Introduction
Open-ended mutual funds (OEMFs) and exchange-traded funds (ETFs) serve as crucial intermediaries between investors and securities markets. Historically, these funds predominantly focused on publicly traded equities. However, since the global financial crisis, they have expanded into less liquid asset classes, including corporate bonds, which tend to be less liquid than, for example, Treasury notes, and can be susceptible to market instability during periods of stress. Yet, corporate bond markets play a vital role in credit provision to the real economy. Therefore, this note focuses on funds’ investments in the US corporate bond market, the largest corporate bond market globally. By the end of 2023, US bond funds (including corporate and sovereign bond funds) managed $6.5 trillion, with ETFs and OEMFs accounting for $1.6 trillion and $4.9 trillion, respectively.1
ETFs and OEMFs can offer similar portfolio exposures but face different vulnerabilities because of their structural features. Especially passively managed ETFs and OEMFs show significant similarities in their portfolio holdings as they aim to track benchmark indices. However, a key feature of OEMFs is that investors can redeem their shares at the fund’s net asset value which gives rise to the so-called payoff complementarities. This creates a first-mover advantage for investors to redeem their shares, potentially incentivizing run-like behavior and contributing to fire sales (Chen, Goldstein, and Jiang 2010, and Cetorelli, Duarte, and Eisenbach 2016). In contrast, ETFs rely on financial intermediaries known as authorized participants, who are incentivized to align the ETF share price with the value of the basket of underlying securities. While this mechanism can partially insulate the underlying securities market from the trading activity in ETF shares, it also introduces potential friction related to the authorized participants’ ability to conduct arbitrage (Gorbatikov and Sikorskaya 2022).
Studies indicate that OEMFs can amplify adverse shocks as their design can encourage early redemptions and runs on funds (see, for example, Goldstein, Jiang, and Ng 2017; O’Hara and Zhou 2021; Bouveret and Yu 2021; Ma, Xiao, and Zeng 2022; International Monetary Fund 2022). Especially during times of stress, OEMFs may face large redemptions forcing them to liquidate assets in already illiquid markets, thereby amplifying asset price volatility. For example, during the COVID-19-related market turmoil in February and March 2020, the average OEMF experienced cumulative outflows of about 10 percent of net asset value (Falato, Goldstein, and Hortaçsu 2021). These vulnerabilities are more pronounced for funds with a larger liquidity mismatch, for example, funds offering daily redemptions while investing in illiquid asset classes (Jiang, Li, Sun, and Wang 2022; International Monetary Fund 2022).
Existing evidence on the impact of ETFs on market fragilities is inconclusive. Falato, Goldstein, and Hortaçsu (2021) document that ETFs were more resilient and experienced fewer outflows than comparable OEMFs during the COVID-19-related market turmoil. They attribute this resilience at least partially to the relatively lower sensitivity of ETF flows to fund-level asset illiquidity. Similarly, the International Monetary Fund (2022) finds that bonds predominantly held by ETFs experienced a smaller increase in return volatility during the COVID-19 market turmoil than comparable bonds held by OEMFs.
These empirical results are consistent with theoretical predictions by Helmke (2023), who argues that ETFs may be less prone to transmitting shocks to underlying security markets because transactions in ETF shares do not necessarily lead to sales in the underlying markets.2 This may temporarily shield the underlying market from the actions of ETF investors. However, Dannhauser and Hoseinzade (2022) find that ETFs created flow-induced pressures in corporate bond markets during the 2013 Taper Tantrum. They attribute this to a clientele effect, whereby ETFs attract high liquidity–demand investors who pass on shocks to securities markets.
Differences in funds’ investor base could drive market fragilities. ETFs and OEMFs likely attract different types of investors. For example, OEMFs may appeal to investors who do not require intra-day liquidity but are wary of potential short-term mispricing in ETFs (Helmke 2023). In contrast, a significant share of ETF trading activity comes from the intra-day trading by liquidity traders as well as the hedging demands of institutional investors. These types of investors are typically absent from OEMFs, which are often held in retirement accounts such as 401(k) plans, where ETFs are often not available. Therefore, to identify the asset price fragilities associated with the fundamental structures of ETFs and OEMFs, it is essential to consider the investor clientele.
This note examines the relationship between funds’ investor profiles and the return volatility of the securities they hold. Based on the security level data of US ETF and OEMF holdings, on aggregate, a higher ETF ownership share is associated with lower bond return volatility. However, there is a notable divergence in the behavior of institutional and retail ETF investors and their impact on the underlying market. When a larger share of a bond is owned by institutional investors through ETFs, the bond’s volatility tends to be higher. The opposite is true for retail investors. The same disparity is not evident for OEMFs.
