1 1 What are Quantitative Modelers Learning from the Crisis? Charles Richard – Cofounder QRM Chicago | London | Singapore
11
What are Quantitative Modelers
Learning from the Crisis?
Charles Richard – Cofounder QRM
Chicago | London | Singapore
22
Introduction: Quantitative model are here to stay
Assumptions
Behavioral
Models
Miss-use of Quantitative Analytical Models
The Way forward
Stress tests
The US Government Stress Test Case Study
Presentation Outline
33
How many of you have been on an plane recently?
Quantitative Models are here to stay
44
By definition, all behavioral models are limited and will never be 100% reliable
Behavioral models are almost always based on historical data
Quantitative modelers know that behavioral models have bounds
History also trained us to focus more on certain types of behavioral assumptions to give short shrift to others:
Big focus on Prepayments, less focus on Defaults and Recovery and the important negative correlation between them
Unfunded commitments (Commercial Lines, helocs, credit cards) impact on liquidity and credit risk were greatly over looked
New volume, run off, reinvestment and its impact on the dynamics of the balance sheet was universally ignored by “silo based risk management organizations”
Behavioral Models
55
By definition, all quantitative analytical models are limited
and will never be 100% reliable
Quantitative analytical models are almost always based on historical
data
Quantitative analytical modelers know that their models have
bounds and cannot be universally applied
Popular Quantitative models such as Value At Risk,
Economic Capital models proved largely ineffective during
the crisis
They are driven by statistical correlation assumptions that by
definition cannot predict behavior that is outside of the sample used
to build them
Analytical Models
66
We failed VAR, VAR did not fail us.
Many organizations used their Var number primarily for compliance
Pre crisis, Daily/monthly/annual limits were never breached
During the crisis, once in a lifetime moves started happening every
week and the models did not keep pace, so limits were adjusted,
assumptions were not
Sound Judgment was not universally appliedModels were either not believed or they were over believed
If we all fail in the worst case, we will get bailed out so why capitalize for
systemic risk anyway?
Miss-use of Quantitative Analytical Models
77
Stress Tests are not limited by History, only by your imagination
They can cover a wide range of possible scenarios and can be
based on multiple market and economic factors
Interest rate, exchange rate, credit spreads, economic drivers
They can be used in conjunction with new volume, runoff and
reinvestment strategies that are consistent with management’s
vision for the organization
They can be used in conjunction with quantitative behavioral and
correlation models to provide the best of both worlds
The results can be used to manage the risk and enhance the
profitability, not just satisfy the regulators
The way forward: Stress Testing
88
Stress Tests Scenarios
Losses and Prepayments
Total Exposure and RWA-Baseline Scenario
Net Income and Tier 1 Capital
New Volume by Credit Score
Losses in SCAP Scenarios
Capital Ratio Under SCAP Scenarios
The US Government Stress Test
99
Stress Test Scenarios
1010
Losses and Prepayments Under SCAP
Scenarios
0
10
20
30
40
50
60
70
80
90
12/3
1/2
008
01/3
1/2
009
02/2
8/2
009
03/3
1/2
009
04/3
0/2
009
05/3
1/2
009
06/3
0/2
009
07/3
1/2
009
08/3
1/2
009
09/3
0/2
009
10/3
1/2
009
11/3
0/2
009
12/3
1/2
009
01/3
1/2
010
02/2
8/2
010
03/3
1/2
010
04/3
0/2
010
05/3
1/2
010
06/3
0/2
010
07/3
1/2
010
08/3
1/2
010
09/3
0/2
