Physical vs Synthetic ETFs: Which Should You Own in 2026?

Buy an S&P 500 ETF and you assume it owns 500 American companies. Often it does not. A large and growing share of index funds hold a swap contract with a bank instead of the underlying shares — and most investors never check which kind they bought.

That single design choice, physical versus synthetic replication, changes the tax you pay, the risk you carry, and how tightly the fund tracks its index. This guide shows you how both methods work, where each one quietly wins, and how to decide which belongs in your portfolio. If you want the structured version, our structured ETF investing course walks through fund construction end to end.

Physical vs synthetic ETFs: what's the difference?

A physical ETF tracks its index by actually buying the shares in it — either every constituent (full replication) or a representative subset (sampling). A synthetic ETF owns almost none of those shares; instead it signs a total-return swap in which a bank agrees to pay the fund the exact index return in exchange for a fee.

So the difference is ownership. With physical, you own a slice of the companies. With synthetic, you own a promise from a counterparty, backed by collateral, to deliver the index return. Everything else below flows from that one distinction.

Sources: BNP Paribas Asset Management, 2024; Lipper (LSEG) European ETF data, December 2024; UCITS Directive counterparty framework.

Read the table top to bottom and the trade-off is clear: physical buys you transparency and simplicity, synthetic buys you a tax and tracking edge on specific indices. The rest of this guide is about when that edge is worth the extra layer.

How does a synthetic ETF actually work?

A synthetic ETF takes your cash and buys a collateral basket — often blue-chip shares or government bonds that may have nothing to do with the index it tracks. It then enters a swap: the bank counterparty pays the fund the index's total return, and the fund pays the bank the return on that collateral basket plus a small fee.

The obvious worry is the bank failing. That is why the structure is regulated hard. UCITS rules cap the fund's net counterparty exposure to any single swap partner at 10% of net asset value, checked daily. In practice providers run well inside that limit.

Three mechanics keep the real exposure small. Collateral is frequently posted above 100% of the swap's value and held at a third-party custodian. Swaps are reset often — sometimes daily — so gains are crystallized and the open exposure resets toward zero. And many funds spread the swap across multiple counterparties so no single bank dominates.

Here is what that looks like with numbers. Say a synthetic fund holds $100 million of investor cash. It posts a collateral basket worth around $105 million at an independent custodian and runs a swap whose net value swings near zero as it resets. If the bank vanished overnight, the fund would still sit on collateral worth more than the exposure — the shortfall you would actually be chasing is a fraction of 1% of the fund, not the full $100 million.

The takeaway: your true "trust the bank" exposure at any moment is a thin slice of the fund, not the whole thing. That is the difference between a synthetic ETF and an unsecured bet on a bank.

The tax edge: why synthetic can win on US indices

Here is where synthetic stops being a curiosity and starts saving money. A physical fund that holds US shares must pay US dividend withholding tax — 15% for an Irish-domiciled fund under the US treaty, and up to 30% for a Luxembourg-domiciled one. That tax is a permanent drag on returns.

A synthetic fund sidesteps it. Under IRS Section 871(m), a swap on qualified US index components can be structured so the fund receives the index's full total return, dividends included, without that withholding. On a dividend-heavy, long-held position that gap compounds.

What this means for you: on a US index with a ~1.3% dividend yield, that 15–30-point withholding gap is worth roughly 0.2–0.4% a year of return — comparable to, and sometimes larger than, the fund's whole expense ratio. It is why synthetic quietly out-tracks physical on the S&P 500.

Put real money on it. Hold $50,000 in a dividend-paying US index at a 1.3% yield and you collect about $650 of dividends in year one. An Irish physical fund loses 15% of that — roughly $98 — to withholding; a Luxembourg one loses up to $195. A synthetic fund keeps it. One year is a rounding error; across two decades of reinvesting the money that was never withheld, the structure can matter more than a few basis points of headline fee. For the broader question of accessing that index from Britain, our guide to investing in the S&P 500 from the UK covers wrappers and fees alongside this.

