#2005-20 December 2005 Voluntary Carve Outs for Social Security Reform: Lessons from the United Kingdom by David Blake Cass Business School, London John Turner AARP Public Policy Institute The AARP Public Policy Institute, formed in 1985, is part of the Policy and Strategy Group at AARP. One of the missions of the Institute is to foster research and analysis on public policy issues of importance to mid-life and older Americans. This publication represents part of that effort. The views expressed herein, which are those of the author, are for information, debate, and discussion, and do not necessarily represent official policies of AARP. © 2005, AARP. Reprinting with permission only. AARP, 601 E Street, NW, Washington, DC 20049 http://www.aarp.org/ppi Foreword In his 2005 State of the Union message, President George W. Bush made Social Security reform the top item on his second-term domestic policy agenda. He proposed that workers be allowed to voluntarily divert part of their payroll taxes into individual accounts. Among high- income countries, only the United Kingdom permits this arrangement. Thus, the United Kingdom’s experience is particularly relevant to the U.S. Social Security policy debate concerning carve-out individual accounts. Voluntary Carve Outs for Social Security Reform: Lessons from the United Kingdom by David Blake of the Cass Business School in London and John Turner of the AARP Public Policy Institute describes the functioning of voluntary carve-out individual accounts in the United Kingdom. Voluntary carve-out accounts enable workers to voluntarily choose to have part of their social security payroll taxes diverted into an individual account. It is an irony of timing that President Bush proposed these accounts for the United States while their popularity was declining in the United Kingdom. In 2004, 500,000 people in the United Kingdom abandoned these accounts and returned to full participation in the British social security program. Though choice is considered to be one of the positive aspects of voluntary carve-out accounts by those who support them, the option of returning to full participation in U.S. Social Security for workers wishing to do so has not been an aspect of proposals in the United States. Blake and Turner pay particular attention to the calculation of the trade-off between the contribution to the individual account and the offsetting reduction in social security benefits. The determination of the trade-off is a complex, technical issue that is at the heart of the feasibility of this type of account. The trade-off is reset every five years in the United Kingdom to adjust to economic and demographic changes. Many people felt that the government made an error in determining the trade-off when it was most recently set, which explains in part the large number of people returning to full participation in social security. Voluntary Carve Outs for Social Security Reform: Lessons from the United Kingdom is one of five new AARP Public Policy Institute papers that examine aspects of social security reform in a number of countries. The other papers deal with the investment of social security reserves in the financial market in Canada, recent developments with partial privatization and individual accounts in Chile, Australia’s system of means-tested benefits and mandatory occupational savings schemes, and Sweden’s move to defined contribution pensions. John Gist, Ph.D. Associate Director AARP Public Policy Institute THIS PAGE INTENTIONALLY LEFT BLANK ii TABLE OF CONTENTS Foreword…………………………………………………………………………………………..i Executive Summary……………………………………………………………………………….v Introduction………………………………………………………………………………………..1 Contracting Out with Individual Accounts in the United Kingdom………………………………2 The Terms of the Offset between the VCO and the Traditional Social Security Benefits………..5 The Effects of Voluntary Carve out Individual Accounts………..…………………………….....8 Conclusions……………………………………………………………………………………....13 References………………………………………………………………………………………..14 iii THIS PAGE INTENTIONALLY LEFT BLANK iv EXECUTIVE SUMMARY Introduction Individual accounts have received considerable attention as part of U.S. Social Security reform. The private accounts can be mandatory or voluntary, and “add ons” to or “carve outs” from Social Security. Although policy analysts have extensively studied individual account plans, of the four basic types determined by the combinations of these characteristics, they have paid least attention to voluntary carve out (VCO) accounts. Purpose The purpose of this paper is to analyze VCOs with special emphasis on the proposals of the President’s Commission to Strengthen Social Security and in President Bush’s 2005 State of the Union message. Lessons are drawn from the experience with VCOs in the United Kingdom, which allows partial privatization with VCOs using individual account plans, similar to the proposals of the President’s Commission. Although the proposals of the President’s Commission left many details unspecified, this paper focuses on two central aspects of the proposed accounts—the individual accounts are voluntary and they are carve outs from the Social Security system. In considering Social Security reform proposals, especially ones that create substantially new structures such as VCOs, it is important to apply what we already know about the functioning of pension systems and their effects on workers to assess the strengths and weaknesses of the reform, rather than analyzing an idealized form of the proposed system. Methodology The methodology of this paper is to investigate the functioning of VCOs like those proposed by the President’s Commission and President Bush using both economic analysis and the experience of the United Kingdom with its system of VCOs. The United Kingdom has nearly twenty years of experience with this type of plan, and has made some adjustments in regulations as it has learned from its experience. Main Findings Some of the main findings from this analysis include the following: VCO accounts destabilize Social Security financing. By withdrawing money that would otherwise be paid into Social Security, these accounts would destabilize U.S. Social Security financing for decades (the transition cost). Many people are abandoning VCO accounts in the United Kingdom and returning to the traditional Social Security system. In 2004, two large insurance companies, Prudential and Norwich Union, sent letters to hundreds of thousands of policyholders with VCO accounts telling them they would be better off to not participate in VCO accounts and that v they should return to the traditional Social Security program because the amount the government was providing to participants in VCOs was insufficient to make them better off than had they stayed fully in the social security program. In 2004, 500,000 people abandoned VCO pensions and returned to the Social Security system. If VCO accounts are offered to U.S. workers on a basis that is neutral with respect to the Social Security benefits they have foregone, then many individuals who take them would be made worse off because of the fluctuations in asset markets. If the government subsidizes VCO accounts, the expense to the government would increase. VCO accounts would probably not be a good arrangement for low-income workers who depend on U.S. Social Security benefits. Social Security benefits represent a much higher percentage of income for most low-income workers than for high-income workers. Thus, the reduction in Social Security benefits for low-income workers when they take VCOs