Introduction

Claims represent the biggest cost within a property and casualty insurance company. Managing the claims process is fundamental to the company's profitability; moreover, an effective claims operation drives customer satisfaction and policyholder retention. Given the importance of the claims management function and the vast amount of information collected on each claim, it is no surprise that analytics have provided substantial return on investment for insurers. Of course, the analytics have to be coupled with a technology platform that can provide seamless delivery to the claims function personnel.

Claims analytics often take two general forms: descriptive and predictive. Descriptive analytics involve robust reporting of trends in key performance indicators. This includes lagging indicators (e.g., payments closure patterns) and leading indicators (e.g., return to work, network penetration). Predictive analytics involve analyzing historical data at each stage in the claims cycle to influence decisions on future claims. Applications of predictive analytics in claims can address any stage of the claims life cycle – from first notice of loss through settlement and recovery.

This chapter discusses a type of predictive modeling application commonly referred to as claims triage. The broad objective of claims triage is to use the characteristics of each individual claim at a specific point in time to predict some future outcome, which then dictates how the claim will be handled. In practice, claims triage might identify simpler claims for fast-track processing or alternatively identify complex claims that require expert handling or intervention. Claims triage models can help assign claims to the right adjuster or inform the adjuster of what actions to take (e.g., when to dispatch an engineer to a claim site or when to assign a nurse case manager to a workers’ compensation claim).

In this chapter, we discuss a specific application that seeks to identify those claims that have a higher likelihood to settle far in excess of early reported loss estimates. This early identification of likely claim escalation should enable the claims department to better manage and potentially mitigate the escalation. Much of our discussion is based on application to workers’ compensation claims, but the methods and issues discussed can be relevant to other long-tailed lines with sufficient claims volume such as general liability.