Creations and Redemptions in Fixed-Income Exchange-Traded Funds: A Shift from Bonds to Cash

Introduction

Fixed-income exchange-traded funds (FI-ETFs) typically create and redeem the bulk of their shares in kind. In-kind transactions consist of exchanging ETF shares for baskets of bonds instead of cash. The ability to create and redeem shares in kind is valuable to ETF investors because it transfers the liquidity risk from the ETFs to the authorized participants (APs), which are typically financial institutions. Consequently, FI-ETFs are not subject to the risk of redemption runs present in less-liquid bond mutual funds (Arora et al. 2019; Chen, Goldstein and Jiang 2010).

In this note, we document a shift in the nature of the creation and redemption activity of FI-ETFs listed in the United States. Using novel data from the US Securities and Exchange Commission (SEC), we find that young FI-ETFs managed by new issuers tend to create their shares almost exclusively in cash. In contrast, older funds of established issuers do most of their creation activity in kind.

The shift toward creations in cash is significant. We find that, in 2018:

  • 50 percent of FI-ETFs created most of their shares in cash.
  • 18 percent of the total market value of FI-ETF creations was done in cash.

Funds that create in cash are more likely to hold less-liquid assets or to be actively managed by new issuers (i.e., they are more likely to seek to outperform an index). Our findings suggest that creating in cash is done more often when:

  • it is more difficult to quote and trade the underlying assets
  • an active fund has incentive to keep information about its holdings private
  • a new issuer launches a fund in a market that is more competitive than before

Because most of the FI-ETFs that have created in cash are young and did not report any redemptions in 2018, it is too early to precisely estimate the aggregate level of redemption activity done in cash. However, we find that among those that did report redemptions in 2018, 80 percent of FI-ETFs that created shares in cash also redeemed in cash. Cash redemption is therefore likely to be of a similar magnitude as cash creation.

The shift toward cash transactions implies that new funds are taking on exposure to liquidity risk (Pagano, Sánchez Serrano and Zechner 2019). Like bond mutual funds, FI-ETFs are now responsible for buying and selling bonds in the market and therefore need to manage the liquidity of their holdings.

Our findings suggest that the liquidity risk mitigating effects of the in-kind mechanism may be weakened as new FI-ETFs choose to create and redeem shares exclusively in cash (Anadu et al. 2018). Cash-only creations and redemptions in FI-ETFs thus warrant continued monitoring, and further analysis is needed to assess the implications for financial stability.

We take note of the new ETF rule adopted by the SEC in September 2019. One objective of this rule is to make the in-kind model more accessible for the funds. Going forward, it will be interesting to see whether the implementation of this rule will soften or end the shift toward cash.

How important are cash creations to FI-ETFs?

To answer this question, we use new data on FI-ETFs from the SEC’s Form N-CEN. Our sample of 294 funds for 2018 covers more than 84 percent, in terms of both number and size, of US-domiciled FI-ETFs (Table 1). The data provide information about the dollar value of creations and redemptions done in cash and in kind for every FI-ETF.

Table 1: FI-ETF market coverage from our sample in 2018

Sample US FI-ETF market Sample coverage (percent)
Number of FI-ETFs 294 348 84
Value of FI-ETFs (US$ billions) 603 627 96

Sources: US Securities and Exchange Commission Form N-CEN and Morningstar Direct

We classify FI-ETFs that did more than 50 percent of their creation activity in cash as “cash” funds and all other funds as “in-kind” funds. Twenty funds did not report any creations in 2018 and are therefore excluded from our analysis. Chart 1 highlights two points about the importance of cash creations in 2018:

  • The panel on the left shows that cash funds account for more than 50 percent of FI-ETFs.
  • The panel on the right shows that 18 percent of the total market value of FI-ETF creations was done in cash.

