Building tools to help balance insurance premiums & claims
As a Health Partner, one of our core services at Alan is health insurance. It is, in essence, rather straightforward: we collect premiums from our insured members and reimburse claims. Our goal is to be near equilibrium, what we call the 100% loss ratio.
loss_ratio = sum(reimbursed_claims) / sum(premiums)Every year, we review our contract loss ratios and adjust the premiums when needed to stay near this 100% loss ratio equilibrium. This blog post (in French), covers the topic in more detail.
While we review the loss ratios every year, we are committed to updating the prices as rarely as possible — so ideally not every year, and to making price changes as foreseeable as possible for our customers. In this article, we are exploring the different levers we use to uphold that commitment.
Striving for price stability
The first step to limiting the number of price updates we do is of course to keep improving our pricing algorithm. We update it every year! This way, when a customer first subscribes to an insurance product, their actual loss ratio is as close to 100% as possible.
However, as accurate as our pricing may be, it is very probable that there will be small variations in the customer health consumption every year. Rather than modifying the prices every year to strictly stick to a 100% loss ratio, we introduce the concept of loss ratio interval:
- We only change the prices if the loss ratio is outside of this interval.
- We use the loss ratio since the last price update (so over several years) rather than the yearly loss ratio.
- The interval becomes smaller and smaller the more years have passed since the last price update.
This loss ratio over a longer period and the interval becoming smaller reflect the fact that, with more history, we are more confident that the loss ratio is representative of the customer’s health consumption, and so we are more confident that prices should be adjusted. This other article of ours dives into the details of our repricing logic.
Finally, we also take into account that exceptional claims, i.e. rare and expensive ones, may have exceptionally degraded the loss ratio. We favor keeping the same prices over increasing them to account for exceptional illness.
Proposing alternative scenarios to customers
When the loss ratio is too far above 100%, rather than increasing the price, we can also decrease the cost of reimbursed claims. In practice, that means changing the health guarantees.
Traditionally, that would mean involving actuaries to do complex computations on a case by case basis and that doesn’t scale well. Plus, account managers are the ones in direct contact with customers, not actuaries. We then decided to build a repricing-specific tool on top of our existing pricing tool, that allows account managers to autonomously build and propose different renewal scenarios to customers.
In this new tool, the renewal scenario becomes the core concept on which account managers collaborate with customers. A scenario is essentially an all-in-one proposal, summarizing all the information relative to all the contracts for one account for the current year and the evolution for the next year. And as the name suggests, you can have several scenarios for each account, so the customer gets to choose what they prefer.
To help account managers get started on each account, we automatically create a base scenario with the suggested new prices to reach a 100% loss ratio without guarantee changes. They can then create copies from the base scenario (or from another scenario they may have already created and modified) and make additional guarantee changes. When they change the guarantees, we automatically recompute the new prices for them.
Account managers can export scenario comparisons into spreadsheets and make several propositions to the account administrators. Once they agree on a scenario, account managers can click on a single button and be redirected to our contracting tool to generate contract amendments already prefilled with the new guarantees and prices.
Finally, by tracking the latest exported scenario for each account, or the scenario approved for contracting, we can know the global state of renewal. We leverage our Data Analytics system to re-aggregate the information from all the account most up-to-date scenarios and we are able to compute the global next year projected loss ratio for all of Alan’s customers. Thanks to this, we are able to build a global real-time report on the sanity of Alan’s portfolio and take quick corrective actions when needed.
Here is a simplified data model of our scenario system:
- There can be several scenarios per account. Within a scenario, there is one renewal offer per contract, so there can be several offers per scenario.
- Rather than working directly with contracts and computing metrics in the backend, contract metrics are provided by our Data Analytics system. Metrics include the base_new_price, i.e. the price required to reach the 100% loss ratio without any guarantee change.
- We store potential changes to the health guarantees and the new price resulting from those guarantee changes at the offer level. We leverage our existing pricer to compute the guarantee_change_cost_increase:
guarantee_change_cost_increase = pricer(new_guarantees) / pricer(current_guarantees)
renewal_offer.new_price = contract_metric.base_new_price * guarantee_change_cost_increase- Note that guarantee_change_cost_increase can be < 1 if the guarantees are decreased, so the effective renewal_offer.new_price can be smaller than the initial contract_metric.base_new_price.
What’s next?
Our repricing process is representative of Alan’s leadership principles:
- We strive to be member-first: we propose alternatives to our customers to let them choose what is best for themselves.
- We are radically transparent: our members know what their loss ratio is and how we use it to modify the insurance prices over the years (and now you know too!).
- We distribute ownership and empower Alan employees accomplish their missions: account managers can autonomously craft alternative proposals for their accounts and don’t need to be insurance experts to do so.
- We simplify: our customers can grasp the principles of our renewal pricing method, independently of their level of technical or actuarial knowledge.
And lastly, a part of Alan’s leadership principles that I personally like a lot is “we haven’t won yet” and “Everything is in draft”. We didn’t build our repricing system in one shot but iterated on it over several years. The version we have today is almost certainly not final. As a team, we always strive to take things to the next level, not for the sake of it, but for the value it creates.
If those values resonate with you, please drop us a line at jobs@alan.com. We’re hiring!
