While the commercial property and casualty industry continues tosearch for signs of a hardening market, several countervailingeconomic factors are combining to slow any positive development.The deepening recession will continue to reduce premiums in linesbased on payroll and sales revenues while generating increasedlosses in many lines sensitive to the economic downturn.

|

Carriers that want to emerge stronger from this toughenvironment must improve underwriting discipline and processefficiency in a cost-effective fashion.

|

Unfortunately, it is not easy in commercial insurance to reduceunderwriting costs while maintaining underwriting control. This ispartly because, unlike personal lines, most commercial insurance isstill written manually and with minimum use of advancedtechnology.

|

There is also a general belief that the sophisticatedunderwriting judgment that goes into reviewing submissions andapplying rating and other pricing tools is too complex to be aidedby analytical tools and predictive models.

|

However, recent innovations in, and applications of, dataanalytics and predictive modeling in other industries as well as inpersonal lines and small-market commercial insurance provide uswith solutions for dramatically improving underwriting performancein the middle-market commercial insurance sector.

|

Utilizing analytical tools with a logical structure designed byin-house underwriting experts, these solutions act as an aid to,not a substitute for, the existing underwriting process. The resultis a more effective and efficient commercial underwritingprogram.

|

Underwriting in commercial insurance tends to include a broadsubjective component. Most underwriting teams experience asignificant level of variability in rating and pricing decisions,because of the diverse knowledge and experience levels of the staffas well as their disparate abilities in managing negotiations withagents and brokers.

|

|

As a result, there is considerable debate about the merits andrisks of using analytical tools for standardizing and improvingunderwriting performance in middle-market commercial insurance.While the perceived benefits are significant, underwriting managerspoint to the various challenges and hurdles faced in this market,including:

|

o Lack of Adequate Data:

|

Because analytical modeling is widely understood to require alarge number of records, there is concern that commercial classesmay be too broad to provide enough data to develop stablemodels.

|

In addition, organizations may not always have the extensiveloss and premium history for each line of business needed todevelop predictive models of profitability on individualsubmissions.

|

o Prohibitive Cost and Disruptive Nature of theProject:

|

Analytical technologies, when implemented in personal lines andin other industries, have typically involved a significantorganizational effort. This has entailed major cost outlays,process changes and infrastructure conversions.

|

o Highly Subjective Nature of CommercialInsurance:

|

Perhaps the key objection to using analytical tools is the viewthat commercial underwriting takes years of experience to performproperly and is too subjective for modeling. By this widelyaccepted view of commercial underwriting, expert judgment is thekey to the decision-making process in delivering the bestunderwriting outcome.

|

That outlook is changing. The above concerns, all valid, can nowbe addressed by modern predictive modeling approaches that includea whole range of techniques–from pure data-driven approaches likecluster analysis and decision trees, to more complex models likemultivariate regression and uplift modeling, to expert drivenmethodologies that are designed by replicating specialists' logicstructures, such as belief-based expert systems.

|

Based on successes in similar applications, a particularlypromising approach to improving middle-market commercial insuranceunderwriting is provided by Bayesian belief-based modeling. Thisapproach relies less on existing data and more on expertjudgment.

|

In effect, the approach models the decision-making thoughtprocess of expert underwriters within the organization relying oninformation available to the underwriter as part of thesubmission.

|

All model outcomes can be analyzed in detail, since the causallogical structures are explicit and transparent.

|

Initially designed based on exhaustive input from in-houseunderwriting experts (and any relevant internal and external datathat may be available), over time the model “learns” from actualsubmission data and underwriting results. This approach isparticularly powerful because it can be designed and tested with arelatively minor investment of time and cost.

|

Full-scale implementation, with the associated infrastructureinvestment and process changes, need occur only after the approachhas proven itself and after the organization has developed a clearunderstanding of the benefits and costs of the requisiteinvestment.

|

The accompanying graphic provides a high-level schematic of acomprehensive approach to improving underwriting performance usingBayesian predictive models. In addition to a technical andperformance reporting facility, the solution comprises three keycomponents: a Risk Appetite Gate, Bayesian Risk Quality PredictiveModels, and a Bayesian Predictive Methodology for PackageSubmissions.

