In a recent paper, David Blanchett found (again) that household spending tends to decrease over the course of retirement. That is, it increases, but not as quickly as inflation. So in “real” terms, it’s gradually going down.

Relative to Blanchett’s earlier work on how retirees change their spending over time, his latest paper had two particularly noteworthy findings.

Median vs Mean

The first especially interesting finding was the difference between the median and mean (average).

For the median retiree, inflation-adjusted spending decreases throughout retirement.

For the average (mean) retiree, however, inflation adjusted spending goes back up at older ages (though it still stays below the initial level of spending).

The difference appears to be significantly due to large health-related costs at older ages, which are included when calculating a mean, but which do not affect the median retiree. For example, as Blanchett writes, “among those who passed away at the age of 95, the median cumulative real lifetime unexpected out-of-pocket medical expenses were only about $50,000 compared to roughly $250,000 at the 95th percentile.”

Reducing Spending by Choice

The second particularly interesting finding is that even households that are “funded or overfunded” still reduce spending. That is, while some retiree households reduce their spending due to limited funds, even households who don’t need to reduce spending nonetheless typically do still reduce spending over time.

As Blanchett writes, “Only those respondents who were the most well-funded and spending at lower levels tended to increase in spending. Average real spending declined for all respondents spending $80,000 or more, regardless of funded status, although spending declines were lower as funded status tended to improve.”

Why People Reduce Spending

To me, it’s not surprising at all to find that people reduce spending over their retirement, even when they aren’t forced to do so.

For example, imagine a world in which there was absolutely no uncertainty. You know exactly what your career earnings and investment returns will be. You know what inflation is going to be. You know exactly how long you (and your spouse, if applicable) will live. You know exactly what your health care costs (and other “lumpy” costs such as home repairs) will be each year.

And so you’re left with some, definitively known, amount of discretionary spending, which you can allocate across the years of your life.

In that world, how would you allocate those dollars, across time?

There’s no right or wrong answer here. But most people would choose to do more discretionary spending in their earlier years and less in their later years, simply because it’s easier to enjoy discretionary spending at a younger age. At 25 it’s easier to have a travel-the-world type of adventure than at 45. It’s easier at 45 than at 65. And it’s easier at 65 than at 85. And the same things goes for most types of discretionary spending. It’s just easier to do it the younger we are.

Some people might choose the classical economics “consumption smoothing” idea of having your spending stay level over time. But it’s hard to imagine many people intentionally choosing an increasing spending path all the way through life (e.g., pinching pennies at age 35 so that you can “live large” at age 85).

Now let’s bring back one type of uncertainty: lifespan. So we’re still assuming no investment risk, no uncertainty as to health care costs or other big expenses. But now we don’t know how long you’ll live. Naturally, that means we need to plan for a scenario where you live longer than your life expectancy, but there’s another aspect here that is often left out. And that is: would you weight earlier years more heavily (i.e., choose to spend more in those years than in the above case) simply because you know you’ll be alive in those years? In other words, separate from the decision you made above about year-by-year spending preferences, when we add longevity uncertainty into the mix, would you now choose to, for example, further shift the spending in the direction of the earlier years, simply because you’re more likely to be alive during those years?

Again, different people will answer differently here. But this factor is either not important to you, or it’s a point in favor of more spending in earlier years. Nobody would say, “I’m less likely to be alive at age 95 than at age 65, and therefore I will plan to allocate more dollars to spending at age 95 than at age 65.”

So we have two factors, both of which point in favor of weighting earlier spending more heavily than later spending (though to differing degrees from one person to another). For most households, that’s broadly the goal that we’re trying to achieve.

“Reducing spending throughout retirement” might sound bad. (And indeed, being forced to do so is probably not what we want.) But “intentionally choosing to spend more in earlier retirement” is another way of saying the same thing, and it is broadly something that people want to do.

Administrative note: there will be no article next week (12/23), as I’ll be taking time off for the holidays.

Last week I linked to my “Roth Conversions Deep Dive” presentation from the recent Bogleheads Conference. And many people wrote in to ask about things that were not included in the presentation.