This note is structured as follows: the first section describes the growth and the nature of US corporate bond funds and then zooms in on how institutional investors use ETFs and OEMFs. The second section analyzes the relationship between fund holdings and bond market volatility, distinguishing between institutional and retail ETF and OEMF investors. The final section concludes and offers a forward-looking eye on future research.
Investment Funds and the Corporate Bond Market
The Role of Open-Ended Mutual Funds and Exchange Traded Funds in US Corporate Bond Markets
OEMF and ETF holdings of corporate bonds have grown markedly since the global financial crisis. While OEMFs still maintain a larger footprint in US corporate bond markets compared with ETFs, the role of ETFs has grown in recent years (Figure 1, panel 1). Combined, ETFs and OEMFs hold around $1 trillion of US corporate bonds, which represents around 12 percent of corporate bonds outstanding and over 4 percent of credit provided to US nonfinancial corporations (Figure 1, panel 2). As a share of the market value outstanding, their holdings are particularly large in the high-yield corporate bond segment (Figure 1, panel 3). The shift from active to passive investing has contributed to the growth of corporate bond ETFs which predominantly pursue index-based investment strategies, while most OEMFs continue to be actively managed (Figure 1, panel 4).3
Growth of ETFs and MFs in Corporate Bond Markets
Citation: Global Financial Stability Notes 2025, 002; 10.5089/9798229003087.065.A001
Sources: EPFR, FactSet, Bloomberg Finance L.P., Bank for International Settlements; Lipper, Refinitiv; and IMF staff calculations. Notes: NFC = non-financial corporation.Notes: In panel 3, the dotted line indicates the cutoff between high-yield and investment-grade ratings. In panel 4, US-domiciled bond funds are shown weighted by assets under management as of December 2022. ETF = exchange-traded fund; MF = mutual fund; OEMF = open-ended mutual fund.How Do Institutional Investors Use Bond Funds?
Institutional investors constitute a significant portion of the OEMF and ETF investor base. In the US, institutional holdings of bond ETFs and OEMFs have gradually increased from around 50 percent in 2012 to over 60 percent in 2023 (Figure 2, panel 1).4 Nearly half of ETF shares are held by investment advisors (which includes OEMFs). Other notable institutional investors include banks, insurance companies, trusts, and hedge funds (Figure 2, panel 2).
Institutional Investors in Corporate Bond MFs and ETFs
Citation: Global Financial Stability Notes 2025, 002; 10.5089/9798229003087.065.A001
Sources: Upper; Refinitiv; FactSet; Bloomberg Finance L.P.; and IMF staff calculations.Notes: Panel 2 shows the investor breakdown by investor type for the ten largest US corporate bond ETFs. ETF = exchange-traded fund; MF = mutual fund.Institutional investors may hold ETFs for various reasons, including for hedging or speculative purposes. For example, OEMFs may use ETF shares to better manage liquidity, potentially leading to elevated trading activity in ETF shares (Sherrill, Shirley, and Stark 2020).5 Consistent with this behavior, Dekker, Molestina Vivar, and Weistroffer (2024) showed that open-ended investment funds, the largest group of ETF investors in the euro area, disproportionally reduced their ETF holdings when facing large outflows in March 2020. Similarly, evidence suggests that OEMFs with higher ownership by other funds experienced substantially higher outflows compared with those owned by retail investors (Allaire, Breckenfelder, and Hoerova 2023). Institutional investors may also use ETFs to take directional positions, with ETFs predominantly held by institutional investors facing higher levels of short positions compared with ETFs held by retail investors (Figure 3, panel 1).
OEMF shares are used by institutional investors for different purposes. OEMFs do not offer intra-day liquidity and cannot be shorted but may have advantages in terms of offering a broader range of active strategies. The institutional investor base for OEMFs may, compared with ETFs, therefore be more tilted toward investors with a longer investment horizon and lower turnover ratios, for example, monthly rebalancing as opposed to intra-day trading. Allaire, Breckenfelder, and Hoerova (2023) find for euro area bond funds that in 2019:Q4, 33.2 percent of domestic (euro area) mutual fund ownership came from other investment funds, 17.2 percent from insurance corporations, and 3.2 percent from pension funds. While not directly comparable with the investor breakdown of US corporate bond ETFs (Figure 3, panel 2), the large ownership share of the insurance sector in European bond mutual funds illustrates that the institutional investor bases for ETFs and OEMFs have a different makeup.
