Download file_3200 - Alfa-Bank

Implementing Risk Management:
Adapting Normative Practice to
Russian Specifics
Challenges and Compensatory Techniques
Adam Smith Conference – Russian Banking
London 6 December 2005
Philip M Halperin
1
Alfa Bank snapshot
•112 branches in Russia, subsidiary banks in
Ukraine and Kazakhstan, Holland, brokers in
London and NYC
•Universal bank - Retail, Commercial, Investment
and Asset Management lines of business
2
Conceptual underpinning of
Unified Risk Management
Credit Risk
Retail
Portfolio
Management
Operational
Risk
Market
Risk
Unified
Risk Management – Overlapping skills sets, avoid silos!
3
Agenda
Institutional issues – transition economies
(Russia, Kazakhstan, Ukraine):
Compensatory Technique – Implementation of
Risk Management and RM Culture
Compensatory Technique – Adaptation of
Normative Best Practice to Meet Challenges
1)
2)
3)

Examples from Risk Management Areas – Market,
Credit, Operational RM.
4
Institutional issues – transition economies
(Russia, Kazakhstan, Ukraine):
Former Soviet Union:
• Lack of standard ways of operating – everything is new
• Lack of historic memory –commercial bank system just
14 years old
• Formalist documentary culture. Partially compensates for
lack of historic memory.
• Lack of standard Western corporate institutions (Internal
Audit, Compliance)
• Special Soviet-legacy institutions (Chief Accountant,
“Internal Control”, Economic Security)
• Transition from Russian Accounting System to IFRS
5
Institutional issues – transition economies
(Russia, Kazakhstan, Ukraine):
• Information-poor environment
• Lack of liquidity, transparency in markets
• Mismatch between risks and controls due to fast
business, processes, and automation change
Most importantly … lack of risk culture … of
managerial responsibility and systemic risks solutions
6
Institutional issues – transition economies
(Russia, Kazakhstan, Ukraine):
• The environment and institutions of a transition
economy are not conducive to standard
technique only!
• One must be creative, and seek compensatory
technique.
7
Achieving Management Buy-In – Document
Ju-Jitsu
• Developed and
promoted Group
level Risk Policies
• Agreed and signed
by every business
leader
• Established
business
responsibility as
Principle #1
•
(Regulators like it as
well!)
8
Examples of Compensatory Techniques Implementation
• Use formal ‘Document Culture’ to implement new risk
culture.
• Convert ‘Economic Security’ to be an active partner in
the loss-collection process.
– Threshold level to ensure business responsibility
• Use former-Soviet style collegial-bodies meetings to
new goals.
9
Market Risk Management - Challenges
Challenges:
 Lack of transparency in pricing
 Fat tails – market moves “squirtier” than in
developed markets
 Illiquidity in pricing
 Crises come along every 7 years or so
10
Compensatory Techniques – Market Risk
VaR (one day) vs. Liquidation Adjusted VaR
VaR, LaVaR (mio USD)
$50
$45
VaR (1% ql), mio USD
$40
Liquidation adjusted
VaR (1% ql), mio USD
$35
$30
$25
$20
$15
$10
$5
n
tio
Po
si
tio
n
To
ta
l
os
i
n
U
kr
P
tio
al
co
w
Po
si
To
t
B
oo
k
To
t
al
D
M
os
er
iv
y
at
iv
e
s
Po
si
tio
n
Fo
re
x
Tu
rk
e
kr
ow
+U
os
c
FI
M
ui
ty
M
os
co
w
+U
kr
U
kr
ui
ty
Eq
Eq
Eq
ui
ty
M
os
co
w
$0
11
Compensatory Techniques – Market Risk
VaR (one day) vs. Stress Testing (one day)
VaR, Stress (mio USD)
$50
$45
VaR (1% ql), mio USD
$40
Stress Testing, mio USD
$35
$30
$25
$20
$15
$10
$5
tio
n
on
os
i
lP
To
ta
Po
kr
To
ta
l
U
ow
M
os
c
To
ta
l
si
ti
si
tio
Po
es
at
iv
iv
er
D
n
k
B
oo
tio
si
Po
Tu
rk
ey
co
os
M
FI
n
x
re
Fo
w
+U
w
+U
co
os
M
ui
ty
kr
kr
U
kr
ty
ui
Eq
Eq
Eq
ui
ty
M
os
co
w
$0
12
Compensatory Techniques – Market Risk
Russian Trading System (RTS) Index:
Empirical vs. Student's t Distribution
7.00%
Empirical
distribution
Normal distribution
Frequency (%)
6.00%
5.00%
Student's t (4 d.f.)
4.00%
3.00%
1% quantile
(Student's t)
1% quantile
(Normal)
2.00%
1.00%
0.00%
-7.00% -5.80% -4.60% -3.40% -2.20% -1.00% 0.20% 1.40% 2.60% 3.80% 5.00% 6.20%
Return (%)
13
Credit Risk Management - Challenges

