How AI can assist the insurance industry in combating fraud
The Covid-19 pandemic has played out in relation to an unprecedented rise in fraudulent activity with regards to insurance claims. The most notable wave within this sea change is that of funeral claims fraud. The real victims of syndicate and individual criminal activity in the insurance industry are honest policyholders, as their premiums rise in relation to wrongfully obtained payouts. This article discusses the general types of insurance claim fraud, and defines the prevalence of funeral claims fraud in South Africa in more detail. Based on this, suggestions are made as to how Artificial Intelligence (AI) powered tools may be of service in detecting and investigating fraudulent activity.
CONTENTS
What is insurance fraud?
Who gains and who loses from insurance fraud?
Soft vs hard fraud cases
Common types of insurance fraud
Healthcare
Property
Automobile
Life
Funeral insurance in South Africa
Covid-19 and funeral insurance fraud
What is AI and how does it relate to insurance?
3 Use-cases for AI in the Insurance industry
Big data management
Pattern recognition to identify syndicate behaviour
Predictive analytics to optimise retention strategies
WHAT IS INSURANCE FRAUD?
Insurance fraud involves knowingly withholding information from your insurer, or fabricating claims information, for financial gain. This is a criminal offence, and in many countries in the world, when found guilty, it results in prison time as well as large fines.
It is important to emphasise the word ‘knowingly’ because if information is withheld, or found to be false, and this is proven to be beyond the knowledge of the claimant, then it is not a case of fraud. There has to be intent to deceive or manipulate insurance policy terms and procedures to benefit financially.
Moreover, insurance fraud points to a circumstance when the value gained from the insurer can be directly linked to the deception or fabrication. For example, if you were to consciously fabricate your degree or qualifications to the insurer, and then make a claim based on loss of property in your business, the misleading information about your degree did not relate or impact the loss of property claim, and therefore you are not considered to have committed fraud. The insurer of course would not look kindly upon the deception, but you cannot be charged with fraud unless you gained financially as a result of the deception.
WHO GAINS AND WHO LOSES FROM INSURANCE FRAUD
Superficially, the gains and losses of a specific case depend on whether or not the fraudulent claimant got away with it. Of course, if they did, they gain monetary value beyond that which they deserve. But it is important to point out that the losers in the game of insurance fraud are not the insurance companies themselves. The losers are always honest policyholders. Insurance companies will aggregate the rates on premiums depending on overall claims. This means that the percentage of money they pay out on claims that relates to fraudulent activity directly impacts how much honest people end up coughing up. Simply put, the more money ‘stolen’ via fraud, the higher the premiums are for everyone.
SOFT VS HARD FRAUD CASES
Insurance fraud is a common crime. It is hard to accurately calculate exactly how frequently fraudulent claims are made, and how much money is lost through those claims because most go undetected. The most prolific form of insurance fraud is ‘soft fraud’; the most damaging and criminal is ‘hard fraud’
Soft fraud
Also known as opportunistic fraud, this consists of policyholders exaggerating claims that are real and legitimate. Soft fraud occurs frequently in the automobile industry. For example, damages after a collision can be over-inflated to increase the pay out. Another example is in medical insurance, when a policyholder may withhold information about existing disabilities or illnesses to gain a lower premium on the insurance policy.
Hard fraud
This refers to an occurrence of deliberate invention of loss, theft, death, or fire in circumstances covered by a valid insurance policy in order to claim payment for damages. This type of fraud is considered a serious criminal offence and is sometimes carried out by syndicates or criminal rings. If successful, this type of fraud can result in significant monetary gain.
In the case of soft fraud, it is likely the insurer will not prosecute and will settle the dispute directly with the client.
COMMON TYPES OF INSURANCE FRAUD
These are the most prevalent types of insurance fraud:
Healthcare
Healthcare insurance fraud refers to when a business or person makes a claim to have a false medical condition so that they can obtain payments or access to prescription medication. In this subset of insurance fraud, a great deal of cases include the wrongdoings of health care providers who may submit claims to health insurance for procedures that never took place.
Property
This refers to home or business related insurance policies that cover personal, business, or real property (land or buildings). In this subset of insurance fraud, claims are made for loss, fire, theft, or other damage to property and there may be two main types of deceptions: One is that the value of damage, or extent of loss may be exaggerated to obtain a higher pay out. The second is that the act of damage, loss, fire, or theft is deliberately induced, or entirely fabricated, to gain a pay out. A classic example may be a claim of theft of a valuable item of jewellery, when the item was actually never stolen. Another example is deliberate arson.
Automobile
This refers to fraudulent claims made on cars, and other automated vehicles. Commonly, after an accident claimants may leverage the help of dishonest mechanics and auto repair providers to exaggerate the extent of damage, thus obtaining a larger pay out. Less common would be a fabricated incident, or even lying to the insurer with regards to details of model of vehicle, mileage and history.
Life
This type of insurance fraud refers to when a person attempts to obtain life insurance and/or funeral plan payouts by staging or fashioning their own, or another’s death. There are a lot of complex factors involved in successfully staging a death, so the risk is high, but the reward can be substantial.
FUNERAL INSURANCE IN SOUTH AFRICA
Life insurance is calculated depending on an individual’s profile and needs. There are fairly complex underwriting requirements involved, and it is a comparatively expensive form of insurance. The evaluation of life insurance also varies quite extensively. Payments occur after the insured person passes away, but it can take up to a year or more for that money to be paid out.