One of the biggest challenges in preparing for a presentation like that is the time limit. In this case it was a 50 minute session, but I was targeting 40 minutes of actual presentation time in order to leave time for audience Q&A. 40 minutes discussing Roth conversions may seem like a lot, but truly, this is a topic where it would be easy to have a three-day continuing professional education seminar for CPAs or CFPs, without running out of things to talk about. There are so many moving parts and so many different ways to look at the question.

Today’s article is meant as a follow-up, to answer a few of the things that people asked about most often in reply to the presentation.

If you haven’t watched the presentation yet, please do, as it provides necessary context for some of the discussion below:

Roth Conversion Deep Dive, with Mike Piper

One other point, by the way, is that the Bogle Center has opted to turn off comments on the YouTube channel so that we don’t have to moderate all the spam that unfortunately comes with financial topics. If you want to discuss any of the videos from the conference, I’m confident you’ll get plenty of discussion if you start a thread on the Bogleheads forum.

Creditor Protection

The details vary by state, but IRAs have some degree of creditor protection, whereas taxable accounts generally do not. So, whenever you’re using taxable account dollars to pay the tax on a conversion (i.e., to “buy more” Roth space), another thing that’s happening is that you’re also likely “buying more” creditor-protected space. But again, the details vary by state. If creditor protection is a major concern for you, consulting an attorney is a good idea.

Estate Taxes: a Point in Favor of Conversions?

With respect to the federal estate tax, it’s not, in theory, a point in favor of conversions, due to the income in respect of a decedent (IRD) deduction. In this context, the IRD deduction essentially says that the beneficiary who inherits a tax-deferred account will receive an income tax deduction for any additional estate tax that was caused by the tax-deferred account.

Example: You have a $1,000,000 traditional IRA and enough other assets that the full IRA is subject to the maximum 40% estate tax. And let’s assume that both you and your beneficiary have a 37% marginal tax rate (i.e., trying to eliminate that as a variable here, in favor of or against the conversion).

If you don’t do a conversion, the $1,000,000 IRA is subject to estate tax. At the 40% rate, that’s $400,000 of estate taxes. The entire $1,000,000 IRA is subject to income tax, but the beneficiary receives a deduction (IRD deduction) for the $400,000 of estate taxes paid. So it’s a net $600,000 of taxable income. At a 37% marginal rate, that’s $222,000 of income tax. So, after estate tax ($400,000) and income tax ($222,000) we have the beneficiary actually inheriting $378,000, net.

Now let’s assume instead that you do convert the whole amount. $1,000,000 conversion, at a 37% rate, leaves us with $630,000 in a Roth account. That $630,000 Roth account is subject to the 40% estate tax ($252,000). And that leaves us with $378,000 that the beneficiary receives. Key point being: same amount as in the scenario above.

However, there are three caveats here:

1. Minor caveat: the IRD deduction is an itemized deduction, so some portion of it may be “wasted” in overcoming the standard deduction for the year. But with the dollar amounts we’re talking about in the case of an estate that exceeds the exclusion amount, that’s not really a big deal.
2. Big caveat: many beneficiaries probably don’t know about the IRD deduction at all. I wouldn’t be surprised to see cases in which that available deduction simply does not get claimed, in which case the conversion (in our example above) would have been very preferable.
3. Big caveat which might not be relevant: with respect to state estate taxes (as opposed to federal), there’s no IRD deduction. So in cases in which we expect a state estate tax to apply, that is a significant point in favor of conversions.

Inherited IRA Rules

One thing I noted repeatedly in the presentation is that money in a Roth IRA can stay in the account for the rest of your life, plus up to 10 years (i.e., before your beneficiaries have to take the money out of the account). For married couples, the length of time in question is really “the rest of the longer of your two lives, plus up to 10 years” (because the surviving spouse can simply roll an inherited Roth IRA into their own Roth IRA and RMDs won’t begin until after that second spouse’s death).

And in some cases it could plausibly be longer than that, if, for example, you’re leaving your IRA to somebody who is disabled. The rules for inherited IRAs are extremely complex (this article — especially the flowcharts — is an excellent reference).

