Weak authorised participant networks create ETF mispricings, research finds

Study shows 15 out of 50 APs are responsible for 98% of all creation and redemptions

Theo Andrew


Mispricing in the US equity ETF market is directly correlated to the number authorised participants (APs) involved in the creation and redemption process, academic research has suggested.

The research, titled Two APs Are Better Than One: ETF Mispricing and Primary Market Participation, found the greater the number of APs involved in the creation and redemption process of an ETF, the lower the level of mispricing in primary markets, particularly during periods of volatility.

The research, published by the London Business School, said ETFs with a diverse AP network mitigate daily mispricing by as much as 15% on a one standard deviation increase in the mean number of APs registered to an ETF.

However, the analysis found the “dense core and sparse periphery” of APs is leading to more mispricing within the US equity market, as fewer larger APs – mostly banks – are responsible for the bulk of the transactions.

“The level of mispricing in a US equity ETF is negatively related to the fund’s network diversity, especially during times of high market volatility,” the research said.

“We observe stronger mispricing co-movement for funds that share more APs. We theoretically show that diverse networks mitigate the effect of shocks to AP-specific arbitrage costs.”

APs are a unique part of the ETF ecosystem that hold the right to create and redeem ETF shares, and play a crucial role in the liquidity of an ETF while ensuring that a price of an ETF stays aligned to its net asset value (NAV).

The research was able to exploit the new regulatory N-CEN filings to understand the annual trading volume of each AP. Since June 2019, ETFs have been required to report the structure and activity of the primary markets to the Securities and Exchange Commission (SEC).

It noted that 15 out of 50 APs are responsible for 98% of all creations and redemptions, with the top three generating half of those transactions, with the mispricing of ETFs using the same ETFs moving in tandem.

“We show that the correlation of mispricing between two ETFs in our US equity sample is related to the commonality in their active AP network,” it said. “On low-stress days, this commonality does not contribute to correlation in ETF mispricing. On high-stress days, however, having twice as many common active APs is associated with six percentage points higher correlation.”

It added mispricing is driven by arbitrage-specific costs, noting the costlier the arbitrage is for secondary market participants, the more mispricing occurs in the primary market. 

“We illustrate that in our model, a larger and more diverse pool of potential arbitrageurs makes mispricing less sensitive to changes in costs for a given arbitrageur or to the exclusion of certain arbitrageurs from the market,” the study said.

Furthermore, the paper concluded the APs exposed to the Federal Reserve’s bond ETF purchasing programme in March 2020 exhibited higher mispricing during the purchase period, while equity ETFs exposed to the same APs also experienced greater mispricing.

“ETFs whose APs were more engaged in the program exhibit higher mispricing during the implementation period. The effect is economically small but statistically significant,” it said. 

“Importantly, we see that the result is concentrated in ETFs with less diverse networks and on days when secondary market arbitrageurs are less likely to step in.”

Related articles


No ETFs to show.