Research Memo
Four placements, and the conditions under which each is correct
Framework memo. Illustrative throughout; no client facts are described. Nothing here is tax advice.
This section stands alone. Each section that follows states a placement decision and then the conditions that must hold for it to be the right one.
The conventional asset location rule, income assets in the retirement account and equities in the taxable account, is correct under a specific configuration of assumptions and incorrect outside it. The assumptions are rarely stated, which is why the rule is applied more broadly than it deserves.
This memo does not argue that the conventional rule is wrong. It argues that it is one answer among four, and that identifying which situation a client is actually in matters more than knowing the rule.
Asset location is not a question about assets. It is a question about the interaction between an asset’s tax characteristics and an account’s tax treatment, evaluated over the client’s actual horizon and terminal disposition. Change the horizon or the disposition and the same asset belongs somewhere else.
Income assets in taxable (§2). Correct when yields are low enough that there is little to shelter, when the income is already tax-exempt, when the horizon is short, or when the position is serving as an accessible reserve.
Income assets in tax-advantaged (§3). Correct when yields are meaningful and the current bracket is high. Interest is the least tax-efficient income a portfolio produces: fully ordinary, received annually, with no control over timing.
Equities in taxable (§4). Correct when the terminal disposition is a bequest or a charitable gift, when the strategy generates tax attributes that only have value in a taxable account, or when turnover is low enough that the preferential rate on qualified dividends and long-term gains does most of the work already.
Equities in tax-advantaged (§5). Correct when the account is a Roth and the assets have the highest expected return in the portfolio, or when the strategy has high turnover and generates short-term gains. Correct far less often in a traditional IRA, for reasons developed in that section.
The matrix is symmetric, but the answers are not. For income assets, the choice between traditional and Roth barely matters, since ordinary income is being sheltered either way. For equities the choice matters enormously, because a Roth is entirely the client’s while a traditional account is shared with the Treasury. That asymmetry is developed in §5.1 and it is the single most useful idea in this memo.
A strategy whose entire return is tax alpha is worthless in a sheltered account. Loss harvesting produces nothing in a retirement account because there is no tax to offset. This is obvious once stated, but it is worth stating as a general principle: shelter space should go to the assets whose returns are taxed most punitively, not to the strategies that are most complex. The two are unrelated. §4.2.
The placement. Bonds, bond funds, and other income-generating securities held in the taxable account rather than sheltered.
This inverts the conventional rule, and there are four distinct conditions under which it is correct. They are unrelated to each other, which is worth noticing: the conventional rule can fail for several independent reasons.
The value of sheltering interest is the interest rate multiplied by the tax rate multiplied by the horizon. When yields are low, the first term is small and the whole benefit compresses toward nothing.
A great deal of asset location convention was formed during a prolonged low-yield period, when the argument for sheltering bonds was weaker than it had ever been and the argument for sheltering equities correspondingly stronger. When yields rose, the arithmetic changed. Any rule of thumb inherited from that period should be re-derived rather than applied.
The practical test is straightforward: compute the annual tax cost of holding the income asset in taxable, and compare it against the opportunity cost of the shelter space it would occupy. Shelter space is finite and it has a next-best use.
Municipal bonds held in a tax-advantaged account are the clearest error in asset location practice. The security’s entire advantage is that its interest is exempt from federal tax. Placing it in an account where all income is already untaxed converts that advantage into nothing, while the investor continues to accept the lower yield that the exemption commands.
Worse, in a traditional account the exempt interest is eventually distributed as ordinary income. The client has taken a below-market yield to purchase an exemption and then converted the proceeds into fully taxable income.
The general principle: do not shelter something that is already sheltered. It applies to munis most starkly, but also to any asset whose return is already tax-favored.
The benefit of tax deferral compounds. Over thirty years it is substantial; over three years it is close to immaterial. A client approaching a spending goal, or one whose planning horizon is genuinely short, gains little from optimized placement and should not accept operational complexity or liquidity constraints to obtain it.
Income assets frequently serve a function beyond return: they are the source of liquidity for spending, the dry powder for rebalancing, and the buffer that prevents forced equity sales in a drawdown.
None of those functions work inside a retirement account for a client under 59½, and all of them are impaired by the ordinary-income treatment of withdrawals for a client above it. A bond position that exists to be spent or deployed belongs where it can be spent or deployed.
This is the condition most often overlooked, because it is not a tax argument at all. Placement optimized purely for tax can defeat the reason the asset was purchased.