Chart 1: Cash creations are significant

Chart 1: Cash creations are significant

Chart 2 shows that FI-ETFs generally do not mix bonds and cash to create ETF shares. Of the 136 in-kind funds (yellow bars), 98 did more than 90 percent of their creation activity in kind. Moreover, of the 138 cash funds (blue bars), 115 did more than 90 percent of their creation activity in cash. This finding is important because it indicates that most cash funds create almost exclusively in cash.

Chart 2: Cash funds create ETF shares almost exclusively in cash

Finally, we find that cash funds are prominent across all investment types: government bonds, corporate bonds, domestic bonds and international bonds. Chart 3 shows that cash funds made up at least 47 percent of FI-ETFs of each type in 2018.

Chart 3: Cash funds exist across all FI ETF types

Chart 3: Cash funds exist across all FI ETF types

Is there a shift toward creating in cash?

Our analysis reveals that creating in cash is more pronounced among new funds and new ETF issuers. These results suggest a possible shift toward creating in cash.

We first look at whether FI-ETF creations in cash vary depending on the issuer’s share of the US FI-ETF market. A fund is classified as “established” if the issuer’s total value of fixed-income assets is greater than the average of fixed-income assets managed by all ETF issuers, and “new” if lower.

Of the 38 FI-ETF issuers represented in our sample, six are established issuers:

  • iShares by BlackRock
  • Invesco
  • PIMCO
  • SPDR ETFs by State Street
  • Schwab ETFs
  • Vanguard

The remaining 32 issuers in the sample are new. On average, new issuers launched their first FI-ETF three years ago, while established issuers launched theirs 11 years ago. By number of funds, 65 percent of FI-ETFs are products of established issuers.

Chart 4 shows that in 2018 cash funds accounted for 80 percent of funds offered by new issuers but only 35 percent of funds offered by established issuers. One possible explanation is that new issuers have less bargaining power with APs than established issuers. New issuers may therefore offer APs the flexibility to create ETF shares either in cash or in kind. Providing more flexibility can allow ETF issuers to build business relationships with APs to gain market share more rapidly.

A complementary explanation could be related to the US regulatory framework for ETFs. Before September 2019, when the SEC adopted a new ETF rule, issuers that chose the in-kind model were required to exchange ETF shares for an exact pro rata representation of their fund’s holdings (SEC 2019). The only way to exchange ETF shares for a custom basket of bonds rather than a pro rata representation was to request an exemption from the SEC. But seeking an exemption could take time and resources, and new issuers generally have fewer resources than established issuers. Consequently, their inability to use custom baskets may help explain why many new issuers have shifted to the cash model.

Chart 4: Across FI-ETFs, new issuers create more in cash

Chart 4: Across FI-ETFs, new issuers create more in cash

We also examine how FI-ETF creation in cash varies depending on the fund’s age. To do so, we further classify funds into “old” and “young” funds. We consider a fund young if it was created in 2016 or later, and old if it was created prior to 2016.

Chart 5 shows that in 2018, cash funds accounted for 64 percent of the young funds of established issuers and 81 percent of the young funds of new issuers. The fact that the young funds of both established and new issuers are largely cash funds suggests a possible shift toward creation in cash.

Chart 5: Across FI-ETFs, young funds create more in cash

Chart 5: Across FI-ETFs, young funds create more in cash

How important are cash redemptions?

In 2018, 8 percent of all FI-ETF redemptions were done in cash. However, 108 FI-ETFs, making up 37 percent of our sample, did not report any redemptions. Most of these (69 funds) were cash funds. We do not have redemption data on these funds yet; but we observe that of the FI-ETFs that reported redemptions in 2018, 80 percent of funds that created in cash also redeemed in-cash. Our results thus suggest that in the future, redemption in cash may become as important as creation in cash.

What factors drive creations and redemptions in cash?

We identify the key factors that drive an FI-ETF to be a cash fund. We analyze creation and redemption activity separately and obtain similar results. The Appendix describes our methodology and lists the factors we consider.

