Reducing CVA exposure using daily coupons
This could be classified as another episode of finance fiction!
This blog starts with a couple of seemingly unrelated paragraphs and then continues in relating them in a way to improve the market infrastructure.
One of the market developments in the last years has been the generalization of the Variation Margin (VM) and Initial Margin (IM). The goal of that framework is to reduce the credit risk in derivatives.
A couple of years ago, the Quant of the Year award was related to a work on the subject of CVA exposure spikes in presence of Variation and Initial Margin. The underlying article, titled Does initial margin eliminate counterparty risk? (subscription required) by Leif Andersen, Michael Pykhtin and Alexander Sokol was published on Risk.Net in May 2017. In some sense, the article indicated that the margin framework, that is working very well in theory, fails in part in practice. The failure is due to the (large) payment of coupons that are taken into account by the VM only on a delayed basis. Following the initial discussion on the subject, I proposed a method to remove those spikes involving one-day forward margin computations and netting between VM and coupons. The method was presented in a blog posted in March 2016 and later published as a working paper available on SSRN.
Another development in the interest rate derivatives market infrastructure is the potential discontinuation of IBORs and the transition to overnight benchmarks as the foundation of interest rate derivatives. The main benchmarks for interest rate derivatives would not be the term IBOR, typically based on three or six-month periods, but overnight rate. Overnight rates are also tenor rates, but with the tenor reduced to the shortest period available in the market: one (business) day.
The most important overnight linked products are Overnight Indexed Swaps (OIS). The typical design of those OISs is based on composition of overnight rates over a certain term, by opposition to daily payments, to avoid too many payments. This feature was introduced for operational reasons, not because this is simpler from a risk management or quantitative finance perspective.
The introduction above can be summarized in the a couple of facts:
Maybe a payment lag of one day between the fixing publication and the payment would be needed to smooth the operational side. A coupon payment for an amount fixed on the same day in the morning may not be the easiest, specially if parties are in different time zones. In a recent blog I mentioned that I don't understand why the ON rates are published T+1. In the main currencies, they have been redesigned recently, are now published by central banks and have moved to T+1 publication. I believe it is fair to blame central banks for this payment lag, for any related convexity adjustment and any T+1 systematic risk in the market.
I wonder if such an elegant solution to credit exposure spikes will ever be implemented by the market. Don't hesitate to leave your opinion or to propose alternative methods to solve the issue.
In a forthcoming document, I will present the technical details of the method, including the convexity adjustment of a potential T+x payment lag and some drawbacks of the approach.
This blog starts with a couple of seemingly unrelated paragraphs and then continues in relating them in a way to improve the market infrastructure.
One of the market developments in the last years has been the generalization of the Variation Margin (VM) and Initial Margin (IM). The goal of that framework is to reduce the credit risk in derivatives.
A couple of years ago, the Quant of the Year award was related to a work on the subject of CVA exposure spikes in presence of Variation and Initial Margin. The underlying article, titled Does initial margin eliminate counterparty risk? (subscription required) by Leif Andersen, Michael Pykhtin and Alexander Sokol was published on Risk.Net in May 2017. In some sense, the article indicated that the margin framework, that is working very well in theory, fails in part in practice. The failure is due to the (large) payment of coupons that are taken into account by the VM only on a delayed basis. Following the initial discussion on the subject, I proposed a method to remove those spikes involving one-day forward margin computations and netting between VM and coupons. The method was presented in a blog posted in March 2016 and later published as a working paper available on SSRN.
Another development in the interest rate derivatives market infrastructure is the potential discontinuation of IBORs and the transition to overnight benchmarks as the foundation of interest rate derivatives. The main benchmarks for interest rate derivatives would not be the term IBOR, typically based on three or six-month periods, but overnight rate. Overnight rates are also tenor rates, but with the tenor reduced to the shortest period available in the market: one (business) day.
The most important overnight linked products are Overnight Indexed Swaps (OIS). The typical design of those OISs is based on composition of overnight rates over a certain term, by opposition to daily payments, to avoid too many payments. This feature was introduced for operational reasons, not because this is simpler from a risk management or quantitative finance perspective.
The introduction above can be summarized in the a couple of facts:
- Daily variation margin is now mandatory for most of market players and daily payments occur in this framework.
- Large coupons payments create a CVA risk even in presence of VM and IM.
- With LIBOR transition, the market is moving to overnight based derivatives.
- Current OIS design is based on periodic payments.
Maybe a payment lag of one day between the fixing publication and the payment would be needed to smooth the operational side. A coupon payment for an amount fixed on the same day in the morning may not be the easiest, specially if parties are in different time zones. In a recent blog I mentioned that I don't understand why the ON rates are published T+1. In the main currencies, they have been redesigned recently, are now published by central banks and have moved to T+1 publication. I believe it is fair to blame central banks for this payment lag, for any related convexity adjustment and any T+1 systematic risk in the market.
I wonder if such an elegant solution to credit exposure spikes will ever be implemented by the market. Don't hesitate to leave your opinion or to propose alternative methods to solve the issue.
In a forthcoming document, I will present the technical details of the method, including the convexity adjustment of a potential T+x payment lag and some drawbacks of the approach.
Comments
Post a Comment