|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
(released July 17, 2000) Contents In the last two months both major party presidential candidates have put forward proposals for a national system of individual accounts. In May, Governor George W. Bush, the presumptive Republican presidential nominee, proposed to establish a system of individual accounts that would be financed by a portion of the designated Social Security payroll tax. In June, Vice-President Al Gore, the presumptive Democratic nominee, put forward a system of individual accounts under which the federal government would provide savings subsidies to the vast majority of the nation's workers. There are important differences between these two proposals, the most significant of which is that the Bush plan would reduce the guaranteed benefit provided by Social Security. By contrast, the Gore proposal would be a supplemental account that workers would have in addition to their Social Security benefit. However, the two proposals also have an important feature in common: they both establish a decentralized system of individual accounts in which workers would invest themselves, subject to certain constraints imposed by the government. By establishing a decentralized system, these proposals would cost considerably more than either the existing Social Security system, or a system of centralized accounts. These additional costs will provide a substantial source of revenue to the financial industry, but will be a significant drain from workers retirement income. This study produces a set of projections for the administrative costs associated with these two plans and the extent to which these costs exceed the cost of alternative options. The projections show that: * The administrative costs of the Bush plan will be between $250-$360 billion over its first twenty years. The cost for the first forty years will be between $1.2-$1.8 trillion. * The administrative costs of the Gore plan will be between $390-$560 billion over the first twenty years. Over a forty year period the administrative costs will be between $1.7-$2.4 trillion. * The accounts proposed under the Bush plan are part of the mandatory tier of savings, where workers have no choice over the amount they contribute. Given their mandatory nature, the existing system can provide the same benefits as the proposed accounts, with no additional administrative expense. Therefore, the administrative costs of the Bush plan can be seen as a complete waste from the standpoint of workers. * The administrative costs of the Gore plan are attributable to an option tier of savings where workers have discretion over the amount they choose to save. While this necessitates some form of individual account, recent research shows that a centralized system, such as that proposed by President Clinton with his Universal Savings Accounts, would be far cheaper. The projections show that the unnecessary costs associated with the Gore proposal -- the difference between the cost of a decentralized system and a centralized system -- will be between $215-$380 billion over twenty years. Over a forty year period the unnecessary costs would be between $850-$1140 billion. This study builds on an analysis of Governor Bush's proposal for individual accounts, published earlier this month by the Century Foundation. That analysis examined the impact of the proposal on future retirement income. This study uses the same methodology as the previous one, but changes three assumptions. The most important change is that the present study uses projections of stock returns that are derived from the projections of profit growth in the Social Security trustees' report. The derivation of these projections is based on a methodology described in earlier work by Dean Baker and Peter Diamond. As a result of high current stock valuations and the low projected rate of profit growth in the trustees' report, the return on stocks in these projections is far below its historic rate, averaging less than 3.6 percent annually. This lower return has a very large impact on the projected income from individual accounts. The accumulation for a worker who has had 35 years to contribute to these accounts will be more than one third less than with the stock returns assumed in the Century Foundation study. With these lower accumulations, all workers will suffer substantial loses in retirement income as a result of the cuts in the guaranteed portion of the program, which are likely to be part of Governor Bush's proposal. It is important to note that the projections of stock returns used in this study are the only ones which are actually derived from the growth projections in the trustees report. The projections used in other analyses are extrapolations from past experience. They do not take into account the trustees projections of profit growth, nor current stock valuations. This paper also examines the effect of changing two other assumptions in the Century Foundation's study. That study has assumed that the individual accounts would cost just 0.4 percent of assets to administer each year. While this may be a plausible figure for a centralized system with little or no individual control over investment options, Governor Bush has indicated that he wants individuals, not the government, to be have control over these accounts. The costs of administering decentralized systems have been shown to be far higher. The study shows that if the costs are 1.0 percent annually, it will reduce the accumulations by approximately 8.0 percent. If the costs are 1.5 percent annually, as has been the case with the privatized British system, then the accumulations will be reduced by approximately 18 percent compared with the cost projection assumed in the Century Foundation study. Finally, the Century Foundation study assumed that these accounts could be converted to annuity payments at no cost. If annuitization was mandatory under a centralized system, this may be plausible. However, Governor Bush has indicated that he wants individuals to have the option to pass their accumulations onto their heirs. This implies that the individuals who choose to buy annuities will have to pay a considerable premium for them. The evidence suggests that this will reduce the annual income by between 10-20 percent compared to the no cost assumption in the Century Foundation study. When the full effect of these three changes are taken into account, the accumulations in the individual accounts are likely to be only about half as large as was estimated in the Century Foundation study. This would mean that virtually everyone will end up with a lower retirement income under Governor Bush's proposal, than under the current system.
