December 3, 1998
Administrative costs are an important part of the debate about the future of Social Security. Even though it is widely acknowledged that Americans could have a much more prosperous retirement if their Social Security taxes earned a better rate of return, some critics claim that the allegedly high administrative costs of individually owned and privately managed Social Security accounts would sharply reduce this gain. As this paper will show, such charges are not true.
Creating and administering a retirement program does not require rocket science, and while the task may be difficult, it is not impossible. To keep administrative costs low, a simple retirement program could start with individually owned accounts with limited investment options. Existing private investment plans--ranging from stock index mutual funds to multi-employer defined contribution retirement plans--could be adapted to a system of individual Social Security accounts. In some cases, administrative fees could be as low as 0.20 percent of fund assets; in all cases, they would continue to decline over time. History shows that administrative costs are highest when a system is first implemented and start-up costs must be covered. As a system matures, costs decline sharply. For instance, administrative costs at the federal Thrift Savings Plan (TSP) declined 76 percent over the TSP's first ten years of operation.
Administering a system of more than 140 million individual Social Security accounts involves little more than processing data. Similar financial data systems already exist. The nation's three largest privately owned credit bureaus, for instance, administer databases that average 190 million accounts, most of which are updated monthly. As computer technology increases, the difficulty of administering such immense databases will continue to diminish. Market competition keeps administrative costs low for the millions of Americans who currently own mutual funds and similar investments. There is every reason to believe that market competition would have the same effect on individual Social Security accounts.
In the upcoming debate on Social Security, Congress and the President should work to structure a simple system of individually owned, privately managed Social Security accounts. They should look beyond the pension industry for examples of technology and management techniques that can be adapted to this task. And they should not assume that any aspect of creating or managing these accounts must necessarily be handled by a federal agency.
A retirement program's costs are determined largely by its structure, and seemingly minor structural changes can affect administrative expenses significantly. A host of proposed programs that would establish individual Social Security accounts are likely to follow the pattern of development that 401(k) plans took, beginning as simple programs and adding features over time.
Although administering a retirement program may be complex, private businesses and the government have run such plans for decades. It is not necessary for Washington to invent either the technology or the investment options to give Americans an opportunity to save more money for retirement. Computers today are capable of tracking the investments of almost 190 million people. Risk-reducing investment programs, such as stock index mutual funds, that could be used to increase the retirement benefits available under Social Security also exist.
Of course, every retirement program, whether privately or governmentally managed, incurs administrative costs. These costs cover such items as the collection of money and information from employees or employers, the investment of that money, and the processing of retirement claims and benefits payments. Costs are directly related to the complexity of the plan, the level of service provided, and the number of available investment options (see page 3). It should go without saying that a simple system with very limited service--such as today's Social Security system--will cost much less to administer than one that is more complex and provides more information and services.
To make this study more meaningful, it is necessary to consider what Heritage expects private accounts to look like. The simplest, and most likely, Social Security accounts would be individually owned and privately managed, and have a limited number of investment options. Participants would be allowed to choose among a Standard & Poor's 500 Index mutual fund, a high-grade corporate bond fund, or a super-safe government bond fund that invests in the new Series I Savings Bonds. These bonds are designed specifically for retirement savings and pay an inflation-adjusted rate of return that is guaranteed for the 30-year life of the investment.2
Administrative costs can be kept
The Social Security system already spends $2 billion a year to administer retirement and survivors benefits. For the average household of two 30-year-old workers with children who each earned just under $26,000 in 1996, however, it will provide a return on their taxes of only about 1.23 percent after inflation.3 If this family had invested the same dollar amount in a portfolio of 50 percent equities and 50 percent government bonds, they could have earned a return of 5 percent or better. Although the administrative costs would be slightly higher in this program than under the current Social Security system, the increase in the couple's return would more than make up for the added administrative expense. In this example, the couple will have an additional return of 3.77 percent on their taxes, and even after paying administrative costs should be able to keep about 90 percent of the increase.
