Wonder

Log in

Research Outline

Prepared for Jian Z. | Delivered November 12, 2019

Property and Casualty Insurance Research

Goals

To understand what the leading practices are to detect fraud and handle potentially fraudulent insurance claims, specifically for property and casualty (P&C) insurance. An ideal response would include the following high priority information: One: What the leading insurance companies are doing to detect potentially fraudulent claims, and what those signals or indicators are. Two: once fraud is detected, what are leading practices to staff the right people to handle fraudulent claims, Three: how do companies operationalize fraud teams, as well as detailing what the pros and cons of outsourcing versus having fraud teams in house, and Four: the role that machine learning will play in this landscape. Specifically, how data and analytics are being used generally to support fraud detection and best practice. Of a lower priority, is having an understanding of whether there are data points on total potential fraud and the impact of losses for property & casualty insurance. As well to verify this data point: "Industry sources estimate that the total cost of insurance fraud in the US is more than $40B/ year and can represent 5-10% of losses paid." Additionally, to find other similar datapoints. Another lower priority goal is to understand the tools or software that insurance companies rely on. For example, some use Carpe Data to detect social media related fraud. Do others?

Early Findings

  • According to a paper by Accenture published in 2018, "insurance companies lose an estimated US$30 billion a year to fraudulent claims. Machine learning helps them identify potential fraudulent claims faster and more accurately, and flag them for investigation." It is thought that "machine learning algorithms are superior to traditional predictive models for this application because they can tap into unstructured and semi-structured data such as claims notes and documents as well as structured data, to identify potential fraud."
  • "Chola MS, one of India’s fastest-growing insurance companies, has adopted mobile technology for its claims survey process. The company’s vehicle surveyor application uses the voice, camera and data connectivity capabilities of the Samsung Galaxy Tablet to capture and store auto survey data in one database. In the past, loss adjusters had to manually match survey notes with e-mail and photos saved in other databases before making a decision on a claim. This initiative helped to speed up the claims settlement process, increased surveyor productivity and improved fraud prevention."
  • For property & casualty (P&C) insurers, and especially for chief claims officers, artificial intelligence (AI) and machine learning are quickly becoming powerful new tools for reducing losses from fraudulent claims.
  • Internal fraud, rate evasion, underwriting fraud, claims fraud, cybersecurity fraud add up to more than $80 billion a year in the US, according to conservative estimates from the Coalition Against Insurance Fraud.
  • Fraud comprises about 10% of property-casualty insurance losses and loss adjustment expenses which is about $34 billion a year.
  • "Unlike rules-based systems, which are fairly easy for fraudsters to test and circumvent", machine learning, which is a form of AI, adapts to changing behaviors in a population through automated model building. With every iteration, the algorithms get smarter and more accurately detect fraud.
  • French AI startup firm Shift Technology "incorporates machine learning and AI in their fraud prevention services, which have already processed over 77 million claims. The cognitive machine learning algorithms have reached a 75 percent accuracy rate for detecting fraudulent insurance claims. The ML algorithms provide details on suspicious claims with potential liability and repair cost assessments and suggest procedures that can resolve and enhance fraud protection."
  •  57 percent of insurers predict an increase in personal-property fraud by policyholders."
  •  18 percent of US consumers believe it’s acceptable to pad a claim to make up for premiums paid in the past.
  •  86 percent of Americans think "insurance fraud leads to higher rates for everyone."
  •  10 percent think 'insurance fraud doesn’t hurt anyone."
  • The total cost of insurance fraud (not counting health insurance) in the US is estimated to be more than $40B a year, according to the FBI. 
  • According to DXC Technology, a global IT services company that leads digital transformations for clients by managing and modernizing mission-critical systems, there are 6 steps to prevent insurance fraud. One: IMPLEMENT A FOUNDATIONAL FRAMEWORK, Two: KNOW THE RELATIVE LEVEL OF FRAUD POTENTIAL, Three: USE DATA ANALYTICS TO DETECT FRAUD, Four: CONTINUALLY REVIEW AND RESCORE CLAIMS, Five: ADOPT A LAYERED APPROACH, and Six: REVISE BASED ON MARKET CONDITIONS.
  • In addition to this public search, we scanned our proprietary research database of over 1 million sources and were unable to find any specific research reports that address the stated goals.

Summary Of Our Early Findings Relevant To The Goals

  • Our first hour of research was able to dig into the role that machine learning will play in P&C insurance, specifically surrounding how data and analytics are being used generally to support fraud detection and best practice.
  • We were also able to provide several data points similar to the one provided to us, and we were able to verify that the FBI stated that "Industry sources estimate that the total cost of insurance fraud in the US is more than $40B a year and can represent 5-10% of losses paid."
  • We assumed a United States focus for this research. If a broader approach is required, for example, a global look, then this would have to be clearly communicated to us in any reply.
  • Because the ask was so large and broad, the vast majority of the ask could not be addressed in the first hour of research.
  • Please select one or more of the options provided in the proposed scoping section below.