In the last two months both major party presidential candidates have put forward proposals for a national system of individual accounts. In May, Governor George W. Bush, the presumptive Republican presidential nominee, proposed to establish a system of individual accounts that would be financed by a portion of the designated Social Security payroll tax. In June, Vice-President Al Gore, the presumptive Democratic nominee, put forward a system of individual accounts under which the federal government would provide savings subsidies to the vast majority of the nation's workers. There are serious differences between these two proposals, the most significant of which is that the Bush plan would reduce the guaranteed benefit provided by Social Security. By contrast, the Gore proposal would be a supplemental account that workers would have in addition to their Social Security benefit. However, the two proposals also have an important feature in common: they both establish a decentralized system of individual accounts in which workers would invest themselves, subject to certain constraints imposed by the government. By establishing a decentralized system, these proposals add significant administrative costs compared to either the existing Social Security system, or a system of centralized individual accounts. These additional costs can be seen as "unnecessary" in the sense that it is possible to accomplish substantially the same goals without incurring these costs. These unnecessary costs would directly reduce the retirement income of workers. While there may be reasons for preferring the decentralized system envisioned in these proposals, it is important to know how much additional expense workers will incur as a result of choosing this option. This paper constructs a set of projections of the administrative costs of both plans. The first section constructs projections for administrative costs of the Bush plan. The second section constructs projections for the administrative costs of the Gore proposal. The third section compares the unnecessary costs associated with both plans.
There are two types of administrative expenses incurred under the Bush plan. The first is the direct administrative expense associated with maintaining the individual account. This includes the cost to the financial institution to maintain the account, as well as its sales, advertising expenses, and profits. The experience of Chile, the United Kingdom and other nations with systems of private accounts suggests that these annual expenses will be between 1.0-1.5 percent of the accumulated assets in the accounts (Mitchell, 1998; Murthi, Orszag, and Orszag, 1999). The second type of cost associated with these accounts will be the cost of obtaining annuities. The current Social Security benefit is paid out as an inflation indexed annuity, with the worker guaranteed that the real value of his or her benefit will remain constant until death. Governor Bush has indicated that he wants individuals to have the option to not buy annuities with the money from their accounts, but instead be able to choose some fixed pay out formula, or pass the accumulation onto their heirs. This creates a problem of adverse selection, where the individuals who choose to buy annuities are likely to be more long-lived than the population average. This will raise the price of an annuity for anyone in the market, including those with life expectancies that are no greater than the average. While adverse selection does raise a serious problem from the standpoint of ensuring that workers are guaranteed a minimal retirement income, it is not directly relevant to the calculation of administrative costs. Recent research indicates that the fee charged for a typical annuity, after accounting for the effects of adverse selection, is approximately 5.0 percent of the worker's accumulation (Poterba and Warshawsky, 1999). This figure is used in the analysis below. Table 1 shows the cumulative administrative fees incurred associated with maintaining the system of accounts proposed by Governor Bush. The table shows the amount of fees under the assumption that administrative costs average either 1.0 percent or 1.5 percent of the assets in the account. The calculations for the accumulations in the accounts assume that the assets are held in a mix of 60 percent equities and 40 percent corporate bonds. This was the mix suggested by Professor Martin Feldstein, a principal advisor to Governor Bush on this topic. The returns on equities assumed in this calculation are derived from current price to earnings ratios and the projections for profit growth in the Social Security trustees’ report. This gives an average real return on equities of approximately 3.5 percent (Baker, 1999). If the return on stocks is closer to its historical average, then the administrative fees would be considerably higher. (The appendix presents a full explanation of these calculation. Administrative Fees
The numbers in table 1 do not include the fees resulting from the purchase of annuities. As noted earlier, Governor Bush has indicated that he would make the purchase of annuities optional, which means that a substantial portion of the population will opt not to purchase them. Since his proposal would establish a new system of accounts, which replaces a portion of the guaranteed Social Security benefit, there is no good mechanism for estimating what percentage of the population will opt to buy annuities. For simplicity, the calculations in this paper assume that half the population will decide to purchase an annuity with their accumulated assets.
