InsurTech’s big ticket items: machine learning on the blockchain

Chris Skinner , 07 Sep 2017 - Blockchain AdoptionMachine LearningOpinion

Chris Skinner is Chair of the European networking forum: the Financial Services Club. He is best known as an independent commentator on Fintech through his blog, and as author of the best selling book Digital Bank and its new sequel ValueWeb.

I don’t usually tell people that I’m a qualified expert in insurance as it’s not relevant, but I’m mentioning it today as I just presented at a private insurance event. There are lots of interesting nuances in insurance. You and I probably think it’s just that once a year renewal of our auto policy, or maybe the regular premiums we pay for life and pensions. That’s an important part of it and is a challenge as people don’t buy insurance, it has to be sold. Think about it. You don’t wake up and think “oh, I might die today! Better get some life insurance”, but you learn through advisory sales that this is what you need to do.

Then there’s a whole other raft of insurance where the real money is made: commercial insurance. Insuring ships, aircraft, offices and employees is where the big-ticket policies arrive. These are more complicated, as the risks have to be calculated in more depth. Under assess the risk and the insurance company loses; over assess and a competitor might steal the business. Equally, the corporate relationship becomes important in this space, as that’s where the strength of understanding the client comes from.

It is not as simple a business as you may first think. For example, how would you calculate the risk of terrorist attack on an aircraft and what price would you put on that policy? Yep, it’s not easy, so insurers have a lot of humans called actuaries who work these things out using math. Now, you know where I’m going with this. Artificially intelligent actuaries.

If my CFO can be an algorithm then so can my actuary and, more and more, is becoming one. In fact, I spoke at an insurance event with Earnix last year and so, rather than me waffle on around AI insurance, I will refer you to their blog to learn more.  They also published some interesting research about how AI will change the insurance industry over the next decade.

Some of the key findings of the survey include:

  • Machine Learning adoption is becoming a significant trend, used by 54% of all respondents and 60% of those from companies with over 5,000 employees.

  • 90% of the survey respondents that use Machine Learning use it in pricing and product management, a result that is likely impacted by a large portion of the respondents coming from these functions.

  • When asked about the top benefits realised by their organizations, as many as 57% of the respondents cited greater analytical accuracy as one of the top three benefits.

  • The most significant promise and expected benefits of Machine Learning moving forward is in greater automation, productivity, and cost savings.

  • Despite the growing adoption, the vast majority of respondents admit significant knowledge gaps when it comes to Machine Learning.

Very good.

I must admit that AI and machine learning for risk management is probably one of the top, if not the top financial application. But that’s more to do with financial institutions’ fragmented data structures that block them from a single customer view than saying it is the best application.

The other big impact technology for insurance for me is blockchain. Yea, sure, blockchain is boring but it’s still going to change the world eventually. I truly believe it is the most underestimated but overhyped technology out there. In fact, it’s all the hype that has made it so boring, as it hasn’t delivered … yet. But when you see the first Ethereum Unicorn appearing and ICOs for new cryptobanks, then you know something is going on and, in insurance, blockchain just makes sense.

This is because one of the underlying principles of insurance is uberrimae fidei, utmost good faith, which is that the insurer asks you a range of questions when you take out the policy or make a claim, and has to trust you are telling the truth. In insurance, it is very different to most industries, in that other industries work on the principle of caveat emptor, buyer beware. If you get screwed by someone who sold you a pup, well, it’s your fault. In insurance, it’s the other way around. They have to trust you, so seller beware. It is why they require full disclosure at the outset and can refuse to pay if you lied. That is their only protection.

Another basic insurance principle is insurable interest: if you are insuring something, you have to show that you have a vested interest in that something. This arose in the 17th century when people took life insurance policies out on the Royalty and celebrities of Elizabethan Britain, on the basis that they thought there was a good chance they might die. It was an old form of betting and as the industry matured, the law changed to say that you could only insure a life that was related to you: a spouse, parent or child; you could not just take a punt on the chance that your grizzly old neighbour might pop it this year.

