Direct Indexing vs ETFs

What it actually is

A separately managed account holds, say, 250 of the S&P 500's constituents, weighted to track the index closely while omitting the smallest and least impactful names (Chapter 37's sampling logic, applied at the individual-investor level). Software rebalances, reinvests dividends, and, the differentiating function, scans continuously for positions trading below their cost basis and sells them to realise losses, replacing them with correlated substitutes to maintain exposure.

The realised losses offset capital gains elsewhere in the investor's affairs: other portfolio gains, a business sale, concentrated stock diversification. This is the product. Everything else is secondary.

The tax-loss harvesting case, honestly quantified

Why the mechanism works: in any year, even a strongly rising index contains individual stocks that fell. A fund holder sees only the fund's price, which is up, and has nothing to harvest. A direct indexer harvests the individual losers while the overall position gains.

How much is it worth? Estimates of "tax alpha" commonly cited by providers run around 0.5–2% a year, and the honest reading of that range requires several caveats:

• It is deferral, not elimination. Harvesting lowers your cost basis, so tax is postponed rather than removed. The benefit is the value of deferral, plus the possibility of lower future rates, plus, significantly, the step-up at death (Chapter 52), which can convert deferral into permanent avoidance.
• It depends entirely on having gains to offset. An investor with no other capital gains gets limited annual benefit, since the offset against ordinary income is capped.
• It decays. Early years produce the most harvesting; as the portfolio appreciates, fewer positions sit below basis and the opportunity shrinks toward zero. Provider estimates often reflect early-year results.
• It requires volatility. Calm markets produce few harvestable losses; the mechanism works best in exactly the periods investors enjoy least.
• Wash-sale rules constrain it (Chapter 52), and the substitutes used to maintain exposure introduce their own tracking noise.

A fair summary: for a large taxable account with substantial ongoing gains, in the early years, in volatile markets, the benefit can genuinely exceed the fee difference. Outside that intersection, it frequently does not.

The customisation case

The second pitch: exclude what you choose. Real applications, in descending order of substance:

Concentrated position management. An executive holding 30% of their wealth in employer stock can direct-index the rest of the market excluding that stock and its sector, avoiding doubling down (Chapter 42's best-use logic, executed precisely). This is the strongest customisation case and, notably, one an ETF genuinely cannot serve.

Values-based exclusion. Removing specific companies rather than accepting a provider's ESG screen (Chapter 46). Real, though a screened index fund achieves much of it at a fraction of the cost.

Tilts. Systematic factor or sector adjustments (Chapter 45's discipline still applies, and the fee now stacks on top).

Charitable giving. Donating the most-appreciated individual lots is more efficient than donating fund shares with a blended basis.

The costs, plainly

• Fees: 0.20–0.40% typical, against 0.03–0.10% for the equivalent ETF. That gap must be earned back every year.
• Tracking will be looser. Sampling, harvesting substitutions and rebalancing frictions mean the account will not match the index as cleanly as a well-run fund (Chapter 35's craft, performed with more constraints).
• Operational load. Hundreds of positions, a long tax report, corporate actions, and a relationship with a manager rather than a ticker.
• Lock-in. As gains accumulate across hundreds of lots, exiting becomes progressively more expensive. Direct indexing is easier to enter than to leave, which is worth understanding before entering.
• It is worth nothing in tax-advantaged accounts. No tax to harvest, so the entire proposition evaporates inside an IRA, 401(k), ISA or pension.

Who it is actually for

A good fit:

• Large taxable accounts, commonly cited thresholds starting around $250,000–500,000 and improving with size.
• Investors with recurring capital gains to offset: other portfolios, business interests, property.
• Concentrated position holders needing complement exposure that excludes specific names.
• Investors with specific exclusion requirements no fund satisfies.
• Those planning to hold to the step-up at death, which converts deferral into avoidance.

A poor fit:

• Anyone investing primarily in tax-advantaged accounts.
• Smaller taxable accounts where the fee difference outweighs plausible harvesting.
• Investors without other gains to offset.
• Anyone who values simplicity; and Chapter 64 argues that this is a financial preference, not merely an aesthetic one.

Why this is largely a US story

Direct indexing's engine is US tax law: the individual-lot harvesting mechanism, the wash-sale rules that shape it, and the step-up at death that can make deferral permanent.

In Europe, the proposition weakens considerably. Loss-offset rules differ by country and are frequently more restrictive; notional taxation regimes (Germany's Vorabpauschale, Sweden's ISK, the Netherlands' box 3 — Chapter 53) can make individual-lot harvesting nearly meaningless; and the wrapper cultures (ISA, PEA, pensions) already remove the tax the strategy exists to manage. There is a customisation case in Europe, but the tax engine is largely absent.

In Asia, jurisdictions without capital gains tax on foreign securities (Singapore, Hong Kong) have no harvesting proposition at all, which removes the entire primary rationale.

Is it a threat to ETFs?

The industry narrative has occasionally framed direct indexing as the ETF's successor. A sober assessment:

• It competes for a specific segment, large taxable accounts with advisers, and not for the broad market.
• It does not scale down. The operational cost per account does not fall with size, so small accounts remain uneconomic even as software improves.
• The ETF's own tax efficiency limits the upside (Chapter 54): the fund already avoids distributing gains, so direct indexing's advantage is confined to the harvesting layer.
• ETF share classes and conversions (Chapter 78) are the industry's competing answer, and they scale.

The reasonable expectation is coexistence: direct indexing as an advised, high-net-worth taxable-account product, and ETFs as everything else: which is roughly where the market already sits.

A worked example

Two US investors, each $600,000 in a taxable account tracking the same index.

• Anna holds an ETF at 0.04%. Annual fee: $240. Her capital gains distributions are essentially zero (Chapter 54), and she harvests only if the whole position falls below basis, which after a few good years it does not.
• Ben direct-indexes at 0.30%. Annual fee: $1,800. In a volatile first year the manager harvests $34,000 of losses, which offset gains from selling a rental property. At his rate, that saves roughly $6,800 in tax: comfortably exceeding the $1,560 fee gap.
• Year six: the account has appreciated substantially, harvesting opportunities have thinned to $6,000 in the year, and Ben has no other gains to offset. The benefit that year is limited by the ordinary-income offset cap; his fee gap remains $1,560, and he now holds 240 positions with embedded gains that make leaving expensive.

Ben's first year justified the product decisively. His sixth did not, and his exit is now costly. That trajectory: front-loaded benefit, decaying opportunity, increasing lock-in: is the shape a prospective direct indexer should have in mind, and it is not the shape the marketing describes.