Marginal Tax Rate ≠ Tax Bracket

The other thing people asked about most often was the category of things that cause your marginal tax rate to be different than your tax bracket. There’s simply no way to cover all of them, because there are too many. I picked three that are very commonly applicable in retirement scenarios (premium tax credit phaseout, Medicare IRMAA, and the unique way Social Security benefits are taxed). But there are so, so, many more things that cause this type of effect. There’s the net investment income tax, medicare surtax, and really anything at all in our tax code that becomes applicable at a specific income level. And that’s a lot of things.

Any of these various provisions in our tax law can potentially be a relevant factor in a conversion analysis. That’s why I always encourage people to use actual tax software rather than DIY calculations.

One thing I run into frequently — both working with clients and via correspondence with readers — is people who have financial goals that, to put it bluntly, are not very good.

And to be clear, I just don’t mean that they’re goals I wouldn’t pick. I get excited about spending money on spring loaded camming devices and trips to the mountains. But if you want to spend your money on beach vacations, playing golf at expensive resorts, or whatever else it is that makes you happy, by all means go for it.

What I mean is that I see a lot of financial goals that are neither personally meaningful nor even useful for actual financial planning. For instance:

• “I want to reduce my RMDs” is not a good financial goal.
• “I want to avoid Medicare IRMAA in the future” isn’t a good financial goal.
• “I want to make sure my Social Security isn’t taxable” isn’t a good financial goal.
• “I want to make sure my LTCGs will be taxed at a 0% rate in retirement” isn’t a good financial goal.
• “I want to convert [a certain amount or percentage] of my tax-deferred accounts by [age/date]” isn’t a good financial goal.

All of those things should be considered in the analysis. It might make sense to do those things. But they shouldn’t be goals.

Goals should be things like:

• “Increase the likelihood that I’ll be able to spend at least $X per year for the rest of my life.”
• “Increase the amount I’m likely to leave to my kids, after taxes.”
• “Be able to donate $X per year while still being able to satisfy our desired level of spending.”
• “Spend $20,000 extra per year in the first 5 years of retirement to take some trips we’ve always wanted to take — without putting our financial security at risk.”

For instance, with the stated goal, “I want to reduce my RMDs,” well, there are various actions we could take that would definitely reduce your future RMDs. But what if we do some modeling and it turns out that those actions would probably increase the likelihood of portfolio depletion during your lifetime? In such a case, why would we want to do it? As soon as “I want to reduce my future RMDs” comes up against a better financial goal (“I don’t want to run out of money”) the RMD-related goal gets discarded immediately.

As an obvious example, imagine that, in your traditional IRA, you chose to use an expensive stock mutual fund that ultimately underperforms the market by 2% per year. Relative to using a low-cost index fund, using the expensive fund would have the result that your traditional IRA is ultimately smaller, thus your RMDs are smaller, your lifetime tax bill is smaller, your Medicare premiums might be lower, and so on. But that’s not a win. You’re just paying less taxes because you have less money.

In the context of more complicated topics (e.g., which dollars to spend in retirement, whether to do Roth conversions, when to file for Social Security, asset location decisions, etc.), this phenomenon is less obvious, given that there are so many moving parts. But it’s still very important to be focused on the correct metrics. If the software you’re using (or your DIY analysis in a spreadsheet) tells you that a given decision has the effect of, for instance, reducing your RMDs or reducing your lifetime tax bill, that’s not good enough. You need to know more than that, otherwise you could be unwittingly making a decision that falls in the category of “paying less taxes because I have less money.”

When setting goals, I find that sometimes it’s helpful to imagine you have a small child playing devil’s advocate. For instance, for some people it might go something like this:

“I want to reduce my future RMDs.”

“Why?”

“I want to reduce the amount of taxes I’ll have to pay.”

“Why?”

“I’m concerned that taxes will make me more likely to run out of money during retirement.”

And that is what this person is actually concerned about.

Admin note: I have the annual Bogleheads Conference coming up this week, followed immediately by some vacation travel and then a finger surgery which will require a couple days of not-very-productive recovery. So there will be a temporary publishing hiatus here, with the next article appearing October 21.