Table 2 shows the projected fees
associated with purchasing annuities. The difference between the fees shown in the two
columns stems entirely from the fact that the accumulations will be somewhat larger in the
scenario with 1.0 percent annual administrative costs. Both columns assume that the cost
of the annuity will be 5.0 percent of the accumulation. As is the case with administrative
fees, the cost of annuities rise substantially through time. By assumption, there is no
cost for the first ten years, because the calculations assume that no workers with
individual accounts retire during this period. After twenty years, the cumulative cost
will be more than $18 billion in either scenario. Over the full forty year period, the
cost will be close to $160 billion in the scenario with 1.0 percent fees, and almost $145
billion in the scenario where annual fees are 1.5 percent. While both figures are
considerably less than the direct administrative fees associated with maintaining the
accounts, they are still quite substantial. Table 3
Table 3 combines the projections for both types of costs. After twenty years the combined fees would be more $250 billion in the low cost scenario, and over $350 billion in the high cost scenario. For the full forty year period the fees would be more than $1.2 trillion in the low cost scenario and more than $1.7 trillion in the high cost scenario. Tables 1 through 3 all present projections in current dollars, which means that they do not control for the effect of inflation. Most of the debate surrounding budget projections and Social Security has been carried through using current dollar figures. Therefore the current dollar figures shown in these tables are comparable to the figures that generally appear in public debate on these issues. However, the use of current dollar numbers is unfortunate, since virtually no one is capable of quickly accounting for the effect of inflation over a long period of time. In order to accurately assess the projected costs of these accounts it is necessary to use inflation-adjusted dollars. Table 4 presents projections of the total fees associated with the accounts proposed by Governor Bush, measured in today's (year 2000) dollars.
Table 4
The numbers in table 4 are considerably smaller than those in table 3, particularly for the more distant periods. Nonetheless they are still quite substantial. If the full costs for forty years are divided over 150 million workers (approximately the average projected size of the labor force over this period), the cumulative cost per worker would be nearly $3,400 in the low cost scenario in real, inflation-adjusted dollars. In the high cost scenario it would be more than $4,700 per worker. These are significant transfers from workers' saving to the financial industry. There is one last point worth making on the projected cost of the accounts proposed by Governor Bush. As noted earlier, these calculations use projections of stock returns that are derived from current stock valuations and the Social Security trustees’ projections for the growth of corporate profits. However, Governor Bush and his aides have assumed a much higher rate of return on stocks. While no economist has been able to show how these returns are possible (given current stock valuations and the trustees growth projections), it is still worth considering the cost of his accounts with a higher assumed rate of return on stocks.
Table 5
As can be seen, the costs of the accounts will be considerably larger if Governor Bush's assumptions on stock returns are correct. This is not surprising, since all the costs are assumed to be proportional to the money in the accounts. With higher stock returns, the accounts will accumulate more money, and therefore generate higher costs. After 20 years the cumulative costs will be nearly $300 billion in the low cost scenario, and over $420 billion in the high cost scenario. For the full forty-year period the costs will be more than $1446 billion in the low cost scenario, and just under $2 trillion in the high cost scenario.