What Are Administrative Costs?
Administrative costs generally include the amount of money that an account owner must pay for maintenance and funds management. In most cases, these costs fall in three general categories: (1) the processing of information related to income and the amount invested; (2) actual funds management; and (3) determining eligibility for benefits and the payment of those benefits. The relative size of each component will vary according to how the plan is structured.
In most cases, administrative costs are measured as a percentage of assets under management. If a fund manages $100 million in assets and its annual expenses come to $500,000, its annual expense ratio would be about 0.50 percent. Percentage of assets is considered the best measure of administrative costs, since it reflects the way that investment management fees are usually assessed and spreads the costs over the entire investment. The one exception is a pay-as-you-go system such as today's Social Security, in which each year's benefits payments are funded by that year's taxes and only the surplus goes into the trust fund. The trust fund is merely a measure of excess taxes collected over time, and not a pool of investments upon which future benefits would be based. Benefits that will be paid to both present and future retirees are determined by a formula that is independent of any measure of the trust fund.
It is also possible to express administrative costs as a percentage of annual contributions to the plan, as a percentage of benefits paid, or as a dollar cost per participant. Comparing one cost measure from one plan against a different cost measure from another is likely to be meaningless. For instance, comparing Social Security's administrative cost of 0.64 percent of benefits paid in fiscal year 1997 with a Standard & Poor's stock index mutual fund's administrative cost of 0.38 percent of assets means as little as comparing the cost of a dozen apples to the cost of a pound of oranges. The comparison is useful only when the same standard of measurement is used.
Administrative costs usually do not include costs that the employer and/or employee incur in sending money or information to the manager of the retirement plan. For both the current Social Security system and private pensions, these costs must be measured separately.
A system of individually owned, privately managed Social Security accounts would be far larger and more complex than any existing retirement plan. The task of setting up and managing such a plan should not be underestimated. But it is not technologically impossible, nor would it have to be administered by the federal government.
This is really a question of data management. The Social Security program today administers about 140 million accounts. By contrast, the three largest private credit bureaus currently administer databases of more than 190 million accounts and, on average, update each active account monthly. These existing private systems are almost a third larger than Social Security, yet they post their account changes much faster. This type of data management could be adapted to a system of individual Social Security accounts.
Moreover, setting up a system of individually owned, privately managed Social Security accounts should become easier over time: The computer industry estimates that the capacity of a computer chip doubles roughly every 18 months.4 And programming ability has kept pace with this rapid growth in hardware capability.
Data processing could be contracted out to a private entity. In the United Kingdom, for example, the administration of the privatized segment of its social security system has been contracted out to the accounting firm of Arthur Andersen. Contracting out could be structured so that the processing contractor uses only the latest computer equipment. This stipulation would afford the manager of the plan the best technology available. The system would not be burdened and slowed by outdated hardware or software owned by the government or a previous contractor.
Transferring money to a funds manager is becoming more efficient and timely in the private sector. Years ago, private financial institutions developed an efficient, low-cost, and fast electronic funds transfer system (EFTS). Today, thousands of companies and financial institutions use EFTS to move billions of dollars daily. This method could be adapted for use with individually owned and privately managed Social Security accounts.
Thus, because of the rapid increases in computer capabilities today, establishing and administering individually owned Social Security accounts will become faster, easier, and less costly in the future.
History shows that administrative costs are highest when a system is first implemented and start-up costs must be covered. As time goes on, administrative costs decline significantly. This is true in the case of 401(k) accounts, the Thrift Savings Plan for federal employees, and even Social Security.
Over the years, for example, the administrative costs of 401(k) plans have decreased despite the growth in investment options and the level of personal service. Although the costs of specific plans vary according to each plan's complexity and size, as well as the type of assets in which the plan is invested, many large companies have been able to keep their annual costs as low as 0.3 percent by offering only a limited number of broad-based funds.5
The federal Thrift Savings Plan, which is a privatized retirement plan open only to federal employees, has seen an even more dramatic reduction in administrative costs. Since the system started in 1988, administrative costs have decreased by 76 percent.
