Tax Policy and Retirement Savings

Key Points

  • Tax subsidies for retirement saving cost $180 billion in 2016 and are one of the largest tax sources of revenue loss for the government.
  • Evidence based on administrative data finds that tax incentives only induce a minority of households to save more. So-called “nudges” might be just as or more important.
  • In 2017, 30 states are exploring different types of retirement savings reforms. State reforms may help inform a national policy to increase household saving.

Tax Policy and Retirement Savings

Editor’s Note: This article is part of a series of tax-related articles sponsored by the Penn Wharton Budget Model and the Robert D. Burch Center at Berkeley. All of the articles in this series are forthcoming in a book by Oxford University Press, co-edited by Alan Auerbach and Kent Smetters.

Tax Policy and Retirement Savings examines tax incentives for retirement savings. Advances in behavioral economics and access to high-quality administrative data has allowed researchers to improve their understanding of the effectiveness of tax incentives for retirement saving. John N. Friedman (2016) draws on evidence from empirical studies to examine new policy options to increase household saving and retirement preparedness.

Retirement saving incentives reduced government revenue by $180 billion in 2016, or just over 5 percent of total federal tax revenue. Saving for retirement increases total saving and investment, which can lead to economic growth. Household saving has fallen over the past four decades. Figure 1 shows that household saving dropped to 5.8 percent in 2015 from 13 percent in 1975. Government incentives to support retirement saving stem from the concern that people save too little for retirement. People might save too little if they suffer from “present bias.” Or, people might rationally save too little because they know that the government will provide for them in their old age, a phenomenon that economics Nobel laureate James Buchanan termed “the Samaritan’s Dilemma.” Researchers find that between about 30 percent and half of households save too little.

Figure 1: Personal Savings Rate as a Percentage of Disposable Personal Income in the U.S.

Source: U.S. Bureau of Economic Analysis, Personal Saving Rate [PSAVERT], retrieved from FRED, Federal Reserve Bank of St. Louis;, February 14, 2017.

Types of Tax Subsidies for Retirement Savings

Households can reduce their tax bill by saving for retirement in qualified retirement accounts. Contributions to traditional IRAs and employer-sponsored plans (traditional pension and 401(k)-style) are not subject to income tax in the year in which they are made. Instead, taxes are deferred until people receive benefits or withdraw money. Roth IRAs and employer-sponsored Roth plans work differently. Contributions are taxed in the year they are made but no tax is paid when people withdraw money from the account when they are retired.

In 2017, the maximum employee contribution to a 401(k) plan is $18,000. When employer contributions are included, total contributions cannot exceed $54,000. Eligible people can contribute up to $5,500 to an IRA. Other types of employer-sponsored plans have different limits. In addition, people aged 50 and older can “catch-up” with higher limits on contributions. On September 30, 2017, households held $25 trillion in retirement savings.

The Saver’s Credit is an additional tax benefit for retirement saving for people with low incomes. In 2017, single taxpayers with income of $31,000 or below ($62,000 for married filers) can subtract up to $1,000 ($2,000 for married filers with double incomes) from their federal tax bill. Saver’s Credit is calculated as a percent (higher for those with lower incomes) of each worker’s first $2,000 of contributions. Households that do not owe federal taxes cannot use the credit because it is non-refundable. In fact, few households make use of the credit. One study found that only 15 to 20 percent of eligible households made contributions to a retirement account.

Effectiveness of Tax Subsidies for Retirement Savings

The impact of tax incentives on total saving and on retirement preparedness will depend on how many people respond to the incentive and the strength of the response.

Friedman finds that tax subsidies for retirement saving likely have a limited effect on how much households save. Some early studies on the growth of IRAs and 401(k)s found that contributions increased total saving. However, a recent study that makes use of administrative data from Denmark finds that only a small portion of contributions to retirement accounts increase total saving. Instead, savings that would have otherwise been made to a taxable account are shifted to a non-taxable retirement account. The same study finds that fewer than 20 percent of people respond to changes to tax incentives by changing their saving. Wealthy and educated people were more likely to change their saving than others.

