Credit Default Swaps and Bank Regulatory Capital
Chenyu Shan, Dragon Yongjun Tang, Hong Yan, Xing (Alex) Zhou
Review of Finance, Volume 25, Issue 1, February 2021, Pages 121 – 152, https://doi.org/10.1093/rof/rfaa021
A major development in bank capital regulation (Basel II and III) is the allowance of capital relief using credit derivatives such as credit default swaps (CDS). Banks can buy CDS protection on their outstanding loans to conduct “capital relief”: i.e., reduce the amount of their risk-weighted assets and as a result, hold less capital. This capital relief role, in both theory and practice, can have important impact on bank behavior. However, empirical investigation on it is scarce.
The authors compile a sample of 105 big US banks and use proprietary transaction-level CDS data from the Depository Trust & Clearing Corporation (DTCC) to provide empirical evidence on the capital relief role of CDS. The DTCC data is detailed, containing full information on the CDS reference name, identity of buyer and seller of each CDS transaction, and transaction time. Therefore, the authors can match it with Dealscan loan issuance data to identify a bank’s CDS buy/sell amount on its borrower’s name at any given time. In this regard, the authors make substantial progress in identification compared with extant literature, which either measures banks’ CDS position at the aggregate level, or measures quantity of CDS contracts on a firm’s name without knowing who buys/sells it.
This paper documents that banks’ RWA ratio, i.e., the ratio of risk-weighted assets to the total unweighted assets or a bank, declines by 0.065, or more than 10% after the bank starts to take net long position in borrower CDS. This effect is on top of bank asset size, profit growth, securization activities, derivatives positions, and other factors that may affect a bank’s RWA ratio. The data advantage allows the authors to differentiate the scenario where the CDS buying bank’s credit rating is better than the CDS reference firm’s rating — this is a power test of the capital relief channel, as capital relief is only allowed by banking regulation when the CDS buyer has a better rating.
Another evidence the authors find is that the RWA reduction effect of CDS is only significant for banks with lower Tier 1 capital ratio, which is also consistent with the capital relief channel. In fact, the authors examine “low Tier 1 capital ratio” banks throughout all tests in the paper, and find consistently stronger effect for these banks.
The more important question is, what is the consequence for banks to do capital relief with CDS? Do banks become riskier after they use CDS for capital relief? The answer is yes. This paper finds that banks that buy more CDS on their borrower names see a riskier profile in terms of tail risk, bank Z-score, and non-performing assets measures. They also generate higher profits than banks that did not conduct capital relief. One explanation for this is that capital reliesf-banks extend credit to risky borrowers. The authors do find higher loan spread and a larger fraction of junk-rated borrowers in capital relief-banks’ portfolio. Such capital relief-induced risk-taking got exploded in the 2007-09 crisis. Banks that did capital relief before the crisis see larger stock price decline when the crisis hit, and are more likely to be rescued by the government.
In a nutshell, banks do exploit capital relief opportunities using CDS, the milestone financial innovation in the past decades. This study finds large-sample evidence for the capital relief practice Such capital relief activities have real consequence on bank risk-taking and the whole banking sector: banks that did capital relief using CDS become more risk-taking, and more vulnerable when crisis hit, although they have restrained on their risk-taking during the crisis.