Social Security showed similar reductions during its formative years. In 1940, when the system first began to pay benefits, its administrative costs equaled 74 percent of all Old-Age and Survivors' Insurance (OASI) benefits paid. In 1945, this figure had declined to 9.8 percent.6 Today, administrative costs make up only 0.64 percent of payments from the OASI trust fund. Even though Social Security's structure has changed over the years so that this is not a perfect comparison, it does give analysts an idea of the possible size of the reduction.
Few people dispute that individual Social Security accounts can work in theory. The question concerns how much they will cost to administer in practice. There is powerful evidence from real-world examples that the increased administrative costs of a private system are small in comparison with the increased income generated by the accounts.
The sources for the information on probable administrative costs of a system of worker-controlled accounts include (1) existing private retirement programs, such as employer-provided pension plans and 401(k) plans; (2) management fees on mutual funds and other private investments; and (3) privatized social security systems in other countries.
Plans offering individually owned and privately managed Social Security accounts would resemble a number of private retirement and investment plans currently available in the United States. Although the actual costs of such accounts would be determined by how a program of individual Social Security accounts was implemented, their administrative costs would approximate those of a fully privatized system.
Defined Contribution Plans. Individual Social Security accounts are likely to be structured like the existing defined contribution retirement plans currently offered by various U.S. employers. Defined contribution plans allow workers to invest up to a certain level of income (which may be matched, either partially or wholly, by the employer) in a specified range of investments. A 1996 U.S. Department of Labor report suggests that the expenses associated with administering defined contribution plans are very low. In 1992, total annual costs for these plans amounted to $34.99 per participant, or just over 0.17 percent of total assets held.7
The expenses associated with a private account can be divided into categories. Investment advisory and management fees generally are set as a fixed percentage of a fund's balance. Other fees, such as record-keeping and legal fees, usually are a fixed amount regardless of the balance in an account. Using these as assumptions and the data from the report, the expenses associated with each employer-provided defined contribution account in 1992 amounted to $24.99 per account, plus an investment fee of 0.05 percent of the account balance. In 1992, this investment fee equaled $10 per account, bringing the total annual cost to the $34.99 cited above.
401(k) Plans. Another fast-growing private component of the U.S. retirement savings system is made up of employer-sponsored 401(k) plans, which range in size and complexity. Initially, most 401(k) plans offered members few investment choices and a low level of service. Over time, however, some added dozens of ways to invest the contributions, and several plans even allow members to rearrange their investments on a daily basis.
Currently, over $1 trillion is invested in 401(k) plans. According to University of Pennsylvania Professor Olivia Mitchell, who recently completed an extensive analysis of the likely costs of a private Social Security system, "On an annual basis, the average per participant cost of administering a 401(k) plan appears to be between $5.00 and $55.00 annually, including non-discrimination testing, quarterly statements, and investor information."8
Today's Social Security Is Not Free
Americans pay a high price for their Social Security retirement benefits, especially when Social Security's rate of return is compared with that of private retirement plans. For example, consider the experience of an average household of two 30-year-old workers with children who earned just under $26,000 each in 1996 and who invest in a retirement plan of 50 percent equities and 50 percent government bonds. After inflation, their savings would earn them a return of at least 5 percent annually. Yet under Social Security, they will see a return of only 1.23 percent on their tax dollars.1
During fiscal year (FY) 1997, the Social Security Administration spent almost $2 billion to administer its Old-Age and Survivors' Insurance (OASI) trust fund,2 which covers only the cost of the retirement and survivors programs. Other benefit plans such as Disability Insurance (DI) are administered by separate trust funds. The billions of dollars Social Security spent on administrative costs in 1997 represented 0.52 percent of the trust fund's income, or 0.64 percent of the benefits paid during FY 1997. But, because Social Security is an unfunded pay-as-you-go program, measuring these costs as a percentage of assets, from an accounting point of view, is meaningless.
