Chairman Smith, Ranking Member Kerry, and Senator Grassley, we appreciate the opportunity to testify before you. We are submitting our testimony as a single joint statement because we believe strongly in the need for a common strategy to expand retirement savings, and in the importance of approaching these issues in a manner that transcends ideological and partisan differences.
At the request of Committee staff, this written statement focuses on our proposal to expand retirement savings for small business workers - the automatic IRA. We are pleased by the positive reaction the proposal has received and are grateful to our colleagues, including those in government and in various stakeholder organizations, who have contributed to these ideas.
With the looming retirement security crisis facing our country, policy-makers from both parties are focused on ways to strengthen pensions and increase savings. Our proposal for automatic IRAs would provide a relatively simple, cost-effective way to increase retirement security for the estimated 71 million workers whose employers (usually smaller businesses) do not sponsor plans. It would enable these employees to save for retirement by allowing them to have their employers regularly transfer amounts from their paycheck to an IRA.
We are by no means suggesting that the automatic IRA proposal is the only step that should be taken to expand retirement savings for small business workers. In fact, we have long believed in the primacy of employer-sponsored retirement plans as vehicles for pension coverage. Additionally, we continue to advocate strongly for the expansion of pension coverage through automatic features in 401(k) and similar retirement savings plans.
The automatic 401(k) approach makes intelligent use of defaults - the outcomes that occur when individuals are unable or unwilling to make an affirmative choice or otherwise fail to act - to enlist the power of inertia to promote saving. Automating enrollment, escalation of contributions, investment, and rollovers expands coverage in several ways. Enrolling employees in a plan unless they opt out increases significantly the number of eligible employees who participate in the plan. Escalating the amount of the default contribution tends to increase the amount people save over time. Providing for a default investment (which participants can reject in favor of other alternatives) reflecting consensus investment principles such as diversification and asset allocation tends to raise the expected investment return on contributions. Finally, making retention or rollover of benefits rather than consumption the default when an employee leaves a job furthers the long-term preservation of retirement savings for their intended purposes. By helping improve performance under the nondiscrimination standards and generally making plans more effective in providing retirement benefits, the automatic 401(k) can also encourage more employers to sponsor or continue sponsoring plans.
The automatic IRA builds on the success of the automatic 401(k). Moreover, as explained below, we would intend and expect the introduction of automatic IRAs to expand the number of employers that choose to sponsor 401(k) or SIMPLE plans instead of offering only automatic IRAs. But for millions of workers who continue to have no employer plan, the automatic IRA would provide a valuable retirement savings opportunity.
The automatic IRA proposal is set out in the remainder of this written statement.
A second major motivation for using 401(k) auto enrollment in many companies is management's sense of responsibility or concern for employees and their retirement security. Many executives involved in managing employee plans and benefits have opted for auto enrollment because they believe far too many employees are saving too little and investing unwisely and need a strong push to "do the right thing" and take advantage of the 401(k) plan. This motivation-by no means present in all employers-is especially unlikely to be driving an employer that merely permits payroll deposit to IRAs without sponsoring a retirement plan.
Third, employers might have greater concern about potential employee reaction to auto enrollment in the absence of an employer matching contribution. The high return on employees' investment delivered by the typical 401(k) match helps give confidence to 401(k) sponsors using auto enrollment that they are doing right by their employees and need not worry unduly about potential complaints from workers who failed to read the notice.
Finally, an employer concern that has made some plan sponsors hesitate to use auto enrollment with 401(k) plans might loom larger in the case of auto enrollment with direct deposit IRAs. This is the concern about avoiding a possible violation of state laws that prohibit deductions from employee paychecks without the employee's advance written authorization. Assuming most direct deposit IRA arrangements are not employer plans governed by ERISA, such state laws, as they apply to automatic IRAs, may not be preempted by ERISA because they do not "relate to any employee benefit plan." For reasons such as these, without a meaningful change in the law, most employers that are unwilling to offer a qualified plan today are unlikely to take the initiative to automatically enroll employees in direct deposit IRAs.
Not Requiring Employers to Use Automatic Enrollment
One possible response would be to require employers to use automatic enrollment in conjunction with the direct deposit IRAs (while giving the employers a tax credit and legal protections). The argument for such a requirement would be that it would likely increase participation dramatically while preserving employee choice (workers could always opt out), and that, for the reasons summarized above, employers that do not provide a qualified plan (or a match) are unlikely to use auto enrollment voluntarily. The arguments against such a requirement include the concern that a workforce that presumably has not shown sufficient demand for a qualified retirement plan to induce the employer to offer one might react unfavorably to being automatically enrolled in direct deposit savings without a matching contribution. (In addition, some small business owners who have only a few employees and work with all of them on a daily basis might take the view that automatic enrollment is unnecessary because of the constant flow of communication between the owner and each employee.)
It is noteworthy, however, that recent public opinion polling shows strong support among registered voters for making saving easier by making it automatic, with 71 percent of respondents favoring a fully automatic 401(k), including automatic enrollment, automatic investment, and automatic contribution increases over time, with the opportunity to opt out at any stage. A vast majority (85 percent) of voters said that if they were automatically enrolled in a 401(k), they would not opt out, even when given the opportunity to do so. In addition, given the choice, 59 percent of respondents preferred a workplace IRA with automatic enrollment to one without.
Requiring Explicit "Up or Down" Employee Elections While Encouraging Auto Enrollment
An alternative approach that has been used in 401(k) plans and might be particularly well suited to payroll deposit savings is to require all eligible employees to submit an election that explicitly either accepts or declines direct deposit to an IRA. Instead of treating employees who fail to respond as either excluded or included, this "up or down" election approach has no default. There is evidence suggesting that requiring employees to elect one way or the other can raise 401(k) participation nearly as much as auto enrollment does. Requiring an explicit election picks up many who would otherwise fail to participate because they do not complete and return the enrollment form due to procrastination, inertia, inability to decide on investments or level of contribution, and the like.
Accordingly, a possible strategy for increasing participation in payroll deposit IRAs would be to require employers to obtain a written (including electronic) "up or down" election from each eligible employee either accepting or declining the direct deposit to an IRA. Under this strategy, employers that voluntarily auto enroll their employees in the direct deposit IRAs would be excused from the requirement that they obtain an explicit election from each employee because all employees who fail to elect would be participating. This exemption-treating an employer's use of auto enrollment as an alternative means of satisfying its required-election obligation-would add an incentive for employers to use auto enrollment without requiring them to use it. Any firms that prefer not to use auto enrollment would simply obtain a completed election from each employee, either electronically or on a paper form. And either way-whether the employer chose to use auto enrollment or the required-election approach-participation would likely increase significantly, perhaps even approaching the level that might be achieved if auto enrollment were required for all payroll deposit IRAs.
