Upload
marius-angara
View
218
Download
0
Embed Size (px)
Citation preview
7/28/2019 Equities as Collateral Study
1/114
March 201
Equities as CollateralIn U.S. Securities Lending Transactions
A Study Implemented byThe RMA Executive Committee on
Securities Lending
7/28/2019 Equities as Collateral Study
2/114
COMMITTEE ONCOMMITTEE
MEMBERS
Chairman
Michael P. McAuleyPremier Global Securities Lending
Patrick Avitabile
Citi
Gene P. Gemelli
Credit Suisse
Sandra L. Linn
Northern Trust Company
Rebecca Nordhaus
Brown Brothers Harriman
Judith Polzer
J.P. Morgan
Jason P. Strofs
Blackrock
Ex-Officio
W. Tredick McIntire
Goldman Sachs Agency Lending
RMA Staff
Christopher Kunkle
Director,
Securities Lending & Market Risk
Fran Garritt
Associate Director
Loretta SpinglerAdministrative Assistant
Kimberly Gordon
Administrative Assistant
SECURITIES LENDING
Attached please find a study with respect to the use of equity securities as collatein agency securities lending transactions. This study was undertaken by the RiskManagement Associations Committee on Securities Lending (the Committee). Thepurpose of the study is to provide statistical and empirical evidence of the importance ofequity collateral as an essential component of a complete and robust risk managementprogram, and to support positive regulatory change that would permit the use of equitycollateral in the US securities lending market. The study is a compilation of the followinthree documents.
Securities Lending: Equity as Collateral: Credit Exposure Analysis Summary by
Alan Laubsch of the RiskMetrics Group.This document compares the credit exposure volatility of securities lending transactionscollateralized with equities to those collateralized with cash; briefly reviews the robustneof major VaR methodologies; backtests daily and weekly credit exposure forecasts; teststhe robustness of equity collateral under stressed conditions; provides a brief analysis ofcash reinvestment risk; and discusses the systemic risk implications of equity as collater
Accepting Equities as Collateral: the European Lenders Experience by Mark C.
Faulkner of Data ExplorersThis document provides empirical data with respect to the practical experiences ofEuropean lenders that had outstanding securities lending and repurchase transactions witLehman Brothers that were collateralized by equities at the time of their default. Itdescribes what actually happened, what the outcomes were and what lessons can belearned.
Equities as Collateral in US Securities Lending Transactions: An Overview of Lega
and Regulatory Considerations by the Legal, Tax and Regulatory Subcommittee of
the Committee.
This document summarizes the legal and regulatory barriers that prevent US marketparticipants from either accepting or pledging equities as collateral.
Our members lend securities as agent for their clients who are generally largeinstitutional investors. Most of these lending agents also provide their clients with aborrower default indemnification pursuant to which they agree to reimburse the client foany shortfall that may occur if collateral is insufficient to purchase replacement securitiein the event that a borrower defaults and fails to return lent securities.
RMA, 1801 MARKET STREET, SUITE 300, PHILADELPHIA, PA 19103 Tel: (215) 446-4003 Fax: (215) 446-4232 E-mail: [email protected]
7/28/2019 Equities as Collateral Study
3/114
Most of these agents lend securities on a global basis and participate in other marketssuch as the European and Canadian markets where equities have long been recognized as anacceptable form of collateral for both securities lending and repurchase transactions. As a result,
many of their clients had outstanding transactions with Lehman Brothers International, Europe(LBIE) when they defaulted in 2008 and had to exercise rights on behalf of these clients toliquidate collateral and purchase replacement securities. It became very apparent to those agentsworking through this process that baskets of equity securities were one of the best and mostliquid forms of collateral. Equity securities performed even better than cash collateral duemainly to the fact that most cash collateral was invested in fixed income securities. Equitysecurities performed extremely well as collateral because they had a single, readily determinedprice at all times; could be sold immediately; were highly correlated to the lent securities; and therequired collateral margins were typically higher than other collateral forms. In addition, inperiods of market stress, baskets of equity collateral that were highly correlated to the lentsecurities significantly reduced the amount of collateral movements needed to meet daily mark-
to-market requirements. These observations by our large US-based members are confirmed bythe similar experiences of European based lenders as documented in the attached Data Explorersstudy of the European experience.
The main regulatory impediment to utilizing equity collateral in the US Market isSecurities and Exchange Commission (SEC) Rule 15c3-3. The requirements of the currentrule effectively prevent US broker dealers from pledging equities as collateral when borrowingsecurities from our members lending clients. The rule deems these lending clients to becustomers of the broker dealer due merely to the fact that they have agreed to lend the brokerdealer a security. This is the case even though the security that is being lent is not in thepossession or control of the broker dealer but rather is in the custody of a third-party custodian.In addition, the lending agent has typically provided a guarantee that the security will be returnedto the lending clients custodial account.
While we view the ability of US broker dealers to pledge equity collateral whenborrowing securities from our members clients as a very important risk management tool, weare not advocating a wholesale change to SEC Rule 15c3-3 that would allow broker dealers topledge equities for all lending transactions. Rather, we urge that changes be considered to thedefinition of customer along the lines of that proposed in a letter to SEC Secretary, JonathanG. Katz, from State Street Bank and Trust Company, dated July 31, 2002 (the State StreetLetter), in response to the SECs request for comments on the subject of whether institutionallenders of securities should be allowed to negotiate collateral agreements other than thoserequired by Rule 15c3-3. A copy of the State Street letter is enclosed and a web link is containedin the following parenthetical (http://www.sec.gov/rules/proposed/s72002/mpmcauley1.htm ). Inaddition, we would suggest that the above-described changes to the rule be applicable only toloans of equity based securities such as equities, ADRs, ETFs and convertible bonds.
In addition to SEC Rule 15c3-3, other regulatory changes would be needed to support theuse of equity collateral in the US market. These include changes to: the US EmployeeRetirement Income Security Act of 1974 (ERISA), as amended, which regulates many pension
http://www.sec.gov/rules/proposed/s72002/mpmcauley1.htmhttp://www.sec.gov/rules/proposed/s72002/mpmcauley1.htm7/28/2019 Equities as Collateral Study
4/114
plans securities lending activities; the U.S. Investment Company Act of 1940 (40 Act), whichregulates registered investment companies securities lending activities; and several U.S. federalstatutes that govern securities lenders rights in the event of the insolvency of a counterpartyincluding a systemically significant counterparty.
In summary, we urge regulators to review the enclosed material and to consider acoordinated approach that would allow institutional participants in the US securities lendingmarket to pledge and accept equities as collateral in transactions involving the lending of equitybased securities. We believe that such a change would benefit all market participants and help toreduce systemic risk in both normal and stressed environments. We also believe that such achange would improve market liquidity, reduce borrowing costs for broker dealers and providebeneficial owners and agent lenders with another important tool to manage the overall risk intheir securities lending programs.
We thank you for your consideration and should you have any questions with respect tothis study or any of the enclosed materials or proposals suggested herein, please call Christopher
Kunkle, Director, at 215-446-4003.
Submitted by:
Christopher R. KunkleChristopher R. KunkleDirector, RMA
Attachments:
Securities Lending: Equity as Collateral
Credit Exposure Analysis
Alan Laubsch, RiskMetricsAugust 9, 2010
Accepting Equities as Collateral
The European Lenders Experience
Mark C Faulkner, Founder & Head of Innovation, Data Explorers
Equities as Collateral In U.S. Securities Lending TransactionsAn Overview of Legal and Regulatory Considerations
Risk Management Association, Legal, Tax, and Regulatory Subcommittee
State Street Letter
7/28/2019 Equities as Collateral Study
5/114
7/28/2019 Equities as Collateral Study
6/114
Securities Lending: Equity AsCollateral
Credit Exposure Analysis Summary
AlanLaubsch
8/9/2010
Thispaperanalyzestheimpactofacceptingdiversifiedequityascollateralforequitylending
transactions.
7/28/2019 Equities as Collateral Study
7/114
TableofContents
Summaryoffindings..................................................................................................................................... 8
Background................................................................................................................................................... 8
RiskEvaluationOverview.......................................................................................................................... 9
CreditExposureAnalysisResults................................................................................................................ 10
AverageCreditExposureVolatility......................................................................................................... 10
MaximumCreditExposureVolatility...................................................................................................... 11
Chartsofdailycreditexposurevolatility................................................................................................ 11
IndustrialCredit
Exposure
Volatility
...................................................................................................
12
ConsumerStaplesCreditExposureVolatility...................................................................................... 12
FinancialsCreditExposureVolatility................................................................................................... 14
ConsumerDiscretionaryCreditExposureVolatility............................................................................ 15
EnergyCreditExposureVolatility........................................................................................................ 16
HealthCareCreditExposureVolatility................................................................................................ 17
InformationTechnologyCreditExposureVolatility............................................................................ 18
MaterialsCreditExposureVolatility................................................................................................... 19
TelecommunicationServicesCreditExposureVolatility.................................................................... 20
UtilitiesCreditExposureVolatility...................................................................................................... 21
CorrelationsDriveHedgeEffectiveness.................................................................................................. 22
ASimpleCreditExposureCalculator................................................................................................... 23
VaRBacktestingResults.............................................................................................................................. 24
AlongtermstudyofequitiesVaRbacktesting................................................................................... 24
Top10dailyDJIAoutliersurprisessince1900.................................................................................... 26
RecentS&P
500
backtesting
results....................................................................................................
