Roth-only?
It is being reported that House Republicans are considering, as part of their tax reform effort, converting the current 401(k) system to “Roth-only,” primarily as a way to raise revenues to finance reductions in marginal tax rates and reform of the corporate tax. In this article we review: (1) the difference between the tax incentives provided by non-Roth and Roth 401(k) contributions; (2) the reasons why House Republicans are considering this change (why converting to Roth-only, or simply increasing the portion of 401(k) contributions that get Roth treatment, increases revenues); and (3) possible effects on retirement savings resulting from such a change.
Background: Roth treatment vs. non-Roth 401(k) treatment
For background on how the current system of tax incentives for retirement savings work, please see our article Retirement savings tax incentives – the current system.
As we said in that article, generally, Roth contributions are taxed on the way into but not on the way out of a 401(k) plan; non-Roth contributions (“regular” 401(k) contributions) are taxed on the way out but not on the way in. But, assuming the participant’s tax rate is the same at the time of contribution and distribution, the participant receives the identical tax benefit whether she makes a Roth or a non-Roth 401(k) contribution.
Let’s illustrate that result. The following table compares the net benefit under non-Roth/Roth contribution treatments, assuming a gross contribution of $18,000, earnings at 3%, a 15-year investment period, and a 39.6% tax rate at the time of contribution and distribution. In this illustration, note that, while the non-Roth contribution is the full $18,000, the Roth contribution is $18,000 minus 39.6% taxes = $10,872.
Non-Roth vs. Roth treatment
Non-Roth | Roth | |
Contribution | $18,000 | $10,872 |
Earnings | $10,043 | $5,573 |
Total | $28,043 | $16,445 |
Minus tax at distribution | $(11,105) | |
Net benefit | $16,938 | $16,445 |
This example simply illustrates what is axiomatic: if the tax rate is the same at the time of contribution and distribution, it doesn’t matter (to the net benefit) whether you collect the income tax at the end of the period (at distribution) or at the beginning of the period (at contribution). So – under these conditions – going to all-Roth has no effect on the participant.
Roth contributions can produce a bigger net benefit
There is one significant difference between non-Roth and Roth tax treatment. If you’ve been following the math so far, you may have noticed that you can in fact generate a greater tax benefit with Roth contributions than you can with non-Roth contributions. That is because, with respect to Roth contributions, the Tax Code dollar limits are applied to the after-tax contribution the participant makes. Thus, a participant making non-Roth contributions is limited to (in 2017) $18,000 in “pre-tax” contributions, and he will ultimately have to pay taxes on this amount plus earnings. But a participant making Roth contributions may make $18,000 in “after-tax” contributions, and he will get that amount plus all earnings on it back tax free at distribution.
Why House Republicans might want to require 401(k) Roth contributions
The interest of House Republicans in converting the current 401(k) tax system, which (more or less) lets the sponsor and participant choose between non-Roth and Roth contributions, to one requiring that all 401(k) contributions be made as Roth contributions, is strictly a function of the way legislation is “scored” for budget purposes.
The Congressional Budget Office scores revenue gain/loss by looking at budget impact over 10 years. While (as we just discussed) the tax effect of Roth and non-Roth programs are (given constant tax rates) equivalent, when the CBO 10-year scoring window is used the budget effect of the two programs produces very un-equivalent results.
Consider the cost to the budget of our example $18,000 contribution by an individual in the highest tax bracket. Under the non-Roth approach, the individual gets an $18,000 deduction in year 1 – cost to budget = $7,128. Ultimately, in year 15, that individual will pay $10,782 in taxes, but since that payment is outside the scoring window, it will not be reflected in CBO’s score. This approach – as many have observed – overstates the cost to the federal budget of non-Roth 401(k) contributions.
On the other hand, under the Roth approach, the individual pays $7,128 tax on the $18,000 in year 1 – cost to budget = $0. Ultimately, in year 15, that individual will get back $16,445 on which she will pay not taxes – cost to the budget (in year 15) = $10,782. But since that payment is outside the scoring window, it also will not be reflected in CBO’s score. So (in a similar fashion) this approach understates the cost to the federal budget of Roth 401(k) contributions.
