Executive Summary
To Roth or not to Roth. It is a question that planners and their clients commonly face, whether making the decision regarding an annual contribution, or about converting (or not) an existing retirement account.
Yet while the appeal for lifetime tax-free growth from a Roth may be appealing, the reality is that the Roth is not always the winning choice, and there are many myths and misunderstands about Roth accounts that make it difficult to know which is best.
The reality is that there are four (and only four!) fundamental factors that determine whether a Roth will or not will be more effective than a traditional pre-tax retirement account. Some factors are always in favor of the Roth account, but others can work against the Roth account; in fact, blindly choosing a Roth and ignoring the relevant factors can actually lead to wealth destruction! By knowing the four factors and avoiding the Roth myths, though, planners and clients can be assured of making an effective wealth-building decision.
The inspiration for today's blog post is financial planner Jeff Rose, who recently called on the personal finance blogosphere to start a Roth IRA movement. Jeff lamented that when speaking to a group of seniors at his alma mater, not one of the 50 students in the audience knew what a Roth IRA was - which means, to say the least, that none of them were likely to contribute to one anytime soon! Jeff was inspired to help get the word out on Roth IRAs.
In the case of financial planners, the existence of the Roth isn't exactly news; it's something we learn about as a basic part of our training. Yet at the same time, while most planners are aware of Roths and their basic features, there is often a remarkable amount of confusion with many myths about what the value really is, and is not, when using a Roth. I've written about this issue extensively in the past, in my May 2009 issue of The Kitces Report, but here are the basic highlights.
The Four Factors Of Roth Vs Traditional IRA
There are only four factors that impact the wealth outcome when choosing between a Roth or traditional IRA (or other retirement account). They are: current vs future tax rates, the impact of required minimum distributions, the opportunity to avoid using up the contribution limit with an embedded tax liability, and the impact of state (but not Federal) estate taxes.
Some of these factors solely benefit the Roth, but others can benefit the pre-tax account; failing to evaluate the situation properly for a client can turn a Roth decision from a wealth creator into a long-term wealth destroyer!
Current Vs Future Marginal Tax Rates
By far, the most dominating factor in determining whether a Roth or traditional retirement account is better is a comparison of current versus future tax rates. Current tax rates means the marginal tax rate that will be paid today (or the marginal tax rate on the deduction that would be received) by contributing or using a pre-tax retirement account versus contribution or converting to a Roth account. Future tax rates means whatever tax rate would apply to the funds in the retirement account when withdrawn in the future - ostensibly in retirement, or possibly even by the next generation if the retirement account is not expected to be depleted during the lifetime of the owner.
The principle of this equation is remarkably straightforward - the greatest wealth is created by paying taxes when the rates are lowest. If rates are low today and higher in the future - e.g., for the young worker, or someone in between jobs - go with the Roth and pay taxes at today's low rates. If rates will be lower in the future - e.g., for someone whose taxable income will drop in retirement, or where the retirement account may be spent by the next generation at their lower personal tax rates - the traditional IRA or similar pre-tax retirement account is the winner. Getting the tax rate equation wrong can result in a significant destruction of client wealth, by unnecessarily paying taxes at high rates!
Impact of RMDs
One important distinction about Roth IRAs (although not Roth 401(k) accounts) is that they are not subject to required minimum distributions (RMDs) during the lifetime of the account owner, while traditional IRAs are. The net result is a slight benefit in favor of the Roth IRA, for the simple reason that it allows more dollars to stay inside their tax-preferenced wrapper. This is an outright benefit for Roths, compared to the traditional IRA that slowly self-liquidates from RMDs, forcing money into taxable accounts where their future growth will be slowed by ongoing tax drag.
Notably, though, this benefit applies only as long as the IRA owner is alive! After death of the owner, all retirement accounts have required minimum distributions for beneficiaries, and the exact same rules apply whether it's an inherited IRA or an inherited Roth IRA (the RMD is the same, even though the tax treatment of the RMD amount may be different). Accordingly, the benefit of avoiding lifetime RMDs applies only as long as the IRA owner is alive, and likewise applies only if the IRA owner actually lives past age 70 1/2 when RMDs begin! Otherwise, the avoiding-RMDs benefit is actually a moot point.
In addition, the benefit of avoiding required minimum distributions for Roth accounts only matters as long as Roth IRAs continue to enjoy the favorable tax treatment - a notable risk, given recent President's budget proposals to eliminate the favorable RMD treatment for Roth accounts!