26
VaRbacktestingresults........................................................................................................................... 27
S&P500backtest................................................................................................................................ 27
SectorBacktesting:CashvsEquityCollateral......................................................................................... 28
FinancialsBacktest:Relative............................................................................................................... 30
7/28/2019 Equities as Collateral Study
8/114
Energy
Backtest:
Absolute
..................................................................................................................
31
EnergyBacktest:Relative.................................................................................................................... 32
MaterialsBacktest:Absolute.............................................................................................................. 33
MaterialsBacktest:Relative............................................................................................................... 34
StressTesting.............................................................................................................................................. 35
Maximumexposurewithcashascollateral............................................................................................ 35
MaximumCreditExposureWithEquityAsCollateral............................................................................ 35
Bearmarketstresstests.......................................................................................................................... 36
StressTest:largestweeklylossesforS&P500................................................................................... 37
StressTest:largestweeklylossesforfinancials.................................................................................. 38
PredictiveHistoricalStressTests............................................................................................................ 38
Stresstestingsummary........................................................................................................................... 39
Cashreinvestmentrisk................................................................................................................................ 40
Systemicriskimplicationsofequityascollateral....................................................................................... 42
Whatifwedontknowourportfolios?................................................................................................... 43
Integrating
VaR
and
Stress
Tests
.............................................................................................................
44
Summary&Conclusions............................................................................................................................. 45
ReferencesandAdditionalReading............................................................................................................ 46
Appendix:AdditionalCreditExposureBacktestingCharts......................................................................... 47
ConsumerDiscretionary:Absolute..................................................................................................... 47
ConsumerDiscretionary:Relative....................................................................................................... 48
HealthCare:Absolute......................................................................................................................... 49
HealthCare:Relative.......................................................................................................................... 50
Utilities:Absolute................................................................................................................................ 51
Utilities:Relative................................................................................................................................. 52
Telcom:Absolute................................................................................................................................ 53
Telcom:Relative.................................................................................................................................. 54
ConsumerStaples:Absolute............................................................................................................... 55
7/28/2019 Equities as Collateral Study
9/114
ConsumerStaples:Relative................................................................................................................ 56
IT:Absolute......................................................................................................................................... 57
IT:Relative........................................................................................................................................... 58
Industrials:Absolute........................................................................................................................... 59
Industrials:Relative............................................................................................................................. 60
Acknowledgements:
Many individuals contributed to this study. This includes RiskMetrics colleagues Marianela Hoz
de Vila, Chen Huang, Denny Yu, Seth Greenberg, Christopher Finger, Carlo Acerbi and JorgeMina. Thanks to the RMA for helping organize this study, and to all members of the securitieslending committee who participated in numerous group and individual discussions and shared
information and insights. Thanks especially to the active contributions of Rajiv Yadlapalli, JudyPolzer, Andrew Peron, Nick Rudenstine of JPMorgan, Sandra Linn and Kristina Richardson of
Northern Trust, Elizabeth Seidel and Neil Hiralall atBrown Brothers Harriman, Tred Mcintire atGoldman Sachs, Glenn Horner and Michael McAuley of State Street, and Francis Garritt at
RMA.
http://www.bbh.com/http://www.bbh.com/7/28/2019 Equities as Collateral Study
10/114
SummaryoffindingsThe following summarizes the RiskMetrics study on the impact of accepting equity as collateralfor equity lending transactions. Our general conclusion is that diversified equity has potential asan acceptable form of collateral if managed appropriately, which includes actively monitoringportfolio risk characteristics. From the lenders view, equity collateral generally reduces creditexposure volatility compared to cash collateral, and responds robustly under stressed conditionsas correlations increase. In particular, systemic spikes in credit exposures are reduced across allindustry sectors, although there are differences among sectors. Finally, allowing diversifiedequity as a form of collateral also can potentially reduce systemic risks and costs in equitylending.
BackgroundU.S. Rule 15c3-3prevents broker dealers from posting equity as collateral when borrowingequities from agent lenders. Equity is a permissible form of collateral between broker dealers,and in other jurisdictions, including U.K., Europe, Australia, and Canada.
In broker dealer default scenarios it is generally understood that equity markets will suffer,leaving one to think that a trade with cash collateral will always be safer. But this isn't the fullstory.
Equity lenders will need to invest cash collateral in assets that yield more than the rebate rate oftheir loaned positions. This generally leaves them in a short-term fixed income portfolio withsome element of credit risk that is riskier than US government securities (e.g., enhanced yield ormoney market fund-like holdings) in order to make the trade economic. In effect, the cashreinvestment into such enhanced yield funds creates a credit, liquidity, and/or duration mismatch.While this study focuses mainly on the impact of equity as collateral, we will include a basicanalysis of cash re-investment risk.
An apples-to-apples comparison between equity versus cash as collateral would be to estimatethe credit exposure profile for the following:
Fromtheperspectiveoftheagentbankinanindemnifiedtransaction:equityloanversusequity
collateralcompared
to
equity
loan
versus
cash
collateral.
Fromtheperspectiveofthelender:equityloanversusequitycollateralcomparedtoequityloan
versuscash,pluscreditexposuretothechosenreinvestmentvehicle(i.e.,thelenderfacestwo
creditexposures:onetotheagentbank,andthesecondtothechosenreinvestmentvehicle).
We will test the hypothesis of whether accepting soundly structured pools of equity collateralwould generally result in lower credit exposure volatility relative to cash collateral, whileperforming robustly in stressed market conditions. In the limit, if both collateral and lent stocks
7/28/2019 Equities as Collateral Study
11/114
were closely matched (or highly diversified), credit exposure would be minimal and the mainchallenges would be operational and execution risk in a potential default event. In practice,however, we would expect some level of sector or size bias mismatch between lent equities andcollateral. Such a mismatch could result in tail risk if these two portfolios diverged significantly.
Our study therefore simulates a range of realistically mismatched portfolios and compares theperformance of equity and cash as collateral.
RiskEvaluationOverviewIn our risk evaluation approach, we take multiple perspectives on credit exposure arising fromequity lending transactions. A brief outline follows below:
1. CreditExposureVolatilityWecomparecreditexposurevolatilityoftransactionscollateralizedwithcashvsequity,andthen
offerabriefreviewoftherobustnessofmajorVaRmethodologies.Wethenapplytheoriginal
RiskMetrics94methodology(RM1994) tobacktestdailyandweeklycreditexposureforecasts.
2. StresstestingWetesttherobustnessofequityascollateralunderstressedconditions.First,weidentifytheten
largestweeklycreditexposures(Jan32005toJan292010)forequityandcashascollateral. We
thenfocusonlargehistoricaldownmarketstresses,whenbrokerdefaultismostlikely
We use the following parameters throughout the study:
1) Lentequities:wewillsimulateaseriesofconcentratedportfolioswithsectororsizebias(e.g.,
financials,smallcap).Forsectorconcentrations,wewillcreateproxiesusingS&P500sectors,
andforsizebiaswewilluseRussell2000and3000growthandvalueETFs.
2) Equitycollateral:wewillassumethatequitycollateralwillbediversified,proxiedbytheS&P500
index.
3) Liquidationhorizon:Weassumethatafivedayhorizonwillbeadequateforperformingan
orderlyliquidationofliquidsecurities.
4) Creditexposurewillbemodeledwithoutcustomarycollateralhaircuts:100%longsector
portfolioand100%shortlentequities.Thisisdoneforsimplicityandconsistency;and,if
desired,effectsofhaircutscouldeasilybefactoredinafterwards.
Whileourstudyfocusesontheperformanceofequityascollateral,wealsoincludeabriefanalysisof
cashreinvestmentrisk.
Finally,wetakeaviewofsystemicriskimplicationsofequityascollateral.
7/28/2019 Equities as Collateral Study
12/114
CreditExposureAnalysisResultsIn this section we show the effects of equity as collateral on credit exposure volatility calculatedusing RM 1994: Exponentially Weighted Moving Average (EWMA) volatility with .94 decay,using log returns of daily closing price data. The results are broadly supportive of equity ascollateral, showing not only a reduction in average volatility across all sectors, but moreimportantly significantly lower maximum volatility and more stable credit exposure volatility.Equity as collateral performs well under stressed market conditions, benefiting from increasedcorrelations. By way of terminology, we will use credit exposure volatility and VaRinterchangeably.
AverageCreditExposureVolatilityCompared to cash collateral, average credit exposure volatility across all sectors is reduced by41% with equity as collateral. The reduction in volatility ranges from 65% (Industrials) to 13%(Consumer Stables).
Figure 1: Average Credit Exposure Volatility (99% Confidence Level VaR)
.
Avg Weekly 99%
VaR*
Cash
collateral Equity collateral Reduction
Industrials 7.6% 2.7% 65%
Cons Disc 7.9% 3.0% 61%
Inf tech 7.7% 3.2% 58%
Materials 9.6% 4.9% 49%
Financials 12.7% 7.3% 42%
Energy 9.8% 6.3% 35%
Telcom 7.5% 4.9% 35%
Healthcare 5.5% 3.9% 29%
Utilities 6.4% 4.9% 24%
7/28/2019 Equities as Collateral Study
13/114
Cons Stpls 4.5% 4.0% 13%
Table 1: Credit Exposure Volatility, ranked by average volatility reduction
MaximumCreditExposureVolatilityEquity collateral is especially effective at reducing credit exposure volatility when it is at its
highest: 56% on average across all sectors. Furthermore, the standard deviation of VaR is 58%lower, on average, meaning that credit exposure volatility estimates tend to be more stable overtime with equity as collateral.
