نقدینگی درون زا در مشتقات اعتباری

We study the determination of liquidity provision in the single-name credit default swap (CDS) market as measured by the number of distinct dealers providing quotes. We find that liquidity is concentrated among large obligors and those near the investment-grade/speculative-grade cutoff. Consistent with endogenous liquidity provision by informed financial institutions, more liquidity is associated with obligors for which there is a greater information flow from the CDS market to the stock market ahead of major credit events. Furthermore, the level of information heterogeneity plays an important role in how liquidity provision responds to transaction demand and how liquidity is priced into the CDS premium.

The infrastructure of the credit default swap (CDS) market has undergone significant transformations in recent years, such as the standardization of documentation and settlement, more streamlined trade processing and confirmation, the mitigation of counterparty risk with central clearing, and improved transparency through the reporting of transaction statistics (Intercontinental Exchange, 2010a; Duffie, Li, and Lubke, 2010). While these developments make it natural to consider a migration toward exchange trading, the CDS market has so far remained an over-the-counter structure dominated by major banks.1 Through direct or electronic communications, often with the assistance of inter-dealer brokers, these banks disseminate bids and offers to potential clients seeking to trade credit protection.2 Consequently, they play a crucial role in providing liquidity to the market. Despite its clear academic and practical relevance, not much is known about the behavior of liquidity provision in the credit derivatives market. In this paper, we offer insights into liquidity determination in the single-name CDS market by exploiting available data on an important dimension of liquidity—the number of distinct dealers providing quotes about a given obligor on any given day, which can be considered as an empirical proxy for market depth. Operating primarily as a quote-driven dealership market, the CDS market nonetheless shares a few characteristics with limit order markets. First, although major banks play a significant role in this market, the barrier to entry is likely to be low because the total number of entities providing quotes can be quite large.3 This blurs the traditional boundary between dealers and non-dealers; rather, it is the decision to supply or take liquidity that distinguishes the participants. Second, electronic trading systems have been developed to facilitate access to dealer quotations. To the extent that investors can obtain quotes from multiple dealers in these trading systems, they may have access to what resembles a small portion of a limit order book. These features suggest that we can think of CDS liquidity from the perspective of the recent literature on endogenous liquidity provision in limit order markets. One potentially important consideration in the decision to supply or demand liquidity is the nature of information possessed by traders. The dominant players in the CDS market, i.e., major banks, may have access to non-public information on CDS obligors through their lending and investment banking activities, and can potentially trade on this information. In an important recent study, Acharya and Johnson (2007, AJ hereafter) analyze the lead–lag relation between the CDS market and the stock market. They show that the information flow from the CDS market to the stock market becomes stronger when a CDS obligor's credit condition deteriorates, and that this information flow is positively related to the number of the obligor's relationship banks.4 This looks, then, like evidence of insider trading or at least informed quote revision by the banks. In classical market microstructure models such as Kyle (1985) and Glosten and Milgrom (1985), an uninformed market maker provides liquidity to other uninformed traders and an informed insider who submit market orders. When the information advantage of the insider increases, the market maker widens the bid-ask spread or decreases the market depth to protect herself against the risk of being exploited by the insider. In other words, an increased presence of information asymmetry is associated with a reduction in liquidity provision by an uninformed market maker. The more recent literature on limit order markets recognizes the decision to provide liquidity as endogenously shaped by the strategic competition among trading agents who might be differentially informed. In this setting, informed traders emerge as natural liquidity providers because their superior information mitigates the adverse selection risk associated with the use of limit orders. In an experimental setting, Bloomfield, O'Hara, and Saar (2005) find that informed traders initially use market orders to exploit the value of their information, but switch to limit orders as this value starts to dissipate. Kaniel and Liu (2006) argue that long-lived private information increases the execution probability of limit orders and makes them more appealing to informed traders. Empirically, they show that limit orders are more informative than market orders on the NYSE. Goettler, Parlour, and Rajan (2009) numerically solve the equilibrium of a dynamic limit order market in which traders decide whether to acquire information about the fundamental value of the traded asset before placing market or limit orders. They show that traders with no inherent motive to trade have the most incentive to acquire information and that they submit most of the limit orders. Overall, these papers confirm that informed traders play an