Western Risk and Insurance Association

INSURABLE LOSSES: SYSTEMATIC OR UNSYSTEMATIC RISK?

Author(s): Dongsae Cho
Source: The Journal of Insurance Issues and Practices, Vol. 10, No. 1 (January 1987), pp. 1-
12
Published by: Western Risk and Insurance Association
Stable URL: http://www.jstor.org/stable/41943214
Accessed: 23-06-2016 07:06 UTC

REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/41943214?seq=1&cid=pdf-reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact [email protected].

Western Risk and Insurance Association is collaborating with JSTOR to digitize, preserve and extend
access to The Journal of Insurance Issues and Practices

This content downloaded from 132.236.27.217 on Thu, 23 Jun 2016 07:06:54 UTC
All use subject to http://about.jstor.org/terms
 INSURABLE LOSSES: SYSTEMATIC OR UNSYSTEMATIC RISK?

Dongsae Cho

ABSTRACT

This paper examines accounting beta estimates for individual lines of insurance
and some correlation coefficients to help apply portfolio theory and the CAPM to the
insurance area. The study result (1) confirms different degrees of diversification
effects among insurance lines and (2) indicates that insurable losses may be either
systematic or unsystematic depending upon the line of insurance under consideration.

INTRODUCTION

The modern portfolio theory and the Capital Asset Pricing Model have been widely
suggested for some insurer decisions such as (1) capacity decisions (Doherty [5]),
(2) underwriting and investment decisions (Hammond and Shilling [9], and Kahane and
Nye [12], and (3) regulation of insurance premium rates (Cummins and Harrington [4],
Fair ley [8], Haugen and Kroncke [10], and Hill [11].

Portfolio theory pursues a mean-variance efficient set of investment assets,

given each component asset's risk- return characteristics. The mean and variance of a
portfolio of investment assets are

/V n
E(Rp) = E XiEÍRi) (1)
i=l

n n n

0 P = + ^

where E(Rp) is expected return on the portfolio,

n is number of investment assets in the portfolio,

Xļ is fraction of the investor's fund put in the i**1 asset,

A ✓ f ļļ
E(Rļ) is expected return on the i asset,

(Tp is variance of returns on the portfolio,

?th is?th
variance of returns
is variance ononthe
of returns theii asset, and
th th
Pļj is correlation coefficient between returns on the i and j assets.

*Dr. Cho is Assistant Professor of Insurance at the University of Minnesota. He

received his Ph.D. from the University of Illinois at Urbana-Champaign. He has
published articles in this Journal, the Journal of Risk and Insurance, and Genea
Papers on Risk and Insurance.

This content downloaded from 132.236.27.217 on Thu, 23 Jun 2016 07:06:54 UTC
All use subject to http://about.jstor.org/terms
Equation (2) indicates that some correlation coefficients play a significant role in
applying portfolio theory to insurer decisions.

The CAPM, rooted in portfolio theory, relates the expected return on an

investment asset to expected excess return on the market portfolio.

E($j) = Rf + /?j[E(Rm) - Rf] (3)

The beta coefficient in Equation (3) measures the nondiversifiable portion of the
security's risk or its systematic risk. The diversifiable portion of the risk or its
unsystematic risk could be eliminated as a large number of securities is added to the
investor's portfolio. The issue is whether insurable losses are diversifiable or
not. Some argue that insurable losses are unsystematic risk independent of market
movements [1, 2, 3, 6, 7, 14, 17, 18, 19] ; others contend that they are systematic
risk [15, 16].

Knowing whether insurable losses are systematic or unsystematic is important

because of the following reasons. Vhen they are systematic the insurance decision of
a corporation is relevant to the equity value of the firm, and also the insurer
should charge their nondiversifiable risk to the corresponding insurance products.
If they are unsystematic and therefore diversifiable, however, corporate insurance
purchasing at the actuarially fair price does not affect its equilibrium stock price,
and the insurer may ignore their risk characteristics in pricing insurance contracts.

This paper attempts (1) to estimate beta coefficient for each line of insurance
to determine if insurable losses are systematic or unsystematic, and (2) to measure
loss correlations among insurance lines to help apply portfolio theory to insurance.
Although overall underwriting betas are examined in detail in their work, Cummins and
Harrington [4] have not looked at underwriting betas for individual lines of
insurance, and also have not pursued the useful application of loss correlations to
insurance decisions.