Figure 2: Maximum weekly 99% confidence level VaR
Max Weekly VaR
(%) Cash collateral Equity collateral ReductionCons Discr 27% 7% 75%
Info Tech 25% 8% 69%
Industrials 24% 8% 68%
Materials 33% 12% 63%
Telecom 28% 12% 58%
Average 29% 13% 56%
Utilities 27% 12% 53%
Energy 40% 19% 52%
Health Care 22% 11% 49%
Cons Stap 20% 11% 43%
Financials 44% 28% 37%
Table 2: Maximum Weekly VaR (Jan 2006 to 2010) for equity and cash collateral.
ChartsofdailycreditexposurevolatilityGraphing historical credit exposure volatility provides useful insights, and confirms the broadeffectiveness of equity collateral.
7/28/2019 Equities as Collateral Study
14/114
Below are charts of each industry sector from 31 May 05 to 29 Jan 10.
IndustrialCreditExposureVolatilityWe will start with a plot of Industrials, which showed the most significant reduction in creditexposure compared to cash collateral: 63% on average, 68% lower peak exposures, 72% lowercredit exposure volatility.
Figure 3: Industrial Credit Exposure Volatility
ConsumerStaplesCreditExposureVolatilityAs a contrast, next we will show the sector with the lowest benefit on average, ConsumerStaples. Due to the relatively low average correlation during normal market conditions, averagereduction in credit exposure volatility is only 13% on average. As correlations increase instressed conditions, however, the hedge becomes more effective. We see a 43% reduction inpeak exposures as well as a 20% reduction in exposure volatility when equity is used ascollateral.
7/28/2019 Equities as Collateral Study
15/114
Figure 4: Consumer Staples Credit Exposure Volatility
7/28/2019 Equities as Collateral Study
16/114
FinancialsCreditExposureVolatilityNext we show financials, which had the lowest reduction in peak credit exposure.
Reduction in credit exposure compared to cash collateral: 43% on average, 37% lower
peak exposure (lowest per industry sector), 38% lower credit exposure volatility
Figure 5: Financials Credit Exposure Volatility
7/28/2019 Equities as Collateral Study
17/114
ConsumerDiscretionaryCreditExposureVolatilitySignificant reduction in credit exposure compared to cash collateral: 62% on average,75% reduction in peak exposures, 75% reduction in exposure volatility
Figure 6: Consumer Discretionary Credit Exposure Volatility
7/28/2019 Equities as Collateral Study
18/114
EnergyCreditExposureVolatilityReduction in credit exposure compared to cash collateral: 37% on average, 52% lower
peak exposures, 54% lower credit exposure volatility
Figure 7: Energy Credit Exposure Volatility
7/28/2019 Equities as Collateral Study
19/114
HealthCareCreditExposureVolatilityReduction in credit exposure compared to cash collateral: 30% on average, 49% lower
peak exposures, 36% lower credit exposure volatility
Figure 8: Energy Credit Exposure Volatility
7/28/2019 Equities as Collateral Study
20/114
InformationTechnologyCreditExposureVolatilityReduction in credit exposure compared to cash collateral: 57% on average, 69% lower
peak exposures, 71% lower credit exposure volatility
Figure 9: Information Technology Credit Exposure Volatility
7/28/2019 Equities as Collateral Study
21/114
MaterialsCreditExposureVolatilityReduction in credit exposure compared to cash collateral: 48% on average, 63% lower
peak exposures, 65% lower credit exposure volatility
Figure 10: Materials Credit Exposure Volatility
7/28/2019 Equities as Collateral Study
22/114
TelecommunicationServicesCreditExposureVolatilityReduction in credit exposure compared to cash collateral: 34% on average, 58% lowerpeak exposures, 58% lower credit exposure volatility
Figure 11: Telecommunication Services Credit Exposure Volatility
7/28/2019 Equities as Collateral Study
23/114
UtilitiesCreditExposureVolatilityReduction in credit exposure compared to cash collateral: 25% on average, 53% lowerpeak exposures, 43% lower credit exposure volatility
Figure 12: Utilities Credit Exposure Volatility
7/28/2019 Equities as Collateral Study
24/114
CorrelationsDriveHedgeEffectivenessThe correlation between lent equities and collateral determines how well the pair performs.Figure 13 shows that correlations between sectors tend to range between 70%-90%, using a 60day lookback period. Table 3 shows that some sectors (Utilities, Telcom, Energy) have thelowest average correlation and the most unstable correlations. In stressed markets, however,correlations tend to converge among all sectors: the difference between the least correlated sector(92%) and the most correlated sector (98%) is minimal. Investors often complain aboutcorrelations going to 1 when markets crash. Equity as collateral actually benefits from this effectas collateral and lent securities are more likely to move in lockstep.
Figure 13: Sector Correlations Vs S&P 500
Correlation Min AverageCorrelation
MaxCorrelation
Std Dev ofCorrelationSummary Correlation
Utilities 21% 70% 92% 14%
Telecom Svc 35% 71% 94% 13%
7/28/2019 Equities as Collateral Study
25/114
Energy 13% 71% 94% 19%
Health Care 47% 79% 96% 11%
Cons Stpl 47% 82% 95% 8%
Materials 53% 83% 96% 10%
Financials 72% 88% 95% 4%
Info Tech 73% 88% 98% 5%
Cons Discr 78% 90% 98% 4%
Industrials 77% 91% 98% 5%
Table 3: Sector vs S&P 500 correlation range
ASimpleCreditExposureCalculatorWe built a simple Excel spreadsheet that allows users to quickly compute credit exposurevolatilities given standard deviations of the long and short portfolios and correlations. Users canalso change time horizon (5 days) and standard deviation multipliers (e.g., 2.61 for 99%Confidence Level assuming a Student t distribution).
Collateral VaR Calculator
Days 5enter
SD multiplier 2.61enter
Equity collateral SD 2%enter
Lent portfolio SD 3%enter
Correlation 80% enter
Cash collateral VaR 17.5%
Equity collateral VaR 10.76%
Volatility reduction 38.5%Figure 14: Simple Credit Exposure Calculator
7/28/2019 Equities as Collateral Study
26/114
VaRBacktestingResultsHow did forecasted volatility compare to realized market movements? The performance of VaRmodels differed significantly depending on methodology used. Historical simulation and otherunresponsive VaR methodologies (e.g., 1 year+ simple moving average volatility) showed poorpredictive results. They adapted too slowly to spikes in volatility; and after volatility had alreadystarted to decline, their forecasts were overly conservative. On the other hand, adaptive models,such as EWMA or the RiskMetrics 2006 generalized ARCH model performed reasonably wellfor equities and other liquid asset classes.
AlongtermstudyofequitiesVaRbacktestingThis section highlights some of the relevant findings from A Historical Perspective On MarketRisks UsingTheDJIAIndexOverOneCentury(2009)byGillesZumbachandChristopher
Finger.Thestudyconfirmedtheoutperformanceofadaptivevolatilityestimatesover
unresponsivemethodologies.ThestudyalsoshowsthattheNormal(Gaussian)distribution
underestimatesdownsideoutliersbeyond95%confidencelevel,andthattheStudentt
distributionwith5degreesoffreedomfitstaillosseswell(aswithotherassetclasses:seeGilles
Zumbach,BacktestingRiskMethodologiesfromOneDaytoOneYear(2007).Figure15below
showsthattheNormaldistribution(dashedblackline)intersectstheStudenttdistribution
(blackline)atthe95%confidencelevel,andisabetterfitbeyondthatconfidencelevel
comparedtotheactualdata.
Figure 15: The probability distribution of DJIA returns rescaled to have unit variance (blue line)
and of the residuals (red line), compared with Gaussian (dashed) and Student t(solid black).
Source: Gilles Zumbach, Christopher Finger: A historical perspective on market risks using theDJIA index over one century Feb 2009
7/28/2019 Equities as Collateral Study
27/114
The figure below shows DJIA returns and annualized volatility from 1900. The rapid phasetransition from low to high volatility environments implies the need for dynamic models toestimate short term credit exposure volatility for equities lending.
Figure 16: DJIA Index level (top) and annualized volatility using RiskMetrics 2006 methodology
(bottom)
7/28/2019 Equities as Collateral Study
28/114
Top10dailyDJIAoutliersurprisessince1900Yet even with the use of dynamic volatility estimates and Student t distribution, we find anumber of unusual outliers, especially on the downside. In Doomed to Repeat it (Nov 2008),Christopher Finger identifies the following top 10 downside surprises for the DJIA since 1990(see Table 4). Interestingly, none of the large absolute loss events during the 2008-2009 GlobalFinancial Crisis (GFC) makes it on the top 10 list (due to the fact that volatility was already atsuch elevated levels during this period).
Return(%)Date Residual Volatility (%) Comment
1 26-Sep-55 -13.30 -6.50 8.10 Eisenhower heart attack
2 19-Oct-87 -12.60 -22.60 32.40 Black Monday
3 29-Jul-27 -10.10 -5.20 8.30 Failure of Geneva conference to naval weapons
4 13-Oct-89 -10.00 -6.90 11.40 Collapse of junk bond market
5 26-Jun-50 -8.10 -4.70 9.30 Start of Korean War
6 27-Feb-07 -7.80 -3.30 6.80 Beginning of subprime, China worries
7 20-Jan-13 -7.00 -4.90 11.40 Tension between Turkey & Europe, naval fight w/Greece
8 30-Jul-14 -6.70 -6.90 16.90 NYSE about to close, WW1
9 28-Jul-14 -6.70 -3.50 8.50 Austrias ultimatum to Serbia, war looming
10 15-Nov-91 -6.60 -3.90 9.60 Program trading losses due to options/futures expiry
Table 4: 10 largest DJIA surprises since 1900
Contrary to popular perception, the most recent top ten surprise did not occur during theGlobal Financial Crisis (GFC), but rather marked its inception on Feb 27 2007. The 3.3% drop inthe Dow Jones was moderate in absolute terms, but highly unusual given the low level ofvolatility at the time. Christopher Finger tagged this outlier as Beginning of subprime and thispoint indeed seemed to mark a phase transition of continually escalating risk culminating in the2008 Lehman default and aftershocks. Large outliers often mark significant events, which may
translate into broader regime shifts.
