Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk...
Transcript of Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk...
![Page 1: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/1.jpg)
Initial Margin
Roland Lichters
Risk Management for Banks 2017, Köln
7. November 2017
![Page 2: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/2.jpg)
Agenda
About Quaternion
Open Source Risk Project Update
Initial Margin
Optimising Initial Margin
Forecasting Initial Margin, Credit Exposure and MVA
![Page 3: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/3.jpg)
Agenda
About Quaternion
Open Source Risk Project Update
Initial Margin
Optimising Initial Margin
Forecasting Initial Margin, Credit Exposure and MVA
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 3
![Page 4: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/4.jpg)
About Quaternion
Risk analytics solution provider: Risk analysis, management, optimisation
Founded in 2010, global presence in four countries (Ireland, UK, Germany, US)
40 staff, deep risk management expertise in the financial industry
Hybrid approach based on specialist consultancy services and software solutions
Focus on transparency: risk analytics software released in part as open source
Partner based approach to analytics distribution
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 4
![Page 5: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/5.jpg)
Clients
39#
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 5
![Page 6: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/6.jpg)
Agenda
About Quaternion
Open Source Risk Project Update
Initial Margin
Optimising Initial Margin
Forecasting Initial Margin, Credit Exposure and MVA
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 6
![Page 7: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/7.jpg)
Open Source Risk Project
The Open Source Risk Project aims at establishing a transparentpeer-reviewed framework for pricing and risk analysis
for benchmarking, validation, training, teaching
as extensible foundation for contemporary pricing and risk solutions
accessible to the end users with transparent interfaces for trade data, market data,system configuration
documentation: user guide, examples, methodology book
Web site with FAQ and Forum: http://www.opensourcerisk.org
First release October 2016, second release May 2017, third release November 2017
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 7
![Page 8: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/8.jpg)
Open Source Risk Project
The Open Source Risk Project aims at establishing a transparentpeer-reviewed framework for pricing and risk analysis
for benchmarking, validation, training, teaching
as extensible foundation for contemporary pricing and risk solutions
accessible to the end users with transparent interfaces for trade data, market data,system configuration
documentation: user guide, examples, methodology book
Web site with FAQ and Forum: http://www.opensourcerisk.org
First release October 2016, second release May 2017, third release November 2017
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 8
![Page 9: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/9.jpg)
Open Source Risk Project
The Open Source Risk Project aims at establishing a transparentpeer-reviewed framework for pricing and risk analysis
for benchmarking, validation, training, teaching
as extensible foundation for contemporary pricing and risk solutions
accessible to the end users with transparent interfaces for trade data, market data,system configuration
documentation: user guide, examples, methodology book
Web site with FAQ and Forum: http://www.opensourcerisk.org
First release October 2016, second release May 2017, third release November 2017
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 9
![Page 10: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/10.jpg)
Open Source Risk Project
The Open Source Risk Project aims at establishing a transparentpeer-reviewed framework for pricing and risk analysis
for benchmarking, validation, training, teaching
as extensible foundation for contemporary pricing and risk solutions
accessible to the end users with transparent interfaces for trade data, market data,system configuration
documentation: user guide, examples, methodology book
Web site with FAQ and Forum: http://www.opensourcerisk.org
First release October 2016, second release May 2017, third release November 2017
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 10
![Page 11: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/11.jpg)
Open Source Risk Engine
Products Scope (release 2): IR, FX (with Equity and Inflation)
Bonds/LoansInterest Rate SwapsCaps/FloorsEuropean and Bermudan SwaptionsCross Currency SwapsFX Forwards and OptionsEquity Forwards and OptionsCPI and Year-on-Year Inflation Swaps
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 11
![Page 12: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/12.jpg)
Open Source Risk Engine
