Transcript Slide 1

The IIASA Modeling Tool
Natural disaster risk management
The CatSim Model
Stefan Hochrainer
Department of Statistics and Decision Support Systems (University of Vienna)
Risk, Modeling and Society Group (IIASA)
Introduction:
Ex-ante measures: Measures undertaken
before the disaster happens:
-
Mitigation
Insurance
Reserve Fund
Contingent Credit
Ex- post measures: Measures undertaken
after the disaster happens
–
–
–
–
Diversion
Assistance
Domestic Credit
MFI loans, Int. borrowing
The CatSim Model
What does the CatSim Model?
It assesses the costs and risks of financial vulnerability and
analysis selected ex-ante financial instruments measures for
reducing vulnerability.
What is unique/new about the CatSim Model ?
First integrated modeling approach to assess financial risk
management strategies for natural disaster.
Includes ex-ante and ex-post measures from an intercorrelated perspective.
User can change interesting parameters and assess the
consequences directly.
Probability based approach and dynamic modeling of
economic effects.
CatSim Model
Disaster risk management for
developing countries as a two stage
decision problem under uncertainty.
• First stage: Ex-ante
• Second stage: Ex-post
Integrative view: The scope of possible
actions at stage two influences the
decision at stage one .
CatSim Model
Model uses Monte Carlo Simulation Technique
(probability based approach)
– Important sampling algorithm to generate the scenarios
from a given damage distribution function.
– Scenarios are stratified samples instead of uniformly
events.
Model evaluates the scenarios dynamically in
the time horizon.
User Interface: Strategies for government
financing of disaster risk can be developed
and its costs and consequences on important
indicators studied.
CatSim Model: Modules
Model consists of two parts:
Module I: Assessment of Financial
Vulnerability for the next year
for various impacts of a disaster.
(Limited information approach)
Module II: Assessment of Financial
Vulnerability for a given time
horizon using ex-ante and ex-post
measures (Probability based approach)
Hazard
Floods, earthquakes etc.
Elements at risk
Capital stock, population
Risk
Potential direct losses
STEP 1
Physical Vulnerability
Susceptibility to physical
damage
Financial
vulnerability/potential
financing gaps
Ability to finance reconstruction of
lost stocks and provide assistance to
households and private sector
STEP 2
Ex-ante instruments
• Mitigation
• Insurance
• Reserve fund
• Contingent credit
Macroeconomic impacts
Effects of losing capital stock and
diverting funds for financing losses
STEP 3
CatSim Model:User Interface
CatSim Model:User Interface: Module I
User-Interface: Module I: Hazard
User-Interface: Module I: Vulnerability
User-Interface: Module I: Elements at risk
CatSim Model:Case Studie Honduras
CatSim Model:Case Studie Honduras
CatSim Model:Case Studie Honduras
CatSim Model:Case Studie Honduras
CatSim Model:Case Studie Honduras
Conclusions
Honduras is highly indepted and highly exposed to
natural disaster.
Honduras is very dependent for borrowing on loans.
Mitigation for the lower year events.
Insurance for the higher events.
End of Presentation
Thank you
Questions?
Decision and Response variables
Decision variables:
Expenses for mitigation.
XL-Insurance.
Contribution to reserve fund.
Fee for contingent credit.
Response variables:
Discounted expected return for the next x (e.g.11) years.
Shortfall probability for the next x years
Expected reduction of the credit buffer in the next x years.
Input Parameters
Economic parameters:
Return on capital
Discount rate for future returns
Depreciation rate
Factor for mitigation
Premium loadings for insurance
Interest rate for reserve fund,
contingent credit, domestic
credit, MFI loan, international
bond
Fee for contingent credit
Maximal Diversion
Maximal Domestic Credit
Initial capital
Initial reserve fund
Fixed budget (planned for
t=1,..,x years)
Credit buffer
Catastrophe parameters:
Probability of first loss
20-years event loss
50-years event loss
100-years event loss
500-years event loss
1000-years event loss
Simulation parameters:
Time horizon
Number of Scenarios
Expenditure length
Mitigation:
Loss
Return period of event
Bold printed line shows the loss as a function of the “hypothetical” loss
without mitigation.
Up to a limit given by the invested mitigation no loss occurs.
If the “hypothetical” loss is larger than this limit, the full loss occurs.
Hence there are two negative effects at once.
Pricing of Insurance Contracts:
Insure against certain "layers" of risk, e.g. insuring against events in
excess of the 100 year up to the 500 year hurricane event.
The XL layer is determined by two points: The attachment point (A) and the
exit point (E). The payment depends on the size of the damage (D).
So the insurance only pays claims if the damage is larger tan (A)
In other words, the insurance pays:
0
if
D<A
D-A
if
A < = D <= E
E-A
if
D>E
claim
damage
A
E
Pricing of Insurance Contracts:
If the damage distribution function is denoted by F(z), the par-price
thencan be calculated by integration:
ParPrice =  A(z) d F(z)
However, the insurance company asks for a risk premium to be
added to the ParPrice, this risk premium must be greater (or equal)
1 and monoton increasing.
To calculate the new price, a function h(p), 0  p  1, is considered,
which has the property stated above, namely:
h(p)  1
h(p) is increasing
Pricing of Insurance Contracts:
AdjustedPrice =  A(z) h(F(z)) d F(z)
=  A(F-1(p)) h(p) dp
=  A(z) h(F(z)) f(z) dz
The function g(z) = h(F(z)) f(z) can be seen as a kind of weight function
7
6
5
4
3
2
1
0
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45