Family Finances

Increasing Retirement Saving: Reforming Tax Incentives and an All-of-the-Above Approach

February 28, 2017  • David Kamin

This guest blog is part of a series related to the Financial Security Program’s event, Should They Stay or Should They Go? Reexamining Retirement Tax Incentives.

Numerous studies find that millions of Americans are under-saving for retirement, and psychology and behavioral economics have taught us that there are well-understood problems in decision-making that can cause people to not put enough away for retirement and other needs. The question is how to correct these failures and avoid the considerable hardship that can come with not having enough savings.[i]

The United States is now using a variety of tools aimed at increasing retirement saving. These include tax incentives through a complex array of tax-preferred retirement saving accounts, defaults and other nudges sometimes deployed to increase retirement saving, and mandates largely through the Social Security system.

It turns out that an “all-of-the-above” approach—using all of these tools—is probably exactly what is needed.  However, the current system—especially the tax-preferred accounts and defaults—are in serious need of fundamental reform.  Quite simply, the tax-preferred accounts and defaults are not helping many of those who need it most. The tax incentives should be retargeted to the middle and bottom of the income spectrum; this system should be made much easier for all Americans to access; and default saving rates should probably be set higher than they are now (where they exist). Unfortunately, initial measures being considered by the new Congress and the Trump Administration would represent steps very much in the wrong direction.

Role for Tax Incentives

The now-renowned study by Raj Chetty, John Friedman, and co-authors from several years ago (the “Denmark study”) looking at the effects of policy changes in Denmark, and it raised serious questions about whether there is any role for tax incentives in encouraging additional saving.[ii]  Many scholars and analysts have interpreted that study as relatively definitive evidence that tax incentives do not increase saving. But, those results have been over-interpreted and, in some cases, misinterpreted.

The Denmark study found that a reduction in the subsidy rate for certain accounts resulted in the substantial majority of those affected not changing how much they saved in those accounts (reacting “passively”) and a minority significantly reducing their private saving as a result.  (Those who say that the study found almost no change at all in overall saving tend to confuse changes in national saving and private saving. In considering the effectiveness of interventions to overcome mistakes in individual decision-making, it is private saving that matters.)  Overall, the Denmark study implies that for every dollar spent on retirement savings incentives, beneficiaries use 50 cents to increase their retirement savings and 50 cents to increase their current consumption

As compared to many other studies of the effects of retirement accounts, the Denmark study stands out for its clean identification of the effects of the policy changes studied—as the policy changes created a quasi-social experiment—as well as the relatively comprehensive information on saving available in the country. But, the study has little to say about the overall effect of retirement saving accounts, as opposed to marginal changes in the subsidy rate; the existence of the tax-preferred accounts and changes in the subsidy rate could have very different effects for a range of behavioral reasons. Further, the Denmark study focused on a relatively tight income range and did not look at effects across the income spectrum as a whole. To consider the overall effect of such accounts and across a broader range of income, we have to look to an earlier literature.

This earlier literature faced considerable methodological difficulties in identifying the effects of the tax-preferred accounts—and findings need to be interpreted with that in mind. These studies found a range of effects, but most concluded that the tax incentives ended up increasing net worth—and that the accounts did not merely represent a shift in wealth from one type of saving to another.  This was especially the case for less financially sophisticated households, with lower wealth and income; those who were more sophisticated tended to simply shift assets rather than actually increasing saving.[iii] In other words, the positive effects of the accounts seemed strongest among those who might otherwise save the least—which would be relatively good targeting, if proven true.

Both the Denmark study and this earlier literature are comprehensively reviewed in a forthcoming article of mine.[iv] The bottom line is that based on what we know now and despite some inconsistent results, there remains a role for tax incentives in increasing retirement saving, especially for low- and middle-income households.

In Need of Reform

However, as has been widely observed, tax incentives at the moment are poorly targeted.  The vast majority of the subsidy goes to those at the top of the income distribution and with the most wealth.  According to the Tax Policy Center, over 60 percent of the tax benefits (in net present value) go to the top quintile, with 6 percent for the top 1 percent.[v]  Especially to the degree that the country has limited resources to deploy toward policy goals and the effects of these retirement accounts are not offset elsewhere in the fiscal system, this intervention is poorly targeted.

It is not just that the top tends to get the largest benefits despite potentially being least sensitive in terms of total saving, but it is also that the current system is far too gameable.  With Roth IRAs and 401(k)s, reports suggest that sophisticated taxpayers at the top are finding ways to shift labor income and rents on intellectual property into the accounts—potentially entirely avoiding taxation of these items. There is no good reason why people should be able to stuff under-valued “founders stock” of a tech start-up into a Roth account.[vi]

All of this is to say that tax incentives are very much in need of reform. This isn’t a call to eliminate them, contrary to some of the recent literature. Evidence suggests they can do real good in helping to address current failures in under-saving, but it is a call to fundamentally change how they are delivered and to whom.

All of the Above

Tax incentives should be a part of an all-of-the-above approach at increasing private saving. The all-of-the-above approach reflects the fact that there is variation in how much people want to save in a given year, sometimes for very good reasons, and there is also variation in how people react to instruments to try to encourage them to save more. As a result, one instrument alone—a tax incentive, a nudge, or a saving floor—is unlikely to produce results as good as these instruments working together as complements.