Different investor clienteles appear to behave differently as seen through the lens of fund flows and trading activity. US open-ended bond funds held by institutional investors exhibit more volatile flows than those held by retail investors, especially during times of stress, proxied by the volatility index (Figure 3, panel 2). Similarly, US corporate bond ETFs with a high share of institutional ownership show larger trading volumes during periods of stress (Figure 3, panel 3).
Institutional Fund Investment and Trading Patterns
Citation: Global Financial Stability Notes 2025, 002; 10.5089/9798229003087.065.A001
Sources: Bloomberg Finance L.P.; Lipper; and IMF staff calculations.Note: ETF = exchange-traded fund; OEMF = open-ended mutual fund. Implied volatility is based on the Chicago Board Options Exchange’s Volatility Index, often referred to as the VIX.Evidence from Fund Flows
Institutional mutual fund flows are more volatile than retail fund flows. Larger outflows pose liquidity risks for mutual funds (Bouveret and Yu 2021), especially if they are forced to liquidate assets in times of stress and impaired market liquidity. From this perspective, institutional mutual funds appear more vulnerable. While 90 percent of weekly retail fund flows range from -0.71 percent to 0.50 percent of assets under management, institutional fund flows range from -0.86 percent to 0.93 percent (Figure 4, panel 1). In addition, institutional OEMF flows tend to be more procyclical than retail OEMF flows (Figure 4, panel 2). On aggregate, funds experience inflows during months with positive index returns and outflows during months with negative returns. This procyclicality of flows could result from end-investors’ reaction to market movements or from fund flows driving market movements. Both effects can contribute to a destabilizing feedback loop, where market movements trigger fund flows, which in turn amplify the initial market movements.
ETF flows may affect the underlying asset markets differently than OEMF flows. ETF flows are more volatile than OEMF flows, with 90 percent of weekly ETF flows in the range of -1.30 percent to 1.81 percent (Figure 4, panel 1).6 In addition, ETF flows show a stronger correlation with bond market returns, indicating higher procyclicality. However, ETF flows, that is transactions with authorized participants, do not necessarily lead to transactions in underlying security markets. The authorized participants observe the relative mispricing between ETF shares and the underlying securities, as well as any balance sheet costs associated with ETF arbitrage trades, before executing trades. In contrast, OEMFs may be forced to buy or sell at less opportune times. As a result, while corporate bond ETF flows are more volatile than OEMF flows, this does not necessarily imply that they have a more adverse impact on markets.
ETF and MF Flow Stylized Facts
Citation: Global Financial Stability Notes 2025, 002; 10.5089/9798229003087.065.A001
Sources: EPFR; Bloomberg Finance L.P.; and IMF staff calculations.Notes: Returns and volatility are based on the Bloomberg-Barclays US High Yield Corporate Bond Index. ETF = exchange-traded fund; MF = mutual fund, AuM = assets under management.Investment Funds and Bond Market Stability
The Relationship between Bond Market Fragility and Fund Holdings
A granular approach is used to study the impact of bonds’ ownership structures on bond return volatility. This note examines the role of fund ownership in bond markets by analyzing, at the security-level, how ETF and OEMF holdings of a bond affect its return volatility. Specifically, the relationship between ETF and OEMF ownership and the volatility of returns is analyzed using panel regressions with the following specification for security i and quarter t:7
The dependent variable VOLit is the standard deviation of daily returns of a US corporate bond within quarter t, annualized and expressed in basis points. The explanatory variables of interest are the share of bond i held by US-based ETFs and the share of bond i held by OEMFs, both expressed as a percentage of the security’s US dollar amount outstanding at the end of each quarter. Control variables include the security’s total market value outstanding and the bond’s time to maturity.
There are empirical challenges associated with estimating the effect of funds’ ownership characteristics on corporate bond market volatility. The relationship between fund flows and asset prices is complex, with periods of high aggregate uncertainty most likely affecting both. Similarly, feedback effects—where asset markets drive flows and flows affect asset markets—complicate the assessment of causality. Moreover, the volatility of daily returns can be influenced by various factors, including broad market developments, market sentiment, news affecting individual issuers, and security-specific characteristics such as maturity, coupon type, or other features. To address these issues, fixed effects at the quarter, bond type, issuer type, and rating level are included alongside other control variables. These fixed effects aim to absorb factors that could distort the assessed relationship between fund holdings and volatility, allowing for “within-group” comparison of bonds that are as similar as possible, aside from their ownership structure. Regression tables can be found in the Annex 2.