Are synthetic ETFs safe? The counterparty risk, honestly

Yes, within the regulated framework — but "safe" is not the same as "risk-free." The honest answer is that a synthetic ETF adds one risk a physical fund does not inherently have: the swap counterparty failing while the fund is owed money.

If that happened, the fund would fall back on the collateral it holds. Because collateral is usually posted above the swap's value and held outside the bank, the loss is designed to be contained to any shortfall above the collateral. The European Central Bank has cautioned that in a default investors still bear risk tied to the quality of that collateral — so the collateral basket, not just the swap, is what you are really trusting.

The practical rule: a synthetic ETF from a major provider, reset frequently and over-collateralized, is a well-mitigated risk, not a reckless one. But if you cannot name the counterparty or read the collateral, that is a reason to prefer physical.

It helps to size that risk against the alternative. A physical fund that lends out a fifth of its shares for extra yield is running a comparable counterparty exposure on that slice — it is just rarely described in those words. The honest comparison is not "risk versus no risk." It is a disclosed, capped swap exposure versus an often-undisclosed lending one.

Physical ETFs aren't risk-free either

It is tempting to treat physical as the "safe" default and stop thinking. Don't. Two risks hide inside physical funds.

First, securities lending. Many physical ETFs lend out their shares to short-sellers for extra income. That introduces counterparty and collateral risk of exactly the kind synthetic funds get criticized for — just less visibly. Check whether your fund lends and how much it returns to holders.

Second, sampling. Funds tracking broad or illiquid indices often hold a subset of constituents rather than all of them. In calm markets that is fine; in a stressed market the statistical relationships sampling relies on can break down and tracking error widens. Full replication avoids this but costs more to run.

Despite all this, physical has decisively won the European market. Synthetic held roughly 46% of European equity-ETF assets back in 2010; by end-2024 synthetic was about 12% of total ETF assets, with physical holding the other ~88%.

Source: Lipper (LSEG) European ETF data, December 2024; Morningstar synthetic-ETF review, 2021.

The message is not that the crowd is always right. It is that synthetic is now a deliberate, niche tool rather than a default — reached for when its specific edge matters. If you want the wider map of how these funds differ from mutual trackers, our breakdown of index funds versus ETFs and of what UCITS versus US ETFs you can actually buy both sit alongside this one.

There is one revealing exception to the decline. Morningstar data show synthetic's share of S&P 500 and MSCI USA tracking has climbed about ten percentage points since 2017, to roughly 30% — the single corner of the market where the tax edge is actively pulling assets back toward swaps. On tracking quality itself, academic work (Broby and Spence, 2020) found the expense ratio remains the dominant driver of tracking difference, with replication method a secondary factor; synthetic funds tend to post marginally tighter tracking error, not looser. So the method is a tie-breaker, not the headline — get the cost and the index right first.

Which replication method should you own?

Match the method to what you are tracking, not to a slogan. Here is the decision rule:

• Broad developed-market equity (world, Europe, FTSE): default to physical, full replication. Transparent, simple, and the tax gap is small on low-yield or non-US indices.
• S&P 500 or MSCI USA as a long-term core holding: a synthetic fund can add 0.2–0.4% a year through the 871(m) dividend-tax edge — worth it if you trust the provider and counterparty.
• Hard-to-access or closed markets (e.g. mainland China): synthetic is often the only clean way in, because physically holding the shares is costly or restricted.
• Leveraged or inverse exposure: almost always synthetic by necessity — understand the daily-reset and swap mechanics before you touch it.
• You can't name the counterparty or read the collateral: choose physical. Never own a structure you can't explain.

One more habit worth building: open the fund's factsheet or KIID and look for the word "swap" or "synthetic." Thirty seconds tells you which machine you own.

Frequently asked questions

Trading and investing involve substantial risk of loss and are not suitable for every investor. This article is educational content, not investment advice.