The relationships between vehicle characteristics and automobile accidents

• Author: Wei‐Jin Wu,Sheng‐Chang Peng,Chu‐Shiu Li
• Published date: 01 December 2020
• DOI: http://doi.org/10.1111/rmir.12163
• Date: 01 December 2020

Received: 27 April 2020

|

Revised: 7 December 2020

|

Accepted: 8 December 2020

DOI: 10.1111/rmir.12163

FEATURE ARTICLE

The relationships between vehicle

characteristics and automobile accidents

Wei‐Jin Wu

1

|Chu‐Shiu Li

1

|Sheng‐Chang Peng

2

1

Department of Risk Management and

Insurance, National Kaohsiung

University of Science and Technology,

Kaohsiung, Taiwan

2

Department of Risk Management and

Insurance, Ming Chuan University,

Taipei, Taiwan

Correspondence

Chu‐Shiu Li, Department of Risk

Management and Insurance, National

Kaohsiung University of Science and

Technology, University Rd., Yanchao,

Kaohsiung City 824, Taiwan, R.O.C.

Email: chushiu.li@gmail.com

Funding information

Ministry of Science and Technology,

Taiwan, ROC, Grant/Award Number:

NSC 98‐2410‐H‐035 ‐025 ‐MY2

Abstract

In this study, the relationships between vehicle char-

acteristics and automobile accidents are examined based

on physical damage insurance claims data from a major

property insurance company in Taiwan. For the whole

sample, we find that the car characteristics index (an

estimate for the replacement cost associated with claims)

is negatively correlated with the claim probability and

positively correlated with claim severity. New cars and

sport utility vehicles (SUVs) (compared with Sedans) are

associated with higher claim probability. However, SUVs

(compared with Sedans) are associated with lower claim

severity. Small (compared with large) vehicles are asso-

ciated with lower claim probability and lower claim se-

verity. Compared with the imported cars from Germany,

policyholders with the imported cars from Japan or the

US are more likely to file a claim but to have lower claim

severity. On subgroup analyses, the relationship between

car characteristics index and claim probability is positive

for domestic cars and negative for imported cars, imply-

ing that imported car policies subsidize domestic car

policies. We also find the impacts of vehicle

Risk Manag Insur Rev. 2020;23:331–377. wileyonlinelibrary.com/journal/rmir

|

331

This is an open access article under the terms of the Creative Commons Attribution‐NonCommercial‐NoDerivs License, which permits

use and distribution in any medium, provided the original work is properly cited, the use is non‐commercial and no modifications or

adaptations are made.

© 2020 The Authors. Risk Management and Insurance Review published by Wiley Periodicals LLC on behalf of American Risk and

Insurance Association.

[Correction added on 21 January 2021, after first online publication: “R. J. Huang et al. (2014)”and “J. S. Huang and

Wang (2016)”have been changed to “Huang et al. (2014)”&“Huang and Wang (2016)”throughout the article.]

characteristics on claim probability and claim severity

differ for domestic cars and imported cars in subsamples.

Our empirical evidence provides important implications

for premium pricing in the automobile physical damage

insurance industry.

1|INTRODUCTION

Many studies have explored factors for predicting automobile accidents. One stream of lit-

erature examines the effects of information asymmetry on the behavior of physical damage

insurance or liability insurance policyholders. Such empirical studies tend to focus on the

problems of adverse selection, moral hazard, and advantageous selection.

1

In addition to de-

mographic factors (such as gender, age, marital status, claim coefficients), another stream of

literature focuses on risk classification factors for automobile accidents, such as biological,

psychological, and behavioral characteristics (Andersson, 2005; Bair et al., 2012; Brockett &

Golden, 2007; Gregersen & Berg, 1994; Iversen & Rundmo, 2002; Oltedal & Rundmo, 2006). In

contrast to the richness of studies on driving behavior related to asymmetric information or

biological, psychological, or behavioral characteristics, there is a sparsity of studies on the

impact of vehicle characteristics on the prediction of automobile accidents.

The purpose of this study is to examine the relationships among various vehicle char-

acteristics and automobile insurance claims. With technological progress and improvements,

new vehicle models and accessories are introduced to attract a broad spectrum of customers.

From previous research, vehicle purchases reflect drivers' personality traits and risk aversion

preferences. For example, Choo and Mokhtarian (2004) explore whether travel attitude, per-

sonality, lifestyle, and mobility factors affect the type of vehicle driven using data collected from

the surveys of randomly selected households in the San Francisco Bay Area in 1998.

2

They

show that people who choose identical vehicles tend to possess similar personality traits and

that travel attitudes, personality, and lifestyle are important factors in the choice of vehicle type.

Lee (2009) examines differences across brands of Korean cars by measuring the brand

personalities of five manufacturers in the Korean automobile industry representing two seg-

ments: functional utilitarian owners and upscale symbolic owners.

