March 22, 2021 – It’s tax season in the U.S. right now! Even though that deadline has just been pushed back to May 17, taxes are on everybody’s mind, so this is a good time to write about the topic in the context of the Safe Withdrawal Rate Series. Until now, I haven’t written all that much about taxes and the main reasons are:
- While I do have a combined 6 letters behind my name (Ph.D. & CFA), I’m missing the three letters “CPA” to write anything truly authoritative about the topic.
- My primary focus is on getting the Safe Withdrawal Rate right. It’s the first issue everyone should worry about. I did some case studies years ago for early retirees and some of them could actually raise their SWR to more than 5% if they do their accounting for future cash flows right. That’s 25% better than the naïve 4% Rule. If you start with a tax plan that’s already somewhat OK and close to optimal, I doubt that you can squeeze out another 25% in after-tax withdrawals through a truly “optimal” tax plan. Hence my approach: get your SWR right and factor in the tax optimization plan afterward to make sure you squeeze maybe another percent or two in the after-tax numbers! (And likewise, if you have a 60-year horizon and not much in the way of supplemental cash flows and you’re looking at a 3.25%, maybe a 3.5% withdrawal rate, you’re not going to “tax-hack” yourself to a 4% withdrawal rate either!)
- Taxes are very personal and it’s difficult to give any generalized advice. As much as I would like to create a spreadsheet like the Google Sheet to simulate safe withdrawal rates (See Part 28 for the details) where you plug in your numbers and the sheet spits out a detailed plan, it’s not so trivial. Very likely, the tax analysis would have to be more custom-tailored! And just to be sure, my Google SWR simulation sheet isn’t trivial either! 🙂
But of course, even if you first do your SWR analysis in before-tax terms, you will want to know how much of a haircut you need to apply to calculate your after-tax retirement budget. Some retirees can indeed make over $110,000 a year and don’t owe any federal tax as I showed in my post in 2019 (“How much can we earn in retirement without paying federal income taxes?“). And in the same post, I showed that to get to a 5% average tax you’ll likely need a $150k annual retirement budget. So, it’s a fair assumption that most of us in the FIRE community will likely get away paying less than 5% of our retirement budget in federal taxes. Add another 0-5% or so for most state tax formulas, and you will likely stay below 10% effective/average tax rate.
But I get the message: because we can’t completely ignore taxes, I wrote today’s post to talk about the general ideas and principles in retirement tax planning. In at least one additional future post (maybe two, maybe three) I will also do a few case studies to see the general principles in action. At that point, I will also include the Excel Sheet I use to perform the tax planning analysis because a lot of readers asked for that tool when I published the Case Studies 3+ years ago! And as I warned before: it’s not as simple as just putting your parameters and Excel automatically spits out your plan. It involves a bit more human input and analysis, stay tuned!
But before we even get to the messy parts, let’s take a look at some general principles…
Principle 1: Try not to compound a tax drag
People often ask me if they should consume or reinvest the dividends in retirement. My recommendation: if you already paid taxes on the income that year, you might as well put it toward the retirement budget. Why would I want to reinvest the cash flow and liquidate more assets and incur even more taxes? Notice that retirement is much easier than saving for retirement in this aspect. You see, for the longest time while working I had to pay taxes on my dividend income in the taxable accounts. And with taxes, I mean Federal plus State (California, arghhh!) as well as 3.8% Obamacare taxes! Only to reinvest those dividends and pay taxes again on those dividends. All else equal, it’s always optimal to defer income for as long as possible and pay taxes only once in the end. Unless you expect a tremendously high tax in the end. Let’s look at the following example:
- A 10-year horizon
- A 10% expected return (all capital gains, no dividends for simplicity)
- A 20% tax on gains
- A $100 initial value
If you were to realize the capital gains every year and pay the 20% tax you’re left with 8% after-tax every year and $215.89 after ten years.
If you defer the capital gains taxes all the way until the end you have $259.37 before tax after 10 years. $100 of which is the tax basis and $159.37 capital gains. After paying capital gains taxes (20% of $159.37 = $31.87) you’re left with $227.50. That’s an annualized after-tax return of 8.57%, 0.57 percentage points higher than if you had realized the gains every year. Not a bad “alpha” purely from better tax planning!