This combined strategy for promoting payroll deposit IRA participation could be applied separately to new hires and existing employees: thus, an employer auto enrolling new hires would be exempted from obtaining completed elections from all new hires (but not from existing employees), while an employer auto enrolling both new hires and existing employees would be excused from having to obtain elections from both new hires and existing employees.
The required election would not obligate employers to obtain a new election from each employee every year. Once an employee submitted an election form, that employee would not be required to make another election: as in most 401(k) plans, the initial election would continue throughout the year and from year to year unless and until the employee chose to change it. Similarly, an employee who failed to submit an election form and was auto enrolled by default in the payroll deposit IRA would continue to be auto enrolled unless and until the employee took action to make an explicit election.
To maximize participation, employers would receive a standard enrollment module reflecting current best practices in enrollment procedures. A nationwide website with standard forms would serve as a repository of state-of-the-art best practices in and savings education. The use of automatic enrollment (whereby employees automatically are enrolled at a statutorily specified rate of contribution - such as 3% of pay -- unless they opt out) would be encouraged in two ways. First, the standard materials provided to employers would be framed so as to present auto enrollment as the presumptive or perhaps even the default enrollment method, although employers would be easily able to opt out in favor of simply obtaining an "up or down" response from all employees. In effect, such a "double default" approach would use the same principle at both the employer and employee level by auto enrolling employers into auto enrolling employees. Second, as noted, employers using auto enrollment to promote participation would not need to obtain responses from unresponsive employees.
Compliance and Enforcement
Employers' use of the required-election approach would also help solve an additional problem-enforcing compliance with a requirement that employers offer direct deposit savings. As a practical matter, many employers might question whether the IRS would ever really be able to monitor and enforce such a requirement. Employers may believe that, if the IRS asked an employer why none of its employees used direct deposit IRAs, the employer could respond that it told its employees about this option and they simply were not interested. However, if employers that were required to offer direct deposit savings had to obtain a signed election from each eligible employee who declined the payroll deposit option, employers would know that the IRS could audit their files for each employee's election. This by itself would likely improve compliance.
In fact, a single paper or e-mail notice could advise the employee of the opportunity to engage in payroll deduction savings and elicit the employee's response. The notice and the employee's election might be added or attached to IRS Form W-4. (As noted, the W-4 is the form an employer ordinarily obtains from new hires and often from other employees to help the employer comply with its income tax-withholding obligations.) If the employer chose to use auto enrollment, the notice would also inform employees of that feature (including the default contribution level and investment and the procedure for opting out), and the employer's records would need to show that employees who failed to submit an election were in fact participating in the payroll deduction savings.
Employers would be required to certify annually to the IRS that they were in compliance with the payroll deposit savings requirements. This might be done in conjunction with the existing IRS Form W-3 that employers file annually to transmit Forms W-2 to the government. Failure to offer payroll deposit savings would ultimately need to be backed up by an appropriate sanction, such as the threat of civil monetary penalties or an excise tax.
Portability of Savings
IRAs are inherently portable. Unlike a 401(k) or other employer plan, an IRA survives and functions independently of the individual saver's employment status. Thus the IRA owner is not at risk of forfeiting or losing the account or suffering an interruption in the ability to contribute when changing or losing employment. As a broad generalization, the automatic IRAs outlined here presumably would be freely transferable to and with other IRAs and qualified plans that permit such transfers. (However, as discussed below, the investment limitations and other cost-containment features of these IRAs raise the issue of whether transferability to other types of vehicles should be subject to restrictions.)
Making a Savings Vehicle Available
Most current direct deposit arrangements use a payroll-deduction savings mechanism similar to the 401(k), but, unlike the 401(k), do not give the employee a ready-made vehicle or account to receive deposits. The employee must open a recipient account and must identify the account to the employer. However, where the purpose of the direct deposit is saving, it would be useful to many individuals who would rather not choose a specific IRA to have a ready-made fallback or default account available for the deposits.
Under this approach, modeled after the SIMPLE-IRA, which currently covers an estimated 2 million employees, individuals who wish to direct their contributions to a specific IRA would do so. The employer would follow these directions as employers ordinarily do when they make direct deposits of paychecks to accounts specified by employees. At the same time, the employer would also have the option of simplifying its task by remitting all employee contributions in the first instance to IRAs at a single private financial institution that the employer designates. However, even in this case, employees would be able to transfer the contributions, without cost, from the employer's designated financial institution to an IRA provider chosen by the employee.
By designating a single IRA provider to receive all contributions, the employer could avoid the potential administrative hassles of directing deposits to a multitude of different IRAs for different employees, while employees would be free to transfer their contributions from the employer's designated institution to an IRA provider of their own choosing. Even this approach, though, still places a burden on either the employer or the employee to choose an IRA. For many small businesses, the choice might not be obvious or simple. In addition, the market may not be very robust because at least some of the major financial institutions that provide IRAs may well not be interested in selling new accounts that seem unlikely to grow enough to be profitable within a reasonable time. Some of the major financial firms appear to be motivated at least as much by a desire to maximize the average account balance as by the goal of maximizing aggregate assets under management. They therefore may shun small accounts that seem to lack much potential for rapid growth.
The current experience with automatic rollover IRAs is a case in point. Firms are required to establish these IRAs as a default vehicle for qualified plan participants whose employment terminates with an account balance of not more than $5,000 and who fail to provide any direction regarding rollover or other payout. The objective is to reduce leakage of benefits from the tax-favored retirement system by stopping involuntary cashouts of account balances between $1,000 and $5,000. (Plan sponsors continue to have the option to cash out balances of up to $1,000 and to retain in the plan account balances between $1,000 and $5,000 instead of rolling them over to an IRA.) Because plan sponsors are required to set up IRAs only for "unresponsive" participants-those who fail to give instructions as to the disposition of their benefits-these IRAs are presumed to be less likely than other IRAs are to attract additional contributions. Accordingly, significant segments of the IRA provider industry have not been eager to cater to this segment of the market. As a result, plan sponsors have tended to reduce their cashout level from $5,000 to $1,000 so that new IRAs would not have to be established.
