Predictive modelling: captive maintenance


Mary Chmielowiec and Paul Marshall

Predictive modelling: captive maintenance

Mary Chmielowiec and Paul Marshall discuss the building and maintenance of a profitable captive.

Some captive managers have an unfair advantage. It’s called predictive modelling, and it is revolutionising the way risk is both predicted and managed. These advanced predictive modelling tools provide captives with the information necessary to quickly and accurately market, price, underwrite and defend claims more effectively without a major investment of capital. Fortunately, through expert third-party service providers, the advantages of such sophisticated modelling tools are available to any captive regardless of in-house technological expertise or available capital. Ultimately, these advantages enable captives to operate more efficiently, thereby lowering overall costs and creating a competitive advantage. For any captive that seeks to improve overall performance, it’s time to understand how predictive modelling can make success a reality.

Know the risks

Every day, healthcare facilities tweak their operations to remain profitable under changing reimbursement policies and evolving regulatory expectations. These tweaks, while important for profitability, often have major implications for underlying risk. To illustrate, consider a healthcare facility that is advised by a consultant to reduce staff and increase its Medicare payer mix in order to improve or maintain profitability. In the short term, thisadvice may assist the facility in reaching such goals, but eventually, this decision will have a dramatic effect on risk. Overtime, rising acuity and diminished staffing ratios will lead to adverse incidents. Traditionally, this variation in underlying risk would go undetected. But with predictive modelling tools and risk analytics, even subtle changes in staffing and acuity can be revealed. Predictive modelling provides insurance professionals with the knowledge of any change to risk drivers, thereby allowing the captive to make pre-emptive changes and to manage risk more effectively.

Once a change to a risk driver is detected, the captive can predict how these changes will affect risk and impact its overall portfolio. From that knowledge, the captive gains deep insight into actual risk and can confidently adjust premiums, offer feedback regarding risk management, and continually monitor before a loss occurs. Without predictive modelling and risk analysis, after an account is written, the policy is generally held in status quo with minimal consideration to any variation in underlying risk, until it’s too late and a major loss develops. Conversely, not all changes to risk are negative. Some operational decisions result in positive effects on risk or provide compensating controls. Consequently, it takes sophisticated tools to parse out the individual effects (both positive and negative), add them up correctly in terms of future risk and obtain the most accurate score for that account.

Operational and financial efficiencies

“Predictive modelling infuses more accuracy and more integrity into the process of predicting losses and identifying the risk drivers that allow insurers to operate more efficiently,” explains Chris Kramer, senior vice president of Atlas Insurance Management. These efficiencies, while evident in every step along the continuum from marketing analytics to claims defence, are most notable in the areas of underwriting and claims management.

"Predictive modelling provides insurance professionals with the knowledge of any change to risk drivers, thereby allowing the captive to make pre-emptive changes and to manage risk more effectively."

It all starts with underwriting. Every account must be analysed to establish the appropriate premium in order for the captive to remain viable for the long term. For that reason, experienced underwriters are critical. However, predictive modelling can provide rapid, critical information to assist underwriters in making more accurate, consistent and timely decisions about the potential for risk. To analyse just one account, an underwriter can spend hours. In comparison, a predictive modelling system can present an analysis in minutes with equal or greater accuracy. In fact, predictive modelling systems have been tested against traditional underwriting approaches and were found to be five to 10 times more accurate.

“There are only so many things an underwriter can look at to assess risk. If he had an infinite amount of time to assess each risk, he’d be almost 100 percent accurate. But, in reality, no underwriterhas infinite time, and we all make judgement calls and mistakes. Predictive modelling allows all the people involved, from underwriters and actuaries to those who set reserves, to use the insight gleaned from hundreds of data points. After the data is entered into predictive modelling software systems, highly reliable results provide underwriters with an enhanced ability to assess risk and set proper premiums and parameters for coverage,” Kramer adds.

