Transcript The Case of Germany - American Risk and Insurance Association

Corporate Risk Management and
the Perception of Terrorism Risk:
The Case of Germany
Christian Thomann*
J.-Matthias Graf von der Schulenburg
Bruno Gas
Razvan Pascalau*
*U of Alabama
Quebec, August, 6th, 2007
Research Question

How do corporations learn about dynamic
risks?
 Terrorism Risk is highly dynamic, potentially
catastrophic and not well known

It thus provides for an opportunity to:



Test if Prospective Reference Theory (Viscusi, 1989)
can be applied in a corporate context
Learn how the inflow of new information changes a
corporation‘s risk management
Understand how corporations weigh new information
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
2
Overview
Introduction:
Terrorism, Insurance and Extremus
Dataset
Corporate Risk Attitudes and
Individual Risk Perception
Hypotheses and Empirical Investigation
Conclusion
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
3
Overview
Introduction:
Terrorism, Insurance and Extremus
Dataset
Corporate Risk Attitudes and
Individual Risk Perception
Hypotheses and Empirical Investigation
Conclusion
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
4
Terrorism
A definition
An Act of Terrorism means an act of any person
acting on behalf of or in connection with any
organisation with activities directed towards
the overthrowing or influencing of any
government de jure or de facto by force or
violence.
Reinsurance (Acts of Terrorism) Act 1993
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
5
Terrorism has changed

Old Terrorism (Wilkinson 1986, Hoffman 1992)



New Terrorism (Enders and Sandler 2000, Hoffman
1997)



Motivation: Separation, Nationalism, Marxist Ideology, Economic
Inequality, Goals are well defined
Organizational Structure: Command and Control. Terrorists are
trained and are committed full-time to their cause
Motivation: Less comprehensible, embrace amorphous religious
aims
Organizational Structure: International networks that can be
diffuse and spontaneous. Terrorists may live normal lives
Terrorism in its present form poses a new and
significant challenge to Risk Managers
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
6
Dynamics of Terrorism
Homeland Security Advisory System
Date
March 12, 2002
September 10, 2002
September 24, 2002
February 7, 2003
February 27, 2003
March 17, 2003
April 16, 2003
May 20, 2003
May 30, 2003
December 21, 2003
January 9, 2004
August 1, 2004
November 10, 2004
July 7, 2005
August 12, 2005
August 10, 2006
August 13, 2006
Threat Level
Introduction: Yellow
Orange
Yellow
Orange
Yellow
Orange
Yellow
Orange
Yellow
Orange
Yellow
Orange financial services sectors in NYC, NJ and DC
Yellow financial services sectors in NYC, NJ and DC
Raised from Yellow to Orange for mass transit
Lowered from Orange to Yellow for mass transit
Red for flights from the UK to the United States
Orange for flights from the UK to the United States
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
7
Dynamics of Terrorism
Homeland Security Advisory System
Date
March 12, 2002
September 10, 2002
September 24, 2002
February 7, 2003
February 27, 2003
March 17, 2003
April 16, 2003
May 20, 2003
May 30, 2003
December 21, 2003
January 9, 2004
August 1, 2004
November 10, 2004
July 7, 2005
August 12, 2005
August 10, 2006
August 13, 2006
Threat Level
Introduction: Yellow
Orange
Yellow
Orange
Yellow
Orange
Yellow
Orange
Yellow
Orange
Yellow
Orange financial services sectors in NYC, NJ and DC
Yellow financial services sectors in NYC, NJ and DC
Raised from Yellow to Orange for mass transit
Lowered from Orange to Yellow for mass transit
Red for flights from the UK to the United States
Orange for flights from the UK to the United States
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
8
Assessing Terrorism Risk

Difficult to assess risk of a terrorist attack

GAO (2004): Homeland Security Advisory System:
No explicit criteria or quantifiable factors are
used to determine the treat level. Threat levels
include a certain amount of subjectivity

Fischhoff et al. (2003): Individuals significantly
overestimate their exposure to terrorism risk:


11/2001: 43 % (19 %) of respondents living within (outside) a 100
miles of the WTC: 50 % or higher chance of themselves being
hurt in a terrorist attack.
In Addition: Information provided by governments and
media is likely to be biased
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
9
Terrorism and Insurance


