Managing Counterparty Credit Risk:

Implementing Real Time CVA Pricing

Dr. Neil Dodgson
VP Business Development Sell Side
February 2012

2012 IBM Corporation

Banking reforms
Risk Management

Finance
IRC

CVA Charge

Stressed VaR

Capital Ratio

RAPM

EPE
VaR

WWR CVA RWA

Credit Exposure
Market Exposure

Basel
III

Fair Valuation
Planning

NSFR
LCR

Collateral
FVA Charge

CVA Price
Front Office

Sensitivities

ALM
Funding
Liquidity

2012 IBM Corporation

Treasury

How is Basel III affecting counterparty credit risk?

Capital requirement for counterparty credit risk using stressed EPE
CVA calculations
Margin period of risk for collateral management
Central counterparties
Wrong-way risk

2012 IBM Corporation

CCR/CVA timeline
In only a few short years, we have seen a shift from passive to more
active and continuous management of CCR requiring CVA:
Before CVA
Firms apply credit limits and
measures such as to limit their
possible exposure to a
counterparty in the future

1998: Asian crisis and LongTerm Capital Management

(LTCM). The unexpected
failure of the large hedge
fund LTCM and Asian crisis
lead to an interest in CCR
from some first tier banks

Passive Management of CVA

Active Management of CVA

Large banks first start using CVA to

assess the cost of counterparty risk
CVA is treated via a passive
insurance-style approach

The credit crisis and resulting failures of

high profile firms generates much more
attention on counterparty risk
Banks are interested in more accurate and
evermore frequent CVA calculations daily,
intra-day, and real-time

2006: New Accountancy

regulations (FASB 157, IAS 39)
mean that the value of
derivatives positions must be
corrected for counterparty risk
All banks must start calculating
CVA on a monthly basis

2012 IBM Corporation

Sept. 10-15, 2008: Lehman

Brothers collapses following a
reported $4 billion loss and
unsuccessful negotiation to find a
buyer, one of Wall Streets most
prestigious firms files for
bankruptcy protection

Risk is changing
Before the credit crisis
Most counterparty risk situations were rather unilateral
The too big too fail concept obscured counterparty risk
Many institutions see their counterparty as being risk-free (at least from their point
of view)
Credit spreads of banks just a few bps
Collateral agreements often one-sided or heavily skewed (independent amounts
etc.)

Counterparty risk was the focus of mainly large global banks (1st tier)
Wrong-way risk was a concept rather than a reality
No-one had ever heard of DVA

2012 IBM Corporation

Counterparty credit risk - a front office issue

Capital Markets
Solution for active management of counterparty credit risk

Credit exposure calculation

Simulated PFE exposures
Notional and add-on measures
Stress testing
What if analysis

FRONT OFFICE /CVA DESK

TRADING

Limit management
Pre deal limit checking
Trade restrictions &
limits
Intra day excess

RISK MANAGEMENT

2012 IBM Corporation

CVA calculation
Unilateral and bilateral
Pre-deal incremental CVA
CVA sensitivities

FINANCE

Counterparty credit risk

Users
Front office

Function
Trading and sales charge clients and
customers for CVA.
May influence trade or counterparty

Requirements
Calculate pre deal exposure and
incremental CVA as amount for deal. Want
portfolio effect (deal and at netting level).

Trade within limit structures

Central CVA Desk

Risk Management

Finance

Charge trading desks

CVA per counterparty for initial charge

Compute internal insurance cost

CVA per deal if advising

Optimizer risk/return via hedging

CVA sensitivities

Monitor and interact with ON CCR and CVA

Use baseline for risk monitoring

Trader compensation includes CVA charger

Use intraday updates for interacting

Monitor risk intraday, on demand risk, whatifs

Run analysis with new trades and new

scenarios

Comply with regulation (e.g., FASB 157)

End of quarter CVA, or CVA -DVA

Calculate fair value of derivative portfolio

2012 IBM Corporation

Why is CVA so complex?

