Liquidity and Pricing of Credit Default Swaps in Japan: Evidence from a Benchmark Index for Corporate Debt Claims

Introducing a CCP to a specific asset, however, has the potential to impede nettings across asset classes. Taking this adverse effect into account, Cont and Kokholm (2014) demonstrate that when a CCP operates for interest rate swaps, adding a CCP for CDSs results in reducing total counterparty risks. By contrast, Duffie and Zhu (2011) show that additional effects of a CCP introduced to CDSs are small when a CCP also operates for interest rate swaps. Such a disagreement comes from assumptions on the distribution of exposures across asset classes. Duffie and Zhu (2011) assume independent and identically distributed Gaussian exposures whilst Cont and Kokholm (2014) relax such Gaussian exposures by taking into account differences in riskiness across asset classes. Loon and Zhong (2014) find for single-name CDSs in the U.S. that an introduction of CCP has contributed towards reducing counterpart risks, and that it has weakened the association of CDS premia with dealers’ credit risk.

In general, financial institutions manage counterparty risks in derivatives transactions by providing collateral with each other, hiring specialists who find some contracts that can be broken without affecting the profiles of credit portfolios, or the so-called ‘trade compression’ (Vause 2010), as well as calculating the value of positions with credit valuation adjustments (Lipton and Sepp 2009). Acharya and Bisin (2014) argue that the lack of transparency in the OTC market encourages insurance seller to take too many short-positions, thereby resulting in increases of the overall counterparty risk.

Although the default of CCPs actually occurred only five times in recent years (Tucker 2011), clearing participants still need to manage the riskiness of a CCP itself. Arnsdorf (2012), Galbiati and Soramäki (2012), and Song et al. (2014) address how to measure the CCP risk. BIS and IOSCO (2015) propose indicators that CCPs have to disclose so as to help clearing participants understand the CCP risk.

A physical settlement had previously been a common method in the case of credit events. In this type of settlement, a protection buyer sells referred claims at face value to its counterpart protection seller. If the buyer does not hold referred claims, it must go to the bond market to buy the claims at a distressed price. In addition, there has been an infrequently-used method, or cash settlement. In this settlement, a protection buyer is provided by its counterpart protection seller with money equal to the difference between the actual and face value of referred claims. When an issuer of a debt claim defaults, different investors may price the claim differently by estimating different recovery rates.

The new ISDA rules came into effect from 8 April 2009, affecting more than 2000 dealers (ISDA 2009c).

When an increase of price volatility of a financial asset is regarded as an increase in the amount of new information to be gathered and processed to price the asset (French and Roll 1986), the increased volatility could require dealers to pay more information costs necessary to trade it. When an increase of price volatility of a financial asset is regarded as an increase of uncertainty on the asset value, the increased volatility could discourage dealers to trade the financial asset by making it difficult for them to hedge their positions (Routledge and Zin 2009).

ADF test statistics calculated for e ₁ , e ₂ , and e ₃ based on the Dickey-Fuller regressions including intercepts but not trends are −6.11, −4.72, and −5.93, respectively. The order of lag (s) is one for all e’s. This is chosen by the SBC. Cheung and Lai (1995)’s critical value is used.

If BAS increases by 1 at τ – 1, ECT1 and ECT2 will simultaneously increase by 1 and 0.20, respectively, and ECT3 will contemporaneously decrease by 1.07. The increase of ECT1 by 1 at τ – 1 will decrease BAS by 0.35 (−0.35 = 1 × −0.35) at τ, which will not affect MS, or will change it by 0 (= −0.35 × 0), and will contemporaneously reduce MD by 0.20 (−0.20 = −0.35 × 0.58). The increase of ECT2 by 0.20 at τ – 1 will decrease MS by 0.05 (−0.05 = 0.20 × −0.26) at τ, which will contemporaneously increase BAS by 0.01 (= −0.05 × −0.21). The decrease of ECT3 by 1.07 at τ – 1 will increase MD by 0.58 (= −1.07 × −0.54) at τ, which will contemporaneously increase BAS by 0.03 (=0.58 × 0.05).

VIF = 1/{1 – (correlation coefficient) ²}. A VIF of MS and PD and that of MD and PD are 1.04 and 1.07, respectively, in the post-CCP period.

For example, a correlation coefficient between yields on five-year government bonds and spreads between the yields and overnight money rates is 0.94 in the full sample period. The data used are weekly averages of daily data.

Arora et al. (2012) use a proprietary dataset that contains not only dealers’ quotes of CDS premia but also actually-agreed CDS premia between a large asset management company and protection-selling dealers. They find that the agreed premia carry premia related to credit risks of protection-selling dealers.

The maturity of the zero-coupon interest rates is the same as the Macaulay duration of the NBE.