To understand the role of investor clienteles, the analysis distinguishes between institutional and retail holdings. For OEMFs, institutional holdings are computed based on Lipper fund share class classifications.8 For ETFs, institutional holdings are computed based on SEC 13F filings by institutional investors. ETF and OEMF holdings are then split into two mutually exclusive components, institutional holdings and non-institutional holdings (referred to as retail holdings), to estimate the following panel regression:
The relationship between fund ownership types and volatility may depend on the level of market stress. To investigate how market stress can play a role, an interaction term between holdings and implied equity market volatility is used. Equation 1 is modified to:
In equation 3, VIX’ is the average quarterly level of the VIX volatility index, which serves as a proxy of market stress. The total effect of OEMF holdings on volatility is reflected by the combination of β1,MF and βl,MF and the level of the VIX. A similar modification can be made to equation 2, by introducing VIX interaction terms for retail and institutional ETF and OEMF holdings. Time fixed effects are dropped from this specification to facilitate the interpretation of the time varying effect of market stress.
Results
ETF ownership is, on average, associated with lower bond return volatility. Consistent with the findings in IMF (2022), bonds held by ETFs experience less of an increase in volatility than comparable bonds held by OEMFs or other investors (Figure 5, panel 1). Specifically, a one percentage point increase in the share of a bond’s value outstanding held by ETFs is associated with a 13 basis points lower return volatility, after controlling for other factors (Annex Table 2.1, column 1).
The investor base matters. ETF holdings attributable to institutional investors are associated with higher bond return volatility while a larger share of ETF holdings attributable to retail investors is associated with lower volatility. Specifically, a one percentage point increase in the share of a bond’s outstanding value held by institutional ETF investors is associated with a 27 basis points increase in return volatility (see Annex Table 2.1, column 2). In contrast, a one percentage point increase in the share of a bond’s outstanding value held by retail ETF investors is associated with an 85-basis-point decrease in return volatility. There is no similar divergence between institutional and retail share classes of OEMFs.
The role of institutional investors is amplified during periods of stress. In line with previous work (International Monetary Fund 2022), OEMF holdings are associated with higher bond return volatility at high levels of market stress (Figure 5, panel 2). By contrast, a higher share of ETF holdings is not associated with an increase in return volatility even at high levels of market stress. However, the divergence between institutional and retail ETF holdings becomes more pronounced during periods of elevated market stress (Figure 5, panel 3). At high levels of market stress, institutional ETF holdings are associated with sizable adverse impacts on volatility, followed by institutional OEMF holdings. This observation is consistent with the hypothesis that institutional investors use ETFs to manage risks and liquidity shocks that materialize during stress episodes, which are then passed on to the underlying markets.
The divergence between institutional and retail holdings is more pronounced for ETFs. For example, the mitigating impact of retail investors on bond return volatility seems to be larger for ETFs than for OEMFs (Figure 5, panel 1). There are two explanations for this observation: first, ETF and OEMF investor types, retail or institutional, may differ in their investment behaviors. Institutional ETF investors may stand out in particular; they might engage in more frequent trading, shorting, and might generally have a shorter investment horizon than institutional investors holding OEMF shares. Second, the investment vehicle’s structural features matter. Even if the retail investors in ETF shares would be very similar to the retail investors in OEMF shares, differences in terms of the relationship between holdings and bond volatility can arise because of differences in the investment vehicles design. As previously noted, the market for underlying securities is partially shielded from the trading activity in ETF shares, while by contrast, OEMFs could be subject to forced sales. If the design features of ETFs indeed explain the benign impact of retail ETF holdings on bond return volatility, then these design features should also help to partially (but not necessarily fully) offset the impact of the more aggressive investment behavior of institutional ETF investors.
Robustness tests suggest that self-selection of bonds into OEMF and ETF portfolios is limited and does not qualitatively alter the conclusions of the analysis. A primary source of endogeneity arises from the “self-selection” of bonds into OEMF and ETF portfolios based on unobserved characteristics. To test for this type of endogeneity, the analysis was performed for various restricted subsamples constructed based on bonds’ ownership structure (Annex Table 2.2). Coefficient signs remain unchanged, and also the magnitudes of coefficients are comparable. Annex Table 2.3 demonstrates that the results are stable under different fixed effect specifications.
