Abstract
This note analyzes the stress experienced (and caused) by open-end mutual funds during the March COVID-19 stress episode, with a focus on global fixed-income funds. In light of increased valuation uncertainty, funds experienced a short period of intense withdrawals while the market liquidity of their holdings deteriorated substantially. To cover redemptions, afflicted funds predominantly shed liquid assets first—for example, cash, cash equivalents, and US Treasury securities. But forced asset sales amplified price pressures in markets and contributed to liquidity falling across fixed-income markets. This drop in market liquidity, as well as the general stress in financial markets, may have led to fund investors becoming even more sensitive to challenging portfolio performance and encouraged further withdrawals. Only after central banks intervened, directly and indirectly supporting asset managers, did liquidity and redemption stress subside. Overall, the March episode validated the financial-stability concerns about liquidity vulnerabilities in the fund industry and calls for further action to address them.
Evolution of Fund Flows During COVID-191
In March 2020 the COVID-19 crisis led to an unexpected increase in uncertainty and rapidly falling asset prices in equity and credit markets. Some market participants “dashed for cash” by selling off relatively safe and liquid assets, while central banks and certain types of money market funds stepped in, as was seen in US Treasury securities markets (Figure 1, panel 1). Funds experienced a period of rapidly intensifying daily outflows in March 2020 (Figure 1, panel 2, left side), which—according to monthly Morningstar data—amounted to outflows from the global fixed-income fund sector of $481 billion in March 2020. Even money market funds, which are generally less exposed to credit risk, were affected: in particular, US prime money market funds faced several weeks of large redemptions, while government money market funds saw inflows as investors sought the safety and liquidity of government securities (Figure 1, panel 2, right side).
Asset Sales, Fund Flows, and Liquidity
Citation: Analytical Notes 2020, 002; 10.5089/9781513563695.064.A001
Sources: Board of Governors of the Federal Reserve System; Morningstar; and Refinitiv Datastream.Note: The monetary authority sector in panel 1 is the group of institutions and financial accounts that supply reserve funds to depository institutions and absorb funds from them. The sector includes the 12 Federal Reserve Banks and their subsidiary offices. Panel 2 shows asset-weighted average cumulative daily net flows for equity, fixed-income, and mixed funds larger than $0.5 billion and all MMFs and hedge funds larger than $50 million. Hedge funds include only open-end funds with alternative strategies. Coverage is limited due to the availability of daily flow and asset data: the figure covers at the end of June 2020 45 percent of the equity funds, 43 percent of the fixed-income funds, 30 percent of the mixed funds, and 57 percent of the money market funds reported by the International Investment Fund Association for the global fund sector, including funds of funds. For hedge funds no coverage data are available, since the available data universe for the hedge fund sector also includes closed-end funds. The sample covers the time from January 1, 2020 to August 12, 2020. In panel 3 portfolio compositions are observed at the beginning of each month only. Bid-ask spreads are based on end-of-day composite bid and ask prices. ETF = exchange-traded fund; FI = fixed-income; MMF = money market fund.At the same time, the market liquidity of securities held by fixed-income funds deteriorated substantially: in a sample of fixed-income funds average bid-ask spreads almost doubled (Figure 1, panel 3). For some fixed-income fund portfolios, average bid-ask spreads more than tripled temporarily, indicating that a few funds bore the brunt of the liquidity impact.
With unprecedented central bank policy support, daily flows returned to their pre–COVID-19 levels in less than a month (IMF 2020), and there have not been any widespread fund closures so far.2 Nevertheless, the episode rekindled a debate among market participants, policymakers, and researchers about financial stability risks emanating from an asset management sector that has grown substantially since the global financial crisis (FT 2020). In response to large outflows, funds holding relatively illiquid assets but offering near-daily redemptions could be vulnerable to runs, forcing them to sell assets in relatively illiquid markets (IMF 2015, 2019; Chen, Goldstein, and Jiang 2010). If multiple funds are affected, fire sales could lead to a further decline in asset liquidity and asset prices (Coval and Stafford 2007).
This note attempts to inform the ongoing discussion about risks and vulnerabilities and their mitigation through preliminary insight into the liquidity and redemption management of global fixed-income funds during the recent COVID-19 episode. For this purpose, the note explores three questions: (1) How did fixed-income funds manage liquidity risk, and how did they react to the jump in redemption requests; that is, how did they raise the cash needed to meet investors’ capital withdrawals? (2) Is there evidence that redemption pressure during the COVID-19 shock led to fire sales by asset managers? (3) What factors made funds more prone to capital redemption by end investors when market conditions were stressed?