Consistent with the discussion above, we find that both younger funds and funds of new issuers create ETF shares more often in cash. Moreover, we show that funds that hold less-liquid assets (i.e., municipal bonds and securitized credit) or that are actively managed (i.e., funds that seek to outperform an index) are more likely to be cash funds (Figure 1).

Figure 1: Factors that drive cash creation and redemption

Factors that drive cash creation and redemption

These findings as well as discussions with industry participants suggest that cash creation is used more in settings where:

  • it is more difficult to quote and trade the underlying assets
  • an active fund has incentive to keep information about its holdings private

Implications for financial stability of cash-only creation and redemption

Cash transactions in the FI-ETF market have both positive and negative implications for financial stability.

On the positive side, the possibility of transacting in cash provides additional flexibility when the bonds required to create or redeem ETF shares trade less frequently or are difficult to buy at the desired price. Fund managers that accept cash when creating shares are compensated by the AP to cover the transaction costs related to the purchasing or selling of the bonds. They also reserve the right to reject this cash-only request if it is not deemed to be in the best interest of the FI-ETF investors. Cash transactions could thus represent normal financial innovations due to heightened competition in the FI-ETF market.

However, the shift toward cash redemptions pushes FI-ETFs away from the in-kind model. This could be concerning because in-kind redemptions help to mitigate the risk of redemption runs that is present in less-liquid bond mutual funds (Arora et al. 2019; Chen, Goldstein and Jiang 2010). When ETF shares are redeemed in kind, the fund manager delivers a basket of bonds and does not have to liquidate its bonds to meet investors’ requests to redeem. Consequently, the first-mover advantage, in which investors have an incentive to redeem ahead of others, does not exist in the in-kind model.

The shift toward cash redemptions could therefore weaken the stability-enhancing effects of in-kind redemptions (Anadu et al. 2018). At the same time, however, FI-ETF managers have options to mitigate liquidity issues in the fund. They can switch back to the in-kind model or adjust the fee they receive from APs for accepting cash transactions. Our analysis motivates the need to better understand how problems around first-mover advantage could arise among FI-ETFs that create and redeem exclusively in cash.

Conclusion

We document a shift in the nature of the creation and redemption activity of FI-ETFs listed in the United States. We find that young funds managed by new issuers tend to create and redeem their shares almost exclusively in cash. In contrast, older funds of established issuers do most of their creation and redemption activity in kind.

The growing importance of cash transactions in FI-ETFs warrants continued monitoring and a better understanding of the potential risks to financial stability. Future research could assess whether the liquidity management of ETF issuers differs when shares are redeemed in cash. Further research could also compare how cash and in-kind creation and redemption activities are affected during periods of stress.

Finally, it will be particularly interesting to see how the new ETF rule adopted by the SEC in September 2019 will affect the shift toward cash. This rule will allow the use of custom baskets for all FI-ETFs and may therefore make the in-kind model more flexible for the funds.

Appendix

Data

We collected novel data from the SEC’s Form N-CEN. Our sample of funds is available for 2018 and includes 294 of the 348 FI-ETFs listed in the United States. The other 54 funds are not included in our sample because they had not submitted their N-CEN form by the time we collected the data at the end of March 2019.

Model

We estimate a probit model using cross-sectional FI-ETF characteristics from December 2018. We perform stepwise (backward elimination) estimations to identify fund characteristics that increase or decrease the likelihood of an FI-ETF being a cash fund.

Table A-1 shows all the fund characteristic variables used in the regression. Table A-2 shows all the statistically significant fund characteristics that increase or decrease the likelihood of an FI-ETF being a cash fund. We can provide detailed regression tables.