And the tax-alpha goes up the longer you defer the gains! In the chart below I plot the IRR comparing the 8% return when you realize the gains every year, the 10% pre-tax return, and in the middle the after-tax return when you realize the capital gains after x years for x between 1 and 40 years. The longer you defer the gains, the higher the IRR because you spread the one-time tax impact over a longer and longer number of years!
Of course, if taxes go up in year 10, the results may change (but refer to Principle 3 below when facing tax changes over time). But the rise in taxes would have to be substantial to overcome compounding the tax drag. In any case, the “don’t compound a tax drag” principle has many different applications in tax planning:
Application 1: Active vs. passive investing
The tax compounding effect is one of the reasons why stock-picking and market-timing don’t belong in a taxable account. In the numerical example above, an active stock investor realizing all gains every year would have to generate a pre-tax return of 8.57%/(1-0.2)=10.71% just to match the buy-and-hold investor’s after-tax return (potentially even more if the gains were short-term and taxed at a higher rate!). 0.71% annualized alpha is not that easy. If you ever want to do any stock picking, market-timing, dividend strategies, etc., try to do those in a tax-deferred account. And some will yell at me for even bringing up such a sacrilegious issue but keep in mind that while I’m mostly a passive investor many of my readers may not be card-carrying Bogleheads.
Application 2: Pre-retirement stock/bond asset location
The question of whether bonds belong in your taxable account or tax-deferred account is as old as the IRS. Or as old as the tax-deferred account, to be precise. I wrote a long 5,000-word post about the topic (SWR Series Part 35) drilling into the mechanics and the tradeoffs you face. Bonds pay interest and the income is (usually) ordinary income, taxed at a higher rate than capital gains and dividends. Besides, there was also a time when bonds still yielded 5%+, so they had a worse compounding issue than stocks with a 2% dividend yield, adding to the pressure to keep bonds in the tax-advantaged account. But that drag is now much lower in light of the lower bond yields, so for some investors, it may be optimal to keep bonds in the taxable account. White Coat Investor has written multiple posts about this issue all referenced in my SWR Part 35, including this one showcasing an example where bonds in the taxable account indeed give you the better expected after-tax outcome. In my analysis in Part 35 of this series, I showed that, quite amazingly, some of the math that drives the pros/cons of bonds in taxable vs. tax-advantaged account is very reminiscent of one of the oldest economic concepts: “Comparative Advantage“, going back to David Ricardo in the early 1800s to formalize how economists think about the gains from (international) trade.
Application 3: High-dividend yield vs. low-dividend-yield stocks
Aside from the stock vs. bond asset location, there is also the separate issue of where you want to hold what stock type. So, within your stock allocation, you’d be better off keeping the lower-dividend-yield stocks in the taxable account and the higher dividend stocks in the tax-advantaged account to minimize the compounding effect of the taxes. I once wrote a post quantifying the inefficiency of broad equity index funds and the benefit of splitting your equity holdings and investing the half with the lowest yield in a taxable account and the upper half in the tax-deferred account (Here’s an idea for a new ETF).
Application 4: But in retirement, it could be OK again to have high-dividend stocks in a taxable account!
Of course, when you are already retired and you withdraw your income every year to live on, the compounding of the tax drag is no longer a concern. You pay the tax on the income and you consume that income and never reinvest it. No need to sweat that tax drag compounding anymore! So, if you want to do a high-dividend strategy in retirement, go ahead. Just don’t come to me crying if you do worse than the the broad index!
Application 5: Keep RMDs low enough to never exceed your retirement budget
The Required Minimum Distributions (RMDs) starting at age 72 are taxable as ordinary income. If the total RMDs exceed your entire retirement budget it’s not the end of the world; you can simply invest the surplus in a taxable account again, but that will lead to the same problem again: You get taxed once on the RMD, and then part of that RMD keeps growing in a taxable account and gets whacked again by the tax-man.
