Introduction

Synthetic – or swap-based – ETFs typically use one of two approaches: the ‘unfunded’ model or the ‘fully funded’ model. The unfunded model uses ‘unfunded’ total return swaps to replicate the performance of an index, while the fully funded model uses ‘funded’ total return swaps. The swap counterparty is typically an investment bank. For investors, the unfunded model is the safer of the two as the ETF owns and has direct access to the underlying collateral. In the fully funded model, on the other hand, collateral is ‘pledged in favour’ of the ETF with a custodian. There is no direct access to it. As Dan Caps, investment manager at Evelyn Partners, put it: “The fully funded swap model sounds great until you realise it is the swap counterparty that is fully funded.”

Safety concerns

A series of crises around 2010 brought ‘counterparty risk’ to the fore and the synthetic share of total European ETF assets fell from around half to just over 10% by the end of 2024. Interest in synthetic ETFs is coming back in areas like US equities, however. This is thanks to good performance and lower tracking error while the safety of synthetic ETFs has increased significantly, too. The unfunded model has improved over time while the fully funded model has all but disappeared.

The ‘unfunded’ model

In the unfunded model, the ETF issuer passes the investor cash on to a swap counterparty and in return receives a substitute, or reference, basket of securities. The ETF owns and has direct access to these securities. The issuer and the counterparty then exchange performance. The issuer pays the returns of the reference basket and receives the returns of the index less a swap spread – essentially a fee paid to the counterparty. This performance is passed on to the end investor following the deduction of management fees. The below schematic from Vanguard captures this in action:

Synthetic ETFs using the unfunded model often carry ‘counterparty risk’. In other words, the difference between the value of the reference basket and ETF’s net asset value (NAV). This ‘valuation gap’ can be no more than 10%, according to the UCITS directive. To ensure this happens, the swap marks-to-market daily. If counterparty exposure moves above of 10%, the swap is reset and the counterparty will have to place additional securities into the reference basket. A further consideration is that the reference basket does not have to be similar to the index securities. As such, if the counterparty defaults, the investor will no longer have exposure to the desired index. Even prior to 2010, most synthetic ETFs employed the unfunded model, but they tended to be heavily exposed to a single counterparty – often the issuer’s parent company – and ‘protected’ by substitute baskets that were opaque and unfit for purpose. Most swap-based ETFs now use multiple counterparties in case one defaults. The reference baskets also tend to bear greater resemblance to the underlying index.

The ‘fully funded’ model

In the ‘fully funded’ model, on the other hand, after receiving the investor cash the counterparty posts collateral with a custodian pledged in favour of the ETF. Unlike with the unfunded model, the ETF does not have direct access to the assets. Here is Vanguard’s depiction:

Like the unfunded model, though, the collateral basket is likely to be different to the securities underlying the index. During periods of stress, the securities in the collateral pool may be difficult to sell – if they can be accessed at all – which brings additional risks to investors.