Funeral insurance is distinct from life insurance in that it covers the cost of a funeral, and the immediate costs incurred by family and loved ones when someone dies. Funeral cover is intended to meet the immediate heavy costs of dealing with death, and for this reason payout is rapid. Some plans promise payout within 48hours of death.
In South Africa funerals are an important part of culture and heritage, and there is a large social expectation to deliver a fairly big ‘send off’. Costs for funerals can be huge! Combine customs and social expectations with the reality that South Africa has, on average, some of the highest funeral costs in the world, and you have a very particular need for financial support at the time of death.
COVID-19 AND FUNERAL INSURANCE FRAUD
Funeral insurance is, in any circumstance, a soft target for fraudsters given how quick and painless the payout process is. However, the Covid-19 pandemic has increased the prevalence of this crime. There have been a lot more deaths, and a lot of economic hardship. With the economy in various stages of shut down, a limited job market has dwindled to devastatingly low levels, and people have been desperate.
Recent reports from South African life insurers have indicated a 12% increase in fraudulent and dishonest claims across all lines of business. The highest incidence being that with regards to funeral insurance, where a total of 2,282 fraudulent claims were made in 2020. In monetary terms, the fraud cost from funeral insurance grew from R54.2 million in 2019 to R80.9 million in 2020.
There are three recent ‘trends’ in funeral claim fraud. They are:
(1) The orchestration of unnatural deaths after a family member has died from natural causes within the waiting period on a funeral plan. Insurers impose a waiting period of 6-12 months on deaths due to natural causes to prevent people from taking our funerals plans when they already know they’re soon to die of an illness or natural cause.
(2) Syndicates are operating with the help of corrupt mortuary employees to purchase unclaimed dead bodies and then claim funeral payouts on policies fraudulently taken out by the syndicates months/years earlier.
(3) Syndicates are also targeting addicts from poor communities to get their personal details on the pretext of employment opportunities, only to use them to apply for funeral policies. In some cases the addicts have been murdered, or if not, the policies are matched with purchased dead bodies.
WHAT IS AI AND HOW DOES IT RELATE TO INSURANCE?
Artificial intelligence (AI) is an umbrella term for intelligent (human-like) computing processes. So AI is the science and engineering of making intelligent machines, especially with regards to software and computer programmes. Often, AI processes mimic human intelligence, but AI is not always confined to methods that are biologically observable.
The insurance industry relies on collecting extensive, complex information about each client, and circumstances they may have been involved in that impact their specific insurance cover. That data collected, over years, or even a lifetime, is enormous. Managing insurance claims and investigations, it then follows, is all about managing big data.
In recent years, AI has been making enormous gains in providing tools and solutions that mimic human intelligence and tasks (sweat hours!) with powerful computer processing in a fraction of the time it takes humans to do. Some AI solutions are now able to provide a scope of insights and depth of analysis that is beyond what a human could viably do.
3 USE-CASES FOR AI IN THE INSURANCE INDUSTRY
These three use cases are constructed to speak to the particular issue of funeral claim fraud, but are applicable in many other areas of insurance claims and investigations.
Big data management
Data management tools powered by AI can reduce the time spent organising, filing, processing, searching, and making sense of large data sets. Insurance providers require multi-layered data organisation, which is highly sensitive and confidential, and yet needs to be accessed at various points over the case lifecycle. AI powered tools can radically reduce the time spent on organisation through intuitively filing and storing, and then providing auto-backlinks to source files.
Searching through data can be a lightning fast and smooth process with machine learning powered AI tools, like Doc Insights, that allow for semantic search results, presenting findings that are nuanced and based on likeness, not just exact keywords. Lastly, AI tools enable various options when it comes to reports and analysis of data; real time reports can be auto-generated from your data set, unearthing instant insights into a case or investigation.
Pattern recognition to determine syndicate behaviour
Syndicate behaviour does have a certain pattern and logic to it. No matter how smart and extensive the syndicate network, there are patterns and traces to activity. Oftentimes, it is impossible for a single human being to get the ‘birds eye’ view of all possible data over time, to be able to detect the logic of syndicate activity. Or, if an investigative team is able to detect this, it takes a really long time and a lot of data analysis before the pattern presents itself.
Machine learning (ML) powered pattern recognition tools can quickly unearth a logic of likeness across a data landscape. These insights can be generated across file types. For example, using facial recognition, a trace of likeness can be drawn across video and image files. ML can unearth contingencies and similarities across other text-based document types, as well as audio. Another example could be a pattern of contingency over a chronology of actions. For example, the contingency between one type of financial activity, an outcome, and the location, or distribution of location of suspects.
Predictive analytics to optimise retention strategies
The term ‘churn’ speaks to the rate at which a client terminates the service agreement or contract. Churn rates are linked to fraudulent activity (notably with hard fraud), because once the wrongfully obtained benefit has been gained, the fraudster usually terminates the contract from which they gained. Insurers can use ML to predict patterns in churn and degrees of effectiveness of retention strategies. This generates a ‘road map’ of highly accurate predictions. The behaviour that rubs against predictions, or defies the logic of behaviour of honest policyholders based on incentives, and other strategies, can then be flagged, and investigated. Moreover, these predictive analytics are also useful for improving services for all policyholders.
For more information about how DocInsights can assist the insurance industry, get in touch with the Doc Insights team and book a demo. Doc Insights offers subscription solutions and custom enterprise solutions.