Calculating the CVA of a derivative is always more complex than pricing the
derivative itself

e.g., CVA of a swap involves volatility, but pricing the swap itself does not

Must account for

Complexities of the trade (cash flows, exercises, resets, ) and market variables

Correlations between market variables

Default probability and recovery value (often more art than science)

Netting (causes exposure to be reduced)

Collateral agreements (as above)

Wrong-way risk (credit derivatives in particular)

2012 IBM Corporation

Current practice: transition to active management

Of the firms surveyed, 50% calculate CVA monthly, 25% daily, and 25%
in real time

Source: Credit Value Adjustment: and the changing environment for pricing and managing counterparty risk, Algorithmics, December 2009

2012 IBM Corporation

Current practice: incremental CVA at deal time

Pre credit crisis: firms that charged CVA were often at a pricing
disadvantage relative to firms that did not
Post credit crisis: firms that charge CVA on an incremental basis are a
competitive advantage vs. firms that cannot

Source: Credit Value Adjustment: and the changing environment for pricing and managing
counterparty risk, Algorithmics, December 2009

10

2012 IBM Corporation

Survey : CVA purpose and management of CVA

The main purpose of CVA today is to facilitate accounting reporting, followed by front
office pricing:

Source: Credit Value Adjustment: and the changing environment for pricing and managing counterparty risk,
Algorithmics, December 2009

11

In the front office, CVA is owned by either a single front office unit (58%), in multiple
groups (25%), or in a single risk group (17%).

50% calculate CVA monthly, 25% daily, and 25% in real time

2012 IBM Corporation

Pre-deal CVA: why use simulation approach?

Simulation approaches enable risk reducing trades to be priced more
competitively than risk increasing trades
End-of-Day
measure

Option 1

Pre-deal analysis results

Option 2

Option 3

Counterparty
Portfolio

Add-on measure

Monte Carlo
Standalone

Monte Carlo
Incremental

95% Peak Exposure

$50M

$2.5M

$1M

($2.5M)

Unilateral CVA

$100K

$5K

$2K

($5K)

Highly conservative
additive estimate of
exposure and CVA

Less conservative
simulated approach to
measure exposure and
CVA

Most accurate measure

capturing diversification,
netting and collateral

If we are able to measure our counterparty risk more accurately, we will be able to use our credit lines and capital more
efficiently. This will allow us to do more business with the same or lower limits, as our current conservative methodologies
constrain the business and may overstate exposure. And because we can understand the CVA in advance of doing a trade,
it lets us be sharper in our prices.1
Source: 1)Risk Magazine, Exposing counterparty risk exposure, March 2010, http://www.risk.net/risk-magazine/feature/1594855/exposing-exposure

12

2012 IBM Corporation

Wrong-way risk
It is typical to assume independence between

Default probability of counterparty

Exposure at default

But, in reality, this is often wrong

Buying out of the money put options

Buying CDS protection

FX products with local currencies

Wrong way risk challenges

13

Correlation and dependency are not the same thing

Wrong-way risk might be quite subtle / indirect

Wrong-way risk can be massive (mono-lines)

2012 IBM Corporation

A realistic example: why wrong-way risk matters (example 1)

Corporate Portfolio
150 Swaps: one-directional (long fixed/short floating), all
denominated in CAD
Maturities: min = 2wks, max = 10yrs
Market factors: short-rate calibrated to swaption vols
Credit modeling: no netting, no collateral
Simulation time steps: quarterly to 10yrs (total of 40)

14

2012 IBM Corporation

A realistic example: why wrong way risk matters (example 2)

Corporate Portfolio
CVA goes from $1.7M - $5M, depending on right or wrong way risk
6,000