The Relationship between ETF and MF Holdings and Bond Volatility
Citation: Global Financial Stability Notes 2025, 002; 10.5089/9798229003087.065.A001
Sources: Bloomberg Finance L.P.; Refinitiv; Upper; and IMF staff calculations.Notes: Regression tables can be found in Annex 2. ETF and MF holdings are computed at security level and expressed as a percentage of the security’s outstanding amount. The dependent variable is the quarterly volatility of daily returns, annualized, and expressed in basis points. ETF = exchange-traded fund; MF = mutual fund; OEMF = open-ended mutual fund.Discussion and Conclusions
This note provides evidence that ETF ownership of corporate bonds is generally associated with lower bond return volatility. It is well documented that OEMFs can transmit non-fundamental demand shocks to security markets, thereby increasing asset price volatility (International Monetary Fund 2022). However, some prior evidence points toward equity ETFs increasing asset price volatility (Ben-David, Franzoni, and Moussawi 2018). This note sheds light on the underlying dynamics and suggests that differences between institutional and retail ETF investors could explain the mixed results in the literature, depending on the ownership structure of the bonds in the considered samples.
The analysis underscores the importance of clientele effects when assessing the impact of ETFs on bond market volatility. While ETFs ownership of bonds is, on aggregate, associated with lower return volatility, there is a significant divergence between institutional and retail ETF ownership. While retail ETF ownership may act as a stabilizing force, institutional ETF holdings are associated with higher return volatility, especially during periods of heightened market stress. These findings corroborate recent work that suggests that institutional investors, notably OEMFs, may use ETF shares as a buffer against outflows, especially during stressed market conditions.
Future research should further explore the nature of these clientele effects. While the analysis in this note distinguishes between institutional and non-institutional investors, a more detailed breakdown of the different types of institutions could provide deeper insight into the mechanisms at play. In addition, examining the role of funds’ management approaches (active versus passive) could shed further light on these dynamics.9 It is also important for policymakers to understand how the structural features of an investment vehicle affect its investor clientele as changes to these structural features could lead to a reallocation of investor clienteles, merely transferring vulnerabilities rather than resolving them.
The role of ETFs for market stability needs to be investigated further. While the results in this note seem to paint a benign picture regarding the relationship between ETF holdings and market volatility, further research is needed. First, the large adverse impact of institutional ETF holdings on market volatility at higher levels of stress needs to be understood better. Second, the role of friction in the share creation and redemption mechanism requires further investigation. Finally, higher frequency data and different market stability measures could shed further light on the impact of ETFs and their different investor types during major selloffs. For example, the relationship between bond returns during flight-like episodes and holdings by ETFs/OEMFs could be informative for assessing financial stability risks.
References
Allaire, Nolwenn, Johannes Breckenfelder, and Marie Hoerova. 2023. “Fund Fragility: The Role of Investor Base.” Working Paper Series, No. 2874.
Ben-David, Itzhak, Francesco Franzoni, and Rabih Moussawi. 2018. “Do ETFs Increase Volatility?” The Journal of Finance 73 (6): 2471–535.
Bouveret, Antoine, and Jie Yu. 2021. “Risks and Vulnerabilities in the U.S. Bond Mutual Fund Industry.” IMF Working Paper, WP/21/109.
Cetorelli, Nicola, Fernando M. Duarte, and Thomas M. Eisenbach. 2016. “Are Asset Managers Vulnerable to Fire Sales?” Liberty Street Economics 20160218, Federal Reserve Bank of New York.
Chen, Qi, Itay Goldstein, and Wei Jiang. 2010. “Payoff Complementarities and Financial Fragility: Evidence from Mutual Fund Outflows.” Journal of Financial Economics 97 (2): 239–62.
Dannhauser, Caitlin D., and Saeid Hoseinzade. January 2022. “The Unintended Consequences of Corporate Bond ETFs: Evidence from the Taper Tantrum.” The Review of Financial Studies 35 (1): 51–90.
Dekker, Lennart, Luis Molestina Vivar, and Christian Weistroffer. 2024. “Passing on the Hot Potato: The Use of ETFs by Open-Ended Funds to Manage Redemption Requests.” ECB Working Paper No. 2024/2963, July.
Falato, Antonio, Itay Goldstein, and Ali Hortaçsu. 2021. “Financial Fragility in the COVID-19 Crisis: The Case of Investment Funds in Corporate Bond Markets.” Journal of Monetary Economics 123: 35–52.
Gorbatikov, Evgenii, and Taisiya Sikorskaya. 2022. “Two APs Are Better Than One: ETF Mispricing and Primary Market Participation.” May 2. https://ssrn.com/abstract=3923503
Goldstein, Itay, Hao Jiang, and David T. Ng. 2017. “Investor flows and fragility in corporate bond funds.” Journal of Financial Economics Volume 126 Issue 3: 592–613.
