How will insurance regulators evaluate AI/Black Box underwriting decisions?

There is quite a bit of auto insurance industry discussion about using AI machine learning models to identify bad risks during the underwriting process. The idea is that AI tools can tease out subtle variations in customer status and behavior that make a significant financial difference. But there's a catch.

The Legal and Regulatory Risks of AI Machine Learning
Given that insurance rates are tightly regulated, using AI machine learning to evaluate customer risk means AI models will be used to deny coverage to customers that do not 'measure up'. Which is a problem. Because customers expect carriers to have a rational reason for why they are being rejected. And regulators and courts become concerned when certain groups, for example minority customers, are rejected at higher rates than others. Unfortunately AI machine learning models cannot explain why they make a particular decision because they are 'black box' solutions. Their creators literally cannot explain how the machine arrives at its decisions, they can only judge the results.
This means relying on AI models to sort the 'good' customers from the 'bad' ones carries large potential regulatory and legal risks should its decisions result or even just be suspected of resulting in disparate impact for a disadvantaged class.

How VeracityID Uses AI
We use AI models to help carriers sort between good and bad customers too. But we only use them as a 'Tip and Lead', as an indicator that a given customer may pose a greater than normal fraud risk. We then use our other tools such as graph analytics, business rules, mobile documentation apps and automated interaction with the customer to identify and resolve specific real risks. When a carrier using VeracityID tools declines to cover a customer they can give a specific reason that makes sense to both customers and regulators:

• "they told us address A was their home but we discovered address B was"
• "they claimed their truck was a personal vehicle but these specific images of it show commercial gear and signage"
• "they quoted a policy with their 17 year old son and then without him and then refused to either add him to, or exclude him from the policy"

Doing so both eliminates legal and regulatory risk and improves customer satisfaction because instead of simply rejecting a customer with no explanation, VeracityID allows carriers to engage in an automated interaction that gives the customer the opportunity to modify their application and still receive coverage. This means that customers who made a mistake or for whom the purchased, 3rd party data is wrong, have the opportunity to correct the record and get back on the path to coverage.

Our motto is 'stop fraud before it starts' but it's also "stop legal problems before they start". 

www.veracityid.com  Twitter: @veracityid.com 

Are Mandated Auto Premium Reductions Justified?

Policymakers in many states - notably including California, Michigan and Louisiana in recent weeks - are considering aggressive steps to drive down auto insurance premiums in the wake of the Covid-19 pandemic and changes in driving behavior.  Even as they contemplate this, many leading carriers have already taken temporary action to reduce rates during the pandemic – both because it was the right thing to do and to forestall damaging policy changes.

On the surface, those initial reductions make some sense.  According to the US Energy Information Administration, gas sales were down 20% in March 2020 over March 2019 – and we all saw the roads were pretty empty.  Gas sales were very likely even lower in April during the height of the shutdown.  With reduced gas sales and empty roads came fewer claims.  And lower claims should drive down premiums in normal circumstances.

But will this still be true as the economy is re-opened in the weeks and months ahead?  Maybe not.

Early data from refineries and distributors suggest gas sales are quickly returning to ‘normal’.  Instead of continued relief from traffic, some experts predict that gridlock is coming to our cities as people choose to return to work by car instead of mass transit.  One forecaster suggests a 30% increase in commute traffic is coming this fall – and that our highways will soon be congested even more than before the virus hit (https://abc7news.com/bay-area-commute-covid-19-coronavirus-traffic/6248486/).  But other forecasters suggest that 25-30% of office employees will continue to work from home at least several days a week – and thereby skip the commute altogether (https://globalworkplaceanalytics.com/work-at-home-after-covid-19-our-forecast).  

At the same time, leading travel analysts observe that 30% of travelers have already switched from flying to driving for work and vacation.  And of course, there is the fear that a second wave could trigger renewed shutdowns.

So, what should we make of this?  How will the shift from bus to car and commute to vacation travel play out in terms of insurance economics?  Will these changes be transitory or permanent? Should policymakers act now – or wait for this settle out before making any dramatic moves?

At VeracityID, we think:

• The pressure to offer mileage-based insurance will increase – at least until there is clarity about driving patterns in the post-Covid economy.  But this may accelerate an adverse selection problem – with low mileage drivers switching from conventional to mileage-based coverage, and high mileage drivers sticking to conventional coverage with unreasonably low estimated miles.  Carriers need to monitor vehicle use in both cases – or will get burned.
• Policymakers should tread lightly for a few more months, at least.  If mileage increases significantly, a forced premium reduction could induce market failures.  If mileage falls, competition will soon drive down rates or risk intervention.  But if traffic gridlock happens, we will likely see a drop in claims severity in the near-term and an acceleration toward alternative work and commute patterns in the mid-term.  But none of this is clear today.
• Carriers should use this period to look for ways to increase customer retention by focusing on improving the customer experience across channels and striving to reduce rates by eliminating fraud, manual work and transaction costs.

Eliminating Underwriting Fraud Can Save Consumers and Carriers Up to 20% of NPW.  

Website: veracityid.com   Twitter: @VeracityID

Coming to an agent near you: direct online auto insurance losses

It is common knowledge that auto insurance distributed via the direct, online channel is much less profitable than agent business. But why is that? And is it likely to persist?