As with any retirement program, Social Security's administrative costs measured in this way declined over time. In 1940, when the system first began to pay benefits, its administrative costs equaled 74 percent of all OASI benefits paid. In 1945, this figure had declined to 9.8 percent.3 Today, administrative costs make up only 0.64 percent of payments from the OASI trust fund in FY 1997.4 Social Security's cost structure reflects the nature of the program. On average, the determination of benefit eligibility and payment of monthly benefits account for 93 percent of administrative costs.5 About 7 percent is spent to collect Social Security taxes and only about 0.01 percent on funds management. Unfortunately, these priorities have resulted in extremely uneven service performance. Furthermore, the current system gives individuals no ability to structure their retirement program to meet their own circumstances. Each retiree simply takes whatever the Social Security program chooses to give.
Social Security takes an average of 17 days to begin benefits payments after receiving an application, yet it regularly takes between 7 and 22 months to post earnings information to an individual's account.6 Furthermore, the U.S. General Accounting Office (GAO) has criticized Social Security for issuing Personal Earnings Benefits Estimate Statements (PEBES) that are confusing and contain inaccurate information.7
The experience of other countries is also instructive. The closest comparisons to the U.S. system are those of Australia and Britain, two developed nations with sophisticated financial markets that have partially privatized their social security systems.
Declining Costs. These private retirement programs cover a variety of products available in the consumer market. A system of private Social Security accounts most likely would be structured so that administrative costs would be even lower than those assessed by private retirement accounts today. This is because a privatized Social Security system operating on a national scale would have many more participants than the private alternatives have. And as the number of accounts increases, the cost-per-participant would tend to fall as "fixed" costs, such as the cost of computer hardware, are spread over a greater number of participants.
Private retirement programs tend to offer a large degree of investment choices, including relatively expensive (and riskier) products such as international equity funds. In a system of individually owned and privately managed Social Security accounts, at least initially, federal prudential regulations likely would confine retirement investments to inexpensive and less risky assets, such as U.S. Treasury bonds and passively managed mutual funds that track broad market indexes like the Standard & Poor's 500 Index.
In this context, perhaps the best indicator of the costs of a system of individual Social Security accounts is the Thrift Savings Plan operated by the federal government for its workers. At the end of 1995, the TSP provided its 2.2 million participants with a comprehensive retirement package that included optional annuities and allowed employees a limited set of investment alternatives. Participants divided their investments among three plans, including a stock index fund, a corporate bond fund, and a government debt fund. In 1995, investment and administrative costs for the TSP totaled 0.09 percent of net assets, or $15.20 for each participant.9
Mutual Funds. Mutual funds are an extremely popular way to save for retirement. The Investment Company Institute estimates that about 35.5 percent of all the assets in retirement plans are invested in mutual funds.10 For individual retirement accounts (IRAs), which in most cases are self-owned and self-directed instead of affiliated with an employer, the figure is higher, with 42 percent of assets invested in mutual funds.11
One of the most suitable forms of investment for retirement is an index fund that passively tracks a broad market index such as the S&P 500, the Dow Jones Industrial Average, or the Russell 2000. Mutual funds following this investment strategy usually have very low administrative and investment costs.
According to Lipper Analytical Services, the principal provider of fund fees and expense data to the mutual fund industry, the median administrative cost of funds that follow the S&P 500 was just 0.38 percent.12 One such fund offered by Vanguard has a cost that is only 0.19 percent of assets.13 Although the median expense ratio for all mutual funds invested in equities is 1.38 percent, this amount includes a number of funds that would probably be unsuitable for individual Social Security accounts, such as those with extremely risky investment strategies.
U.S. Treasury Bonds. U.S. Treasury bonds constitute an extremely inexpensive and risk-free retirement savings option. According to Lipper Analytical Services, the median mutual fund invested in U.S. Treasury bonds has an expense ratio of 0.83 percent.14 But it may not even be necessary for an individual to use a mutual fund to hold bonds.