Table A-1: Variables and data sources

Fund characteristic Source Computation and units
Active fund Morningstar Direct Binary categorical variable
Annual expense ratio Morningstar Direct In percentage points
Age of the fund Morningstar Direct In years
Net assets of the ETF share class Morningstar Direct Log dollar value
Percentage of assets held in the top 10 largest holdings Morningstar Direct In percentage points
Average credit quality Morningstar Direct Ordinal scale
Number of bond holdings Morningstar Direct Discrete values
Asset allocation in foreign bonds Morningstar Direct In percentage points
Asset allocation in cash Morningstar Direct In percentage points
Asset allocation in corporate bonds Morningstar Direct In percentage points
Asset allocation in sovereign bonds Morningstar Direct In percentage points
Asset allocation in municipal bonds Morningstar Direct In percentage points
Asset allocation in securitized products Morningstar Direct In percentage points
Asset allocation in derivatives Morningstar Direct In percentage points
Turnover ratio Morningstar Direct In percentage points
Collateral required by ETF issuer SEC Form N-CEN Binary categorical variable
ETF issuer fixed-income market experience Morningstar Direct and Bank of Canada calculations In years (number of years since the ETF issuer launched its first fixed-income ETF)
Fixed-income ETF net assets managed by ETF issuer Morningstar Direct and Bank of Canada calculations Log dollar value
Dollar value of creation unit Morningstar Direct, SEC Form N-CEN, Bank of Canada calculations Dollar value (ETF price multiplied by number of shares required for creation unit)
Herfindahl-Hirschman Index (HHI) of the APs for the ETF SEC Form N-CEN, Bank of Canada calculations Measure of AP competitiveness within an ETF (HHI calculated as the sum of the squared market shares of the APs in the ETF)

Table A-2: Variables and impact on likelihood of cash creations

Fund characteristic Direction and intuition
Active fund Positive—active funds are more likely to create in cash
Age of the fund Negative—older funds are less likely to create in cash
ETF issuer fixed-income market experience Negative—more established issuers are less likely to create in cash
Percentage of assets held in the top 10 largest holdings Negative—funds concentrated in a few bonds are less likely to create in cash
Asset allocation in foreign bonds Positive—funds with more foreign bonds are more likely to create in cash
Asset allocation in municipal bonds Positive—funds with more municipal bonds are more likely to create in cash
Asset allocation in securitized products Positive—funds with more securitized products are more likely to create in cash
Asset allocation in derivatives Positive—funds with more derivatives are more likely to create in cash

Note: These fund characteristics are all statistically significant at 5 percent.

References

  1. Anadu K., M. Kruttli, P. McCabe, E. Osambela and C. Hee Shin. 2018. “The Shift from Active to Passive Investing: Potential Risks to Financial Stability?” Federal Reserve Bank of Boston Working Paper No. 18-04.
  2. Arora, R., G. Bédard-Pagé and G. Ouellet Leblanc and R. Shotlander. 2019. “Could Canadian Bond Funds Add Stress to the Financial System?” Bank of Canada Staff Analytical Note No. 2019-9.
  3. Chen, Q., I. Goldstein and W. Jiang. 2010. “Payoff Complementarities and Financial Fragility: Evidence from Mutual Fund Outflows.” Journal of Financial Economics 97 (2): 239–262.
  4. Pagano, M., A. Sánchez Serrano and J. Zechner. 2019. “Can ETFs Contribute to Systemic Risk.” European Systemic Risk Board, Report of the Advisory Scientific Committee No. 9.
  5. Securities and Exchange Commission (SEC). 2019. “SEC Adopts New Rule to Modernize Regulation of Exchange-Traded Funds.” Press release (September 26).

Acknowledgements

We thank Guillaume Bédard-Pagé, Jean-Philippe Dion, Jean-Sébastien Fontaine and Virginie Traclet for helpful comments and suggestions. We are thankful to Narahari Phatak and James McLoughlin for helpful discussions about the data. Finally, we are grateful to Carole Hubbard and Colette Stoeber for editorial assistance.

Disclaimer

Bank of Canada staff analytical notes are short articles that focus on topical issues relevant to the current economic and financial context, produced independently from the Bank’s Governing Council. This work may support or challenge prevailing policy orthodoxy. Therefore, the views expressed in this note are solely those of the authors and may differ from official Bank of Canada views. No responsibility for them should be attributed to the Bank.