In today's tumultuous economic landscape, the insurance industry is grappling with a pressing concern: the rampant rise of fraudulent activities. As economies face recessionary pressures and uncertainties, insurance fraud has emerged as a critical issue affecting insurers and policyholders alike.

For instance, in a recent case, a Singaporean has been charged with multiple counts of cheating for allegedly duping an insurance company into paying out more than $8.9 million in false claims by depositing money to third-party bank accounts. This incident could have been potentially prevented with the use of a fraud detection tool which uses data analytics to identify and respond to unwanted and suspicious behaviour in real time.

From an insurer’s perspective, analytics helps shave a substantial amount of time needed to detect fraud cases. A leading global insurance company, Aksigorta used to take their investigators six months to expose cases of organised fraud, with analytics it now allows them to do it in just 30 seconds.

The digital transformation of processes within the industry has provided fraudsters with new opportunities to exploit vulnerabilities and deceive insurance companies. Recent estimates reveal that fraud cases now comprise a staggering 10% to 15% of all insurance claims, posing significant challenges for insurers in safeguarding their businesses and ensuring the financial well-being of honest policyholders.

On top of this, though the Singapore insurance market is saturated, it is still poised for substantial growth. It is forecasted to grow by nearly 46% from $4.7 billion in 2021 to $6.8 billion in 2026, according to GlobalData. While it presents opportunities, this growth is not without challenges. As highly regulated as the Singapore insurance market is, there is still the potential for greater insurance fraud with insurance growth.  

Below are seven steps that insurers can take to enhance their fraud detection capabilities, protect their businesses, and provide a more secure environment for their policyholders:

See also: <a class='gtm_see_also_1_inread' style='text-decoration: underline;' href='/news/digital-economy/ceo-nadia-omer-move-wants-travel-brand-recognition-post-airasia-era?utm_source=website&utm_medium=article&utm_campaign=ONPAGESEO'>CEO Nadia Omer of Move wants travel brand recognition in post-AirAsia era</a>

Step 1: Resolve the Data

Data is the cornerstone of analytically detecting and preventing insurance fraud. However, insurers often face challenges due to information silos within their organisations. To overcome this, advanced analytical techniques such as probabilistic matching should be employed to resolve different data variations and identify potential fraud that would be difficult for humans to detect easily or quickly.

Step 2: Create a Three-Pronged Data Strategy

See also: <a class='gtm_see_also_2_inread' style='text-decoration: underline;' href='/digitaledge/digital-economy/how-aws-empowers-asean-organisations-innovate-generative-ai?utm_source=website&utm_medium=article&utm_campaign=ONPAGESEO'>How AWS empowers Asean organisations to innovate with generative AI</a>

A well-planned data strategy allows analytics to work effectively at three levels: event level, entity level, and network level. By utilising previous history and external data, insurers can analyse claims, new business applications, and other events more comprehensively. Additionally, considering various entities such as individuals, addresses, service providers, and IP addresses enhances fraud detection capabilities. The analysis of interrelated networks can also unveil patterns and relationships that may indicate fraud.

For North Carolina Department of Insurance’s Criminal Investigation Division (CID), using SAS Visual Investigator and SAS Visual Analytics has enabled them to monitor trends in real time, discover hidden connections, conduct link analysis and investigate every case that comes in. Prior to the ICIS, investigators hovered around a 60% success rate in complying with case deadlines.

With the new system, that number jumped to 98.7%. Because of this efficiency boost, the average case resolution time has dropped from 90 days to 57, allowing the CID to investigate more cases. Within 7 months of implementation, the agency was able to recover US$6.9 million, highlighting the success of the system.

Step 3: Search the Sources

Making the right data available to investigators is crucial for improving speed and efficiency during the triage and investigation processes. Internal data, such as customer records, should be easily searchable, allowing investigators to uncover additional information relevant to specific cases. Integrating external data sources, including insurance consortium data, credit scores, and social media platforms, can provide valuable insights and strengthen the investigative process.

Step 4: Harness Business Rules

Rule-based systems play an important role in detecting known types of fraud based on specific patterns of activity. However, these systems may generate high false-positive rates and fail to detect emerging fraud modes of operation. Regular review and updates of business rules are essential to maintain effectiveness and adapt to evolving fraud techniques. While business rules serve as a critical part of the fraud detection process, they should be complemented by other analytical approaches.

To stay ahead of the latest tech trends, <a class='gtm_see_also_3_inread' style='text-decoration: underline;' href='/section/digitaledge?utm_source=website&utm_medium=article&utm_campaign=ONPAGESEO'>click here for DigitalEdge Section</a>

Step 5: Detect Anomalies

Anomaly detection, powered by machine learning, allows insurers to establish baselines for key performance indicators associated with claims or new business applications. By continuously monitoring activities and identifying outliers or unusual patterns, insurers can detect both known and emerging fraud with greater accuracy. The automatic nature of anomaly detection ensures prompt identification and correction of fraudulent activities.

Step 6: Use Predictive Modelling

Predictive modelling, a form of machine learning, enables insurers to build models that produce fraud propensity scores. These scores help assess the likelihood of fraud for new business applications and claims, providing valuable insights for investigators. However, it is essential to decode these scores effectively and communicate them to claims handlers, ensuring a clear understanding of the fraud assessment.

Step 7: Automatically Uncover Networks

Organised fraud rings pose a significant threat to insurers. Network link analysis, an automated approach that models relationships between entities, can help identify fraud-related connections and patterns. By analyzing interrelated networks and detecting unexplained relationships, insurers can proactively identify and combat organised fraud activities.

In times of economic recession and pervasive uncertainties, insurance fraud poses a formidable threat to the stability of the industry and the financial security of honest consumers. As fraudulent activities become increasingly sophisticated and organised, traditional fraud detection methods are proving inadequate in curbing this alarming trend. To outsmart fraudsters and protect their organisations and customers, insurers must embrace advanced solutions that leverage technologies such as artificial intelligence, machine learning, and data analytics.

Lim Hsin Yin is the managing director for Singapore at SAS Institute

Seven steps for effective fraud analytics for insurers

Related News

Highlights

MSCI boss warns of climate risks

Related News

The cybersecurity parasite hiding in plain sight

Temasek-StarHub JV Ensign InfoSecurity comes down from Mount Olympus after four stealthy years

Why sustainability and digital transformation must go hand-in-hand