general, regardless of the system’s structure, the principle of economies of scale suggests that as individuals’ accounts grow over time, the administrative costs per participant dollar should decrease. Other design features, such as the flexibility in selecting and changing investments, can also affect system costs. For example, administrative costs would be lower in a system that offered primarily index funds, such as those offered to federal employees under the TSP. These funds hold securities in proportion to their representation in the stock or bond 13 Centralization does not guarantee low costs from economies of scale. Achieving such economies requires planning, management, and oversight. Page 9 GAO/HEHS-99-131 Social Security Reform B-282845 markets and do not require significant research on individual companies or securities. However, this approach would result in relatively limited choices for investors. Conversely, when a wide spectrum of investment choices is offered, individual choice is enhanced but administrative costs are likely to rise, especially if the choices include more actively managed investments. These investments are accompanied by higher management fees because the investment manager spends more time and money on researching, selecting, buying, and selling investments. In addition, systems that offer individuals the option to transfer funds from one investment to another can have higher administrative costs. When workers frequently transfer their account balances from one investment to another, they may also incur extra costs to cover the additional administrative tasks and costs associated with buying and selling investments. A system of individual accounts that allows relatively free choice among different investment funds could encourage competition and lower costs. However, experiences in other countries have demonstrated that, under certain circumstances, competition may not achieve lower costs. For example, a recent study of the United Kingdom system, which includes accounts that are voluntary and decentralized, found administrative costs to be as high as 36 percent of an account’s value.14 The study linked these high costs, in part, to competition among providers that resulted in high marketing costs and frequent switching between investment providers.15 A fundamental decision for paying retirement benefits would be how much flexibility to offer individuals in the choice of payout options. The options to pay retirement benefits include lump sum payments, timed withdrawals, and annuities. Under a lump-sum payment option, individuals could liquidate their accounts through a single payment at retirement and choose when to spend or save that money. In a timed withdrawal, retirees specify a withdrawal schedule with the investment manager. Each month, they receive their predetermined amount while the balance of the individual account remains invested. Annuities can be structured in many ways and, therefore, may be more complex to administer and hence more costly than the other methods. However, they provide more long-term security because they ensure that benefits are available for the entire retirement lifetime. Permitting individuals to choose among all three 14 The 36 percent includes costs attributed to fund accumulation (that is, management and administrative costs) and costs of switching from one financial provider to another or stopping contributions altogether; it does not include annuitization costs. 15 Mamta Murthi, J. Michael Orszag, and Peter R. Orszag, “The Charge Ratio on Individual Accounts: Lessons From the U.K. Experience,” Birkbeck College Working Paper 99-2, University of London, London, Eng., Mar. 1999. Page 10 GAO/HEHS-99-131 Social Security Reform B-282845 options in the current market could further increase overall administrative complexity and cost by requiring systems to explain and keep track of the various choices. Finally, the types and level of customer service provided also affect the costs of a system of individual accounts. Customer service includes a range of activities designed to accommodate individuals’ investment choices or to inform them about the system or their accounts. These activities include providing statements of account balances, answering questions and providing educational material, facilitating transfers of balances between different investments, and calculating the gains or losses on investments at different intervals. When services are offered in greater numbers or with more frequency, the costs and administrative complexity of managing the accounts increase. For example, contributions can be deposited into the accounts at varying intervals, ranging from daily to annually. If contributions were made frequently, workers would benefit from earlier investment of funds, but the administrative recordkeeping costs would be likely to rise. Similarly, the frequency and means of providing information about the system or an individual’s account affect cost. Beginning in fiscal year 2000, SSA will be required to provide statements estimating future benefits to nearly all U.S. workers annually. Providing these statements will cost SSA more than $130 million per year. Some private pension plans may provide account statements monthly with higher attendant costs. In addition, the more personal the contact offered, the more expensive the service. For example, handling a call to a toll-free number can cost about five times as much as sending out an annual statement. Finally, if contributors are permitted to borrow from their accounts, administrative costs could increase because loans add a number of administrative tasks, including accepting applications, determining eligibility, and monitoring repayment. A number of means are used to calculate and report administrative costs for defined contribution plans. Some of the most common measures include • Expense ratio. This is a ratio that reflects total annual operating expenses as a percentage of accumulated fund assets. It is expressed either as a percentage of assets or in basis points. One basis point equals 0.01 or 1/100 of a percent; thus, 100 basis points equals 1 percent of assets. • Dollar cost per plan participant. This is usually a flat fee that is charged to each participant. It could be an annual assessment or a one-time charge. It Page 11 GAO/HEHS-99-131 Social Security Reform B-282845 may be the only fee charged or it may be charged in addition to fees reflected in a percentage fee, such as an expense ratio. • Other ratios. Administrative costs can also be calculated as a percentage using bases other than accumulated assets, including total benefits paid, number of workers and retirees covered by the system, income per capita, or contributions. Not all Social Security reform proposals that include individual accounts More Specifics on provide explicit detail on how the accounts would be implemented and System Design Are managed, and this makes it difficult to determine accurately the Needed to Accurately administrative costs of a system of individual accounts. Studies of the possible costs are available but have limitations for a number of reasons. Estimate Some are based on the reported costs of existing systems, which often fail Administrative Costs to capture the full administrative cost of those systems. Further, the studies do not include the costs of added responsibilities that could be required under a new system, such as the cost to the government for monitoring and oversight. Despite their limitations, however, the studies help shed light on the possible range of administrative costs. They vary in their approaches, but overall, as expected, they predict that costs would increase when accompanied by decentralized recordkeeping, more and varied investment options, and greater levels of customer service. Cost Estimates Increase To better understand the possible costs of administering a system of With Design Complexity individual accounts, we identified a number of studies that provide either but May Be Understated estimates of these costs or information on the actual costs of roughly comparable systems. These studies provide a useful starting point for thinking about what the costs might be under various system structures and