Before 9/11 Terrorism is commonly included in Standard
Insurance Policies
9/11



Governments Intervene on Terrorism insurance markets




Highest Insured Loss Due to Terrorism
Results in Exclusion of Terrorism from Standard Insurance
Contracts
United States (TRIA)
Germany (Extremus)
France (GAREAT)
Interventions in Germany and US must be prolonged in
2007
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
10
Terrorism Insurance in Germany
After September, 11th
Government Intervention: Primary Insurer (EXTREMUS)
Cover


Risks > 25 million €
Buildings, Content, Business Interruption and Clean up
Costs
Exclusions

NBC, War, ...
Limit for Compensation: 1.5 bn €
100 % reinsurance by insurance industry & German Government
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
11
Overview
Introduction:
Terrorism, Insurance and Extremus
Dataset
Corporate Risk Attitudes and
Individual Risk Perception
Hypotheses and Empirical Investigation
Conclusion
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
12
Data Set
Overview

Data on all government-reinsured terrorism
insurance purchases between 11/2002 and
3/2007 in Germany (n>5000)




Name, Industry and Location of Policyholders
Amount of Coverage Purchased
Price of Insurance Coverage
Start and End of Insurance Coverage
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
13
Data Set, Descriptive Statistics
Policyholder
2002/3
Mean
2004
2005
2006
2007
20022007
554
377
378
365
415
419
p50
48.30
54.20
55.80
58.60
61.10
55.00
Mean
71.20
73.00
75.60
80.70
85.80
77.20
p50
43.30
46.40
47.60
50.00
50.30
47.80
Degree of
Coverage
Mean
0.85
0.86
0.85
0.87
0.87
0.86
p50
1.00
1.00
1.00
1.00
1.00
1.00
Net
Premium (€)
Mean
89,014
73,207 58,934
57,151 56,220
67,095
p50
9,558
11,246
11,654 12,858
11,201
MPL (m €)
UL (m €)
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
11,389
14
Data Set, Descriptive Statistics
By Industry and Year
Banks Asset Managers
Construction
Utilities
Airports
Real Estate
Real Estate Inv Funds
Churches, Foundations
Hospitals
Logistics
Media, IT
Local Authorities
Heavy Industry
Transportation
Stores, Art, Fairs, Other, Tourism
Insurance
Total
2002/3
99
19
19
26
490
96
21
9
11
34
32
28
10
92
194
1180
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
2004
111
19
16
22
452
100
20
5
15
33
18
31
11
100
118
1072
2005
106
31
16
20
460
107
19
7
14
31
18
36
12
112
127
1117
2006
105
23
14
22
530
120
21
7
15
31
18
35
11
122
121
1195
15
2007
92
13
13
17
492
107
17
7
12
24
15
23
12
101
105
1050
Data Set, Descriptive Statistics
By Industry and Year
2002/3
99
19
19
26
490
96
21
9
11
34
32
28
10
92
2004
111
19
16
22
452
100
20
5
15
33
18
31
11
100
2005
106
31
16
20
460
107
19
7
14
31
18
36
12
112
2006
105
23
14
22
530
120
21
7
15
31
18
35
11
122
2007
92
13
13
17
492
107
17
7
12
24
15
23
12
101
Insurance
194
118
127
121
105
Total
1180
1072
1117
1195
1050
Banks Asset Managers
Construction
Utilities
Airports
Real Estate
Real Estate Inv Funds
Churches, Foundations
Hospitals
Logistics
Media, IT
Local Authorities
Heavy Industry
Transportation
Stores, Art, Fairs, Other, Tourism
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
16
Data Set, Descriptive Statistics
By Month
ΔPRInc
Policies Sold
Month
Mean
SD
Min
Max
Mean
SD
Min
Max
1
924
186
599
1054
-329
2432
-3083
3513
2
40
63
9
153
126
274
-6
616
3
30
40
2
101
306
841
-284
1792
4
30
37
5
84
273
500
4
1022
5
13
6
5
17
32
17
17
52
6
16
12
9
33
26
10
19
41
7
21
8
11
30
12
89
-107
107
8
13
9
7
26
18
27
-22
42
9
12
4
7
16
23
15
4
39
10
14
8
6
23
43
43
14
107
11
9
6
3
19
22
10
14
37
12
24
22
10
62
386
778
21
1778
Total
106
273
2
1054
82
790
-3083
3513
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
17
Data Set, Descriptive Statistics
By Month
ΔPRInc
Policies Sold
Month
Mean
SD
Min
Max
Mean
SD
Min
Max
1
924
186
599
1054
-329
2432
-3083
3513
2
40
63
9
153
126
274
-6
616
3
30
40
2
101
306
841
-284
1792
4
30
37
5
84
273
500
4
1022
5
13
6
5
17
32
17
17
52
6
16
12
9
33
26
10
19
41
7
21
8
11
30
12
89
-107
107
8
13
9
7
26
18
27
-22
42
9
12
4
7
16
23
15
4
39
10
14
8
6
23
43
43
14
107
11
9
6
3
19
22
10
14
37
12
24
22
10
62
386
778
21
1778
Total
106
273
2
1054
82
790
-3083
3513
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
18
Overview
Introduction:
Terrorism, Insurance and Extremus
Dataset
Corporate Risk Attitudes and
Individual Risk Perception
Hypotheses and Empirical Investigation
Conclusion
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
19
Corporate Risk Management
Modigliani and Miller (1958)
Mayers and Smith (1982), Stulz (1984),
Froot, Scharfstein, and Stein (1993)