Analytics Scope (release 2)
Portfolio pricing and cash flow projection
Derivative portfolio analytics based on a Monte Carlo simulation framework
Credit exposure evolution with netting and collateral
Collateral modeling with Dynamic Initial Margin (DIM)
Derivative value adjustments (CVA, DVA, FVA, COLVA, MVA)
Market risk measures, sensitivity analysis, stress testing
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 12
![Page 13: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/13.jpg)
What’s new in Release 3
CMS Swaps with Caps/FloorsAmortization structures (fixed, percentage, annuity)Inflation simulation and XVA, calibration to CPI Caps/FloorsCredit derivatives (CDS)Extended sensitivity framework (adding inflation, equity and credit)Basic parametric Value at Risk (Delta and Delta-Gamma Normal VaR)Additional unit tests and examplesExtended user guide
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 13
![Page 14: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/14.jpg)
Agenda
About Quaternion
Open Source Risk Project Update
Initial Margin
Optimising Initial Margin
Forecasting Initial Margin, Credit Exposure and MVA
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 14
![Page 15: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/15.jpg)
Background
2009: Following the financial crisis, the G20 agreed to initiate a reformprogramme to reduce systemic risk from OTC derivatives
trading of standardised derivatives through exchanges and cleared through centralcounterparties (CCPs)
reporting of OTC contracts to trade repositories
higher capital requirements for non-cleared derivatives
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 15
![Page 16: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/16.jpg)
Background
2009: Following the financial crisis, the G20 agreed to initiate a reformprogramme to reduce systemic risk from OTC derivatives
trading of standardised derivatives through exchanges and cleared through centralcounterparties (CCPs)
reporting of OTC contracts to trade repositories
higher capital requirements for non-cleared derivatives
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 16
![Page 17: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/17.jpg)
Background
2011: The G20 agreed to add margin requirements on non-centrally clearedderivatives
called upon BCBS and IOSCO to develop a global standard
initial proposal in 2012 (BCBS 261)
latest version March 2015
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 17
![Page 18: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/18.jpg)
Background
2011: The G20 agreed to add margin requirements on non-centrally clearedderivatives
called upon BCBS and IOSCO to develop a global standard
initial proposal in 2012 (BCBS 261)
latest version March 2015
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 18
![Page 19: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/19.jpg)
BCBS 261
Key features:
Variation and Initial Margin (VM/IM) for non-cleared derivatives
Margin to cover counterparty risk at a high level of confidence
Collateral to be highly liquid
IM to be exchanged without netting, to be held default-remote, norehypothecation
Because IM will impact market liquidity, it is phased in between 2016 and 2020, andhave a threshold of 50 Mio. EUR
IM model to be calibrated to historical data that contains a period of stress
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 19
![Page 20: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/20.jpg)
BCBS/IOSCO Timeline
Initial Margin is phased in based on group aggregated average notionalexceeding
Sep 2016: 3 trillion EUR (from Feb 2017 in the EU)
Sep 2017: 2.25 trillion EUR
Sep 2018: 1.5 trillion EUR
Sep 2019: 0.75 trillion EUR
Sep 2020: 8 billion EUR
In September 2016, about 20 US and and Japanese banks started posting IM
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 20
![Page 21: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/21.jpg)
Phase 2, 2017
Danske
ING
Nordea
Santander
ANZ (Australia)
SMBC (Japan)
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 21
![Page 22: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/22.jpg)
Phase 3, 2018 (excerpt)
Banca IMIBank of New York MellonCommonwealth Bank AustraliaCommerzbankHSBC Security Services and HSBC LifeLloyds BankHBOSNational Australia BankState StreetUni Credit AGToronto DominionWells FargoWestPac
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 22
![Page 23: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/23.jpg)
BCBS/IOSCO - IM Calculation
Equivalent to VaR at 99% confidence level, 10 business day horizon
Based on historical data with a period of stress
Model to be approved by the regulator
Alternative: Standardised IM schedule, i.e. fixed percentage of notional based onasset class and maturity
Model may account for diversification within asset classes, but not across
risk.net reported in June 2016 that Standardized IM is 12-15 times higher thanISDA SIMM (see next slide) in some scenarios.