If you have read about retirement tax planning, you have likely read about how people’s tax rate in retirement is often much higher than they’d expected. You may have even heard it referred to as a “tax torpedo.”*

The idea, broadly speaking, is that there are assorted provisions in our tax code that result in situations where, as your income goes up, not only does your tax go up in keeping with your tax bracket, it also causes something else to happen (e.g., some other tax kicks in) — which leads to your actual marginal tax rate being much higher than just your tax bracket.

In some cases we can see marginal tax rates in retirement exceeding 50%, especially when we include state income tax.

That sure sounds scary, doesn’t it? Sounds like it could be a major risk to your financial security in retirement.

But here’s the key point: those high tax rates are marginal tax rates. And at the levels of income in question, the effective tax rate is usually still very low.

Just to back up a step and make sure everybody is clear on definitions:

• Your marginal tax rate is the tax rate you would pay on an additional amount of income. (For example if your income went up by $100 and your tax bill went up by $30, your marginal tax rate for that $100 of income was 30%.)
• Your effective tax rate is the total amount of income tax you pay, divided by your total amount of income.

Let’s run through a few simple examples of relatively common retirement income profiles. You’ll see what I mean about high marginal tax rates and not-that-scary effective tax rates.

For all examples we’re assuming it’s 2024, using current tax law. We’re looking only at federal taxes. And for all examples we’re assuming a married couple filing jointly, both age 65 or older. (The exact same concepts apply to people filing as “single.” I’m simply keeping the filing status the same from one example to another in order to make it easier to compare.)

The tax calculations were done using Holistiplan.

Example 1:

• $40,000 of ordinary income (e.g., any combination of taxable interest or taxable distributions from a traditional IRA)
• $10,000 qualified dividends/long-term capital gains
• $40,000 Social Security benefits

Their marginal tax rate for additional ordinary income is 22.2% (12% bracket, but each dollar of income is also causing $0.85 of Social Security to become taxable at a 12% rate). But they have $90,000 of income, and their total federal income tax is just $3,832. That’s an effective tax rate of just 4.3%.

Example 2:

• $80,000 of ordinary income (e.g., any combination of taxable interest or taxable distributions from a traditional IRA)
• $10,000 qualified dividends/long-term capital gains
• $44,000 Social Security benefits

Their marginal tax rate for additional ordinary income is 27% (higher than just their tax bracket, because each additional dollar of ordinary income is causing a dollar of QD/LTCG income to get pushed from the 0% tax rate range into the 15% tax rate range). But they have $134,000 of income, and their total federal income tax is $9,906. That’s an effective tax rate of just 7.4%.

Example 3:

• $60,000 of ordinary income (e.g., any combination of taxable interest or taxable distributions from a traditional IRA)
• $10,000 qualified dividends/long-term capital gains
• $80,000 of Social Security benefits

Their marginal tax rate for additional ordinary income is 49.9% (!!!). This is because each additional dollar of income is causing $0.85 of Social Security to become taxable at a 22% rate, while also pushing a dollar of QD/LTCG income into the 15% rate range. And yet, on $150,000 of income, their total tax is $11,175. An effective tax rate of 7.5%.

Example 4:

• $250,000 of ordinary income (in this case, assuming taxable distributions from a traditional IRA)
• $40,000 qualified dividends/long-term capital gains
• $70,000 Social Security benefits

Their marginal tax rate for additional ordinary income is 24%. That’s because, in this case, we’re past the “weird” stuff that’s relevant in the examples above. That is, 85% of their Social Security is already included in their gross income, so that effect is no longer a concern for additional income. And all of their capital gains are already being taxed at the 15% rate. So they’re just in the regular 24% bracket. They have $360,000 of total income, and their total federal income tax is $60,133. That’s an effective tax rate of 16.7%. Definitely higher than in the previous examples, but that’s what you’d expect with a much higher level of income.

For tax planning (especially with respect to retirement accounts), we generally care about marginal tax rates. Marginal tax rates are what matter when we’re trying to figure out things like:

• Whether to contribute to Roth or tax-deferred accounts while still in our earning years.
• Whether to spend from Roth or tax-deferred accounts in our retirement years.
• Whether to do a Roth conversion in any given year.

For those decisions, we exclusively care about marginal tax rates. We don’t want to consider effective tax rates at all. And that’s why you hear so much about these high marginal tax rates in retirement. They are an important factor in many decisions we have to make.