Greater Efficiency in the allocation of risk among a company‘s
stakeholders
Bankruptcy Costs / Costs of financial distress
Taxes
Agency Problems
 Corporations act as if they were risk averse
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
20
Corporate Risk Management
Empirical Studies and Extension

CRM: Empirical Studies (Insurance):

Mayers and Smith (1990), Hoyt and Khang (2000), Kleffner
and Doherty (1996), Cole and McCollough (2006)

CRM and Bankruptcy: Marin (2007)
 CRM: Extension to Ambiguous Risks:

Kunreuther et al. (1995): Ambiguity aversion describes prices
set by insurance underwriters.

CRM of dynamic risks: has not been analyzed
empirically
Further Similarities between individuals’ and
corporations’ behavior
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
21
Prospective Reference Theory
Assessing Dynamic Risks

Viscusi (1989): Prospective Reference Theory




Generalization of the EU model
Decision makers assess Probability with a Bayesian process
Empirically tested Viscusi and Evans (2006), Viscusi and
O’Connor (1984)
Posterior Probability (p*) is the weighted average of:
Prior Probability q (weight: γ)
 Probability p of the outcome observed (weight: ξ= number of trials)

q  p
p* 
 
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
22
Overview
Introduction:
Terrorism, Insurance and Extremus
Dataset
Corporate Risk Attitudes and
Individual Risk Perception
Empirical Investigation
Conclusion
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
23
Events Studied
Event Window Event
3/2004
(-0;+3)
March 11th 2004: Attack on Commuter Trains in
Madrid. Attacks took the lives of 190 people and
wounded approximately 1500.
7/2005
(-0;+3)
July 7th 2005: Attack on Busses and Trains,
London. Attacks took lives of 52 people and
wounded 700.
7/2006
(-0:+2)
Terrorist bombs were discovered on July 31 and
August 1, 2006. Germany
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
24
Time Series Models
( A) Polt  0  1JANUARY  i ATTACKi  i INTERAi   a
( B) PRINCt  0  1JANUARY  i ATTACKi  i INTERAi  b
JANUARY = dummy variable for January
ATTACKi=
Dummy variable: Attacks on Madrid (i=1),
London (2) and the attempts to bomb two
trains in Germany (3)
INTERAi =
Interaction term, denotes a January after
an attack
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
25
OLS Estimates
Δ Polt
JANUARY
-139.39
Δ PRINC
***
(5.55)
ATTACK1
ATTACK2
ATTACK3
INTERA1
4.60
*
INTERA3
_cons
R²
9831.609
(2.64)
(26587.53)
4.10
35944.01
(2.91)
(29271.9)
1.60
46391.62
(3.31)
(33267.96)
62
82
***
1656964
***
(77446.18)
***
2986772
(7.71)
(77446.18)
1
2529190
(7.71)
(77446.18)
9.39
***
(55732.04)
(7.71)
INTERA2
-3092249
***
***
***
9093.249
(1.03)
(10349.18)
0.97
0.99
Absolute value of t statistics in parentheses, * significant at 10 %; ** significant at 5%; ***
significant at 1%
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
26
OLS Estimates
Δ Number of Policies
JANUARY
-139.39
Δ Premium Income
***
(5.55)
ATTACK1
ATTACK2
ATTACK3
INTERA1
4.60
*
INTERA3
_cons
R²
9831.609
(2.64)
(26587.53)
4.10
35944.01
(2.91)
(29271.9)
1.60
46391.62
(3.31)
(33267.96)
62
82
***
1656964
***
(77446.18)
***
2986772
(7.71)
(77446.18)
1
2529190
(7.71)
(77446.18)
9.39
***
(55732.04)
(7.71)
INTERA2
-3092249
***
***
***
9093.249
(1.03)
(10349.18)
0.97
0.99
Absolute value of t statistics in parentheses, * significant at 10 %; ** significant at 5%; ***
significant at 1%
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
27
Results