designs. However, in some cases, both the estimates and actual costs of current systems may understate the full range of costs that may be associated with individual accounts. Table 2 provides a range of costs for a system of individual accounts under the three basic administrative structure options that are currently being discussed. The costs are expressed in expense ratios, which identify the percentage of fund assets that are deducted from the fund total for administrative expenses before gains or losses are posted to individual accounts. The estimates vary in how they were constructed and in the assumptions they make about the structure of the proposed individual account system, in part because there is a wide range of possible options for structuring a system of individual accounts. Estimates for a centralized Page 12 GAO/HEHS-99-131 Social Security Reform B-282845 structure generally rely on continuing SSA’s current centralized recordkeeping system rather than creating a new system and assume that SSA or some central clearinghouse would contract for a limited number of investment options and provide a basic level of customer service, similar to the TSP.16 To approximate the possible administrative costs for a decentralized employer-sponsored system of individual accounts, we used costs of current 401(k) plans.17 Studies of costs in current 401(k) retirement plans can provide useful information because a nationwide system of employer-sponsored plans could be similar to 401(k)s, depending on the size of the plans and any new requirements that might be imposed. The costs of a decentralized individually managed system are based largely on the administrative costs of mutual funds, because they are the most common retail investment instrument individuals use in their IRAs. (For more information on the studies and why we chose these estimates, see appendix I.) Finally, this section of the report is about the ongoing administrative costs associated with building up an account balance. Because individuals often do not incur the cost of purchasing an annuity until they retire, we discuss the effects of annuity costs later in this report. 16 Unlike the TSP, most estimates do not include the costs associated with permitting contributors to borrow from their accounts before retiring. 17 Some employers, while supporting the individual account concept, have expressed reluctance to take on additional administrative responsibilities, and there has been less discussion and analysis of employer-sponsored individual accounts than of the two other options. Page 13 GAO/HEHS-99-131 Social Security Reform B-282845 Table 2: The Range of Administrative Costs Under Discussion Administrative Annual cost Additional information on the source and study structure Sourcea as % of assets approach Centralized Advisory Council on 0.11 The Report of the 1994-1996 Advisory Council on Social Security Social Security considered a centralized individual account option. The administrative cost estimate was a consensus of the Council members’ opinions. Employee Benefit Low = 0.10 EBRI, a nonprofit nonpartisan organization dedicated Research Institute (EBRI) High = 2.0 to public policy research on economic security and employee benefits, considered two possible systems—one with a low level and one with a high level of service features—based on costs of 401(k) plans and other investment companies and on discussions with other experts. James and others Low = 0.14-0.18 The authors, experts on employee benefit plans, High = 0.49-0.79 analyzed data on institutional funds to estimate the costs for a centralized system with passively managed (low) and actively managed (high) funds. They constructed a “total fund expense profile,” which includes all the costs for mutual funds.b State Street Corporation 0.19-0.34 State Street Corporation, a private financial services firm, based its estimates on the unit costs of the various administrative activities that would be required under a system of individual accounts, such as recordkeeping and computer system maintenance. The estimates shown here are for year five of the Corporation’s proposed system. Decentralized Mitchell 0.28-1.88 The author, an expert on employee benefit plans, employer-sponsored published a number of studies of the costs of existing 401(k) and other retirement plans. Pension and Welfare 0.3-3.0 Department of Labor’s PWBA is responsible for Benefits Administration overseeing and regulating the nation’s 401(k) plans. (PWBA) The cost range reflects the administrative costs of 401(k) plans and the opinions of PWBA officials. (continued) Page 14 GAO/HEHS-99-131 Social Security Reform B-282845 Administrative Annual cost Additional information on the source and study structure Sourcea as % of assets approach Decentralized James and others 0.32-1.50 The authors used the costs of retail mutual funds to individually manged estimate the costs for a decentralized system. They constructed a “total fund expense profile,” which includes all the costs for mutual funds.b Investment Company 0.46-1.49 ICI, the national association of the American mutual Institute (ICI) (Rea and fund industry, studied trends in costs for equity mutual Reid, 1998, 1999) funds, bond mutual funds, and money market mutual funds. It used a cost measure called the “total shareholder cost,” which incorporates all the costs for a mutual fund.b Advisory Council on 1.0 The Report of the 1994-1996 Advisory Council on Social Security Social Security contained a more decentralized option that would permit workers considerable flexibility in their investment decisions and assumed a contribution of 5 percent of taxable payroll. The administrative cost estimate was based on the costs of existing similar systems, such as mutual funds. a Full bibliographic data for these sources are given in the bibliography. b The cost to buy and hold mutual funds includes two primary categories: shareholder transaction fees, which are one-time fees that can be imposed when the funds are bought or sold, and annual operating expenses, which include the costs of operating the funds on an ongoing basis, such as the cost for administrative expenses, compensation for the funds investment adviser, and advertising costs. The Costs of a Centralized As shown in table 2, the estimates of administrative costs for centralized Administrative Structure systems of individual accounts range from 0.10 percent to 2 percent of assets. Overall, a centralized system is expected to be less expensive than a decentralized system when customer service and investment choices are held constant, because of the economies of scale that could arise from having centralized contribution collections, recordkeeping, and communications with participants. However, costs generally increase as the number and type of investment options expand. For example, James and others estimated that administrative costs would be between 0.14 percent and 0.18 percent of assets with passively managed index portfolios and between 0.49 percent and 0.79 percent with actively managed portfolios. Alternatively, costs may rise from variation in the customer service features provided. For example, the Employee Benefit Research Institute (EBRI), a private nonprofit research organization, assumed that providing a higher level of service, such as the daily valuation of accounts, allowing loans before retirement, investor education, and other services, could raise administrative costs from a low of 0.10 percent to a high of 2 percent of assets. State Street Corporation, a private financial services company, provided the most detailed analysis of Page 15 GAO/HEHS-99-131 Social Security Reform B-282845 costs per administrative function based on known costs. Its estimate relied on a number of customer service assumptions, including the assumptions that participants would have little reason to call and that a large percentage of inquiries would be handled through means that cost less than person-to-person contact, such as automated telephone menus and the Internet. The estimates for a centralized system may be understated because they do not take into account changes required in the administrative foundation. The estimates all rely on the current payroll reporting and recordkeeping