Well, of course, if you are 72 this year and you have to withdraw $200,000 due to RMDs and your budget is only $100,000 each year you will have no choice. But most retirees will benefit from planning before they even reach age 72. It may make sense to avoid the RMD issue or at least alleviate it before you get there. Use the time before claiming Social Security, when you likely have relatively little ordinary income to perform Roth conversions and/or tap the 401k and IRA already to fund your living expenses to avoid the over-accumulation in the tax-deferred accounts. (see some guidance in Principle 3 below)
Principle 2: It’s likely optimal to first liquidate shares with a high cost basis
Before we even get into the question of what account types to liquidate first (taxable vs. tax-deferred vs. tax-free) there is one question that people often raise: if you have a taxable account, what tax lots should you sell first? Notice that this has nothing to do with stock picking. Even if you own only one single mutual fund (like Fidelity FZROX or Vanguard VTSAX) you still need to determine which tax lots should go first!
Here’s my answer: If you have losses (short and/or long-term), liquidate all(!) of those lots. If that’s more than you need for your retirement budget, you still liquidate them all and reinvest the residual in a similar but sufficiently different asset (=tax loss harvesting), i.e., you sell an S&P 500 index fund at a loss and you buy a U.S. Total Market Index Fund. Or vice versa. This avoids the IRS wash-sale rule. See my post “Be Your Own DIY Zero-Cost Robo-Adviser!” for more details.
Considering that you’re in retirement and you had a long history of accumulating assets, we can probably discard the possibility of sizable short-term gains. So within the remaining tax lots, now all with long-term gains, we’ll likely optimize our taxes when we liquidate tax lots with a higher cost basis first, assuming we face a constant tax rate on capital gains over time (not necessarily true, I know, but see principles 3&4 below). To prove this point, let’s look at the following simple numerical example:
A retiree has two equity lots, each worth $100,000. To make the math as easy as possible, let’s assume that the first lot has a cost basis of $100,000 (i.e., zero gains) and the second has a basis of $0 (assume that it’s close enough to $0 that it’s OK to round down, e.g., you bought $50 worth of AMZN in 1997, now worth $100,000). You also expect an 8% return over the next year and face a 15% capital gains tax. You have an after-tax $100,000 cash flow need this year and like to maximize the expected after-tax cash flow next year. How much of a difference would the order of liquidation make?
- Option 1: If we sell Tax Lot 1 first we get $100,000 and pay no taxes. Tax Lot 2 will be worth $108,000 after a year, which is $91,800 after paying the 15% capital gains taxes.
- Option 2: We sell Tax Lot 2 first and receive $85,000 after tax. We liquidate another $15,000 from Lot 1 to make it to $100,000 after-tax. Our $85,000 lot grows to $91,800 in year 2. But unfortunately, we still owe 15% taxes on the $6,800 in gains. That’s a $1,020 tax bill. So we are left with only $90,780.
So, it would have been wiser to let the lot with the larger capital gains run even longer and liquidate the high-cost-basis lot first. It may not sound like much but if you can increase your year 2 after-tax income by 1.12% that’s nothing to sneeze at. And remember this is not from stock picking, not from market-timing, it’s pure tax arbitrage.
I should also state that the results are qualitatively the same if I assume less extreme tax lots, say Lot 1 with a $70,000 basis and Lot 2 with a $30,000 basis. The math will get a bit messier but the advantage is still on selling Lot 1 first, though at a slightly smaller tax arbitrage percentage. If you drill down into the math, it has some of the same flavors as the “avoid the tax drag compounding” in Principle 1! Try to let capital gains run as long as possible! And the advantage of holding on to the highly-appreciated shares is even larger when you factor in the possibility of the step-up basis, see Principle 4 below!
The situation gets a bit ambiguous when you have both short-term and long-term gains. On the one hand, if one of the lots has a short-term gain much smaller than even the smallest gain in all of the long-term gain lots, it might be optimal to liquidate the short-term gain lot first, even if that means you need to pay ordinary income taxes on the gains. On the other hand, if the difference in gains is small it might be better to wait until the short-term gains turn into long-term gains. It’s hard to make a generalized recommendation, though. But as I said in the beginning: retirees will probably not have a lot of short-term gains, especially once you’re several years into retirement.
