Demographic Factors and Price Distortions in Insurance

• Author: Kathleen A. McCullough,David A. Macpherson,Ron Cheung,Charles Nyce,Cassandra R. Cole
• DOI: http://doi.org/10.1111/rmir.12027
• Published date: 01 March 2015
• Date: 01 March 2015

Risk Management and Insurance Review

C

Risk Management and Insurance Review, 2015, Vol.18, No. 1, 1-28

DOI: 10.1111/rmir.12027

DEMOGRAPHIC FACTORS AND PRICE DISTORTIONS

IN INSURANCE

Ron Cheung

Cassandra R. Cole

David A. Macpherson

Kathleen A. McCullough

Charles Nyce

ABSTRACT

Few papers have analyzed the potential linkages between price distortions and

the specific demographic and political traits of customers. The existence of price

distortions may have adverse and potentially unintended impacts on certain

demographic groups, leading to significant public policy concerns. The current

study uses census tract data and rating factors to examine age, income, and race

demographics in an effort to determine if any subgroups of the population are

adversely impacted by the price distortions of property insurance. The results

suggest that Hispanics, lower income households, and specific age groups pay

relatively more for insurance coverage than comparison groups.

INTRODUCTION

Price distortions occur when some groups of insureds are paying a higher premium rel-

ative to their expected losses than other groups. In some cases, this may stem from the

ratings factors used by insurers. For example, pooling individual risks into a homoge-

neous group based on a limited number of observable characteristics (rating variables)

may result in some specific risk factors being overlooked, as well as the misclassification

Ron Cheung is an Associate Professor of Economics at Oberlin College, Rice Hall 233, 10 N.

Professor Street, Oberlin, OH 44074; phone: 440-775-8971; fax: 440-775-6978; e-mail: rche-

ung@oberlin.edu. Cassandra R. Cole is the Robert L. Atkins Professor in Risk Management

and Insurance at Florida State University, 821 Academic Way, RBA 525, Tallahassee, FL 32306;

phone: 850-644-9283; fax: 850-644-4077; e-mail: ccole@cob.fsu.edu. David A. Macpherson is the

E.M. Stevens Professor of Economics at Trinity University, One Trinity Place, San Antonio, TX

78212; phone: 210-999-8112; fax: 210-999-7255; e-mail: David.MacPherson@trinity.edu. Kathleen

A. McCullough is an Assistant Department Chair and State Farm Professor of Risk Manage-

ment/Insurance at College of Business, Florida State University, 821 Academic Way, RBB 150,

Tallahassee,FL 32306; phone: 850-644-8358; fax: 850-644-4077; e-mail: kmccullough@cob.fsu.edu.

Charles Nyce is an Assistant Professor of Risk Management/Insurance at College of Business,

Florida State University,821 Academic Way,RBB 349, Tallahassee,FL 32306; phone: 850-645-8392;

fax: 850-645-8391; e-mail: cnyce@cob.fsu.edu.

1

2RISK MANAGEMENT AND INSURANCE REVIEW

of some individuals or properties. In other cases, regulatory or political pressure may

create a price distortion as regulators and legislators exert outside influence on the rating

process. While much has been written on the impact of these issues on affordability and

availability of coverage, the potential varying impact across the population is often not

incorporated. In many cases, this is due to data availability. However, it is an important

question given that the impact of price distortions may not be uniform across all sub-

groups of the population and price distortions may have a disparate impact on some

groups.

This study uses a unique and detailed data set to overcome some of the information

difficulties inherent in prior studies seeking to analyze issues of price distortion and

demographic factors. Specifically, we use policy-level information on the wind-only

policies from Citizens Property Insurance Corporation’s (Citizens) Coastal Account, the

state-run property insurer in the state of Florida, and census tract-level demographic

data.1This provides several advantages. First, we have detailed information on the

location of the policy and characteristics of the property. Second, we have modeled

average annual losses (AALs) and actual premium information for each insured loca-

tion. The combination of data allows us to isolate the measure of price distortion to

a single coverage, property coverage for wind. Additionally, the catastrophe exposure

of the account creates a great deal of potential variation, even in a small geographic

area, due to the differences in housing construction and distance to the coast. Further,

given the detailed information on the location of the properties and the extent of overlap

of the Coastal Account counties and census tracts, we are able to match the housing

and policy data to the sociodemographic characteristics of the population.2Collectively,

the data allow us to consider whether there is evidence of price distortions within

three particular demographics of the population (racial groups, age cohorts, and in-

come levels) while controlling for the potential impact of spatial autocorrelation. Taken

together, this provides a unique environment in which to test for evidence of price

distortion as it removes concerns identified in prior studies related to demographics

and insurance pricing. Specifically, we are able to remove moral hazard and adverse

selection problems by focusing on wind-only policies that provide a coverage required

by mortgage lenders. In addition, using a single provider removes the strategic use

of contract design concerns that may be correlated with hidden risk factors. Thus, by

1Citizens is the residual market property insurer in the state of Florida. Although designed as

a residual insurer, it is the leading property insurer in the state of Florida and has the largest

market share in several of the coastal areas used in this study.

2Policy-level demographic data related to the race, age, and income of policyholders were not

available from Citizens. Race information is not collected by property insurers in Florida, and

for privacy concerns of their policyholders, age and income are not part of the data set. For

this reason, matching data to the census tract level provides the closest available proxy for this

information. The data provide a reasonable means to test whether there are price distortions

that impact race, age, and income groups in different ways. Given the relatively high percent-

age of policies written by Citizens in these areas, we have a reasonable proxy for the price

discrepancies. Even with potential noise from the census tract data, we are still able to show

evidence of the fact that political actions may have unintended consequences related to price

distortions in insurance tied to factors such as age, income, and race rather than the underlying

risk.