But for budgeting, we care about effective tax rates. If we want to know whether taxes pose a risk to your financial security in retirement, we care about your effective tax rate in retirement. That is, how much total tax are you paying? Is it likely to be so much that it puts you at risk of depleting your savings?

*I really dislike the term “tax torpedo,” because it gives the impression that it’s a big danger to people, when, for the reasons discussed above, it tends not to be. Still, it’s the term that has become popular, so I’m acquiescing here.

The topic that I help clients with most often is retirement tax planning. And my approach to that work includes modeling different tax strategies in financial planning software.

When I do that, what I find is that a plan for tax-efficient spending (i.e., which dollars to spend each year in retirement) often has a significant impact on retirement safety. For example, it often has the result of:

• Reducing the projected probability of portfolio depletion by several percentage points (e.g., from 20% to 14%), while also
• Pushing back the date at which those depletion scenarios do occur (e.g., so that in the “unlucky” scenarios, the household now runs out of savings in their mid 90s instead of late 80s).

But once you have a tax-efficient spending plan in place, a Roth conversion plan does not usually result in any meaningful improvement to retirement safety. That is, conversions don’t typically improve either of those metrics above by a meaningful amount.

And that has been my experience with a broad range of clients — ranging from super financially secure to more borderline, with varying ages upon retirement, and with a wide variety of portfolio sizes and compositions. And that has been the case using a broad variety of assumptions (e.g. varying assumed lifespans, average investment returns, etc.).

Roth conversions simply do not tend to increase retirement safety.

Why?

This is a simplification, but I often think of Roth conversions as a tool for solving/alleviating two “problems.” (For more on these topics, please see this article: The 4 Effects of Roth Conversions.)

1. Required minimum distributions (RMDs) from tax-deferred accounts (both during your lifetime and after your death).
2. The ongoing tax drag that occurs in taxable accounts (i.e., by letting you use taxable dollars to pay the tax on a conversion — thereby effectively using your less-tax-efficient taxable dollars to buy very tax-efficient Roth space).

But neither RMDs nor tax drag in taxable accounts is on the list of “stuff that’s likely to cause portfolio depletion in retirement.”

RMDs simply do not cause people to go broke. (In fact, RMDs are often recommended as a strategy for spending from a portfolio.)

And in unlucky retirement scenarios (e.g., poor investment returns early in retirement or a major spending shock early in retirement), the problem of tax drag in the taxable account typically solves itself, because the taxable account usually gets spent down pretty quickly in those scenarios anyway.

So Why Bother with Roth Conversions?

Given the above, you might ask why a person would bother with Roth conversions. The answer is that, in cases in which Roth conversions are suitable (which, for many but not all households, is in the years after retiring but before collecting Social Security and before RMDs kick in), they provide a significant increase in the the after-tax bequest that is likely to be left to heirs. (Again, please see this article for a discussion of the mechanics.)

In other words, a well-crafted Roth conversion plan typically:

1. Does not make the unlucky outcomes significantly better or worse and
2. Makes the lucky outcomes significantly better.

When I say “a well-crafted Roth conversion plan,” what I mean is a plan that carefully considers whether conversions should be done each year, and if so, to what threshold. Roth conversions are not suitable for everybody. For instance, households with large charitable intent are much less likely to benefit from conversions, because they can use qualified charitable distributions to satisfy RMDs and because the after-death tax rate on their tax-deferred accounts will be 0% to the extent the balances are left to charity.

A reader writes in, asking:

“We are getting closer to retirement and beginning to adjust our asset allocation. Recently we rebalanced our asset allocation from 90/10 stocks/bonds to 70/30. It was the first time we rebalanced in about 7 years. We think given our time horizon we should consider 50/50 or even 40/60. It’s a very difficult decision.

In addition, we’re trying to figure out how often we should be rebalancing going forward as we move into retirement.

How do we figure out what is the best rebalancing frequency for our funds held at Vanguard: Total Stock Market Index Fund, Total Bond Market Index Fund and Intermediate Term Bond Index Fund? Those funds are our complete retirement portfolio…trying to make you proud…KISS. You helped us so much in the 10+ years that we have been following you.”