RecentS&P500backtestingresultsFigure 17 below shows daily 95% VaR backtesting results for the S&P 500 index from Jan 07to April 10. The Feb 27 outlier event is circled on the left. Backtesting results show thedownside skew in markets during the GFC (7.12% downside vs 4.51% upside outliers), butresults are robust overall with residuals generally not much greater than 3 standard deviations(sd). However, the intra-day data, reveals that the May 6 Flash Crash would have represented
7/28/2019 Equities as Collateral Study
29/114
7.78 sd fall, close in magnitude to the Feb 27 8.5 sd outlier. Outliers are important early warningsignals, often marking regime shifts. Risk managers should therefore pay close attention to suchsignals when managing their counterparty credit exposures.
Figure 17: S&P 500 level (purple) overlayed with daily log returns (blue) with 95% VaR bands
(red)
VaRbacktestingresultsThe following charts will illustrate daily and weekly 99% VaR backtesting results around oursample S&P 500 sector portfolios from 3 Jan 06 to 29 Jan 10. Weekly forecasts showed feweroutlier surprises than daily returns, indicating that tail risk generally became more normal atweekly horizons. Furthermore long/short combination of equities generally showed lowerdownside outliers than outright long positions, which further supports the use of equity as robustcollateral.Again, the results below are based on the basic RM 1994 methodology, using log returns andsquare root of time scaling of volatility from 1 day to 1 week. We use RM 1994 as the referencemodel as it is the most transparent and easy to understand, while still providing robust results.
S&P500backtestThe S&P 500 backtest below shows significant daily downside outliers of 2.92%, but weeklydownside outliers are close to expectation at 1.32%.
7/28/2019 Equities as Collateral Study
30/114
Figures x. S&P 500 daily (top) and weekly (bottom) 99% VaR backtest
SectorBacktesting:CashvsEquityCollateralThe sector backtesting section is organized as follows. First, we show daily and weekly creditexposure backtests with cash collateral (i.e., equivalent to short sector & long cash), and thenshow the results substituting equity as collateral (i.e., short sector & long S&P 500). For weeklyforecasts, we simply scale daily volatilities by the square root of 5. For simplicity we assume nomargin haircuts (i.e., initial equity and cash collateral is assumed to be 100%, not 102% orhigher).By way of terminology, we useAbsolute when cash is assumed as collateral, andRelative whenequity is assumed as collateral.
7/28/2019 Equities as Collateral Study
31/114
FinancialsBacktest:AbsoluteWe start with an analysis of financials, which transitioned from being one of the lowest volatilitysectors prior to the Feb 27 07 outlier to the most volatile sector with absolute weekly VaRexceeding 40% for several months following the Lehman meltdown.
Figure 18: Financials (Absolute), daily and weekly backtests
7/28/2019 Equities as Collateral Study
32/114
FinancialsBacktest:RelativeWhen looking at relative financial returns, we can see the same pattern of escalating waves ofrisk commencing with the Feb 27 07 outlier as in the absolute chart. This makes sense, sincefinancials were the primary cause of the systemic risk that slowly infected broader markets.Equity collateral does reduce average volatility by 42%, but peak volatility is only 37% lower.Weekly excessions are the highest of any industry sector at 2.36% when using the Normaldistribution. Substituting the Student t distribution (i.e., with a standard deviation multiplier of2.61 instead of 2.33) reduces weekly excessions to an acceptable 1.32%.
Figure 19: Financials (Relative), daily and weekly backtests
7/28/2019 Equities as Collateral Study
33/114
EnergyBacktest:AbsoluteEnergy was the second most volatile sector, and showed a sudden spike in VaR whichquadrupled from 10% to 40% at the onset of the Lehman crisis. Downside outliers (whichrepresent upside movements in Energy) are moderate for daily and weekly returns.
Figure 20. Energy (Absolute), daily and weekly backtests
7/28/2019 Equities as Collateral Study
34/114
EnergyBacktest:RelativeBy comparison, relative energy volatility increases more gradually when equity is used ascollateral; and, at its weekly peak volatility is 20%, about half of the absolute sector volatility.
Figure 21: Energy (Relative), daily and weekly backtests
Backtesting charts with summary statistics for all other sectors are presented below.
7/28/2019 Equities as Collateral Study
35/114
MaterialsBacktest:Absolute
Figure 22: Materials (Absolute), daily and weekly backtests
7/28/2019 Equities as Collateral Study
36/114
MaterialsBacktest:Relative
Figure 23: Materials (Relative), daily and weekly backtests
For backtesting results for all other sectors, please see Appendix 1.
7/28/2019 Equities as Collateral Study
37/114
StressTestingWe take two perspectives on stress testing. First, we consider the maximum historical creditexposure for each sector with cash vs equity as collateral. Secondly, we focus on credit exposurein historical market downturns.
MaximumexposurewithcashascollateralTable 5 shows that maximum credit exposure with cash as collateral occurred during the 27 Nov08 and 3 Nov 08 rally, and equity as collateral would have resulted in minimal exposure. Whilebroker default is unlikely in rising markets; if such an event did occur, it would result insynchronized losses across all major sectors and therefore result in large aggregate credit losses
throughout the securities lending industry.
Date Absolute CE Relative CE Sector ReturnS&P 500Return
Financials27-Nov-
08 -30.4% -12.2% 30.4% 18.3%
Energy27-Nov-
08 -24.8% -6.6% 24.8% 18.3%
Materials27-Nov-
08 -23.5% -5.3% 23.5% 18.3%
Cons Discr27-Nov-
08 -23.1% -4.9% 23.1% 18.3%
Telecom Svc27-Nov-
08 -20.1% -1.8% 20.1% 18.3%
Industrials27-Nov-
08 -16.0% 2.3% 16.0% 18.3%
Info Tech27-Nov-
08 -14.6% 3.6% 14.6% 18.3%
Utilities27-Nov-
08 -11.3% 6.9% 11.3% 18.3%
Cons Staples 3-Nov-08 -10.7% 3.0% 10.7% 13.6%
Health Care 3-Nov-08 -9.7% 3.9% 9.7% 13.6%
Table 5. Worst Sector Credit Exposures (CE) For Cash As Collateral (Relative)
MaximumCreditExposureWithEquityAsCollateralTable 6 shows the maximum weekly credit exposures between 7 Feb '05 to 29 Jan '10, assumingequity as collateral. These weeks represent the largest one week S&P 500 outperformance foreach sector. While some of the credit exposures are sizeable (assuming no haircut), the resultsare broadly robust. The maximum exposure week for each sector is unique, indicating thatindustry specific factors drove exposures . The single largest credit exposure (financials, -22.8%)arose during a large bull market when broker default is unlikely. Furthermore, for any default in
7/28/2019 Equities as Collateral Study
38/114
a bull market, credit exposure assuming cash collateral would have been larger (-35%). Creditexposures during significant down market are highlighted in red. As expected the two largestdown market credit exposures are to defensive sectors that fell less than the S&P 500 (Utilities &Consumer Staples), and clearly cash collateral outperforms equity in down markets (assuming noreinvestment risk).
Sector Worst weekRelative CE(equity)
Absolute CE(Cash)
S&P 500return Sector Return
Financials 13-Mar-09 -22.3% -35.0% 12.8% 35.0%
Cons Staples 20-Nov-08 -9.8% 0.0% -20.2% -10.4%
Utilities 17-Nov-08 -9.4% -1.1% -8.3% 1.1%
Energy 22-Sep-08 -8.8% -9.8% 1.0% 9.8%Health Care 13-May-09 -7.6% -3.6% -4.0% 3.6%
Materials 12-Dec-08 -7.2% -8.4% 1.2% 8.4%
Telecom Svc 28-Oct-08 -7.2% -5.7% -1.5% 5.7%
Info Tech 12-Dec-08 -7.1% -4.1% -3.0% 4.1%
Cons Discr 26-Nov-08 -6.9% -18.1% 11.2% 18.1%
Industrials 10-Oct-08 -6.9% 0.0% -19.7% -12.9%
Table 6. Worst Sector Credit Exposures (CE) For Equity As Collateral (Relative)
In summary, when cash is used as collateral, credit exposures are driven largely by broad marketrallies. In contrast, equity as collateral results in more diversified and idiosynchratic creditexposures across sectors.
BearmarketstresstestsIn this section we examine how diversified equity performs as collateral during large equityselloffs.We consider the following scenarios:
Top 10 worst drops in the overall equity market over the last 5Y
Top 10 drop weeks in financials over the last 5Y.