Contribution Limits and the Embedded Tax Liability
Another factor that favors the Roth IRA is the interaction between the IRA contribution limits and the future tax liability of a pre-tax account.
For example, imagine a client in the 28% tax bracket. If the client has $1,000 in an IRA, the reality is that the client actually has $720 in the IRA for themselves, and $280 in the IRA that's "on hold" for the IRS and the Federal government in the form of future taxes. If the IRA doubles to $2,000, then the client's share grows to $1,440 and the IRS's share grows to $560; the IRS still has 28% of the account earmarked.
In general, this isn't necessarily a "problem" as the benefit is still grow on the IRS' share before they have to be paid (that's the benefit of tax-deductible contributions); the goal is simply to pay the IRS its share whenever the tax rate is lowest, as noted earlier.
However, if the client wishes to make a maximum $5,000 contribution, now it's a problem. Because the client can't make a full $5,000 contribution; in practice, the client makes a $3,600 contribution for themselves and a $1,400 contribution on behalf of the IRS. On the other hand, if the client makes a Roth contribution, the entire $5,000 amount is held for the client, because the IRS' share is paid with outside investment dollars. So as long as the client intends to contribute the limit, it's better (all else being equal) to contribute to a Roth and pay the taxes with outside dollars, than contribute to a traditional where the IRS' share crowds out some of the contribution limit, while tax-inefficient dollars are still growing on the side. Notably, the same effect applies for a Roth conversion where the tax liability is paid with outside dollars; just pretend that the current balance of the traditional IRA is effectively the "contribution limit" to a Roth.
State Estate Taxes
The final factor that can favor a Roth IRA is estate taxes, for the simple reason that it's bad news to pay estate taxes on a retirement account when part of it isn't even yours in the first place - it's earmarked for Uncle Sam!
For example, a client with a $1,000,000 traditional IRA and a $1,000,000 investment account has to report $2,000,000 on their estate tax return (combined with any other assets); however, the client who converts the account has a $1,000,000 Roth IRA and only a $650,000 investment account (assuming a 35% tax rate on the conversion), for a total estate value of $1,650,000. At a 35% estate tax rate, making $350,000 of value "disappear" can result in $122,500 of estate tax savings!
The caveat is that such conversion strategies don't necessarily help for Federal estate taxes, because of the Income in Respect of a Decedent (IRD) deduction, which allows beneficiaries to deduct any estate taxes attributable to the IRA from the income they must report when they take withdrawals from the IRA. The net result is that whether the IRA is converted before death (reducing estate taxes due by paying the income taxes) or is passed on as a pre-tax IRA (reducing the income taxes due by paying the estate taxes), the heirs end out with the same amount of money.
However, that's only for Federal estate taxes. Most states that have a state-level estate tax do not have a state IRD deduction. As a result, clients who are exposed to state estate taxes will find that the Roth allows them to leave more money for the next generation, at least to the extent of the 6% to 16% estate tax rate applicable in most states who have such a tax. Of course, clients should be cautious not to push up their tax rate so far with a big conversion that the adverse income tax impact outweights the state estate tax savings!
The Bottom Line
In the end, avoiding lifetime RMDs, state estate taxes, and paying IRA tax liabilities with outside dollars are all benefits of using a Roth IRA over a traditional IRA. However, the reality is that the driving force on wealth creation - or destruction - is still a comparison of current versus future tax rates. As a result, even with all the other factors working favorably, a Roth can actually still be a wealth destroyer if future tax rates were going to be much lower for that individual client when the funds are withdrawn (either by the client in retirement, or by heirs in the next generation).
So while the Roth IRA may be favorable in many situations, it's hardly automatic in all of them! It's important to evaluate the details of the client's situation, and look at each of the four factors, to determine whether and to what extent each will have an impact, and make a decision accordingly.
For further information on the details of the four factors, see the May 2009 issue of The Kitces Report.