would be considerably larger as a percentage of their total retirement income than for high-income workers. The change in the risk that low-income workers would bear by switching to a VCO would be considerably larger than for high-income workers. To the extent that lower-income workers are more risk averse than higher-income workers, they would likely invest in low- risk assets and have a lower expected return. VCO accounts are a particularly expensive way of providing individual accounts. Because of the individualization of decision making, VCO accounts are a particularly expensive way of providing individual accounts. One U.K. study found that the associated administrative expenses have reduced benefits by 40 to 45 percent. The administrative costs of providing annuitized benefits would be relatively high. One aspect of the high administrative costs for VCOs is the cost of annuitizing them. The cost of converting each individual account to an annuity, with private insurance companies providing the annuities, would be relatively high because of the high cost of individualized transactions, especially with many of the accounts being relatively small. One U.K. study found that the cost of annuitizing the accounts reduced their value by 10 percent on average. VCO accounts differ from 401(k) accounts in that they are like a loan from the government. VCO accounts are like a loan from the government because the worker must accept reduced Social Security benefits in exchange for the diversion of Social Security payroll taxes into the individual account. VCO accounts would be difficult to administer. With VCO accounts, it is difficult to determine the appropriate relationship between the reduction in the worker’s payment to Social Security and the reduction in that person’s Social Security benefits. Every five years, the U.K. government resets that relationship. Insurance companies in the United Kingdom are now arguing that the government made an error in setting the current rate and that workers therefore should not take the VCO. vi Conclusion VCO accounts would likely not replicate key features of U.S. Social Security or 401(k) plans. They would not provide quick crediting of contributions as do 401(k) plans, but might require an 18-month delay in crediting contributions from the start of the calendar year in which the contributions were received. With VCOs, a key difficulty is determining the offset rate between the individual account and the social security benefits. Many people are abandoning VCO accounts in the United Kingdom and returning to the traditional social security system because of a change in the offset rate between the individual account and the social security benefits. VCO accounts would destabilize social security financing by reducing the amount of contributions it received while having no effect on benefit payments in the near term. VCO accounts are a particularly expensive way of providing individual accounts because of the extra record keeping required to determine the effect of those accounts on the traditional social security benefit and because of the voluntary nature of the accounts. vii THIS PAGE INTENTIONALLY LEFT BLANK viii 1. Introduction Individual account plans1 have received considerable attention as part of Social Security reform proposals (World Bank 1994; Gillion et al. 2000; Feldstein 1998). They can be mandatory or voluntary, and “add ons” to or “carve outs”2 from Social Security. Although policy analysts have written extensively about individual account plans, of the four basic types determined by the combinations of these characteristics, the least attention has been paid to voluntary carve out (VCO) accounts. In recent years, however, there has been increased interest in them (see Blake 2003; Turner and Rajnes 1995; Gustman and Steinmeier 1998; Kotlikoff, Smetters and Walliser 1998; Disney, Palacios and Whitehouse 1999; Orszag and Greenstein 2001; Gustman and Steinmeier 2003; Williamson 2000; National Academy of Social Insurance 2005). VCOs can be the most complex type of individual account. With a VCO, the worker has a choice. The worker can remain in the traditional Social Security system or withdraw from it--either partially or fully, depending on the structure of the VCO. VCOs in the United Kingdom can involve either a defined benefit3 or a defined contribution plan4 as the alternative to Social Security. We focus on defined contribution plans because they have been the subject of recent proposals in the United States. The United Kingdom is the only high-income country that allows VCOs with the worker establishing an individual account defined contribution plan. A worker choosing a VCO pays the same amount in payroll taxes5 as a worker not taking a VCO, but the disposition of the funds differs. With a VCO, part of the worker’s payroll taxes (the contracting-out rebate) is diverted to an individual account from which he or she would receive future benefits, and future Social Security benefits are in turn reduced. The Commission to Strengthen Social Security that President George W. Bush appointed in his first term (President’s Commission 2001) was required to make VCO accounts a key aspect of Social Security reform. The final report of the President’s Commission contains three proposals, all of them involving VCO accounts. President Bush indicated in his 2005 State of the Union message that he intends to propose Social Security reform involving this type of individual account. 1 These are accounts based on the defined contribution principle. 2 Carve outs are financed with Social Security taxes and substitute for part of Social Security benefits. 3 They can be defined benefit if they are arranged via the worker’s company. This option was historically important, but most private sector companies have terminated their defined benefit plans. 4 If these defined contribution plans are arranged by the worker’s company, they are known as a contracted-out money purchase plan. If they are arranged by the worker directly, they are known as a contracted-out or appropriate personal pension plan or a contracted-out stakeholder pension plan. Contracted-out stakeholder pension plans are low cost personal plans in 2001 having annual charges capped at 1.5 per cent for the first 10 years and 1 percent annually thereafter, and imposing no entry or exit charges. For further details, see Blake (2003). 5 Payroll taxes in the United Kingdom are called National Insurance Contributions. This paper discusses VCOs with special emphasis on the proposals of the President’s Commission and the 2005 State of the Union message. Although the commission report, the State of the Union address, and associated background information left many details unspecified, this paper focuses on two central aspects of the accounts proposed—they are voluntary, and they are carve outs from Social Security. VCOs can be structured in different ways (National Academy of Social Insurance, Chapter 9, 2005). This paper assumes, consistent with President Bush’s proposal, that VCOs are available as an option to all workers, financed by diverted payroll tax payments, and offset by reduced future Social Security benefits. These VCOs require that • The full amount of the reduction in payroll taxes is transferred to an individual account. • Workers have some choice as to how their individual accounts are invested. • The fees for managing the accounts are charged as a flat percentage of account balances with no front loading. • Pre-retirement access to the funds is not allowed. • At least part of the account