City Research Online

Abstract

The insurance market contains systemic sources of risk and bias which emerge from the interactions of the agents operating within that market. This type of risk is often not considered in traditional statistical or competitive pricing models. In this thesis, agent-based simulation models (ABMs) are used to investigate and analyse sources of emergent systemic risk.

Customers’ opinions of insurer service quality influence their loyalty and are often spread via word-of-mouth networks. An ABM is used to examine patterns that might arise from this phenomenon and parameterised with empirical data. The existence of the network acts as a persistent memory, causing a systemic bias whereby an insurer’s early reputation achieved by random chance tends to persist and leads to unequal market shares. This occurs even when information transmission rates are low. This suggests that newer insurers might benefit more from a higher service quality as they build their reputation. Insurers with a higher service quality earn more profit, even when customer preference for better service quality is small. The impact of this systemic effect is exacerbated under a new regulation which bans the practice of charging renewing customers more than new customers.

The winner’s curse is a systemic under-estimation of risk caused by imperfect information. Insurers that have under-estimated risk are more likely to be willing to offer lower prices and therefore win more business. The systemic estimation bias caused by the winner’s curse also impacts stochastic capital models commonly used by insurers to assess risks and manage capital. This leads to capital requirements which are more often underestimated than overestimated. ABM simulations show that there is increased parameter uncertainty in capital estimation when there are either more competitors or fewer customers. Features such as higher customer heterogeneity, higher renewal rates, and increased customer tendency to seek quotes from a greater number of insurers, all functionally create a similar situation and worsen the impact of the winner’s curse. An insurer should consider the impact of the systemic estimation bias caused by the winner’s curse when setting risk and capital management strategies.

An ABM is used to investigate heterogeneous insurer strategies for a market where premium is determined by the balance of supply and demand. Insurers follow either a boundedly rational strategy, or a chartist strategy where market premium is extrapolated from recent trends. As the presence of chartists is increased, the model demonstrates that the market becomes more volatile. Chartist insurers often take better advantage of this disruption and make a higher profit than the rationalists. However, chartist performance is also notably much more volatile. As a result, rationalists remain the dominant choice in an adaptive market where agents may dynamically select strategies. This model suggests that which strategy is ‘best’ depends on the current situation in the market. For insurers primarily driven by profit, a chartist strategy may be optimal. Insurers who value stability may prefer a rationalist strategy.

Finally, an ABM is constructed as an extension to the model produced by Taylor [North American Actuarial Journal, 12(3): 242–26 (2008)] with the aim of establishing a market framework with minimal parameters for use with future work. The model allows for entrants and exits, customer loyalty and price sensitivity, as well as regulatory interventions such as solvency requirements. It also allows for insurers choosing to move either towards or away from the market average, and a strategy where insurers are more willing to take risks when they have a higher capital adequacy. The insurance market simulated by this ABM retains similar dynamics to actual insurance markets including reasonable market premium rates with emergent cyclicality along with stable individual insurer assets. However, the cycles display a slower periodicity than a real-world market.

Publication Type:	Thesis (Doctoral)
Subjects:	H Social Sciences > HD Industries. Land use. Labor > HD61 Risk Management H Social Sciences > HG Finance
Departments:	Bayes Business School > Actuarial Science & Insurance Bayes Business School > Bayes Business School Doctoral Theses Doctoral Theses

Export

Downloads