Similar estimate by BCBS-IOSCO for US banks, SIMM: 315bn $, Standard: 3600bn $.
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 23
![Page 24: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/24.jpg)
ISDA SIMM
ISDA has published the Standard Initial Margin Model (SIMM)
SIMM defines how to compute sensitivities and vegas
SIMM consists of 3 components, Delta, Vega and Curvature Margin
Delta/Vega Margin are computed analogously to FRTB standard, with prescribedcorrelations between buckets and risk factors
SIMM’s Curvature Margin is “implied” from vegas
Aggregation of SIMM margin types similar to FRTB
Model was approved by US regulator contingent on enhancement
Risk factors added by March 2017: Dividend swaps, Cross-currency basis, Inflationvolatility, Credit base correlation
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 24
![Page 25: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/25.jpg)
Initial Margin Impact, Bank’s Perspective
IM reduces credit risk, but does not eliminate it completely,reduces CVA/DVA and CVA risk charge, computational burden remains
IM increases funding cost: Gives rise to the Margin Value Adjustment (MVA)
Experts expect CVA/DVA/FVA to become very small, KVA and MVA to become thenew drivers of XVA P&L
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 25
![Page 26: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/26.jpg)
Initial Margin Impact, New Processes
ISDA SIMM implementation (→ ORE+)
Daily reconciliation with counterparties
Model validation, benchmarking, backtesting (→ ORE+)
Initial Margin Optimisation (→ ORE+)
Option: Outsource SIMM calculation, reconciliation, backtesting and benchmarking(→ AcadiaSoft Service, ORE+)
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 26
![Page 27: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/27.jpg)
Initial Margin Impact, New Processes
ISDA SIMM implementation (→ ORE+)
Daily reconciliation with counterparties
Model validation, benchmarking, backtesting (→ ORE+)
Initial Margin Optimisation (→ ORE+)
Option: Outsource SIMM calculation, reconciliation, backtesting and benchmarking(→ AcadiaSoft Service, ORE+)
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 27
![Page 28: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/28.jpg)
Initial Margin Impact, New Processes
ISDA SIMM implementation (→ ORE+)
Daily reconciliation with counterparties
Model validation, benchmarking, backtesting (→ ORE+)
Initial Margin Optimisation (→ ORE+)
Option: Outsource SIMM calculation, reconciliation, backtesting and benchmarking(→ AcadiaSoft Service, ORE+)
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 28
![Page 29: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/29.jpg)
Initial Margin Impact, New Processes
ISDA SIMM implementation (→ ORE+)
Daily reconciliation with counterparties
Model validation, benchmarking, backtesting (→ ORE+)
Initial Margin Optimisation (→ ORE+)
Option: Outsource SIMM calculation, reconciliation, backtesting and benchmarking(→ AcadiaSoft Service, ORE+)
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 29
![Page 30: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/30.jpg)
Initial Margin Impact, New Processes
ISDA SIMM implementation (→ ORE+)
Daily reconciliation with counterparties
Model validation, benchmarking, backtesting (→ ORE+)
Initial Margin Optimisation (→ ORE+)
Option: Outsource SIMM calculation, reconciliation, backtesting and benchmarking(→ AcadiaSoft Service, ORE+)
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 30
![Page 31: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/31.jpg)
Initial Margin Impact, New Processes
ISDA SIMM implementation (→ ORE+)
Daily reconciliation with counterparties
Model validation, benchmarking, backtesting (→ ORE+)
Initial Margin Optimisation (→ ORE+)
Option: Outsource SIMM calculation, reconciliation, backtesting and benchmarking(→ AcadiaSoft Service, ORE+)
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 31
![Page 32: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/32.jpg)
Agenda
About Quaternion
Open Source Risk Project Update
Initial Margin
Optimising Initial Margin
Forecasting Initial Margin, Credit Exposure and MVA
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 32
![Page 33: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/33.jpg)
Optimising Initial Margin
To minimise funding cost for IM, a consortium of banks is establishing a SIMMoptimisation process where a central party (AcadiaSoft) is provided with the requiredinformation (sensitivities) by all banks involved.