Helmke, Anna. 2023. “Will ETFs Drive Mutual Funds Extinct?” Jacobs Levy Equity Management Center for Quantitative Financial Research Paper, December.
International Monetary Fund. 2022. “Asset Price Fragility in Times of Stress: The Role of Open-End Investment Funds.” Global Financial Stability Report, Chapter 3, October.
Jiang, Hao, Yi Li, Zheng Sun, and Ashley Wang. 2022. “Does Mutual Fund Illiquidity Introduce Fragility into Asset Prices? Evidence from the Corporate Bond Market.” Journal of Financial Economics (JFE) 143 (1): 277–302.
Ma, Yiming, Kairong Xiao, and Yao Zeng. October 2022. “Mutual Fund Liquidity Transformation and Reverse Flight to Liquidity.” Review of Financial Studies 35 (10): 4674–711.
O’Hara, Maureen, and Xing (Alex) Zhou. October 2021. “Anatomy of a Liquidity Crisis: Corporate Bonds in the COVID-19 Crisis.” Journal of Financial Economics 142 (1): 46–68.
Sherrill, D. Eli, Sara E. Shirley, and Jeffrey R. Stark. 2020. “ETF Use among Actively Managed Mutual Fund Portfolios.” Journal of Financial Markets, Elsevier 51 (C): 100529.
Todorov, Karamfil. 2021. “The Anatomy of Bond ETF Arbitrage.” BIS Quarterly Review, March.
Annex 1. Dataset
The empirical analysis is based on a dataset comprising the following data sources.
Description of Dataset
Description of Dataset
| Data Source | Level | Notes | Key |
|---|---|---|---|
| Bloomberg Finance L.P. | Fund (ETF) | Fund strategy; estimate of institutional holdings; short positions | Fund ticker/ISIN |
| Refinitiv | Security | Bond price, volatility, market value outstanding, static security characteristics | Bond ISIN |
| Lipper | Fund (MF) | Share class sizes (institutional, non-institutional) | Fund ISIN |
| FactSet | Fund × Security | Fund holdings by ISIN | Fund ISIN, security ISIN |
Description of Dataset
| Data Source | Level | Notes | Key |
|---|---|---|---|
| Bloomberg Finance L.P. | Fund (ETF) | Fund strategy; estimate of institutional holdings; short positions | Fund ticker/ISIN |
| Refinitiv | Security | Bond price, volatility, market value outstanding, static security characteristics | Bond ISIN |
| Lipper | Fund (MF) | Share class sizes (institutional, non-institutional) | Fund ISIN |
| FactSet | Fund × Security | Fund holdings by ISIN | Fund ISIN, security ISIN |
Securities are included in the sample if they are classified as corporate bonds. Corporate bonds need to meet both of the following criteria:
1. Denoted as fixed income securities.
2. Issued by nonfinancial corporations and financial services companies.
Description of Variables
Description of Variables
| Variable | Description | Units |
|---|---|---|
| ETF | ETF holdings of security | Holdings expressed as a share of the security’s outstanding amount for the given quarter, in percent |
| ETF institutional | Share of ETF holdings of security attributable to institutional investors | |
| ETF retail | Share of ETF holdings of security attributable to non-institutional investors, which are assumed to be retail | |
| MF | MF holdings of security | |
| MF institutional | Share of MF holdings of security attributable to institutional investors | |
| MF retail | Share of MF holdings of security attributable to non-institutional investors, which are assumed to be retail | |
| Active | Share of holdings of security by actively managed funds (MFs and ETFs) | |
| Passive | Share of holdings of security by passively managed funds (MFs and ETFs) | |
| Value outstanding | Market value outstanding of the given security in the given quarter | |
| Time-to-maturity | Time to maturity in years | |
| Time FE | Quarterly Fixed Effect | |
| Issuer type FE | “FIN,” “CORP” | |
| Bond type FE | “STR,” “FLO,” “CVT,” “IXL,” “ZER” | |
| Rating FE | “NR,” “A,” “BBB,” “BB,” “B,” “CCC,” “AA,” “D,” “AAA,” “CC,” “C” | |
Description of Variables
| Variable | Description | Units |
|---|---|---|