The placement. Bonds, bond funds, REITs, and other ordinary-income-generating assets held in a traditional IRA, Roth IRA, or employer plan.
This is the conventional rule, and it is right more often than it is wrong. The conditions are straightforward, so this section spends most of its length on the distinction between account types.
The case is strongest when yields are meaningful, the client’s current bracket is high, and the horizon is long. Interest is the least tax-efficient income a portfolio produces: taxed at ordinary rates, received annually whether wanted or not, with no ability to defer recognition or control timing. Every structural advantage a taxable account offers, deferral, preferential rates, timing control, step-up, is unavailable to it.
The case extends naturally to assets that produce ordinary income without being bonds. REITs distribute the majority of their income as ordinary dividends. Commodity strategies and many alternatives generate ordinary income, short-term gains, or both. High-turnover strategies of any kind generate short-term gains taxed at ordinary rates.
These are frequently the cleanest asset location decisions available, and they are rarely the ones that get debated.
For income assets, the choice between traditional and Roth is genuinely less consequential than it is for equities. Both shelter the annual interest from current taxation, which is the primary benefit being sought. The remaining difference concerns when tax is paid rather than whether the income is sheltered.
This is not a reason to be indifferent, but it is a reason to allocate attention accordingly. The traditional-versus-Roth decision does substantially more work for equities, for reasons developed in §5.1.
One configuration deserves its own treatment because the reasoning is not obvious.
Consider a client in a high bracket during their working years who expects a materially lower bracket in retirement. The standard analysis says contribute to a traditional account: deduct at the high rate, withdraw at the low one.
There is a further step. Holding modest-growth assets in that traditional account, and converting to a Roth after the bracket falls, can improve the outcome beyond the simple bracket arbitrage, and the reason is mechanical rather than about returns.
Conversion income stacks. A Roth conversion is ordinary income in the year it occurs, layered on top of everything else. A client with a low retirement bracket has room to convert cheaply, but only a limited amount of room before the conversion pushes them into higher brackets. A large traditional balance cannot be converted efficiently for exactly that reason: it takes many years of partial conversions, or it spills into brackets that eliminate the advantage.
Keeping the traditional balance modest therefore preserves the option to convert it inside the low brackets. It also keeps required minimum distributions manageable, which matters because RMDs are involuntary and can push a client into brackets and surcharge thresholds they were otherwise below.
A traditional account is not simply a tax-deferred bucket. It is a future stream of ordinary income whose size the client partly controls today through what they place in it.
Placing the highest-growth assets there maximizes that future stream, which is the opposite of what a client planning conversions wants. This runs directly counter to the intuition that the retirement account should hold whatever grows fastest. Whether it applies depends entirely on the client’s bracket path, which is why the diagnostic in §6 asks about it first.
The placement. Equity exposure, individual stocks, index funds, ETFs, held in the taxable account.
The conventional rule reaches this conclusion by elimination: bonds go in the shelter, so equities go in taxable. That reasoning is weak, but the conclusion is frequently right for reasons of its own.
This is the strongest argument in the memo, and it has nothing to do with the asset’s characteristics during the holding period.
Step-up at death. Appreciated securities held in a taxable account receive a basis adjustment to fair market value at death. The embedded gain is eliminated and never taxed. No retirement account offers this. A traditional IRA is taxed in full as ordinary income to the beneficiary; a Roth passes tax-free but had its tax paid on contribution.
For any client whose intention is to hold an appreciated position until death, the taxable account is not merely acceptable. It is the only account where the gain disappears.
Charitable disposition. A gift of long-term appreciated securities to a public charity or donor advised fund produces a deduction at fair market value while the embedded gain is never recognized by anyone. Assets held inside a retirement account cannot be given this way. A qualified charitable distribution from an IRA is available to eligible clients but is subject to an annual limit and is a different mechanism entirely.
Both of these convert what looks like a liability during the holding period, accumulated unrealized gain, into an advantage at disposition. Which means the tax drag measured over the holding period is the wrong measure if the position is never sold.
A tax loss harvesting or direct indexing strategy produces its return in the form of realized losses that offset gains elsewhere. That return exists only in a taxable account. Inside a retirement account there is no tax to offset, no gain to shelter, and no mechanism by which a harvested loss produces anything at all.
A strategy whose entire excess return is tax alpha is worth exactly zero in a sheltered account.
The instinct to place the most sophisticated or most actively managed strategy inside the retirement account is strong and, in this case, precisely inverted. The same logic applies to any strategy whose value derives from tax attributes: foreign tax credits from international holdings are creditable only against tax actually owed, and are forfeited inside a retirement account.