Your conclusion that there are only 4 factors impacting the decision to choose a traditional versus Roth IRA is not correct. You have ignored the variable of what will be the future yield on the earning assets. By choosing a Roth IRA you are buying a tax exempt asset going forward with the price being the value of the tax deduction you are giving up on day 1. This tax shelter only has value if there are future earnings !! (i.e future tax rates have no relevance if there are no future earnings) What this means for the conservative investor with low future earnings expectations is that a Roth may make no sense. (i.e. why pay for tax shelter when there is little income to shelter) Also one must look at the person who invests his/her money in the stock market. Making a much higher rate of return going forward is not a sure thing. Take a look at the S&P 500 for example. Currently the index is still 3.5% lower than where it was at 12/31/99. This is a period of over 12 years where the IRR is negative excluding dividends.
Stan,
With all due respect, this is not correct, as you’ll see if you look at the underlying analysis I provided in the attached newsletter.
The future return on assets does not determine WHICH strategy will be better. It determines the magnitude by which the strategy will be better.
In other words, in any scenario the Roth wins, it will win by more if the return is higher. But there is no return between between 0% and 1,000% that CHANGES the outcome. Similarly, if the traditional is better, it’s better at ALL returns (at least, all positive returns). The higher the return, the MORE the strategy wins, but it doesn’t influence WHICH strategy wins.
It is true that the outcome changes if the expected return is negative. However, no one invests for an expected negative return. The return may BE negative – that’s risk – but we at least invest with the expectation of a positive return, and the rule holds true for all positive returns.
In the context of conservative investors, this similarly means that whether you’re a conservative cash investor going for 1%, or an aggressive stock investor going for 10%, these factors are still the ones that determine WHICH is a success, Roth or traditional. It is true that the magnitude of wealth created by choosing the right strategy will be more limited in a lower return portfolio, but it doesn’t change the fact that SOME wealth is created for any return greater than zero by observing the four factors.
Respectfully,
– Michael
Michael I have two comments re your comment:
#1 You should have said upfront that your conclusion only is valid if positive returns are achieved. Obviously people invest with the goal of making a positive return but people need to know what happens if negative returns are realized. Like I said in my comment ask the people who invested in the S&P 500 12 years ago how that investment is working out for them.
#2 Although your comment is mathematically correct that the “dollar value” of the Roth always will exceed the “dollar value” of the traditional IRA at any yield greater than zero this statement misses the point as to the proper way to evaluate the alternatives. You need to figure the IRR from choosing the Roth alternative where your investment is the tax benefit given up and your return on investment is the tax savings achieved in the future. Only this approach takes into account all of the variables. If you run iterations with low future positive investment returns you will see that the IRRs are so low that a person could easily achieve a higher IRR from an alternative investment. So even in a positive return environment the point is not solely that the dollar value of the Roth always will be higher but rather could I get a better return on an alternative investment versus making the investment in a Roth. When you evaluate the decision this way the difference in the positive returns makes a big impact on the IRR of the Roth investment.
Stan,
The issue is what “account wrapper” to invest inside, not what the underlying investment should be.
The point here is WHATEVER better investment, alternative investment with a better IRR, or whatever else you wish to hold, you can determine whether THAT should be held in a traditional IRA or a Roth IRA with the framework presented here.
In other words, the idea that “a person could achieve a higher IRR from an alternative investment” is not relevant in this context. If you think that’s a better investment, own it. THEN decide WHERE you should own it, in a pre-tax traditional IRA or a Roth IRA, using the framework I have presented here.
WHAT to invest and WHICH type of account to invest with are separate issues. You don’t invest IN a Roth. You invest IN an investment; you may or may not HOLD that investment in a Roth.
Respectfully,
– Michael
Michael – I agree, return changes nothing. I’d add that with the opportunity to convert, there is more than just the two endpoints to consider. It’s easy to find times in one’s life when their income drops for one of a dozen reasons. Those times can be used to convert.
Great writing.
Joe,
Absolutely. In practice, we find “opportunistic” rebalancing in low income years to be an appealing strategy.
