With an increasing use of electronic underwriting engines, how do reinsurers compare the quality of different provider's rulebooks? Andy Smith investigates risk assessment and operational performance
Either may be valid, or both companies may even agree one set of criteria is better than the other, but one has been unable to implement changes for varying reasons. It is not intended here to argue which approach would get to the heart of the risk more successfully, but rather to demonstrate that if data was provided from both companies on such a risk, the content would be entirely different.
It would not, therefore be possible to run data from these rules through a master rule-set without bending the criteria and therefore compromising credibility. Fundamentally, we are comparing apples with pears.
A further problem with comparisons of rulebooks has been the emphasis on Straight Through Processing (STP) performance and speed of issue. Such figures may well detract from qualitative issues on how risks have been assessed. For example, how do we know that a rulebook achieving 75% automated acceptance or more, is accepting the right cases? Are standard risks being identified accurately or are any substandard risks being inadvertently allocated to the standard pool? STP performance figures do not address such matters.
So, how can we accurately assess the risk selection quality if the STP rates do not help and there is no magic solution that perfectly compares rulebook quality?
focus on the nitty gritty
The solution as it stands is to brush operational performance figures temporarily aside and focus on the qualitative content of the rulebook. That means the nitty gritty; the core branches, questions, answer options and actions that form the rule trees.
This by no means needs to be random nor sporadic; data can be used to drive such an analysis. Ultimately, however, the assessment will depend on core rule designing proficiency, along with recognition of the potential loopholes that so haunt anyone who has been held responsible for signing off rules as fit for purpose.
Perhaps to date, the industry has been overly concerned with headline grabbing STP figures without spending sufficient time on the quality of those decisions. The ultimate test of how well or otherwise underwriting engines are performing will only come when adequate claims data is available. Until then, insurers and reinsurers need to continue investing time and resources into optimal rulebook design, from the perspective of both risk and performance.
Nobody needs reminding that while underwriting engines are excellent tools for processing large numbers of cases accurately and efficiently, they are just as efficient at processing errors. If this is to be avoided, regular data analysis combined with the right skills to interpret what is within, and importantly, outside that data, must be given due priority in the modern protection organisation.
Andy Smith is underwriting development manager at Munich Re UK & Ireland Life