must be annuitized based only on age (and not gender), and must be available at the same age as Social Security benefits. • For married couples, all contributions and earnings on those contributions during marriage and all earnings on account balances brought into marriage must be divided equally in the event of divorce. • Workers will be given a one-time choice as to whether to participate in the VCO and will not be allowed to reverse that choice. In considering Social Security reform, especially a reform approach that creates substantially new structures such as VCOs, it is important to apply what we already know about the functioning of pension systems and their effects on workers to assess the strengths and weaknesses of the proposed reform, rather than analyzing an idealized form of the proposed system. This paper draws on the United Kingdom’s experience with VCOs. A major hurdle in instituting a VCO system is that it worsens the financial status of Social Security during a long transition period by taking contributions from the current Social Security system and redirecting them to the VCO. If the VCO is also subsidized by the Social Security system, the problem is even worse. This paper focuses on the structure of VCOs. It does not analyze issues related to solving the long run financing shortfall for the U.S. Social Security system. 2. Contracting Out with Individual Accounts in the United Kingdom In the United Kingdom, workers can “contract out” of Social Security, replacing part of their Social Security benefits with private pensions, either through a defined benefit plan or, since 1988, an individual account defined contribution plan. The genesis of voluntary carve outs in the U.K. was motivated by completely different reasons from those motivating President Bush’s proposal. The U.K. was quite late in establishing an earnings-related Social Security program, which was not done until the 1970s. At that time, a well-established private pension 2 sector was already in place. Voluntary carve outs were permitted in the U.K. not to reduce a pre- existing Social Security program, but to protect a pre-existing defined benefit private pension system. Later, for ideological reasons, workers were allowed to establish private accounts to reduce their participation in Social Security. Since the early 1990s, there has been little growth in the number of people contracted out, which has resulted in a decline in the percent of the workforce contracted out from 69 percent to 61 percent. Contracting out has been much more popular through defined benefit plans than defined contribution plans. Although in the private sector nearly everyone with a defined benefit plan is contracted out, only one-third of those with defined contribution plans are contracted out (Pensions Policy Institute 2004). Because contracting out through individual defined contribution plans, called appropriate personal pensions or stakeholder pensions, is similar in some ways to the individual accounts proposed for the United States by the President’s Commission (2001), this paper focuses on that approach. The U.K. government serves as a clearinghouse for contributions to VCO accounts. Workers and employers pay the full amount of Social Security contributions to the government. For workers taking the VCO, part of the Social Security contributions (the contracting-out rebates) are diverted by the government to the workers’ individual account, in exchange for which the workers receive lower Social Security benefits at retirement. Once annually, approximately 18 months following the beginning of the tax year, the government pays the workers’ VCO contributions for the entire previous tax year to the workers’ individual account. During the 18 months before the government makes the payment to the workers’ individual accounts, the government pays no interest on the accumulated contributions it holds. This long delay between workers making contributions and the contributions being credited to the workers’ accounts is done to reduce administrative expenses. It is also a feature of the Swedish mandatory individual account system, where the delay is also about 18 months (Turner 2004). This compares with the requirement for U.S. 401(k) plans that all contributions be credited to the workers’ accounts by the 15th business day following the month in which the contribution was made. Although the U.K. government receives and disburses contributions to individual accounts, it does not have a record keeping function for the VCO accounts. In the U.K. system, rather than the government serving as the record keeper for the entire system, each worker has an account with the company managing the investments of the VCO, usually an insurance company. The amount of the VCO payment relative to total Social Security contributions can act as an incentive or disincentive to contracting out depending on whether the resulting benefit is more or less than the Social Security benefit that is given up. That result depends on the rate of return the worker receives on the VCO. Workers can contract out (take the VCO) and contract in as many times as they like, but they must do so for complete tax years. 3 Since 1993, an age-related payment to the individual account has been used in the U.K. The payment to the individual account rises with age to provide a financial incentive not to contract back into the Social Security system at later ages. Workers must take the benefit between ages 60 and 75. Because men cannot receive Social Security benefits until age 65 (women can receive them at 60), the availability of contracted-out benefits at age 60 may provide an incentive to men to contract out. Benefits can be received as a phased payout up to age 75 and then must be annuitized on a unisex basis. Workers are not required to annuitize their benefits when they first take them but are given flexibility as to when they annuitize. This flexibility permits workers to pick a time to annuitize when they think that interest rates are relatively high, and thus they would receive relatively high annuitized benefits. In this way, they can try to mitigate interest rate risk, with low interest rates at the time of annuitization reducing their benefits. However, in practice most workers annuitize when they retire and do not take advantage of the option to wait until a time when interest rates are more favorable. Workers must contract with an insurance company to receive their annuitized benefits. The workers are responsible for investigating the prices charged by insurance companies. Generally, workers annuitize with the insurance company managing their pension account. Insurance companies must calculate annuities on a unisex basis. The pension benefit bought from the insurance company with the individual account must be price indexed (up to 3 percent per year). If the individual is married, the company must provide a 50 percent survivor’s benefit. However, an unmarried worker can use the entire amount to purchase a single life pension, and receive a benefit that could be 10 to 15 percent higher than the joint life pension received by a married person. The payments the U.K. government currently provides to individual accounts for contracted-out workers are designed to be the actuarial equivalent of the benefits given up by a married person. The government payment to the individual accounts is calculated assuming personal pension fees of one percent per year. Experience in the U.K. indicates that inertia in contracting out is an important consideration. Once people contract