Period includes Bear near collapse and Lehman collapse
Predictive historical stress tests (e.g., 87 crash, LTCM, Asian Crisis, 911,DotCom)
AssumptionsWe used daily observations and chose worst weeks based on non-overlapping
weekly log returns
7/28/2019 Equities as Collateral Study
39/114
StressTest:largestweeklylossesforS&P500Equity as collateral performs robustly during the largest weekly losses observed in recent history,aided by increased correlations during stress events. Table 7 highlights credit exposures in excessof 2% in red.The following patterns emerge:
Minimal credit exposure to high beta sectors: Financials, Energy, Materials, Industrials,
Cons Disc
Largest credit exposure to low beta sectors, which tend to fall less than the broad market:
Consumer Staples (9.8% max), Health Care, Telcom, Utilities
Relative returns (long S&P 500 short sector)
Worst weeksAbsoluteS&P loss
ConsDisc Cons Stpls Energy
Financials
HealthCare IndustrialsInf Tech Materials
TelecomSvc Utilities
120-Nov-08 -20.2% 1.1% -9.8% 0.7% 16.2% -3.0% -1.3% -3.0% 7.0% -2.0% -8.5%2 9-Oct-08 -19.8% 1.8% -3.2% 0.9% 14.7% -2.2% -4.8% -4.6% -5.6% 0.1% -1.4%3 27-Oct-08 -14.9% 1.6% -3.9% 4.2% 3.0% -3.9% -0.8% -0.3% 11.0% -4.5% -3.3%412-Nov-08 -11.8% 3.8% -5.9% 0.1% 8.3% -5.4% 0.1% 0.6% 3.9% -4.1% -7.3%5 23-Feb-09 -11.7% -3.2% -7.7% 2.2% 10.9% -5.7% 3.2% 0.3% 2.0% -6.0% -1.0%6 5-Mar-09 -11.3% -3.4% -6.2% -1.7% 20.5% -1.3% 2.3% -7.1% -2.3% -6.0% -0.1%7 20-Jan-09 -9.4% -2.2% -7.1% -1.9% 24.8% -6.9% -2.1% -2.0% -2.9% -4.8% -8.8%
8
3-Oct-08
-9.4%
2.3%
-6.9% 4.0% 0.7% -5.4% 3.2% 2.6%
7.1%
-3.5% -2.9%9 19-Feb-09 -8.2% -1.3% -6.9% -1.3% 14.7% -5.0% 0.0% -0.4% 0.3% -4.0% -2.3%10 29-Sep-08 -7.9% -0.5% -5.5% 5.6% 3.2% -4.0% 1.0% 0.2% 8.0% -2.9% -3.6%
Bear 6-Mar-08 -4.4% 1.0% -2.8% -0.2% 4.8% -0.9% -1.2% -0.5% -1.4% 0.2% -1.6%Table 7. Top 10 worst drops in the overall equity market (2005-2010) plus Bear Stress
7/28/2019 Equities as Collateral Study
40/114
StressTest:largestweeklylossesforfinancialsThe largest weekly drawdowns for financials showed a similar pattern: overall moderate creditexposure, with the same maximum credit exposure of 9.8% to consumer staples. Minimal creditexposure to high beta sectors.Table 8. Top 10 worst drops for financials (2005-2010)
Relative returns (long S&P 500 short sector)
Weeks S&P Loss Cons DiscrConsStpl Energy
HealthCare Industrials
InfoTech Materials Telecom Utilities Financials
20-Nov-08 -20.2% 1.1% -9.8% 0.7% -3.0% -1.3% -3.0% 7.0% -2.0% -8.5% 16.2%
9-Oct-08 -19.8% 1.8% -3.2% 0.9% -2.2% -4.8% -4.6% -5.6% 0.1% -1.4% 14.7%
20-Jan-09 -9.4% -2.2% -7.1% -1.9% -6.9% -2.1% -2.0% -2.9% -4.8% -8.8% 24.8%
5-Mar-09 -11.3% -3.4% -6.2% -1.7% -1.3% 2.3% -7.1% -2.3% -6.0% -0.1% 20.5%
19-Feb-09 -8.2% -1.3% -6.9% -1.3% -5.0% 0.0% -0.4% 0.3% -4.0% -2.3% 14.7%
11-Nov-08 -11.7% 3.4% -6.0% -2.8% -5.5% 0.2% 1.2% 3.5% -2.0% -6.1% 9.7%
20-Feb-09 -8.3% -2.2% -6.6% 1.2% -4.8% 1.8% -0.8% -1.4% -5.9% 0.0% 12.2%
27-Oct-08 -14.9% 1.6% -3.9% 4.2% -3.9% -0.8% -0.3% 11.0% -4.5% -3.3% 3.0%
19-Nov-08 -6.2% 2.9% -4.2% -8.7% -1.9% 0.8% -0.1% 2.6% -0.6% -6.9% 11.7%
4-Feb-09 -5.6% 0.9% -0.6% -1.6% -5.5% 1.8% -3.4% 0.6% -1.6% -4.7% 11.4%
PredictiveHistoricalStressTestsIn addition to recent stresses, we wanted to consider more distant stress scenarios. RiskMetricsresearch identified five worst case daily and weekly scenarios that resulted in the largestportfolio loss for global equity/fixed income investors prior to 2000:
1. Black Monday (1987)
2. Gulf War (1990)
3. Mexican Crisis (1995)
4. Asian Crisis (1997)
5. Russian Devaluation (1998)
We applied a predictive stress testing methodology (Kupiek) to estimate potential creditexposure if we saw a repeat of these scenarios on our current sector exposures, given specificcorrelation assumptions:AllanThe transition to this paragraph might be better if you changed the order:If we saw a repeat of these scenarios on our current exposures (given specific correlationassumptions), we applied a predictive stress testing methodology to estimate potential creditexposure.
7/28/2019 Equities as Collateral Study
41/114
To simulate stressed conditions, we used correlations observed at the height of the Lehmancrisis, Nov 20 2008, with .94 decay and 1 year lookback as this represented an actual brokerdefault eventHistorical stress tests show relatively moderate credit exposures across all sectors, includingsmall caps. Credit exposures in excess of -2% are highlighted in red, smaller exposures inorange.
Long S&P shortsector % return Asian Crisis '97
BlackMonday '87
RussianDevaluation '98
MexicanCrisis '95 Gulf War '90
Cons Discr 5.22 3.46 0.53 -1.92 -0.21
Cons Stap 0.2 -3.5 -3.88 -1.64 -1.82
Energy 1.67 -3.22 0.87 2.75 1.73Financials -5.3 11.77 4.06 5.33 12.37
Health Care 0.25 1.97 -1.26 2.18 -0.71
Industrials -1.15 -3.9 -0.86 1.52 -3.97
Info Tech -0.66 -6.26 -1.05 -3.94 -5.56
Materials 4.74 0.02 2.82 -0.94 -5.18
Telcom -3.4 11.77 3.76 -4.52 1.67
Utilities 5.3 -6.93 -1.15 -5.58 -0.93
Russell 2000 Growth 6.94 -3.61 2.3 1.27 -6.98
Russell 2000 4.42 0.71 2.67 0.85 -4.11
Russell 2000 Value 4.39 -2.14 3.23 0.1 -3.63Russell 3000 Growth 1.41 1.17 -1.96 1.26 -4.33
Russell 3000 1.72 -4.22 0.3 1.48 -1.83
Russell 3000 Value -1.57 -1.32 0.39 1.89 5.35
Table 8: Predictive Stress Tests: .94 decay set on 20 Nov '08 (Lehman crisis)
StresstestingsummaryStress results are broadly supportive of the potential for equity as collateral.1. Equity as collateral performs well for high beta vulnerable sectors (e.g., financials, energy,
materials, industrial) that represent highest systemic risk to market.
2. Correlations between risky assets tend to increase during stressed conditions, making theequity collateral more effective when it is most needed.
7/28/2019 Equities as Collateral Study
42/114
CashreinvestmentriskThe analysis of cash as collateral thus far has assumed that cash would hold 100% of its value. Inreality, most enhanced yield funds lost value, in some cases significantly. In the wake of theLehman default, the Reserve Fund froze redemptions and posted a-3% weekly loss, which thenamplified the run on credit markets. AAA rated subprime backed securities that were assumed tobe cash equivalents proved impossible to liquidate and plunged in value. For example, theSchwab YieldPlus, which lost over 30% over a 1 year period due to concentrated investments inAAA rated subprime bonds. On the other hand, if cash was invested in securities backed by theU.S. government and other highly rated sovereigns, a flight to safety would have resulted in again. Unfortunately, Treasurys were not a common option as lenders sought to maximize yield.
Furthermore, assumptions about the safety of government bonds are again being revisited in thelight of the escalating sovereign bond crisis.From a systemic risk perspective, there are two major issues with cash collateral reinvestment:
1. Cash collateral isprocyclical reinvestment adds liquidity in normal markets, forcedunwind in stressed markets drains liquidity. Simply put, it is a form of bad liquidity.
2. Cash as only collateral option results in a major systemiccrowded trade.From the perspective of the lender, cash reinvestment incurs the following risks:1. CreditRisk
Reinvestmentofcashdoublescreditexposure
Exposureto
credit
downgrade
and
default
risk,
both
corporate
and
sovereign
2. Pricetransparencyandcomplexity Fixedincomeinstrumentsnormallyfoundinthereinvestmentportfolioaremoredifficultto
price
3. Liquidityrisk Creditinstrumentscanbecomehighlyilliquidunderstress,andliquidityisdifficultto
measureandmonitorfortheassetclass(duelargelytolackoftransparency)
4. Durationmismatch Giventhatequitylendingtransactionscanbeunwoundearly,itisnotpossibletomatchthe
maturityanddurationprofileoffixedincomereinvestments.