For somewhat similar reasons, IRA providers might expect payroll deposit IRAs to be less profitable than other products. As a result, employers and employees might well find that providers are not marketing to them aggressively and that the array of payroll deposit IRA choices is comparatively limited.
The prospect of tens of millions of personal retirement accounts with relatively small balances likely to grow relatively slowly suggests that the market may need to be encouraged to develop widely available low-cost personal accounts or IRAs. Otherwise, for "small savers," fixed-cost investment management and administrative fees may consume too much of the earnings on the account and potentially even erode principal.
A Standard Default Account
Accordingly, to facilitate saving and minimize costs, we believe that a strong case can be made for a default IRA that would be automatically available to receive direct deposit contributions without requiring either the employee or employer to choose among IRA providers and without requiring the employee to take the initiative to open an IRA. Under this approach, for the convenience of both employees and employers, those who wish to save but have no time or taste for the process of locating and choosing an IRA would be able to use a standard default, or automatic, account. If neither the employer nor the employee designated a specific IRA provider, the contributions would go to a personal retirement account within a plan that would in some respects resemble the federal Thrift Savings Plan (the 401(k)-type retirement savings plan that covers federal government employees).
These standard default accounts would be maintained and operated by private financial institutions under contract with the federal government. To the fullest extent practicable, the private sector would provide the investment funds, investment management, record keeping, and related administrative services. To serve as a default account for direct deposits that have not been directed elsewhere by employers or employees, an account need not be maintained by a governmental entity. Given sufficient quality control and adherence to reasonably uniform standards, various private financial institutions could contract to provide the default accounts, on a collective or individual institution basis, more or less interchangeably-perhaps allocating customers on a geographic basis or in accordance with other arrangements based on providers' capacity. These fund managers could be selected through competitive bidding. Once individual default accounts reached a predetermined balance (e.g., $15,000) sufficient to make them potentially profitable for many private IRA providers, account owners would have the option to transfer them to IRAs of their choosing.
Both the direct deposit IRAs expressly selected by employees and employers and the standardized direct deposit IRAs that serve as default vehicles would be designed to minimize the costs of investment management and account administration. It should be feasible to realize substantial cost savings through index funds, economies of scale in asset management and administration, uniformity, and electronic technologies.
In accordance with statutory guidelines for all direct deposit IRAs, government contract specifications would call for a no-frills approach to participant services in the interest of minimizing costs. By contrast to the wide-open investment options provided in most current IRAs and the high (and costlier) level of customer service provided in many 401(k) plans, the standard account would provide only a few investment options (patterned after the Thrift Savings Plan, if not more limited), would permit individuals to change their investments only once or twice a year, and would emphasize transparency of investment and other fees and other expenses.
Specifically, costs of direct deposit IRAs might be reduced by federal standards that, to the extent possible,
- Exclude brokerage services and retail equity funds from the investment options available under the IRA
- Limit the number of investment options under the IRA.
- Allow individuals to change their investments only once or twice per year.
- Specify a low-cost default investment option and provide that, if any of an individual's account balance is invested in the default option, all of it must be.
- Prohibit loans (IRAs do not allow them in any event) and perhaps limit pre-retirement withdrawals.
- Limit access to customer service call centers.
- Preclude commissions.
- Make compliance testing unnecessary.
- Give account owners only a single account statement per year (especially if daily valuation is built into the system and is available to account owners).
- Encourage the use of electronic and other new technologies (including enrollment on a web site) for fund transfers, record keeping, and communications among IRA providers, participating employees, and employers to reduce paperwork and cost. Electronic administration has considerable potential to cut costs.
The availability to savers of a major low-cost personal account alternative in the form of the standard account may even help, through market competition, to drive down the costs and fees of IRAs offered separately by private financial institutions. Through efficiencies associated with collective investment and greater uniformity, the standard account should help move the system away from the retail-type cost structure characteristic of current IRAs. It should also help create a broad infrastructure of individual savings accounts that would cover most of the working population.
In conjunction with these steps, Congress and the regulators may be able to do more to require simplified, uniform disclosure and description of IRA investment and administrative fees and charges (building on previous work by the Department of Labor relating to 401(k) fees). Such disclosure should help consumers compare costs and thereby promote healthy price competition.
Another approach would begin by recognizing the trade-off between asset management costs and investment types. As a broad generalization, asset management charges tend to be low for money market funds, certificates of deposit, and certain other relatively low-risk, lower-return investments that generally do not require active management. However, it appears that limiting individual accounts to these types of investments would be unnecessarily restrictive. As discussed below (under "Default Investment Fund"), passively-managed index funds, such as those used in the Thrift Savings Plan, are also relatively inexpensive.
A very different approach to cost containment would be to impose a statutory or regulatory limitation on investment management and administrative fees that providers could charge. One example is the United Kingdom's limit on permissible charges for management of "stakeholder pension" accounts-an annual 150 basis point fee cap for five years that is scheduled to drop to 100 basis points thereafter.  As another and more limited example, the U.S. Department of Labor has imposed a kind of limitation on fees charged by providers of automatic rollover IRAs established by employers for terminating employees who fail to provide any direction regarding the disposition of account balances of up to $5,000. Labor regulations provide a fiduciary safe harbor for auto rollover IRAs that preserve principal and that do not charge fees greater than those charged by the IRA provider for other IRAs it provides.
Presumably, a mandatory limit would give rise to potential cross-subsidies from products that are free of any limit on fees to the IRAs that are subject to the fee limit -- a result that could be viewed either as an inappropriate distortion or as a necessary and appropriate allocation of resources. We would view a mandatory limit as a last resort, preferring the market-based strategies outlined above.
Default Investment Fund
Both the IRAs offered independently by private financial institutions and explicitly selected by employees or employers and the default IRAs would serve the important purpose of providing low-cost professional asset management to millions of individual savers, presumably improving their aggregate investment results. To that end, all of these accounts would offer a similar, limited set of investment options, including a default investment fund in which deposits would automatically be invested unless the individual chose otherwise. This default investment would be a highly diversified "target asset allocation" or "life-cycle" fund comprised of a mix of equities and fixed income or stable value investments, and probably relying heavily on index funds. (The life-cycle funds recently introduced into the federal Thrift Savings Plan are one possible model.) A portion or all of the fixed income component could be comprised of Treasury inflation protected securities ("TIPS") to protect against the risk of inflation.