LITERATURE REVIEW

To apply portfolio theory to insurance, some correlation coefficients should be

computed. Mayers and Smith [18] claim that insurance decisions are an integral part
of overall protfolio decisions since the risk-return characteristics of overall
investments are jointly determined by those of (1) marketable assets, e.g., stocks
and bonds, (2) nonmarketable assets, e.g., human capital, and (3) losses from pure
risks. In the framework of portfolio theory, therefore, the insurance decision of a
corporation should be made based upon, among others, correlation coefficients among
all possible combinations of these three sets of assets.

Doherty [5] inspects insurer capacity in the context of portfolio theory.

Insurer underwriting capacity is evaluated after considering risk-return
characteristics of (1) an additional insurance policy under study, (2) the existing
underwriting portfolio of the insurer, and (3) their investment portfolio. The means
and variances of the return on the total portfolio of the insurer, before and after
the additional policy in the portfolio, are then calculated for insurer underwriting
decisions.

Hammond and Shilling [9] treat insurance lines as individual investment assets
in seeking the optimum mix of insurance business for an insurer. The standard
deviation of its overall underwriting returns are calculated based upon (1) the

This content downloaded from 132.236.27.217 on Thu, 23 Jun 2016 07:06:54 UTC
All use subject to http://about.jstor.org/terms
standard deviations of returns on component underwriting lines, and (2) return
correlation between each pair of insurance lines. Kahane and Nye [12] demonstrate
diversification effects of underwriting and investment activities of an insurer by
using correlations among their annual returns.

Benston and Smith [1] and Mayers and Smith [17, 18] argue that losses from
insurable - pure risks are specific to the firm, so that they could be diversified
away. Meyer and Power [19] further claim that the risk from insurable losses could
be removed without insurance through portfolio diversification. Main [15, 16]
contends that corporate insurance purchasing at the actuarially fair premium has no
value to stockholders. Also Cummins [2] uses the CAPM for some risk management
decisions of a firm such as the optimum retention level. Since a deductible amount
has no impact on the correlation between firm and market returns under certain
conditions, losses covered by deductible insurance are considered to be
diversifiable.

On the other hand, Main [15, 16] argues that some insurance whose payoffs are
correlated with market returns is nondiversifiable, reducing the systematic risk of a
corporation. If a loading charge is calculated as the market price of risk times
reduction in such systematic risk, however, this excess charge over the actuarially
fair premium renders insurance a matter of no concern to corporate investors.

METHODOLOGY

Vriters have used the CAPM and APT in exploring equilibrium insurance pricing
models for property and liability insurance to arrive at competitive insurance
premiums and fair underwriting profits [4, 8, 11, 13]. Such fair price or just
profit rate must be predicted for each line of property and liability insurance
business to be applicable to the real world.

The problem, however, is that most insurers are involved in more than one line,
making it impossible to estimate market betas for individual lines of insurance.
Furthermore insurer overall beta either market beta determined in a competitive
market (Hill [11]) or underwriting beta used by Cummins and Harrington [4], and
Fairley [8]) does not serve this purpose either. Only Hill [11] attempts to measure
accounting betas for individual lines with some accounting data (i.e., profit rates
disaggregated by lines accumulated through Best's Aggregates and Averages'). In this
study, therefore, the Hill's approach is used to estimate accounting betas for all
lines of insurance.

In order to measure accounting betas for individual lines of insurance, the

following two regression models are run (Hill [11]). Model 1 is selected under the
assumption that performance fluctuations of the stock market affects insurer
underwriting results immediately. Model 2 is tested because good or bad economic
conditions may affect the insurance industry with some time lags.

Model 1: Rļt - <4 + + eļt (4)

Modelī! *lt -«i + ii«., ♦ «lt ♦#31«.(t_2 ♦ «it (5)
where Rj is underwriting profit rate (1-loss ratio - expense ratio) for the 1 line
of insurance during period t,

/9 j is accounting beta for the i^ line of insurance,

3

This content downloaded from 132.236.27.217 on Thu, 23 Jun 2016 07:06:54 UTC
All use subject to http://about.jstor.org/terms
 R is return on the market portfolio during time t, and

e^t is error term.

