The evolving world of fraud detection in underwriting
Today’s underwriting environment is under constant siege. New and emerging risks are mandating new underwriting and portfolio risk management techniques that demand improved or enhanced technologies to bolster performance.
We know that technology such as automation, digital applications, and advanced-analytics engines are further transforming operations such as underwriting, claims and marketing—principal drivers of corporate performance. And while we know that insurers continue to struggle with technology and operations that operate separately, underwriters are faced with expense ratio pressures, a sticky measure of profitability. In our current challenging and competitive environment, mistakes made during the underwriting process that are the result of fraud can cut deeply into an insurer’s bottom line.
“Analysis suggests that a substantial proportion of claims fraud is perpetrated through illegally obtained policies,” notes insurance fraud and security intelligence veteran James Ruotolo CFE, senior manager, Fraud Risk Mitigation, Grant Thornton.
This is true in workers compensation, where the underwriter faces layers of complexity tied to uncovering factual information about the prospect’s business operations. The goal, of course, is to employ a risk management and loss control scenario that most uniquely fits the needs of the business owner. Since accepting new applicants without proper evaluation and assessment of risk classification can have disastrous results, during the application process the underwriter plays part-time detective, often without the tools necessary to thwart fraud.
What Kind of Storage, Exactly?
So, what happens when an insurer experiences application fraud at the point of sale? Here is a fictional example of how this works: In this case, a young junior underwriter under pressure to build her book of business falls prey to carefully planned, deceitful practices of the applicant. The company applying for workers’ compensation insurance identified itself as a newly formed LLC whose focus was to “clean out storage units.”
Because the company was just being established, the underwriter could not review the company’s yet-to-be-launched website or social media pages, so relied solely on NCCI data and the owner’s description of his operations. Without the availability of an X-Mod to reference or other means of obtaining additional information, the underwriter took the word of the owner and issued the policy, assigning a premium of $20 thousand to $25 thousand a year.
The old saying, “we learn by living,” could be adjusted here to “we learned the hard way.” Shortly after policy issuance, the company launched its website, started advertising on social media, and drew the attention of the insurer’s risk control department, which called the underwriter to report that the company had been flagged for review as misclassified. The “storage” units that the company’s owner referred to were, in fact, portable toilets, and the “clean out” meant employees were dealing with dark water, bio-hazard risks, flatbed risks during hauling, and more.
Had the underwriter been able to use additional data referencing other similar fraud attempts, she would have been able to better determine the accuracy of the class codes, setting the premium upwards toward $70 thouand to $80 thousand a year.
What’s Required to “Get it Right”
The complicated system of data collection, and identification of risk exposure is predicated on having accurate information about the subject (who) and the object (what). This means a review of who has single or shared interest, a history of financials, evaluation of the building, employees and their roles, location, etc. equipment, vehicles (on site and off) and on and on, followed by proper categorization of risk, loss cost development and ultimate rate development based on “getting it right.”
“It comes down to being able to mitigate risks with a uniform and data-driven screening to maximize accuracy,” says Christian van Leeuwen, FRISS co-founder and CTO. “And it’s about being able to conduct real-time risk assessments to enable underwriting decisions in a split second—and in the process—build a healthier insurance portfolio.”
Since we know that the underwriting process is tasked to verify proper risk placement, underwriters must have at their disposal data sources that inform their underwriting procedures, technologies that support uncovering hidden risk exposure, inspection reports, loss histories, third-party resources such as rating bureau information, and safety program information to establish how well the policyholder is already managing workplace risk.
Reflecting on the 2019 biennial “State of Insurance Fraud Technology” report, Dennis Jay, executive director at the Coalition Against Insurance Fraud, offers some optimism. “More insurers than ever — 40 percent surveyed — say they will expand and upgrade their toolbox of anti-fraud technology this year,” he says, adding that slightly more than a third (34 percent) plan to add technology to address underwriting such as false insurance applications.
“There is an invitation to develop a competitive advantage by establishing a robust technology-based underwriting fraud detection program,” says Ruotolo. “Data prefill, cross-referencing public records and real-time risk analytics can help address the threat.” Data integration from internal and external sources, predictive modeling and the use of advanced analytics to detect anomalies can also play a key role in the approve/refer/deny decisions critical to the onboarding of a new policyholder.
Insurers will need to remain vigilant to protect their honest customers, because, unfortunately, there are fraudsters ready to scam the system in all economic environments and across all lines of business. Insurers that can provide their underwriters with a powerful source of relevant information based on claims AI and specific fraud models to proactively determine if a certain type of risk is “worth it” will retain competitive advantage.