Tax incentives may help people save enough for retirement. About 80 percent of households have retirement savings by the time they near retirement age. However, many of those households may not have enough. Friedman estimates that changing tax incentives may not significantly change saving for retirement. For instance, when employees change jobs, their retirement contributions remain steady, despite different employer match policies. When administrative data is used, higher match rates may increase participation in retirement accounts by only five percent.

Friedman finds that tax incentives for saving largely benefit wealthy households who would save anyway. Studies that take into account retirement resources such as progressive Social Security benefits find that many households undersave. Research by EBRI and the Center for Retirement Research and RAND find that households with low incomes and less education are less likely to have adequate savings for retirement than those with higher incomes and more education. However, other research from the University of Wisconsin and the Federal Reserve find that households with low incomes are no more likely to fall short in retirement than those with high incomes. These results may suggest that many households might be rationally undersaving, consistent with Buchanan’s Samaritan Dilemma hypothesis.

Effectiveness of Nudges for Savings

In addition to tax incentives, hard mandates and even softer “nudges” may also be important for increasing total saving and retirement preparedness. For instance, total saving increased when mandated savings of 1 percent of income were introduced in Denmark. Automatic retirement saving enrollment is an example of a nudge: participants are defaulted into saving but can always opt out. Empirically, when work plans introduce automatic enrollment, participation can increase by 50 percent. In addition, automatic enrollment is most likely to increase participation in retirement savings plans for low income workers. However, automatic enrollment is not a panacea. If default contribution rates are low, some higher income workers may take that as a signal to contribute at lower rates and save less.

Retirement Saving Policy Reforms

There have been many recent tax reform proposals that aim to take advantage of new research. For example, a nudge in the form of an auto-enrollment IRA has been proposed by American Association of Retired Persons (AARP). In this proposal, workers would be automatically enrolled in an IRA with payroll deduction if their employer doesn’t offer them a retirement plan.

Other reforms try to make tax incentives for saving for retirement more rewarding for lower income households than those with higher incomes. One such reform includes the presidential candidate Hillary Clinton’s plan to place an upper limit on all deductions to a 28 percent rate, even if a taxpayer’s income is taxed at a higher rate. This reform requires households in the top three tax brackets to pay some tax on contributions to retirement saving. President Barack Obama proposed to cap the amount a household can save for retirement also limits the use of tax incentives. Friedman proposes a tax credit for retirement contributions rather than the current tax deduction. Figure 2 shows that while the value of a tax deduction increases with income, the value of a tax credit is the same across income groups. Friedman (2015) proposes a tax credit for firms who offer retirement savings plans and include automatic enrollment. Other proposals, such as the Retirement Enhancement and Savings Act of 2016, would allow small employers to band together to offer retirement plans.

Figure 2: Average Tax Benefit of a Contribution to a Retirement Account in the Year the Contribution is Made, 2013

State Initiates for Retirement Saving

About 30 states are exploring reforms to retirement policy and a number have passed bills. These state initiatives may provide a testing ground to evaluate the impact of different reform proposals.

There are four broad categories of state initiatives for retirement saving policies. The first type of plan, as in Illinois, requires employers who don’t sponsor a plan to automatically enroll employees in a state selected retirement plan where employees can select investments. The second type of reform, proposed in Washington, requires employers to choose a retirement plan from a marketplace. The third, as proposed in California, requires employers to offer a plan and the option of having money in the plan collectively invested rather than having employees make investment selections. Utah has proposed a fourth type of reform where an IRA is offered with no automatic enrollment and no employer mandate.


Retirement saving incentives reduced government revenue by $180 billion in 2016. Friedman finds little evidence that tax incentives increase total saving. Rather, nudges seem more effective at encouraging additional saving, especially for lower income households. Recent state experiments with retirement saving policy may help provide the federal government with evidence of the types of reforms that are most effective at increasing retirement saving.

A discussion of this paper is provided by Brigitte C. Madrian.