Caveat: It might be a good time to also get rid of the high-fee mutual funds with built-in gains once you reach a lower tax bracket in retirement. They may not have the highest basis, but if the fees are high enough, it might be time to get rid of the “stinkers!” See this post for more details: Stuck With a High-Expense-Ratio Fund? Here’s a Google Sheet to Weigh the Pros and Cons of Dumping that “Stinker” in Your Portfolio!
Principle 3: “Smooth” your marginal tax rates across time
In the U.S., as in most countries, we face a progressive tax system, which means that not only will taxes increase with income, but they (normally) increase at an increasing rate. Thus, the marginal tax rates also increase with income. Below is a chart that demonstrates this artifact. The blue line plots the federal income taxes as a function of the AGI (Adjusted Gross Income). There are some special rules for long-term capital gains, so I plot this chart assuming that all your income is ordinary income. Notice how the blue line has straight/linear sections (actually “affine” is mathematically more accurate) and kink points where you jump into the next tax bracket. So the blue line forms a (weakly) convex function, so it keeps going up straight and occasionally makes slight “left turns”.
One implication from this tax function convexity is that variations in your income can potentially raise the average tax burden. Take the following example. A married couple filing jointly with a $100,000 annual (ordinary) income. At the $100,000 annual baseline, you’re looking at a tax burden of $8,590 per year. Small fluctuations don’t hurt you. For example, the +/-$1,000 variation still leaves you in the 12% federal tax bracket, so the tax burden is +/-$120. But if the fluctuations get larger, the high-income scenario pushes you into a higher tax bracket and you lose more on the upside (e.g. +$1,585 in taxes when making $10,000 more) than you gain from the lower taxes when your income is lower (e.g. -$1,200 in taxes when you make $10,000 less). This effect from the curvature of the tax function gets worse when the fluctuations get larger.
And for the math geeks: Yes, I know, I haven’t invented anything new and revolutionary here. It’s all a result of Jensen’s Inequality, appropriately named after the famous FIRE blogger Carl Jensen. Uhm, no, just kidding, that was a different Jensen.
In any case, here are a few applications of this method:
Application 1: Roth 401k vs. Regular 401k
This is relevant mainly for younger folks still working. But it’s a nice application of this principle. Let’s look at the pros and cons of channeling money into the (regular) 401k vs. the Roth 401k. If the (expected) marginal tax rate is lower in retirement than while working, it’s advantageous to invest in the regular 401k. See the diagram of the payouts below. The 401k vs. Roth 401k edge is determined entirely by relative tax rates if we assume the portfolio is invested in the same asset!
But that doesn’t have to be true for every single dollar you contribute. We always have to think at the margin and if your increased 401k contributions eventually push you into a lower tax bracket today and/or into a higher tax bracket while in retirement then the advantage might flip and you ought to reoptimize: you might stop contributing to the 401k and continue with the Roth 401k!
Side note: As a commenter, Martin Silfen pointed out below: There one slight complication in the Traditional vs. Roth 401k tradeoff. The Roth 401k offers a slight advantage because if you are bound to max out all of your tax-advantaged then the Roth 401k is able to accommodate more money (pre-tax equivalent) than the 401k. So if you have a lot extra savings left over and all that money has to be invested tax-inefficiently in a taxable account it would make the Roth 401k slightly more attractive. Sometimes this can make a difference, especially if the above comparison is very close. I built a comparison tool where you can compute this effect for your particular situation.
Application 2: Capital Gains Harvesting (CGH)
On your federal return, a married couple filing jointly can generate $80,800 in long-term gains at a 0% tax rate (2021 values, according to the Tax Foundation). And those $80,800, the top of the 0% tax bracket, is in addition to the standard deduction ($25,100 in 2021). Capital gains harvesting means that you might sell appreciated assets above your cash flow needs to fill up unused space in that 0% tax bracket. This will raise your cost basis and hopefully lower your tax burden in future years where you might again overshoot the 0% bracket and get hit with the 15% rate (or even 20% or worse yet, 20% + 3.8% ACA taxes when you realize capital gains way into the six-figures).
And the idea behind CGH is exactly this tax smoothing approach. It’s inefficient to have 0% marginal rates on capital gains today and 15%+ later. It’s better to max out the 0% today and push your capital gains all the way to the beginning of the 15% rate if you expect a 15% rate (or higher) in the future. Smooth your marginal tax rate! You could potentially even justify CGH beyond the 0% bracket and harvest gains into the 15% bracket if you project landing in the 20% bracket or even the 23.8% bracket later.