CVA depending on correlation

Wrong way

5,000

4,000

Right way

3,000

2,000

1,000

0
-1

-0.75

-0.5

-0.25

-1,000

15

2012 IBM Corporation

0.25

0.5

0.75

Portfolio credit risk framework

16

2012 IBM Corporation

Wrong-way risk

Joint simulation of market and credit risk factors

Market, systemic & specific risk factors

Utilises existing economic capital models

Migrations and defaults modelled

Reduced number of micro/macro factors

Conditional scenarios

Enhanced performance

17

2012 IBM Corporation

Market
Factors

Systemic Specific
Credit
Credit

Credit exposure modeling: Monte Carlo simulation approach

18

MARKET RISK FACTOR SCENARIO GENERATION

POSITION VALUATION

AGGREGATION & NETTING

COLLATERAL ADJUSTMENT

OBTAIN EXPOSURE PROFILE FOR EACH SCENARIO

OBTAIN EXPOSURE METRICS

2012 IBM Corporation

CVA: Natural extension of credit exposure modeling

MARKET RISK FACTOR SCENARIO GENERATION

CVA Use Risk Neutral Scenarios

POSITION VALUATION

AGGREGATION & NETTING

COLLATERAL ADJUSTMENT

OBTAIN EXPOSURE PROFILE FOR EACH SCENARIO

OBTAIN EXPOSURE METRICS

6
19

CVA - Calculate Discounted Expected Exposure

2012 IBM Corporation

Pre-trade/deal-time pricing of CVA

New trade info

F/O
Product
System

Update portfolio
Inception pricing

CVA Engine

CVA overrides

Nonstandard
queue

Book
trade

View/adjust model inputs

What-if trade

What-if CVA
Diagnostic GUI

CVA

What-if trade
CVA
CVA desk

Marketer
20

2012 IBM Corporation

CVA
override

CVA desk: day in the life

Pre-Deal
Calculate incremental impact of new trade on market and counterparty exposure,
compare to limits
Calculate incremental impact of new trades on CVA, use this information for pricing
Manage excesses

Intra-day
Update market and credit exposure profiles to include impact of new trades, as trades
are booked
Investigate counterparty exposures
Ad-hoc stress tests on market factors, CSA parameters, modeling assumptions

End-of-Day
Measure market and counterparty exposures for limit and reporting purposes,
including VaR, stress tests and counterparty PFE (shortfall above 90%), and CVA
Model counterparty portfolio diversification, netting, and collateral
Cover all trades including exotics
Calculate CVA across market scenarios to hedge P/L volatility
Manage excesses
21

2012 IBM Corporation

Pro-active CVA management

22

2012 IBM Corporation

CVA desk: key challenges

Positioning of CVA desk

Centralised or decentralised

Profit centre or utility

Hedging policy

23

Basis, proxies, liquidity, market gaps

Overtrading due to unstable sensitivities

Divergence between business practice and regulation (Basel III)

DVA (Debt Value Adjustment)

Should you monetise your own default?

Link to funding

Wrong-way risk

How to minimize wrong-way risk

How to create right way exposures

Tight operational integration and fast analytics are both essential

2012 IBM Corporation

CVA sensitivities
Credit
FX
Rates
Commodities
Equities
Volatility
Correlation
Cross gamma sensitivities
Full simulation approach
High performance computing

24

2012 IBM Corporation

Real time architecture

Credit
Derivative

Interest
Rate

Equity

FX

Trade Data Feeds

2

Control

Desktop
1

Collateral

Real
Time

Batch

Analysis

Trades
Hierarchy
Limits

Scenario
2

Simulation

Aggregation
& Limit Check

3
csv

Presentation

Mkt Data

Database

Netting

Calculation Engines

Market

Cpty

Transient Intra Day

Cache

Real Time

Scenario Builder

Limits

Report

If using batch
load
Positions
Exposure

Historical Report Db

Report

Analysis

25
2012 IBM Corporation

BI Tools

Case study - energy

RESULTS - ENERGY
All counterparties (PFE)

38% reduction

Top 5 counterparties (PFE)

61% reduction

5 counterparties (PFE)

300% increase!

CVA

33% reduction

PERFORMANCE - ENERGY
End-of-day elapsed time

3.5 hours

Grid nodes

48 engines / 6 machines

Computational time

113 CPU hrs

Shortest simulation

7.89 seconds

Longest simulation

4631.17 seconds (~1.5 hours)

97% of portfolio simulates in less than 30 minutes

26

2012 IBM Corporation

Case study equity derivatives

RESULTS - EQUITY
All counterparties (PFE)

38% reduction

Top 5 counterparties (PFE)

53% reduction

CVA

No model was used to calculate the

value before

Combined exposures simulated

10% of the counterparties shared

PERFORMANCE - EQUITY

27

End-of-day elapsed time

2 hours

Grid nodes

48 engines / 6 machines

Computational time

18 CPU hours

2012 IBM Corporation

What are we seeing?

CVA is calculated in middle office/finance

CVA desks being established

Real time CVA pricing (sub second response time)

CVA sensitivities (over 100 sensitivities required)

Capital charge affect on RWAs

Liquidity/funding value adjustment

28

2012 IBM Corporation

Summary

CVA enables, in fact, requires a strategic change in risk culture and

practice. It needs to fit within a broader business vision.

CVA should be managed actively: CVA desk

Short cuts will not work: wrong way risk

Best practice implementations are achievable they are about both,

tight operational integration and fast and accurate analytics

CCR and CVA approaches need to be consistent and should leverage

each other. Regulatory CVA is a separate stream.

29

2012 IBM Corporation