Insurer underwriting profits seem to follow a systematic pattern over a
relatively long period of time, e.g., 10 years. This pattern, called underwriting
cycle, is already contained in the model because it uses underwriting profit rate as
its dependent variable, and as a result using time-series data would not create any
problem.

Instead of return correlation between each pair of insurance lines, loss

correlations are computed and tested to examine diversification effect because of the
following three reasons. * The first reason is that there is little consensus as to
how to allocate insurer expenses among individual lines (such allocation is necessary
to compute returns on separate lines). The second reason is that an expense ratio
for the insurance industry has little to do with individual insurers, if scale
economies or diseconomies are present. The third reason is that loss trends of
individual lines would also give an idea with respect to the degree of their
respective diversification effects.

THE DATA

To achieve the objective of the study, the data listed below are collected on an
annual basis. ^

1. Stock price indexes - composite (500 stocks) as a proxy of the market

portfolio.
2. cost of living indexes - all items.
3. Per capita income in real dollars.
4. U.S. population.
5. Life insurance and annuity benefit payments (i.e., death benefits, matured
endowments, and annuity payments).
6. Health insurance benefit payments including both group and industrial
health insurance, 3 and disability payments.
7. Fire losses.
8. Homeowners' multiperil losses.
9. Commercial multiperil losses.
10. Ocean marine losses.
11. Inland marine losses.

^Although this argument may also applyto accounting betas in Equations (4) and
(5), returns are used instead to estimate betas since to alternative to them is
available.

^Accumulating quarterly instead of annual data would be desirable to produce

more reliable study results. However, life and health loss data are unavailable on a
quarterly basis, although limited property and liability loss data could be obtained
from A. M. Best Company. For this reason the study uses only annual loss data.

^Health insurance benefits include hospital, surgical, and other medical

expenses, and cost of dental care, excluding disability income and accidental death
and dismemberment payments.

This content downloaded from 132.236.27.217 on Thu, 23 Jun 2016 07:06:54 UTC
All use subject to http://about.jstor.org/terms
 12. Vorkers' compensation losses.
13. Burglary and robbery losses.

The term Rmt

„in v
(4)' v ' is calculated as

SPt+l - SPt (6)

SPt

where SP is stock price index. Data items (1) and (2) are accumulated from Standard
and Poor's Statistical Service. Security Price Index Record. 1984 Edition. Item
(3)is collected from Statistical Abstract of the United States: 1985. 105th Edition,
U.S. Bureau of the Census, Washington, D.C., 1983, and Historical Statistics of the
United States. Colonial Times to 1970. U.S. Department of Commerce. Item (4) is
taken from Statistical Abstract of the United States: 1985. 105th Edition. Data
items (5)and (6) are accumulated from Life Insurance Fact Book. American Council of
Life Insurance, various years. Items (7) - (130 are collected from Best's Aggregates
and Averages. Property - Casualty. A. M. Best Company, various years. Data items (7)
- (13) include only aggregate insured losses reimbursed by U.S. stock and mutual
insurers, excluding other forms of insurers such as reciprocals, and alien insurers
due to data unavailability. For data items (7), (10), (11), (12), and (13),
aggregate data are taken over the period of 1951-1983, while shorter periods are
selected for (8)and (9)due to data unavailability. Losses to automobiles and from
auto accidents are excluded not only because the sample size is insufficient but also
because only a small segment of such aggregation is available, e.g., only losses paid
by stock insurers.

To increase the degree of homogeneity of these time-series data, they are

adjusted for (1) growth of population, (2) cost of living indexes- all items, and (3)
per capita income in real dollars. Since property and liability, and life and health
losses in aggregate terms are closely related to U.S. population, insured losses from
all lines are adjusted for the growth of population. Because most life insurance
benefits (e.g., death benefits, matured endowments, and annuity payments) would
remain fixed regardless of changes in CPI,^ they are not modified for CPI. This
study modifies disability income payments for CPI, since insurers usually promise to
reimburse disabled workers a certain percentage of their salaries immediately before
disabilities. Also health insurance benefits are adjusted for CPI because health
care costs tend to move together with it. All property and liability losses, i.e.,
data items (7) - (13), are also modified for changes in such indexes because the
value of real properties, and inventories, and workers' salaries are likely to be
highly correlated with price levels. Finally losses from all lines of insurance are
adjusted for real economic growth by per capita income in real dollars.