| ETF | ETF holdings of security | Holdings expressed as a share of the security’s outstanding amount for the given quarter, in percent |
| ETF institutional | Share of ETF holdings of security attributable to institutional investors | |
| ETF retail | Share of ETF holdings of security attributable to non-institutional investors, which are assumed to be retail | |
| MF | MF holdings of security | |
| MF institutional | Share of MF holdings of security attributable to institutional investors | |
| MF retail | Share of MF holdings of security attributable to non-institutional investors, which are assumed to be retail | |
| Active | Share of holdings of security by actively managed funds (MFs and ETFs) | |
| Passive | Share of holdings of security by passively managed funds (MFs and ETFs) | |
| Value outstanding | Market value outstanding of the given security in the given quarter | |
| Time-to-maturity | Time to maturity in years | |
| Time FE | Quarterly Fixed Effect | |
| Issuer type FE | “FIN,” “CORP” | |
| Bond type FE | “STR,” “FLO,” “CVT,” “IXL,” “ZER” | |
| Rating FE | “NR,” “A,” “BBB,” “BB,” “B,” “CCC,” “AA,” “D,” “AAA,” “CC,” “C” | |
Dataset Descriptive Statistics
Dataset Descriptive Statistics
| Periods covered | 2013:Q3–2022:Q4 |
| Number of unique ETFs | 317 |
| Number of unique mutual funds | 1,926 |
| Number of unique bond ISIN | 15,887 |
| ETFs: assets under management (interquartile range) | $193.2 million to $5,867.2 million |
| MFs: assets under management (interquartile range) | $552.7 million to $11,720.6 million |
Dataset Descriptive Statistics
| Periods covered | 2013:Q3–2022:Q4 |
| Number of unique ETFs | 317 |
| Number of unique mutual funds | 1,926 |
| Number of unique bond ISIN | 15,887 |
| ETFs: assets under management (interquartile range) | $193.2 million to $5,867.2 million |
| MFs: assets under management (interquartile range) | $552.7 million to $11,720.6 million |
Variable Descriptive Statistics
Variable Descriptive Statistics
| Variable | Unit | 25th Percentile | Median | 75th Percentile |
|---|---|---|---|---|
| MF | Holdings as a percentage of security’s outstanding | 3.0 | 7.4 | 14.9 |
| ETF | 0.1 | 1.2 | 3.1 | |
| MF institutional | 1.6 | 4.0 | 8.4 | |
| MF retail | 0.8 | 2.5 | 6.1 | |
| ETF institutional | 0.1 | 0.8 | 2.1 | |
| ETF retail | 0.1 | 0.3 | 0.9 | |
| Outstanding value | Million US dollar | 334 | 522 | 871 |
Variable Descriptive Statistics
| Variable | Unit | 25th Percentile | Median | 75th Percentile |
|---|---|---|---|---|
| MF | Holdings as a percentage of security’s outstanding | 3.0 | 7.4 | 14.9 |
| ETF | 0.1 | 1.2 | 3.1 | |
| MF institutional | 1.6 | 4.0 | 8.4 | |
| MF retail | 0.8 | 2.5 | 6.1 | |
| ETF institutional | 0.1 | 0.8 | 2.1 | |
| ETF retail | 0.1 | 0.3 | 0.9 | |
| Outstanding value | Million US dollar | 334 | 522 | 871 |
Annex 2. Regression Tables
This annex provides regression tables for the main result presented in this note. Table 2.1 shows the full results presented in Figure 5, panel 1. Annex Tables 2.2 and 2.3 provide additional regressions to test robustness of the results.
Annex Table 2.2 tests whether self-selection skews the results. Self-selection could undermine the causal interpretation if—for example—funds would seek out less or more volatile bonds or if other factors affect both a security’s volatility and its likelihood to be held by ETFs and mutual funds. The regression is performed in subsamples, whereby subsamples are bound by the share of a security’s outstanding that is held by either ETFs or mutual funds. The results are consistent across various buckets, suggesting that self-selection does not play a major role.
In Annex Table 2.3, different combinations of fixed effects are used to test the sensitivity of the results to different groupings. Once again, the results show a consistent picture.