Application 3: Roth Conversions and Preparing for RMDs
As mentioned above: We want to cushion the effect of Required Minimum Distributions (RMDs) later in retirement. The potentially large boost in ordinary income might put us into a higher tax bracket starting at age 72. So, let’s take a look at an example retiree facing this issue, see the chart below. Before claiming Social Security, this retiree gets away with paying zero income taxes by simply liquidating capital gains and staying in the 0% bracket for gains. Once Social Security kicks in we do end up in the 12% federal bracket. But leaving the IRA/401k accounts untouched for so long we are now getting hit with the 22% federal bracket once RMDs commence. Well, that seems really inefficient. We would have done better filling up some of the lower tax brackets in the early years. Certainly the 10% and likely the 12% brackets:
And the math and the mechanics are the same again as in Application 1: As long as the conversions/withdrawals are done at a smaller marginal rate than the 22% in the future we gain from the transactions. Once we leave the future 22% bracket and land in the 12% we would be indifferent if we reach the 12% bracket today. Well, I would forecast tax increases later in retirement and certainly still max out the 12% bracket. If we’re still in the 22% future tax bracket then we might even do some additional small conversions into the 22% bracket today to hedge against the possibility of rising tax rates in the future!
But I also like to point out an example where the Roth Conversions clearly went too far. If we get impatient and push the conversions too hard now and catapult our marginal tax rates into the 22% or even 24% bracket early on, only to find we have converted the entire(!) IRA/401k leaving us in the 12% bracket in the future, then that’s a losing proposition. You’d only go overboard like that if you believe that taxes on the middle class (federal brackets 1+2) jump to 24% in the future, but I’m not that risk-averse. It’s possible but not very likely.
Application 4: Withdrawals from Roth vs. Traditional IRA
Before the 2017 federal tax reform, distributions of an inherited Roth IRAs could be spread over an extended period. But with the tax reform, most non-spouse beneficiaries of IRAs must liquidate those accounts within 10 years, making an inherited Roth IRA slightly less valuable. This applies to both IRAs and Roth IRAs. Your beneficiaries will be taxed on the (traditional) IRA liquation. Sure, the Roth IRA liquation will be tax-free, but the beneficiary loses the tax shield for potentially hundreds of thousands of dollars in the Roth and will be forced to reinvest that cash in a taxable account if it’s not used for their immediate living expenses. And of course, there are some exceptions to the 10-year Rule but most inter-generational transfers are subject to that rule. [an earlier version of my post stated that before the 2017 tax reform, the inherited Roth could have been bequeathed indefinitely for as many generations as you like, but I corrected this misstatement. Thanks to reader Martin Silfen for pointing this out! You see: I’m not a CPA! 🙂 ]
So, the old recommendation to hold on to that Roth for as long as possible not as valid now. It might be advantageous for retirees to dig into the Roth during the last few years of their lifespan if liquidating too much of their Traditional IRA would push the marginal tax rate higher than what your heirs are expected to pay. Specifically, you want to play some of those tax smoothing schemes with intergenerational(!) marginal tax rates: if your heirs have much higher marginal tax rates than you, then leave the Roth to the kids. But if it’s the other way around, absolutely, dig into your Roth and leave the traditional IRA to the kids! Smooth your taxes, even across generations!
