Semi-Transparent and Non-Transparent Structures

The bind these structures address

An active manager's objection to the ETF wrapper is specific and not unreasonable:

• Front-running. If tomorrow's basket shows the fund accumulating a position, others can buy ahead of the remaining purchases, raising the manager's cost.
• Free-riding. Competitors and copycats can replicate a research-driven portfolio for nothing.

The counter-argument, equally serious: daily transparency is what makes ETF arbitrage tight, and tight arbitrage is what gives investors a fair price and a narrow spread. Any structure that weakens disclosure weakens the mechanism, and the cost lands on the fund's own shareholders in basis points.

So the question was never "is secrecy nice for the manager" but "is the strategy's protected edge worth more than the execution quality shareholders give up?"

The approved structures, and how each works

1. Proxy basket (substitute basket) models. The fund publishes a basket that is not its portfolio but is constructed to track it closely: typically containing many of the actual holdings plus substitutes and cash, with disclosure of the degree of overlap rather than the composition. Market makers hedge with the proxy and accept the residual mismatch.

2. Verified intraday indicative value (VIIV) models. The fund keeps holdings confidential but publishes a per-second, independently calculated indicative value with pricing verification. Market makers hedge statistically against correlated instruments, pricing off the VIIV rather than the basket. This asks more of the market maker, which shows up in the spread.

3. Trusted-agent (confidential account) models. An agent knows the full portfolio and executes creations and redemptions on behalf of an AP inside a confidential account, so the AP transacts without ever seeing the holdings. This preserves in-kind mechanics, and therefore the tax treatment of Chapter 54, while keeping the portfolio private.

4. Periodic disclosure. Holdings published monthly or quarterly with a lag, which is a mutual-fund convention transplanted into an ETF, with the arbitrage consequences you would expect.

The structures are proprietary, licensed to issuers, and differ in detail; what they share is a mechanism for letting market makers hedge something correlated when they cannot hedge the actual thing.

What the live evidence shows

Several years of trading data support a reasonably firm assessment:

They function. Funds using these structures trade continuously, market makers quote them, creations and redemptions occur, and premiums and discounts stay within tolerable bands. The structural fear that arbitrage would break has not materialised.

The predicted trading-cost penalty largely failed to materialise. Issuer filings warn that shielded funds may trade wider, and price-to-NAV bands do loosen around volatile opens. At the category level, though, the gap has been unimpressive: in NYSE's data for the first nine months of 2023 the median semi-transparent spread was 22.5 basis points against 22.9 for fully transparent active ETFs, essentially level. Both trade far wider than index flagships; the shielded structure itself has not been visibly punished on spread.

Adoption has been modest. The semi-transparent category remains a small fraction of active ETF assets, while fully transparent active ETFs have taken the overwhelming majority of the growth (Chapter 76). That is the most informative datum in this chapter: given the choice, most active managers concluded that the benefit of secrecy did not justify the execution cost or the added complexity, and most investors did not require secrecy to buy an active fund.

The mutual fund conversion route competed it away. Managers wanting the wrapper largely converted their existing funds wholesale (Chapter 78) and accepted transparency, rather than adopting a shielded structure.

The honest argument on both sides

For the structures. The front-running concern is real and measurable for concentrated, high-conviction equity strategies with meaningful position-building. If a manager's edge genuinely erodes on disclosure, protecting it protects shareholder returns too, and a few basis points of spread is a defensible price for that. These structures also expanded choice: strategies that would never have entered the wrapper otherwise are now available with ETF tax and cost characteristics.

Against. The premise assumes an edge worth protecting, and Chapter 6's evidence says most strategies do not have one: which makes the trade-off a real cost, in complexity and opacity, paid for an uncertain benefit. There is also an accountability angle worth naming: opacity is precisely what index funds abolished, and reintroducing it in a wrapper marketed on transparency invites the question of what else is not being seen. And whatever execution cost does arise lands on shareholders who are not the ones benefiting from the secrecy, at least not directly.

Our position, labelled as such: the structures are legitimate engineering, and the market's own verdict, overwhelming preference for full transparency, is the more informative signal. For an investor, the practical implication is not to avoid them, but to require the same evidence you would require of any active fund (Chapter 76's six questions) plus an explicit accounting of the extra trading cost.

What it means for the wider ETF story

Two structural observations worth carrying forward:

Transparency is load-bearing, not cosmetic. Part I described daily disclosure as a nice feature for investors. This chapter shows it is actually a pricing mechanism: the easier a fund is to replicate, the cheaper it is to quote. That reframes several earlier chapters: the reason index ETFs trade so tightly is not only their liquid holdings but the fact that anyone can replicate them exactly; the shielded structures had to rebuild that confidence through engineering, and largely managed it.

The wrapper has boundaries. The ETF structure is unusually accommodating, but it cannot host a strategy that requires secrecy without imposing a cost somewhere. Chapter 97's discussion of tokenisation and the wrapper's future runs into the same constraint from a different direction: mechanisms that depend on open information degrade when information is withheld.

A worked example

An investor compares two active US equity ETFs from the same manager, same strategy, launched two years apart.

• The fully transparent version: daily holdings, spread 0.09%, premium/discount within ±0.05% on ordinary days.
• The semi-transparent version: proxy basket, holdings quarterly, spread 0.21%, premium/discount band roughly ±0.15%.

The difference of about 12 basis points of round-trip cost, plus a wider band, is what the confidentiality costs him in this pairing; category-wide, NYSE's 2023 medians were essentially level, so a gap this size is possible rather than typical. For a buy-and-hold holder trading twice a year, the annual impact is a few basis points: genuinely small. For anyone rebalancing more often, or entering and exiting at size, it compounds. His conclusion: the structure is acceptable if the strategy is worth owning, and irrelevant if it is not, which returns the decision to Chapter 76's six questions and leaves this chapter as a cost adjustment rather than a verdict.