Dynamic pricing was one of our very first thoughts as an add-on: how do you develop mechanisms that create risk-specific loans prices. This also expands across to health insurance, wealth advisory and other financial services.
The Problem: charging different interest rates for similar loans based upon borrower risks isn’t easy.
This sounds perfect. Use a Big Data engine to price loans that has parallels to Advanced IRB credit risk models (how banks measure risk themselves), insurance risk or other financial modelling on a per customer basis. Lower risk, by our own metrics, would lead to a lower interest rate or premium. However, we quickly decided it wouldn’t work and we have several reasons why:
- Customers will shop around
You are naïve to think that customers will not. They will, either through the lazy or delegated method (comparisons sites, mortgage brokers), deep research (gathering quotes from different lenders) or talking to people casually (this still happens!). Your risk specific pricing starts to fall apart as each lender has a separate and unique risk model. They do: look at the various bank Advanced IRB credit or statistical risk models and variation in what drives risk is evident. As a lender, in a dynamic pricing universe, you’re mostly guaranteed to attract the customers where you have underestimated the risk (you offered them the cheapest loan).
- Financial resources are not finite
Price discrimination works well when the product or service runs out. Time is a great enabler (e.g. I must fly to Ireland on the 17th March), meaning you can discriminate. Loans, on the other hand, don’t always have time maturity, so the price doesn’t have to be monotonic increasing with an inverse time decay.
Borrowers can simply wait until prices fluctuate down or a lower offer comes along. All loans have mostly marginal cost (whereas flights or cinema tickets are mostly fixed and perishable).
- Customer segregation to remove the deadweight loss is harder in finance
Cinema ticket pricing is an easy starting point. Retirees and students, often with a lower ability and willingness to pay, obtain discounts whilst affluent professionals have to pay higher prices. This allows separate pricing. It is structurally much harder to split up pricing based upon risk: the customers cannot say to each other “hey, I’m a student, you’re not”. The division between low and high risk is often based upon wealth, which creates a dangerous PR mix and may not be legal. Friends also talk and they will quickly get annoyed with variability in offers as being unfair.
- People will think harder for the larger transactions
Mortgages and wealth products, where the accumulated interest charges, fees or potential outcomes are of a higher order than small items, are harder move pricing dynamically. Customers will do a greater level of research. Small items, such as annual car insurance premiums, are easier to play around with and may be purchased in one sitting. Less thought, more ability to be dynamic.
Look at the market
Aside from these above items, market structure can play a significant part. Is pricing determined on day one? Can it be changed over time by the lender (such as Australian home loans), by a formula (e.g. bank covenants) or do prices have annual adjustments (such as Australian health insurance)? These vary from country to country and have a similar impact as tax treatment does for wealth advisory.
Finally, financial products are also influenced by macro factors as well as micro. Long-term interest rates and business cycles play a very large part in this. Bank regulatory capital, a key part in loan pricing, should be measured on through-the-cycle metrics (such as through-the-cycle observed probability of default) whilst short term dynamic pricing may be heavily influenced by point-in-time. In other words, you shouldn’t be too dynamic as long-run trends are very important and pricing in the short-run will likely lead to problems if recession arrives.
There are several ways to break down the above problems. Deciding what you show your customers and what you don’t is a key part. You also need to pick the right market and right product to operate in.
Solution coming later this week…
Picture from https://en.wikipedia.org/wiki/Deadweight_loss