Thus, converting to “all-Roth” will produce an (artificial) revenue gain. Some would call this a gimmick. But it makes a very tempting target when a policymaker is (1) trying to reduce marginal tax rates without (2) increasing the budget deficit.
Objections to an “all-Roth” approach
Many have objected, and are objecting, to converting the current system to all-Roth. Let’s review some of the problems the all-Roth approach raises:
Participants don’t like Roth contributions. There is some data to support the claim that participants, and especially lower-paid participants, don’t like the idea of Roth contributions. That result is somewhat counterintuitive – among other things, it seems reasonable to assume that lower paid/low (or no) tax-paying participants would be less motivated by tax concerns. But the result is there in the data.
Here are some attempts at explaining that result:
Current contribution withholding/matching contributions are calculated based on gross pay. If you convert to all-Roth and continue to make calculations based on gross pay then, e.g., in our example above, the Roth contribution would be $18,000, and the participant would have to pay an up front tax of $11,801. The participant’s “take home pay” would be reduced not by (as in our example) $18,000, but by $29,801: $29,801-39.6% taxes = $18,000. And where there is a 3% match, the participant would have to contribute 3% of gross pay to get the full match, and then pay taxes on that 3%-of-gross-pay contribution. All of this appears to be an artifact of recordkeeping, and it’s not clear that a policymaker would consider such a recordkeeping issue to be an insurmountable obstacle.
Participants may not understand the math. This may be true. But that would appear to be an issue of communication/education. And, similarly, it’s not clear that a policymaker would consider such a communication/education issue to be an insurmountable obstacle.
Participants may expect to be paying taxes at a lower rate when they retire than when they make their contribution. Indeed, if their non-Roth contribution is big enough, it may currently move them into a lower tax bracket. Given data about how participants are not “saving enough” for retirement, this feature of regular 401(k) contributions – the ability to move current income into a future year in which the participant is paying a lower tax rate – may in fact be very valuable to many participants.
Whatever the case, if it is true that participants don’t like Roth contributions, and that that dislike can’t be overcome (e.g., by education or defaults), then converting to all-Roth will reduce retirement savings, which most would view as a bad thing.
Switching to all-Roth will reduce taxpayers’ ability to smooth income. In a sense this is just another version of the last reason why participants don’t like Roth contributions. The current system allows income smoothing in both directions: taxpayers who expect their future tax rate to be lower can make non-Roth contributions, reducing their current income/tax rate; and taxpayers who expect their future tax rate to be higher can make Roth contributions, reducing their future income/tax rate. Roth-only will only allow the latter. This is not a point about retirement policy – income smoothing is not a retirement policy value. But many would argue it is an important feature of income tax policy, especially where marginal rates are steeply progressive.
Switching to all-Roth will result in a negative shock to capital formation. Under the current (predominantly non-Roth) system, most taxes are collected on the back end of the savings period, when the participant, in effect, pulls his money out of the capital markets and “consumes it.” Under an all-Roth system, taxes will be collected on the front end.
Assuming constant tax rates (which for a number of reasons may be a problematic assumption), a change to all-Roth may not, over the long run, affect total tax revenues or total capital flows. In our example, the same tax benefit is delivered to the Roth saver, she just has to wait till year 15 to get it. The same should be true for the capital markets. But the year an all-Roth system becomes effective, all the money that will be paid (up front) as taxes will come out of money that used to flow into the market. Many policymakers may regard that as a significant problem.
Semi-Roth
Many have speculated that the House Republicans will ultimately wind up with something like the proposal made in 2014 by Congressman Dave Camp (R-MI), then-Chairman of the House Ways and Means Committee. Under his proposal, except for certain plans of small employers (100 employees or less), the dollar limit on regular (non-Roth) 401(k) contributions would be reduced by one-half. Thus, based on 2017 limits, under such a proposal the limit on non-Roth contributions would be $9,000. Current limits would remain in place for the combination of non-Roth and Roth contributions – a participant could still make the maximum 401(k) contribution, but half of that contribution would have to be made on a Roth basis.
Such a hybrid approach would address (at least) concerns being raised about the effect of an all-Roth system on lower paid participants. It would also mitigate the effects of some of the other issues being raised, although it would not eliminate those effects.
* * *
We will continue to follow this issue.