Funds’ Portfolio Composition in Response to Redemption Shocks
With only a handful of funds suspending redemptions, most fixed-income funds resorted to a mix of strategies to deal with outflows. The most afflicted funds used their relatively ample liquidity buffers and shed liquid assets such as cash and cash equivalents to cover redemptions; the small number of funds that received inflows hoarded cash and delayed investment, presumably because of uncertain market conditions (Figure 2, panel 1).3
Fixed-Income Funds Met High Outflows with Liquid Assets if Possible, but Massive Liquidation Needs Still Lowered Portfolio Risks
Citation: Analytical Notes 2020, 002; 10.5089/9781513563695.064.A001
Source: Morningstar.Note: All panels, except panel 4, show balanced samples; that is, only funds that do not enter or exit during January 2017 and April 2020 are included. In panel 1, Treasury bills and other securities that mature in less than 92 days are included in cash equivalents. In panels 1–3, the sets of funds reporting respective portfolio components differ, while the set of funds reporting flows remains unchanged. Samples include available monthly data from January 2015 to April 2020 for fixed-income funds with assets over $0.5 billion.As funds facing outflows were burning through liquid assets, the proportion of less liquid assets in their portfolios, such as corporate bonds and to a lesser extent sovereign bonds, increased (Figure 2, panels 2 and 3). Such increases were particularly pronounced in funds with high exposure to corporate credit, which faced heavy outflows.
Despite these portfolio reallocations observed in March 2020, funds’ portfolios saw marginal improvements in their average rating quality profiles in April 2020, suggesting that while asset managers sold predominantly liquid assets, they did not let the risk profile of their holdings deteriorate, at least over the short time period assessed in this note (Figure 2, panel 4).
Quantile regressions confirm that, in normal times, fixed-income funds experiencing high outflows are likely to cover redemptions with cash, cash equivalents, and sovereign bonds while actually holding onto corporate exposures. More specifically, in normal times, outflows are accompanied by declines in cash, its equivalents, and sovereign bonds, especially for funds facing either pronounced out- or inflows (Figure 3, panels 1–2). Corporate bond positions, on the other hand, are negatively correlated with fund net flows, especially at the tails of the fund flow distribution; that is, outflows are accompanied by increases in holdings of corporate bonds (Figure 3, panel 3). The tendency for more pronounced flow reactions at more extreme quantiles may stem from delays in portfolio re-optimization as well as differing time needs in the purchase of assets of varying liquidities.
Fund Flows and Portfolio Characteristics
Citation: Analytical Notes 2020, 002; 10.5089/9781513563695.064.A001
Sources: Bloomberg Finance L.P.; and Morningstar.Note: Numbers reported are coefficients significant at the 5 percent level in unconditional quantile regressions of fund net flows, measured in percent of assets, on portfolio shares of cash, corporate bonds, and sovereign bonds and on contemporaneous returns, fund size, fund age, a quarter dummy, and a COVID-19 dummy, as well as interactions of the latter with cash, corporate bonds, sovereign bonds, and returns and a set of macro-financial variables, including the Chicago Board Options Exchange Volatility Index, a term spread, a credit risk spread, a proxy for US interest levels, and a basket of major exchange rates versus the US dollar. Fund fixed effects are included. Samples include available monthly data for fixed-income funds with assets over $0.5 billion from January 2015 to May 2020. In panels 1–3 coefficients display the individual impact of each variable on net flows, implying that net effects are generated by the accumulation of the individual coefficients.However, as outflows accelerated during the COVID-19 episode for the more vulnerable tail of funds (Figure 3, panel 4), funds modified their behavior. In the period starting February 2020 sensitivities observed for the lower-flow deciles increased in some cases for cash and cash equivalents, more than doubled for sovereign bonds, and lost substantially in strength for corporate bonds (Figure 3, panels 1–3).
Taken together, this evidence is for the most part consistent with funds following a “waterfall” approach; that is, instead of keeping portfolio weights unchanged, funds sold the most liquid assets first. However, although funds facing extreme outflows tended to sell less risky, higher-quality assets first, during the COVID-19 period—compared with previous periods of stress—they were less reluctant to shed riskier assets, such as corporate bonds.