out, they tend to continue participating in the individual account in spite of market downturns. Experience suggests that few people attempt to time their participation in the individual account based on their expectation as to future market returns. The U.K. Government Actuary’s Department (GAD) reevaluates the payment rate for individual accounts every five years, taking into consideration changes in life expectancy and interest rates. The payment rate fell during the late 1990s, but rose in 2002-03. Changes over time in the government’s payment rate for taking individual accounts may affect the workers’ decisions as to whether to take the VCO. In the U.K., as the terms of the tradeoff have changed for the VCO compared to staying in Social Security, past arrangements have been grandfathered, so that there are separate 4 calculations for benefits accrued at different times. This creates a complex system for determining benefit payments out of Social Security, and makes it practically impossible for workers to determine what their Social Security benefits will be. The value of the government’s payments to the individual accounts, as periodically calculated by the GAD, is affected by changes in real interest rates. If real interest rates fall, the expected investment returns that workers receive will fall, and the payment calculated in the U.K. would rise. In 2002, at the time the payment rate was last set, real interest rates had fallen. However, the GAD expected that fall to be temporary so it did not fully adjust upwards the payment rate. The GAD is under competing pressures. On the one hand, keeping the payments low saves the government money. On the other, if it sets the payment rate too low, people would not take VCOs, and so it thus may be under some pressure to set a generous payment rate to encourage people to take the VCO. In 2004, two large U.K. insurance companies, Prudential and Norwich Union, sent letters to hundreds of thousands of policyholders telling them that the GAD had set the benefit offset rate too low. The policyholders were advised that they would be better off to not participate in their VCO accounts and should return to the traditional Social Security program. As a result of that advice and for other reasons, in 2004, 500,000 people abandoned VCO pensions and returned to Social Security (Cohen 2005). This decline occurred in part because some insurance companies advised their clients that they would be returned to full participation in Social Security unless they advised the company otherwise. The calculation of the government’s payments to the individual accounts assumes the continued existence of the U.K. Social Security benefit in its current form. If further cutbacks occur in the generosity of the benefit structure, as is anticipated given population aging, the payments may turn out to have been overly generous for the benefit actually received (Pensions Policy Institute 2004). 3. The Terms of the Offset Between the VCO and Traditional Social Security Benefits A VCO account is like a loan from the government because the worker must accept reduced future Social Security benefits in exchange for the diversion today of Social Security payroll taxes into the individual account. The terms of the offset between contributions to an individual account and reductions in future Social Security benefits are a key aspect of the structure of VCOs because they affect the decisions workers make as to whether to take the VCO. For U.S. workers choosing the VCO, their reduction in traditional Social Security benefits at retirement would be determined by the Social Security Administration (SSA). For each person taking a VCO, SSA would create and track a hypothetical individual account. SSA would credit to the hypothetical account the value of the payroll tax payments diverted from Social Security. It would credit interest earnings to the hypothetical account at an interest rate that in President Bush’s proposal is fixed at 3 percent real, but in practice could vary in real as well as nominal terms. When the worker claims Social Security benefits, SSA then would convert the amount in this hypothetical account to a hypothetical annuity using a unisex 5 mortality assumption. The worker’s annual Social Security benefits would be reduced by the annual benefit amount of this hypothetical annuity. The key element in this calculation is the real interest rate credited to the hypothetical account. The higher the interest rate, the greater is the reduction in the worker’s Social Security benefit. Thus, the higher the interest rate used, the less generous are the conditions under which the worker takes the VCO. The interest rate on government bonds is a benchmark for comparing the generosity of the terms of the tradeoff between the VCO and Social Security benefits. The worker could invest his or her VCO account in government bonds and be guaranteed a risk-free rate of return on the VCO account. If the interest rate credited to the hypothetical account is less than the interest rate on government bonds, the worker could make a virtually risk-free investment in government bonds and be guaranteed that the amount in the VCO account at retirement would exceed the amount in the hypothetical account at retirement. Thus, the amount in the VCO account would exceed the amount by which Social Security benefits would be reduced. Generosity. The smaller the reduction in the worker’s future Social Security benefits when taking the VCO, the more favorable to the worker is the VCO but the greater is the cost to the Social Security system of offering the VCO. A generous VCO with a low assumed rate of return for calculating the offset may result in the traditional Social Security system providing a substantial subsidy of individual accounts, therefore worsening the long-term financing of Social Security and increasing the need for additional funding. The problem of determining the generosity of the offset is highlighted in the Report of the President’s Commission (2001). In its three proposals, the commission sets three different rates. To determine the offset, it reduces future Social Security benefits the worker would receive by an amount based on the reduction in payroll taxes compounded by a real interest rate of 2 percent, 2.5 percent, or 3.5 percent. These three rates result in three levels of reduction in traditional Social Security benefits. The amount of the benefit offset determines the effect of the VCO on the long-range solvency of Social Security and the degree of needed general revenue financing, benefit cuts, tax increases, or added debt. If the worker is required to forgo a portion of benefits actuarially equivalent to the amount of benefits that would have been paid for by the reduction in his or her Social Security payroll taxes, there is no long-run effect in the aggregate on Social Security finances. If the benefit offset deviates from actuarial equivalence, it would affect Social Security’s long-run finances. Since the U.S. Treasury is expected to receive a real rate of return of 3 percent on its bonds over the long term, the rates of 2 and 2.5 percent for determining the reduction in Social Security benefits are less than a worker could receive if the worker invested his or her individual account in U.S. Treasury bonds. Thus, the Social Security system would subsidize the individual accounts if the offset were set at those lower rates. However, U.S. Treasury bonds in 2005 were earning less than a 3 percent real interest rate, and some analysts expected that lower rate to persist. 