An extremely low-cost retirement investment with no management fees and no risk would be the U.S. Treasury Department's new Series I Savings Bonds. Designed for retirement savings, these bonds pay an inflation-adjusted rate of return that is guaranteed for their 30-year life. They can be purchased for no cost from almost any local bank branch in denominations as low as $50. Savings bonds issued through April 1999 will pay a guaranteed real rate of 3.3 percent for the next 30 years.15
According to government statistics, annual administrative costs for Australia's system of private accounts totaled 0.85 percent of fund assets in the first quarter of 1998--the equivalent of an annual average cost per participant of $70.20 in Australian dollars (or US $44.71). A number of funds in Australia offer participants total fixed annual costs of $52 (or US $33.12 per year).16
Britain's system illustrates the importance of requiring a simple, easy to understand disclosure of fee schedules. Although several major banks advertise that their annual fees will be 1.0 percent of assets, fees for the popular Group Personal Pensions (GPP) tend to be confusing. An October 1998 survey of these plans17 showed that, in addition to annual management charges, all plans also charged commissions and various annual policy fees. While the total cost may be clear to financial professionals, it can be highly confusing to the average investor. However, GPP plans usually are offered through the employer, which often pays the fees, and appear to be negotiable for larger employers.18
Although individual Social Security accounts may cost slightly more to administer, the benefits they generate will far outweigh these costs. This is particularly important to remember in light of the low rate of return offered by the current Social Security system. An administratively "cheap" system may offer a lower level of retirement security than a system that offers higher rates of return, but it also costs more to administer. Consider the following example.
The cost of administering this account is $50 (which is the worst-case scenario--an annual fee of $50 lies at the upper range of the data on average costs presented earlier in this paper);
Costs are allocated on a flat basis per account so that workers with $1 million in their accounts pay the same fixed charge as workers with $50 in their accounts (in practice, costs should vary with the size of the account, so costs of the relatively small account would be much less than $50); and
The worker allocates 50 percent of the account to U.S. Treasury bonds yielding 2.8 percent per year after inflation, and 50 percent to a broad market equity index fund yielding 7 percent per annum after inflation. Earnings are assumed to grow at one percent after inflation.19
Although critics of individual Social Security accounts usually focus on the impact of administrative costs on the first year of earnings, this is a long-term investment, and fees should be considered the same way. In this case, the $50 administrative cost absorbs just over 6.5 percent of the account's balance during its first year. However, administrative costs fall rapidly and amount to less than 1.0 percent of the account balance within six years. Thus, the claim that the higher returns of private accounts would be wiped out by their administrative costs is demonstrably false, especially if one takes the medium- or long-term perspective that is appropriate for retirement planning.
The Social Security Administration's own projections imply that the mixed bond/equity portfolio described above would yield an after-inflation return of 4.9 percent per annum before administrative costs are included.20 By contrast, according to the Social Security Administration's own calculations, a male child born in 1997 who earns a low income can expect to receive only 1.95 percent from Social Security after inflation.
Chart 1 compares the performance of the high yield/"high" administrative cost private account with the performance of a plan that invests in an asset with the same yield as Social Security. The Social Security Administration has already considered administrative costs in calculating this yield. In the first scenario, the worker saves 3 percent of income in a private account and earns a return of 4.9 percent per annum. In the second scenario, the worker makes a similar investment decision but pays an annual administrative fee of $50. In the third scenario, his money is retained in a Social Security-type investment and yields a post-inflation return of 1.95 percent per annum.22
Over the long term, the power of compound interest and the higher return from private investments entirely negates the initial impact of higher administrative costs within six years. At the end of the 40-year period, administrative costs have reduced the difference in returns between the low cost/low return Social Security-type asset and the private account by only 11 percent.
Are Small Accounts a Big Problem?
