![Sample our Economics, Finance,Business & Industry journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5931/TOC-Subject-Sample-2021-Economics-Finance-Business-Industry.jpg"})
ABSTRACT
Closeout procedures enable central counterparties (CCPs) to respond to events that challenge the continuity of their normal operations, most frequently triggered by the default of one or more clearing members. The procedures typically entail three main phases: splitting, hedging, and liquidation. Together, these ensure the regularity of the settlement process through the prudent and orderly liquidation of the defaulters’ portfolios. Traditional approaches to CCPs’ margin requirements typically assume a simple closeout profile, not accounting for the ‘real life’ constraints embedded in the management of a default. The paper proposes an approach to assess how distinct closeout strategies may expose a CCP to different sets of risks and costs taking into account real-life frictions. The proposed approach enables the evaluation of a full spectrum of hedging strategies and the assessment of the trade-offs between the risk-reducing benefits of hedging and the transaction costs associated with it. Using an unexplored set of transactional level data, the proposed framework is evaluated assuming the hypothetical default of a real CCP clearing member. We consider the worst-case loss of a large interest rate swap portfolio observed over the past 10 years (i.e. 2005–2015) and show that an efficient hedging strategy which minimises risk may not be optimal when transaction costs are taken into account. The empirical analysis suggests that transaction costs are a significant factor and should be accounted for when designing a hedging strategy. Specifically, it is shown that the risk-reducing benefits arising from more tailored hedging strategies may introduce higher transaction costs, and therefore may change the effectiveness of the strategies.
Acknowledgements
The views expressed in this paper are those of the authors and not necessarily those of the Bank of England, its committees, or those of ICE Clear Europe, BlackRock or the University of Sussex. We would like to thank the two anonymous referees and the associate editor for their helpful comments and constructive suggestions. We also appreciate the feedback provided by Arnaud Faure, Luis Antonio Barron G. Vicente, Sebastiano Daros, Paul Bedford, Bjørn Bakke, Eric Schaanning and colleagues at the Bank of England. We also thank participants at the 7th International Conference of the Financial Engineering and Banking Society (FEBS) at the University of Strathclyde, Glasgow.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
* This work was conducted while Fernando Cerezetti, Ujwal Shreyas, and Anannit Sumawong were working at the Bank of England.
1. European Market Infrastructure Regulation, EMIR, is the common name for Regulation (EU) No 648/2012 of the European Parliament and Council, 4 July 2012, on OTC derivatives, central counterparties and trade repositories. Dodd–Frank Wall Street Reform and Consumer Protection Act, Dodd–Frank, represents the US Public Law no 111-203, from 21 July 2010.
2. Differently from other financial institutions, where one of the main sources of contagion is derived from deleveraging, for CCPs it is their ability to properly function in case risks materialize that avoids propagation of shocks to financial intermediation.
3. As argued by Vicente et al. (Citation2015), these constraints may relate to settlement procedures of each individual contract, trading costs associated with the market architecture in which closeout occurs, and funding needs arising from the liquidation process.
4. Fire-drills are simulation exercises conducted by CCPs to test their own, and also the members’, readiness to handle a default.
5. Closeout procedures are typically comprised of three phases, i.e. splitting, hedging and liquidation. However, CCPs may organise their procedures differently, according to the products they clear. For instance, CCPs that only clear exchange-traded instruments may choose to liquidate contracts without the hedging phase if market conditions are favourable.
6. Due to this relation, the terms transaction costs and market liquidity risk will be used interchangeably when referring to the costs associated with hedging strategies. Please, refer to Section 3 for a detailed discussion on transaction costs.
7. See Section 3 for a detailed discussion.
8. The adjustment is identical to including the roll-yield when trading futures contracts. See Alexander, Prokopczuk, and Sumawong (Citation2013) for more details.
9. Please, access http://www.cmegroup.com/clearing/risk-management/span-overview.html/ for more details.
10. The value at risk approach, VaR, was introduced in the early 1990s by J.P. Morgan with the publication of the RiskMetrics technical document.
11. In particular, different splitting configurations constrain the set of potential trading strategies to be pursued during the hedging phase. Likewise, liquidation performance determines the necessity of further recovery interventions.
12. There are a number of other legal and operational aspects associated with the hedging process that are not discussed in the paper but that have an important role when designing the strategies. Access https://jenner.com/lehman, volume 5, page 1871 onwards, for an example of those regarding the Lehman Brothers Holding Inc. case.
13. For the purposes of the paper, it is assumed that the hedging process can happen in any day of the closeout period and, as such, hedging horizon and closeout horizon may be used interchangeably.
14. For example, when the original portfolio is formed of interest rate swaps, one possible specification for the hedging components would be: hedging assets, defined as all configurations of plain vanilla interest rate swaps expiring in less than 5 years; hedging quantities, defined as the respective notional to be traded in each swap; hedging timings, defined as the 5 days following the assessment date (i.e., the date when the strategy is designed).
15. See Section 4 for one possible definition of liquidity to be used.
16. DV01 stands for the dollar value of one basis point. It represents the price sensitivity of a financial contract, measured in monetary units, with respect to the yield of the instrument.
17. Conscious of the nuances that may differentiate the two concepts, funding need is introduced here to capture the funding liquidity risk.
