October 25th, 2011

Predictive Modeling Where You Need It Most

Posted at 1:00 AM ET

tedeschi_john_gcciJohn Tedeschi, Head of GC Analytics® - Americas
Contact

Taming the “long tail” beast has never been easy, and reinsurance intermediaries - including Guy Carpenter - have long focused on helping cedents quantify these risks. For events that happen only infrequently but come with severe financial consequences, planning, mitigation and measurement of outcomes can be notoriously difficult.

Yet, despite the significant threat posed by the long-tail risks, the bulk of companies’ overall risk exposure - for example, the high-frequency, low-impact events that represent 50 percent to 60 percent of premium income - are where intermediaries have been able to provide only limited tools and advice. The focus on the long tail, it seems, has come at the expense of the basic risk and capital management that addresses most of an insurer’s daily needs.

Fortunately, change is afoot. New developments in the reinsurance intermediary space are poised to help carriers. The support will help them take charge of their portfolios, manage high-frequency smaller risks more effectively and take decisive action to improve underwriting, hit return on equity (ROE) targets and grow firm value.

There is a new collaborative effort between Guy Carpenter and EagleEye Analytics, the global leader in predictive analytical solutions for insurance companies. The effort now delivers the capabilities necessary to improve underwriting results in areas where smaller losses, historically, have eroded capital a little bit at a time, accumulating to significant declines.

EagleEye’s proprietary data analytics and modeling process enable carriers to segment, select and price insurance risks more effectively. And as a result of the exclusive strategic relationship, Guy Carpenter can provide EagleEye software and expertise within its overall client offering - and help our clients manage a much higher proportion of risk exposure than ever before.

The EagleEye approach is a salient step forward relative to the predictive modeling methodologies that insurance companies have adopted in recent years. Most existing approaches, usually built around a generalized linear modeling framework, focus on analyzing pre-defined parameters and isolating them from others - it is a siloed process. As a result, carriers spend unnecessary time identifying up front what the parameters of an analysis should be. In the end, the investment in rigor comes at the expense of not being able to respond to changing market conditions.

EagleEye’s process, on the other hand, is not rigid about which parameters (or how many) are analyzed, and it does not fixate on pre-defined parameters. Rather, it mines all exposure and claim data to identify the risk characteristics that generate the most predictive signal.

EagleEye’s system is built upon non-linear machine learning algorithms and allows for a rapid review of parameters that can be monitored over time. An insurer can typically receive indications of key portfolio drivers within 60-90 days, compared to the costly processes prescribed by large consulting firms that can take more than a year.

This process provides greater differentiation and can be used to augment an existing modeling platform. In fact, insurers using EagleEye have seen a 2 percent to 10 percent reduction in loss ratios while better focusing their marketing, claims and loss control efforts to help reduce expenses.

Guy Carpenter is rolling out its predictive modeling capabilities through EagleEye immediately, particularly in many regions where historical claim and policy level detail exists. Early success has already been seen in the United States, Canada, the United Kingdom, Europe, Australia and Asia. Guy Carpenter provides this exclusive service to its clients at reduced fees compared with those charged by alternative predictive analytical consultants - making it easier and quicker to reach ROE targets than ever before.

Click here to read additional GC Capital Ideas stories on modeling >>

Click here to register to receive e-mail updates >>

AddThis Feed Button
Bookmark and Share


Related Posts