Defend yourself

In terms of claims management, predictive modelling accelerates the acquisition of knowledge and helps insurers put claims into the proper context. “The quicker we can investigate, understand and evaluate the claim, the quicker we can reach our decision points,” explains Paul Hamlin, founder and president of Hamlin and Burton Liability Management, Inc. and an expert in the field of claims management. “Does the claim have value? What is the potential value of this claim? Are we comfortable defending it at trial? Risk analytics helps us tailor the investigations, because if we can identify problem areas, we can dig deeper and perform more pointed and specific investigations. Any time we have access to critical information about the facility, staffing levels, potential problem areas, etc. early in the claims-handling process, we improve our chances of being able to evaluate the claim accurately and resolve it in an optimal fashion.”

Moreover, predictive modelling can chart the course for improved negotiations with plaintiffs and, ideally, lower overall settlements. “Predictive modelling and risk analytics have allowed us to focus on critical claims by gaining rapid clinical and analytical insight to help us put claims into a broader context and obtain more favourable outcomes,” notes Jonathan Swann, underwriter, CareSurance Nursing Home Programme at Lloyd’s. At a minimum, the information guides claims professionals and quickly gives them the wisdom to determine which claims to defend and which to settle. Too many captives are satisfied with loss ratios of 50 percent because, for the industry as a whole, this is acceptable. But predictive modelling is going to shake up the perception of ‘acceptable’ loss ratios and has already demonstrated its ability to deliver significantly lower loss ratios.

The end result is improved accuracy and lower claims administration costs. Claims managers who research files in the traditional approach will spend 25 to 50 percent more time completing the legwork necessary to obtain the data necessary to evaluate a claim. “Most data in the comprehensive risk analysis we could find out on our own, but it would take a tremendous amount of legwork and would be done slowly, expensively and inconsistently,” says Hamlin. If predictive modelling and risk analysis can save an insurer just five percent of total claims management expense, that could easily be converted into a significant competitive advantage.

It’s flexible—pick and choose applications

Predictive modelling tools are available for any step along the continuum, including marketing analytics, underwriting, risk management and loss mitigation. To illustrate, consider a captive created to meet the insurance needs of nursing homes belonging to a particular faith-based association. Let’s assume that the president of the association has several nursing homes in the pool andbelieves his facilities deserve a reduced rate. In this instance, the captive can use an outside, objective third-party expert and access models that quickly evaluate each facility and provide a one-page summary of the comparative risk.

Such analysis can also provide the quantitative information necessary to allocate premiums more fairly and more accurately across the captive. Essentially, predictive modelling can help eliminate the human and emotional response that naturally occurs in the underwriting process. As explained by John Henry, principal owner of the Boston Red Sox: “People operate with beliefs and biases. To the extent that you can eliminate both and replace them with data, you gain a clear advantage. Actual data means more than individual perception/belief.”


Predictive modelling and public data can also be used to target preferred risk and steer the acquisition of new business. By taking the offensive position in finding new business that meets a desired set of risk parameters, a tremendous amount of savings can be realised. On average, 50 percent of the first year’s premium is paid out in commission, underwriting and the sales application process—not to mention that finding just one suitable account in the traditional approach may take underwriting the effort of reviewing and analysing more than 10 accounts before finding one that is right for the programme.

Another advantage of predictive modelling is the ability to establish more accurate reserves. With improved accuracy in identifying overall risk, carriers can establish reserves that are commensurate with the underlying risk. Such financial efficiencies allow an organisation to direct limited resources to the right place. If too much premium is allocated to reserve, then not enough is available for operations, and vice versa. Again, predictive modelling helps an organisation be both fiscally and operationally efficient.

Consider the evidence To remain competitive in the industry, insurers need to capitalise on advances in technology and the growing availability of data by implementing data-driven predictive modelling tools. Evidence clearly indicates that insurers that embrace the power of sophisticated predictive modelling tools experience significantly lower loss ratios than industry averages. The reason is simple: by gaining greater insight into what is happening today in their risk pool, action can be taken to address risk, appropriately price the premium and operate more effectively.

Mary Chmielowiec is executive vice president for insurance at PointRight Inc. She can be contacted at:

Paul Marshall is director of insurance business development at PointRight Inc. He can be contacted at:

The company website is:

Captive International