HYPOTHESES TESTING

Two sets of hypotheses are tested in this study: (1) accounting beta
coefficients for some lines of property-liability insurance, and (2) correlation
coefficients among insured losses. The null and alternative hypotheses are

^Benefit payments from some life insurance instruments are affected by general
market movements such as variable and universal life insurance, and variable
annuities. However, these products constitute only a small proportion of the life
and annuity market in the U.S.

This content downloaded from 132.236.27.217 on Thu, 23 Jun 2016 07:06:54 UTC
All use subject to http://about.jstor.org/terms
 Hjq! Underwriting betas for all insurance lines are zero.
HjjJ Underwriting betas for all insurance lines are different from zero.
H2Q: Loss correlations among different lines of insurance are zero.
HjļS Loss correlations among different lines of insurance are different from
zero.

A t-test is performed for hypotheses H.q and In order to test hypotheses

H»n and H.., a z statistic could be used by assuming tnat the term
(T/2)ln[ [f i+/»ļj)/(l-Pjj)l is normally distributed with its mean of
(l/2)ln{[(l+Pjj)/(l-Pļj)], and variance of l/(n-3). Then
1+? 1+0
(l/2)ln(_ii) - (l/2)ln(i±Si) 1+0
TT7 1-0
ij (7)
2 *

EMPIRICAL RESULTS AND IMPLICATIONS

Beta coefficients and their standard errors for individual lines disintegrated
by stock and mutual insurers are shown in Table I; loss correlations are presented in
Table II. Table I shows that although accounting beta coefficients for Model 1 are
not significantly different from zero except for fire insurance handled by mutual
insurers, more accounting beta coefficients for Model 2 are significantly different
from zero. The result implies that the importance of corporate insurance purchasing
depends upon the type of insurance lines. For systematic insurable losses , the
insurance decision of a firm is relevant to its value, and also a rational insurer
should charge this systematic risk to the corresponding insurance contracts.

Statistics from Table II indicate that generally insurable losses are highly and
significantly correlated among themselves. Although exactly the opposite events are
insured against by annuity contracts and life insurance» a positive correlation
between them (.73) makes a lot of sense because they cover different sets of
insureds. Generally health insurance is positively correlated with the others
because good economic conditions would lead to (1) improved employment rates, (2)
increased demand for insurance in general, and (3) more group health coverages. The
same reasoning also applies to other correlations (e.g., ocean marine versus
disability income insurance).

Annuity and commercial multiperil are mostly negatively correlated with other
lines, and positively correlated with each other at the 99 percent confidence
interval. Among the insurance lines with positive correlations, some are very highly
correlated with other lines (death benefits, endowment payments, disability income
payments, ocean marine, inland marine, and workers' compensation), whereas some
others have nearly zero correlations (homeowners' multiperil).

This empirical result may be interpreted in the context of portfolio theory.

Annuity and commercial multiperil have extremely powerful diversification effects,
creating underwriting capacity for other lines of insurance. Insurance lines with

This content downloaded from 132.236.27.217 on Thu, 23 Jun 2016 07:06:54 UTC
All use subject to http://about.jstor.org/terms
high positive correlations reduce their capacity substantially, but homeowners'
multiperil has negligible diversification effect with little impact on the
underwriting capacity for the insurer.

SUMMARY AND CONCLUSION

Underwriting beta coefficients and loss correlations are hypothesized and tested
in this paper to apply portfolio theory and the CAPM to insurance. Generally
correlations between pairs of insured loss groups are significantly different from
zero; the mixed results from the significance test of accounting beta estimates
indicate that insurable losses may be either systematic or unsystematic.

The results of the study have implications for the insurance industry as well as
for risk managers. Correlation analyses are useful to the insurance industry because
they show differing degrees of diversification effects of individual lines on the
limited underwriting capacity of insurers. They also maybe used by risk managers for
retention decisions in an inefficient insurance market. Some nonzero underwriting
betas imply that a risk manager can increase the value of a firm through a good
insurance decision in an inefficient insurance market. They also suggest that the
insurance industry should charge an insurance premium based upon the systematic risk
of the insurable losses under consideration rather than based upon their actuarial
cost.

