Why Retirement Plan Tax Preferences are Not as Expensive as You Might Think

Filed Under (Retirement Policy, U.S. Fiscal Policy) by Jeffrey Brown on Dec 13, 2012

Retirement plans such as the 401(k) receive favorable tax treatment under the U.S. income tax system.  Historically, this favorable tax treatment was provided to increase individual saving.  Recent research has called the efficacy of this approach into question, suggesting that individual saving rates may not be all that responsive to marginal tax rates.

Last week, I wrote about the danger of drawing the conclusion that tax incentives do not matter and that we should therefore look to eliminate the tax preference for retirement saving.  My focus was on the role that tax preferences play in providing an incentive for employers to offer plans, and to design them in a way that uses behavioral nudges to increase saving.

This week, I want to focus on a different aspect of this issue, the public discussion of which has been misleading – how much this tax preference costs the U.S. Treasury.  My contention is that the cost figures being bandied about (including my own use of the $100 billion figure in last week’s post) are substantially overstated.  The point of today’s post is to note that the amount of revenue that the government would receive by eliminating the preferential tax treatment for retirement saving would be much less than what it might appear.

To understand this, one must understand (1) how retirement plans are treated under U.S. tax law, (2) how the government actually accounts for the foregone revenue, and (3) how the government ought to account for the foregone revenue.  These are complex topics, but some simple exposition is sufficient for seeing the main point.

(1)   How are retirement plans treated under U.S. tax law?  In a nutshell, the income tax on retirement plan contributions is deferred, not eliminated.  This is an important distinction.  If I receive an additional $1000 in cash salary, and I am in a 35% tax bracket, I owe the government an additional $350 in taxes.  If, however, I receive this additional $1,000 in the form of a contribution to a 401(k) plan, I owe no taxes today.  However, I will owe taxes on the money when I withdraw it during retirement.  Of course, there is financial value to deferring my taxes (what we economists call tax free “inside build-up”), but it is not as if the initial contribution escapes the tax system entirely.

(2)   How does the government account for the foregone revenue?  The U.S. Department of Treasury and the Congressional Joint Committee on Taxation prepare annual estimates of what they label “tax expenditures.”  These tax expenditures are basically just an estimate of how much additional tax would be collected if a particular activity went from being untaxed to being taxed, assuming no behavioral response to the tax.  (As an aside, the fact that they do not account for a behavioral response is why they are careful to always note that “a tax expenditure estimate is not the same as a revenue estimate.”)  In the case of retirement plan contributions, they roughly calculate the amount of money being deferred, apply the relevant marginal tax rates to it, and obtain a rough estimate of how much revenue is not being collected as a result of this tax preference.  However, a key point is that they do not estimate this over the entire life of the account, but rather use an arbitrarily truncated time horizon to estimate the effects.

Going back to my simple example: suppose I contribute an additional $1,000 today to a 401(k) plan.  That saves me $350 in taxes today, and costs the government $350 in foregone revenue in the current tax year (assuming I would save the same amount either way).  So far, so good.  But suppose that I plan to pull the money out in 20 years.  I will pay income taxes on the amount I withdraw.  The present discounted value of the tax that I pay in 20 years will likely be less than $350, but it will be much greater than zero.  For the sake of example, suppose it is worth $150 in present value.  If so, then the net gain to me (and the net cost to government) over my lifetime is $200.  The problem is that the government does not use a present value method.  Instead, it looks at just the front end, and thus overstates the value of the deduction.

(3)   How should the government account for tax expenditures?  Ideally, the government would compute these tax expenditures using the “present value” concept just explained.  A number of experts have made this suggestion.  For example, a paper by the American Society of Pension Professionals and Actuaries (ASPPA) boldly states “tax expenditure estimates for retirement savings provisions should be prepared on a present-value basis” because this “would allow an ‘apples to apples’ comparison” with other tax deductions.

What does all this imply?  A paper written by two Treasury Department officials and published in the December 2011 National Tax Journal found that “the long-run NPV cost can be dramatically different if measured using relatively short time horizons.”  The calculations are a bit tricky because one must make assumptions about rates of return, the appropriate discount rate, current and future marginal tax rates, and so on.  And the extent to which estimates differ depends on the time horizon being examined.

But, these caveats aside, the ASPPA study concludes that “the present-value tax expenditure estimates of contributions made in the first five years are 55 percent lower than the JCT five-year estimates and 75 percent lower than the Treasury five-year estimates.”  That is a huge wedge.

How does all this matter for policy?  The fiscal cliff has DC policymakers scouring the four corners of the earth looking for ways to boost revenue without raising marginal tax rates.  One way to do this is to eliminate tax expenditures.  However, some of those tax expenditures exist for good economic reasons, and the provision of favorable tax treatment for retirement saving is one of them.

As noted last week, the elimination of this provision could have serious unintended consequences for the availability of retirement savings programs through employers.  Now add to that the fact that any revenue implications of such a policy change are substantially overstated and what you get is the potential for good intentions (closing the fiscal gap) to lead to bad policy.

