The impact of central clearing on counterparty risk, liquidity, and trading: Evidence from the credit default swap market

Abstract

This paper examines the impact of central clearing on the credit default swap (CDS) market using a sample of voluntarily cleared single-name contracts. Consistent with central clearing reducing counterparty risk, CDS spreads increase around the commencement of central clearing and are lower than settlement spreads published by the central clearinghouse. Furthermore, the relation between CDS spreads and dealer credit risk weakens after central clearing begins, suggesting a lowering of systemic risk. These findings are robust to controls for frictions in both CDS and bond markets. Finally, matched sample analysis reveals that the increased post-trade transparency following central clearing is associated with an improvement in liquidity and trading activity.

Introduction

In the aftermath of the 2008–2009 financial crisis, the US Congress passed the Dodd-Frank Wall Street Reform and Consumer Protection Act, which mandates central counterparty (CCP) clearing for eligible over-the-counter (OTC) derivatives.² Assessing the impact of central clearing on the OTC derivatives market is a pressing concern. First, the sheer size of the OTC derivatives market means that any effect brought on by central clearing is likely to be economically significant. According to the Bank for International Settlements (2012), the market had a combined notional principal of $648 trillion at the end of 2011. Second, the likely effects of CCPs on counterparty risk, liquidity, and trading activity have generated much discussion among academic scholars, policy makers, and market participants. However, to date, a paucity of direct empirical evidence exists on these important issues. Our study fills this gap by examining the North American single-name credit default swap (CDS) contracts that were voluntarily cleared by ICE Clear Credit (ICECC), the first operational CCP for North American CDSs.³ The advent of ICECC created a hybrid structure in the CDS market in the sense that market participants can either voluntarily clear their trades through a CCP or rely on existing bilateral counterparty risk reduction arrangements. We exploit the unique nature of this hybrid environment, as well as several newly available data sets, to test hypotheses concerning the effects of central clearing.

The first hypothesis we examine is whether central clearing reduces counterparty risk, which is the expected loss sustained by the party to a contract when the other party fails to fulfill its contractual obligations. When a CCP clears a derivatives contract, such as a CDS, it replaces the original contract with two new contracts. In one contract, the CCP becomes the seller to the original buyer; in the other, the buyer to the original seller. Through this process of novation, the CCP becomes the counterparty to both the original buyer and seller of the contract and guarantees contractual fulfillment if either party defaults. In addition, through multilateral netting, a CCP reduces clearing members' exposures and losses in the event of a counterparty default.⁴ Through position limits and margin requirements, a CCP can also prevent the excessive build-up of risky exposures by market participants, a cause of counterparty default especially in opaque OTC markets. In the context of the CDS market, this lowering of counterparty risk should raise the value of credit protection, i.e., lead to an increase in the CDS spread.

To test this hypothesis, we conduct an event-study using the commencement date of central clearing as the event date (day 0). Our sample consists of 132 reference entities (obligors) for which voluntary central clearing was initiated between 2009 and 2011. We find that CDS spreads of centrally cleared reference entities increase around the initiation of central clearing. In the 10-day period prior to the commencement of central clearing (day –10 to –1), the mean (median) CDS spread increase is 1.4% (1.6%). For the 11-day period starting with the clearing date (day 0 to 10), the mean (median) CDS spread increase is 1.7% (0.8%).⁵ The plot of cumulative abnormal percentage changes in CDS spreads reveals a pattern of price appreciation that shows no sign of reversing after day 10 (Fig. 1). Thus, our event-study results are consistent with central clearing reducing counterparty risk. To highlight the economic significance of central clearing's impact on CDS pricing, we repeat the analysis using a rate of return (measured in terms of the notional value of the CDS contract). We find a mean cumulative abnormal return (CAR) of 26 basis points (bps) in the 21-day period centered on day 0, which translates into a change in value of $2.6 billion per trillion dollar of notional value.⁶

We explore the robustness of these results along several dimensions. Because both a reduction in counterparty risk and an improvement in CDS liquidity could influence CDS spreads, we specifically control for the change in CDS liquidity in the event-study to separate the relative contribution of these two effects on CDS spreads. While the results indicate that an improvement in CDS liquidity is associated with an increase in CDS spread, we continue to find significant evidence of a CDS spread increase resulting from the reduction in counterparty risk. Finally, our results are also robust to controls for changes in bond liquidity and other frictions related to the CDS-bond basis arbitrage.

The hybrid structure in the CDS market also offers a rare opportunity to test the counterparty risk reduction hypothesis and directly measure the pricing effect of counterparty risk. For the same reference entity, the CDS spreads used by the clearinghouse for daily mark-to-market valuations (settlement spreads) should be higher than CDS spread quotes aggregated from a range of market participants. This pricing gap emerges because settlement spreads do not suffer a discount for counterparty risk so long as ICECC can effectively guarantee the performance of all cleared contracts. Market quotes, meanwhile, reflect pricing by different market participants, many of which might not use a CCP and, thus, still face counterparty risk. Fig. 2 clarifies the conceptual link between the event-study and pricing gap analyses. The CDS spread increase shown in the event-study is represented by the rising solid line around the start of central clearing. The comparison of ICECC settlement spread and CDS spread quoted in the market (the difference between solid line and the dash–dotted line) captures the pricing gap.⁷ We find that ICECC settlement spreads are consistently higher than the spreads quoted in the broader market. In the 11-day period beginning with day 0, settlement spreads are, on average, 0.4% higher than market quotes and the difference is statistically significant at the 1% level. If central clearing is perceived to be an improved method of counterparty risk mitigation, market participants should migrate to ICECC over time and gradually close the pricing gap (convergence of the solid line and the dash–dotted line in Fig. 2). Consistent with this prediction, we find that the average pricing gap decreases from 0.4% to 0.3%, 0.2%, and 0.1% in days [0, 20], [0, 60], [0, 180], respectively. The event-study and pricing gap analysis suggest that the overall pricing effect of counterparty risk ranges from 0.4% to 3.5% [price increase of 1.4% (day –10 to –1)+price increase of 1.7% (day 0 to 10)+pricing gap of 0.4%]. This estimated range is in line with theoretical CDS pricing models of counterparty risk.⁸