important role in providing liquidity to the market. Much of this literature assumes the sequential arrival of traders and focuses on the resulting dynamics of the limit order book. Meanwhile, the nature of private information is greatly simplified, typically, by assuming common information about the value of the asset among informed traders. In the credit derivatives market, however, major dealers' information about credit risk could be diverse. For instance, the differential information major banks had about the state of the U.S. housing market likely resulted in different exposures to subprime mortgages during the recent credit crisis (e.g., Lehman Brothers vs. Goldman Sachs). A model of endogenous liquidity provision that incorporates heterogeneously informed traders is provided by Boulatov and George (2010). In an extension of the Kyle (1985) model, they assume that informed traders can submit price-contingent supply schedules (which are essentially a collection of limit orders) or market orders. In their model, diverse private information results in imperfect competition among informed traders, who want to earn additional rents by providing liquidity. However, the limit orders that they place can partially reveal their information to the rest of the market. Boulatov and George show that, when the market is “dark” (the consolidated supply schedule is not made known), all informed traders will act as liquidity providers; when the market is “open,” only some of the informed traders will provide liquidity, but the number of liquidity providers is still proportional to the total number of heterogeneously informed traders.5 Given the intuition of “more insiders, more insider trading” in the lightly regulated CDS market (Acharya and Johnson, 2010), one would expect a positive association between the amount of informed trading and the number of quote providers.6 Besides information heterogeneity, liquidity provision in the CDS market is also expected to depend on the frequency of uninformed trading and the riskiness of the CDS obligor. A larger uninformed trading volume, perhaps as a result of higher hedging demand by bond market investors, implies a higher level of profits to be shared among the dealers. Riskier obligors, on the other hand, may be associated with more mispriced dealer quotes that are susceptible to being picked off by market participants (Goettler, Parlour, and Rajan, 2009). Since these determinants of CDS liquidity—information heterogeneity, the frequency of uninformed trading, and the riskiness of the CDS obligor—are potentially correlated with each other as well as common firm attributes, we begin our empirical analysis by simply documenting the relation between the quarterly average number of CDS quote providers and lagged quarterly firm characteristics. Using comprehensive CDS market data spanning 2001–2008 and 732 North American obligors, we identify robust cross-sectional relations between CDS liquidity and several important firm characteristics. First, larger firms and firms with more stock trading volume have more CDS quote providers. Intuitively, such firms are associated with a greater demand for CDS trading from uninformed investors. Second, when conditioning on firm size, we uncover an inverse U-shaped relation between liquidity provision and credit rating, with the peak of liquidity centered around the investment-grade/speculative-grade cutoff. This interesting nonlinear relation could reflect investors' hedging demand and/or dealers' risk of supplying limit orders. Third, the number of CDS quote providers increases with the number of banking relationships of the CDS obligor, which suggests that endogenous liquidity provision in the CDS market is increasing in the level of information heterogeneity. With regard to the connection between liquidity provision and informed trading, we also attempt to estimate a direct link. Specifically, we follow AJ's methodology to measure informed trading by the amount of information flow from the CDS market to the stock market under adverse credit conditions (up to 90 days before a one-day increase in the CDS premium greater than 50 basis points, among others). We find that this conditional information flow is indeed increasing in the number of CDS quote providers, and moreover, our result is robust to controlling for firm characteristics that might be correlated with hedging activities. While such a finding goes against the intuition of classical market microstructure models, it is consistent with the endogenous provision of liquidity by informed traders. We further explore the marginal dealer's decision to start or stop supplying liquidity by analyzing the time-series behavior of the number of quote providers. Using the lagged CDS premium change to proxy for transaction demand, we find that a higher demand leads to more liquidity provision when the market is relatively calm. During the period preceding an abrupt increase of the CDS premium, however, the positive link between transaction demand and liquidity provision is weakened, and can even turn negative among obligors with a large number of quote providers. Since the number of liquidity providers is positively related to the amount of informed trading in the CDS market, especially prior to major credit events, this result suggests that a given increase in demand matters less for the marginal dealer when the existing level of information heterogeneity is high. If CDS liquidity is provided by potentially informed dealers, then it is not clear how the number of quote providers will impact the level of the CDS premium. On one hand, more dealers would bring greater competition, which can lower the premiums customers have to pay, particularly