system as the administrative foundation; however, SSA officials told us that depending on the structure and expectations of a centralized system of individual accounts, the agency might need to make significant and costly changes to its recordkeeping system. Under SSA’s current recordkeeping system, it can take as long as 7 to 22 months from the time FICA taxes are withheld to the time earnings are posted to individual records. Under a system of individual accounts, this time lag could result in lost returns on investments in cases in which the value of an individual’s chosen investment rises before the individual’s contribution can be invested.18 Also, SSA currently does not follow up with employers for reporting errors under a certain dollar threshold, since benefits are not significantly affected by these errors. In addition, each year SSA cannot post as many as 1.5 percent of the earnings reported to any individual record because of missing or erroneous identifying information. Under a system of individual accounts, in which the benefits would rely on the dollar amounts contributed, these errors could be problematic. It is not yet clear whether any of these practices would need to be revisited or would be acceptable under a new system.19 Finally, depending on system design, SSA officials said they could also incur significant additional customer service costs, such as an increase in calls from individuals inquiring about contributions or account balances. Changes in reporting and recordkeeping requirements could also affect the IRSand employers. The IRS could incur additional administrative costs as it collects and reconciles the FICA taxes, especially if a new system required tracking individual account contributions separately from these taxes. Moreover, employers could be affected if the system required changes to 18 Some proposals contain alternative measures that could mitigate the effect of this time lag. For example, contributions could be pooled together and invested in a safe investment vehicle, such as a money market fund, until they are allocated to individual accounts, at which time the investment earnings could also be credited to the individual accounts. 19 For more detailed information on these and other recordkeeping factors to consider when designing individual accounts, see GAO/HEHS-99-122. Page 16 GAO/HEHS-99-131 Social Security Reform B-282845 the current wage-reporting documentation and procedures or if employers were required to prepare and submit information on individuals more frequently than the current annual reporting requirement in order to hasten the posting of information to individual accounts. We did not find any estimates of these possible additional costs, and it is not clear who would bear these costs under a new system. Costs to SSA and the IRS could be funded through general revenue, or they could be deducted from individual accounts. Costs to employers under a centralized system could be included in their normal costs of operation. However, these costs could be passed along to individuals through other means, such as a reduction in other employer-sponsored retirement benefits. Finally, the TSP has been cited as a model for a centralized system of individual accounts. Administrative costs for the TSP were 0.08 percent of assets in 1998.20 However, a former official from the Federal Retirement Thrift Investment Board noted that managing the TSP differs in important ways from managing a national system of individual accounts. The federal workforce and the federal government, as a single employer, differ substantially from the group that would be covered under a nationwide system. For example, the federal workforce experiences less job turnover, tends to be older, and has higher average earnings than the general workforce. In addition, federal agencies experience greater stability and have greater access to automation than the employer population at large. Serving a more diverse population of investors with a wide variety of employers would likely result in higher administrative costs, according to this former official. In addition, the administrative costs for the TSP do not include the services federal agencies provide on behalf of the plan, such as enrolling individuals and working with the recordkeeper. The Costs of a Decentralized Mitchell reported that the costs for 401(k) plans holding mutual funds Employer-Sponsored Structure ranged from 0.28 to 1.88 percent of plan assets. These costs include the costs of small and mid-sized 401(k) plans, which ranged from 0.28 to 1.32 percent, and the costs of plans that held mostly mutual funds, which ranged from 0.84 to 1.88 percent. These estimates, however, do not include all possible expenses. In some cases, they exclude money management fees, while in others they exclude recordkeeping fees. Mitchell concluded that privatization options would be expected to have somewhat higher 20 When TSP participants leave the federal government before they are entitled to their retirement benefits, the accumulated government contributions and the earnings on them are used to help offset administrative costs, which decreases the gross administrative cost from 0.08 percent of assets to a net cost of 0.06 percent. Page 17 GAO/HEHS-99-131 Social Security Reform B-282845 administrative costs than the current system but that these higher costs might be offset by economic benefits from a privately managed system. Officials from the Department of Labor’s Pension and Welfare Benefits Administration (PWBA) provided a wider range of administrative costs: In 1998, the 401(k) market contained about 250,000 plans covering approximately 25 million individuals, with costs ranging from 0.3 percent to 3 percent of assets. Although the costs of existing 401(k) plans provide a basis for approximating costs for a decentralized employer-sponsored system of individual accounts, the full cost of administering existing employer-sponsored plans is difficult to measure. In 401(k) plans, for example, employers often contract with a plan administrator to provide needed services, which may include keeping records, managing investments, or providing information.21 These arrangements and the services provided vary widely among plans and may result in underreporting of the plans’ full costs. For example, in some employer plans, much of the payroll collection, recordkeeping, and benefit payment activities may be handled by the employer in-house and are not necessarily billed to the pension plan’s accounts. Further, 401(k) costs included in annual reports submitted to PWBA do not include investment management expenses debited directly from the earnings that accrue in the participants’ accounts. Under a national system of individual accounts, these differences in services and how costs are allocated could raise questions of fairness. More uniformity in the way plan costs are allocated between the employer and the individuals might be called for in order to ensure more equitable benefits across plans and to facilitate public understanding and oversight of the system. Finally, the demographic characteristics of employees covered by 401(k) plans or other employer-sponsored retirement plans differ from those of the group of individuals who would be covered under a mandatory nationwide system. According to EBRI, employees covered by employer-sponsored plans have higher earnings and lower job turnover than the general workforce. The Costs of a Decentralized Analysts estimate that the administrative costs for a decentralized Individually Managed Structure individually managed system would range from 0.32 percent to 1.5 percent of assets. Variation in the estimates stems, in part, from differences in investment strategy and the range of investment choices offered, the level 21 Some recordkeeping costs faced by 401(k) and other employer-sponsored plans, such as costs resulting from compliance with plan participation requirements, may not be applicable under a mandatory nationwide system. Page 18 GAO/HEHS-99-131 Social Security Reform B-282845 of service provided to the investor, and the level of marketing and communication done by investment firms. It is important to note, however, that this range of estimates does not reflect the lowest