The mix of equities and fixed income would be intended to reflect the consensus of most personal investment advisers, which emphasizes sound asset allocation and diversification of investments-including exposure to equities (and perhaps other assets that have higher-risk and higher-return characteristics), at least given the foundation of retirement income already delivered through Social Security and assuming the funds will not shortly be needed for expenses. The use of index funds would avoid the costs of active investment management while promoting wide diversification.
This default investment would actually consist of several different funds, depending on the individual's age, with the more conservative investments (such as those relying more heavily on TIPS) applicable to older individuals who are closer to the time when they might need to use the funds. Individuals who selected the default fund or were defaulted into it would have their account balances entirely invested in that fund. However, they would be free to exit the fund at specified times and opt for a different investment option among those offered within the IRA.
The standard automatic (default) investment would also serve two other key purposes. It would encourage employee participation in direct deposit savings by enabling employees who are satisfied with the default to simplify what may be the most difficult decision they would otherwise be required to make as a condition of participation (i.e., how to invest). Finally, the standard default investment should encourage more employers to use automatic enrollment (thereby boosting employee participation) by saving them from having to choose a default investment. This, in turn, would make it easier to protect employers from responsibility for IRA investments, especially employers using automatic enrollment (as discussed below).
We would not fully specify the default investment by statute. It is desirable to maintain a degree of flexibility in order to reflect a consensus of expert financial advice over time. Accordingly, general statutory guidelines would be fleshed out at the administrative level after regular comment by and consultation with private-sector investment experts.
An additional and major design issue is whether the standard, limited set of investment options for payroll deposit IRAs should be only a minimum set of options in each IRA, so that the IRA provider would be permitted to provide any additional options it wished. Limiting the IRAs to these specified options would best serve the purposes of containing costs, improving investment results for IRA owners in the aggregate, and simplifying individuals' investment choices. At the same time, such restrictions would constrain the market, potentially limit innovation, and limit choice for individuals who prefer other alternatives.
One of the ways to resolve this tradeoff would be to limit direct deposit IRAs to the prescribed array of investment options without imposing any comparable limits on other IRAs, and to allow owners of direct deposit IRAs (including default IRAs) to transfer or roll over their account balances between the two classes of accounts. Under this approach, the owner of a direct deposit IRA could transfer the account balance to other (unrestricted) IRAs that are willing to accept such transfers (but perhaps only after the account balance reaches a specified amount that would no longer be unprofitable to most IRA providers). While such a transfer to an unrestricted IRA would deprive the owner of the cost-saving advantages of the no-frills, limited-choice model, such a system would still enable individuals to retain the efficiencies and cost protection associated with the standard low-cost model if they so choose.
Employers Protected from any Risk of Fiduciary Liability
Employers traditionally have been particularly concerned about the risk of fiduciary liability associated with their selection of retirement plan investments. This concern extends to the employer's designation of default investments that employees are free to decline in favor of alternative investments. In the IRA universe, employers transferring funds to automatic rollover IRAs and employer-sponsored SIMPLE-IRAs retain a measure of fiduciary responsibility for initial investments.
By contrast, under our proposal, employers making direct deposits would be insulated from such potential liability. These employers would have no liability or fiduciary responsibility with respect to the manner in which direct deposits are invested in default IRAs or in nondefault IRAs (whether selected by the employer or the employee), nor would employers be exposed to potential liability with respect to any employee's choice of IRA provider or type of IRA. This protection of employers is facilitated by statutory designation of standard investment types that reduces the need for continuous professional investment advice. To protect workers against inappropriate IRA providers or inappropriate employer selection of IRA providers while continuing to insulate employers from fiduciary responsibility, employers could be precluded from imposing a particular IRA provider on its employees other than the government-contracted default IRA or could be constrained to choose among an approved list of providers based on capital adequacy, soundness, and other criteria.
Public Opinion Polling
Recent public opinion polling has shown overwhelming support for payroll-deduction direct deposit saving. Among registered voters surveyed, 83 percent of respondents said they would be agreeable to having their employer offer to sign them up for an IRA and allow them to contribute to it through direct deposit of a small amount from their paycheck to help them save for retirement. Similarly, 79 percent of registered voters expressed support (and 54 percent expressed "strong" support) for giving taxpayers the option to have part of their income tax refund deposited into a retirement savings account such as an IRA by just checking a box on their tax return.
In addition, the polling shows very strong support for a requirement that goes far beyond our proposal, that every company offer its employees some kind of retirement plan-such as a pension or 401(k), or at least an IRA to which employees could contribute. Among registered voters surveyed in August 2005, 77 percent supported such a requirement (and 59 percent responded that they were "strongly" in support). As discussed, the approach described in this paper would not require employers to offer their employees retirement plans, but would give firms a financial incentive to offer their employees access to payroll deduction as a convenient and easy means of saving, and would require firms above a certain size and maturity to extend this offer to their employees.
The Importance of Protecting Employer Plans
Employer-sponsored pension, profit-sharing, 401(k), and other plans can be particularly effective - more so than IRAs -- in accumulating benefits for employees. As noted earlier, the participation rate in 401(k)s, for example, tends to range from two thirds to three quarters of eligible employees, in contrast to IRAs, in which fewer than 1 in 10 eligible individuals participates. Employer plans tend to be far more effective than IRAs at providing coverage because of a number of attributes: for one thing, pension and profit-sharing plans, for example, are funded by employer contributions that automatically are made for the benefit of eligible employees without requiring the employee to take any initiative in order to participate. Second, essentially all tax-qualified employer plans must abide by standards that either seek to require reasonably proportionate coverage of rank-and-file workers or give the employer a distinct incentive to encourage widespread participation by employees. This encouragement typically takes the form of both employer-provided retirement savings education efforts and employer matching contributions. The result is that the naturally eager savers, who tend to be in the higher tax brackets, tend to subsidize or bring along the naturally reluctant savers, who often are in the lowest (including zero) tax brackets.
Employer-sponsored retirement plans also have other features that tend to make them effective in providing or promoting coverage. As noted, the proposal outlined here seeks to transplant some of these features to the IRA universe. These include the automatic availability of a saving vehicle, the use of payroll deduction (which continues automatically once initiated), matching contributions (further discussed below), professional investment management, and peer group reinforcement of saving behavior.
The automatic IRA must thus be designed carefully to avoid competing with or crowding out employer plans and to avoid encouraging firms to drop or reduce the employer contributions that many make to plan participants. Owners and others who control the decision whether to adopt or continue maintaining a retirement plan for employees should continue to have incentives to sponsor such plans. The ability to offer employees direct deposit to IRAs should be designed so that it will not prompt employers to drop, curtail, or refrain from adopting retirement plans.