3

Lee (2009) concludes that

when making purchasing decisions consumers choose a brand that reflects their personality

and promotes their self‐concept. Erdem et al. (2010) indicate that educational level, risk tol-

erance, income, and attitude toward alternative energy influence the decision to purchase a

hybrid electric–petroleum vehicle (hybrid vehicle).

1

Early empirical studies on automobile insurance investigate the presence of adverse selection (Dionne et al., 2001; Puelz and Snow 1994) and moral hazard

(Abbring et al., 2003; De Meza and Webb, 2001; Dionne and Gagné, 2002; Li et al., 2007; Richaudeau 1999; Wang et al., 2008; Li et al., 2013b). However, adverse

selection and moral hazard may concurrently exist in the case of high risk insured with high insurance coverage. Thus, more recent empirical studies focus on

the issue of asymmetric information (Chiappori et al., 2006; Chiappori and Salanie, 2000; Cohen, 2005; Cohen and Siegelman, 2010; Saito, 2006; Wang, 2004).

In addition, some studies demonstrate advantageous selection in automobile insurance (Bair et al., 2012; Li et al., 2013a).

2

In the study by Choo and Mokhtarian (2004), there are nine vehicle categories: small, compact, midsize, large, luxury, sports, minivan/van, pickup, and sport

utility vehicle (SUV).

3

Brand personality is measured using 17 items as proposed by Aaker (1997). There are five automobile manufacturers in Korea: Hyundai, Kia, Daewoo,

Samsung, and SsangYong.

332

|

WU ET AL.

Psychology is increasingly applied to the study of risk behavior and traffic safety, especially

the role of personality in accidents. Gregersen and Berg (1994) find that lifestyle characteristics

are highly associated with road traffic accidents. Iversen and Rundmo (2002) show that those

who score high on questions related to sensation seeking, normlessness, and driver anger are

more likely to be involved in accidents.

4

Brockett and Golden (2007) provide a complete and

detailed survey of the literature related to the classification of risk factors for both risky driving

and financial decision making based on biological, psychological, and behavioral character-

istics. They conclude that credit scores provide information that is distinctive from demo-

graphic factors (such as gender, age, marital status in automobile insurance), such as an

individual's biopsychological characteristics, which yields useful underwriting information

about the risk of automobile accidents.

5

Huang et al. (2014) also find that accident probability is

heterogeneous with respect to vehicle characteristics such as car age and car size.

Why do vehicle characteristics affect automobile accidents? On the basis of a theory put

forth by Brockett and Golden (2007), vehicle types reveal information concerning an in-

dividual's personality (Choo & Mokhtarian, 2004; Erdem et al., 2010; Lee 2009) and psycho-

logical profile. This, in turn, provides data on risk‐taking behavior associated with driving an

automobile (Bair et al., 2012; Gregersen & Berg, 1994; Iversen & Rundmo, 2002).

The aim of this study is to explore the associations among various vehicle characteristics

and automobile claims including the claim probability and claim severity from physical damage

insurance policies. In Taiwan, the car characteristics index and the insured factors

6

are im-

portant factors in the official automobile premium formula for physical damage insurance. The

official car characteristics index refers to a weighted average of replacement costs that reflect

the car characteristic when policyholders file claims. The car characteristics include type,

brand, size, age, number of seats, and country origins. In general, a higher replacement cost is

related to a higher car characteristics index and a higher insurance premium. However, drivers

who own vehicles with high replacement costs are more likely to maintain prudent driving

behavior to avoid property loss.

7

To the best of our knowledge, there are no studies that analyze

the effects of car characteristics index on automobile claims. Therefore, we first examine the

association between the official car characteristics index and vehicle damage insurance claims

to understand whether the current car characteristics index is in line with the principle of fair

insurance premium pricing. We also examine domestic cars versus imported cars. Since the car

characteristics index is an estimate of replacement cost associated with filed claimed, we

further explore whether vehicle characteristics affect claim probability or claim severity. In

addition, we analyze the interactions between new car and vehicle characteristics.

For the whole sample, we find that the car characteristics index is negatively correlated with

claim probability and positively correlated with claim severity. This is inconsistent with the

“fairness principle”

8

as it leads to advantageous selection for the insurance company. Higher

4

Iversen and Rundmo (2002) use data from a self‐completed questionnaire from a survey carried out in 2000–2001 among a sample of Norwegian drivers

randomly selected from the driver's license register.

5

Bair et al. (2012) provide further support for Brockett and Golden's (2007) hypothesis by demonstrating that people who are more responsible in terms of

vehicle maintenance are also more responsible in terms of driving behavior, even after adjusting for standard underwriting variables. Using previous years'

maintenance records, they study kilometers, the time between each maintenance, and items maintained. Their empirical evidence shows that proper vehicle

maintenance, measured in kilometers, and recommended maintenance items are negatively correlated with loss probability.

6

This paper does not focus on risk factors that serve as control variables.

7

We thank the referee who provided this comment.

8

In general, insurance companies establish an insurance premium rate system by offering lower premium rates to low‐risk policyholders and higher premium

rates to high‐risk policyholders.

WU ET AL.

|

333