010
10/3
1/2
010
11/3
0/2
010
12/3
1/2
010
Mil
lio
ns Losses Under the SCAP Scenarios
CAP Baseline CAP More Adverse
0
50
100
150
200
250
12/3
1/2
008
01/3
1/2
009
02/2
8/2
009
03/3
1/2
009
04/3
0/2
009
05/3
1/2
009
06/3
0/2
009
07/3
1/2
009
08/3
1/2
009
09/3
0/2
009
10/3
1/2
009
11/3
0/2
009
12/3
1/2
009
01/3
1/2
010
02/2
8/2
010
03/3
1/2
010
04/3
0/2
010
05/3
1/2
010
06/3
0/2
010
07/3
1/2
010
08/3
1/2
010
09/3
0/2
010
10/3
1/2
010
11/3
0/2
010
12/3
1/2
010
Mil
lio
ns Prepayments Under the SCAP Scenarios
CAP Baseline CAP More Adverse
1111
Total Exposure and Risk Weighted Assets
Baseline Scenario
-
100,000
200,000
300,000
400,000
500,000
600,000
-
100,000
200,000
300,000
400,000
500,000
600,000
12
/31
/20
08
01
/31
/20
09
02
/28
/20
09
03
/31
/20
09
04
/30
/20
09
05
/31
/20
09
06
/30
/20
09
07
/31
/20
09
08
/31
/20
09
09
/30
/20
09
10
/31
/20
09
11
/30
/20
09
12
/31
/20
09
01
/31
/20
10
02
/28
/20
10
03
/31
/20
10
04
/30
/20
10
05
/31
/20
10
06
/30
/20
10
07
/31
/20
10
08
/31
/20
10
09
/30
/20
10
10
/31
/20
10
11
/30
/20
10
12
/31
/20
10
Cre
dit R
WA
Th
ou
san
ds
Ex
po
su
re b
y C
red
it S
eg
me
nt
Th
ou
san
ds
Total Exposure and Risk Weighted Assets - Baseline Scenario
700-799 600-699 500-599 RWA
1212
Net Income and Tier 1 Capital
0
2
4
6
8
10
12
-7,000
-5,000
-3,000
-1,000
1,000
3,000
5,000
7,000
01
/31
/20
09
02
/28
/20
09
03
/31
/20
09
04
/30
/20
09
05
/31
/20
09
06
/30
/20
09
07
/31
/20
09
08
/31
/20
09
09
/30
/20
09
10
/31
/20
09
11
/30
/20
09
12
/31
/20
09
01
/31
/20
10
02
/28
/20
10
03
/31
/20
10
04
/30
/20
10
05
/31
/20
10
06
/30
/20
10
07
/31
/20
10
08
/31
/20
10
09
/30
/20
10
10
/31
/20
10
11
/30
/20
10
12
/31
/20
10
Cap
ital R
ati
o
Th
ou
sa
nd
s
Net Income and Tier 1 Capital Under Baseline Scenario
Net Income Tier 1
0
2
4
6
8
10
12
-7,000
-5,000
-3,000
-1,000
1,000
3,000
5,000
7,000
01
/31
/20
09
02
/28
/20
09
03
/31
/20
09
04
/30
/20
09
05
/31
/20
09
06
/30
/20
09
07
/31
/20
09
08
/31
/20
09
09
/30
/20
09
10
/31
/20
09
11
/30
/20
09
12
/31
/20
09
01
/31
/20
10
02
/28
/20
10
03
/31
/20
10
04
/30
/20
10
05
/31
/20
10
06
/30
/20
10
07
/31
/20
10
08
/31
/20
10
09
/30
/20
10
10
/31
/20
10
11
/30
/20
10
12
/31
/20
10
Cap
ital R
ati
o
Th
ou
san
ds
Net Income and Tier 1 Capital Under More Adverse Scenario
Net Income Tier 1
1313
New Volume Income by Credit Score
0
1
2
3
4
5
6
7
8
9
10
Base More Adverse
Base More Adverse
Base More Adverse
Base More Adverse
Base More Adverse
Base More Adverse
Base More Adverse
Base More Adverse
Q1/2009 Q2/2009 Q3/2009 Q4/2009 Q1/2010 Q2/2010 Q3/2010 Q4/2010
Mil
lio
ns
New Volume Income by Credit Score in the SCAP Scenarios
Rating 1-4 Rating 5-8 Rating 9-12 Rating 13-16
60% of the income is being generated by
mid-low tier credit quality obligors, which
increases the risk
1414
Losses in SCAP Scenarios
0
10000000
20000000
30000000
40000000
50000000
60000000
70000000
80000000
Losses in SCAP Scenarios
CAP Baseline CAP More Adverse
1515
Capital Ratio Under SCAP Scenarios
4
5
6
7
8
9
10
11
Capital Ratio Under SCAP Scenarios
Reinvest - Baseline Reinvest - More Adverse Growth - Baseline Growth - More Adverse
Actual growth strategy is critical
and in this example will result in
increased capital needs
1616
If you would like a copy of the QRM’s Government Stress
test case study or any additional information please contact:
Thank you for your attention
Please enjoy the Asian Bankers Summit!
Additional QRM information
1717
World Headquarters:181 West Madison Street, 41st Floor
Chicago, IL 60602+1 (312) 782 1880
London Office:288 Bishopsgate Suite 1.04
London, EC2M 4QP+44 (0) 207 959 3075
Singapore Office:80 Raffles Place, #36-08 UOB Plaza 1
Singapore 048624+65 6248 4545
www.qrm.com