It simply makes the point that different parts of the IRA may have different endpoints for the “starting tax rate vs ending tax rate” framework. No one ever said you have to take all the money out or convert it on the same day or in the same year. 🙂
– Michael
To avoid prolonging this discussion this will serve as my final comment on the subject. If one simply looks at the ending value of the two investments, given a positive yield, then the four factors mentioned in your article represent the only variables that can change which investment is larger than the other. If, on the other hand, one also chooses to consider the magnitude by which one alternative exceeds the other then the investment return achieved on the assets becomes a 5 th factor that must be considered. To make an analogy most people would advocate making an investment decision on two alternatives via an IRR approach versus the payback method. Thus in making this decision there are four factors to consider using the payback method and 5 factors to consider using an IRR approach. Thus each answer is correct depending on how you want to evaluate the investment decision. (i.e. 4 factors for a simple approach and 5 factors for a more complicated approach) From my personal perspective, given that you give up cash flow on day 1 by picking a Roth, a decision in favor of a Roth requires more than simply knowing its end value is higher. You also need to know the degree to which it is higher. The answer to this question then requires an investment return assumption or a 5 th factor that must be considered. Given what has happened in the stock market since 1999 I always present to my clients a sensitivity analysis that includes different return assumptions. Based on this information and my clients’ views as to future stock market returns some of my clients have steered away from the Roth as they value the upfront cash flow more than the minimal advantage that a Roth may yield sometime in the future.
Stan,
You’re still missing the point.
The return is still irrelevant to determining WHICH option is better.
I show the math clearly in the underlying newsletter that is attached. Since the discount rate you use to evaluate the future tax liability is the same as the growth rate for the account – by definition – the NPV of the tax liabilities are always the same. And you can generate the same return in either account. That’s why the rate of return never changes WHICH outcome is better. It only affects the magnitude of how MUCH better the winning strategy turns out to be.
In any scenario where the traditional wins, ANY rate of return yields the same decision. In any scenario where the Roth wins, ANY rate of return yields the same decision.
Just look at the math shown in the newsletter itself for more detail.
– Michael
And don’t forget the complication that Traditional IRA distributions can cause Social Security benefits to be taxed where Roth distributions will not.
David
David,
Indeed, in evaluating the future tax rate, anything/everything that impacts the marginal tax rate on future IRA withdrawals is relevant, including the taxation of Social Security.
– Michael
All income-based tests should be considered in evaluating “future tax rates.” For example, Medicare premiums rise with AGI levels so Roth distributions do not impact Medicare premiums.
Chuck,
Absolutely agreed. Evaluating the future tax rate includes everything that goes into the future tax rate, from marginal tax brackets, to the phaseout of certain potentially relevant deductions at higher income levels, to the taxation of Social Security benefits, to the adjustment of Medicare Part B premiums based on AGI.
Future tax rate estimates should be a true reflection of all the factors that go into a future tax rate, not merely an estimate of a tax bracket alone.
– Michael
Michael:
Obviously we have different views on this matter. (including your last comment that the discount rate has to equal the earnings rate which just happens to make the PV of the tax liabilities always the same and thus serving your argument)
Let’s just leave it at we agree to disagree !!
I used to be a big believer in Roths & it probably is a good idea for most people who have no idea where their money is going each month.
For those who are very disciplined & maximize their dollars I think the pre tax option is better.
Let’s say I put in $100 pretax & it doubles to $200.
I am assuming 25% for easy math. I pay $50 & I am left with $150.
If I put in a Roth today instead I can only afford $75 & that doubles to $150 = same ending amount.
I have no idea if my bracket will be higher in retirement.
Additionally, I never understood the paying taxes withoutside investment dollars.
That is for people with alot of extra money.
For most people extra dollars can be used to make sure all pre tax options – 401k, 403b, 457, traditional IRS’s etc are used up before saying that extra dollars will be used to pay taxes on the Roth contribution/conversion.
Maybe a too simple way to look at it but until I can afford to max out all pre-tax options & assuming the simple bracket equation I mentioned earlier (same in retirement as now)Pre-tax is the way to go as everything will be equal & probably the rate will be lower in retirement. This way I build up a much bigger nest egg & worry about all taxes when I retire.
the only way I lose I lose is if taxes are much higher in retirement or I blow money that I could have used to pay taxes now & contribute the same $100 to a Roth.
What about the ability to withdraw contributions to the Roth without penalty or taxable income prior to the age of 55? Maybe that should be another important factor to consider.
The observations regarding “Contribution Limits and the Embedded Tax Liability” are somewhat misleading when the impact on Social Security becomes a factor. The Tier 1 and Tier 2 levels for taxable Social Security income yield defacto tax rates up to 46%. This goes a long way towards increasing the benefits of a Roth IRA.
The old adage about putting money into a 401k plan when your tax rates are high and taking it out when your rates are lower after you retire is a myth for most taxpayers receiving Social Security. The rates quickly jump from 15% to a defacto 28% to a defacto 46%.