in light of the fact that dealers are net-sellers of CDS to customers. On the other hand, more dealers could also indicate greater informed trading (protection buying) ahead of negative credit news, and this can increase the CDS premium. To explore these distinct possibilities, we examine how lagged changes in liquidity provision predict changes in the CDS premium. Generally, we find that an increase in liquidity provision leads to a small reduction of the CDS premium. However, this effect can go the other way when the existing number of dealers is large. Therefore, it appears that the degree of information heterogeneity also plays an important role in how liquidity is reflected in the CDS premium. To the best of our knowledge, our paper is the first to explore the determinants and implications of endogenous liquidity provision in the CDS market. The prior literature has mainly focused on estimating the liquidity component of CDS premiums or expected CDS returns by exploiting the relation between the CDS premium and various liquidity proxies, such as the bid-ask spread (Tang and Yan, 2007 and Bühler and Trapp, 2010; Bongaerts, de Jong, and Driessen, 2011). Our paper also complements the empirical literature on the determinants of the number of Nasdaq market makers. For instance, Wahal (1997) finds that both the level and change of the number of market makers are explained by trading activities and the riskiness of the stock, and Ellis, Michaely, and O'Hara (2002) further show that exiting market makers have low volume and profit rankings relative to other market makers. Our paper extends this literature by identifying information heterogeneity as a determinant of the level and change of CDS market liquidity. Currently, new regulations regarding over-the-counter (OTC) derivatives market transparency have been proposed by both the Commodity Futures Trading Commission and the Securities and Exchange Commission in an effort to implement the Dodd-Frank Act.7 The proposed changes would put CDS trading on so-called “swap execution facilities” (SEFs), with real-time trade reporting and dealer quotes disseminated to a wider set of investors. These changes are meant to increase market liquidity by intensifying competition among dealers and encouraging participation among non-dealers. However, to the extent that increased transparency reduces information heterogeneity, the results of our paper suggest that this could lead to a reduction of liquidity provision in the CDS market. The rest of the paper proceeds as follows. Section 2 describes the data used in our analysis as well as related summary statistics. Section 3 presents our empirical results. We first analyze the cross-sectional determinants of CDS liquidity, then focus on establishing a direct connection between liquidity and informed trading, and finally explore the time-series determinants of liquidity as well as the pricing of liquidity. Section 4 concludes.

In this paper, we examine the determinants of liquidity provision in the over-the-counter market for credit default swaps. As highlighted by Acharya and Johnson (2007), this is a market dominated by major dealers, in which informed trading or quote updating plays an important role. The recent literature on limit order markets has shown that, when given a choice, informed traders will often choose to act as liquidity providers. Therefore, we treat the CDS market as a laboratory for studying the general properties of endogenous liquidity in dealership markets, in particular how it relates to the level of information heterogeneity. Using the number of quote providers as a measure of CDS market depth, we explore the behavior of CDS liquidity across 732 firms over 2001–2008. Cross-sectionally, we find that large firms and firms near the investment-grade/speculative-grade boundary tend to be the most liquid. This is generally consistent with these firms having the largest amount of hedging demand from investors. We also demonstrate that the amount of information flow from the CDS market to the stock market is increasing in the number of CDS quote providers. This shows that the total amount of CDS liquidity is positively related to the level of informed trading/quote updating in the CDS market. In the time-series, we examine the marginal dealer's decision to start or stop providing liquidity as a function of transaction demand. We find that CDS liquidity normally responds positively to transaction demand, but this response is weakened significantly during a short window prior to major credit events, particularly when the existing number of dealers is large. We interpret this as evidence of dealers turning away from supplying liquidity when they are concerned about other dealers having superior information. Given the endogenous nature of CDS liquidity and especially its close relation with information heterogeneity, we also study the effect of CDS liquidity on CDS pricing. We find that the effect is normally negative, suggesting that better liquidity generally leads to a lower CDS premium. However, this effect could become positive when the existing number of dealers is large, consistent with the latter proxying for the level of information asymmetry in the CDS market. Overall, our findings underscore the important role of information heterogeneity in the determination of CDS market liquidity. In particular, they imply a potential downside to the current regulatory reform calling for increasing transparency in the dealer-dominated CDS market. While this reform may result in greater participation by uninformed investors, it could also reduce the incentive for privately informed dealers to act as liquidity providers, leading to lower CDS market liquidity.