or highest cost that an investor could incur. The estimates are based on average mutual fund costs. In a system offering a wide range of investment choices, some individuals could incur costs lower than these averages, while some could incur higher costs because some mutual fund companies charge higher administrative fees. In our analysis of 1998 data provided by Morningstar, Inc., we found that administrative costs for more than 9,300 mutual funds with more than $1 million in assets ranged from as low as 0.01 percent to as high as 7.34 percent of assets, with an average of 1.33 percent of assets.22 Other Costs Are Not In addition to the limitations discussed above, most of the cost estimates Included in Reform do not capture the significant costs associated with starting up a new Discussions system and those for the additional responsibilities, including government oversight and public education, that would probably result from a system of individual accounts. These additional costs could be borne by employers, the government, individuals, or some combination, depending on the structure and design of the account system. Start-Up Costs Since no current system is available to handle a national system of individual accounts, some additional costs would be incurred to create such a system, regardless of the structure selected. These start-up costs include the costs of developing or adapting computer systems, establishing electronic links between recordkeepers and investment firms, informing and educating the public about the changes and about available investment options and their risks and costs, hiring and training new staff, and establishing or expanding an infrastructure for communicating with and serving the public. Any of these services could be provided through contractual arrangements. Under a centralized system, the majority of these costs would be borne by the government. Alternatively, under a decentralized employer-sponsored structure, a significant number of employers would incur start-up costs because they do not provide pension plans. Currently, about 57 percent of private sector workers are not covered by an employer-provided pension plan. Their employers would need to develop an infrastructure to deposit contributions and convey employees’ choices to investment managers, unless provisions were made 22 We analyzed the data Morningstar, Inc., reported as annual expenses and did not include those reported as one-time fees. (Morningstar, Inc., is a private investment research firm that maintains a proprietary database on U.S. mutual funds, stocks, and other financial vehicles.) Page 19 GAO/HEHS-99-131 Social Security Reform B-282845 to permit the uncovered employees to invest through a different vehicle, such as an IRA. Finally, under a decentralized individually managed structure, financial service providers could incur some start-up costs to increase their overall capacity. Little historical information is available on start-up costs for very large retirement systems. Separate funding was provided for the start-up of the TSP, which included costs for hiring and training new staff, software development, printing materials, and other initial activities. According to EBRI, start-up costs for the TSP averaged about $5.00 per participant in the first year and, when translated into 1998 dollars, would equate to a start-up cost of $1.08 billion for a similar system provided on a national scale. In addition, these costs could be understated because TSP was able to rely on the already established National Finance Center for its recordkeeping, which had a computer system in place that included records on a portion of the federal workforce. However, a national system of individual accounts would be much larger than the TSP and could benefit from economies of scale in start-up costs as well. The State Street Corporation estimated that costs for the first year of operation of a national system of individual accounts, including start-up costs, would range from 0.7 to 1.34 percent of fund assets. The Corporation’s cost estimates then decrease gradually to 0.27 to 0.51 percent of assets in year three and to 0.19 to 0.34 percent in year five. Costs to the Government Under a new system of individual accounts, the government would be taking on additional management activities, expanding oversight and regulatory responsibilities, or both, depending on the administrative structure selected. Under a centralized system, the full costs of the government’s taking on a new role are difficult to predict, as stated earlier. In addition, if any kind of decentralized structure were adopted, the government would be likely to incur additional costs for oversight and enforcement activities needed to ensure compliance, and these costs are not reflected in any of the estimates. Depending on the structure, different agencies would be affected. Under an employer-sponsored system, the government would be likely to play a larger role in ensuring that employers properly transfer an individual’s contributions to the investment manager. For example, PWBA officials estimated that they would need to dramatically increase their investigative staff of 350 if a system of individual accounts had a 401(k) structure and if they were responsible for oversight. Further, if the structure of a system of individual accounts involved more open-market investments, more trading activities would occur, thus increasing the need for oversight by the Securities and Page 20 GAO/HEHS-99-131 Social Security Reform B-282845 Exchange Commission (SEC) or another government entity. SEC has broad responsibility over the securities markets, as well as the market intermediaries who provide brokerage services and operate mutual funds. SEC officials stated that if an individually managed structure were selected, the government would need to enhance its oversight efforts to protect new investors.23 Moreover, if the individually managed system were mandatory, some federal agency would likely be responsible for monitoring individual compliance. Finally, how the contributions are collected or distributed could create additional government responsibilities. For example, if contributions were provided through a tax rebate, new systems for providing the rebate and monitoring the process would be required. Officials in key federal agencies that could be affected by the creation of a system of individual accounts told us that their agencies had not yet developed a full estimate of the cost of oversight for these possible changes. In addition, depending on the design, a new centralized system could require SSA to keep two parallel systems running at once—one for tracking and paying traditional Social Security benefits and one for tracking and paying individual accounts under a new system. Also, if the government offered any sort of minimum benefit guarantee, SSA would most likely be required to monitor benefits under both systems and calculate benefit payments accordingly. The costs of these activities are also difficult to estimate, according to SSA officials. Costs to Educate the Public Regardless of the design of a new system of individual accounts, changing Social Security would require educating the public about the new program’s purpose and features. The costs of this initial education would most likely be borne by the government but could be shared by the employers or investment managers. A national system of individual accounts would require educating some workers who have never invested before.24 According to SSA staff, information on changes to the program would most likely be sent to every working individual through the mail. They estimate that the minimum mailing cost would be $0.50 per letter, which totals more than $70 million per mailing. Because individual accounts would include new types of information, SSA believes that it would also need to significantly redesign the personalized annual statement it currently sends. In addition, the government would probably 23 The need for enhanced oversight could vary, depending on the investment options available. For example, permitting individuals to choose among a few mutual funds would require less government oversight than if individuals were given a wide range of choices among different investment vehicles. 