6 The possibility of a government subsidy to privatized individual accounts resulting from VCOs is not purely hypothetical. The British government publicized that it initially established a favorable offset for workers to encourage them to choose VCOs. It subsequently estimated that the present value of the reduction in future state benefits was £5.9 billion (in 1988 prices) lower than its cost in incentives provided to take the VCO. The government’s cost in incentives was roughly twice as great as the government’s savings through reduced benefit payments (Budd and Campbell 1998). The perceived generosity of a carve out depends on the state of financial markets. When prices in financial markets are rising, the VCO would be more attractive than when they are falling. If financial markets have a prolonged period of poor performance, political pressure may grow for increasing the generosity of the VCO or for allowing workers to leave the VCO and return to Social Security, if they were not otherwise permitted to do so. Age Neutrality. When contracting out into personal pensions was first allowed in the U.K., the government provided a single rate for workers of all ages for determining its payments to individual accounts. With the single rate, the payments were particularly generous at young ages, but were generally viewed by workers and policy analysts as insufficient at older ages. Since the tax year 1993-94, the payments have been age-related, with more generous payments at older ages. However, the age-related payments to the individual accounts are capped, so that for workers at older ages they are still insufficient. Younger workers receive lower governmental payments to their individual accounts than older workers. Individual accounts are more favorable relative to the Social Security defined benefit plans for younger workers because the defined benefit Social Security plan in the U.K. is backloaded. In 2001-02, a 20-year-old received a 4 percent payment rate, although a 50-year-old received the maximum payment rate of 9 percent of covered salary. Age-related payment rates designed to keep the contracting-out arrangements age neutral are complex, expensive to administer and poorly understood by workers. Gender Neutrality. A further difficulty with VCOs is the problem of structuring a gender neutral benefit offset. Because women have a longer life expectancy than men, a gender- blind offset would not be gender neutral in its effect on workers who are deciding whether to take it. The annuitized benefit from a defined benefit Social Security program is worth more to women than to men because women tend to live longer. For this reason, women need a higher payment than do men into their individual account to provide them with an annuitized benefit of equal annual value to the annuitized benefit forgone from Social Security. If the payment into the individual account is the same for men and women, women will be less likely to take the individual account. The offset in the U.K. is the same for men and women, but it is not gender neutral. In the 45-54 age range, 93 percent of eligible men choose the individual account pension versus 32 percent of eligible women. The explanation for this difference presumably is the longer life expectancy of women, but it could also be explained by greater risk aversion among women than men causing women to be less likely to choose an individual account. 7 It is difficult for VCO accounts to replicate features of the benefits provided by Social Security, such as inflation protection. Social Security benefits in the United States are fully price indexed. Price indexation of benefits would probably not be a feature of the benefits provided by U.S. individual accounts. Price indexation has not been provided to any extent by U.S. insurance companies providing annuities. Price indexation of benefits of up to 3 percent per year in individual accounts is provided in the U.K., with the government paying for price indexation beyond 3 percent if inflation exceeds that amount. Taxation. The tax treatment of VCO individual accounts versus traditional Social Security can affect workers’ decisions as to which to take. The tax treatment relates to the treatment of contributions as well as benefits. Contributions to individual accounts taken out of workers’ payroll taxes are entirely after-tax. Because the remaining contributions to Social Security are primarily from before-tax earnings, the benefits received from Social Security and from the individual account would need to be taxed differently so as to offset the advantage to the pre-tax contribution to Social Security. 4. The Effects of Voluntary Carve out Individual Accounts VCO accounts may affect the behavior and finances of workers, employers, and the Social Security program. Workers’ Issues Workers who take VCOs make a number of decisions that could affect the level and security of their retirement benefits. Although workers are known to make investment errors in managing individual accounts (Turner 2004), this section focuses on the additional problems facing workers because their accounts are VCO accounts. Information Costs and Decision Errors. Information costs and the complexity of choice are greater for VCOs than for other types of individual accounts. Not only do workers need to understand the fundamentals of investing, as is necessary for other types of individual accounts, but they also need to be able to compare the risks and returns of individual accounts with those of traditional Social Security. Workers would need to be able to understand how benefits accrued in Social Security at different ages and how their marital status, the earnings of their spouse, and their expected mortality affect the decision to take a VCO. An argument offered in favor of voluntary individual accounts is that workers will opt for the accounts only if they expect to be made better off by them since they are not forced to participate in them. That argument assumes that expanded choice is always desirable. An advantage of greater choice is that it may allow workers to choose a retirement arrangement that better suits their risk preferences. Whether an expanded range of options offered to workers is a social good depends on both the costs and benefits of the additional choices. Expanded choices increase the chances of decision errors, and a proper assessment of alternative reforms involving choice should take into 8 account potential errors and misperceptions (Sheshinski 2000). Decision errors and other costs associated with choosing may outweigh the benefits for many workers, particularly if they are unable to reverse their choice. This situation happens particularly when workers lack expertise and their evaluation requires complex calculations and comparisons. Long-range discounting of future benefits by workers and the value of annuitization to them depend on uncertain personal attributes such as future earnings, health, and mortality experience, as well as on market factors such as future annuity yields, rates of return on investment assets, and inflation (Blake 2003). It is difficult for workers early in life to evaluate the uncertainties affecting the decision to take a VCO. Later in life when mistakes in decision making become apparent, it might be too late to rectify the situation. Risk-Return Tradeoff. A key aspect of whether workers make good choices concerns how well workers are able to assess the relative risks and returns of the alternatives. In deciding whether to take the VCO, workers would need to consider that the