Transcript Stress Testing: An Approach to Making Forward

Stress Testing: An Approach to Making
Forward-looking and Objective Scenarios
Deloitte Touche Tohmatsu LLC
Tsuyoshi Oyama
This paper does not include the official view of Deloitte Touche
Tohmatsu LLC but only reflects the lecturer’s own opinion.
Post-crisis Stress Testing -- Required Elements
 Required elements
Deep involvement of board and senior management
Clarification of degree of stresses
Transparent and accountable process of stress testing
Forward-looking scenarios
Objective scenarios
Dynamic scenarios
Comprehensive scenarios based on root causes
Use of the outcome of stress testing for business strategy as well as risk
management
Flexible update corresponding to changes in external environments
Elements required
for stress
conditions
Elements required
for scenarios
Consistency with
Management’s risk
appetite
Objective
Exceptional but
plausible scenarios
Dynamic
Forward-looking
Comprehensive
Macro perspective
Firm-wide scenarios
Market micro
structure perspective
2
Elements required for
the use of stress
testing
Effective integration into
risk management and
business judgments
Transparent and
accounting process
A Way of Making Stress Scenarios 1/2
(Step 1) Scenario conditions
1.1 Determine the degree of stresses (severity and frequency) to be assumed in
stress scenarios in line with risk appetite of senior managers
1.2 Identify all important risk factors for own portfolio and pay higher attention to
the scenarios that could impact them significantly
(Step 2) Prepare the tools to make forward-looking scenarios
2.1 Make several enterprise-wide forward-looking stress scenarios using, for example, the
following tools




3
Several early warning indictors of banking crisis to select the scenarios
The future event database that controls the flood of information of emerging
stress events from international/national agencies, media, academics, etc. on
timely manner
The stress event database that comprises big stress events in the past, of
which causes and root causes are analyzed and categorized for supporting the
scenario-making
Global macro economic modeling techniques that forecast the development
of macro-economies of major countries under the certain scenarios over the
coming 2-3 years
A Way of Making Stress Scenarios 2/2
(Step 3) Determining forward-looking stress events
3.1 Confirm the countries with high probability of banking crisis using the early
warning indicator
3.2 List up 4—5 stress scenarios highlighted by the market, regulators, media and
academics using the future event database and the information of 3.1 and 3.2
3.3 Narrow down the number of stress scenarios by using the conditions identified by
Step 1
3.4 Determine the severities and frequencies of stress scenarios based on the risk
appetite confirmed by Step 1
(Step 4) Narrating the stress scenario stories
4.1 Make the stories of stress scenarios using the stress event database
4.2 Determining the paths of macro-economic indicators of major countries over the
coming 2-3 years under the stress scenarios using the global macro-economic
modeling techniques
(Step 5) Translating the macro stress scenarios into risk parameter
5.1 Translate macroeconomic indicators assumed under the stress scenarios into risk
parameters
5.2 Identify the impacts of stress scenarios on FIs’ portfolios
4
Example 1 Early Warning Indicators
 This tool is expected to enhance forward-looking, objective and macro-oriented
scenarios
The case of early warning indicators of financial crisis occurring in a major country
 International and national agencies, as well as academics have already developed various early
warning indicators relating to financial crisis, which will be used
 These indicators Identify the region(s) with probability of crisis occurring above a certain
threshold
Crisis
prob. 1
Crisis
pro. 2
Japan
16.2
27.3
US
29.5
15.7
UK
15.5
22.2
EU
35.6
43.2
Check!
China
23.1
34.2
Check!
Korea
25.6
12.2
LA
17.8
22.5
Others
…
…
Input data
Macro economic
imbalance
Bank lending
Asset prices
Financial system
stability
Market liquidity
Exchange rate
5
Example 2 Global Risk Heat Map based on the Future Event Database
 This tool is expected to enhance forward-looking, objective, comprehensive and
macro-oriented scenarios
Changes in future event heat map (From August to September 2010)
US
Real
economy
EU
China
WEO (April): Worsening fiscal balance, BOJ Outlook (April): Historically high level of
public debt
Media: Frequent negative comments
ECB (June): Funding cost
hike due to the sovereign
crisis
Sovereign
Japan
Asia
Others