Probably the single most important protection for employer plans is to set maximum permitted contribution levels to the automatic IRA so that they will be sufficient to meet the demand for savings by most households but not high enough to satisfy the appetite for tax-favored saving of business owners or decision-makers. The average annual contribution to a 401(k) plan by a nonhighly compensated employee is somewhat greater than $2,000, and average annual 401(k) contributions by employees generally tend to be on the order of 7 percent of pay. A $3,000 contribution is 7.5 percent of pay for a family earning $40,000, and 6 percent of pay for a family earning $50,000.
Yet IRA contribution limits are already higher than these contribution levels. IRAs currently allow a married couple to contribute up to $8,000 ($4,000 each) on a tax-favored basis, and an additional $1,000 ($500 each) if they are age 50 or older. By 2008, these figures are scheduled to rise to $10,000 plus $2,000 ($1,000 each) for those age 50 or older. These amounts-the current $9,000 a year for those age 50 and over ($8,000 for others) and the post-2007 $12,000 annual amount for those age 50 and over ($10,000 for others)-may well be enough to satisfy the desire of many small-business owners for tax-favored retirement savings. Even some small-business owners that might consider saving somewhat more than $10,000 or $12,000 per year might well conclude that they are better off not incurring the cost of making contributions and providing a plan for their employees because the net benefit to them of having a plan for employees is not greater than the net benefit of simply saving through IRAs and giving their employees access to IRAs.
Accordingly, at the most, payroll deposit IRAs should not permit contributions above the current IRA dollar limits, and could be limited to a lower amount such as $3,000. (A 3% of pay contribution would remain below $3,000 for employees whose compensation did not exceed $100,000.) Imposing a lower limit on the payroll deduction IRA would reduce to some degree the risk that employees will exceed the maximum IRA dollar contribution limit because of auto enrollment, combined with possible other contributions to an IRA. That is already a risk under current law, but the automatic nature of auto enrollment increases the risk, especially if auto escalation is implemented. There is a tradeoff between the desirability of limiting the contribution amount (to mitigate both this risk and the risk of competing with employer plans) and the simplicity of using an existing vehicle (the IRA) "as is".
In any event, the employee - not the employer - would be responsible for monitoring any of all of their IRA contributions to comply with the maximum limit (in part because employees can contribute on their own and through multiple employers). The ultimate reconciliation would be made by the individual when filing the federal income tax return.
In addition, the automatic IRA should be designed to avoid reducing ordinary employees' incentives to contribute to employer-sponsored plans such as 401(k)s. If workers perceive a program such as direct deposit savings to IRAs as a more attractive destination for their contributions than an employer-sponsored plan (for example, because of better matching, tax treatment, investment options, or liquidity), it could unfortunately divert employee contributions from employer plans. This in turn could have a destabilizing effect by making it difficult for employers to meet the nondiscrimination standards applicable to 401(k)s and other plans and therefore potentially discouraging employers from continuing the plans or their contributions. While a detailed discussion of these points is beyond the scope of this paper, it is important to maintain a relationship between IRAs and employer-sponsored retirement plans that preserves and protects the employer plans.
Automatic Payroll Deduction Can Promote Marketing and Adoption of Employer Plans
Our approach is designed not only to avoid causing any reduction or contraction of employer plans, but actually to promote expansion of employer plans. Consultants, third-party administrators, financial institutions, and other plan providers could be expected to view this proposal as providing a valuable new opportunity to market 401(k)s, SIMPLE-IRAs and other tax-favored retirement plans to employers. Firms that, under this proposal, were about to begin offering their employees payroll deduction saving or had been offering their employees payroll deduction saving for a year or two could be encouraged to "trade up" to an actual plan such as a 401(k) or SIMPLE-IRA.
Especially because these plans can now be purchased at very low cost, it would seem natural for many small businesses to graduate from payroll deduction savings and complete the journey to a qualified plan in order to obtain the added benefits in terms of recruitment, employee relations, and larger tax-favored saving opportunities for owners and managers.
The following compares the maximum annual tax-favored contribution levels for IRAs, SIMPLE-IRA plans and 401(k) plans in effect for 2006:
Under age 50
$4,000 per spouse ($5,000 after 2007)
Age 50 and above
$4,500 per spouse ($6,000 after 2007)
In addition, as noted, small employers that adopt a new plan for the first time are entitled to a tax credit of up to $500 each year for three years. As discussed, the proposed tax credit for offering payroll deposit would be smaller, so as to maintain the incentive for employers to go beyond the payroll deduction or direct deposit IRA and adopt an actual plan such as a SIMPLE, 401(k), or other employer plan.
Encouraging Contributions by Nonemployees
The payroll deposit system outlined thus far would not automatically cover self-employed individuals, employees of the smallest or newest businesses that are exempt from any payroll deposit obligation, or certain unemployed individuals who can save. A strategy centered on automatic arrangements can also make it easier for these people to contribute to IRAs.
Encouraging Automatic Debit Arrangements
For individuals who are not employees or who otherwise lack access to payroll deduction, automatic debit arrangements can serve as a counterpart to automatic payroll deduction. Automatic debit enables individuals to spread payments out over time and to make payments on a regular and timely basis by having them automatically charged to and deducted from an account-such as a checking or savings account or credit card-at regular intervals on a set schedule. The individual generally gives advance authorization to the payer that manages the account or the recipient of the payment, or both. The key is that, as in the case of payroll deduction, once the initial authorization has been given, regular payments continue without requiring further initiative on the part of the individual. For many consumers, automatic debit is a convenient way to pay bills or make payments on mortgages or other loans without having to remember to make each payment when due and without having to write and mail checks.
Similarly, as an element of an automatic IRA strategy, automatic debit can facilitate saving while reducing paperwork and cutting costs. For example, households can be encouraged to sign up on-line for regular automatic debits to a checking account or credit card that are directed to an IRA or other saving vehicle. With on-line sign-up and monitoring, steps can be taken to familiarize more households with automatic debit arrangements and, via Internet websites and otherwise, to make those arrangements easier to set up and use as a mechanism for saving in IRAs.
