Transcript Which Firm Gets Burned?

Which Firm Gets Burned?
Rachel J. Huang
Larry Y. Tzeng
Kili C. Wang
1
The main purpose of our
paper
• We intend to examine the theory of De
Meza and Webb (2001) directly.
Risk Probability
Self-Protection
Market Insurance
2
Asymmetric information:
theory
• Hidden Information
– Adverse selection
• Hidden Action
– Moral hazard
• Both of them predict that the higher
the coverage, the higher the
probability of risk.
3
Empirical evidence
• Many papers didn’t find the evidence
of adverse selection or moral hazard.
For example,
– Auto insurance: Chiappori and Salanie
(1997, 2000), Dionne, Gourieroux and
Vanasse (2001)
– Life insurance: Cawley and Philipson
(1999)
– Health insurance: Cardon and Hendel
(2001)
4
Some papers even found
that…
• A negative correlation between the
coverage of the insurance and the
occurrence of the risk
– Finkelstein and McGarry (2003): longterm care insurance
– Gronqvist (2004): dental care insurance
5
Another theory:
advantageous selection
• De Meza and Webb (2001)
– hidden risk preference
– A separating equilibrium:
• Timid individuals purchase market insurance
as well as make an effort to reduce the loss
probability, whereas bold individuals neither
purchase market insurance nor make an
effort to reduce the loss probability.
6
Make sure the negative
correlation is not reflecting
the underwriting ability…
Risk Probability
-
(+)
Self-Protection
(+)
+
()
Market Insurance
7
Our data set
• Target: Commercial Fire Insurance
• Through the Fire Department, we hand-collect
the data:
– market insurance
– self protection activities
– records of fire accidents
– other relevant variables on which insurance
companies usually collect data to underwrite
commercial fire insurance
• Observations: 2,592.
8
Are firms risk aversion?
• Most of the literatures focus on individual
insurance.
• Risk attitude of a firm:
– The classic finance literature: risk neutral
– Recent studies:
• The value function of a firm could be concave due to
– bankruptcy costs (Greenwald and Stiglitz, 1990;
Stiglitz, 2002)
– taxes (Eeckhoudt, Gollier, Schlesinger, 1997)
• Thus, a risk-neutral firm will behave as if it is risk
averse.
9
Why not from insurance
companies?
• Drawbacks to the data obtained from
insurance companies:
– Insurance companies do not have
information on the non-insured.
– Insurance companies may not have
information on individuals’ self
protection activities.
10
Self-protection variables
• The proxy must satisfy two criteria:
– A non-market risk reduction activity to reduce
the probability of a loss.
– This activity to reduce risk is unobservable to
insurers.
• Two proxy variables:
– the operations of self-defense fire
organizations
– the qualified operation of fire safety
equipment
11
Summary statistics
• 42.47% of the firms purchase commercial fire
insurance.
• 5.79% of the firms face a fire accidence in the
last year.
• 74.6% of firms have self-defense fire
organization operated in the last half year
• 75.3% of firms have qualified operation of fire
safety equipment in the last half year
12
Summary statistics:
business type
• Class-A places
13.46%
0.83%
24.48%
9.49%
30.55%
9.32%
11.87%
Movie projection places, singing room, club, restaurant, etc.
Bowling alley, indoor screen-type golf training field, game-playing ground etc.
Sightseeing hotel, restaurant, guesthouse, rest house (with bedrooms).
Store, market, department store, supermarket, retail market, exhibition field
Dining room, catering shop, coffee bar, tearoom
Hospital, sanitarium, long-term nursing home, health care institution, elderly service institution, etc.
Sauna, public baths
13
Summary statistics:
Location
806
804
807
812813
811
801
100
103
800
104
116
115
114
105
112
106
111
110
108
100
103
104
105
106
108
110
111
112
114
115
116
800
801
804
806
807
811
812
813
14
Empirical Methodology
and Results
15
Conditional dependency test
• Gourieroux, Monfort, Renault, and Trognon
(1987)
• Probit regressions:
Prob( yi  1)  ( X i  y )
(1)
Prob( zi  1)  ( X i  z )
(2)
Prob( si  1)  ( X i  s )
(3)
• Estimated residuals:
(Xi j )
(Xi j )
ˆi  E ( i | ji ) 
ji  (1  ji )
( X i  j )
 ( X i  j )
j
j
(4)
j  y, z , s
16
W statistic
n
W jk 
j k 2
ˆ
(  i ˆi )
i 1
~  (1)
2
n
j 2
k 2
ˆ
ˆ
(

)
(

 i i)
i 1
j , k  y, z , s
jk
• If W jk is significant, we reject
j
k
H0: cov(  i ,  i )  0
17
Conditional correlation
coefficient
• We predict that
Risk Probability
 yz  0
 zs  0
Self-Protection
 ys  0
Market Insurance
18
Table 3
Items
Panel A: Self-Defense Fire Organization as a Proxy forWSelf-Protection
jk
 jk
Market insurance vs. Loss occurrence
122.035***
-0.1252***
Self-protection vs. Loss occurrence
398.908***
-0.0475**
Self-protection vs. Market insurance
30.5324***
0.4426***
Panel B: Qualified Operation of Fire Safety Equipment as a Proxy for SelfProtection
Market insurance vs. Loss occurrence
122.035***
-0.1252***
Self-protection vs. Loss occurrence
460.523***
-0.3481***
Self-protection vs. Market insurance
10.1094***
0.2042***
19
4. Conclusion (1/2)
• Our paper contributes to the literature by
providing more direct evidence to support
the existence of the advantageous
selection theory derived by De Meza and
Webb (2001).
• We simultaneously examine the pair-wise
correlations among the purchase of market
insurance, the efforts made to engage in
self-protection, and the occurrence of
accidents.
20
4. Conclusion (2/2)
• We find that
– (1) firms that purchase market insurance have
a greater tendency to make an effort to
engage in self protection;
– (2) firms that make an effort to engage in self
protection are less likely to suffer a fire
accident;
– (3) firms with commercial fire insurance have a
lower chance of suffering a fire accident than
those without such insurance.
21
Thank you