Nanyang Business School Forum on Risk Management and Insurance

Testing for information asymmetry in liability coverage in China’s automobile insurance market

by | Sep 13, 2018 | Adverse Selection, Economics, Market, Non-life insurance | 0 comments

Tags: information asymmetry, adverse selection, auto insurance, liability coverage, China, IRFRC
More from: Yi Yao, Yinglu Deng, Hao Zheng

Editor’s Note: Posted by Yi Yao, Associate professor at Department of Risk Management and Insurance, School of Economics, Peking University; Yinglu Deng, Assistant professor at PBC School of Finance Tsinghua University; Hao Zheng, PhD candidate at Department of Risk Management and Insurance, School of Economics, Peking University

Information asymmetry is an important problem in insurance market. It has important and complicate implications on pricing, contract design and regulation. Since Akerlof’s theoretic work on this question and get a prediction on a positive relationship between risk and coverage, there are a lot of empirical research to find the evidence to prove this result especially in automobile insurance market. However, scholars from the world get different results. No matter which county their data is from and what the result is, most of those papers pay attention only on different characteristics of insureds (for example, newly insureds, experienced drivers etc.), and less frequently on different types of claims. However, it is unreasonable to treat different kinds of accidents as the same, such as accident with bodily injury and accident with only property damage. Intuitively, the former one has a higher claim amount but a smaller frequency than the second one, and this may has lead different performance of insurance contract on information asymmetry.

To prove this, we make use of a unique and complete dataset from China with detailed claim information, which enable us to test for information asymmetry across different lines of coverage, i.e. claims for liability coverage versus physical damage (collision) coverage, and to further compare degree of information asymmetry across different types of liability claims, i.e. those with bodily injury versus those with only property damage.

Liability coverage in automobile insurance is an important layer of protection given that the potential magnitude of loss is typically much larger than property itself. In addition, liability claims with bodily injury typically represent a large proportion of claim payments (in our sample, the claim payments due to those with bodily injury cases count for 57% of claim amounts). Yet, the probability of having a liability claim resulting from bodily injury is relatively small in general (0.93% in our sample). In China, the pricing of automobile insurance is based on an almost identical set of variables for both physical damage (collision) and liability coverage, including bonus-malus system, insured’s demographic characteristics, sales channel, type of car, usage type, local car dummy, fleet car dummy, used car and customer loyalty. The major difference lie in that age of car is only used in pricing physical damage (collision) coverage, while the price of car is used to serve as coverage amount in physical damage (collision) line. And, these two variables are not considered in liability coverage pricing. We seek, therefore, to study the degree of information asymmetry in different types of claims, in order to shed light on the design of pricing for different lines of coverage of automobile insurance.

We analyze the data with two classical empirical models which are two-stage model and bivariate probit model, and compare the degree of information asymmetry among liability claims with bodily injury and those with only property damage, as well as in collision coverage. From the results, firstly, we find evidence for asymmetric information in the third party liability coverage in the auto insurance product in general. Second, on the severity, we find the information asymmetry is severer in claims involved with bodily injury than claims with only property damage. These results proves that it is unreasonable to treat claims with bodily injury and claims with only property damage as the same. Also, these results call for a reform in the pricing mechanism of the third party liability coverage, as well as practice in underwriting and claim adjusting in automobile insurance policy in China.

Our study contributes to the existing literature in several aspects. First, we present the first study analyzing asymmetric information in the auto insurance market in China using large and complete policy level dataset within a representing province. The auto insurance market in China is one of the largest in the world, and the premium income of auto insurance takes up to 78% of property and liability insurance premium in China. For property and liability insurers in China, thus auto insurance has been the most important line of business. We had unique access to the dataset of an entire province for all policy sold for every company operating during that year.

Second, the existing literature mostly focus on different characteristics of insureds, for example younger drivers, newly insureds, and less frequently on different types of claims. We instead focus on different types of claims both within liability coverage (i.e. claims with and without bodily injury), as well as across different lines of coverage (i.e., claims under liability coverage versus those under physical damage coverage).

Third, the empirical literature reaches a consensus that asymmetric information is not prevalent in auto insurance markets. We show, however, that asymmetric information exists in liability coverage but not in physical damage coverage in this market. When we take into account of different types of claims, those with bodily injury involves higher degree of asymmetric information in general, which call for a better design of pricing system for auto insurance product in China.

The complete paper is available for download at:
http://ssrn.com/abstract=3247724.

0 Comments

Submit a Comment

Your email address will not be published. Required fields are marked *

Skip to toolbar