Facilitating Automatic Debit IRAs Through Professional or Trade Associations
Professional and trade associations could facilitate the establishment of IRAs and the use of automatic debit and direct deposit to the IRAs. Independent contractors and other individuals who do not have an employer often belong to such an association. The association, for example, might be able to make saving easier for those members who wish to save by making available convenient arrangements for automatic debit of members' accounts. Association websites can make it easy for members to sign up on line, monitor the automatic debit savings, and make changes promptly when they wish to. Although such associations generally lack the payroll-deduction mechanism that is available to employers, they can help their members set up a pipeline involving regular automatic deposits (online or by traditional means) from their personal bank or other financial accounts to an IRA established for them.
Facilitating Direct Deposit of Income Tax Refunds to IRAs
Another major element of a strategy to encourage contributions outside of employment would be to allow taxpayers to deposit a portion of their income tax refunds directly into an IRA by simply checking a box on their tax returns. Currently, the IRS allows direct deposits of refunds to be made to only one account. This all-or-nothing approach discourages many households from saving any of the refund because at least a portion of the refund is often needed for immediate expenses. Allowing households instead to split their refunds to deposit a portion directly into an IRA could make saving simpler and, thus, more likely.
The Bush administration has supported divisible refunds in its last three budget documents; however, the necessary administrative changes have yet to be implemented. Since federal income tax refunds total nearly $230 billion a year (more than twice the estimated annual aggregate amount of net personal savings in the United States), even a modest increase in the proportion of refunds saved every year could bring about a significant increase in savings.
Extending Direct Deposit to Independent Contractors
Millions of Americans are self-employed as independent contractors. Many of these workers receive regular payments from firms, but because they are not employees, they are not subject to income tax or payroll tax withholding. These individuals might be included in the direct deposit system by giving them the right to request that the firm receiving their services direct deposit into an IRA a specified portion from the compensation that would otherwise be paid to them.
Compared to writing a large check to an IRA once a year, this approach has several potential advantages to independent contractors, which might well encourage them to save. These include the ability to commit themselves to save a portion of their compensation before they receive it (which, for some people, makes the decision to defer consumption easier); the ability to avoid having to make an affirmative choice among various IRA providers; remittance of the funds by the firm by direct deposit to the IRA; and, where payments are made to the independent contractor on a regular basis, an arrangement that, like regular payroll withholdings for employees, automatically continues the pattern of saving through repeated automatic payroll deductions unless and until the individual elects to change.
In many cases, the independent service provider will not have a sufficient connection to a firm that receives the services, or both the independent contractor and the firm will be unwilling to enter into a payroll deposit type of arrangement. In such instances, the independent contractor could contribute to an IRA using automatic debit (as discussed above) or by sending together with the estimated taxes that generally are due four times a year.
Matching Deposits as a Financial Incentive
A powerful financial incentive for direct deposit saving by those who are not in the higher tax brackets (and who therefore derive little benefit from a tax deduction or exclusion) would be a matching deposit to their direct deposit IRA. One means of delivering such a matching deposit would be via the bank, mutual fund, insurance carrier, brokerage firm, or other financial institution that provides the direct deposit IRA. For example, the first $500 contributed to an IRA by an individual who is eligible to make deductible contributions to an IRA might be matched by the private IRA provider on a dollar-for-dollar basis, and the next $1,000 of contributions might be matched at the rate of 50 cents on the dollar. The financial provider would be reimbursed for its matching contributions through federal income tax credits.
Recent evidence from a randomized experiment involving matched contributions to IRAs suggests that a simple matching deposit to an IRA can make individuals significantly more likely to contribute and more likely to contribute larger amounts.
Matching contributions-similar to those provided by most 401(k) plan sponsors-not only would help induce individuals to contribute directly from their own pay, but also, if the match were automatically deposited in the IRA, would add to the amount saved in the IRA. The use of matching deposits, however, would make it necessary to implement procedures designed to prevent gaming-contributing to induce the matching deposit, then quickly withdrawing those contributions to retain the use of those funds. Among the possible approaches would be to place matching deposits in a separate subaccount subject to tight withdrawal rules and to impose a financial penalty on early withdrawals of matched contributions.
American households have a compelling need to increase their personal saving, especially for long-term needs such as retirement. This paper proposes a strategy that would seek to make saving more automatic-hence easier, more convenient, and more likely to occur-largely by adapting to the IRA universe practices and arrangements that have proven successful in promoting 401(k) participation. In our view, the automatic IRA approach outlined here holds considerable promise of expanding retirement savings for millions of workers.
The Heritage Foundation is a public policy, research, and educational organization recognized as exempt under section 501(c)(3) of the Internal Revenue Code. It is privately supported and receives no funds from any government at any level, nor does it perform any government or other contract work.
The Heritage Foundation is the most broadly supported think tank in the United States. During 2013, it had nearly 600,000 individual, foundation, and corporate supporters representing every state in the U.S. Its 2013 income came from the following sources:
The top five corporate givers provided The Heritage Foundation with 2% of its 2013 income. The Heritage Foundation’s books are audited annually by the national accounting firm of McGladrey, LLP.
Members of The Heritage Foundation staff testify as individuals discussing their own independent research. The views expressed are their own and do not reflect an institutional position for The Heritage Foundation or its board of trustees.
David John is a Senior Research Fellow for Retirement Security and Financial Institutions at the Thomas A. Roe Institute for Economic Policy Studies at The Heritage Foundation. Mark Iwry is Senior Advisor to The Retirement Security Project, a Nonresident Senior Fellow at the Brookings Institution, Research Professor at Georgetown University, and formerly the Benefits Tax Counsel, in charge of national private pension policy and regulation, at the U.S. Department of the Treasury.
The Retirement Security Project is supported by The Pew Charitable Trusts in partnership with Georgetown University's Public Policy Institute and the Brookings Institution.
The views expressed in this testimony are those of the two witnesses alone and should not be attributed to The Heritage Foundation, the Brookings Institution, Georgetown University's Public Policy Institute, or The Pew Charitable Trusts.
 We have previously written and testified before Congress on various aspects of employer-sponsored retirement plans. David John has written and testified about the funding problems faced by defined benefit pension plans and about the United Kingdom's pension situation. Mark Iwry led the Executive Branch efforts in the 1990s to develop the SIMPLE plan for small business, the startup tax credit for small employers that adopt new plans, and the saver's credit for moderate- and lower-income workers, as well as the Executive Branch initiatives to define, approve and promote 401(k) automatic enrollment, automatic rollover to restrict pension leakage, and automatic 401(k) features generally. See also William G. Gale, J. Mark Iwry and Peter R. Orszag, "The Saver's Credit" (The Retirement Security Project, Policy Brief No. 2005-2; available at http://www.retirementsecurityproject.org).