Funds’ Liquidity Conditions and Selling Pressure in Fixed-Income Markets4
During the COVID-19 stress episode, funds’ liquidity profiles deteriorated together with overall market conditions. While all funds saw the liquidity of their portfolios decline, liquidity reached the lowest levels for funds facing strong outflows (Figure 4, panel 1).
Fixed-Income Funds: Liquidity and Fund Flows
Citation: Analytical Notes 2020, 002; 10.5089/9781513563695.064.A001
Sources: Morningstar; and Refinitiv DatastreamNote: Panel 1 shows value-weighted bid-ask spreads and average net flows by flow quintile. Cash and cash equivalents are assumed to have no bid-ask spread. Panel 2 shows the bid-ask spread of the assets bought and sold in a given month relative to the bid-ask spread of the fund’s portfolio. The bid-ask spread of assets sold and bought is the average bid-ask spread in the month the assets were sold or bought. Cash positions are not considered.As shown above, funds facing outflows sold safe and liquid assets first in order to avoid realizing losses. Figure 4, panel 2, shows the trading behavior of fixed-income funds expressed as the bid-ask spread of assets bought and sold relative to the average bid-ask spread of a fund’s portfolio.
For funds facing outflows, the liquidity differential between assets sold and assets bought increased in the first quarter of 2020. For funds facing modest outflows or inflows on the other hand the bid-ask spreads of assets bought are generally close to the bid-ask spreads of assets sold, even though assets bought tend to be less liquid than assets sold. Hence, some funds’ willingness to absorb relatively illiquid assets even under uncertain market conditions worsened their portfolios’ liquidity positions further but may have also helped to marginally mitigate price pressures.
Even though fixed-income funds seem to have sold predominantly liquid assets first, the large off-loading of assets in March 2020 may have had some adverse effect on both asset prices and market liquidity, although mitigated by the unprecedented and speedy policy support.
In March 2020 the bid-ask spreads of assets most heavily sold across all fixed-income funds facing large redemptions increased more than the bid-ask spreads of assets not facing such selling pressure (Figure 5, panel 1). As a consequence, during March 2020, cumulative returns of assets under selling pressure declined more than those of assets experiencing no selling pressure. Hence, funds’ sales of liquid assets are likely to have contributed to price pressures and liquidity strains observed in core fixed-income markets such as US Treasury markets. Similarly, funds increasingly having to sell corporate bonds may have amplified the price dislocations observed in risky credit markets in March 2020.
Selling Pressure and Fire Sales
Citation: Analytical Notes 2020, 002; 10.5089/9781513563695.064.A001
Sources: Morningstar; and Refinitiv Datastream.Note: Panels 1 and 2 are based on detailed portfolio holdings data of 390 fixed-income funds holding approximately 13,000 identifiable securities in March 2020. Prices and bid-ask spreads are computed based on Refinitiv composite end-of-day bid and ask prices. Selling pressure of security i in March 2020 is defined similarly to the definition in Coval and Stafford (2007) as the fraction of flow-motivated trading in a security’s average monthly trading volume. Flow-motivated trading is the difference between a security’s purchases by funds experiencing inflows higher than 90 percent of their peers and sales by funds facing outflows higher than 90 percent of their peers. The mentioned fraction defines a security as experiencing high selling pressure if it is in the bottom decile of the ratio’s distribution across all securities; it is considered to experience no pressure if this ratio is positive.Determinants of Investor Flows
The evidence so far suggests that funds tend to prioritize a waterfall approach when liquidating assets in periods of severe redemption stress, selling relatively liquid assets first. While this approach may have prevented liquidity conditions from deteriorating even further in already illiquid assets, it also left funds’ remaining investors with fewer liquid asset holdings. Moreover, the sheer size of the outflows observed in March 2020 required sales of riskier portfolio components, which may have also contributed to selling pressure in this market segment.
Funds holding relatively illiquid assets but offering near-daily redemptions are inherently fragile because investors who redeem their shares do not necessarily absorb all the costs from trading. This leads to strategic complementarities that can amplify outflows following, during, or in anticipation of bad performance (Chen, Goldstein, and Jiang 2010). As a consequence, the relationship between flows and poor performance is expected to be stronger for funds facing outflows, at least during periods of extreme stress, when market liquidity deteriorates, such as the COVID-19 period or other stress periods of comparable severity.
