Elsevier

Journal of Financial Economics

The use of credit default swaps by bond mutual funds: Liquidity provision and counterparty risk☆

Abstract

Corporate bond mutual funds increased their selling of credit protection in the credit default swaps (CDS) market during the 2007–2008 financial crisis. This trading activity was primarily in multi-name CDS, greater among larger and established funds, and directed toward counterparty dealers in financial distress. Funds that sold credit protection during the crisis experienced greater credit market risk and superior post-crisis performance, consistent with higher expected returns from liquidity provision. Funds using Lehman Brothers as a counterparty experienced abnormal outflows and returns of –2% immediately following Lehman's bankruptcy, suggesting that funds' opportunistic trading in CDS exposed investors to counterparty risk.

Introduction

The assets under management of corporate bond mutual funds expanded dramatically in recent years from $388 billion in 2000 to $2.43 trillion in 2016.1 Bond funds, which primarily invest in investment grade and high yield corporate bonds, often use credit default swaps (CDS) to manage credit risk exposure and gain transactional efficiency.2 However, the CDS market structure also creates opportunities for bond funds and other credit market participants to provide liquidity during periods of market stress. CDS transactions and positions are concentrated among a select group of large dealers (Getmansky et al., 2016). Swap dealers experiencing capital losses could be less willing to bear the risks associated with selling credit protection and, instead, buy protection to actively reduce risk. A sudden shift in dealers' demand for credit protection can increase CDS prices above fundamentals if new capital is slow to move into the markets. Consequently, those who are not capital constrained could be attracted by opportunities to sell credit protection at a premium and, in this sense, provide liquidity. We propose that bond funds played such a role during the 2007–2008 financial crisis.

Our empirical analysis uses a large sample of quarterly portfolio disclosures by corporate bond mutual funds over 2004–2009. The data set is drawn from public regulatory filings and contains detailed information about the CDS positions held by bond funds, including the underlying reference entity or index, position type (i.e., buyer versus seller), and identity of the counterparty dealer. Importantly, our data set covers a period—the 2007–2008 financial crisis—during which many swap dealers and bond insurers experienced capital losses due to their high leverage and exposure to the historical downturn in the mortgage market. As a result, the market's capacity to bear risk was diminished as many levered institutions, like swap dealers, were forced to delever and/or reduce asset risk (Shleifer and Vishny, 2011). We use our data on CDS positions to investigate whether bond funds help provide a stabilizing force during periods of market stress by selling credit protection to dealers in financial distress.

We find that the notional amount of bond funds' net selling of credit protection rises steadily from a bottom of about –1% to 1% of total fund assets over 2007Q2–2009Q1 (see Fig. 1). This increase is mainly driven by an increase in funds' selling of credit protection (see Fig. 2) or, equivalently, from swap dealers' buying of protection from bond funds. Why did swap dealers increase their buying of credit protection from bond funds during the crisis? The highly leveraged nature of dealers exposes them to significant capital losses and even greater leverage when asset prices fall. Consequently, following declines in their net worth, broker-dealers actively manage their risk by delevering or reducing the riskiness of their asset mix (Adrian and Shin, 2010). Our evidence suggests that dealers managed their risk during the crisis, in part, by purchasing credit protection from bond funds in the CDS market.

We present additional corroborating evidence. First, mutual funds' selling of credit protection was primarily in multi-name rather than single-name CDS markets. Multi-name contracts are contingent on the default events for a pool of reference entities, rather than the default of any specific firm. The fact that dealers were mainly hedging non-diversifiable market risk (versus firm-specific risk) aligns well with the fact that the dealers in our sample are well diversified across single-name reference entities.

Second, the net selling of credit protection by bond funds during the crisis is more pronounced among funds with more stable investor capital, as measured by funds with greater assets under management and managers with longer tenure (Spiegel and Zhang, 2013). These funds are better equipped to bear the risk associated with protection selling, avoid investor redemptions, and realize a liquidity premium when market conditions normalize post-crisis.

Third, dealers in greater financial distress were more active in purchasing credit protection from bond funds during the crisis (and, hence, were counterparties to funds' selling of protection). Such dealers were plausibly less willing to bear credit risk and more active in reducing their risk exposure by purchasing credit protection.

Fourth, we find that an increase in protection selling during the crisis predicts greater fund performance over the post-crisis period. Specifically, a 10% (of fund assets) increase in notional value underlying the selling of multi-name CDS is associated with a 3.14% increase in abnormal fund performance over 2009Q2–2009Q4. This estimate is a risk-adjusted measure that goes beyond the returns on similar funds in the same style category, funds with the same investment grade and high yield credit risk exposure, and funds with the same expense ratios. This evidence illustrates the primary benefit to fund investors from opportunistic CDS writing during the crisis.