Of course, the system as it now stands includes all three types of instruments. It’s just that the tax incentives and nudges, in particular, are broken in a number of ways. More than 40 percent of Americans do not have access to an employer-based retirement plan, creating an immediate barrier for them to access the related tax benefits.[vii] Among those that do, a substantial share are not defaulted into those retirement accounts,[viii] and those that are defaulted into accounts are often defaulted at relatively low rates of saving that run the risk of actually reducing private saving rather than the opposite.[ix]

As result, reforms should be focused on several key priorities:

  1. Refocusing the current tax incentives on low and middle income Americans whose saving behavior would be most sensitive to these incentives. There have been a number of proposals in this vein including enhancing or replacing the current subsidy system with refundable tax credits.[x]
  1. Simplifying the system by establishing universal savings accounts that are easily portable from workplace to workplace and into which all workers are defaulted.[xi]
  1. Carefully studying optimal default saving rates and using ones that are probably higher than now exist in many workplaces.

Steps in the Wrong Direction

Given these priorities, it is disappointing to see the new Congress and the Administration considering steps in the wrong direction.  There are no apparent efforts to refocus who benefits from retirement tax incentives.

And, Congress is now considering undermining state efforts to default workers into retirement saving accounts where employers have not made ones available. Five states have enacted plans that default workers into a state-run IRA plan with employers deducting contributions—and employees always having the option of choosing not to participate. Many millions of workers are expected to gain access to an easy way to save for retirement through these state-based auto-IRAs,  The Department of Labor rules issued toward the end of the Obama administration that clarified that there were no legal blockades to these plans, and Congress is now considering passing a resolution of disapproval under the Congressional Review Act that would eliminate these Obama administration regulations and potentially subject the state plans to legal challenge. Contrary to this, Congress should be building on the state efforts and considering ways to make sure every employer in the country defaults their workers into savings accounts, whether through an employer plan or otherwise, and hopefully into simpler and more portable accounts than exist today.

Congress is doing this even as the administration has suggested it will roll back the Obama administration’s fiduciary rule, which requires retirement planners to act in the best interests of their clients and to no longer take into account backdoor payments they receive from financial product providers when deciding what product to recommend. As the Obama Administration Council of Economic Advisers calculated, each year $17 billion in investment returns that should be ending up in people’s retirement accounts are actually ending up in the hands of financial advisers and others in the financial markets.  We know that ensuring access to good financial advice and improving the decision-framework, as this rule does, can help people with retirement saving decisions.

Based on these early signals, there is a risk that Congress and the President will take steps to aggravate some of the system’s greatest weaknesses, rather than doing what they should and improving the broken all-of-the-above system that we now have.

[i] Parts of this blog post are adapted from a forthcoming article of mine, Getting Americans to Save: In Defense of Tax Incentives, 70 Tax L. Rev. (forthcoming 2017),

[ii] See Raj Chetty et al., Active Vs. Passive Decisions and Crowd-Out in Retirement Savings Accounts: Evidence from Denmark, 129 Q.J. Econ 1141 (2014).

[iii] Daniel J. Benjamin, Does 401(k) Eligibility Increase Saving? Evidence from Propensity Score Subclassification, 87 J. Pub. Econ. 1259, 1281 (2003) (finding “that 401(k)s are more effective as saving incentives for less financially sophisticated or less patient households.”); Victor Chernozhukov & Christian Hansen, The Effects of 401(k) Participation on the Wealth Distribution: An Instrumental Quantile Regression Analysis, 86 Rev. Econ. & Stat. 735, 749 (2004) (“[T]he results suggest that there is substitution between assets held in 401(k)s and other components of wealth in the upper tail of the wealth distribution, but that most financial assets held in 401(k)s in the lower tail of the distribution represent new saving.”); Eric M. Engen & William G. Gale, The Effects of 401(k) Plans on Household Wealth: Differences Across Earnings Groups 33 (Nat’l Bureau of Econ. Research, Working Paper No. 8032, 2000) (finding that, for those with low earnings or low wealth, 401(k)s “may tend to be net saving.”)

[iv] See note 1.

[v] Urban-Brookings Tax Policy Center, “Table T16-0161:  Tax Benefit of Certain Retirement Savings Incentives (Present-Value Approach) by Expanded Cash Income Percentile, 2016,” August 16, 2016,

[vi] See Govt. Accountability Office, Individual Retirement Accounts: IRS Could Bolster Enforcement on Multimillion Dollar Accounts, but More Direction from Congress Is Needed 29 (2014) (describing this type of gaming).

[vii] See Pew Charitable Trusts, A Look at Access to Employer-Based Retirement Plans in the Nation’s Metropolitan Areas (2016).

[viii] For instance, a Vanguard survey of its retirement plans found that as of the end of 2015, only about 40 percent of plans had an automatic-enrollment feature.  Vanguard, How America Saves 2016 (2016).

[ix] See, e.g., Ryan Bubb & Richard H. Pildes, How Behavioral Economics Trims Its Sails and Why, 127 Harv. L. Rev. 1593, 1624 (2014) (“[I]t might be that automatic enrollment has so far exacerbated, rather than eased, the retirement savings problem.”)

[x] See, e.g., Gene B. Sperling, A 401(k) for All, New York Times, July 22, 2014.

[xi] See, e.g., John N. Friedman, Building on What Works: A Proposal to Modernize Retirement Savings (2015).

The views and opinions of the author are their own and do not necessarily reflect the view of the Aspen Institute Financial Security Program or our funders.

This blog comes from the Aspen Financial Security Program