|

The Risk Appetite Gate is a tool managed by a lower-levelanalyst or technical resource and ensures that only submissionsthat fall under the organization's specific underwriting riskappetite definitions are evaluated by its underwriters–all othersare either declined or returned to the producer for clarificationor changes.

|

Risk appetite rules are absolute criteria that can be based onSIC codes, class, location, line of business, and so on. A standardBusiness Rules Engine is used for defining and managing therules.

|

Existing risk exposure of the organization's book of business isan input into the tool. Consequently, the tool is dynamic, ensuringthat underwriting does not review undesired submissions.

|

There is a Bayesian model for each product line, and allsubmissions are evaluated on a line-by-line basis. The followingare key characteristics of these models:

|

o Models evaluate Risk Quality (not profitability) of each lineof submission.

|

o Models are designed and developed with critical input fromin-house underwriting experts, and “trained” with historical dataif available, or with ongoing data for an initial period ifnot.

|

o After initial training, the models are “tested” with anadditional set of data.

|

Once fine-tuned with these two groups of data, the modelscontinuously “learn” in a “model learning loop” with livesubmissions data and underwriting results. Model performance andpredictability improves over time.

|

A model's output is a score evaluating the risk qualityprobabilities for the respective line of business. Based onthresholds applied because of internal risk quality guidelines andresource constraints, these are divided into specificrecommendations including: “Decline,” “Pass to Rating,” or “Referto Underwriting for Further Review.”

|

All entries referred to underwriting include guidelineindicators, identifying the key reasons for the referral based onthe model's underlying analysis.

|

These models and the associated processes are designed andconstructed based on the market segments, underwriting processesand internal data availability specific to each carrier.

|

After the models have identified a score for each individualline of a package submission, these scores are provided to aBayesian model that assesses the overall score of thesubmission.

|

Similar to the Risk Quality predictive models, this model isdesigned with expert input to evaluate package submissions based onweights of the relative risk quality scores of each line as wouldbe assessed by internal experts.

|

In addition, any outside rules–such as a temporaryorganizational desire to grow the workers' compensation book, forexample–can be added to the model to ensure proper treatment of allkinds of package submissions.

|

Similar to monoline entries, any multiline submission referredto underwriting includes guideline indicators providing reasons whythe package was referred and what the underwriter should befocusing on in their evaluation.

|

What are the advantages of applying these methods?

|

Internal underwriting profitability models, if available, canprovide further enhancement of the process. These models, incombination with the risk quality scores, can help determine aweighted expectation of the profitability of each packagesubmission, thereby helping prioritize submissions further forunderwriting.

|

Carriers looking to dramatically improve underwriting quality byusing this solution as a supplement to expert underwriting judgmentwill realize several advantages, including:

|

o Increased efficiency of valuable underwritingresources, as they focus only on the most complex and promisingsubmissions.

|

o Ability to implement the solution even withlimited initial data by reliance on models built to replicateexpert underwriting judgment.

|

o Continuous improvement, as the models learnand grow more sophisticated over time.

|

o Relatively low up-front cost to test theapproach before full implementation.

|

o Rapid acceptance by underwriting, sincemodels are designed with underwriting input and because each modelrecommendation can be analyzed for root causes.

|

o Use as a documentation, quality improvement andtraining tool for inexperienced underwriters.

|

o Valuable technical/modeling and underwritingperformance-related reporting facility for management analysis anddecision-making.

|

Advanced analytical tools have proven successful in manyapplications where expert judgment is a key element of the decisionprocess.

|

In today's challenging environment, commercial carriers focusedon the middle-market can significantly reduce underwriting expenseand maintain underwriting discipline by adopting these tools.Embracing these methods will translate into a sustained competitiveadvantage through all market conditions.

|

Aamer Mumtaz is director of business analyticsat TNC Management Group based out of Chicago, Ill. He may bereached at [email protected].

Want to continue reading?
Become a Free PropertyCasualty360 Digital Reader

  • All PropertyCasualty360.com news coverage, best practices, and in-depth analysis.
  • Educational webcasts, resources from industry leaders, and informative newsletters.
  • Other award-winning websites including BenefitsPRO.com and ThinkAdvisor.com.
NOT FOR REPRINT

© 2024 ALM Global, LLC, All Rights Reserved. Request academic re-use from www.copyright.com. All other uses, submit a request to [email protected]. For more information visit Asset & Logo Licensing.