Physical vs Synthetic Replication (Intro)

Physical replication: own the bricks

The intuitive method. An MSCI World ETF using full replication buys all ~1,300 stocks in the index, at index weights, and holds them. The fund's return is the index's return, minus fees and the small real-world frictions of trading and taxes.

For very broad or illiquid indices, buying every line is inefficient: an index with thousands of small positions or hard-to-trade bonds would cost more to replicate perfectly than the precision is worth. So many physical funds use sampling (also called optimization): they hold a carefully chosen subset, perhaps 1,800 of an index's 3,000 names, engineered to behave like the whole. Done well, a sampled fund can track better after costs than a stubbornly complete one. Chapter 37 covers how to judge the craft; for now, "physical" spans a spectrum from owning everything to owning a smart subset.

One more feature of physical funds to file for later: many lend out a portion of their holdings for extra basis points of return (securities lending, Chapter 36). That is its own small, managed counterparty exposure, worth remembering when synthetic funds get criticized for theirs.

Synthetic replication: contract for the outcome

A synthetic ETF takes a different route to the same destination. The fund holds a substitute basket of quality assets, which may have little to do with the target index, and enters a total return swap with one or more banks: the fund pays away its basket's return and receives the index's exact return in exchange. The tracking is, by construction, near-perfect: the swap delivers the index number, whatever the basket did.

Why would anyone add a bank contract to a fund? Three practical reasons:

1. Access. Some markets are hard or impossible to hold directly at reasonable cost (historically China A-shares, certain emerging and frontier markets, commodity indices where holding barrels of oil is not an option). A swap reaches what warehouses of securities cannot.
2. Precision. For dense or fast-moving indices, contracting for the return can beat physically chasing it.
3. A specific tax edge. The famous case: US equity indices. Physical UCITS funds domiciled in Ireland lose 15% withholding tax on US dividends. Swap counterparties, under US rules applying to qualifying index swaps, can deliver the gross dividend inside the swap. Result: well-run synthetic S&P 500 UCITS ETFs have persistently tracked a few tenths of a percent per year better than their physical twins. Chapter 51 covers the withholding machinery; the headline here is that "synthetic" sometimes simply means "keeps more of the dividends."

The honest risk conversation

Synthetic's cost is counterparty risk: if the swap bank failed, the fund could lose whatever the bank owed it at that moment. The right response is neither dismissal nor panic; it is knowing the safeguards.

• The exposure is capped by law. UCITS rules limit net exposure to any single swap counterparty to 10% of fund assets; in practice, funds reset swaps frequently and hold or receive collateral, keeping actual exposure far below the cap, often near zero.
• The substitute basket belongs to the fund. In a counterparty failure, the fund still owns its basket (or holds collateral); the loss is limited to the uncollateralized swap value, not the fund.
• Symmetry matters. Physical funds that lend securities also carry collateralized counterparty exposure. The two structures' risks rhyme more than marketing on either side admits.

After 2008 put counterparty risk on every front page, European providers moved heavily toward physical replication for mainstream equity indices, and physical now dominates assets. Synthetic survives where its advantages are structural: commodities, some emerging markets, and the US-dividend edge above.

How to tell what you own, in ten seconds

Every factsheet states the replication method: "physical (full replication)", "physical (sampling/optimized)", or "synthetic (swap-based/unfunded/funded)". European provider naming sometimes hints too (ranges historically labeled "Swap" or "1C"), but the factsheet line is the authority. Synthetic funds additionally disclose swap counterparties and collateral policy; a provider that makes this hard to find is telling you something.

A worked example

Two Irish UCITS ETFs track the S&P 500. Fund P is physical, TER 0.07%. Fund S is synthetic, TER 0.05%.

• The index's gross dividend yield is ≈1.1% (mid-2026). Fund P receives dividends at 85% (15% US withholding): a drag of ≈0.17% per year versus gross. Fund S's swap delivers the gross-dividend index return.
• Expected tracking versus the gross-return benchmark, before other frictions: Fund P ≈ −0.28% (fee + withholding); Fund S ≈ −0.05% (fee). A ~0.2%/yr structural gap in synthetic's favor, visible in published tracking differences year after year.
• The price of that gap: living with a collateralized, capped, reset-daily bank exposure instead of direct stock ownership.

Reasonable investors land on both sides. What is not reasonable is deciding on vibes: the factsheet gives you the method, Chapter 35 gives you the realized tracking numbers, and Chapter 38 gives you the collateral details if you want to go all the way down.