 William G. Gale, J. Mark Iwry and Peter R. Orszag, "The Automatic 401(k): A Simple Way to Strengthen Retirement Savings," (The Retirement Security Project, Policy Brief No. 2005-1; available at http://www.retirementsecurityproject.org); William G. Gale and J. Mark Iwry, "Automatic Investment: Improving 401(k) Portfolio Investment Choices" (The Retirement Security Project, Policy Brief No. 2005-4; available at http://www.retirementsecurityproject.org).
1.This testimony does not address any issues relating to Social Security reform. The proposal is intended to have no implications, one way or the other, regarding proposals to finance individual accounts using Social Security taxes or to offset Social Security benefits by individual accounts. Also outside the scope of this testimony are potential reforms to the private pension system (including employer-sponsored defined contribution and defined benefit plans).
2.This testimony is intended only to outline the proposal, not to resolve all of the specific but significant design and implementation issues that cannot readily be addressed within the limited scope of this testimony.
3.Craig Copeland, "Employer-Based Retirement Plan Participation: Geographic Differences and Trends: Employee Benefit Research Institute Issue Brief No. 286," October 2005 (referred to below as "Copeland, EBRI Issue Brief No. 286"), Figure 1, p. 7. The nonparticipants include those who are not eligible for their employer's plan as well as those who are eligible but who fail to participate. Among the subset of approximately 92 million full-time, full-year wage and salary workers between the ages of 21 and 64, 65 percent work for an employer that sponsors a plan, and 57 percent participate in an employer-sponsored plan. Id.
4.Copeland, EBRI Issue Brief No. 286, Figure 1, p. 7.
5.See, for example, Alicia H. Munnell and Annika Sunden, Coming Up Short: The Challenge of 401(k) Plans(Brookings Institution Press, 2004).
6.In the Conference Report to the Tax Reform Act of 1997, Congress stated that "employers that choose not to sponsor a retirement plan should be encouraged to set up a payroll deduction [IRA] system to help employees save for retirement by making payroll-deduction contributions to their IRAs" and encouraged the Secretary of the Treasury to "continue his efforts to publicize the availability of these payroll deduction IRAs" (H.R. Rep. No. 220, 105th Cong., 1st Sess. 775 ).
7.Department of Labor Interpretive Bulletin 99-1 (June 18, 1999), 29 C.F.R. 2509.99-1(b); IRS Announcement 99-2.
8.Neither the IRS nor the Department of Labor guidance addressed the possible use of automatic enrollment in conjunction with direct deposit IRAs (discussed at length below).
9. William G. Gale, J. Mark Iwry, and Peter R. Orszag, "The Automatic 401(k): A Simple Way to Strengthen Retirement Savings," (The Retirement Security Project, Policy Brief No. 2005-1; available athttp://www.retirementsecurityproject.org);
10.In 2004, the IRS affirmed that plans are permitted to increase the automatic contribution rate over time in accordance with a specified schedule or in connection with salary increases or bonuses. See letter dated March 17, 2004, from the Internal Revenue Service to J. Mark Iwry. The idea of coordinating automatic contribution increases with pay increases was developed by Richard Thaler and Shlomo Benartzi. See Thaler and Benartzi, "Save More Tomorrow: Using Behavioral Economics to Increase Employee Saving," Journal of Political Economy 112, no. 1, pt.2, pp. S164-87.
11.Brigitte Madrian and Dennis Shea, "The Power of Suggestion: Inertia in 401(k) Participation and Savings Behavior," Quarterly Journal of Economics 116, no. 4 (November 2001): 1149-87; and James Choi and others, "Defined Contribution Pensions: Plan Rules, Participant Decisions, and the Path of Least Resistance," in Tax Policy and the Economy, vol. 16, edited by James Poterba (Cambridge, Mass.: MIT Press, 2002), pp. 67-113. See also Sarah Holden and Jack VanDerhei, "The Influence of Automatic Enrollment, Catch-Up, and IRA Contributions on 401(k) Accumulations at Retirement," Employee Benefit Research Institute Issue Brief No. 283 (July 2005).
12.Any such statutory provision could usefully make clear that automatic enrollment in direct deposit IRAs is permitted irrespective of any state payroll laws that prohibit deductions from employee paychecks without the employee's advance written approval. Assuming that most direct deposit IRA arrangements are not employer plans governed by ERISA, such state laws, as they apply to automatic IRAs, may not be preempted by ERISA because they do not "relate to any employee benefit plan."
13.The absence of an employer match might make some employers more willing to offer auto enrollment on direct deposit IRAs because increased participation would not come at the cost of increased employer matching contributions. On the other hand, the absence of the match tends to make participation in the plan less attractive to workers, which could exacerbate employee concerns or complaints about having been enrolled in a program that reduces their take-home pay without their explicit prior written authorization. As a result, the absence of a match might also make employers more apprehensive about possible complaints from employees who failed to read the auto enrollment notice.
14. Between August 28 and 31, 2005, in a survey commissioned by The Retirement Security Project, The Tarrance Group, in conjunction with Lake, Snell, Mermin/Decision Research, interviewed 1,000 registered voters nationwide about retirement security issues. A full report of the survey findings can be found at http://www.retirementsecurityproject.org.
15.James Choi, David Laibson, Brigitte Madrian, and Andrew Metrick, "Active Decisions" NBER Working Paper No. 11074 (January 2005).
16.Employers that sponsor a SIMPLE-IRA plan may deposit all employee contributions in IRAs at a single designated financial institution selected by the employer (IRS Notice 98-4, 1998-2 I.R.B. 25).
17.Considerable challenges are involved in building and implementing a workable universal saving system based on employer direct deposits of contributions to IRAs. These challenges include dealing with the contingent workforce, with employees who have multiple jobs, who work part-time, and often who earn relatively low wages, and with small employers. A somewhat different and thoughtful approach to designing such a system can be found in the evolving work of the Conversation on Coverage, a collaborative effort among individuals (including one of the authors) drawn from a diverse range of stakeholder organizations. See Conversation on Coverage, "Covering the Uncovered," Report of Working Group II (2005). For a recently published analysis by a non-partisan expert panel (including one of the authors) of the issues involved in designing arrangements for distributions from individual accounts, see National Academy of Social Insurance, Uncharted Waters: Paying Benefits from Individual Accounts in Federal Retirement Policy (2005). There have been various other efforts to design such systems or programs, which this testimony does not attempt to catalogue.