To quantify the risk of fixed income portfolio, relying on credit spread VaR is not sufficientbecause such an approach only considers historical changes in spreads and not default or ratingsmigrations. A better practice would be to integrate spread VaR with default simulations toproduce credit portfolio VaR and Expected Shortfall (ES) statistics. However, at short horizonsof 1 week, even relatively high confidence credit VaR/ES of 99.5% is likely to be miniscule.Portfolio credit VaR should be viewed at longer horizons, such as one year, to reflect theongoing risk that default lenders face as part of engaging in the business of lending.
7/28/2019 Equities as Collateral Study
43/114
From a short term perspective, it makes most sense to run credit stress tests to quantify potentiallosses in the event of a major default as part of a broader systemic crisis. We have illustratedsuch an approach on a representative yield enhanced portfolio.We have assumed the default of one obligor within the portfolio with minimal recovery rateassumption, combined with a broad increase in credit spreads. The results are largely in line withthe 3% instant drawdown experienced by the Reserve fund, with around 2.2% loss attributed todefault and 65.3bp to spread movements. Two defaults result in total loss levels of 4.5%.Thestress scenario could be elaborated by differentiating between different types of assets, but giventhe relatively short duration of the portfolio, the bulk of losses would still be driven by defaultand recovery rate assumptions.
Figure 38: sample reinvestment portfolio credit stress testIn the future, this study could be expanded to analyze a broader range of standard reinvestmentportfolios.While we have noted that spread VaR is not sufficient to quantify the tail risk for creditinstruments, it is worth noting that spread movements can provide early warning signals ofescalating risk. For a topical application of this analysis to sovereign risk, please see EarlyWarning Case Study: European Divergence.
7/28/2019 Equities as Collateral Study
44/114
SystemicriskimplicationsofequityascollateralAllowing diversified equity as collateral for equity lending transactions should have a net effectof reducing systemic risk for the following three main reasons:
1. Equityascollateralisnotprocyclical,andunlikecashcollateral,thereisnoincentivetounwindtradesforcashliquidityduringcrises.
2. Intheeventofabrokerdefaultandforcedunwindofequitylendingtransactions,diversifiedequityrespondsrobustlyundersystemicstress:
Selling S&P 500 takes liquidity in the most liquid market but adds it back in less
liquid sectors
The net effect would be security and sector divergence and thus lowering
correlation within the equity market as a whole, which reduces a driver of
systemic risk
Price impact on the overall equity market would be negligible as sales are
matched with buys, which would lessen the wild swings seen in the imbalanced
cash collateral unwind process
3. An additional collateral option for investors would reduce the single major crowded trade
effect:The new smaller crowded trade based on sector mismatch could be more readily
monitored in the transparent equity markets, as opposed to the opaque fixed income
markets associated with cash reinvestment
Nonetheless, any crowded trade would result in realized credit exposure growing larger
than expected in the event of a forced unwind. Crowded trades should therefore be
proactively monitored and penalized with extra margin and costs to discourage these
from getting larger
7/28/2019 Equities as Collateral Study
45/114
Whatifwedontknowourportfolios?Some securities lending programs executed by third parties do not provide full transparency onportfolio and collateral composition, and will require the use of generic rules for margining (e.g.,by number of securities, maximum single issuer and industry concentrations, minimum tradingliquidity, etc.). While there are ways to estimate the likely range of credit exposures bysimulating large numbers of randomly generated portfolios, the results will be suboptimal.Lenders will require a more conservative aggregate estimation of margin for the additionaluncertainty, but are still likely to be less protected than if the specific portfolio characteristicswere known. Even with higher average collateral levels, margin is likely going to be too low forthe riskiest combinations where it matters most. Furthermore, given that other equity lending
platforms will have the ability to determine more appropriate pricing and margin levels relativeto risk, platforms without transparency will attract a disproportionate amount of underpriced highrisk transactions and a low proportion of overpriced risk transactions.In summary, while it is possible to run no-transparency platforms on a risk controlled basis, veryconservative levels of margin and pricing are required, and a greater amount of capital must betied up by securities lenders to protect against this uncertainty. Costs for both lender andborrower will be high given the additional uncertainty due to lack of transparency.If systems were upgraded to provide security identifiers, appropriate margin and pricing levelscould be calculated on a real-time basis. Collateral could even be customized to better matchlent securities. In short, risk transparency for such platforms would be a big win, lowering risksand costs for lenders and borrowers.
7/28/2019 Equities as Collateral Study
46/114
IntegratingVaRandStressTestsWe used VaR and Stress Tests to analyze the risk of equity lending transactions. One frequentdiscussion point is how to prioritize and use these statistics, given that both are useful but partialmeasures of risk.
Dynamic VaR backtests robustly for equity as collateral, with less fat-tailed surprises than
outright long equities positions, especially if the Student t distribution is used. VaR provides
useful early warning signals, and responds to escalating short term risk. On the other hand,
VaR only represents realized volatility, not the hidden pressures that often build up and cause
a sudden jump (events). For example, VaR was at very low levels just prior to the escalation
of the subprime crisis, and more recently the sovereign debt crisis.
Stress tests allow flexible analysis of tail risks, but choosing appropriate scenarios and
aggregating the results is a challenge. Historical stress scenarios as we have shown offer
useful perspectives, but hypothetical scenarios should also be considered. For example, many
firms were surprised by the extent of synchronized losses incurred by AAA rated subprime
securities which had no precedent. Reverse Stress Testing is also a useful technique, which
searches for portfolio specific large loss scenarios. This can be particularly useful to perform
across the entire portfolio of counterparty exposures, to identify scenarios that could cause
large aggregate credit losses. One advantage of Reverse Stress Testing is that it requires
continual searching for new scenarios as the portfolio changes, as opposed to re-running ofstandard scenarios which can often build false confidence. But even with the best stress
testing approaches, we are still faced with the challenge of integrating and interpreting an
extensive collection of single point estimates which do not respond to changing market
conditions.
In considering the strengths and weaknesses of both VaR and Stress Testing, its clear that eachapproach adds important dimensions to risk. Stress testing allows risk managers to probe forstructural vulnerabilities, while VaR provides a dynamic pulse on markets and is useful for earlywarning signals. Dynamic VaR is suitable to use as a primary day-to-day risk measure, and isappropriate given the short term nature of credit exposures. But stress tests should be performed
on a regular basis and may at times override VaR based perspectives, for example when VaR isunusually low.Furthermore, VaR analysis could be adjusted to account for a dependency between counterpartydefault and equity moves through the calculation of Conditional VaR statistics. One might, forexample, calculate Conditional VaR based on the assumption of 2 standard deviation downshock or a 10% weekly drop in equities. Christopher Finger describes such an approach in hisToward A Better Estimation of Wrong-Way Credit Exposure.
7/28/2019 Equities as Collateral Study
47/114
Summary&ConclusionsIn analyzing credit exposure arising from equity lending, we have observed significantdifferences by sector and market cap and dramatically time varying volatility. Nonetheless,diversified equity generally performed robustly as a form of collateral for equity lending. Belowwe discuss the most important themes to focus on in implementing collateral managementprograms.Risk is not static and varies greatly depending on market conditions. This calls for the use ofresponsive risk methodologies, such as EWMA or more sophisticated ARCH models.Throughout the crisis, much attention has been paid on the prevalence of fat-tails in financialmarkets. Indeed, the use of fat-tailed distributions such as the Student t can help improve
backtesting results. It is worth emphasizing, however, that the use of dynamic volatility forecastsis much more important than the application of fat tailed distributions.Managing risk requires being proactive at identifying and responding to emerging concentrationsof risks across the financial network. Monitoring concentrations by sector, issuer, market cap,liquidity, and other standard factors is a fundamental discipline. It is also important to vigilantlysearch for hidden concentrations; for example, securities that are exposed to the same underlyingdriving factors or vulnerable to similar scenarios. A leap forward in systemic risk managementwould involve sharing of risk transparency between market participants to better monitor andaddress crowded trades. Even though agent lenders compete, they have much more to gain thanlose by sharing risk transparency. As the financial crisis has illustrated, any risk managementfailure can have dire implications in our interconnected financial network.
This study is not exhaustive, and additional analysis should be conducted as agent lendersimplement equity as a form of collateral. It would make sense to conduct more extensive analysisof concentrated portfolios. Furthermore, advances in liquidity risk management could advancerisk management practices by considering exogenous market liquidity and endogenous portfoliocharacteristics (see Valuing Liquidity: An equity portfolio case study(2010) by Carlo Acerbi &Christopher Finger).Nonetheless, this study demonstrates that equity as collateral has great potential as a sound formof collateral. With security level risk transparency, collateral could even be tailored to moreclosely match lent equities and further lower risk. And most importantly, both costs and systemicrisk could be reduced by the inclusion of an alternative form of collateral that is not procyclicaland responds well under stress.
7/28/2019 Equities as Collateral Study
48/114
ReferencesandAdditionalReadingFinger,Christopher(2000).TowardABetterEstimationofWrongWayCreditExposure.
RiskMetricsGroup.Availableat
http://www.riskmetrics.com/publications/working_papers/wrongway_creditexposure.html
Zumbach,Gilles(2007).BacktestingRiskMethodologiesfromOneDaytoOneYear.RiskMetrics
Group.Availableat
http://www.riskmetrics.com/publications/working_papers/backtesting.html
Finger,Christopher(2008).DoomedtoRepeatit.RiskMetricsGroup.Availableat
http://www.riskmetrics.com/publications/research_monthly/20081100
Zumbach,GillesandFinger,Christopher(2009).AHistoricalPerspectiveOnMarketRisksUsing
TheDJIAIndexOverOneCentury.RiskMetricsGroup.