These two core principles of insurance are so overwhelmingly pervasive for a blockchain application that it almost makes me drool. This is because (a) I can immediately find out through your authenticated data whether you really are 31 years old with a clean driving license, living at the address you entered and with the car you state; (b) I can therefore feel comfortable that you are telling the truth and that it belongs to you.

In other words, blockchain deals at the heart of two of the six core principles of insurance. No wonder the largest insurance companies have formed a blockchain alliance.

The Blockchain Insurance Industry Initiative B3i was launched in October 2016 to explore the potential use of distributed ledger technology, and has 15 members: Achmea, Aegon, Ageas, Allianz, Generali, Hannover Re, Liberty Mutual, Munich Re, RGA, SCOR, Sompo Japan Nipponkoa Insurance, Swiss Re, Tokio Marine Holdings, XL Catlin and Zurich Insurance Group. (The original five members were Aegon, Allianz, Munich Re, Swiss Re and Zurich.)

Again, rather than me talking through the opportunities for blockchain in insurance, Zurich Insurance Group recently wrote an interesting summary as part of B3i:

The widely-held hope is that blockchain technology could deliver a shared and transparent record of industry-wide contract and claims-related information.

While smart contracts and claims automation are technically feasible using alternative technology, the addition of blockchain adds a level of transparency and trustworthiness.

Crucially, blockchain can remove the need for a central authority to control operations, and places the customer in control of the relationship with the service supplier.

For example, automated claims-handling processes based on blockchain technology, smart contracts and publicly available data can be developed. When a certain set of conditions is met, such as a flight being cancelled, the contract can be resolved automatically, without requiring additional assessment.

Not only does this approach increase customer trust, the technology has the potential to streamline communications and transactions by allowing for automatic data transfers in seconds.

“From a customer perspective, they should be getting a better service if we can operate and process claims more efficiently,” adds Alessandro.

Fraud detection and pricing

One of the most exciting possibilities for the insurance sector is the possibility that blockchain could assist in creating an incorruptible, international and cross-industry database that can be used to flag possible signs of insurance fraud.

Given that the ABI estimates that fraud adds an average of £50 to every UK policyholder’s insurance bill, the potential for blockchain to reduce the chances of fraud is intriguing.

The database could be used to detect identity fraud, and to check claims history and police reports, alongside a host of other information that may be required from insurers and brokers when dealing with contracts and claims.

In this manner, contracts and claims could be recorded onto the blockchain, ensuring that only valid claims are made and multiple claims for one accident are rejected. The addition of smart contracts can also ensure that payments are triggered when certain conditions are met and validated.

Reducing admin costs

On a practical level, there is the potential to reduce the cost of insurance by simplifying transactions and minimising the administration burden. By storing contracts and transactions on a shared ledger, not only is contract consistency and execution ensured, but the administrative burden is reduced for multiple stakeholders.

Automation of workflow, while reducing the administration burden, also helps reduce the chances of data entry duplication, and the disputes and delays this can lead to. In addition, blockchain can save time by opening up the potential for the automation of policyholder identity verification and contract validity checking.

“This should result in less administration and effort for brokers”, explains Alessandro, adding that, “the aspiration is that these savings will trickle down all the way through the value chain.”

Cyber security

Another of the main advantages of blockchain, which should help to allay fears of personal data loss, is that the technology is secure by design. The decentralised ledger, which serves to record transactions across multiple computers, removes the central hub that functions as an obvious target for attack.

Crucially, this means that data in a ‘block’ cannot be retroactively altered by any party, ensuring consistent and transparent contracts. According to Alessandro, this means that: “Blockchain can deliver faster cycle times and reduce disputes and errors.”

This presents a significant advantage, as the reconciliation following a major claim can be expensive and time consuming if disparities in the fine print are uncovered following a claim. In the aftermath of 9/11, disparities led to an insurance dispute of $3.5bn.

Ultimately, blockchain technology has the potential to transform how transactions, policies and claims are recorded and reconciled. If B3i ultimately proves the validity of the technology, blockchain could serve to reduce error rates in contracts and claims, boost security and deliver significant cost reductions across the insurance value chain.

All in all, times are very interesting I guess, especially for insurers.

Benediximus amicis meis epularer (so old school these insurance chaps).