18.Until recently the federal Thrift Savings Plan had five investment funds: three stock index funds (S&P 500, small and midcapitalization U.S. stocks, and mostly large-capitalization foreign stocks), a bond index fund consisting of a mix of government and corporate bonds, and a fund consisting of short-term, nonmarketable U.S. Treasury securities. Effective August 1, 2005, the Plan added a set of life-cycle funds, each one of which is composed of a mix of the other five investment funds.19. This was part of the impetus behind the 2001 statutory provision to the effect that the Secretaries of Labor and Treasury may provide, and shall give consideration to providing, special relief with respect to the use of low-cost individual retirement plans for purposes of automatic rollovers and for other uses that promote the preservation of assets for retirement income (Economic Growth and Tax Relief Reconciliation Act of 2001, Public Law 107-16, 115 Stat. 38, Section 657[c][B]). In a similar vein, one of the co-authors has proposed a strategy for States to act as a catalyst in expanding coverage under standardized, low-cost payroll-deposit IRAs, SIMPLE-IRA plans, and 401(k) plans by facilitating the pooling of small businesses to offer these vehicles. The proposal has been outlined in "Expanding Retirement Savings at the State Level," Written Statement of J. Mark Iwry to the Legislature of the State of Washington (April 2003), and is more fully described in a separate written statement by Iwry, separately submitted for the record, and scheduled to be published in the NYU Review of Employee Benefits and Executive Compensation 2006 and the BNA Tax Management Compensation Planning Journal.
20.The difference in expense between passively managed index funds and actively managed mutual funds has been estimated to be-as a broad generalization-roughly 100 basis points (1 percent) a year (William F. Sharpe, "Indexed Investing: A Prosaic Way to Beat the Average Investor" presented at the Spring President's Forum, Monterey Institute of International Studies (May 2002).
21.One of the authors has testified before Congress regarding the British retirement plan system and has been critical of the UK's attempt to impose a limit on charges. See David C. John, testimony before the Subcommittee on Social Security of the Committee on Ways and Means, U.S. House of Representatives (June 16, 2005); David C. John, "What the United States Can Learn from the UK's Pensions Commission Report" (forthcoming).
22.As noted, the federal Thrift Savings Plan consists mainly of index funds, which are the building blocks for the recently added life-cycle funds. The Thrift Savings Plan informational materials state that the life-cycle funds " provide a way to diversify your account optimally, based on professionally determined asset allocations. This provides you with the opportunity to achieve a maximum amount of return over a given period of time with a minimum amount of risk. . . " (Federal Thrift Savings Plan website, www.tsp.gov). To the extent that a professionally run "managed account" could achieve similar results at no greater cost, that might be another attractive option, and managed accounts are growing in popularity as an option in 401(k) plans. A question may be raised as to whether, managed accounts are a better fit for 401(k) plans than for automatic IRAs, because 401(k)s tend to have more substantial account balances and greater flexibility to accommodate individual preferences while allocating costs to individuals who opt for costlier alternatives.
23.The question of how best to fit the direct deposit IRAs, with their improved and simplified investment structure, into the larger IRA universe is related to a broader issue: the potential simplification of IRAs. We favor simplification and revision of the current array of IRA options. However, the specifics of any such proposals are beyond the scope of this testimony.
24.The retirement security poll referred to in note 14, above, had a margin of error of 3.1 percent. The question that elicited these results was as follows: "Would you support or oppose a requirement that every company offer their employees some sort of retirement plan-either a traditional pension, a 401(k) or an IRA that the employer sets up but does not contribute to. The company would choose which one they wanted to offer employees. Would you support or oppose requiring every employer to give employees at least one of these options?" A full report of the survey findings can be found at www.retirementsecurityproject.org.
25.See Craig Copeland, "Retirement Plan Participation and Retirees' Perception of Their Standard of Living," Employee Benefit Research Institute Issue Brief No. 289 (January, 2006), pp. 1-6, Figure A4.
26.It is conceivable that the risk of exceeding the IRA dollar limit could be mitigated to some degree through enrollment procedures that cap automatic enrollment at, say, $250 a month (for an annual total of $3,000) or $300 a month. However, because automatic enrollment would be administered at the employer level and might be based on paychecks provided weekly or every two weeks, the maximum dollar amount would need to be adjusted accordingly (e.g., $60 if weekly, $120 if every two weeks, or $250 if monthly).
27.J. Mark Iwry, "Using Tax Refunds to Increase Savings and Retirement Security" (Retirement Security Project, Policy Brief No. 2006-1, Jan. 2006; available at www.retirementsecurityproject.org).
28.Among the issues such an approach would need to address is the means of reimbursing those private financial institutions that have no federal income tax liability to offset because they are tax exempt or in a loss position.
An alternative mechanism would modify the existing saver's credit (a federal income tax credit to households with income below $50,000 for contributing to an IRA or employer plan) to convert it to a direct matching deposit to an IRA or other savings account. (As currently structured, the saver's credit reduces the household's federal income tax liability and is nonrefundable; thus, it is not automatically saved.) A variation would be to have such a direct matching deposit delivered by the financial institution that sponsors the IRAs or serves as financial provider to the 401(k) plan to which the individual contributes. One of the authors was involved in developing the Saver's Credit and, in congressional testimony and writings, has advocated its extension and expansion. See, e.g., William G. Gale, J. Mark Iwry, and Peter R. Orszag, "The Saver's Credit: Expanding Retirement Savings for Middle- and Lower-Income Americans" (Retirement Security Project Policy Brief No. 2005-2, March 2005; available at www.retirementsecurityproject.org). However, issues relating to the Saver's Credit and its potential expansion are beyond the scope of this testimony. Another significant asset-building approach targeted to lower- and moderate-income households is reflected in the Individual Development Accounts (IDAs). See, e.g., Michael Sherraden, Assets and the Poor: A New American Welfare Policy (M. E. Shapre, 1992), and Ray Boshara, "Individual Development Accounts: Policies to Build Savings and Assets for the Poor" (Brookings, Policy Brief, March 2005).
29. Esther Duflo, William Gale, Jeffrey Liebman, Peter Orszag, and Emmanuel Saez, "Saving Incentives for Low- and Middle-Income Families: Evidence from a Field Experiment with H&R Block" (Retirement Security Project, May 2005; available at http://www.retirementsecurityproject.org).
30.A detailed treatment of the matching deposit option is beyond the scope of this testimony.