Coefficients for the dummy Attack-variables are
not significant
 Yet terrorists’ activity during the previous 12
months has a significant influence and
apparently not declining importance on the
demand for terrorism insurance:


Interaction terms are positive and significant.
Negative coefficient for the JANUARY variable:
Captures the drop in demand that resulted from the
absence of major terrorist attacks between November
2002 and January 2004.
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
28
Further Evidence

The data on the renewal of policies also supports the
importance of recent terrorist activity for the demand for
terrorism insurance in Germany
2002/3 2004
% of Contracts
renewed in t+1
% of contracts
renewed until
2007
2005
2006
20022006
76*
62
78
86
80
39
58
71
80
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
29
Overview
Introduction:
Terrorism, Insurance and Extremus
Corporate Risk Attitudes and
Individual Risk Perception
Hypotheses and Empirical Investigation
Conclusion
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
30
Conclusion
Corporations’ Risk Assessments are very
dependent upon recent experience with
dynamic risks
Overweighing of recent events will be a serious
obstacle for the development of a private
insurance markets against highly dynamic risks
Necessary to model corporate management of dynamic
risks with the help of concepts that allow for dynamic
updating
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
31
Terrorism Insurance
Expectations and Experience in Germany
When Extremus was founded in 2002 its shareholders expected
to generate premium income of € 250 millions.
2002/3
2004
2005
2006
2007
20022007
Contracts sold
1180
1071
1116
1195
1050
5612
Premium Income
(€ m)
105
78.4 65.8
68.3
59
377
Sum of Max. Poss.
Losses (€ bn)
653
407
422
436
436
2350
Sum of Upper Limits
(€ bn)
84
78.7
84.5
96.5
90.1
434
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
32
Thank you
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
33
Results
Recent Terrorist Activity has a strong and not declining importance for a
company’s insurance decision.
The demand for terrorism insurance decreases in the absence of
terrorist attacks
Recent terrorist attacks stabilize the demand for terrorism insurance
When making Risk Management Decisions Corporations place a great
and not declining weight on their recent experience with terrorism risk.
Necessary to model corporate management of dynamic risks with the
help of concepts that allow for dynamic updating
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
34
Corporate Risk Management
Empirical Studies and Extensions
Mayers and Smith (1990)
Demand for reinsurance by US insurers
Hoyt and Khang (2000)
Corporate demand for primary insurance (US)
Kleffner and Doherty (1996)
Supply of earthquake insurance in California
McCollough/
McCollough/
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
35
Terrorism Insurance
Experience in Germany and the U.S.
80%
Pick-up Rate
USA 2004
70%
60%
50%
40%
30%
Market Penetration
Extremus Germany
2004
20%
10%
0%
> 25 m.€
> 5 bn. €
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
<100 Mio. $ 5 - 10 bn $
36
Terrorismus
Definition
„Terrorakte sind jegliche Handlungen von Personen
oder Personengruppen zur Erreichung politischer,
religiöser, ethnischer oder ideologischer Ziele, die
geeignet sind, Angst [...] in [...] Teilen der
Bevölkerung zu verbreiten und dadurch auf eine
Regierung oder staatliche Einrichtungen Einfluss zu
nehmen.“
Extremus Allgemeine Bedingungen für die Terrorversicherung
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
37
Absence of Terrorism Insurance
Economic Effects
Hubbard et al. 2005

