Beware of Retirement Tax Gimmicks

This week, House Republicans are set to reveal their tax legislation, and reports suggest that they are planning to use proposed changes in retirement saving limits as a way to partially hide the true effects of their legislation. The retirement proposal could game both the cost estimate for the legislation and, perhaps even more importantly, the distributional estimate. Republicans seem likely to claim that their retirement proposals increase taxes significantly at the top end helping to offset the bill’s tax cuts for them, when in fact the proposals would not do that and might even cut taxes for high-end Americans.

The gimmick is not a new one. Smaller retirement account changes have been used in the past in this way, but this is super-sizing the game — and potentially with large implications for the tax system going forward.

I’m writing this as an explainer and as a warning to those trying to interpret the Republicans’ tax plans in the coming days and, especially, any cost or distributional estimates that they release. Do not fall for the games. Any cost or distributional estimates that reflect the effect of these changes on a cash basis will mislead by significant amounts. Estimates that either remove the effects altogether or make calculations on a net present value basis — taking into account the true effects of these changes to the accounts over time — should be given much greater weight.

Traditional Accounts, Roth Accounts, and Timing

The key to the Republican game is apparently to reduce contribution limits for traditional saving accounts but not Roth saving accounts — and perhaps to even increase the Roth limits.

Right now, when it comes to accounts like 401(k)s, employers can establish two kinds of accounts: traditional saving accounts and/or Roth saving accounts. There is then a unified limit for employee contributions to these accounts — currently, standing at $18,000. (Employers can then make contributions on top of that.) So, if an employee contributes $10,000 to a traditional 401(k), she can only contribute $8,000 to a Roth 401(k) if one is available (totaling $18,000) and so on.

The two accounts provide a similar — and, under certain conditions, exactly the same — tax benefits, though with very different timing. The traditional account is a “front-loaded” account referring to when the tax benefits are delivered, while the Roth account is “back-loaded” account.

With a traditional account, there is a write off upfront for any amounts contributed, and a tax break is delivered then. However, when amounts are being withdrawn in retirement, the balances are subject to tax at ordinary rates. By contrast with a Roth, there is no write off upfront, but withdrawals in retirement aren’t subject to any tax. (Note to the wise: I said before that the there is a unified contribution limit for Roths and traditional accounts. That is true but the limit for Roths is effectively higher because dollars are contributed on after-tax basis — with contributions not subject to any further tax — whereas the traditional account is pre-tax.)

In economics parlance, both types of accounts exempt the returns to saving from taxation that are “normally” available in the market (returns that aren’t extraordinary, large returns coming from scarce assets like stock in a start up). And, if tax rates are constant and the investor is just earning this normal return, there is no difference at the end of the day in how big a subsidy that person gets from either account; there is just very different timing. If you’d like to see an example worked through of how the accounts can end up in the exact same place, see addendum 1. And, if you’d like a quick exploration of how the accounts aren’t exactly the same under certain, sometimes important circumstances (like taxation of extraordinary returns where Roths can give a big benefit to rich, sophisticated players), see addendum 2.

Gaming the Score

So, with that basic set up out of the way, it becomes clear how Republicans may be trying to game the score — and, in this case, both cost estimates and distributional estimates. Reporting on the exact changes being considered are confused, but, as best I can tell, they are potentially planning two moves:

First, reports suggest that Republicans might propose limiting contributions to traditional accounts to a very low number, perhaps $2,400 — shunting contributions that would previously have gone into traditional accounts into Roths.

Second, at the same time, they might increase the limit on Roths, allowing people to contribute even more money overall to these tax subsidized accounts. I’m interpreting that as the idea that was floated last week as a “compromise” with the White House: potentially, increasing the limit to $20,000 (from $18,000) — probably even as they restricted the traditional accounts. This would only benefit the small share of the population now facing the upper limit on contributions. Currently, less than 10 percent of people making elective contributions to accounts are at the upper limit.

This would have a large effect on metrics that analyze tax changes on a cash basis (looking at cash flows by year). For cost estimates, these two changes would show up as an upfront infusion of cash to the government within the budget window (ten years), even as that might be more than given away over time outside the budget window. When it comes to distributional estimates and despite the long-run effects, this would be reflected as a high-end tax increase initially — since the benefits of retirement accounts are concentrated there. And this would help “offset” the plan’s other tax cuts for the wealthy.

But, here’s the thing: This would be a fake tax increase paying for a very real tax cut. The overall effect of limiting contributions to traditional accounts but increasing the Roth limits certainly isn’t a significant tax increase and might very well be a tax cut for the wealthy over time. And so to put it a bit more sharply: Republicans might be trying to use a tax cut to pay for a tax cut. (We’re in the upside down!)

When it comes to the deficit effect, this game might face some constraints in the Senate. That’s because of the Senate Byrd rule that doesn’t allow reconciliation legislation to add to the deficit outside the budget window, and, while this game generates revenue in the budget window, it likely loses revenue outside. While there is one precedent of the Senate ignoring the effects of these kinds of retirement account timing shifts in judging long-run deficit effects, the size of this shift is of a different order of magnitude, and should make life somewhat harder in the long-run for Republican tax writers — even if there may still be other ways of gimmicking the long-run estimates.

But, when it comes to distributional tables, Republicans seem likely to cherry pick one or two years on which to focus and claim that their plan is fair pointing to Joint Committee on Taxation estimates. In that analytical game, there’s no Byrd rule to restrict them. Unfortunately, the Joint Committee on Taxation does its distributional estimates on a “cash” basis, and so this possible tax cut for the high end will show up as a significant tax increase at the top in initial years.