24 In a forthcoming report, we will discuss in more detail the need for public education and the effect of individual accounts on national savings. Page 21 GAO/HEHS-99-131 Social Security Reform B-282845 arrange for public service announcements on television and radio to heighten people’s awareness. Furthermore, since most proposals for individual accounts provide investment choices, additional education would be needed to help individuals understand their investment options and their associated risks or costs. In addition, depending on the system’s design, it would be important for individuals to understand how increased customer service and other options, such as frequently changing investments, could affect administrative costs. The government, employers, and investment managers might each have some role and incur some costs for this ongoing investment education, depending on system design. The cost of public education and who would be responsible for providing it are not included in all the cost estimates we reviewed. Although the precise cost of a system of individual accounts is difficult to Costs Affect Account predict, available information can be used to illustrate the effect that Accumulation and different levels of administrative costs could have on individual account Retirement Benefits accumulation and retirement benefits. Our analysis shows that the level of administrative costs passed on to individuals could have a significant effect on the balance of funds that would accumulate in their accounts, as well as on the retirement benefits their accounts would provide. Account Accumulations To illustrate the effect of different levels of administrative costs, we used a Decrease as Costs Increase model of the Social Security system to simulate the balances that would accumulate in a system of individual accounts for selected workers, given four different annual administrative costs—0.1, 0.25, 1, and 2 percent of assets. We selected these costs because they fall within the range of possible costs presented in table 2.25 We assumed a system of individual accounts that was established beginning in 2002 and simulated the balances that would accumulate by retirement for workers born in 1984, who would participate in the new system throughout their careers with low, average, and high gender-specific annual earnings. We further assumed that these workers made annual contributions of 2 percent of their taxable earnings that started at age 22 and ended with retirement at 25 We did not simulate the effect of a 3-percent cost because this amount was notably higher than estimates by others. Page 22 GAO/HEHS-99-131 Social Security Reform B-282845 age 67.26 We use 1998 dollars to report our simulation results. (See appendix I for further details on our methodology and the model.) Figure 2, which illustrates the change in accumulations for a working man with average annual earnings throughout a 45-year career, shows that the accumulated balances would decrease significantly as the annual administrative cost increased. In our simulation, changing from an administrative cost of 0.1 percent to 1 percent would reduce the account accumulations by more than 22 percent, and changing from an administrative cost of 0.1 percent to 2 percent would reduce the account accumulation by almost 40 percent. For example, a man born in 1984 with average annual earnings who worked from age 22 to age 67 would accumulate $125,430 in a system with 0.1 percent annual administrative costs and $75,995 if the administrative costs were 2 percent annually. The proportionate effect on accumulations was approximately the same for all workers, regardless of whether they had low, average, or high annual earnings. 26 We did not attempt to address the financing issues related to a system with 2-percent contributions. We treated the individual accounts as an addition to the current Social Security program. Page 23 GAO/HEHS-99-131 Social Security Reform B-282845 Figure 2: The Effect of Administrative Cost Changes on Accumulated Account Balances for a Man With Average Annual Earnings Throughout a 45-Year Career It is important to note here, and throughout this discussion, that higher administrative costs could be associated with more customer service and, potentially, with higher investment returns or investment portfolios that more closely matched individual needs. However, higher investment returns are not consistently correlated with higher administrative costs. Many actively managed investment options have not been able to generate higher returns than broad market indexes. It should also be noted that our simulation made a number of simplifying assumptions, which, if changed, would further affect accumulations. For instance, we assumed that individuals would have earnings every year from age 22 until the normal retirement age of 67, when in reality many individuals have discontinuous work histories and retire before the normal age. To continue our illustration of a man born in 1984 with average annual earnings in a system with a 2-percent annual administrative cost, he would accumulate $65,214 by the early retirement age of 62, as opposed to Page 24 GAO/HEHS-99-131 Social Security Reform B-282845 the $75,995 balance he would accumulate by the full retirement age of 67. In addition, we chose to evaluate the effects on individual accounts regardless of the person’s marital status. We found, as expected, that women’s accumulations would be smaller than men’s because the average annual earnings for women are significantly lower than those for men. However, some proposals would allow earnings-sharing between married persons, which could help mitigate some of the disparity between women and men. Our analysis also assumed administrative costs would be withheld from the earnings for each account through an annual percentage fee. However, for small account holders, the method used to assess administrative costs can make a difference. Under a system of individual accounts involving contributions of 2 percent of taxable earnings, many individuals would have small account balances. For example, individuals who earned $30,000 annually would contribute only $600 into their individual accounts each year. More than 64 percent of the working population earned less than $30,000 in 1997. If individuals were charged a flat fee per account for administrative costs, accumulations in small accounts would be affected to a greater extent than if they were charged an annual percentage. Other alternatives to mitigate the effects of administrative costs on small accounts are available. For example, some analysts have suggested pooling the funds of small accounts into one single fund until the accounts reach a certain minimum balance, where they would be less vulnerable to the effects of administrative costs. Additional Payout Costs When individuals make the decision to retire, they may bear all or some of May Affect Retirement the costs associated with the payout from a system of individual accounts. Benefits For our analysis, we assumed that each individual would pay a one-time fee to purchase his or her retirement annuity. In the current market, the average for the costs to cover the administration of an annuity is a one-time charge of about 5 percent of the amount being converted into the annuity.27 It is important to note that these costs vary widely and can be lower or much higher than the average. The administrative cost of purchasing annuities includes maintaining records, making payments, and providing services to the annuitant as well as some profit margin for the 27 James M. Poterba and Mark J. Warshawsky, “The Cost of Annuitizing Retirement Payouts from Individual Accounts.” National Bureau of Economic Research, Cambridge, Mass., Jan. 1999. Page 25 GAO/HEHS-99-131 Social Security Reform B-282845 annuity provider.28 In addition to these administrative costs, individuals who currently purchase annuities in the private market pay additional costs because of “adverse selection.” In the current market, individuals who expect to live a long time are much more likely to purchase annuities than are individuals whose life expectancies are shorter. As a result, to cover the risk of having