Corporate bond exchange-traded funds (ETFs) have been increasing rapidly and becoming an emerging financial instrument in the past two decades. They allow investors to build low-cost bond portfolios. Investors can adjust positions in real-time instead of trading corporate bonds over the counter or investing in bond mutual funds with minimums and limits. With ETFs targeting credit quality or maturity, investors can also form portfolios with more specific preferences.
One attractive feature is that ETF share prices and net asset values (NAVs) are presumably close via the creation/redemption mechanism. For example, if the share price is lower than NAV, authorized participants (APs), or market makers, send back ETF shares. ETF sponsors then provide a basket of corporate bonds to APs to fulfill this redemption requirement.
Figure 1: The percentage difference between ETF share prices and NAVs of Vanguard Total Bond Market ETF (ticker: BND), January 2019 to April 2020. The shaded area represents the COVID-19 pandemic. The difference during the COVID-19 period was -0.29%, and the number was 0.06% before the pandemic. Data source: The Center for Research in Security Prices.
However, the COVID-19 pandemic broke the balance and depicted a puzzling picture: Share prices were unprecedentedly and persistently lower than NAVs. For example, BND experienced large difference between ETF share prices and NAVs during the COVID-19 pandemic (-28 basis points). This number was almost five times greater than the pre-COVID difference (6 basis points), which also lasted for at least one month (See Figure 1). The large discounts reemerged in June 2022, when bonds posted historic losses due to spikes in inflation. This puzzling fact prompted a broad debate between policymakers and practitioners. In a Bank for International Settlements policy paper, Karamfil Todorov argued that ETFs deliberately handed low-quality bonds to APs, who were no longer willing to initiate redemptions. Practitioners disagreed with the statement and claimed that ETF sponsors from BlackRock kept providing proper baskets and market makers continued redeeming ETF shares throughout the crisis and accepting baskets of bonds that did not harm either investors who remained with the fund or those who chose to redeem.
The arguments in this debate concentrate on two questions: (1) what motivates ETFs and APs to negotiate baskets in redemptions and (2) what impacts these redemptions have on investors and market demand. In my research, I tackle these two questions to reconcile the debate. I investigate economic incentives underlying the choice of bonds in corporate bond exchange-traded funds (ETFs) redemptions. In the primary market between ETF sponsors and authorized participants (APs), sponsors fill baskets with high price pressure exposed bonds, and APs select assets that negatively co-move with illiquidity in their own portfolios. In the secondary market, where ETF shares are publicly traded, days with redemptions are associated with lower ETF returns, liquidity, price efficiency, and demand elasticity. APs profit from redemptions by correcting discrepancies between ETF share prices and portfolio values.
First, to understand ETF’s motivations, I show that ETFs negotiate with APs and fill the redemption baskets with corporate bonds under high price pressures.1 Price pressures are defined using changes in corporate bond mutual funds holdings with extreme fund flows (top and bottom ten percent). If a mutual fund had to sell corporate bonds to meet redemption obligations, these bonds would suffer from selling pressures at low prices. In this negotiation, ETFs are cautious about the price pressure of their portfolios because higher pressures would be associated with lower future NAVs, which might encourage further redemptions.
I separate corporate bonds of ETFs with redemptions into two parts, baskets and remaining portfolios. Baskets contain those corporate bonds handed to APs; in the remaining portfolios, the number of corporate bond holdings is the same as the number before the redemption. I calculate the average price pressures for both baskets and remaining portfolios and compare the sample means. It turns out that the price pressures in baskets are always higher than those in the remaining portfolios. By keeping corporate bonds under low price pressure, the ETF portfolio value decreases by $38,570, which helps to correct the difference between ETF share prices and portfolio values.
To study the effect of negotiation on the incentives of ETFs, I construct hypothetical ETFs, which keep both baskets and remaining portfolios upon redemptions. The price pressures of hypothetical ETFs will be larger if baskets contain bonds with high pressures. The strong negotiation between ETFs and APs implies baskets have a small amount of different corporate bonds. I find that strong negotiation leads to greater price pressures in hypothetical ETFs.
This result shows that price pressure is a vital element of ETF incentives in the redemption mechanism. Corporate bond ETFs do dispose of bonds but not deliberately; instead, they care about price pressures and send bonds with high pressures to APs such that the remaining portfolios have a relatively stable value, which facilitates narrowing the discounts. Hence, the result is consistent with the explanations provided by policymakers.