This comes back to my warning: Do not take these estimates at face value. They have a big gimmick built into them. Instead, give much greater weight to those estimates that pull out the effects of these contribution limit changes or calculate the actual long-run effects on the budget and distributional estimates using a “net present value” analysis. Net present value sums the effects over time taking into account the time value of money. For instance, Tax Policy Center distributional tables often do this with retirement accounts. Such tables will capture real effects rather than fakery.

The Missed Opportunity: Actual Reform

We’ll have to dive in to see exactly how Republicans are proposing to change retirement tax incentives. But, these initial signs suggest that Republicans are missing the opportunity to fundamentally reform retirement tax incentives — reforms which are desperately needed.

I’m not one who believes these incentives should be eliminated entirely. I have written a defense of tax incentives for retirement elsewhere. (Anyone want to read many, many pages on the topic. Well, if so, see here.)

But, the incentives should be reformed. The value of the incentives at the top should be decreased significantly rather than increased by reducing contribution limits and other measures; we should stop the gaming of these retirement accounts — especially Roths — by the super-rich; we should increase the value of subsidies for the middle- and low-end of the income spectrum; and we should simplify the accounts and make participation automatic for the entire population; and the list could go on.

News reports have suggested that Republicans may pair these changes in retirement accounts with an expansion in the Saver’s Credit which is now an extremely limited and complicated benefit for the low- to middle-end of the income spectrum. Unto itself, that sounds like a step in the right direction, but there is danger it ends up being used as a fig leaf to cover up how little the legislation offers for low- and middle-income Americans more broadly. And, there is so much more to be done when it comes to reforming saving incentives. Gaming Roths and traditional accounts to get a better score while still possibly delivering an even larger tax cut at the top isn’t going in the right direction.

Addendum 1: Working Through an Example to Show Equivalence (Under Some Circumstances)

To give a simple example of how traditional and Roth accounts can deliver the same tax subsidy: Compare what happens if someone wants to put $10,000 on a pre-tax basis in a retirement saving account, using a traditional account or a Roth. I’ll assume the person will hold the money in the account for 20 years making a 5 percent annual rate of return, and faces a constant 40 percent tax rate (like the top tax bracket now).

In year 1 with a traditional account, the person gets a $10,000 write off and doesn’t face any tax. $10,000 goes into the account. With a 5 percent rate of return, the account balance grows to $26,533 by the end of the 20th year. But, then, there’s a 40 percent tax at withdrawal. $10,613 in taxes are paid at that point, leaving $15,920 for consumption.

Compare that to the Roth. With a Roth, there is no upfront write off. Amounts contributed must come from post-tax dollars. And so the 40 percent tax applies upfront. $4,000 goes to the government, and $6,000 goes into the account. That $6,000 then makes a 5 percent rate of return, resulting in a balance of $15,920 at the end of year 20. The amounts are then withdrawn without facing any tax — and the person ends up in the exact same position as he would have using a traditional account.

As compared to investing outside a retirement account, the tax subsidy is the same: in both cases, the return to saving has been exempted from tax and the person ends up with the same resources. But, the timing of tax payments is very different.

Addendum 2: Roths and Traditional Accounts Aren’t Actually the Same, But That Doesn’t Fundamentally Change the Game

Roth and traditional accounts don’t always provide the same tax subsidy. There are a variety of differences, though none of this changes the fundamentals of the game that Republican tax writers appear to be playing. Just to run through a few of these:

Roths, in some important ways, provide a more generous tax subsidy. They not only exempt the “normal” return to investment from taxation that is available widely in the market; they also exempt “extraordinary” returns — above-market rates of return available on scarce assets (i.e. scarce in that if you had more money upfront, you couldn’t buy more of these amazing assets). What are these scarce assets paying “extraordinary” returns? Well, they might mix in returns to labor with returns to saving or reflect under-valuation of the initial assets being contributed to the accounts. This is why reports of some very sophisticated players stuffing assets like “founders stock” into Roths is so egregious. They know the stock is likely worth a lot though the IRS has no real ability to challenge the initial very low valuation. And, they then could entirely escape taxation of these extraordinary rates of return.

Or, to put this in terms of one recent presidential candidate: As became public back in 2012, Mitt Romney apparently has a super-sized traditional IRA since he stuffed assets like “carried interest” into it. But, as compared to the new super rich playing around with Roths, Romney is in a far worse position. He’ll still face tax on those huge returns upon withdrawal. Imagine a modern-day Romney now setting up a Roth account and stuffing in carried interest. The huge return could entirely escape taxation.

Roths and traditional accounts are also different if a person’s tax rate doesn’t remain constant over time. Holding all else constant, Roths are better for those facing higher tax rates in the future, and traditional accounts are better for those in the opposite position.

From a broader policy perspective, this is probably yet another strike against Roths. On average, tax rates seem more likely to rise over time perhaps especially at the top rather than the opposite given the government’s fiscal shortfall, and so Roths become more expensive if that is the case. Further, as William Birdthistle and Daniel Hemel have pointed out, Roths suffer from greater mistargeting since they give the greatest benefit to those in the highest tax brackets (and so earning the most income) specifically in retirement — an odd approach.

On the other hand, Hemel has also noted how Roths can minimize the fees going to financial services firms and increase government revenues as a result (again, holding all else constant).

Finally, there is an important question of how people respond to a similar tax incentive but delivered in a very different way. I won’t go into the details here, but there are arguments going different directions as to which set up may most increase saving and empirical evidence is so far relatively scant.

These are all interesting and, in some cases (like the non-taxation of extraordinary returns), important issues. But, none of this changes what is motivating the Republican effort to push people into Roths: Which is to game cost and distributional estimates. Nor does any of this change the fact that in doing so Republicans are probably ignoring many of the most glaring problems with the existing retirement tax incentives.