Absent Major Attack: “…GDP may be $ 53 billion (0.4 %)
lower, household net worth may be $ 512 billion (0.9 %)
lower, and roughly 326,000 (0.2 %) fewer jobs may be
created.”

In Case of an Attack “… tens of thousands more jobs
could be lost due to the lack of insurance coverage and
thousands of additional bankruptcies could occur compared
to the 9/11 event, which was covered by the insurance
industry.”
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
38
Individual Risk Perception
Individual’s perception of risk not systematically
governed by probability of event:
Biases

Overestimation of Dread Risk, Unknown Risk
(Slovic, 1987)
Probabilities

Assessed with Availability Heuristic
(Tversky and Kahneman, 1973)
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
39
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
40
Country
Name of Insurance Scheme
Israel *
Victims of Enemy Action
Northern
Ireland*
Criminal Damage Compensation Scheme)
Spain *
Consorcio de Compensación de Seguros (CCS)
France
Gestion de l’assurance et la Réassurance des Risques Attentats
et Actes de Terrorisme (GAREAT)
South Africa *
South Africa Special Risks Insurance Association. (SASRIA))
Great Britain *
Pool Reinsurance Company
USA
Terrorism Risk Insurance Act
Germany
Extremus
*=founded before 2001
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
41
Data Set, Descriptive Statistics
Number of Policies Sold by Month
Month
1
2
3
4
5
6
7
8
9
10
11
12
Total (n=5611)
Mean
923.6
40.4
30
30.25
13
15.75
21
13.25
12
13.5
9
24.2
105.9
Std. Dev.
186.24
63.04
40.22
36.54
5.66
11.53
7.87
8.96
3.92
7.59
6.32
21.71
272.61
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
Min
599
9
2
5
5
9
11
7
7
6
3
10
2
Max
1054
153
101
84
17
33
30
26
16
23
19
62
1054
42
Data Set, Descriptive Statistics
Number of Policies Sold by Month
Month
Mean
Std. Dev.
Min
Max
1
923.6
186.24
599
1054
2
3
4
5
6
7
8
9
10
11
12
Total (n=5611)
40.4
30
30.25
13
15.75
21
13.25
12
13.5
9
24.2
105.9
63.04
40.22
36.54
5.66
11.53
7.87
8.96
3.92
7.59
6.32
21.71
272.61
9
2
5
5
9
11
7
7
6
3
10
2
153
101
84
17
33
30
26
16
23
19
62
1054
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
43
Terrorism

Terrorism is the premeditated use, or threat of
use, of extra-normal violence or brutality to gain
a political objective through intimidation or fear
against a targeted audience (United States
Department of State, 2000)
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
44
International Terrorist Attacks
1990-2003
600
500
400
300
200
100
0
0
9
19
2
9
19
4
9
19
6
9
19
8
9
19
0
0
20
2
0
20
Source: US Department of State
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
45
Extremus
Market Penetration in Germany
2003
2004
2005
2006
% of corporations that are customers of Extremus
< 1 bn €
2.9
2.6
> 5 bn €
21.7
15.0
% of total sums insured by Extremus
< 1 bn €
4.6
4.5
2.7
18.3
2.8
16.7
4.7
5.2
> 5 bn €
39.0
38.0
-
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
40.6
46
Private Public Partnership
„Omar, I think the boat is not going
straight“
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
47
Terrorism and Insurance
After September 11th, 2001 Terrorism has been excluded
from many standard insurance contracts
Terrorism poses significant problems for insurers:

Dynamic Uncertainty
 Asymmetric Distribution of Information between
Government and Private Sector
 Potential of Catastrophic Losses
Thomann / Graf von der Schulenburg / Gas / Pascalau ● August 6th 2007
48