to pay benefits to those who live longer, annuity providers charge more for annuities than they would if every individual purchased an annuity at retirement. The cost of insuring against adverse selection could cost an individual as much as an additional 12 percent (above the 5-percent administrative costs) of the amount being converted into the annuity.29 If a new system of individual accounts were mandatory, the adverse selection cost might be somewhat mitigated, but there would still be a cost to administer the annuity. Therefore, continuing our example, if the man with average annual earnings who accumulated $75,995 in a system with a 2-percent annual administrative cost were to purchase an annuity with a 5-percent fee, he would pay about $3,800 in administrative costs to purchase an annuity that would provide an average annual benefit of $5,584. Administrative costs are but one of many important issues to consider Conclusions when deciding whether and how to create a system of individual accounts as part of Social Security reform. However, because they can affect the amount of savings individuals are able to accumulate through individual accounts, they can be a key element of a reform package that includes individual accounts. Because program design affects the level of administrative costs and who bears them, policymakers will need to assess the possible costs and trade-offs for each of the various options under consideration. Individual account structures with lower administrative costs are often associated with more restricted investment choices and more centralized management, while increasing individual choice and decentralizing the management structure could result in increased costs. Finding the right balance will depend, in part, on the goals of the new program. Moreover, steps can be taken under any system to help mitigate the effect of administrative costs, especially on small account-holders. These include limiting customer service options or pooling small accounts. 28 In a forthcoming report, we will provide a more detailed discussion of the factors that affect the costs associated with purchasing an annuity and how these costs may factor into a system of individual accounts. 29 Poterba and Warshawsky, “The Cost of Annuitizing Retirement Payouts from Individual Accounts.” Page 26 GAO/HEHS-99-131 Social Security Reform B-282845 Finally, although the effect of administrative costs may be offset by higher returns on investments, it may not be. If a system of individual accounts is implemented, the public will need to fully understand how its choices will affect the administrative costs it will incur and how these costs affect account accumulations. If a decentralized structure is chosen or if diverse investment choices and varied customer service options are offered, it will be especially important to ensure that the public has access to easy-to-understand information on the costs of investment options and on the effect the costs can have on its accounts and retirement benefits. We provided a draft of this report to SSA, IRS, SEC, PWBA, the Department of Agency and Other the Treasury, and the Federal Retirement Thrift Investment Board, as well Comments and Our as to external reviewers who are experts in related areas. In commenting Response on our report, the reviewers generally agreed with our characterization of the factors that influence administrative costs and the possible range of costs under a system of individual accounts. They provided comments to us in either oral or written form. These comments were primarily technical and clarifying in nature. In addition to submitting technical comments, SSA stated that we should expand our discussion of the costs of compliance and customer service. We expanded our discussion of these issues. Furthermore, SSA and others suggested that we provide additional detail on the costs associated with annuities. Because this issue is a key focus of a forthcoming report, we did not expand on it in this report. The written comments are printed in appendix II. As agreed with your office, unless you publicly announce its contents earlier, we plan no further distribution of this report until 30 days from the date of this letter. At that time, we will send copies to the Honorable Bill Archer, Chairman of the House Ways and Means Committee; other interested congressional committees; the Honorable Kenneth S. Apfel, Commissioner of Social Security; the Honorable Alexis M. Herman, Secretary of Labor; the Honorable Arthur Levitt, Chairman of the Page 27 GAO/HEHS-99-131 Social Security Reform B-282845 Securities and Exchange Commission; the Honorable Roger W. Mehle, Executive Director of the Federal Retirement Thrift Investment Board; and the Honorable Robert E. Rubin, Secretary of the Treasury. Copies will be made available to others upon request. GAO contacts and staff acknowledgments are listed in appendix III. If you have any questions concerning this report, please contact me on (202) 512-7215. Sincerely yours, Cynthia M. Fagnoni Director, Education, Workforce, and Income Security Issues Page 28 GAO/HEHS-99-131 Social Security Reform Page 29 GAO/HEHS-99-131 Social Security Reform Contents Letter 1 Appendix I 32 Objectives, Scope, and Methodology Appendix II 36 Comments From the Social Security Administration Appendix III 38 GAO Contacts and Staff Acknowledgments Glossary 39 Bibliography 40 Related GAO Products 44 Tables Table 1: Factors That Affect Administrative Costs 9 Table 2: The Range of Administrative Costs Under Discussion 14 Figures Figure 1: Options for Account Administration and Recordkeeping 6 Figure 2: The Effect of Administrative Cost Changes on 24 Accumulated Account Balances for a Man With Average Annual Earnings Throughout a 45-Year Career Page 30 GAO/HEHS-99-131 Social Security Reform Contents Abbreviations EBRI Employee Benefit Research Institute FICA Federal Insurance Contributions Act IRA individual retirement account IRS Internal Revenue Service PWBA Pension and Welfare Benefits Administration SEC Securities and Exchange Commission SSA Social Security Administration TSP Thrift Savings Plan Page 31 GAO/HEHS-99-131 Social Security Reform Appendix I Objectives, Scope, and Methodology This appendix provides detail about our review and analysis of the estimates of administrative costs for individual accounts. For this report, we addressed three key questions: 1. What factors influence administrative costs? 2. What estimates are available for administrative costs associated with a system of individual accounts? 3. How might administrative costs affect the accumulation of savings in individual accounts and the retirement benefits they provide? To address the first two questions, we met with officials from the federal agencies that would be affected by a system of individual accounts, including the Social Security Administration, Internal Revenue Service, Securities and Exchange Commission, Department of Labor’s Pension and Welfare Benefits Administration, Pension Benefit Guarantee Corporation, and the Department of the Treasury. In addition, we met with experts in the areas of Social Security and pension reform, as well as employer representatives, payroll processors, investment managers, and annuity providers, to obtain a more detailed understanding of the costs involved in managing accounts. We also reviewed the experiences of organizations and other countries related to the administrative costs of individual accounts. Further, we reviewed the actual costs for administrative activities of some defined contribution pension plans, and we reviewed several studies of the estimated administrative costs of a new system of individual accounts. The actual and estimated costs we reviewed, with the exception of the estimates of the Advisory Council on Social Security, are generally not associated with any particular reform proposal but are, rather, an effort to predict the administrative costs under a variety of proposals. To address the third question, we selected a range of cost estimates from those under discussion and, using a Social Security policy simulation model, projected their effects on account accumulations and retirement benefits. To identify administrative cost estimates and the factors that