Actuarial pricing, capital modelling and reserving

Pricing Squad

Issue 31 -- January 2020

Welcome back to Pricing Squad

Pricing Squad is the newsletter for fellow pricing practitioners and actuaries in general insurance.

Today's issue is about IFRS17.

Get in touch

If you are looking for a contractor or need actuarial help in 2020, get in touch.

Contractual Service Margin in IFRS17

We are drifting in a sea of machine-made information. As an actuary you can teach others how to navigate it profitably.

Here is a recent example. International Financial Reporting Standard #17 introduces a requirement to calculate Contractual Service Margin. CSM is the expected profitability of an insurance policy discounted for time value of money. As long as an actuary provides an estimated claim cost for each and every policy, you can calculate CSM for each and every policy.

However, this can lead to a dangerous "granularity conflict". I talked about granularity conflicts in Pricing Squad 11 so you might already be familiar with this concept.

The problem emerges because CSM is calculated using two separate models with different resolutions (or granularities). The premium model has maximal resolution, providing the exact premium charged for each policy is known. But the claim cost model is only a statistical prediction. Since premiums normally depend on more rating factors (or interactions) than included in claim cost models, your premium model typically has higher resolution than your claim cost model.

Even if the claim cost model can be scored onto each individual policy, the actual information value of this scored prediction is limited to the low resolution of the original claim cost model.

When you combine the two resolutions together in a single CSM, users will be misled. The diagram below explains how.

The top four squares represent premiums and claims for 16 hypothetical policies. The bottom two squares represent CSM.

The claim cost model is built with low resolution (for instance it is based on sum insured and geographical factors but it ignores bespoke underwriting factors). The claim cost model is then scored onto 16 policies resulting in spuriously granular predictions in the red square. This is where deception lurks. These granular predictions are not real. Claim cost predictions can only be trusted on the aggregate level of granularity for which the model was originally developed.

If you simply rely on the red square for your CSM then, for example, policy 14 will appear profitable (green) in the CSM. However, in reality this policy is a member of the low resolution segment C which has lower-than-average premiums and higher-than-average claims.

So CSM is only safe to use for the resolution corresponding to the lowest resolution model used in the CSM calculation. Only provide users with CSM for aggregates corresponding to the low-resolution model.

Copyright © 2020 Jan Iwanik, All rights reserved. You are receiving this email because you subscribed to updates from We publish data and analysis for informational and educational purposes only. You can unsubscribe from this list by emailing us.