This content downloaded from 132.236.27.217 on Thu, 23 Jun 2016 07:06:54 UTC
All use subject to http://about.jstor.org/terms
 DEFINITION OF TERMS

DTH is death benefits,

ENDV is endowment payments,

DISA is disability income payments,

ANTY is annuity payments,

HI is reimbursement of medical expenses, excluding disability income payments,

FIRE is fire losses,

HOME is homeowners' multiperil losses,

COMM is commercial multiperil losses,

OCN is ocean marine losses,

INLD is inland marine losses,

VKRS is workers' compensation losses, and

BGLY is burglary and robbery losses.

This content downloaded from 132.236.27.217 on Thu, 23 Jun 2016 07:06:54 UTC
All use subject to http://about.jstor.org/terms
 *
* * o
*
OK z
*n o o
2
>-« 3
ms « on X Tļ ZTr /
* *■ P" 3? r Z 2 Z 1? Uì 3 3 (fl
< (fl O 3mm C íB C rr
- 1 rr 0
tn en a) o# o c n
ro Z' 3 ** * X
o > vû »û Kj ^ n ►-
3 3 • • IÜ 0
(I M- - ^
in • n m h h
n n o J3 TT

"n flj cu ^ c ib ro ^ i ^ ^ ^
3 3 Tili •• •• •• •• •• •• •• Q-)
vû fr fr (B rr < O O O GO h» ** »■* O O O O O rr
r til 1 Hi O »* O K) vi O »-vi vi 00 O O 00 řO 0# 3
Iflifli 13 0 H ^ M xi O n* <1 Ul W *0 »- vi ©vi O >0 rr 0
iB rr rr 03T3C Ul ÜIO O O N) ■* <0 vi 4* O O vi 01 a
3 Q IB w w w w w • (ļj
•* rr rr a "1 01 »-
OTT •* ** rr (/) Z
IM 3 (ji (0 ^ Oí 0 ►-
*0 w CL 3 5 • (fi
rr *Q *Q ' • W W 0-4 W řO M W O rr
T >Û ül T) WW ř0 ÍO vi vi ro ►- Q 0
<B o cr au m -K n
XJ T3 Ul IB ^ m jT n
Q. IB (B 01 p rr (D n
(D 1 1 IB Oí 3 ^ o
<* n n ai m m rr ^ i ^ ^ j ^ ' ^ ^ j 3 c
IB ÍD c T •••• • i •••• •• •• (ļļ m 3
3 3 3 ai 3 IB WW >- W *• W N) IO N) Ul ^ O ^ O rr C rr
*•• rr rr "1 «* Ul *0 4* O •- »>4 ^ *-* >- O »>4 OK) 00 fM 3 T
rr ID 0 0 ID 01 Ul Ih ^ (J' & O Ul Ul 00 00 GO Ul^ rr 0 ID 3
-• n n «* i oi Ul >4 u un w to <*) *-» -u ^ ro ro oí a i 0
0 0 0 Q. ^ ^ w . ro oi
3 3 3 - rr CT ro ¿u oi
-* -* < T 3 ID o) Z
0 **• m- ib íB 0 ** CU 0 ro ÍB
rrS *
*
-h a a i oí a o ■ 3 .
ft íB Ol (B IB X 04 0-4 C4 C-4 W f-0 <A D * m
rr 3 3 >" n W Kj N) v4 vi K) »- 0 m
(B n n -h n t q -h ai ^
1 (B (B O iO Ul rr
3 Ol (B NJ rr

i/l M- T -H fl|
3
* 3 3 *■* ^ ^ ^ ^ /s ^
Oí
it ft H h- j| (j •• ••■ • « •• i i * * y} cr
iB fB • n 01 ** *+ O O O O ** O h* O h* »■* o rr ib
1 *1 »-• (B rr vi (.H vj vi GD 04 LH O 0) QD ÛD 0» 3 01
<< IB ODOW»- v<43 N) GD ^ O 0DU1 CH ^ rr O
*
Qf 0i 3 rr 3 ^0^0 v4 ^ ■£* 00 O 04 O O 0 v^ r'j vi 01 Cl
rr 3T CU ^ ^ ^ ^ w w ^ ¡ķ m ro *
Ol II rr »-* *