Regression: Institutional versus Retail Bond Holdings of ETFs and MFs
Regression: Institutional versus Retail Bond Holdings of ETFs and MFs
 |
Regression: Institutional versus Retail Bond Holdings of ETFs and MFs
|
Robustness: Self-Selection
Robustness: Self-Selection
| Main Result (all F) | 0% < F < 5% | 5% < F < 10% | F > 10% | |||||
|---|---|---|---|---|---|---|---|---|
| ETF institutional | 27.42*** | (5.78) | 53.52*** | (10.29) | 39.53*** | (10.14) | 27.97*** | (5.07) |
| MF institutional | -1.18** | (0.53) | 1.76 | (1.96) | 5.10* | (2.89) | -0.34 | (0.80) |
| ETF retail | -85.29*** | (7.24) | 150.40*** | (21.67) | -144.56*** | (13.95) | -75.76*** | (6.93) |
| MF retail | -1.48*** | (0.37) | 1.91 | (2.52) | -1.86 | (2.58) | -0.92 | (0.60) |
| Value outstanding | -0.00 | (0.01) | -0.04*** | (0.01) | 0.02** | (0.01) | 0.00 | (0.01) |
| Time-to-maturity | 33.17*** | (2.10) | 34.33*** | (2.15) | 35.05*** | (2.49) | 30.29*** | (1.95) |
| Constant | 435.77*** | (25.99) | 427.87*** | (24.53) | 368.14*** | (41.70) | 452.96*** | (27.58) |
| Time FE | yes | yes | yes | yes | ||||
| Issuer type FE | yes | yes | yes | yes | ||||
| Bond type FE | yes | yes | yes | yes | ||||
| Rating FE | yes | yes | yes | yes | ||||
| Clustering | T | T | T | T | ||||
| Observations | 245,773 | 69,655 | 56,224 | 117,727 | ||||
| R2 | 0.55 | 0.53 | 0.64 | 0.52 | ||||
| Adjusted R2 | 0.55 | 0.53 | 0.64 | 0.52 | ||||
Robustness: Self-Selection
| Main Result (all F) | 0% < F < 5% | 5% < F < 10% | F > 10% | |||||
|---|---|---|---|---|---|---|---|---|
| ETF institutional | 27.42*** | (5.78) | 53.52*** | (10.29) | 39.53*** | (10.14) | 27.97*** | (5.07) |
| MF institutional | -1.18** | (0.53) | 1.76 | (1.96) | 5.10* | (2.89) | -0.34 | (0.80) |
| ETF retail | -85.29*** | (7.24) | 150.40*** | (21.67) | -144.56*** | (13.95) | -75.76*** | (6.93) |
| MF retail | -1.48*** | (0.37) | 1.91 | (2.52) | -1.86 | (2.58) | -0.92 | (0.60) |
| Value outstanding | -0.00 | (0.01) | -0.04*** | (0.01) | 0.02** | (0.01) | 0.00 | (0.01) |
| Time-to-maturity | 33.17*** | (2.10) | 34.33*** | (2.15) | 35.05*** | (2.49) | 30.29*** | (1.95) |
| Constant | 435.77*** | (25.99) | 427.87*** | (24.53) | 368.14*** | (41.70) | 452.96*** | (27.58) |
| Time FE | yes | yes | yes | yes | ||||
| Issuer type FE | yes | yes | yes | yes | ||||
| Bond type FE | yes | yes | yes | yes | ||||
| Rating FE | yes | yes | yes | yes | ||||
| Clustering | T | T | T | T | ||||
| Observations | 245,773 | 69,655 | 56,224 | 117,727 | ||||
| R2 | 0.55 | 0.53 | 0.64 | 0.52 | ||||
| Adjusted R2 | 0.55 | 0.53 | 0.64 | 0.52 | ||||
Robustness: Fixed Effects
Robustness: Fixed Effects
 |
Robustness: Fixed Effects
|
Based on EPFR statistics. Other funds not classified as bond funds, for example, mixed funds, may also hold bonds.
In addition, ETFs can select the securities baskets that they want to exchange for ETF shares. Differences between the creation and redemption baskets can help to absorb shocks to the bond market (Todorov 2021).
Based on the London Stock Exchange Group Lipper classification. Index funds are classified as passively managed. Note that while there are more active US bond exchange-traded funds (ETFs) than passive ones, active ETFs tend to be smaller than passive ETFs, and weighted by assets under management, passive ETFs dominate.
Institutional holdings of open-ended mutual funds are measured by the share of assets under management held in institutional share classes. Institutional holdings of ETF shares are based on 13F filings. Form 13F is a quarterly report required by the US Securities and Exchange Commission for “institutional investment managers” who control over $100 million in assets.
See Andrew McCollum, “ETFs: Valuable Versatility in a Newly Volatile Market”, Greenwich Associates, 2018. Of the survey respondents, 56 percent indicated that they use ETFs for liquidity management purposes.
ETF flows reflect transactions between the ETF and its authorized participants; secondary market trading on the exchange does not constitute a flow in or out of the ETF.
Details about the dataset are presented in Annex Table 1.1, Annex Table 1.2, Annex Table 1.3 and Annex Table 1.4.
Importantly, institutional OEMF share classes tend to include investment through defined contribution retirement accounts.
In unreported results, we find that passively managed fund holdings tend to be associated with higher bond return volatility. This may be driven by active funds’ greater discretion to avoid trading certain bonds at times when liquidity is limited.