This central party then performs an analysis across all participants and suggeststransactions between the participants which reduce the amount of posted collateral by theconsortium and the overall funding cost.
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 33
![Page 34: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/34.jpg)
Objective Function
Optimisation over the total initial margin to be posted, objective function
f (x) =∑i ̸=j
[SIMMij(x)− T]+
whereT is a constant threshold given by the regulatorSIMMij is the SIMM as seen by party i due to business with party j
x represents a parameter vector
Task: Vary x and find the minimum of f (x)
Problem: Dimension of x is large, quadratic in the number of entities N, proportional to thesize of the sensitivity vector
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 34
![Page 35: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/35.jpg)
Constraints
Assumptions
1 Net sensitivity position per party facing all others must not change
2 SIMM can increase for some if the overall objective is met, i.e. we do not haveadditional constraints on the individual entity IM amounts
3 Ignore individual risk limits that might inhibit some risk transfers within the consortium
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 35
![Page 36: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/36.jpg)
Example: Three Entities
∆k is a shift in the k’th component of a sensitivity vector
2
∆k
3
∆k
1
∆k
1
−∆k
3
−∆k
2
−∆k
Applying the same sensitivity shift in a loop of entities keeps the net sensitivity positionsunchanged.
This case can be generalised to N entities.
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 36
![Page 37: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/37.jpg)
ExampleThree entities with randomly generated sensitivity vectors that can becompressed to zero.
Solution:Global optimisation with a genetic algorithm (Differential Evolution) → ORE+
��
������
������
��������
������
��������
������
��������
������
��������
������
�� ���� ���� ���� ���� ����
���������
���������
������
�����������
����������������������������������������������������������������������������
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 37
![Page 38: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/38.jpg)
Agenda
About Quaternion
Open Source Risk Project Update
Initial Margin
Optimising Initial Margin
Forecasting Initial Margin, Credit Exposure and MVA
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 38
![Page 39: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/39.jpg)
Initial Margin Impact
To quantify the
desired credit exposure and credit capital reduction
undesired increase in future funding cost (MVA)
one needs to forecast Initial Margin in a Monte Carlo simulation framework
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 39
![Page 40: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/40.jpg)
Credit Exposure
Future dynamic IM(t) needs to be computed on each simulation pathto adequately offset future credit exposure E(t):
E(t) = [ NPV(t) + U(t)− VM(t)− IM(t)]+
whereNPV(t) is the simulated nettings set NPV at time t
U(t) is the value of unpaid cashflows up to time t
VM(t) is the variation margin collateral value covering the portfolio value NPV(t)
→ ORE
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 40
![Page 41: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/41.jpg)
Funding Cost, Margin Value Adjustment
Expected discounted future Initial Margin is required to compute MVA
MVA =
n∑i=1
(fb − sI) δi × SC(ti)× SB(ti)× E [IM(ti)D(ti)]︸ ︷︷ ︸expected discounted IM
withborrowing spread fb and spread sI received on initial marginSB,C(t) cumulative survival probability of the two partiesD(t) stochastic discount factor
→ ORE
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 41
![Page 42: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/42.jpg)
Forecasting Initial Margin
Approaches:
Project today’s SIMM into the future as ’lowest order’ estimate.Overestimates IM, unlikely to pass backtests and regulatory approval
Sensitivities and SIMM under Monte Carlo simulation (→ ORE+).Possible, but challenging
Monte Carlo simulation frameworks in Banks typically do not provide simulatedsensitivities