interest rate credited to their hypothetical account is a risk-free interest rate, while the interest rate earned on their actual account is a risky rate. Individual accounts may appear to provide higher rates of return than Social Security, but that comparison may mislead some workers. They might not consider the higher financial market risk of individual accounts. Less frequently recognized, the comparison does not take into account the financing of the transition cost. Because VCOs withdraw money from Social Security, additional financing would be needed to pay the Social Security benefits already promised to current and future retirees. Financial Advice. The lack of financial sophistication of many workers creates the potential for abuse of a VCO system by pension service providers with a financial interest in workers choosing those accounts. That problem occurred in the U.K. with the personal pensions “mis-selling” scandal. More than two million people were mis-sold individual accounts, meaning they were sold accounts when they would have been better off remaining in their company’s defined benefit plans or in Social Security. More than 40 percent of those who initially took VCOs with personal pensions were affected. Financial service providers have had to pay £12.5 billion in compensation to those who were misled, who were primarily lower-wage workers (Gillion et al. 2000)6. Regulation is needed to control financial advice given when service providers have a conflict of interest. Fees. VCO accounts are more expensive to administer than mandatory individual accounts because of the expanded element of choice. High charges are a key weakness of the U.K.’s first attempt at introducing individual VCO accounts. These plans are subject to both high front-loaded charges and high annual charges. In 1998, the combined effect of these charges was equivalent to an average reduction in yield of 3.2 percent per year for 10-year plans and 1.7 percent per year for 25-year plans (Blake and Board 2000: Table 3). Further, plans change their fee structures on a regular basis (see Money Management’s annual Personal Pensions publications), which makes it difficult to compare funds over time. Also, it raises questions as to whether particular fee structures, and changes to them, are used to conceal the true impact of fees. One study found that fees reduced the value of VCO accounts by between 40 and 45 percent (Murthi, Orszag and Orszag 2001). This compares to a reduction of 28 to 33 percent in 6 Personal pensions are similar to individual retirement accounts. 9 Chile (Gil, Packard and Yermo 2005), and a reduction of approximately 20 percent on average in Sweden (Turner 2004). In the U.K., market forces alone were not sufficient to drive down charges to competitive levels on pensions for individual account holders. Because of this market failure, the U.K. government introduced “stakeholder” pension plans in 2001 that are subject to a statutory maximum annual fee of 1 percent of asset values with no entry or exit fees.7 As a consequence of this regulatory limit, providers of stakeholder pensions have greatly reduced the amount of “free” advice they provide to workers trying to decide whether to choose these accounts (Bolger 2001). Low Persistency with Voluntary Arrangements. When workers have the option to end their participation in the VCO and return to full participation in the Social Security system, there may be problems arising from low persistency of participation in the VCO. Low persistency suggests dissatisfaction with VCOs. A regular premium pension plan involves a substantial commitment of time and resources by both the plan provider and its members if the desired objectives are to be achieved. As argued by the U.K. Personal Investment Authority (PIA)8: “if investors buy policies on the basis of good advice,… they would not normally be expected to cancel premiums to their policies unless forced to do so by unexpected changes in their personal circumstances. This means that persistency can be a powerful indicator of the quality of the selling process.” (Personal Investment Authority 1999, p. 3). The PIA defines persistency as “the proportion of investors who continue to pay regular contributions to their personal policies, or who do not surrender their single premium policy” (p. 3). Persistency rates (i.e., the percentage of policies that have not lapsed) after 4 years of membership are between 57 percent and 67 percent in the U.K. (Personal Investment Authority 1999, Table 1), and industry estimates suggest that the persistency level after 25 years is just 16 percent (Blake and Board 2000). These figures suggest that very few personal pension plan members are likely to maintain plan membership for long enough to build up an adequate pension. The PIA regards these persistency rates as “disturbing” (1998, p. 10) and offers a number of explanations: members were mis-sold pensions that were either unsuitable or too expensive; regular premium policies might be unsuitable for those with irregular earnings or uncertain long- term employment; a change of employment may lead to members joining employer-provided pension plans and abandoning their individual account VCOs; and adverse general economic conditions could worsen persistency rates. Investment Choices and Moral Values. One of the issues in considering the investment choices that would be offered U.S. participants in VCOs is the moral or ethical dimension of investing. It is likely that participants’ choices would be limited to index funds because these funds offer broad diversification with relatively low fees. However, participants may object to these funds on moral or ethical grounds for a variety of reasons because some companies engage 7 This was subsequently raised in 2004 to 1.5 percent for the first 10 years as a result of complaints to the government from stakeholder plan providers that they could not make any profit with a one percent fee. 8 Replaced in 2001 by the Financial Services Authority. 10 in activities contrary to their values. Depending on their values, some people may object to investing in companies that engage in stem cell research; or companies that receive a substantial portion of their revenue from gambling; or that produce weapons, alcohol, or cigarettes; or that produce movies, television programs, video games, computer software, music or publications that they consider to have objectionable content, including entertainment that popularizes the occult; or in companies known to discriminate against women, exploit child labor, are anti-union, or engage in other unfair labor practices; or in companies that pollute and despoil the environment; or in companies that do business with terrorists or with countries that support terrorists; or with countries that have objectionable policies, such as the current situation of mistreatment of black Africans in the Sudan. Most broad equity or bond index funds would include companies in some of these categories of business, and thus some people may decide not to participate in a VCO because of ethical or moral objections to the companies in which they would be required to invest. Treatment of Low-Income Workers. Several features reduce the likelihood of low- income workers taking a VCO, as is seen to be the case in the U.K. First, lower-income workers tend to be more risk averse concerning the investment of their retirement savings than upper- income workers (Hinz, McCarthy, and Turner 1996). Although a VCO invested in equity could help lower-income workers achieve better diversification