The PCF and Basket Construction

What is actually in it

A PCF is a machine-readable file, distributed before each trading session through the National Securities Clearing Corporation in the US and equivalent channels in Europe, containing:

• The security list with exact share quantities per creation unit.
• A cash component, positive or negative, balancing accrued dividends, income, expenses and any residual so the basket's value equals the creation unit's NAV precisely.
• Creation unit size (commonly 25,000–100,000 shares; some funds use 10,000 or 200,000).
• Estimated cash amounts and any per-order transaction fees.
• Substitution and cash-in-lieu instructions for securities the AP cannot deliver.

Two properties make it powerful. It is published in advance, so every market maker can price the fund precisely and hedge with the actual constituents. And it is binding, so arbitrage is riskless in the operational sense: the AP knows exactly what it must deliver and exactly what it will receive.

Why the basket is not simply "the portfolio"

The intuitive assumption is that the creation basket is a pro-rata slice of the fund's holdings. Often it is. But issuers hold real discretion, and it is used for good operational reasons:

Cash in lieu. A constituent may be untradeable for the AP: a foreign market closed today, a stock suspended, a security with local ownership restrictions, or an IPO not yet settled. The PCF specifies a cash amount instead, and the fund buys the security itself. Cash-in-lieu is routine and usually trivial; heavy use signals a fund whose basket is hard to assemble, which shows up as wider spreads (Chapter 57's basket-cost ingredient).

Representative rather than complete baskets. For funds tracking thousands of securities (global aggregate bond funds, all-cap equity funds), demanding delivery of every line would make creation impossible. The basket is a sampled, representative subset: the same craft as Chapter 37's replication sampling, applied at the transaction level rather than the portfolio level.

Different creation and redemption baskets. This is the one worth pausing on. Nothing requires the two to be identical. An issuer may accept a broad basket on creation while specifying a different, narrower basket on redemption: typically loaded with the fund's lowest-cost-basis positions. That is precisely the mechanism behind Chapter 54's US tax efficiency: gains leave the fund through the redemption door. The discretion is disclosed and legitimate; knowing it exists is what lets you understand why two funds on the same index can have systematically different tax outcomes.

The tuning knobs, and what they cost you

Each element of basket design is a lever the issuer pulls, with visible consequences:

The unifying principle, worth stating once: the basket exists to keep flow costs with the investor causing the flow. Every design choice above is, at root, an implementation of that idea.

Bonds: where basket construction gets hard

Equity PCFs are largely mechanical. Fixed-income baskets are a craft, because a global aggregate index may hold 30,000 bonds, most of which have not traded this week. The practical adaptations:

• Small, representative baskets: a few hundred liquid bonds matching the portfolio's duration, sector and credit profile (Chapter 37's stratified sampling, again).
• Heavy use of cash creations where assembling bonds is impractical, with transaction-fee adjustments so the flowing party pays.
• Negotiated and custom baskets (Chapter 70's subject), where the AP and issuer agree a specific list, which is where the flexibility becomes genuinely strategic.

The consequence for the fund: bond ETFs realize more internal gains than equity ETFs (Chapter 54's boundary) and their arbitrage is looser, which is exactly why their premiums and discounts breathe in wider bands (Chapter 33).

Transparency as a design constraint

Publishing the basket daily is what makes the whole system work, and it also constrains what an ETF can be. A fund whose value depends on secrecy, a genuinely proprietary active strategy, cannot publish tomorrow's basket without giving away its positions, which is why:

• Most active ETFs are fully transparent and accept that constraint (Chapter 76).
• Semi-transparent structures exist specifically to break it, using proxy baskets, delayed disclosure or trusted-agent models so market makers can hedge without seeing the portfolio (Chapter 77).
• Non-transparent approaches trade arbitrage precision for secrecy, and the price appears as wider spreads and looser tracking: an honest trade, visible in the numbers.

The PCF, in other words, is the hinge between an ETF's transparency and its tightness. You cannot loosen one without loosening the other.

A worked example

Two funds, same MSCI World index, same day.

• Fund A publishes a 1,340-line basket covering 96% of holdings, with cash in lieu for a handful of suspended and market-closed names. Its cash component is 0.8% of the unit. Market makers hedge precisely; spread 0.05%.
• Fund B, smaller and newer, publishes a 420-line representative basket with a 6% cash component and frequent substitutions across Asian constituents. Market makers carry more residual risk between the basket and the index; spread 0.14%, premium/discount band roughly twice as wide.

Both are legitimate, well-run funds. The difference in what you pay to trade them was decided in a file neither investor will ever open: which is the argument for judging funds on realized outcomes (Chapter 35's tracking difference, Chapter 57's spreads) rather than on structure alone, while understanding the structure well enough to know why the outcomes differ.