« en z
it mm o» o ^
ü C ā • 3
C 3 0¡ W W W U NJ W Oi O C
0»1#T W W íO M Nj >J TO »-0 rr
►•*10 iīi -h r
» i» • »
0 y
Xl ^ Ä ^ ^ I ^ ^ H
N . M* •• •• •• •• •• •• •• (/) 3
IB *•• 1 WW ' O W *+ *+ WW W »- ro vi ^ C.0 rr oí
1 <• (si ^4 00 LH ^ Ul IO Ul O 00 ro 00 W O OÍ 3 C
0 Ol W O N) W W Ul W W W 00 LH 00 rr O 1
Tjl rr Ul O W W -0 vi 00 >0 O ro M 00 CO tfi a íB
3MÜi ^ ^ ^ w w ^ ^ # fB "1
m ^'3 * * H-m
CU « Cl

^
lit. 5
Di1 . en
1 a,z o w
3 a W W W 04 NJ i.) 3 •
U W NJ KJ>4 Ni KJ D
2T I» o
t rr J*
rr w w ^
~~ y)10 *
o
**. t

rr m. oí
01 3 •

This content downloaded from 132.236.27.217 on Thu, 23 Jun 2016 07:06:54 UTC
All use subject to http://about.jstor.org/terms
 * * * s^gnxTixsomo
* 33 £ z z z zi -i œ o z
en cr» en 10 ° 3 m m < » *
ro »- »
(0 tO 'Q
33 m
rr m. m. . m y
3" -* •* >o f7
* 3" ä -* ñ •* <•', >o f f7
"H SU fl#
- 3 3 ** *
*
<û n- ri- 0

m
m řO
řO?»■
(+ (+ ï-
rr Orr
■*i iI>
5
■h r+ rf î Í
0 3" T *
1 í I ^
î 5 -o 01 ,Ü OD La ^ H
J ro -o 01 KJ o <A <
ro _ „ ♦ * *
«
« Q. m ÍÜ « ffl Q. m ÍÜ « * * *
a> 1 1 X
^ n n « • « « M
? 2 2 m o w ní
5. 1,
rP
►-(T
O W £
|>4 n
rf £ Ī 0
1
*-• n n * "i
0 0 0 I íD
3 3 3 • • • • » *jļ
-e« ^4 o -o « Oí
2. ñ « ui w co o» .* 35 rf
S"« « ♦♦♦♦♦m H»'

_
_ »T J J 2 J ? * * * * * 3
0
« n n tn
1 m ib
3Iß**"
**"**'
X
...... Q
cr
(D
* £. i o o tn ►- w en 2 ri-
S
5 ~ 2 * ♦ * IB
~ ^ * * *
< < * * * ro
D
2L
^ M
^ Sí.
• •
M
• •
11
i »
iii
• Q Tl I>
-i

■* ņj *- oo tn tn m o CU lil
vi O ►- C.1 vj C.1 M z r
* * * * * 2 1 m
* * * * ui
H-#

0 ~
I I
• •••••••q
■&w«ootn>o*ooon 1-4

en ►- ■* o o tn o» en z 3
* ****** U1
****** r

fD
i i a
* • • • i « i • « H
«0tfl»~*0Ultfl<0«0(DZ r
NiM'Ovioaj^^otnr o
♦ ♦ ****** o ai
* * ****** in
0)
m
i i *
•••••• i • i i C *
-O *0 C-4 0-4 *0 0* O ^3 *0 :? *
<-A G* M *0 o O O C'J en •- iņ
* * *******m
* * * * * * *

i i
* • 1 •••••••• Ijj
"0*0>0viO>0.£O>0«0vlQ
"V >» •* o . œ -û m so tn m >0 r
* * * * ******<
♦ * * * * ****

10

This content downloaded from 132.236.27.217 on Thu, 23 Jun 2016 07:06:54 UTC
All use subject to http://about.jstor.org/terms
 REFERENCES

1. Benston, G.J., and C.W. Smith, Jr., "A Transaction Cost Approach to the Theory
of Financial Intermediation," The Journal of Finance. Volume XXXI No. 2 (May
1976), pp. 215-231, Papers and Proceedings of the Thirty-fourth Annual Meeting
of the American Finance Association, Dallas, Texas, December 28-30, 1975.