of their sources of retirement income than by relying solely on Social Security (Schieber and Shoven 1999), the replacement of low risk Social Security benefits with more volatile financial market assets would increase the risk as well and this may not be viewed as desirable by low-income workers. Second, Social Security in the United States represents a larger percentage of retirement income for lower-income workers than for upper-income workers. For this reason, a VCO has a larger effect on the total risk borne by lower-income workers. Social Security provides 82 percent of the total income of the aged in the bottom fifth of the income distribution. For 18 percent of beneficiaries, Social Security is their only income. By comparison, for the aged in the top fifth of the income distribution, Social Security accounts for only 18 percent of total income (Social Security Advisory Board 2001). Further, if taking a VCO were to reduce Social Security benefits by 10 percent, that would be a change in the risk characteristics of 10 percent of the retirement income for the bottom 18 percent of the income distribution, although it would be a change of less than 2 percent for the top 20 percent of the income distribution. Thus, a voluntary carve out would cause a considerably larger change in the retirement income sources for low- income workers than it would for high-income workers. Third, transactions costs unrelated to transaction size, such as information costs in determining the best option, may outweigh any presumed advantages of a VCO. For low-income workers, the absolute advantage of taking a VCO would be smaller, other things equal, because their level of benefits would be lower (Disney, Palacios and Whitehouse 1999). Higher-income workers, with established personal savings and participation in 401(k) plans, more likely have knowledge of investments, so the information costs of setting up VCOs would be lower for them. Fourth, the insurance features of Social Security benefits may be more important to low- income workers than to high-income workers. For example, the feature that permits full career workers to drop their five lowest years of earnings when calculating U.S. Social Security benefits 11 may be more valuable to low-income workers who have a greater risk of unemployment. The survivors and disability insurance provided by Social Security may be more important to low- income than to high-income workers. The U.K. has encouraged workers to choose a VCO, and permits defined benefit plans to be used for the carve out. Of those that remain fully in Social Security, 65 percent are women and 70 percent earn less than half of average full-time male earnings (Budd and Campbell 1998). Thus, most of the workers who remain fully in the Social Security system are lower-paid workers. Because a carve out would reduce their Social Security benefits, workers might invest their individual accounts more conservatively than if the individual account were an add-on to Social Security. Thus, U.S. workers’ investments might be more conservative than those of workers currently investing their 401(k) balances. If workers were to invest more conservatively, their expected returns on their individual accounts would be lower. Employers’ Issues Increased Complexity of Employee Benefit Design. With some U.S. workers choosing the VCO and others choosing to remain fully in Social Security, employers would face increased complexity in designing retirement benefits for their workforce. For those workers who chose the VCO, the amount of retirement benefits they would receive through a defined benefit plan— the traditional Social Security plan—would be reduced, and the amount they would receive through a defined contribution plan—the individual account—would be increased. This change in benefits might cause offsetting effects—it might reduce the demand by workers for employer- provided defined contribution plans and increase the demand for defined benefit plans. At the same time, the demand for employer-provided plans by those workers remaining in Social Security would presumably be unchanged. Since the participation rates would likely be higher for high- than for low-income workers, VCOs might complicate nondiscrimination testing of pension benefits for employers. Nondiscrimination testing is required by the Employee Retirement Income Security Act (ERISA) to ensure that moderate and lower-income workers receive adequate pension benefits through employer-provided pension plans, relative to the benefits received by upper-income workers. Effects on Social Security Social Security Administrative Costs. VCOs would increase the administrative costs of Social Security in the United States. A tradeoff exists between the amount of choice allowed workers and the level of administrative expense. It tends to be more expensive to administer a voluntary carve out system than a mandatory system. The record keeping requirements of the proposed individual accounts are considerably greater than for the U.S. Thrift Savings Plan or for the British or Swedish mandatory individual account systems. Such requirements are increased because of the complexities arising from the considerably larger size of the U.S. labor force compared to the British or Swedish labor force. The record-keeping requirements are also 12 greater because of the complexities in the system that has been proposed, with the need to track both actual and hypothetical accounts for all workers. If the Social Security Administration served as a clearinghouse, it might be unable to credit the funds going into VCOs in a timely manner. This was a major problem faced by the British Social Security system, and it was dealt with by taking the relevant agency (the Contributions Agency) away from the Department of Social Security and placing it within the Treasury (Blake 2003). The record-keeping for the hypothetical accounts would increase the administrative costs. The government would need to maintain a hypothetical account for each worker, as well as tracking the account balance in each individual account. In the hypothetical account, the government would have to track each worker’s contribution and the amount of interest that it credits for determining the reduction in future Social Security benefits. In addition, for all married couples, the government would need to be able to determine the amount of contributions made by each during the marriage, and the amount of actual (not hypothetical) investment earnings received during marriage for both partners, so that the amount might be split in the event of divorce. An important aspect of administrative costs is administrative feasibility. The computer capability required for managing large numbers of individual accounts is considerable and the U.K., Sweden and the federal Thrift Savings Plan (Thrift Savings Plan 2001) have encountered major problems with the computer systems to administer mandatory individual accounts. Problems with the calculations done by the National Insurance computer system in the U.K. have caused difficulty in paying thousands of VCO benefits that it took months to sort out (BBC 1998). 