Optimising the multilateral netting of fungible OTC derivatives

Abstract

Multilateral netting, carried out via a procedure known as ‘compression’, is used to reduce counterparty exposure in over-the-counter derivatives markets. In compression, market participants share trade information via a third-party company, which then proposes a set of trades which will use multilateral netting to reduce counterparty exposures. In this paper, we propose and analyse a set of multilateral netting algorithms based on exposure minimization. As we assume fungibility, these methods are appropriate for derivative markets with wide-scale product standardization. We find that these compression algorithms all perform extremely well across a range of criteria and we discuss their relative attributes. We strongly favour compressions based on the -norm as we find that they eliminate a high fraction of bilateral connections and retain the greatest common divisor of existing positions. We argue that multilateral netting is an effective counterparty risk mitigation technique in OTC derivative markets if done optimally, and the benefits increase with the number of participants.

Keywords:

• Counterparty risk
• OTC derivatives
• Credit risk
• Optimization
• Graph theory

AMS Subject Classifications:

• C44
• C61
• D40
• G01
• G28

Acknowledgements

The author would like to thank Paul Glasserman for discussions and both referees for constructive comments. He would also like to thank Sam Morgan, David Murphy, Matthew Livesey, Christopher Randall, and TriOptima for their comments.

Notes

No potential conflict of interest was reported by the author.

1 See http://www.bis.org/publ/otc_hy0905.pdf.

2 See http://www.icap.com/news/2013/130116-trioptima-eliminates-usd84-trillion-in-otc-derivatives.aspx.

3 Source: Private communication.

4 The CCRO is a US-based independent non-profit corporation which advises the use of best practices across the energy companies and the energy industry.

5 The G14 dealers are Bank of America-Merrill Lynch, Barclays Capital, BNP Paribas, Citi, Credit Suisse, Deutsche Bank, Goldman Sachs, HSBC, JP Morgan, Morgan Stanley, RBS, Societe Generale, UBS and Wells Fargo Bank.

6 The exact quantity that TriOptima minimizes is not public. It could also be the number of trades.

7 We actually used the global minimization algorithm described later.

8 This open source project can be found at http://lpsolve.sourceforge.net.

9 The compression cycle is fragile in the sense that if one party rejects the proposal made by TriOptima, the whole process must fail and be restarted. This event becomes more likely the more counterparties there are involved.

10 We can show this by writing the fixed point condition as . This is satisfied by the global solution, i.e. .

11 This is on an Intel Core i5-2320 CPU running at 3.00GHz with 8GB of RAM.

Additional information

Funding

This research benefited from the support of the French Banking Federation (FBF) Chair on Banking regulation and innovation under the aegis of the Louis Bachelier laboratory in collaboration with the Foundation Institut Europlace de Finance (IEF) and EDHEC.