influence Studies We Reviewed them, we reviewed the literature and evaluated estimates from a number of different sources. We limited our review to studies that were relatively Page 32 GAO/HEHS-99-131 Social Security Reform Appendix I Objectives, Scope, and Methodology comparable in their assumptions, methods, and formats.30 This led us to focus on a set of estimates provided by a few specific studies.31 These studies varied in the data that they used to calculate their estimates, as well as in the assumptions they made about the structure of the proposed individual account system. Studies that estimated administrative costs for a centralized system of individual accounts included work by the Advisory Council on Social Security, work published by the Employee Benefit Research Institute, and work by Estelle James of the World Bank and others and by State Street Corporation. Studies that estimated administrative costs for a decentralized individually managed system of individual accounts included work by James and others and by the Advisory Council on Social Security. We also used work published by the Investment Company Institute. We found no studies that estimated administrative costs for a decentralized employer-sponsored system of individual accounts; however, we reviewed available studies by the Pension and Welfare Benefits Administration and Olivia S. Mitchell of the Wharton School of the University of Pennsylvania on the costs of 401(k) pension plans, and we used the administrative costs they provided as approximations of the cost of a decentralized employer-sponsored system. Each of the costs included in table 2 covers to some extent the costs incurred for collecting contributions, managing records, investing money, and determining eligibility. The following list describes the basis of each study’s cost estimates and other information covered in its report: • The Report of the 1994-1996 Advisory Council on Social Security based its cost estimate for its decentralized individual-managed system on the costs of passively managed index funds and mutual funds. The cost estimate for its centralized system was a consensus of the Council members’ opinions. • The Employee Benefit Research Institute report discussed the options and difficulties in administering individual accounts and the lack of comparability between current retirement savings plans and a system of individual accounts. The authors based their discussion of administrative costs on the current costs of 401(k) plans and other investment companies and their discussions with other experts. In addition, the authors used the 30 For example, we did not use estimates from the Cato Institute because they were based on a substantially higher contribution rate. Also, Peter Diamond of the Department of Economics at the Massachusetts Institute of Technology estimated the cost of a centralized system of individual accounts in the form of dollars per worker per year rather than as a percentage of assets. While it is possible to convert from a dollar estimate to a percentage, it requires making other assumptions about account size and rate of return. In addition, he provided an estimate for a decentralized system of “at least 1% of assets,” which lacked the specificity of the other cost estimates we selected. 31 We also gathered a number of other estimates of administrative costs; however, the assumptions they were based on were unclear. Nevertheless, most of them fell within the range of costs we discuss in this report. Page 33 GAO/HEHS-99-131 Social Security Reform Appendix I Objectives, Scope, and Methodology SSASIM-2 model to simulate the effect of administrative costs on individual account benefits. • In the report by James and others, data on institutional funds—funds limited to institutional investors—were used to estimate administrative costs for a centralized system and data on retail mutual funds were used to estimate costs for a decentralized system. • The State Street Corporation’s report constructed a model of a market-based individual account system designed to ensure reasonable costs and minimize the administrative burden for employers. It then calculated an estimate for administrative costs, using unit costs from its current operations. As a result, its estimated costs depend greatly on the assumptions it made about the volume of calls for information and the number of transactions that the system would be required to handle. The authors structured their proposed system so that individuals would have limited investment choices and could change their investments only once per year; therefore, they assume that individuals will have little need to call, and that a large percentage of calls will be handled through voice and Internet technology. • The Mitchell report provides an analysis of current administrative costs for mutual funds, employer-sponsored defined contribution pension plans, and annuities provided by life insurance companies. • The Investment Company Institute reports analyzed trends in the cost of investing in equity mutual funds, bond funds, and money market mutual funds from 1980 to 1997. We used the 1997 costs. In order to answer the question of how administrative costs affect account Social Security accumulations and the retirement benefits they provide, we used a Social Simulation Model Security simulation model (SSASIM-2) that was recently developed by the Policy Simulation Group.32 The model has the capability of analyzing the implications of adding individual defined-contribution accounts to Social Security’s existing defined-benefit structure. Incorporated into the model is the dynamic interaction of the population, the economy, and Social Security programs. In our analysis, we made a number of assumptions. With respect to population and economic projections, including returns on investment and projected wages, we used the same assumptions as those used to produce the intermediate-range estimates of the 1999 Annual Trustees Report of the Social Security Administration. These resulted in a 10.3-percent 32 We consulted with Martin Holmer of the Policy Simulation Group in using the model to run our own simulations. Page 34 GAO/HEHS-99-131 Social Security Reform Appendix I Objectives, Scope, and Methodology nominal rate of return for corporate stocks (or about a 7-percent real return) and a 6.3-percent nominal return on Treasury bonds (or about a 3-percent real return). With respect to the structure of the individual accounts themselves, we assumed that the account was funded by a contribution of 2 percent of taxable earnings, and we isolated the effect of administrative costs on individual accounts from any changes to the Social Security program. We assumed that allocations of portfolios were based on a lifecycle model of investing and that the accounts would be created in 2002.33 Our simulations of account accumulations were done for workers born in 1984, who would enter the system at the age of 22. We simulated accumulations for two groups, those who would retire by the normal retirement age of 67 and those who would retire by the early retirement age of 62. For each group, we ran separate sets of simulations using estimates of administrative costs of 0.10, 0.25, 1, and 2 percent of assets per year. We picked these estimates because they cover the spectrum of most of the cost estimates we identified in table 2. For each set, we prepared six simulations segmented by gender and by low, average, and high earnings levels. We assumed that individuals with low earnings earned 45 percent of the average annual earnings each year throughout their careers and that those with high earnings earned 160 percent of the average, which is consistent with the assumptions used by the Advisory Council on Social Security and SSA’s Office of the Actuary. We chose to produce our results separately for men and women because their earnings patterns are significantly different. In addition, to simulate the effects of the cost of purchasing annuities, we assumed that individuals would pay a one-time fee of 5 percent of the balance of their account to purchase a gender-specific, indexed annuity. 