2. Cummins, J. David, "Risk Management and the Theory of the Firm," The Journal of
Risk and Insurance. Volume XLIII No. 4 (December 1976), pp. 587-609.

3. Cummins, J. David, "Risk Management and the Theory of the Firm: Author's
Reply," The Journal of Risk and Insurance. Volume L No. 1 (March 1983), pp. 145-
150.

4. Cummins, J.D., and S.E. Harrington, "Property-Liability Insurance Rate

Regulation: Estimation of Underwriting Betas using Quarterly Profit Data," he
Journal of Risk and Insurance. Volume LII No. 1 (March 1985), pp. 16-43.

5. Doherty, Neil A., "A Portfolio Theory of Insurance Capacity," The Journal of
Risk and Insurance. Volume XLVII No. 3 (September 1980), pp. 405-420.

6. Doherty, N.A. , and S.M. Tinic, "Reinsurance under Conditions of Capital Market
Equilibrium: A Note," The Journal of Finance. Volume XXXVI No. 4 (September
1981), pp. 949-953.

7. Doherty, Neil A., "Portfolio Efficient Insurance Buying Strategies," The Journal
of Risk and Insurance. Volume LI No. 2 (June 1984), pp. 205-224.

8. Fairley, William B., "Investment Income and Profit Margins in Property-Liability

Insurance: Theory and Empirical Results," The Bell Journal of Economics. Volume
10 No. 1 (Spring 1979), pp. 192-210.

9. Hammond, J.D., and N. Shilling, "Some Relationships of Portfolio Theory to the

Regulation of Insurer Solidity," The Journal of Risk and Insurance. Volume XLV
No. 3 (September 1978), pp. 377-400.

10. Haugen, R. A., and C. 0. Kroncke, "Rate Regulation and the Cost of Capital in
the Insurance Industry," Journal of Financial and Quantitative Analysis. Volume
VI No. 5 (December 1971), pp. 1283-1305.

11. Hill, Raymond D., "Profit Regulation in Property-Liability Insurance," The Bell
Journal of Economics. Volume 10 No. 1 (Spring 1979), pp. 172-191.

12. Kahane, Y., and D. Nye, "A Portfolio Approach to the Property-Liability
Insurance Industry," The Journal of Risk and Insurance. Volume XLII No. 4
(December 1975), pp. 579-598.

13. Kraus, A., and S.A. Ross, "The Determination of Fair Profits for the Property-
Liability Insurance Firm," The Journal of Finance. Volume XXXVII No. 4
(September 1982), pp. 1015-1028.

14. Main, Brian G. M., "The Firm's Insurance Decisions: Some Questions Raised by
the Capital Asset Pricing Model," Managerial and Decision Economics. Volume 3
No. 1 (March 1982), pp. 7-15.

11

This content downloaded from 132.236.27.217 on Thu, 23 Jun 2016 07:06:54 UTC
All use subject to http://about.jstor.org/terms
15. Main, Brian G. M., "Risk Management and the Theory of the Firm: Comment," The
Journal of Risk and Insurance. Volume L No. 1 (March 1983), pp. 140-144.

16. Main, Brian G. M., "Corporate Insurance Purchases and Taxes," The Journal of
Risk and Insurance. Volume L No. 2 (June 1983), pp. 197-223.

17. Mayers, D., and C. W. Smith, Jr., "Contractual Provisions, Organizational

Structure, and Conflict Control in Insurance Markets," Journal of Business.
Volume 54 No. 3 (July 1981), pp. 407-433.

18. Mayers, D. , and C. W. Smith, Jr., "The Interdependence of Individual Portfolio

Decisions and the Demand for Insurance," Journal of Political Economy. Volume 91
No. 2 (April 1983), pp. 304-311.

19. Meyer, R. L. , and P. B. Power, "The Investment Value of Corporate Insurance,"

The Journal of Risk and Insurance. Volume L No. 1 (March 1983), pp. 151-156.

12

This content downloaded from 132.236.27.217 on Thu, 23 Jun 2016 07:06:54 UTC
All use subject to http://about.jstor.org/terms