In line with this hypothesis, sensitivities of flows to contemporaneous returns deviated in the COVID-19 episode substantially from what was observed over the previous five years: sensitivities increased for funds with very high outflows (Figure 6, panel 1). For the bottom decile of net flows the respective sensitivity increased from 0.06 percent to 0.54 percent for each 1 percent increase in returns. Thus, during the COVID-19 episode, it appears that investors started sanctioning the inferior performance of funds already facing large outflows with additional redemption requests.5
Flow-Performance Relationship
Citation: Analytical Notes 2020, 002; 10.5089/9781513563695.064.A001
Sources: Bloomberg Finance L.P.; Morningstar; and Refinitiv Datastream.Note: Displayed coefficients are significant at the 5 percent level. The model specification for panel 1 is the same as the one reported in Figure 3. Panel 2 shows the coefficient of a piecewise linear regression of daily net flows on daily returns over the period January 2019–May 2020. Returns are first grouped by date and global category group and ranked on a scale from 0 to 100, then “Low” is defined as min(0.2, rank), “Medium” as min(0.6, rank), and “High” as (rank-(low+mid)).Similarly, looking at daily fund net flows and differentiating flow sensitivities across performance levels reveals that flows tend to react generally more to returns for funds showing either high or low performance levels, relative to those in the center of the return distribution (Figure 6, panel 2).6 During the COVID-19 pandemic, however, the sensitivity of flows of funds facing low returns more than doubled, to 0.86 percent for each 1 percent increase in returns.
These results thus point to an increased inclination of investors to redeem their shares when perceiving mounting valuation risk for their investments as they did during the COVID-19 episode. Such behavior suggests that feedback mechanisms from the devaluation in securities prices to capital withdrawals by investors may have intensified and could have led to runs on funds, had liquidity conditions deteriorated further.
Conclusions
Assets under management by investment funds have grown rapidly since the global financial crisis (FSB 2020). As previously highlighted (IMF 2015, 2019), funds can be subject to fragilities because of the mismatch between the liquidity of funds’ underlying assets and end investors’ ability to redeem on a near-daily basis. Second-round effects can occur when redemptions cause selling pressure that then reinforces price declines and leads to further redemptions. Recognizing these risks, policymakers have called for the implementation of various liquidity and redemption management tools, such as swing pricing, redemption fees, and other anti-dilution methods (FSB 2017; IOSCO 2018), which until recently had been largely untested in a market-wide stress scenario. Similarly, fund sector-wide liquidity stress scenario exercises, used in numerous reports of the IMF’s Financial Stability Assessment Program and by other institutions as well, picked up on these particular risks.
The decline in prices across various asset classes and subsequent large portfolio losses and significant outflows observed in March 2020 presented an unparalleled stress test for the fund industry. Central banks reacted with unprecedented speed to the market turmoil and implemented wide-ranging policy measures to preserve market functioning and maintain the flow of credit to the economy. Among these measures were lending against risky assets, including commercial paper, mortgage-backed securities, corporate bonds, and exchange-traded fund shares, and offering credit to primary dealers and money market funds. Supported by rebounding asset prices and improved investor confidence, fund flow patterns normalized relatively quickly.7
Even though the stress episode in March 2020 was relatively short-lived, this note highlights important weaknesses funds faced. Most funds sold relatively liquid assets first, which could have deepened their redemption stress if liquidity conditions had not improved following massive central bank interventions. When looking at the flow-performance relationship, fund investors seem to have responded to this risk very consciously, by reducing their holdings in underperforming funds already exposed to high redemption pressure.
Even though funds only marginally engaged in large-scale sales of low-liquidity assets there was some evidence of price declines in markets following selling pressure, which highlights the risk of fire sales if funds must unload relatively illiquid assets quickly.
Looking ahead, a comprehensive review of available prudential tools in the investment fund sector, including a broad set of tools to better manage redemptions—such as enhanced use of minimal requirements for liquidity buffers; more widespread adoption of swing pricing; or, if that is not possible, requiring funds to better match redemption periods to the liquidity profiles of their portfolios—could help mitigate vulnerabilities revealed during the COVID-19 market turmoil (FSB 2020; IMF 2019).8