Carriers that distribute through agents have historically earned their profits through "seasoning", first year business is unprofitable but profitability improves as carriers run off bad customers and renew  the rest. This contrasts with the direct channel where customers tend to shop their insurance frequently because carriers have made shopping online so easy. And since it's easy to shop, direct customers find it easy to use multiple online quotes to figure out just which combination of less than honest data can generate the lowest premium for them.

So good news for agent-centric carriers right? Not really. We know that more and more agent customers are using online quote sites to shop for their insurance. Once they find the policy they want, they buy it from an agent. But this means that they are both learning how easy it is to shop at every renewal and how easy it is to manipulate their rate.

If you want to see where all of this is going, look at the United Kingdom: there, virtually all shopping happens on line either using carrier quote systems or more often, aggregator sites with both frequent shopping and rate manipulation being common. But most customers still buy their policies from agents. The result is a terrible, unprofitable auto insurance market with average loss ratios much higher than in the US..

And this "British disease" is now becoming a problem for US carrier. With older baby boomer customers being replaced by their younger, much more internet savvy offspring, the level of rate manipulation and churning will inevitably rise. Indeed in our work with carriers we've found that direct channels have 20 to 30 points of preventable fraud baked in at the point of sale.

Therefore even Agent-centric carriers must aggressively fight point of sale insurance fraud. If they don't, they'll discover that more and more of their business is 'first year' business.

We're Veracityid (www.veracityid.com) and we'd love to show you how we're saving carriers up to twenty points of loss ratio by eliminating fraud before it's baked into a policy.

Six auto insurance trends driven by the pandemic.

It's been a century since the United States has experienced a pandemic on the scale of the current COVID19 outbreak, resulting in radical short term changes to the economy and people's lifestyles But the crisis is also driving six trends in the auto insurance industry whose impact will continue long after the virus is gone.

Social distancing is accelerating direct channel sales growth. Stuck at home, more customers are transacting their insurance business directly and finding it easy. This is accelerating the already established trend towards direct sales as the baby boomers are replaced by digitally adept customers from the Millennial and Generation Z cohorts.

Cooped up customers are spending more time shopping for insurance. With many families hurting financially and having time on their hands, customers are spending more time and effort shopping for the best rate. They are using aggregators, carrier quotation websites and traditional agents to solicit record numbers of quotations.

Customers are using the knowledge they gained from shopping to manipulate rates more. Modern quotation systems make it easy for customers to run multiple rate scenarios, omitting drivers, switching addresses or transferring commercial vehicles to personal policies. Today's extreme financial pressures and the relative anonymity of direct channels are making these types of frauds more common.

Pressure from direct channel manipulation is pushing agents to cut more corners. To stay relevent, agents need to demonstrate that they can get their customers  'better rates' whether justified or not. This is shrinking the difference in  new business first year loss ratios between the historically unprofitable direct and more profitable agent models. We see both getting worse.

Insurance card scams are proliferating. More customers are  signing up for insurance using a pay plan simply to get the insurance card and and then failing to pay the premium installments. Unless, of course, they have a claim.

Pre-existing damage claims are set to grow. Many consumers are saving money by cancelling their comprehensive and collision coverage. But the casualty events will still happen. This is resulting in even higher levels of 'single vehicle' claims without police reports being filed in the early weeks of new policies.

None of these challenges are new to the industry but the pandemic's extreme isolation and financial distress are making them worse. And the intensifying shift to (largely unprofitable) direct channels means that dealing with these challenges will be more difficult and failing to do so will be more expensive.

idFusion solves these problems. The idFusion Fraud Identification, Intervention and Management platform is designed to identify and resolve these types of frauds in real time at the point of sale regardless of channel. In future posts we will describe each of these fraud challenges and our solutions for them in more detail. Learn more at www.veracityid.com

How to fight fraud and please sales at the same time

One of the biggest issues in insurance fraud prevention is managing error rates. The goal is to detect as many of the frauds hiding in the quote/application stream as possible. But this must be balanced against the risk that false positives will drive away good customers.

The devil is in the Type 1 (false positive) and Type 2 (false negative) error details. Fraud fighters can increase the 'tightness' of their business rules to minimize the chance of missing a fraudster only at the cost of increasing the number of good customers who are mistakenly turned away. It's a tough trade-off but we have developed a solution in our idFusion fraud management platform that allows carriers to substantially reduce false negative errors without losing good customers.

idResolve - our highly flexible, real time, in transaction intervention solution - allows carriers to instantly reach out to customers/agents and resolve false positives, putting good customers back on the path to coverage while causing the dishonest to abandon. Because of this, carriers can tighten their rules to minimize the undetectable false negatives and then using idResolve, sort through and recover the resulting large number of false positives. By doing so we can maximize fraud reduction at the lowest possible cost to customer conversion.

Why is this? When a customer who is attempting fraud is challenged by idResolve they almost always abandon their session, seeking a less observant carrier. On the other hand, when a customer simply makes a mistake they don't abandon. Instead, they welcome help to correct the error. Indeed while we haven't fully tested it, we believe that the best customers actually will be more satisfied with carriers that help them get their details right at the point of sale.

So a major advantage of idFusion over simple "one and done" data panels or black box fraud scoring systems is the ability to radically reduce the error rate by constructive, real time interaction with customers.