Acerbi,CarloandFinger,Christopher(2010).ValuingLiquidity:Anequityportfoliocasestudy
(presentation). RiskMetricsGroup.
Acerbi,Carlo(2010).MarktoLiquidity:quantifyingportfolioliquidityrisk.Availableat
http://www.riskmetrics.com/on_the_whiteboard/20100515
http://www.riskmetrics.com/publications/working_papers/wrongway_creditexposure.htmlhttp://www.riskmetrics.com/publications/working_papers/wrongway_creditexposure.htmlhttp://www.riskmetrics.com/publications/working_papers/backtesting.htmlhttp://www.riskmetrics.com/publications/working_papers/backtesting.htmlhttp://www.riskmetrics.com/publications/research_monthly/20081100http://www.riskmetrics.com/publications/research_monthly/20081100http://www.riskmetrics.com/on_the_whiteboard/20100515http://www.riskmetrics.com/on_the_whiteboard/20100515http://www.riskmetrics.com/on_the_whiteboard/20100515http://www.riskmetrics.com/publications/research_monthly/20081100http://www.riskmetrics.com/publications/working_papers/backtesting.htmlhttp://www.riskmetrics.com/publications/working_papers/wrongway_creditexposure.html7/28/2019 Equities as Collateral Study
49/114
Appendix:AdditionalCreditExposureBacktestingChartsThe following pages show credit exposure backtesting results for all industry segments notincluded in the main document.
ConsumerDiscretionary:Absolute
Figure 24: Utilities (Absolute), daily and weekly backtests
7/28/2019 Equities as Collateral Study
50/114
ConsumerDiscretionary:Relative
Figure 25: Consumer Discretionary, relative, daily and weekly backtests
7/28/2019 Equities as Collateral Study
51/114
HealthCare:Absolute
Figure 26: Health Care (Absolute), daily and weekly backtests
7/28/2019 Equities as Collateral Study
52/114
HealthCare:Relative
Figure 27: Health Care (Relative), daily and weekly backtests
7/28/2019 Equities as Collateral Study
53/114
Utilities:Absolute
Figure 28: Utilities (Absolute), daily and weekly backtests
7/28/2019 Equities as Collateral Study
54/114
Utilities:Relative
Figure 29: Utilities (Relative), daily and weekly backtests
7/28/2019 Equities as Collateral Study
55/114
Telcom:Absolute
Figure 30: Telecom (Absolute), daily and weekly backtests
7/28/2019 Equities as Collateral Study
56/114
Telcom:Relative
Figure 31: Telecom (Relative), daily and weekly backtests
7/28/2019 Equities as Collateral Study
57/114
ConsumerStaples:Absolute
Figure 32: Consumer Staples (Absolute), daily and weekly backtests
7/28/2019 Equities as Collateral Study
58/114
ConsumerStaples:Relative
Figure 33: Consumer Staples (Relative), daily and weekly backtests
7/28/2019 Equities as Collateral Study
59/114
IT:Absolute
Figure 34: Information Technology (Absolute), daily and weekly backtests
7/28/2019 Equities as Collateral Study
60/114
IT:Relative
Figure 35: Information Technology (Relative), daily and weekly backtests
7/28/2019 Equities as Collateral Study
61/114
Industrials:Absolute
Figure 36: Industrials (Absolute), daily and weekly backtests
7/28/2019 Equities as Collateral Study
62/114
Industrials:Relative
Figure 37: Industrials (Relative), daily and weekly backtests
7/28/2019 Equities as Collateral Study
63/114
7/28/2019 Equities as Collateral Study
64/114
Accepting Equities as CollateralThe European Lenders Experience
Mark C FaulknerFounder & Head of Innovation, Data Explorers
NewYork75RockefellerPlaza19thFloorNewYorkNY10019+12127102210London2SeethingLaneLondonEC3N4AT+44(0)2072647600
7/28/2019 Equities as Collateral Study
65/114
Executive Summary
The question that this short paper seeks to address is whether the experience of Europeanlenders accepting equities as collateral from Lehman Brothers at the time of their default
should re-enforce the status quo with regards to the current US regulatory environment orchallenge it.
Currently there are restrictions on the acceptability of equities as collateral for many USlending programmes.
The European lending experience, where equities have been acceptable for many years,combined with the default of Lehman Brothers provides us with a useful test caseenvironment in which to explore this important issue. This experience lies at the core of thispaper.
Conclusion & Recommendation
Based upon the positive experience of European lenders accepting equities as collateral fromLehman Brothers at the time of their default,
our conclusion is based on there being potentially significant positive risk-mitigationbenefits for the lenders of securities in accepting equities as collateral.
As with all potential changes in regulation, the details will need to be considered carefully.We would not want to see any further unintended consequences impacting the securitiesfinance industry or the broader capital markets.
Reduced cost and capital can be seen as re-enforcement for the argument in favour ofconsidering regulatory change rather than as the primary rationale.
Executed properly which means with adequate controls and within a robust operationalinfrastructure - the argument for accepting equities as collateral, as part of a portfolioapproach to collateral management, is compelling.
Background
The purpose of this paper is to provide the reader with an insight into the practicalexperiences of several European lenders who were accepting equities as collateral fromLehman Brothers at the time of their default.
Our focus will be upon what actually happened, what the outcomes were and what lessonscan be learned. We will leave it to others to conduct a vital and more scientific analysis.
The default of Lehman Brothers put the global capital markets under sustained andconsiderable pressure. That this pressure was felt in the international securities lendingmarkets comes as no surprise. Global markets are inextricably linked and Lehman Brotherswas a major player in the securities lending market.
The term perfect storm has joined black swan and tipping point in the popularvernacular and we beg your indulgence for using it once again here. We feel that it is anappropriate term to describe the circumstance in which the following occurred: -
Lehman Brothers was a major global player in the securities finance market:
Afirmthatengagedgloballyineverymajormarket,acrossallassetclasses
Afirmwithapropensitytodealinnoncashcollateral whereverpossible
AninnovatorinthefinancingfieldwithextensiveTripartyrelationships
7/28/2019 Equities as Collateral Study
66/114
Afirmwithatrulyinternationalclientbasewithover350lendingrelationships
In this paper we will explore what happened when this perfect storm hit EuropeanLenders taking equities as collateral.
Outline
To set the scene properly for the readers of this paper we will cover the following maintopics:
Anintroductiontothesecuritieslendingprocessandthedifferentcollateralmodels
Themanagementofsecuritieslendingriskandhowtheriskscanbemitigated
AprofileofLehmanBrotherspositioningwithinthesecuritieslendingmarketplace
BackgroundontheEuropeanlendersinterviewed
Ananalysisofwhathappenedoncethedefaultoccurred
Observationsandrecommendationsbasedontheirexperiences
We recognise that many practitioners and regulators (our target audience for this paper) areexperienced in this field and have therefore assumed a fundamental knowledge of thesecurities lending activity. For readersrequiring more background information we suggestAn Introduction to Securities Lending1.
Disclaimer
Although Data Explorers has made every effort to ensure the information and data herein are correct, nevertheless no guarantee is givenas to accuracy or completeness. All opinions, views and estimates expressed herein are those of Data Explorers on the date it was
prepared and are subject to change without notice; however no such opinions, views or estimates constitute legal, investment or otheradvice. You must therefore seek independent legal, investment or other appropriate advice from a suitably qualified and/or authorizedand regulated adviser prior to making any legal, investment or other decision. This material is intended for information purposes onlyand is not intended as an offer or recommendation to buy, sell or otherwise deal in securities.
Copyright 2009 Data Explorers. All rights reserved.
An Introduction to Securities Lending Collateral Models
In this section we will introduce the different securities lending models and demonstrate the
differentrolesofcollateral assecurityandasapotentialsourceofrevenue.Inthecaseofnon
cashcollateralitissolelysecurity,inthecaseofcashcollateralitisboth.Giventhesubjectmatter
inhand wemakenoapologiesforfocussingonthenoncashcollateralmodel.
Asabyproductofbeingappropriatelydocumented forregulatorypurposesandasaresultof
prudentriskmanagement,securitiesloansintodaysmarketsaremadeagainstcollateralinorder
to protect the lender against the possible default of the borrower. There are effectively two
distinctcollateralmodelsthatneedtobeconsidered.Collateralcaneitherbecash,ornoncash.
NonCashSecuritiesLendingTransaction:
1www.dataexplorers.com/consultancy
7/28/2019 Equities as Collateral Study
67/114
Loan
Loan Terminates
BorrowerLenderCollateral
Loan Commences
Lender Borrower
Loan
Collateral
Loan
Loan Terminates
BorrowerLenderCollateral
Loan
Loan Terminates
BorrowerLenderCollateral
BorrowerLenderCollateral
Loan Commences
Lender Borrower
Loan
Collateral
Loan Commences
Lender Borrower
Loan
Collateral
Lender Borrower
Loan
Collateral
Inmanycases theownerof theassetsbeing lentwillappointanagent tomanage their lending
activity.Thisagentcanbeacustodian,a3rdpartylenderoranassetmanagementlendingagent.
Daily marking to market will result in collateral flows in either direction during the term ofthe loan.