Relevant Disclosures:  I serve as a trustee for TIAA, a provider of retirement plans to the not-for-profit sector.  I have also received compensation as a consultant or speaker for a wide range of other financial services institutions.  The opinions expressed in this blog (and any errors) are my own.

 

Tax Subsides for 401(k)’s Work, But Not for the Reasons You May Think

Filed Under (Finance, Retirement Policy, U.S. Fiscal Policy) by Jeffrey Brown on Nov 30, 2012

Earlier this week, the New York Times Economix Blog wrote a piece “Study Questions Tax Breaks’ Effect on Retirement Savings.”  The article summarizes the findings of a fantastic research paper issued by the National Bureau of Economic Research (NBER).  A quick summary of the paper written by the authors themselves can be found here.  The short version is that the researchers used data from Denmark (where much better date is available) to provide evidence that tax subsidies have little effect on overall savings rates.

Their main finding is that “when individuals in the top income tax bracket received a larger tax subsidy for retirement savings, they started saving more in retirement accounts.  But the same individuals reduced the amount they were saving outside retirement accounts by almost exactly the same amount, leaving total savings essentially unchanged. We estimate each that $1 of government expenditure on the subsidy raised total savings by 1 cent.”

The policy implications of their finding are extremely important given the current debate about fiscal policy in the U.S.  After all, if tax subsidies for saving do not actually increase saving, then perhaps we should re-think the $100 billion per year that we forego in tax revenue by exempting retirement savings from the income tax base?  Such a conclusion would be quite tempting to politicians who are desperately seeking ways of raising revenue without raising tax rates.

But I say “not so fast.”  Although I do not disagree with the empirical findings of the study, I strongly disagree with the assertions being made by some that this finding justifies the elimination of the tax preference for 401(k) and other retirement vehicles.

The study itself is an outstanding intellectual contribution, and one that will likely (and deservedly) end up being published in a leading scholarly journal.  I can personally vouch for the high intelligence and research integrity of the two U.S. authors.  Raj Chetty was named a MacArthur “Genius” earlier this year, and is widely expected to be awarded the prestigious John Bates Clark medal sometime in the next 6-8 years.  John Friedman of Harvard is also an emerging research star in the economics profession.

So, the researchers are top notch, the study is extremely well done, and the conclusion is that tax subsidies do not generate net much net savings.  So, why not simply eliminate the tax preference for 401(k) plans in the U.S. and raise a trillion dollars of revenue over the next decade?

Because of the important role of plan sponsors, that is why.

For better or for worse, the employer plays a central role in the U.S. retirement system.  Although there are several reasons that employers offer retirement plans and other employee benefits (e.g., to differentially attract certain types of workers, to help manage retirement dates, to motivate workers, etc.), there is little question that the large tax subsidy  looms very large in their decision to use retirement plans – as opposed to other types of benefits – to achieve these outcomes.

To qualify for favorable tax treatment, employer provided retirement plans, including the 401(k), must meet a long list of “plan qualification requirements.”  These requirements are what provide Congress and regulators the ability to influence the design of retirement plans.

An important example is the set of “non-discrimination rules” designed to ensure broad-based participation in an employer’s plan.  These rules provide incentives for plan sponsors to find innovative ways of encouraging saving by their employees.  Indeed, it is not much of a stretch to suggest that these rules are the reason we have seen the widespread adoption over the years of employer matching contributions, automatic enrollment, automatic escalation of contributions, and numerous other innovations in the retirement plan space that have been shown to increase saving.

The authors themselves note that “automatic enrollment or default policies that nudge individuals to save more could have larger impacts on national saving at lower fiscal cost.”  I agree that behavioral nudges have had an enormous impact.  But in an employer based retirement plan system, the only way to get employers to offer those nudges is to provide them with a compelling financial reason to do so.  In essence, tax subsidies are the nudge for employers to provide the nudge for employees.

Of course, this does not necessarily mean that the existing system should be treated as sacrosanct.  It may be that employers would continue to offer 401(k)’s – along with their numerous savings nudges – if the financial incentive were provided in a less expensive way (e.g., by capping deductibility).  That is a debate we ought to have (hopefully informed by evidence of the same high quality as the NBER study).  My point is simply that any policy discussion should recognize the very important role that employers play as trusted sponsors of the plan, and be careful not to throw out the baby with the bathwater.

Indeed, given that only about half of US workers have opportunities to save through their current employer, we should be looking for ways to encourage more employers to sponsor plans.  If we go after the tax incentives for retirement saving, we must be careful not to inadvertently destroy the plan sponsor infrastructure that is the foundation of retirement security for millions of Americans.

 

Relevant Disclosures:  I am a Research Associate of the NBER (through which the study above was released) and Associate Director of the NBER Retirement Research Center (through which the authors have received some funding for their study).  I am also a trustee for TIAA CREF, a provider of retirement plans to the not-for-profit sector.  I have also received compensation as a consultant or speaker for a wide range of other financial services institutions.  The opinions expressed in this blog (and any errors) are my own.