33 A lifecycle model of investing assumes that younger individuals are able to assume more risk in their investment strategy than individuals who are much closer to retirement age. For example, individuals in their twenties would invest 100 percent of their portfolio in the stock market and their investments in stocks would decrease incrementally to 71 percent in their forties and 23 percent in their sixties. The remainder of their portfolio would be invested in Treasury bonds. Page 35 GAO/HEHS-99-131 Social Security Reform Appendix II Comments From the Social Security Administration Page 36 GAO/HEHS-99-131 Social Security Reform Appendix II Comments From the Social Security Administration Page 37 GAO/HEHS-99-131 Social Security Reform Appendix III GAO Contacts and Staff Acknowledgments Barbara D. Bovbjerg, Associate Director, (202) 512-7215 GAO Contacts Kay E. Brown, Assistant Director, (202) 512-3674 In addition to the persons named above, the following team members Staff made important contributions to our work and this report: R. Elizabeth Acknowledgments O’Toole, Alicia Puente Cackley, Abbey Frank, Gerry Grant, William McNaught, Deborah Moberly, Valerie Rogers, George Scott, Roger J. Thomas, and Rodina Tungol. Page 38 GAO/HEHS-99-131 Social Security Reform Glossary 401(k) Plan An employer-sponsored defined contribution plan that allows participants to contribute, before taxes, a portion of their salary to a qualified retirement account. Annuity A form of contract sold by life insurance companies that guarantees a fixed or variable payment made periodically (usually monthly) to the annuitant at some future time, usually retirement. Basis Point The smallest unit of measure for administrative costs. One basis point equals 0.01 or 1/100 of a percent; thus, 100 basis points equals 1 percent. Charge or Load A fee paid by an investor for buying and selling shares in a mutual fund or annuity. Defined Contribution Plan A pension plan in which the contributions are allocated to individual accounts by a predetermined formula and benefits vary, depending on the contribution level and the return from the investment of these contributions. Expense Ratio Expenses as a percentage of accumulated fund assets. Commonly used when referring to the administrative costs of mutual funds. Individual Retirement A personal, tax-deferred retirement account set up by an individual Account worker. Page 39 GAO/HEHS-99-131 Social Security Reform Bibliography Advisory Council on Social Security. Report of the 1994-1996 Advisory Council on Social Security, Vols. I and II. Washington, D.C.: U.S. Government Printing Office, Jan. 1997. Diamond, Peter. “Administrative Costs and Equilibrium Charges with Individual Accounts.” Paper presented at the National Bureau of Economic Research Conference, Cambridge, Mass., 1998. Revised Feb. 1999. Employee Benefit Research Institute. “Individual Social Security Accounts: Issues in Assessing Administrative Feasibility and Costs.” Special Report SR-34 and Issue Brief 203, prepared by Kelly A. Olsen and Dallas L. Salisbury, Washington, D.C., Nov. 1998. Genetski, Robert. “Administrative Costs and the Relative Efficiency of Public and Private Social Security Systems.” Social Security paper 15. Washington, D.C.: The Cato Institute, Mar. 9, 1999. James, Estelle; Gary Ferrier; James Smalhout; and Dimitri Vittas. “Mutual Funds and Institutional Instruments: What is the Most Efficient Way to Set Up Individual Accounts in a Social Security System?” Paper presented at the National Bureau of Economic Research Conference, Cambridge, Mass., 1998. Mitchell, Olivia S. “Administrative Costs in Public and Private Retirement Systems.” In Privatizing Social Security, ed. Martin Feldstein. Chicago, Ill.: University of Chicago Press, 1998. Pp. 403-56. Murthi, Mamta, J. Michael Orszag, and Peter R. Orszag. “The Charge Ratio on Individual Accounts: Lessons from the U.K. Experience.” Birkbeck College Working Paper 99-2, University of London, London, Eng., Mar. 1999. National Academy of Social Insurance. Report of the Panel on Privatization of Social Security. Washington, D.C.: Nov. 1998. Pension and Welfare Benefits Administration. Study of 401(k) Plan Fees and Expenses. Prepared by Economic Systems, Inc., and the Hay Group. Washington, D.C.: Department of Labor, Apr. 13, 1998. Page 40 GAO/HEHS-99-131 Social Security Reform Bibliography Poterba, James M., and Mark J. Warshawsky. “The Cost of Annuitizing Retirement Payouts from Individual Accounts.” National Bureau of Economic Research, Cambridge, Mass., Jan. 1999. Rea, John D., and Brian K. Reid. “Trends in the Ownership Cost of Equity Mutual Funds.” Investment Company Institute Perspective, Vol. 41, No. 3 (Nov. 1998), pp. 1-15. Rea, John D., and Brian K. Reid. “Total Shareholder Cost of Bond and Money Market Mutual Funds.” Investment Company Institute Perspective, Vol. 5, No. 3 (Mar. 1999), pp. 1-8. State Street Corporation. Administrative Challenges Confronting Social Security Reform. Boston, Mass.: 1999. Page 41 GAO/HEHS-99-131 Social Security Reform Page 42 GAO/HEHS-99-131 Social Security Reform Page 43 GAO/HEHS-99-131 Social Security Reform Related GAO Products Social Security Reform: Implementation Issues for Individual Accounts (GAO/HEHS-99-122, June 18, 1999). Social Security: Criteria for Evaluating Social Security Reform Proposals (GAO/T-HEHS-99-94, Mar. 25, 1999). Social Security: Individual Accounts as an Element of Long-Term Financing Reform (GAO/T-HEHS-99-86, Mar. 16, 1999). Social Security Reform: Experiences of the Alternative Plans in Texas (GAO/HEHS-99-31, Feb. 26, 1999). Social Security and Surpluses: GAO’s Perspective on the President’s Proposals (GAO/T-AIMD/HEHS-99-95, Feb. 23, 1999). Social Security and Minorities: Current Benefits and Implications of Reform (GAO/T-HEHS-99-60, Feb. 10, 1999). Social Security: What the President’s Proposal Does and Does Not Do (GAO/T-AIMD/HEHS-99-76, Feb. 9, 1999). Social Security Reform: Implications for Women (GAO/T-HEHS-99-52, Feb. 3, 1999). Social Security: Different Approaches for Addressing Program Solvency (GAO/HEHS-98-33, July 22, 1998). Social Security Reform: Raising Retirement Ages Improves Program Solvency but May Cause Hardship for Others (GAO/T-HEHS-98-207, July 15, 1998). Social Security Financing: Implications of Government Stock Investing for the Trust Fund, the Federal Budget, and the Economy (GAO/HEHS-98-74, Apr. 22, 1998). Social Security Financing: Implications of Stock Investing for the Trust Fund, the Federal Budget, and the Economy (GAO/T-AIMD/HEHS-98-152, Apr. 22, 1998). 401(k) Pension Plans: Loan Provisions Enhance Participation but May Affect Income Security for Some (GAO/HEHS-98-5, Oct. 1, 1997). (207048) Page 44 GAO/HEHS-99-131 Social Security Reform Ordering Information The first copy of each GAO report and testimony is free. Additional copies are $2 each. Orders should be sent to the following address, accompanied by a check or money order made out to the Superintendent of Documents, when necessary. VISA and MasterCard credit cards are accepted, also. Orders for 100 or more copies to be mailed to a single address are discounted 25 percent. Orders by mail: U.S. General Accounting Office P.O. Box 37050 Washington, DC 20013 or visit: Room 1100 700 4th St. NW (corner of 4th and G Sts. NW) U.S. General Accounting Office Washington, DC Orders may also be placed by calling (202) 512-6000 or by using fax number (202) 512-6061, or TDD (202) 512-2537. Each day, GAO issues a list of newly available reports and testimony. To receive facsimile copies of the daily list or any list from the past 30 days, please call (202) 512-6000 using a touchtone phone. A recorded menu will provide information on how to obtain these lists. For information on how to access GAO reports on the INTERNET, send an e-mail message with "info" in the body to: firstname.lastname@example.org or visit GAO’s World Wide Web Home Page at: http://www.gao.gov PRINTED ON RECYCLED PAPER United States Bulk Rate General Accounting Office Postage & Fees Paid Washington, D.C. 20548-0001 GAO Permit No. G100 Official Business Penalty for Private Use $300 Address Correction Requested
Social Security Reform: Administrative Costs for Individual Accounts Depend on System Design
Published by the Government Accountability Office on 1999-06-18.
Below is a raw (and likely hideous) rendition of the original report. (PDF)