We also identify two costs from selling protection to distressed dealers during the crisis. First, compared to other funds, net sellers of CDS were more exposed to market-wide credit risk during the crisis (but not before). Funds did not liquidate their corporate bond positions to offset the credit risk obtained from writing CDS. As a result, their investors were more exposed to credit market risk as credit conditions worsened over 2008.

Second, by directing more of their CDS writing to dealers in greater financial distress, funds exposed investors to counterparty credit risk. Using high frequency daily data on mutual fund returns and flows, we find that funds using Lehman Brothers as a CDS counterparty experienced 2% lower returns over the two weeks immediately following Lehman's bankruptcy (but not before). To help establish a causal link between a fund's exposure to Lehman and its post-bankruptcy returns, we find that returns are particularly lower among funds with greater notional value underlying CDS positions arranged with Lehman. In addition, Lehman-exposed funds experience greater outflows over the two-week post-bankruptcy period (but not before), an indication of investors in Lehman-exposed funds reacting negatively to the news of Lehman's default. Overall, our evidence suggests that Lehman's demise impacted Lehman-exposed funds via 1) direct losses in the economic value of funds' CDS positions arranged with Lehman and 2) indirect losses from costly, flow-motivated trading.

Our paper contributes to recent evidence on the pricing of CDS. Stanton and Wallace (2011) find that the prices of ABX index CDS, which are based on baskets of subprime residential mortgage-backed securities, were higher than what would be consistent with reasonable assumptions for mortgage default rates. They argue that the price of default insurance rose above "fair value" due to an increase in the demand for credit protection and a limited amount of capital behind the providers of mortgage bond insurance. Siriwardane (2016) finds that when mega-sellers of credit protection experience a shock to their capital, their risk bearing capacity declines and CDS spreads increase. We take these arguments a step further and show that bond funds that are not capital constrained sell credit protection during periods of market stress, and in this sense, provide liquidity to the CDS market.

Our analysis relates to the literature on how asset managers use derivatives.3 Jiang and Zhu (2016) find that bond funds with higher liquidity risk (e.g., fund flow volatility, illiquid asset holdings) are more likely to either buy or sell CDS, consistent with derivatives providing a liquid alternative to transacting in the underlying bond market. Similarly, in our sample, we find that CDS usage is positively correlated with liquidity risk. However, during the crisis, we find that funds' net selling (i.e., buys minus sells) of CDS is actually greater among funds with more stable investor bases and, hence, lower liquidity risk. This is consistent with such funds having a greater capacity to bear credit risk during a crisis. Jiang and Zhu (2016) also find that funds herd together in selling credit protection on reference entities that are "too large to fail." Our paper adds to this literature by showing that bond funds use multi-name CDS to trade opportunistically as sellers of market-wide credit protection during a crisis period.

Finally, we contribute to the literature on counterparty risk. Existing studies find that the effects of counterparty risk on CDS prices are economically small (Arora et al., 2012, Loon and Zhong, 2014), and that buyers of credit protection avoid transacting with lower quality sellers rather than bearing the risk of a counterparty default (Copeland et al., 2014, Du et al., 2016). We find that bond funds directed more of their protection selling towards lower quality dealers who, presumably, were willing to pay a premium for credit protection. In addition, our evidence that Lehman-exposed funds experienced negative returns and abnormal outflows following the Lehman bankruptcy highlights a risk to bond funds from trading CDS with lower-quality counterparties.

The remainder of the paper is organized as follows. Section 2 discusses the related literature and hypotheses. Section 3 describes the data and provides summary statistics of our bond funds sample. Section 4 discusses the results from our main analysis. Section 5 concludes.

Section snippets

Background and hypothesis development

The extensive use of leverage by swap dealers contributed to significant declines in their net worth as asset prices fell during the crisis.4 Fig. 1 illustrates the rise in financial distress among swap dealers over our sample period. The median cost

Data sources and summary statistics

Here we describe the data used in the empirical analysis and provide summary statistics of bond funds and their CDS positions.

Analysis and results

In this section, we present our main analysis. We investigate the selling of credit protection by bond funds during the crisis and consider whether this activity is associated with a post-crisis premium in fund returns. We also study whether protection selling entails greater exposure to credit market and counterparty risks for fund investors.

Closing remarks

We decipher six years of quarterly portfolio disclosures to provide new evidence on the role of bond mutual funds in credit markets. Bond funds increased their selling of credit protection during the 2007–2008 financial crisis, primarily in multi-name CDS. This protection was purchased mainly by swap dealers in greater financial distress and sold by larger funds and more established fund managers. Protection selling during the crisis also predicts greater abnormal fund returns post-crisis. We

Cited by (17)

View full text

© 2018 Elsevier B.V. All rights reserved.