Critics of individually owned accounts often point to an apparently large number of low-wage workers. Particular attention has been given to the claim that in 1998, 30 percent of all wage and salary earners earned $10,000 or less. It has been argued that the small amounts of money flowing into the private accounts of these earners are likely to be taken entirely to pay the accounts' higher administrative costs. However, a closer look at the evidence suggests that the problems implied by small accounts may not be as large or intractable as defenders of the status quo claim.
According to U.S. Census Bureau data cited by the Employee Benefit Research Institute (EBRI), 53 percent of these low earners consist of workers either at the very end of their careers (older than age 60) or at the very beginning (aged 25 and under).1 A system of private accounts could be structured to avoid the compulsory inclusion of those over mandatory retirement age or of very young workers who are in the educational system and work only on a part-time basis.
The overemphasis on small accounts is the result of the failure to consider income mobility. Many workers' incomes fluctuate widely from year to year. This fluctuation will not be captured in the annual data. Indeed, there is strong evidence that a large proportion of "low-income" workers in any year are experiencing atypical years, and their earnings in that year do not reflect their long-term income.
A 1996 Syracuse University study examined the income mobility experience during the 1980s of workers aged 25 to 55 (this group excludes most college students whose incomes rise upon graduation and older workers who are nearing retirement).2 Within one year, 19 percent of all workers in the bottom 20 percent of the population had moved to a higher level. Within five years, 36 percent of workers in the poorest 20 percent of the population moved into the upper 80 percent. These data show that better than one-third of those who, in their prime working ages, are recorded as earning $10,000 or less are experiencing an atypical and temporary downward fluctuation in earnings.
Not only is the scale of the problem of small private accounts overstated, but the solution to this problem could be generated by sound policy decisions. Pricing could be designed so that fund managers are required to charge expenses on the basis of a proportion of fund balances, rather than as flat fees that are proportionately more burdensome on small accounts.
Administrative costs are not a barrier to creating a system of individually owned and privately managed Social Security accounts. It is not that difficult to structure the new system to keep accounts simple and costs low. Existing technology allows for the creation of a method to track these investments, and the experience of the United Kingdom shows that the administration of such accounts could be contracted out successfully.
Because administrative costs are determined largely by the structure of the accounts, investment options for these worker-controlled accounts could be limited initially to a stock index mutual fund, a high-grade corporate bond fund, and a government bond fund that invests in the new Series I Savings Bonds. As the system matures and costs drop, additional investment options and consumer features could be added.
All Americans could be earning a higher rate of return on Social Security taxes than is possible today if Washington allowed them to invest in individually owned and privately managed accounts. The low administrative costs of these accounts over time, along with their returns, would provide most Americans with a more secure and prosperous retirement.
--David C. John is a Senior Policy Analyst for Social Security at The Heritage Foundation.
2. For more information about Series I bonds, see the U.S. Treasury's Web site at http://www.publicdebt.treas.gov/sav/sbiinvst.htm.
7. Olivia S. Mitchell, "Administrative Costs in Public and Private Retirement Systems," in Martin Feldstein, ed., Privatizing Social Security (Chicago, Ill.: University of Chicago Press, 1998), p. 433.
15. U.S. Department of the Treasury, Bureau of the Public Debt, "U.S. Savings Bond Home Page," at www.publicdebt.treas.gov/sav/sav.html.
16. Australian Government Publishing Service, Insurance and Superannuation Commissions Statistical Bulletin, various issues. U.S. dollar valuations based on exchange rate of 1.57 (quoted on http://www.cnnfn.com on November 25, 1998). See also Daniel J. Mitchell and Robert P. O'Quinn, "Australia's Privatized Retirement System: Lessons for the United States," Heritage Foundation Backgrounder No. 1149, December 8, 1997.
22. A direct comparison with Social Security is not possible, as individual balances do not compound over time within the Social Security system. This example is constructed only for the purpose of illustrating the point that a low-return/low administrative costs is often inferior to an investment that is more expensive to administer but that yields a higher return. However, 1.95 percent is the rate of return the Social Security Administration estimates that a low-income male born in 1997 will receive from Social Security.