The eligible collateral will be agreed between the parties, as will other key factorsincluding:
Notionallimits:
The
absolute
value
of
any
asset
to
be
accepted
as
collateral
Initialmargin:Themarginrequiredattheoutsetofatransaction
Maintenancemargin:Theminimummarginleveltobemaintainedthroughoutthetransaction
o Concentrationlimits:
Themaximumpercentageofany issuetobeacceptable,e.g. lessthan5%ofdaily
tradedvolume
Themaximum percentage of collateral pool that can be taken against the same
issuer,i.e.thecumulativeeffectwherecollateralintheformoflettersofcredit,CD,
equity,bondandconvertiblemaybeissuedbythesamefirm
Inalargenumberofsecuritieslendingtransactions,collateralisheldbyaTriPartyAgent.
7/28/2019 Equities as Collateral Study
68/114
ASecuritiesLendingTransactionInvolvingaTriPartyAgent
Lender Borrower
Tri PartyAgent
Loan
Collateral
ReportingReporting
Loan Commences
Tri PartyAgent
BorrowerLenderLoan
Collateral
Loan Terminates
Lender Borrower
Tri PartyAgent
Loan
Collateral
ReportingReporting
Loan Commences
Lender Borrower
Tri PartyAgent
Loan
Collateral
ReportingReporting
Loan Commences
Tri PartyAgent
BorrowerLenderLoan
Collateral
Loan Terminates
Tri PartyAgent
BorrowerLenderLoan
Collateral
Loan Terminates Thi
sspecialistagent (typicallya largecustodianbankor InternationalCentralSecuritiesDepository)
willreceiveonlyeligiblecollateral fromtheborrowerandhold it inasegregatedaccounttothe
order of the lender. The Tri Party Agent will mark this collateral to market, with information
distributedto
both
lender
and
borrower.
Typically
the
borrower
pays
afee
to
the
Tri
Party
agent.
There is debate within the industry as to whether lenders that are flexible in the range ofnon-cash collateral they are willing to receive are rewarded with correspondingly higherfees. Some argue that they are, others claim that the fees remain largely static but thatborrowers are more prepared to deal with a flexible lender and therefore balances andoverall revenue rise.
Transactions collateralised with cash
Cash collateral is, and has been for many years, an integral part of the securities lendingbusiness, particularly in the United States. The lines between two distinct activities -securities lending and cash reinvestment - have become blurred; and to many USinvestment institutions securities lending is virtually synonymous with cash reinvestment.This is much less the case outside the United States but consolidation of the custodybusiness and the important role of US custodian banks in the market means that thispractice is becoming more prevalent.
Lender Borrower
Money
Markets
Loan
Collateral
Cash
Cash
Loan Commences
Lender Borrower
Money
Markets
Loan
Collateral
Cash
Cash
Loan Terminates
Lender Borrower
Money
Markets
Loan
Collateral
Cash
Cash
Loan Commences
Lender Borrower
Money
Markets
Loan
Collateral
Cash
Cash
Loan Commences
Lender Borrower
Money
Markets
Loan
Collateral
Cash
Cash
Loan Terminates
Lender Borrower
Money
Markets
Loan
Collateral
Cash
Cash
Loan Terminates The importance of this point lies in the very different risk profiles of these increasinglyinterrelated activities. Crucially, cash reinvestment is not typically covered by an indemnity(of which more later) which can be argued to create a conflict of interest for the cashmanager they earn fees but do not share the direct financial risk of this activity. They do,
7/28/2019 Equities as Collateral Study
69/114
however, run considerable reputational and commercial risk if they do not manage thispotential conflict.
Cash reinvestment was traditionally dominated by unitized funds which pooled thecollateral for ease of management and to achieve the various economies of scale availablein money
market investment. However, segregated accounts with client specific risk profiles are nowbecoming much more common.
Note that the securities lending loan term (i.e. time to maturity) will determine thebenchmark rate that is to be paid on the cash. Most loans can be recalled at any time, so thecash will generally have an overnight rate benchmark. It is common for the interest to bephysically settled monthly.
Below we provide an example of the relative importance of cash and non-cash to differentfiscal locations and as you can see the US domiciled lenders are overwhelmingly takingcash as collateral whilst other jurisdictions have a predilection for non-cash collateral.
The relative scale and importance of the US lending community brings the overall
percentage of collateral taken as cash up to 57%. However, the contrast between the UStrends and the UK, Canada and the Netherlands in particular is dramatic.
These statistics owe a great deal to historic tax legislation and inertia but have served manyof the non US lenders well in recent turbulent times.
The revenue generated from cash-collateralised securities lending transactions is derived ina different manner from that in a non-cash transaction. It is made from the difference orspread between interest rates that are paid and received by the lender.
Some securities lending agents offer bespoke reinvestment guidelines whilst others offerreinvestment pools.
Summary
There are different securities lending models globally. These models have developed for awide variety of reasons, including: regulation; client preference; and tax
The US model is dominated by cash collateral. The European lenders have a higherpropensity to accept non-cash collateral. Tri - party providers have established an importantcapability to facilitate collateralisation. The acceptance of cash collateral offers potentialrevenue opportunities. Different securities lending models have different risk profiles.
7/28/2019 Equities as Collateral Study
70/114
In the next section we will explore the risks inherent in securities lending and how theymay be measured, managed and mitigated.
7/28/2019 Equities as Collateral Study
71/114
The Management of Securities Lending Risk
Inthissectionwewillfocusupontheriskmanagementissuesthatareandshouldbeontheminds
of all parties with fiduciary or risk management oversight responsibility for securities lending
programmes.A
more
detailed
description
of
risk
issues
emanating
from
securities
lending
is
containedinourBestPracticepaper2.
Properlyconfigured,securities lendingshouldbea lowrisk, lowreturn,highriskadjustedreturn
activity. This is very often the case. However there are risks which must be understood and
managed. Both market risks and operational/legal risks can result in financial losses for the
lenders,but the financial risksaremore readilyquantifiable than those in thenonfinancial risk
category.Alloftherisksidentifiedcanbemeasuredtosomedegreeandmanagedandmitigated
asubjectthatweexploreindetailinthissection.
The aftermath of the Lehman Brothers default showed that financial risks can be more
difficult to measure than previously thought; and that non-financial risks can play asignificant part in the total risk taken.
Non Financial Risks
These include:
Legal Risk covers both contractual risk and enforceability risk. Contractual risk refers to
ensuring the terms and conditions for securities lending between two institutions are
comprehensive and appropriate. One recent important example: problems following the
Lehman default in the legal setoff calculation failing tomatch the reality of the sales and
purchases required to close positions. Enforceability risk refers to the enforceability of the
contractunderprescribednational laws. In theeventofadefault situation, the rightof the
nondefaultingpartytonetthecollateralvalueandtheunderlying loanvalue isarguablythe
mostimportant
example
of
this.
OperationalRisk the risk that the securities lendingagentdoesnothaveadequatecontrols
and infrastructure inplacetomanageasecurities lendingprogram.Itshouldbenotedthat if
the lender does not or cannot engage in the process of collateral liquidation and loan
repurchase,theyendupwithadefactochange inassetallocation;thismaybedramaticand
unexpectedtosomebeneficialowners(theownersofthesecuritiesbeinglent).
RecallRisktheriskthattheborrowerdoesnotreturnrecalledsecurities intimetoenablea
saleorcorporateactiontobemet.
Reputational Risk the risk that by virtue of being engaged in a securities lending
program the reputation and standing of the lender is somehow damaged. This may resultfrom an operation failure or more likely in recent times, the headline risk of being inthe paper on the wrong side of a story.
2www.dataexplorers.com/bestpractices
http://www.dataexplorers.com/bestpracticeshttp://www.dataexplorers.com/bestpractices7/28/2019 Equities as Collateral Study
72/114
Tax risks for example, crystallized profits on disposals of assets following a defaultmay attract Capital Gains Tax.
Non financial risks can be mitigated by appropriate legal agreements industry standard
documentationgives
more
security
that
they
are
enforceable,
particularly
when
reinforced
by
external legalopinion.Netting isan importantelementofriskmanagement;marketparticipants
willoftenhavemanyoutstandingtradeswithacounterpart.Ifthereisadefault,standardindustry
master agreements ensure that, postdefault, various payments are accelerated, i.e. payments
becomedueatcurrentmarketvalues.
Financial Risks and Mitigants
Thiscoversanumberofareasandisbestsummarisedinatableshowingpossiblemitigants.Given
thetopicofthispaperwehavegoneintomoredetailonthoserisksmorelikelytooccurinanon
cashenvironment:
FinancialRisk
Mitigant
Default or Credit risk the risk that a counterpart or issuer
cannotmeet itsobligations. Inacollateralised loan,adefault
triggers the process of collateral liquidation and loan
repurchase,exposing theprogram tomarket risk; inoutright
reinvestment the loss is immediate and prospects of asset
recoveryareusuallypoor.
CounterpartyorIssuerSelection/ApprovalbyLender an
area of increasing focus and onewhich needs to be kept
under constant review. Agents who can demonstrate a
dynamic approach and remove most of the burden of
counterparty vetting are finding that their client
relationshipsarestrengthenedthisisparticularlyrelevant
wheretheagentmaybeunabletoofferastrongindemnity.
MarketRisk (Mismatch risk,occurringonly in theeventof a
default)
the risk that themarketpriceof theunderlying securityor
collateralmovesadversely inashortperiodof time. Thiscan
arisebecause
of
changes
in
yield
curves
(i.e.
interest
rates),
currencies,creditspreads,orequitymarkets.
the risk that t