Scenario 2: Euro crisis
⇒a little moderated

Scenario 3: China
bubble bursting ⇒no
change

Scenario 4: JGB
bubble bursting ⇒no
change

New scenario?
⇒Basel III shock
scenario
GFSR (April), ECB (June), BOJ/FSR (March): Increase in Sovereign risk
WEO (April): Sovereign risk, Rewinding leverage and Stabilization of CDS spreads
Banking
system
WEO (April): Emerging risk of holding government bonds
Financial
market
ECB (June): Risk of interest rate hike due to widening fiscal deficit
GFSR (April): Roll-over of banks’ short-term debts Media:
Banks credit crunching
Media: Pressures
on Renminbi
appreciation
Ex market
BOJ FMR (July): Yen
appreciation
Stock
market
BOJ FMR (July):
High volatility due to
the Yen’s
appreciation
Media: Pressures
on currency
appreciation
GFSR (April): Asset bubbles
BOJ Outlook (April) Asset bubbles
BOJ FSR (March):
Deteriorating quality of
commercial mortgage loans
Credit
Media: Export limits
on wheat/ Russia
Commodity
Bank
regulation
6
Scenario 1: US
economy double dip
recession ⇒probability
increases
WEO(April) : Capital inflow
GFSR (April): Inflation
BOJ Outlook (April):
Decaying the fiscal
stimulus impacts
GFSR (July), BOE (June):
Repercussion of emerging
sovereign risk
Real estate

Media: Basel III framework set, discussions on capital surcharge for SIFI
Dodd-Frank
EU/UK regulations
Example 3 Stress Event Database 1/2
 This tool is expected to enhance objective, dynamic and macro scenarios
 There were so many financial and economic crises in the past
 While the same event with the past could not occur, the events having similar elements with the
past tend to occur repeatedly
7
Example 3 Stress Event Database 2/2
 The trigger event, cause and root cause of each event are analyzed and classified into several
categories so as to facilitate the process of making the scenarios with a modern combination of
some elements of past crises
Stress event DB structure
8
Example 4 Macroeconomic modeling
 This tool is expected to enhance objective, dynamic and macro-oriented scenarios
Stock price
Real GDP growth rate
4.0
1,500.0
3.0
1,400.0
1,300.0
2.0
1,200.0
1.0
1,100.0
1,000.0
0.0
2009
2010
2011
2012
2013
▲ 1.0
900.0
800.0
▲ 2.0
700.0
▲ 3.0
600.0
2009
Baseline/Japan
ベースライン（日本）
Stress/Japan
ストレス（日本）
ベースライン（米国）
Baseline/US
ストレス（米国）
Stress/US
Long-term interest rate
7.0
6.0
5.0
4.0
3.0
2.0
1.0
0.0
2009
9
2010
2011
2012
2013
2010
2011
2012
2013
Example 5 Stresses associated with market micro structure
 This tool is expected to enhance objective and micro-structure based scenarios
Fundamental
Macroeconomic Causes
Macro
Market Micro-structure
Unexpectedly immobilized position
Market Crash
Market liquidity
evaporation
Huge losses
Delayed
recognition of
market
position
Market instability
Market Micro-structure
Micro
10
Fund-raising problem
Unexpected
correlation
Typical factors
behind
historical huge
losses of
investment
banks
Example 6 Translation of the development of macro economic
variables under the stress scenarios into risk parameters
Estimating the effects of macro economic change on credit cost and credit risk: PD case
Parameter
Approach
Some examples
Direct Approach
Directly adjust PD
Indirect Approach
Putting stress on
migration matrices
PD
【Concepts of PIT and TTC】
PIT (point in time)
債務者数
景
気
改
善
時
良
低
１
２
３
４
５
６
７
悪
８ 格付
TTC (through the cycle)
債務者数
モ
デ
ル
構
築
時
景
気
悪
化
時
一定
良
低
１
２
３
４
５
６
７
８
好景気
一定
高
11
モデル構築時
高
実績デフォルト率
Putting stress on credit
scoring
実績デフォルト率
不景気
デフォルト率
格付毎で安定
デフォルト率
景気により変動
債務者数
景気により変動
債務者数
格付毎で安定
Putting stress on
financial data of each
obligor
① Specification of the relationship
between economic cycle/GDP growth
rate and PD by industry
② Specific hypothesis in accordance
with shock scenario (e.g., double the
PD for lowly rated bonds, setting
PD=100% for top X% of major client)
① First confirming the state dependency
of rating migration and then applying
the rating migration under economic
downturn
② First estimating the relationship
between credit scores and macro
economic variables and then using the
result to calculate PD under the stress
scenario
③ First estimating the sensitivity of
financial data of major obligors to main
macro economic variables and then
using this information to estimate the
financial state of these obligors under
the stress scenario