For many recipients of Social Security for every dollar withdrawn from an IRA there is a $1.85 increase in taxable income.
David,
The impact of higher marginal tax rates is covered in the section regarding the comparison of current to future tax rates.
If the future tax rate is projected to be higher due to the taxation of Social Security benefits, then that would be part of the analysis in comparing current to future tax rates. Notably, though, that depends on the overall income picture; in some cases, income may be low enough that Social Security won’t be taxed regardless, while in others income may be so high (e.g., from pensions or other sources) that Social Security benefits will already be included in income at the maximum 85% regardless of Roth conversion (or not). Thus, as always, the evaluation of current versus future tax rates is very specific to individual circumstances.
Respectfully,
– Michael
Another factor which should be considered is the time duration of which you can reasonably expect to receive benefits from the IRA or the Roth IRA. As you get older, the rmd on the conventional ira increases substantially, forcing you into an even higher bracket if your account has a substantial value. Beyond 10 years, the Roth value increases almost exponentially. My mother is 94 and still kicking. MY dad was 86, and the gm of my wife made 100. I expect at least one of us to make 100.
You must also consider what you will do with the money that you keep by not converting to a Roth. Under new rules, it may be subject to the 3.8 medicare tax, 20% longs term gain, etc. It will not be worth much in 10 years. Who knows what Congress will do with the Roth plan, and this is the fly in the ointment.
Here are two other factors I see in choosing between the Roth vs. traditional IRA.
First is conversion opportunities set by the government. Back in 2010 the government allowed the conversion to be distributed over the next two years, 2011 & 2012. Actually when the Roth was first created in the late 90’s, the government on a one time basis at the time allowed taxes to be paid over a three year period. This happened to work out perfectly for myself on a $25K windfall payment from an ESOP stock buyout payment from the company where I worked at the time. One should look closely at these opportunities when they present themselves.
The second factor I see is the person’s age. I see the Roth favoring the young. A person in their 20’s or 30’s have a greater life expectancy than someone older near retirement age. This factor works in favor of the younger person giving the “Rule of Seventy Two’s” time to work its magic on doubling their tax-free savings.
All is not lost with the traditional IRA, however. Someone in a high tax bracket today, most likely will be in a much lower tax bracket when that person retires. This was the main selling point on the IRA when it was first introduced years ago. I will find out in about 2-1/2 years when my required distributions kick in.
This is a great post. I will take some issue to the statement that someone in the 28% tax bracket has to earmark a full 28% for the government. Our progressive tax code makes it so that the effective rate on the withdrawals would likely be much lower than 28%, unless there is a significant pension in the picture. I realize the SS can cloud the picture as well, but there will still be lower tax brackets to fill.
I see people compare marginal rates now and later in this evaluation all of the time, and it just rings false. You will save at the marginal rate with a traditional contribution now, but only pay at an effective rate later. There can be a big difference.
Matt,
You are correct that someone in the 28% tax bracket has not paid 28% on all their income up to that point. Tax brackets are progressive, and the prior income will have filled a series of brackets. However, the NEXT $10,000 that is withdrawn/converted/whatever IS subject to that marginal tax rate. So if that last $10,000 (or whatever amount) is what you’re planning around, that’s the rate that should be used for it.
The same is true in the future analysis. The proper approach is still using the MARGINAL tax rates that will apply at that time. If I earn $150,000 this year to be replaced with a $150,000/year pension next year, the reality is that a $10,000 IRA withdrawal this year and next year are at the same marginal rates. To use marginal this year and effective next year would mischaracterize the impact of shifting $10,000 of IRA income to this year versus next year. Certainly, for most people, their base income will change over time and in retirement, so where that last $10,000 (or whatever amount) of IRA income falls will shift, but that’s the POINT of why it’s important to compare marginal tax rates now with marginal rates in the future.
Of course, underlying that is the importance of actually calculating the marginal rate CORRECTLY in the first place, which is an entirely different discussion. I cover it to some extent here: http://www.kitces.com/blog/archives/446-Understanding-Marginal-Tax-Rate-Vs-Effective-Tax-Rate-And-When-To-Use-Each.html
Respectfully,
– Michael
Thanks for the excellent and thought-provoking article!
I am surprised that recharacterization was not included as an advantage for the Roth IRA. The ability to move an asset from traditional to Roth – and then move it back if it drops in value – is a key benefit of making a Roth conversion.