In June, Vice-President Gore outlined a proposal to establish a national system of subsidized savings accounts which would serve as a supplement to the defined retirement benefit provided by Social Security. While there are few details available on the specifics of the proposal, Gore did describe the subsidy schedule, as well as the savings limits for the accounts. He also indicated that he intended for the accounts to be administered through a decentralized system, similar to the one that presently exists for individual retirement accounts. This information can provide a basis for projecting the administrative cost of these accounts. The subsidy schedule was designed to be very progressive. Workers who earned less than $15,000 a year would receive a three to one match for each dollar saved, up to $500. This means that a worker earning less than $15,000 a year who is able to save the maximum of $500 would receive a match of $1,500; for a total annual savings of $2,000. Couples could pool their earnings and savings, so that a couple that earned less than $30,000, per year could have up to $1000 matched at a three-to-one rate. Workers earning between $15,000 and $30,000 would have their savings matched one to one, with the total again capped at $2,000. Workers earning between $30,000 and $45,000 would receive a match of one to three: a worker that put aside $1,500 a year would receive a match of $500. Again, the total would be capped at $2,000 per worker per year. While the Gore campaign estimated that the budgetary cost of this plan would be $200 billion over ten years, it is likely that it would be considerably higher. The generous matches for low income workers will provide a substantial incentive for these workers to participate in the program. Although low income workers with families will find it difficult to put money aside regardless of how generous the subsidy is under this proposal, many low income workers are students or young people just out of school. These people would probably be able to borrow money from family members. Since the returns on the money placed in these accounts would be over 300 percent, it is likely that there would be a very high participation rate from this group of workers. At higher income levels most workers are already placing money in a retirement account of some form. There would be a strong incentive for these workers to redirect a portion of their current savings into the accounts proposed by Gore in order to take advantage of the government subsidy. Without more details on how the plan would be administered, it is difficult to have much basis for determining the number of people who would participate, but the generosity of the subsidies should attract a high percentage of those who are eligible. The calculations in this paper assume that half of those who are eligible participate in the plan. Given the current size of the labor force, this would be approximately 63 million workers. (Approximately 90 percent of the work force earns an income that places them below the cap and therefore enables them to qualify for a subsidy.) For simplicity, it is assumed that each worker contributes the full amount allowed under the law, in each year that he or she is eligible. In reality, most workers would probably participate to some extent in some years, and in other years, not at all, but for the purpose of calculating administrative costs, the distribution of the accounts across workers does not matter. Gore has indicated that the ceilings and subsidies would be adjusted in step with inflation. Therefore it is assumed that the size of the average contribution moves in step with the rate of inflation. The assumptions on rates of return, administrative fees, and the costs of annuities are the same as were used in the calculation of the cost of the Bush plan. (A full description of the cost estimates for the Gore plan is in the appendix.)
Table 6
Table 6 shows projections for the total cost of the Gore proposal including both annual fees and the cost of annuities. The numbers shown in table 6 are considerably larger than the projected cost of the Bush plan in table 3. After twenty years, the cumulative cost of the Gore accounts would be nearly $400 billion in the low cost scenario and almost $560 billion in the high cost scenario. After forty years the cost would be more than $1.7 trillion in the low cost scenario, and close to $2.4 trillion in the high cost scenario. The difference between the cost of the Gore and Bush plans is entirely attributable to fact that these calculations assume more money will be placed in the accounts proposed by Vice President Gore than the 2 percent of payroll that would be placed in the accounts proposed by Governor Bush.