Second, APs agree on the basket components such that their liquidity negatively co-moves with the liquidity of their bond portfolios. The liquidity co-movement is the correlation between the changes in liquidity of two portfolios.
Previous literature has discussed the liquidity of baskets and remaining portfolios in either ETFs or mutual funds. It has shown that portfolios are more liquid in redemptions. However, this evidence does not hold during the 2008 and COVID-19 crises. The reason is that the co-movement among different assets increases, especially in crisis periods, making the liquidity level comparison fail. Therefore, I focus on the co-movement in liquidity between baskets and AP portfolios. It is also related to liquidity management in financial institutions, an essential concern to financial markets after the bankruptcy of Lehman Brothers and the recent global financial crisis in 2008.
The liquidity co-movement between ETF baskets and AP portfolios is significantly negative. The negative number suggests that APs select components of baskets such that liquidity moves in the opposite direction from their own corporate bond portfolios. For example, a decrease in basket liquidity leads to an increase in AP portfolio liquidity. During the COVID-19 pandemic, I show that the magnitude of negative co-movement is even more prominent. It suggests the incentive of APs in the negotiation is more robust in the COVID-19 crisis.
The evidence suggests that APs target liquidity in the redemption mechanism. During the COVID-19 period, it became more challenging to select negative co-moving securities because most assets are correlated, which is consistent with the puzzling large discount in Figure 1 and aligns with the anonymous market maker’s claim.
The incentives of ETFs and APs suggest that, in redemptions, their negotiation determines the final composition of the baskets. ETFs dispose of bonds with price pressures, but APs accept them because they can benefit from these transactions. The result reconciles the argument between policymakers and practitioners.
Regarding the second research question, I investigate the economic impacts of redemptions on ETF investors, APs, and market demands. I compare the ETF performance on redemption days and normal days, where no creation/redemption happens. The comparison between redemption and normal days emphasizes who benefits or suffers from redemptions. For example, compared to normal days, if an ETF investor yields lower returns on redemption days, then we conclude that this investor suffers from redemption activities.
Redemptions decrease ETF investors’ returns by two basis points per day or 5% per year. After redemptions, these investors also pay a 10% extra transaction cost in the less efficient secondary market. The evidence suggests that investors face a worse trading environment and suffer losses from redemptions. In contrast, APs gain more from the difference between ETF share prices and NAVs. In particular, in an ideal environment, APs profit $127,660 from redemption.
In terms of market demand, I focus on price demand elasticity, the ratio between the percentage changes of quantity and price. On redemption days, the elasticity is lower than that on normal days. The result suggests that the corporate bond ETF market becomes less elastic on redemption days than on normal days. Specifically, if the demand for bond ETF shares increases by one dollar, the ETF market value will change by $10.5 upon redemptions. The large elasticities on redemption days allow to correct the discrepancies between the low ETF share prices and the high portfolio values. One potential explanation is that investors have difficulty gathering information from the opaque corporate bond market, so they become concerned about the ETF’s performance. Thus, the demand is inelastic upon redemption, and investors would suffer from welfare loss.
These results show that ETF investors suffer from redemptions because their returns and market conditions are worse. In contrast, redemptions provide APs with more arbitrage opportunities.
I also apply the new monetary policy to explore whether the incentives of ETFs are sensitive to these changes. On March 23, 2020, Federal Reserve announced the Secondary Market Corporate Credit Facility (SMCCF) to support the corporate bond market. The SMCCF allowed the Federal Reserve, for the first time in its history, to directly purchase qualified corporate bonds and U.S-listed investment-grade corporate bond ETFs. I show that the new policy supports the incentive of ETFs, where they keep filling redemption baskets with bonds of high price pressures.
To emphasize the importance of APs in the corporate bond ETF market, I compare the index bond mutual funds and bond ETFs. The difference between these two types of products is that index funds directly trade bonds in the market without APs. I compare the average price pressures of baskets and remaining portfolios in redemptions. Index funds do not have significantly lower pressure in baskets relative to remaining portfolios. It implies that APs act as a buffer for ETFs to prevent further redemptions.
Han Xiao is a PhD candidate in Finance at the Smeal College of Business at Pennsylvania State University.
This post is adapted from his paper, “The Economics of ETF Redemptions,” available on SSRN.