5. Conclusions Although the strengths and weaknesses of individual accounts as part of Social Security reform have received considerable attention, most analysis has focused on mandatory accounts. Relatively little attention has been paid to voluntary carve out accounts. This article draws lessons for the United States from the experience of the U.K. VCO accounts destabilize Social Security financing. By withdrawing money that would otherwise be paid into Social Security, these accounts would destabilize U.S. Social Security financing for decades (the transition cost). Many people in the U.K. are leaving VCOs and returning to the traditional Social Security program. Two large insurance companies have sent letters to hundreds of thousands of policyholders telling them that they are better off not participating in their VCO accounts and should return to the traditional Social Security program. Proposals for the United States have not contained the feature that workers would be allowed to return to Social Security, but the British experience suggests that there would be political pressure at times for that to be allowed. 13 VCO accounts would not be a good arrangement for low-income workers who depend on Social Security benefits. The reduction in Social Security benefits when a low-income worker takes a VCO would be considerably larger as a percent of their total retirement income than for a high-income worker for whom Social Security is a much less important source of benefits. Thus, the change in the risk that a low-income worker would bear by reducing Social Security benefits and taking a VCO would be considerably larger. To the extent that lower-income workers are more risk averse than higher-income workers, they would likely invest in low-risk assets and have a lower expected return. The administrative costs of providing annuitized benefits would be relatively high. The cost of converting each individual account to an annuity, with private insurance companies providing the annuities, would be relatively high because of the high cost of individualized transactions, especially with many of the accounts being relatively small. In periods of financial market downturns, there would be pressure on the government to let people participating in VCO accounts return to Social Security. In part because of the poor performance of financial markets, workers are being encouraged in the U.K. to not take a VCO. If the goal is to offer VCO accounts on a basis that is neutral with respect to the Social Security benefits workers have foregone, then many workers who take them will be made worse off because of the fluctuations in asset markets. If the government subsidizes VCOs, the first problem will be lessened, but the expense to the government will increase. In conclusion, while a number of issues have arisen with VCO accounts in the U.K., perhaps the most fundamental problem is that it is difficult to determine the appropriate relationship between the reduction in the worker’s payment to Social Security and the reduction in that person’s Social Security benefits. Because of this problem, insurance companies in the U.K., the only country with such a system, are now arguing that the government made an error in setting the current rate and that workers should not take the VCO. References British Broadcasting Corporation (BBC). “No End to Pensions Problems.” November 4, 1998. Available at: http://news.bbc.co.uk/hi/english/uk_politics/newsid_208000/208136.stm. Blake, David. Pension Schemes and Pension Funds in the United Kingdom. Oxford, England: Oxford University Press, 2003. Blake, David, and Board, John. “Measuring Value Added in the Pensions Industry.” The Geneva Papers on Risk and Insurance 25 (October 2000): 539-67. Bolger, Andrew. “Warning Over Lack of Advice on Low-cost Pensions.” Financial Times, November 7, 2001 Budd, Alan and Campbell, Nigel. “The Roles of the Public and Private Sectors in the U.K. Pension System.” In Privatizing Social Security, edited by Martin Feldstein. Chicago: 14 University of Chicago Press, 1998. Cohen, Norma. “A Bloody Mess.” The American Prospect online edition, January 11, 2005. Disney, Richard; Palacios, Robert; and Whitehouse, Edward. “Individual Choice of Pension Arrangement as a Pension Reform Strategy.” The Institute for Fiscal Studies, Working Paper Series No. W99/18. Feldstein, Martin (ed.). Privatizing Social Security. Chicago: University of Chicago Press, 1998. Gill, Indermit S.; Packard, Truman; and Yermo, Juan. Keeping the Promise of Social Security in Latin America. Washington, DC: The World Bank, 2005. Gillion, Colin; Turner, John; Bailey, Clive; and Latulippe, Denis, (eds.). Social Security Pensions: Development and Reform. Geneva, Switzerland: ILO, 2000. Gustman, Alan L. and Steinmeier, Thomas L. “Privatizing Social Security: First-Round Effects of a Generic, Voluntary, Privatized U.S. Social Security System.” In Privatizing Social Security, edited by Martin Feldstein. Chicago: University of Chicago Press, 1998. Gustman, Alan L. and Steinmeier, Thomas L. “Retirement Effects of Proposals by the President’s Commission to Strengthen Social Security.” NBER Working Paper No. 10030, October 2003. Hinz, Richard; McCarthy, David; and Turner, John. "Are Women Conservative Investors? Gender Differences in Participant Directed Pension Investments." In Positioning Pensions for the Year 2000, edited by Olivia Mitchell. Philadelphia: University of Pennsylvania Press, 1996. Kotlikoff, Laurence J.; Smetters, Kent A.; and Walliser, Jan. “Opting Out of Social Security and Adverse Selection.” NBER Working Paper No. 6430, February 1998. Money Management: Personal Pensions, Financial Times Business, London. Murthi, Mamta; Orszag, J. Michael, and Orszag, Peter R. “Administrative Costs Under a Decentralized Approach to Individual Accounts: Lessons from the United Kingdom.” In New Ideas about Old Age Security, edited by Robert Holzmann and Joseph E. Stiglitz. Washington, DC: The World Bank, 2001. National Academy of Social Insurance. Uncharted Waters: Paying Benefits from Individual Accounts in Federal Retirement Policy. Study Panel Final Report. Co-chaired by Kenneth S. Apfel and Michael J. Graetz. Washington, DC: National Academy of Social Insurance, 2005. 15 Orszag, Peter R. and Greenstein, Robert. “Voluntary Individual Accounts for Social Security: What Are the Costs?” Washington, DC: Center on Budget and Policy Priorities, August 21, 2001. Pensions Policy Institute. “Pensions Primer.” April 2004. Personal Investment Authority (1998), Fourth Survey of Persistency of Life and Pension Policies, October, London. Personal Investment Authority (1999), Fifth Survey of Persistency of Life and Pension Policies, October, London. President’s Commission. “Strengthening Social Security and Creating Personal Wealth for All Americans: Report of the President’s Commission.” December 11, 2001. Schieber, Sylvester L. and Shoven, John B. The Real Deal: The History and Future of Social Security. New Haven: Yale University Press, 1999. Sheshinski, Eytan. “Bounded Rationality and Socially Optimal Limits on Choice: An Example.” The Hebrew University of Jerusalem, January 2000. Social Security Advisory Board. “Social Security: Why Action Should be Taken Soon. (revised edition).” July 2001. http://www.ssab.gov/actionshouldbetaken.pdf Thrift Savings Plan. “Progress Report on the New Record Keeping System.” Highlights November 2001. Turner, John. “Individual Accounts: Lessons from Sweden.” International Social Security Review 57 No. 1 (2004): 65-84. Turner, John and Rajnes, David M. “ ‘Pay or Play’ Pensions in Japan and the United Kingdom.” In Social Security: Time for a Change, edited by Kevin Stephenson, JAI Press, 1995. Williamson, John B. “Social Security Privatization: Lessons from the United Kingdom.” Center for Retirement Research at Boston College CRR WP 2000-10, November 2000. World Bank. Averting the Old Age Crisis: Policies to Protect the Old and Promote Growth. Washington, DC: World Bank, 1994. 16