TPC Tax Topics

E-mail Newsletter

Enter your e-mail address to receive periodic updates on TPC publications and events.

Individual retirement accounts

Originally published in the NTA Encyclopedia of Taxation and Tax Policy, Second Edition, edited by Joseph J. Cordes, Robert D. Ebel, and Jane G. Gravelle. The encyclopedia is available in paperback from the Urban Institute Press. Order online at www.uipress.org or call toll-free 1-877-847-7377

Paul Burnham and Larry Ozanne
Congressional Budget Office

Savings accounts with specific income tax incentives and penalties designed to encourage working people to save for retirement.

Individual Retirement Accounts (IRAs) were first authorized in the Employee Retirement Income Security Act of 1974 (ERISA) as a means of encouraging working people to save for retirement. Recently, preretirement withdrawals from IRAs for major medical expenses, higher-education costs, and a firsttime home purchase have been authorized. Unlike 401(k) and related salary reduction plans, IRAs are not run by employers. The enactment of IRAs extended to workers without pensions the same kind of tax advantages already granted to pension funds and the self-employed.

History

Starting in 1975, individuals were allowed to set up separate accounts at financial institutions and deduct the value of their contributions from their current taxable income. Only employees without pensions were eligible to contribute, and their annual contributions were limited to 15 percent of pay or a maximum of $1,500. The investment returns of these accounts were also excluded from taxable income in the year earned, but withdrawals were to be included in taxable income in the year they occurred. To encourage use of the accounts for retirement saving, ERISA set a penalty of 10 percent additional tax on withdrawals by taxpayers before age 59 1/2.

The Economic Recovery Tax Act of 1981 expanded IRA eligibility and increased maximum contributions. Starting in 1982, all persons with earnings could contribute to IRAs, whether or not they were in a pension plan, and the maximum contribution was increased to 100 percent of earnings or $2,000. In response, tax returns with IRA contributions jumped from 4 percent of all returns with wage and salary income in 1981 to 14 percent in 1982 and 18 percent in 1986. Deductible IRA contributions were restricted by the Tax Reform Act of 1986. Starting in 1987, the act allowed deductible contributions only by an individual who was not covered by an employer pension plan (and whose spouse was not covered), and who had adjusted gross income between $25,000 and $35,000 (or, for joint returns, $40,000 to $50,000). Those not qualifying for deductible contributions could still make nondeductible contributions, thereby benefiting from the exclusion of investment returns from taxable income. The restriction caused tax returns with deductible IRA contributions to drop to 8 percent of all returns in 1987 and to decline slowly from there, reaching 4 percent in 1997.

The Taxpayer Relief Act of 1997 substantially raised the income limits applying to taxpayers covered by an employer plan. Eventually, those limits will be $50,000 to $60,000 for single taxpayers (2005 and after) and $80,000 to $100,000 for married taxpayers (2007 and after). Even higher limits ($150,000 to $160,000) were enacted for married taxpayers previously disqualified from making deductible contributions solely by virtue of a spouse being covered by an employer plan. The act also allows individuals to elect back-loaded IRAs, called "Roth" IRAs, in which contributions are not deductible but withdrawals are not taxed-a treatment similar to that of a tax-exempt bond. These IRAs are phased out between $95,000 and $110,000 of income for singles, and $150,000 to $160,000 for joint returns. All but the highestincome taxpayers will thus be eligible for some type of fully tax-favored IRA. Regulations imposing penalties on premature withdrawals were also relaxed, so that the accounts may be used to save for higher education expenses or the first purchase of a home.

The Economic Growth and Tax Relief Reconciliation Act of 2001 (EGTRRA) increased the maximum allowable contribution to both deductible IRAs and Roth IRAs. For taxpayers under age 50, the limit will reach $5,000 in 2008 and then will be indexed for inflation. Taxpayers age 50 and above are allowed to make additional contributions up to a limit that will reach $1,000 in 2006. Like all provisions of EGTRRA, the higher limits are scheduled to expire in 2011 and revert to $2,000. EGTRRA also provides lower-income taxpayers with a nonrefundable credit of up to 50 percent of IRA contributions up to $2,000 through 2006.

Effect on saving

IRAs have been popular among taxpayers. For example, during the years in which all taxpayers were eligible to contribute to IRAs, up to $40 billion of savings was deposited into IRA accounts annually. The large volume of IRA contributions, however, does not necessarily indicate that IRAs increased saving. IRAs increased personal saving only to the extent that people funded their contributions by reducing their consumption. Contributions funded by saving less in other accounts and by paying fewer taxes do not raise saving.

In principle, the incentive structure of IRAs could increase or decrease saving. The deferral of taxation on contributions and investment earnings until withdrawal increases the rate of return to saving. Because a higher return to saving has offsetting substitution and income effects, the net effect could be to increase or decrease saving. In addition, the limit on contributions means that persons able to contribute more than the limit without increasing their saving have only an income effect, which could reduce saving. Finally, features of the IRA and its marketing by financial institutions may have increased saving by making it easier for people to save. The heavy advertising of IRAs between 1982 and 1986 may have helped people focus on the importance of saving for retirement, and the deduction on contributions and the penalty on withdrawals may have helped people make and preserve their contributions.

Empirical studies have not been able to resolve the uncertainty about how IRAs affect saving, although many attempts have been made. The evidence for the full population is contradictory, but a limited consensus suggests that IRAs increased saving for nonelderly and less-wealthy families. The range of results is indicated by three pairs of analysts who have each measured saving for samples of families, estimated models of saving behavior, and used those models to predict how IRAs affected saving between 1982 and 1986. One pair, Venti and Wise (1986, 1990, 1991), estimate in three separate studies that if the contribution limit had been raised above $2,000, between 57 and 66 percent of additional IRA contributions would have come from reduced consumption. A second pair, Gale and Scholz (1994), estimate that the level would be at most 2 percent, and that IRAs possibly increased consumption. The third pair, Joines and Manegold (1991), find that the expansion of eligibility in 1982 had no statistically significant effect on consumption for the full population.

Although Gale and Scholz and Joines and Manegold find no increase in saving for the full population, each pair finds an increase for a subpopulation. Gale and Scholz find that people under age 60, who could not withdraw from IRAs without paying the penalty, reduced their consumption by up to 35 cents per dollar contributed. Joines and Manegold find that people who had less than $25,000 of taxable financial assets in 1982 reduced their consumption by 63 cents per dollar contributed. Because Venti and Wise find large increases in saving for the full population, all three studies present evidence that IRAs increased saving for sizable subpopulations. Studies finding an increase in saving for subpopulations but not for the full population indicate either that the increases found for some subpopulations were small relative to saving for the full population or that other subpopulations reduced their saving.

More recently, Attanasio and DeLeire (2002) found little evidence of an IRA savings effect, using a methodology similar to Joines and Manegold.

ADDITIONAL READINGS