The previous two sections
provided projections of the costs of the Bush and Gore plans. This section examines the
extent to which these costs can be seen as unnecessary in light of the purpose of the two
proposals. The sense in which the costs can be viewed as unnecessary of course depends on
what the actual goals of the proposals are. This discussion will therefore assume specific
goals for the proposals, and compare the cost estimates in the previous two sections with
alternative methods of meeting these goals. In the case of Governor Bush's proposal the calculation of unnecessary costs is a very simple exercise. The entire administrative cost can be viewed as unnecessary, since it is possible to provide the same retirement income to workers through the existing administrative structure without incurring the additional administrative expenses of individual accounts. The cost of the individual accounts is a pure addition to the expenses incurred under the current system, since it would be necessary to maintain the exact same administrative structure to run the defined benefit portion of the Social Security system, even after the two percentage points of the payroll tax has been diverted to individual accounts. It is important to note that the Bush proposal does not give workers any additional flexibility in the amount of money they choose to contribute to the system. Workers would still be obligated to pay 12.4 percent of their wages. The only difference to the worker is that the size of the guaranteed benefit is being reduced. The individual accounts will replace some portion of this benefit, with the exact amount depending on the individual worker's luck with his or her investments. While it could be argued that the individual accounts will allow for greater returns on average, this is incorrect for two reasons. First, given the projected growth path of profits and current stock valuations, there will be very little difference between the returns available from stocks compared with government bonds. While Governor Bush and his aides have assumed significantly higher returns from stock, they have been unable to demonstrate how this is possible. There is another reason why higher returns cannot be a rationale for individual accounts: it is possible to hold higher yielding assets, such as stocks or corporate bonds, in the trust fund itself, thereby saving the administrative costs of individual accounts. While there are political economy issues associated with government holdings of corporate stock, these are not obviously different from the issues that would arise from stock holdings in individual accounts, since Governor Bush has indicated that these accounts would be heavily regulated. In short, since the individual accounts proposed by Governor Bush offer no apparent gain over the guaranteed benefit provided by the current system, their entire cost can be viewed as unnecessary. In fact, the tables in the first section understate the true cost of these accounts, since they ignore the demands on workers' time required by these accounts. In contrast to Social Security, which in a typical year will not require any time from employees participating in the program, these accounts will require that workers exert some effort to oversee them. There has not been any research that has attempted to quantify the amount of time that such accounts would require, but even if it is as little as one hour per year, this would still be a substantial cost in aggregate. Under these circumstances, imputing an economy-wide average compensation figure of $15 per hour would add more than $2.1 billion annually to the administrative cost of these accounts. If workers spent an average of five hours a year overseeing their accounts, the imputed value of this time would be more than $10.5 billion annually. While additional research would be needed to determine how much of workers' time will be required to oversee these accounts, it is important to keep this factor in mind. The demands of raising children and maintaining a home have left many workers pressed for time. A government policy that makes additional demands on workers' time should be not be adopted without good cause.
The purpose of the Gore plan is qualitatively different than the Bush plan, in that it is intended to promote an optional second tier of savings to supplement the guaranteed benefit provided by Social Security. While the guaranteed benefit could simply be enlarged, the rationale for a voluntary second tier is that workers should be given an option as to how much of their current income they wish to put aside in order to increase their income after retirement. Allowing this option requires some type of individual account, and therefore some additional administrative cost. However, the administrative costs could be significantly reduced if the accounts were managed centrally, as President Clinton had proposed in his Universal Savings Account plan. By proposing a decentralized system, Vice-President Gore is unnecessarily increasing the administrative costs of his plan. The model for a centralized system of accounts is the Thrift Savings Plan for federal employees. This plan is currently being operated with administrative expenses of less than 0.1 percent of the accumulated assets (www.tsp.gov/features/tspsf.html). The Thrift Savings Plan benefits from the fact that it gets its contributions from a single employer. A national plan would likely incur higher expenses for this, and other, reasons. Professor Martin Feldstein, a leading expert on Social Security and a top advisor to Governor Bush, has assumed that a centralized system of individual accounts could be administered with an annual fee of 0.4 percent of assets (e.g. Feldstein and Liebman, 2000). This figure will be used in calculating the unnecessary costs associated with the Gore proposal. The cost of issuing annuities in a decentralized system is also far higher than in a centralized system. A recent study found that the Thrift Savings Plan offers annuities that pay close to 5.0 percent more than those available from other financial institutions (Poterba and Warshawsky, 1999). This implies that the annuities offered by the Thrift Savings Plan are close to actuarially fair, with only a very small fee being subtracted for administrative costs. In other words, virtually the entire cost of annuities under the Gore Plan can be seen as an unnecessary expense, since the cost would fall close to zero in a centralized system.
Tables 7A-7B
Table 7B (Bush Proposal)
Table 7A presents estimates of the portion of the administrative costs of the Gore proposal which can be viewed as unnecessary. Even in the low cost scenario the majority of the administrative costs fall into this category. After twenty years unnecessary administrative costs would exceed $200 billion in the low cost scenario and be over $380 billion in the high cost scenario. Over the full forty year period unnecessary administrative costs would be nearly $860 billion in the low cost scenario and more than $1.1 trillion in the high cost scenario.
