Skip to content


Research Interest

  • Insurance Economics and Regulations
  • Industrial Organization
  • Corporate Risk Management
  • Public Policy

Working Papers

Multidimensional Barriers to Entry in the Insurance Industry


In this paper, we propose and estimate a multi-agent model of entry. Estimations are conducted at each market-year level such that we could disentangle the relative importance of barriers to entry across three dimensions: geographic, product, and time. Barriers to entry exist and are quite substantial in the insurance industry. Overall, we find de novo entrants are the ones that most subject to barriers to entry across all markets. On average, expanding within a state is as costly as expanding within a product line. With further examinations, we discover that product-specific knowledge plays a critical role in successful expansions, and it is relatively more important than state-specific connections. Among all product lines, expertise in mortgage guaranty insurance creates the most barriers, and these barriers are most subjected to impacts of the financial crisis.

Full paper

Insurance Bundling and Market Power

with Cameron M. Ellis


The goal of this paper is to estimate an equilibrium model of insurance demand and the impact of product bundling on price elasticity in the P&C insurance industry, without having the individual-level data. We employ a random coefficients logit model that allows flexible substitution patterns. Moreover, we propose an innovative instrument, the retaliatory premium tax rates, which are a unique feature of the insurance industry, to address the endogeneity of price. It is very difficult to identify supply-side cost shifters in the insurance industry because most “costs” of insurance are claim-related, which is obviously not exogenous to the demand. In the insurance industry, states are allowed to discriminate against foreign insurers, and they can also defense these discriminatory taxes. Under the system of retaliatory taxes, the premium tax rate of a foreign insurer is the higher of the tax rate in domicile state and that of the operating states. Thus, we derive identification from the following scenario: if a state marginally changes the local tax rate, this will impact firms domiciled in the same state and firms domiciled in states with a lower rate; it will not impact firms domiciled in states with a higher rate. This change in tax rate will create firm-market specific shocks that are plausibly exogenous to local demand. In this paper, we also estimate the market power and markups of product bundling.

Do Hand-held Cellphone Bans Improve Road Safety? Evidence from the Insurance Industry


Road safety is one of the central public health concerns in the past few years. The Association for Safe International Road Travel (ASIRT) forecasts traffic injuries to be the fifth leading cause of death by 2030 if no proper actions are taken. Recent research has drawn attention to the relation between distracted driving and road safety. A few studies shed light on the effect of distracted driving on the automobile insurance market. In this paper, I employ a multivariate differences-in-differences design and synthetic control method to study the impact of hand-held cellphone bans on automobile insurers, regarding losses incurred, loss frequencies, and loss severities. The preliminary results indicate that the hand-held cellphone bans do not have an impact on auto insurers’ losses incurred at the state-level. However, it is negatively associated with collision frequencies and positively associated with collision severities. Furthermore, our results suggest that insurance rate regulation is holding back the effect of hand-held cellphone bans.

Peer Effects and Envy: Measurement and Impact on Retirement Decisions

with Jeffrey Boles and Jacqueline Volkman-Wise


In this paper, we experimentally examine the impact of peer effects upon the individual investment decision-making process for retirement planning. Participants make portfolio allocation decisions, knowing the decision made by their peers. We find there are significant differences in portfolio selections when individuals know the choice made by their peers as compared to a group of individuals that do not know the choice made by peers. The method implemented here allows us to determine which factors influence individuals to be more affected by decisions made by their peers or alternatively, in which individuals are more envious of their peers. Finally, theoretical predictions in previous work conjecture that the functional form for envy is convex. The experimental design of this paper allows for the determination of the proper functional form for modeling envy, and we find the support that the functional form should be convex.

Perseverantia Vincit