Lack of reliable balance sheet data (IFRS desirable),
difficulty of correct factors choice

Lack of default data for calibration

Few borrowers have quoted liquid bonds

Few borrowers have liquid stocks

Asset market value and its volatility are usually
unobserved
14
Credit Expected Loss Estimation
Components of credit risk
estimation:
1.
2.
3.
Credit exposure (CE)
Probability of default (PD)
Loss given default (LGD) or
Recovery rate (RR)
LGD=1-RR
«Event tree» of the credit:
CE
1-PD
PD
Credit is paid off:
loss=0
CE*RR
Default:
loss=CE*(1-RR)
Credit Expected Loss estimation:
Expected Loss = (1-PD)*0+PD*CE*(1-RR) = PD*CE*(1-RR)
15
Approaches to credit risk estimation (1)
1. Altman Z-model.
Construct a formula for a borrower’s ‘score’ (which determines
default rate) depending on balance sheet ratios.
Doesn’t suit Russia:
- Lack of reliable balance sheet data (IFRS desirable), difficulty
of correct factors choice
- Lack of default data for calibration
2. Market bond prices as an indicator of the issuer’s credit risk.
Corporate bond spread can be transformed into market
estimation of default probability.
Limited applicability to Russia:
- Few borrowers have quoted liquid bonds.
16
Approaches to credit risk estimation (2)
3. Structural models (Merton).
Default process is endogenous and relates to borrower’s capital
structure. (Analogous to option pricing B-S model for Market
Value of borrower’s assets). Default happens when value of
assets fall below face value of debt.
Limited application to Russia:
- Few borrowers have liquid stocks
- Asset market value and its volatility are usually unobserved
17
Credit RM – Compensatory Technique
4. Reduced-form models.
Default probability depends on exogenous factors (borrower’s
characteristics, general economic variables); the
dependency can be specified by logit or probit model.
This works in Russian environment.
• Statistically significant estimates of default probability
were obtained using Bank’s proprietary credit portfolio
payments and default information
18
Credit RM – Compensatory Technique
Goal: Credit VaR Model: Loss Distribution
12.0%
Квантиль 1% (3.9)
Квантиль 5% (6.8)
Квантиль 95% (104.0)
Квантиль 99% (148.2)
Среднее (39.6)
Станд.откл. (31.5)
10.0%
6.0%
4.0%
2.0%
244
234
224
214
204
194
184
174
164
154
144
134
124
114
104
93.5
83.5
73.5
63.5
53.5
43.5
33.5
23.5
13.5
0.0%
3.52
Частота
8.0%
Потери, $млн
19
Operational RM - Challenges

Loss collection structure reveals native collusion
- Many low impact events
- Few high impact events
- Medium-size events missing (!)
• Essentially no external loss data in our region –
difficult in the existing environment
• Consortia not helpful
• Press monitoring – low usefulness
20
Operational RM – Compensatory Techniques
Compensation:
We got External
Data from
Insurance brokers:
21
Operational RM – Compensatory techniques
Compensation:
Work with
‘Economic
Security Forces’
Patch and match
external
Insurance Broker
sources
22
Operational RM – Compensatory Techniques
• Redesigned BBB Insurance Policy
• Were able to keep reasonable limit, instead of silly
coverage suggested by cost-cutting activists, for
similar money (higher limit, higher excess).
• This actually paid off
Разница цены – $214 000USD
0.7
0.6
0.5
0.4
90%
0.3
99%
0.2
0.1
0
0
1
20 25 40 50
60
млн USD
80
100
120
140
160
23
Summary
• Russia is a market like any other market,
and different from all others like any other.
• We must be
– creative in our implementation.
– compensatory in our adaptation of
technique.
24
Contact Information
Alfa-Bank
http://www.alfabank.com/
http://www.alfabank.ru/
Philip M Halperin, Director of Risk Management
http://www.btinternet.com/~phalperin/
[email protected]
25