First a note on terminology, because it may cause some misunderstandings when reading the links I’m about to provide: the change that you describe having recently made is not rebalancing. Rebalancing is when you bring your allocation back to the intended (target) allocation. For example, if the target is a static 60/40 allocation and every quarter you make adjustments to bring the portfolio back to the 60/40 allocation, that’s rebalancing. If you change the target (e.g., deciding instead that a 40/60 allocation is the new target), that’s not rebalancing.

This is not to say that changing the target is a bad idea. Sometimes it’s a good idea — especially as your life circumstances change. I’m just belaboring this terminology point, because when reading about rebalancing in more technical writing, it’s important to know very specifically what is being discussed. (This is a common terminology mix-up, by the way. People get it wrong constantly on the Bogleheads forum for instance.)

And with that out of the way, the following are a few things you may want to read.

• Momentum (classic 1999 article from Bill Bernstein)
• An In-Depth Look at Portfolio Rebalancing Strategies (2015 research paper from Michael Kitces)
• Finding The Optimal Rebalancing Frequency – Time Horizons vs Tolerance Bands from Michael Kitces

A takeaway from reading the three articles above is that rebalancing more often than annually is likely not a great idea. In very brief, the reason is that the stock market has historically exhibited a slight degree of momentum over periods shorter than a year. That is, if yesterday was a good day, today is more likely than usual to be a good day. And if yesterday was a bad day, today is more likely than usual to be a bad day. And the same goes for monthly periods.

And the result is that rebalancing daily or monthly would mean that, in a market downturn, you’re constantly buying more stocks as they keep falling, resulting in an overall loss that’s worse than if you hadn’t been rebalancing. And during upward markets, you’re constantly selling stocks, resulting in less of a gain than if you hadn’t been rebalancing.

The following two links are runs from PortfolioVisualizer, comparing monthly vs annual rebalancing, for a basic 3-fund portfolio using a “4% rule” spending strategy. Rebalancing annually worked out slightly better in terms of return, maximum drawdown, and standard deviation. (Note that I’m simply using the earliest start date available here, and letting it ride until today. If interested, you could instead test with rolling 30-year periods, for instance, to see how reliable this outcome is. You could also test with different target allocations or with different spending strategies.)

• Rebalancing annually
• Rebalancing monthly

Plot Twist: Contrary Evidence

There’s also, however, a 2010 paper from Vanguard (no longer on their website, but here’s a Web Archive link), which found essentially no difference between rebalancing monthly, quarterly, or annually — other than the time (and potentially transaction costs) involved in doing so.

Also, as always, anything based on historical data — as all of the above is — should be treated with a healthy degree of skepticism. Sometimes, trends that persisted for a very long period, even many decades, eventually disappear, as the markets themselves change (e.g., as the participants in the market shift, as products available change, as laws/regulations change, etc.)

And indeed, per a 2022 paper, it appears that that’s exactly what has happened:

• Autocorrelation of stock and bond returns, 1960–2019 from Jonathan Lewellen of Dartmouth College

In this paper, the author found that the autocorrelation of stock returns (i.e., the correlation from yesterday’s returns to today’s returns) declined over the period 1960-2019 and actually became significantly negative in the second half of the sample. That is, yesterday being a good day would mean today is more likely than usual to be a bad day, and vice versa. And that would mean that rebalancing everyday (as you would see in a target-date or LifeStrategy fund) would actually be helpful.

So, where does all of this leave us?

Frankly, I really don’t know, other than to say that there’s some good evidence in favor of just about any option. My personal thinking at this point can be summarized as follows:

• If you have a target-date fund, LifeStrategy fund, or anything similar which is rebalancing for you daily, that’s probably fine. (Though it can create tax costs in a taxable account.)
• If you’re using a DIY allocation, and you want to rebalance quarterly, annually, or every two years (or “annually but only if the allocation is off-target by at least x%”), that’s probably fine too.
• More frequent rebalancing means more work, if you’re doing the rebalancing yourself.
• I wouldn’t worry too much about this topic overall. Nor would I put too much faith in Strategy A instead of Strategy B. It’s more along the lines of “pick one approach that seems reasonable, and stick with it.”