Equity returns arrive in tax-favored forms. Qualified dividends and long-term capital gains are taxed at preferential rates, and unrealized appreciation is not taxed at all until sold. A low-turnover equity position therefore carries a modest and largely deferrable tax burden even without any shelter.
Placing that asset inside a traditional account converts its preferential-rate returns into ordinary income on withdrawal. This is a genuine cost that the conventional framing, which treats the retirement account as unambiguously advantageous, obscures entirely.
High-turnover equity strategies do not qualify. A strategy realizing short-term gains annually produces ordinary income with no deferral and no preferential rate, and it belongs in a sheltered account for the same reasons bonds do. The characteristic that matters is not the asset class but the tax character of the return it generates.
The placement. Equity exposure held inside a Roth IRA, traditional IRA, or employer plan.
This is where the traditional-versus-Roth distinction becomes decisive, and where the memo’s central asymmetry lives.
A Roth account is entirely the client’s. A traditional account is shared with the Treasury: a balance of $X with a future withdrawal tax rate of t is economically worth $X(1−t) to the client. The Treasury holds a claim on the remainder that grows in exact proportion to the account.
The consequence for placement is direct. Growth inside a Roth accrues entirely to the client. Growth inside a traditional account is shared with the Treasury at whatever rate applies on withdrawal, a rate that could rise as the balance grows.
So if one account in the portfolio should hold the assets with the highest expected return, it is the Roth. Not because the return is sheltered, it is sheltered in both, but because the client owns all of the upside in one and a fraction of it in the other.
This is the argument that most cleanly overturns the conventional rule, and it produces a specific ordering: highest expected return assets in the Roth, ordinary-income assets in the traditional account, and assets destined for bequest or charity in taxable.
Three costs attach to holding high-growth equity in a traditional account, and none of them appear in the simple deferral framing.
Character conversion. Long-term capital gains and qualified dividends are taxed at preferential rates in a taxable account. Withdrawn from a traditional account, the same economic return is ordinary income. The shelter has converted favorably-taxed returns into unfavorably-taxed ones, a real cost that partially or wholly offsets the deferral benefit.
No step-up. The single largest advantage available to appreciated equity is unavailable inside any retirement account. An asset the client intends to hold until death is close to the worst possible use of traditional account space.
RMD amplification. Required minimum distributions are calculated from the account balance. Placing the fastest-growing assets in the traditional account maximizes the future balance and therefore maximizes involuntary ordinary income later, potentially at rates the client cannot control and in years when it pushes them across bracket and surcharge thresholds.
The conclusion is not that equities never belong in a traditional account. It is that the case is considerably weaker than the conventional rule implies, and that it rests on the client’s bracket path rather than on any property of equities.
High-turnover equity strategies, and any equity strategy generating short-term gains, belong here. So does equity exposure a client expects to trade actively, rebalance frequently, or reposition, since the shelter removes the tax friction that would otherwise penalize the activity.
Shelter the assets whose returns would otherwise be taxed at ordinary rates, and place in taxable the assets whose returns are already preferentially taxed or will never be taxed at all.
Four questions identify which of the four placements applies. They are ordered by how much they move the answer.
1. What is the terminal disposition of the asset: spent, bequeathed, or given? A bequest makes step-up available and pushes appreciated equity toward taxable. A charitable disposition does the same. Spending during life removes both arguments and returns the decision to a comparison of tax drag against deferral.
2. What is the bracket now, and what is it expected to be at withdrawal? A falling bracket path favors the traditional account and opens the conversion planning described in §3.3. A rising or flat path favors the Roth. This assumption does more work than any assumption about the assets themselves.
3. How long is the horizon? The benefit of optimized placement compounds with time. Under roughly a decade, the differences narrow enough that liquidity, simplicity, and constraint management should take precedence.
4. Does the strategy generate tax attributes that only have value in a taxable account? Harvested losses and foreign tax credits are worth nothing inside a shelter. If the strategy’s return depends on them, the placement question is already settled.
A fifth question is practical rather than analytical, and it frequently overrides all four: what placements are actually available given existing balances, contribution limits, and the client’s current holdings? Asset location is a constrained optimization, and in most real portfolios the constraints bind before the theory does.
Illustrative framework prepared for discussion. Not tax advice. Placement decisions depend on individual circumstances and should be confirmed with qualified tax counsel.
CFA® and Chartered Financial Analyst® are registered trademarks owned by CFA Institute.