Calculation of “bump and revalue” sensitivities in a Monte Carlo framework is acomputational challenge
More sophisticated approaches to produce sensitivities efficiently (e.g. Adjoint AlgorithmicDifferentiation, AAD) require a complex revision of the entire Monte Carlo Simulationsystem
Regression approaches leveraging existing Monte Carlo infrastructure (→ ORE)
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 42
![Page 43: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/43.jpg)
Forecasting Initial Margin
Approaches:
Project today’s SIMM into the future as ’lowest order’ estimate.Overestimates IM, unlikely to pass backtests and regulatory approval
Sensitivities and SIMM under Monte Carlo simulation (→ ORE+).Possible, but challenging
Monte Carlo simulation frameworks in Banks typically do not provide simulatedsensitivities
Calculation of “bump and revalue” sensitivities in a Monte Carlo framework is acomputational challenge
More sophisticated approaches to produce sensitivities efficiently (e.g. Adjoint AlgorithmicDifferentiation, AAD) require a complex revision of the entire Monte Carlo Simulationsystem
Regression approaches leveraging existing Monte Carlo infrastructure (→ ORE)
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 43
![Page 44: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/44.jpg)
Forecasting Initial Margin
Approaches:
Project today’s SIMM into the future as ’lowest order’ estimate.Overestimates IM, unlikely to pass backtests and regulatory approval
Sensitivities and SIMM under Monte Carlo simulation (→ ORE+).Possible, but challenging
Monte Carlo simulation frameworks in Banks typically do not provide simulatedsensitivities
Calculation of “bump and revalue” sensitivities in a Monte Carlo framework is acomputational challenge
More sophisticated approaches to produce sensitivities efficiently (e.g. Adjoint AlgorithmicDifferentiation, AAD) require a complex revision of the entire Monte Carlo Simulationsystem
Regression approaches leveraging existing Monte Carlo infrastructure (→ ORE)
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 44
![Page 45: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/45.jpg)
References
Fabrizio Anfuso, Daniel Aziz, Paul Giltinan, Klearchos Loukopoulos: A SoundModelling and Backtesting Framework for Forecasting Initial Margin Requirements,RISK (May 2017)
Peter Caspers, Paul Giltinan, Roland Lichters, Nikolai Novaczyk: Forecasting InitialMargin Requirements - A Model Evaluation; Journal of Risk Management in FinancialInstitutions Volume 10 Number 4 (September 2017)
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 45
![Page 46: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/46.jpg)
Regression: EUR Swap
ATM Vanilla Swap in EUR, 10Y maturity, flat market, regression in 4Y
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 46
![Page 47: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/47.jpg)
Evolution of Expected DIM
ATM Vanilla Swap in EUR, 10Y maturity, flat market
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 47
![Page 48: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/48.jpg)
Thank you
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 48
![Page 49: Roland Lichters Risk Management for Banks 2017, Köln€¦ · Initial Margin Roland Lichters Risk Management for Banks 2017, Köln 7. November 2017](https://reader033.fdocuments.us/reader033/viewer/2022051812/602d62c07fb4c16a665de91e/html5/thumbnails/49.jpg)
Firm locations and details
Quaternion™ Risk Management is based in four locations:
Ireland London USA Germany54 Fitzwilliam Square 29th floor 24th floor Maurenbrecherstrasse 16Dublin 2 Canada Square World Financial Centre 47803 KrefeldIreland Canary Wharf 200 Vesey Street Germany
London E14 5DY NY 10281-1004+353 1 678 7922 +44 2077121645 +1 646 952 8799 +49 2151 9284 800
quaternion.com
Quaternion™ is a sponsor of opensourcerisk.org along with Columbia University and Tullet Prebon Information
DISCLAIMER:
This document is presented on the basis of information that is not publicly available. Quaternion™ is not liable for its contents.The presentation is for the named recipient only and is private, confidential and commercially sensitive.
© 2017 Quaternion™ Risk Management Ltd. Roland Lichters 49