18. In practice, represents the last time the portfolio was marked to market, which may not always coincide with the moment of the default. However, for the purposes of the paper, these two concepts are being used interchangeably.
19. See page 123 of Vicente et al. (Citation2015) for a formal definition of these metrics. The reader should note that the signs of the formulas were inverted here to represent risk as a positive variable.
20. Even though the market risk and funding need also depend on the original defaulter’s portfolio, to simply notation is omitted from the formulas.
21. Different terminologies for the concept of fundamental price can be found in the literature, such as, equilibrium price, fair price, or expected price.
22. The authors are not suggesting that ‘dislocations from the fundamental values’ and ‘intraday changes in the fundamental value’ represent the same concept. What is being proposed is that in terms of pay-ins and pay-outs, both concepts will impact the total cost of the hedge, and will have to be funded by the owner of the position.
23. Please, see Section 4 for an alternative proposal on how to specify the trade price of a contract to be used in the metric.
24. Reference prices are commonly determined towards the end of the trading session, in a pre-specified time window, seeking to represent to the most extent possible the prevailing market conditions at the time. As such, this process could be seen as an attempt to represent the fundamental value of a contract for a particular point in time.
25. Please, see Cielinska et al. (Citation2017) for one pioneering publicly available study conducted using trade repository data.
26. Data is submitted to a TR in the form of a state report, and participants should inform the entry, modification and the termination of all derivatives contracts (over-the-counter and exchange traded).
27. Please, see https://www.esma.europa.eu/policy-rules/post-trading/trade-reporting for further details on the data fields.
28. For instance, when fixed rates were being considered (i.e., data fields 2.33, and 2.34), different number representations were encountered (e.g., 0.015, 1.5%, 1.5, etc.), and standardisation was needed before the use.
29. For instance, maturity dates (i.e., reported in the YYYY-MM-DD format) had to be converted into days to maturity (e.g., 365 days), such that the market value of swaps could be calculated. Likewise, payment frequencies expressed in an alphanumeric form (e.g., 10D, 3M, 5Y) had to be converted into number of days, accounting respectively for the day-count convention associated with each leg of the swap.
30. Over-the-counter contracts typically entail trading relationships that assume the form of dealer-to-dealer or dealer-to-customer bilateral arrangement. Price formation depends on the network established among participants, with demand and supply induced from interactions with dealers. In addition to information asymmetry and inventory holding premium, search costs for investors and bargaining power for dealers are relevant aspects in the determination of prices. These frictions can make the price formation opaque, creating extra costs for trading. See the classical work of Garman (Citation1976), and Amihud and Mendelson (Citation1980) for more details on inventory premium; and Duffie, Garleanu, and Pedersen (Citation2005), and Feldhütter (Citation2012) on search costs.
31. See Cruz Lopez and Manning (Citation2017), and references therein for an ongoing debate on the topic.
32. The defaulter’s portfolio is modified and scaled to avoid identification of the clearing member.
33. The total number of calculated prices is a function of the number of contracts (i.e., original portfolio and hedging portfolio), the length of the closeout period, the number of days in the lookback period, and the number of re-pricings for DV01 purposes.
34. Although some CCPs allow hedging to be performed with financial instruments outside the range of contracts they clear, as this is not a universal practice, this possibility was excluded from the analysis. As such, it is assumed that if a member defaults with a portfolio of GBP interest rate swaps, only this type of contract will be available for hedging.
35. Although contractual characteristics of each hedging asset are equal to those observed in the market, the value of the swap notional may differ. This is because under the hedging process the amount traded is determined by the optimization algorithm discussed above.
36. The second and third constraints aim at reflecting the realistic condition where the market may not be able to absorb the required hedges in a relatively short period of time.
37. See Manganelli and Engle (Citation2001) for a review of the literature on historical simulation methods used in finance.
38. Ideally the two lookback periods would be aligned, but current data availability, resulting from the reporting structure to the trade repositories, limits the implementation. Therefore, transaction costs were estimated using the more recent data, and adjusted to account for the different levels of the interest rates prevailing in the past.
39. As stated previously, results presented in this and the following sub-sections should be interpreted illustratively, as the original portfolio was altered to avoid identification of the clearing member.
40. This evidence is consistent with findings from Johnson (Citation2008) and Foucault, Pagano, and Roell (Citation2013), which report that greater volumes may not in always be associated with lower spreads.
41. Figure C10 in the Appendix, also shows that the more tailored the hedging process is, the less dispersed is the distribution of profit and losses of the portfolio.
42. For a technical approach on how to assess hedging performance see Alexander, Prokopczuk, and Sumawong (Citation2013), and references therein.
43. Transaction cost figures derived by the paper’s proposed approach were compared with those estimated surveying market participants for the cost of hedging under stress situations. Although distinct, results for both approaches were not far apart.
44. For this type of test to be statistically significant, the comparison should be repeated for several different periods. However, the testing of one single round should show that the approach can be back-tested, similarly to other risk models.
45. Please note that the terms market risk and funding need are being used loosely here, as there is only one state of the world and the calculated figures represent real modifications on the value of the assessed portfolio on those dates.
46. Hedging efficiency is assessed against DV01 (instead of total risk) due to the risk profile of the portfolio. Under the prevailing market conditions at the time, this would imply that the hedging trades impose losses to the portfolio, distorting the purpose of the exercise. See Table for more details.