Here's a self-description of someone who should probably go Roth:

  • I am in a very low tax bracket
  • I qualify for tax deductions/credits that I won't qualify for in retirement (like child tax credit, mortgage interest deduction, etc.)
  • I expect to have lots of taxable retirement income, such as social security, pension, annuities, and traditional IRA/401k distributions
  • I max out all my tax-advantaged contributions and still save additional money for retirement in a fully-taxed account
  • I think that tax rates will be higher in the future
  • I expect to need some of my retirement savings before age 65
  • I currently work in a state with low income tax, and/or I will retire to a state with high income tax

Here's a self-description of someone who should probably go Traditional:

  • I am currently in the highest tax bracket (or a very high one)
  • I don't expect much retirement income besides IRA/401k distributions
  • I expect tax rates to be lower in the future
  • I will qualify for tax breaks in retirement that I don't qualify for now (e.g., my current high income phases me out of them)
  • I work in a state with high state income tax, and/or I'll probably retire to a state with lower income tax

No single item is definitive. Read the information on Wikipedia and in what I consider to be the definitive r/PF post on the topic. There was another recent post that covered similar concepts here.

And remember,

  • the closer you are to retirement,
  • the less you save for retirement,
  • the lower your tax bracket (both now and in retirement),

the less it matters which you choose.

FYI, employer contributions to a 401k are traditional, no matter what your contributions are.

For a plot and some Matlab code to compare Roth/Traditional/401k/IRA/matched/unmatched/taxable options, go here.

Some analysis of the "Roth lets you save more" argument

tl;dr: Roth lets you save more in tax-advantaged accounts, but that rarely (never?) affects whether Roth is right for you.

Howdy r/PF. Until RedRedditRabbit's post a few months ago my generic advice to everyone was to go Roth unless they were in a very high tax bracket one year. That post convinced me that for many people (especially single professionals, couples with two mid/high incomes, and people who expect little or no pension or social security income) traditional is actually better. One caveat I add to my current recommendation, however, is that if you are maxing out all tax-advantaged savings, Roth lets you save more, since it has the same dollar limits, but uses after-tax dollars.

Lately I've been thinking about whether that's actually true, and my conclusion is that under a few reasonable assumptions to simplify the analysis, it's NOT true. I am by no means a mathmatician, though, so I'd love it if you double-checked my analysis, and my assertion that the asusmptions are reasonable. Plus at least one person has indicated they're interested in this sort of analysis.

So first, the assumptions:

  1. Lacking a crystal ball or influence of Congress, I assume that aside from inflation adjustments, future tax brackets and tax laws will look essentially the same as they do now. I know that in reality they might be higher, but they also might be lower, so I assume they'll be the same. Historically, the bottom brackets have been fairly consistent. If you're in the top bracket, you're on your own.
  2. I assume that enough tax-advantaged savings are available to you to put all bonds and fixed-income savings in tax-advantaged accounts. This lets us consider taxable accounts in retirement using just the long-term capital gains tax (which also applies to qualifying dividends along the way)
  3. My analysis ignores situations where Roth is the right answer for reasons besides the "Roth lets you save more" argument. Specifically, it only considers situations where the current marginal income tax rate is higher than the tax rate on distributions in retirement.

With that out of the way, some notation:

  • CMIT is the Current Marginal Income Tax rate.
  • RMIT is the tax rate on distributions in retirement. It's the average rate over all distributions, not necessarily the marginal tax rate in retirement.
  • LTCGT is the long-term capital gains tax rate, assumed the same both now and in retirement.
  • GF is the growth factor in the tax-advantaged account. If you ignore inflation, it's the expected market return over a certain number of years.
  • GFT is the growth factor in the taxable account. I'm assuming that you realize gains every year, so there is no tax due when you take the money out in retirement, and the LTCGT is baked into GFT. In practice, GFT is usually a little less than GF. If the tax-advantaged account is a 401k with lousy fund choices, GFT might actually be greater than GF, but again, for simplicity I'll ignore that scenario. I also assume that both GF and GFT are greater than 1 (otherwise, you shouldn't be investing). I'll talk more about GFT later.
  • L is the limit on how many dollars can go into the tax-advantaged account this year.
  • S is the total number of dollars you want to save. This analysis is only relevant when S>L.

So, the analysis is as follows:

Let's assume the extreme case, where S*(1-CMIT) = L, meaning all our savings will barely fit in a Roth account. We then compare whether $L in Roth is worth more than $L in traditional plus CMIT*$L in taxable.

We seek a situation (values of RMIT, CMIT, RCGT, GF, and GFT) where [Roth Value in Retirement] is greater than [Traditional Value in Retirement] + [Taxable Value in Retirement] - [Capital Gains Tax on Taxable], or:

[L*GF] > [(1-RMIT)*L*GF] + [L*CMIT*GFT]

Canceling L and adding (1-RMIT)*GF to both sides gives:


Dividing through by R*GFT gives:


Remember, when this inequality is satisfied, Roth is preferred over Traditional.

changed March 11, 2015 delete history edit