Michael,
I’m of the opinion that question of which one isn’t the correct question to ask, rather we should be asking how much in each. I feel like its an impossibility to accurately project progressive tax rates going forward, but if we build taxable diversity within our plans than at the very least we have greater control. Just my two cents. Thanks for the post.
Ken
Not enough emphasis on the fact that you contribute at marginal rates and withdraw at much lower effective rates, barring a pension or other taxable income in retirement. If you’re contributing at 25% or higher, you would need a ridiculous amount of withdrawals to actually get to a 25% effective rate, especially in in the years before Social Security and if you have significant Roth assets.
You contribute at marginal rates and withdraw at marginal rates. Marginal rates are how all tax strategies at the margin should properly be analyzed.
Yes, your marginal tax rate may be different in retirement, depending on whether or not it is boosting by an underlying pension or other taxable income. But that’s THE POINT of the article. If you don’t have other income boosting your marginal rate in retirement, your retirement rates will be lower than current rates, and a Roth conversion now is a losing transaction.
Even if income is expected to be low in retirement, taxation of Social Security income can have the effect of multiplying your marginal tax rate by 1.5 or even 1.85. Distributions from a regular IRA (RMDs or otherwise) trigger taxation of Social Security money. Distributions from Roth IRAs do not. So, in your example, if one is at an income level where each marginal $ of income triggers $0.85 of Social Security income to become taxable, then an effective marginal rate of 25% means you’re in the 25/1.85 = 13.5% bracket.
Indeed, which is why there are entire articles on this site discussing the importance of calculating the marginal tax rate properly, including for the impact of the taxability phase-in of Social Security benefits. See https://www.kitces.com/blog/the-taxation-of-social-security-benefits-as-a-marginal-tax-rate-increase/
Your comment here actually makes my point, which is that a proper Roth analysis requires evaluating the true marginal tax rate, and not merely looking at “income” in the abstract. 🙂
– Michael
I hope you saw what I wrote as supportive of your statement(s). That was my intent. Your article (URL above) makes the case a lot better than I did. You put things very clearly.
Well I do both. I nearly max out my 401k contributions and max out my roth ira contributions simultaneously each year. I have budgeted my life style to allow for this kind of savings and so it don’t feel much pain from it. I tried to ask my financial advisor at t. rowe price recently where i would be in terms of tax rate when i retire given the growth projections for my assets. Not surprisingly they dodged the question saying i need to consult a tax professional and that it was impossible to determine with very good certainty what my taxes would be in retirement. Probably some of that is true but I also know they don’t want me to stop contributing to my roth ira account with them. In the end i decided to continue with both the 401k and roth ira’s since I am able to do it now and may not be able to do it in the future. As long as I can I will contribute to both but if I fall on very hard financial times then i will probably have to axe the roth ira contributions. Either way i really liked this article. It does help me understand the relative merits of both type of accounts a little better.
Hi Michael,
Excellent piece and very informative. Linked over here from Morningstar. My wife and I come close to maxing out our 401k contributions and still put a little in the Roth’s each year. I have traditional’s that I never converted as well. I like the point you make about having at least some control over your taxable income come retirement time, that’s the exact reason I contribute to both.
Regards,
Rick
I agree that current vs. future tax rates are the most important criteria in determining the appropriate pre/post tax contributions to retirement assets. Your comments below not withstanding, the appropriate comparison is the marginal rate while working vs. the average rate in retirement.
Your gain in using pre-tax funds is that your taxes are reduced – typically at the highest marginal rate. This rate might be impacted by a contribution that lowers your tax bracket..
When these funds are withdrawn – they are taxed at the average rate. Often that rate is minimal. For instance a withdrawal of 20K a year from 401k might come tax free for a couple over 65. Even for a couple with a 90K/yr. income (half from SS and half from traditional IRA/other taxable sources) would expect a tax bill of less than 7000 dollars.
The standard deductions for elderly couples (22,400) should not be ignored. One needs quite a bit of taxable income to bring the total taxes paid in retirement even close to the savings gained by saving “off the top” while working.
The appropriate comparison is the present value tax dollars saved vs the tax dollars paid. The tax dollars saved are marginal rate times savings (adjusted for the time to retirement). The tax dollars paid are, well, the tax dollars paid: Average tax rate time withdrawal amount.