This paper has constructed a set of projections for the administrative costs of the individual accounts proposed by both Governor Bush and Vice-President Gore. These projections use many simplifying assumptions, but should provide a reasonable first approximation of the cost of these two proposals. The projections show that the administrative costs for both plans will be quite large. In the case of the Bush plan the costs over the first twenty years will be between $257 and $359 billion. Over the full forty year period for which projections were made the cost would be between $1.2 and $1.8 trillion. The projections of the costs for the Gore plan were slightly higher, since the accounts were assumed to be larger on average. In the first twenty years the costs were projected as being between $393 and $559 billion. Over a forty-year period the costs were projected to be between $1.7 and $2.4 trillion. The paper then examined the extent to which these costs can be viewed as unnecessary. The entirety of the costs projected under the Bush plan fall into this category, since the existing program can provide the same benefits, risk free, with no additional cost. The Gore plan is designed to be a voluntary second tier, which necessitates some type of individual account. The expenses projected for his plan were compared with projected expenses for a centralized system such as those proposed by President Clinton in his Universal Savings Account plan. Over a twenty-year period the unnecessary expense of the Gore plan was projected to be between $215 and $381 billion. For the full forty-year period the unnecessary expense was projected as between $859 and $1134 billion. All of these costs are quite significant, and represent a large amount of wasted taxpayer’s dollars. However, these unnecessary costs have not yet entered into the debate over either plan. If policy-makers, journalists, and voters are to be fully informed about the merits and minuses of these proposals, these real costs must be taken into account.
Baker, D. 1999. "Letter to Martin Feldstein of May 15, 1999." (www.cepr.net/Social_Security/letter_to_feldstein2.htm) Mitchell, O. 1998. "Administrative Costs in Public and Private Pension Systems." in Martin Feldstein ed., Privatizing Social Security, Chicago: University of Chicago Press. Feldstein, M. and J. Liebman, 2000. "The Distributional Effects of an Investment-Based Social Security System." Cambridge, MA: National Bureau of Economic Research Working Paper 7492. Murthi, J., P. Orszag, and M. Orszag. 1999. "The Charge Ratio on Individual Accounts: Lessons from the U.K. Experience." Birbeck College Working Paper 99-2 (University of London). Poterba, J., and M. Warshawsky, 1999. "The Costs of Annuitizing Retirement Payouts From Individual Accounts." Cambridge, MA: National Bureau of Economic Research Working Paper 6918.
The calculations for the administrative expenses of the accounts in the Bush plan assume that 2 percent of covered payroll is contributed to the accounts each year. Since Governor Bush indicated that older workers will remain in the current system, it is assumed that 75 percent of covered wages are earned by workers that will contribute to the accounts in the first year. This percentage rises by 2.5 percentage points each year, so that after the tenth year all workers in the system are contributing to the accounts. The calculation assumes that no money is removed from the accounts until after the tenth year. The first age cohort pulls its money out of the system in the 11th year. As a simplifying assumption, all age cohorts currently in the system are assumed to be of equal size and workers of all ages are assumed to earn the same annual wage. In subsequent years the workforce is assumed to grow at the rate projected by the Social Security trustees, and all workers wages rise in step with the projected growth rate of the average wage. It is assumed that accumulations continue over 35 years, but after a point no further money is added to an account. This in effect assumes that workers will typically have 35 full years of work over their lifetime. It is assumed that half of the money pulled out of the system each year is used to purchase annuities. The return on the accounts is based on the assumption that they will be held 60 percent in stocks and 40 percent in corporate bonds. The returns for stock are taken from Baker (1999). The real returns on corporate bonds are assumed to be 4.0 percent annually. The administrative costs assumed in each projection are deducted from this return. The calculations for the Gore plan assume that 90 percent of the workforce is eligible to participate in the plan, and that half of those who are eligible actually do participate. The calculations assume that each participant places the maximum amount of $2,000 (combined worker contribution and subsidy) in his or her account each year. This ceiling is assumed to rise in step with the projected rate of inflation. The labor force is assumed to grow at the rate projected by the Social Security trustees. The withdrawal pattern is assumed to be the same as with the Bush plan, as is the mix of assets and assumed rates of return. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
