Just like any other business, an insurer is only profitable if its revenue is not outpaced by costs. However, this relatively simple equation becomes much more complicated when we factor in the nuances associated with the insurance industry. Where other businesses may be able to raise prices to offset higher costs, insurers must abide by rules and regulations that stipulate not only how much, but also when premiums can be raised and when those increases can take effect. As such, even when permitted to raise premiums, insurers can find themselves paying out for claims that are covered under a lower premium.
As important, claims losses, which make up a significant portion of combined ratios, are impacted by a variety of factors outside the control of insurers. For example, the technology found in today’s automobiles adds greatly to the cost of repairs. Inflation and supply chain issues can make it more expensive to repair a home or apartment following an incident. And as we see play out time and time again, weather events are becoming more frequent, and more severe, increasing claims frequency and severity.
There are steps that insurers can take to positively impact combined ratios that are unrelated to increasing premiums. One of these strategies is to avoid paying out on fraudulent claims. And while this may sound simple in theory, it is much more complicated in practice. Organizations like the National Insurance Crime Bureau (NICB) and the Coalition Against Insurance Fraud report that up to 10 percent of all insurance claims include some element of fraud, representing billions of dollars in losses a year. These losses attributed to fraud are a significant contributor to the combined ratio problem. Shift’s own research has indicated that 2-3% of an insurer’s combined ratio can come from fraudulent claims. Yet, when suspicious claims are mixed in among legitimate claims, all of which are being handled by different claims handlers, the suspicious claims can be incredibly difficult to spot.
And it’s not only the volume of fraud being perpetrated against insurers that makes it difficult to spot and stop. Bad actors are constantly evolving their methods. They are adopting new technologies to commit fraud. They are inventing new schemes and recruiting new accomplices. The fight against claims fraud may seem daunting. It may feel like fraud is simply a cost of doing business. The truth is quite different from the perception.
As mentioned earlier, insurance is an industry of both nuances and regulations. Bad actors understand this and constantly develop schemes to take advantage of these industry quirks. For example, in many jurisdictions, once a policy has been written, even if the policyholder has not yet paid on the premium, claims made under that policy must be covered. In some jurisdictions insurers may even be liable for claims made even if the policy is eventually cancelled. Not surprisingly, fraudsters are taking full advantage of this gap, typically in the auto insurance market.
The scam begins when a policy is taken out using a Vehicle Identification Number (VIN) taken from a photo of a vehicle for sale online and the premium paid using an illegitimate form of payment (most likely a stolen credit card). Almost immediately, a claim is filed against the bogus policy using bogus evidence (e.g. doctored photos and documents). What is interesting in this case is that the insurer is dealing with fraud in both the underwriting and claims processes, meaning that the insurer is exposed twice. Not only do they take a hit on the premium, but they may also end up paying out for a fraudulent claim. This double dip by the fraudster can have an outsized impact on the combined ratio. And while the scenario described may be one of the best examples of why it is best practice to detect the risk of potential fraud at underwriting, AI-powered claims fraud detection is a formidable defense. If the insurer is well positioned to discover the suspicious claim and avoid having to pay, they are down only the unpaid premium, not the illegitimate claim made against the policy.
Reducing operational costs has long been a lever insurers could pull to reduce combined ratios. As automation technology became more mainstream, the insurance industry sought opportunities to introduce new efficiencies into the claims process. They adopted online and mobile FNOL, set caps (sometimes surprisingly low) for which claims required no human intervention, and drove toward no touch/low touch and straight through processing for claims. And costs did come down. At the same time incidents of fraud began to climb.
The reasons for this are many. When it comes to opportunistic fraud, it is simply easier for the policyholder to lie if a human insurance professional is not part of the process. Claimants are less reluctant to overestimate values, or submit falsified documents to support the claim when they do not have to defend themselves to another human being. For organized fraud networks, it does not take long to figure out the value for which claims are eligible for automation. Once that threshold is known, bad actors are free to take advantage of it by submitting claims they know will not be reviewed and thus not identified as suspicious. Any combined ratio gains achieved via automation are now negated by fraud.
"Analyzing claims earmarked for automation can spot irregularities… that indicate a claim may be illegitimate."
Fortunately, this is yet another area where claims fraud detection, powered by AI can effectively mitigate the risks introduced by automation. Analyzing claims earmarked for automation can spot irregularities, such as doctored or reused photos and documents, that indicate a claim may be illegitimate. Those claims can then be removed from the automation workflow for further review and investigation. Making advanced claims fraud detection an integral part of any claims automation strategy is one of the best ways to preserve the combined ratio benefit derived from operational efficiency, while at the same time boosting it even further by avoiding paying on fraudulent claims.
Even legitimate accidents can lead to illegitimate claims. This is especially true in cases where bodily injury is reported. And paying out for fraudulent bodily injury claims can wreak havoc on combined ratios. Depending on the jurisdiction, it is conceivable that bodily injury could increase the value of a claim by $15,000 at minimum, and go up from there based on the parameters of the policy. Bad actors have seized on this to create networks of providers — attorneys, doctors, other medical services, etc. — who are willing to submit exaggerated, if not outright false, medical bills associated with an accident. What makes these fraudulent claims so difficult to uncover is that they may be associated with an otherwise legitimate claim for an auto accident or a slip and fall.
"AI makes it easier to identify these provider networks and see how they are impacting an insurer’s book of business."
AI makes it easier to identify these provider networks and see how they are impacting an insurer’s book of business. Connections that may not be evident to an individual claims handler become clear when hundreds of thousands of claims are being analyzed. As important, AI can quickly and efficiently identify those claims where medical expenses may be misaligned with industry norms.
When you take into account the fact that 10% of claims include some element of fraud, the impact on combined ratios is quite clear. Fortunately, what was once considered a “cost of doing business” has evolved into something that insurers can proactively look to thwart. New technologies have given insurers the ability to understand the nature of a claim in seconds and make a determination about how to proceed - pay it, review it, or investigate it - just as quickly and efficiently. If you can shave 2-3 points from your combined ratio through effective claims fraud detection, it seems like a natural investment.
This is the fourth in a series of commentaries on the combined ratio problem facing the insurance industry. For further discussion of this topic and how the application of AI is addressing this challenge throughout the policy lifecycle, continue reading.