Want more Insights from SAS? Subscribe to our Insights newsletter. Or check back often to get more insights on the topics you care about, including analytics, big data, data management, marketing, risk and fraud.
The state of insurance fraud technology
Highlights from the 2016 Coalition Against Insurance Fraud Report
As Hurricane Matthew marched up the southern Atlantic coast in October 2016, adjusters were bracing for the aftermath – the flood of insurance claims soon to come. They knew from industry research and experience that about 10 percent of those claims would likely contain some flavor of fraud – but which ones?
Fraud erodes the bottom line for insurance companies. Whether it’s internal fraud, rate evasion, underwriting fraud, claims fraud or cybersecurity breaches, insurance fraud (not including health insurance) adds up to more than $40 billion per year in the US, according to the FBI. Virtually all insurers are fighting back with some form of anti-fraud technology.
In 2016, the Coalition Against Insurance Fraud, in partnership with SAS, conducted a survey-based study to better understand how insurers are using technology to detect and prevent these losses. The 86 participating companies represented a significant share of the property/casualty market. The study is a follow-up to similar studies conducted in 2012 and 2014, which enables comparisons over time. To what degree are insurers succeeding with technology? What are their challenges, results and future plans?
Here are some key findings from the report:
Hardly anyone is trying to fight fraud without technology.
The 2012 study concluded that roughly half of insurers had fully integrated technology into their anti-fraud systems. By 2016, that percentage was closer to 75 percent. Only 2.5 percent of respondents reported using no anti-fraud technology, down from 8 percent in 2012.
Insurers are reporting higher fraud rates.
A total of 61 percent report that the fraud incidents increased either slightly or significantly in the last three years (compared to 51 percent of respondents in the 2014 study). What’s unclear is whether this increase is because fraud is actually on the rise or because insurers are getting better at detecting it. For example, using advanced analytics, one insurance company using SAS® has doubled its capacity in terms of fraud referrals and investigations.
Insurers are adopting more sophisticated fraud detection methods.
More than half of insurers surveyed are now using predictive modeling, up 35 percent from just four years ago (from 40 to 54 percent). However, systems based on red flags and business rules are more common, used by 90 percent of insurers who have some form of anti-fraud technology. Real-time scoring technology can help investigators detect fraud immediately, and risk- and value-based scoring models help prioritize claims for investigators. Insurers are likely to get better results from whatever technologies they use, because they have more data sources and larger amounts of data available to them.
Claims fraud detection continues to be the leading area for technology.
Insurance companies use anti-fraud analytics for various purposes, such as to detect rate evasion and underwriting fraud, but claims fraud leads all other uses. Now 76 percent of insurers use technology to detect suspect claims, up from 65 percent four years ago.
Technology is responsible for more referrals.
In 2012, only 55 percent of respondents said they were receiving more than 10 percent of their referrals from technology. By 2014, that figure had risen to 66 percent; in 2016 it’s 70 percent. Two years ago, nobody reported getting 60 percent or more of their referrals through their anti-fraud technologies; now 6 percent of insurers do.
These referrals are better quality.
Nearly 60 percent of respondents said technology is not only producing more referrals, but better quality ones – much earlier in the cycle. That means fewer false positives, more effective use of investigators’ time and mitigation of losses.
Preferred success metrics remain unchanged from years past.
When asked how their organizations measured success, 50 percent cited fraud detection rate, followed by the number of referrals received. Still, one in five said their organizations do not use any metrics to gauge the success rate of their technologies.
IT resources are still a constraint.
More than 75 percent of respondents cited lack of IT resources – both budget and in-house expertise – as the biggest challenge in implementing anti-fraud technology. That may explain why only 64 percent maintain their systems in-house and the rest rely on hosted or remotely managed solutions from companies such as SAS. While IT resources may be tight, the same technology used for one area of fraud can be useful in other departments to spread the total cost of ownership and seek benefits in other ways – such as preventing claims before they are paid and providing better customer service. Because insurance companies may be able to stop paying substantial sums to settle fraudulent claims, their policyholders won’t see rate hikes due to fraud. Insurers can maintain their focus on superior customer service and customized plans.
Investment in anti-fraud technology will likely continue to grow.
Fully 85 percent of insurers say their technology budgets will hold steady or increase in 2017, up significantly from 2014. Where will that budget go? Most say they will invest in predictive modeling systems, followed by link analysis and social media software, and then text mining.
The growing interest in advanced analytics is significant because analytics can make fraud detection far more accurate and effective – and analytics can evolve as fraud schemes shift. The 2016 study showed that the most notable areas of growth were in these analytics technologies:
- Predictive modeling reveals patterns among data elements that point to a high propensity for fraud. With predictive insight, insurers move from pay-and-chase to prevention.
- Social network analysis, also called link analysis, reveals previously hidden connections among entities to expose organized fraud rings or collusive activities.
- Machine learning adapts to changing behaviors in a population through automated model building. With every iteration, the algorithms get smarter and deliver more accurate results. It’s easy to see the value of machine learning in keeping pace with changing fraud tactics.
Find out more in the full 2016 report, The State of Insurance Fraud Technology: A Study of Insurer Use, Strategies and Plans for Anti-Fraud Technology.