Counterparty risk is relevant not only because it affects the valuation of financial contracts, but also because of its implication for systemic risk. One dimension of systemic risk is the concern that a financial institution's default can damage its counterparties and precipitate a cascade of destabilizing failures. A potential benefit of a CCP is its ability to insulate counterparties from each other's default. In the current hybrid environment, the lowering of default correlation is expected to dampen, but not eliminate, the impact of counterparty risk on CDS spreads (e.g., Hull and White, 2001). Because dealers are the most likely counterparties in the CDS market, we expect the initiation of central clearing to weaken the impact of dealers' credit risk on CDS spreads.⁹ To test this prediction, we estimate panel regressions relating cleared obligors' market-adjusted CDS spreads to CDS dealers' credit risk and an interaction between dealers' credit risk and the initiation of central clearing. The coefficient on dealers' credit risk is negative and significant, which is consistent with the idea that counterparty risk is priced in the CDS market before central clearing. Of greater interest to our study is the coefficient on the interaction term, which is positive and statistically significant, but smaller in magnitude than the credit risk coefficient. Across all specifications, the coefficient on dealers' credit risk ranges from –0.19 to –0.27, and the coefficient on the interaction term ranges from 0.07 to 0.09. As an illustration, prior to central clearing, a 60 basis points decrease in dealers' credit risk results in about a 12 basis points increase in cleared obligors' market-adjusted CDS spreads. But, after central clearing, the same decrease in dealers' credit risk results in only about an 8 basis points increase in cleared obligors' market-adjusted CDS spreads. Thus, dealers' credit risk has a smaller impact on the price of credit protection after the start of central clearing, suggesting that the availability of a clearinghouse reduces investors' concerns about systemic risk.

A comprehensive evaluation of the impact of central clearing on the CDS market must also consider a CCP's effects on liquidity and trading because a market's viability depends not only on the mitigation of counterparty risk, but also on the efficiency of the trading process. Furthermore, market liquidity could influence the onset of a systemic crisis if financial institutions hold the same or similar derivatives. A financially distressed institution that conducts a fire sale of its derivatives positions in an illiquid market can cause sharp drops in prices, which in turn can reduce the values of derivatives held by other institutions and potentially threaten their solvency. To the extent that liquidity is priced in fixed income markets such as the CDS market, any changes in CDS liquidity due to central clearing could ultimately affect CDS spreads. Through the public dissemination of daily trading volume, open interest, and settlement prices, ICECC increases the post-trade transparency of centrally cleared obligors. Prior to central clearing, the only publicly available information on CDS trading activity (e.g., open positions) comes from weekly reports of aggregate trading activity published by the Depository Trust and Clearing Corporation (DTCC). Because the market microstructure literature predicts that greater post-trade transparency can improve liquidity and trading through reduced adverse selection, improved dealer risk sharing, and better pricing information for investors, we test the hypothesis that initiation of central clearing increases CDS liquidity and trading.

Using a variety of liquidity measures, we find that the liquidity of cleared reference entities improves after the commencement of central clearing. For example, the mean relative quoted spread falls from 8% in the pre-event period to 6.8% after central clearing begins and the decline is statistically significant at the 1% level. Mean price impact exhibits a percentage drop of 21%, which is statistically significant at the 1% level. Price dispersion, which reflects dealer inventory cost and investor search cost, registers a statistically significant decline from an average of 2.1% to 1.8%. The number of unique dealer quotes, a depth-based measure of liquidity, exhibits a statistically significant improvement in liquidity after central clearing begins. We continue to find an improvement in liquidity when we conduct a difference-in-differences analysis using a matched sample of noncleared reference entities. Extending the difference-in-differences analysis to trading activity measures reveals that trading activity in cleared obligors increases relative to that of matched noncleared obligors after ICECC initiates central clearing. For example, trading in each cleared obligor outstrips trading in its matched sample by an average of $1.2 billion in gross notional value per annum. Overall, our results show that central clearing not only reduces counterparty risk, but also increases CDS liquidity. In turn, this improvement in cleared obligors' liquidity could further interact with CDS spreads. Therefore, by improving liquidity, central clearing also indirectly affects CDS spread, which underscores the importance of including both event dummies and changes in CDS liquidity in our analysis. As shown in the event-study, the event dummies are positive and statistically significant, reflecting the effects of counterparty risk reduction, and the positive (negative) coefficients of CDS liquidity (illiquidity) measures capture the additional effects on CDS spreads due to the improvement in CDS liquidity following the initiation of central clearing.

The rest of the paper proceeds as follows. Section 2 reviews various strands of research related to our study. Relying on insights from the literature survey, Section 3 develops the hypotheses of the paper. Section 4 describes our data sources, defines variables, and provides descriptive statistics. Section 5 examines the effects of central clearing on CDS counterparty risk. Section 6 investigates the effects of central clearing on CDS liquidity and trading. Section 7 concludes the paper.