Caveats: There are probably many different caveats to this approach, but the obvious ones are all related to the fact that Jensen’s Inequality will not apply if the tax function isn’t convex. And there are at least several different quirks in the U.S. tax code and the ACA health care subsidies that create non-convexities:
- If your ordinary income pushes more of your previously untaxed capital gains into the 15% bracket, you will face a 27% (=12%+15%) marginal tax rate not just a 12% marginal tax on ordinary income. Thus, the marginal tax rate on ordinary income can go from 12% to 27% and then back down to 22% once your ordinary income goes into the third federal tax bracket, creating a non-convexity. I wrote about this quirk in 2016. A similar issue comes about again when the capital gains tax goes from 15% to 20% and then to 23.8%, and you might have an additional 5%/3.8% back-door marginal tax on ordinary income. Michael Kitces has more on this: Navigating The Capital Gains Bump Zone: When Ordinary Income Crowds Out Favorable Capital Gains Rates
- The “Tax Torpedo”: Not all of your Social Security income will be taxable on your federal return. The share that is subject to taxation will depend on your other income. And if you generate more income you will crowd out some of the deductions in the Social Security taxation equation, creating a jump in the marginal taxes and then a reduction in marginal taxes once the phase-out is complete. Another non-convexity. I wrote about this issue in 2019: Taxation of Social Security: The Tax Torpedo & Roth Conversion Tightrope
- The ACA subsidy cliff: If your income is even $1 above the maximum income amount you completely lose the subsidy. To account for this non-convexity it might be worthwhile to plan your taxable income accordingly and push it below the ACA cliff for several years but then “go to town” one year and make up for the lost income. The efficiency loss from the convexity of the U.S. tax code might be smaller than the gain from ACA subsidies! (as readers pointed out: the cliff is not an issue in 2021/2022 and might even go away completely. Good news because this eliminates one of the non-convexities and it will make the case for tax-smoothing stronger!)
Principle 4: Plan for the step-up basis
The step-up basis means that if you inherit appreciated assets, you can readjust the cost basis to the market value at the time of the inheritance. You don’t need to pay taxes on the capital gains accumulated during the time the deceased held the assets. In effect, the government forgives you those capital gains!
Planning for the step-up basis comes with a bit of a caveat because there is some political uncertainty about whether this tax hack for the rich will survive the massive deficits we’re piling up these days. But under the assumption that a lot of the folks writing the laws like to use this tax hack in their own estate planning, maybe it’s not so far-fetched to assume it will persist?! Here are some applications:
Application 1: The step-up basis is even better for rental real estate!
The step-up basis works not just for stocks. The same principle applies to real estate, whether owner-occupied or investment property. But in the latter, there is an additional sweetener: The initial owner will likely “depreciate” the asset (for tax purposes) and use the depreciation allowance as a “business expense” to offset the rental income on their tax return. If that owner were to sell the asset then the accumulated depreciation will be “recaptured”, i.e., added to the capital gains, and it’s thereby taxable. Not so if the owner dies. The heirs will inherit the asset and also reset their basis. It’s as if all of the depreciation allowances to the original owner are forgiven! I think that this tax hack is almost too frivolous to stay in place over the next decades. But as long as it’s in place it would be a wise decision for current retirees who like to hand over the maximum (after-tax) inheritance to their heirs to hold on to their rental properties, especially if they have significantly appreciated and/or they have been subject to significant depreciation over the years!
Application 2: What tax lots to sell in a taxable account
In Principle 2, I showed that even with constant marginal capital gains taxes, you want to sell the high-basis (low capital gains) tax lots first. If you can bequeath your highly appreciated assets to your heirs and their capital gains are reset back to zero it’s as if they face a 0% capital gains tax. So, if Principle 2 wasn’t enough of a rationale to hold on to the low-cost-basis shares, the step-up basis certainly makes it a no-brainer!
Application 3: Sequence of withdrawals: Tax-advantaged vs. Taxable accounts
For retirees planning to leave a sizable bequest, it might be worthwhile to hold on to the taxable accounts as well as the highly appreciated real estate assets and live off the retirement accounts during the last golden years in retirement. Especially if your kids face high tax burdens! This is often the case because most bequests move from the parents to kids when the latter are in their prime earning years and facing high marginal tax rates! Getting valuable assets with a step-up basis help a lot in that situation!
So much for today! We’re getting close to 5,000 words! I wanted to set the stage for doing some more specific retirement tax optimization post(s) in the future. There is no one-size-fits-all solution to the tax-optimization. It’s best to stay flexible and do the optimization on a case-by-case basis with all of the principles above in the back of my mind. Stay tuned for some case studies in the next few weeks! And an Excel Sheet to help with that, as promised! 🙂
Thanks for stopping by today! Please leave your comments and suggestions below! Also, make sure you check out the other parts of the series, see here for a guide to the different parts so far!
Title picture credit: Pixabay.com