The annual renewal is the moment when a carrier has the most direct control over its book composition. A policyholder who is unprofitable can be repriced or released. A policyholder who is profitable can be retained at a margin that reflects their value. What most carriers do instead is apply a uniform rate change — a percentage increase driven by aggregate loss trend — that treats every policyholder identically regardless of their individual loss profile, their price sensitivity, or the competitive alternatives available to them.
The consequence is predictable. Profitable policyholders with good loss histories and multiple competitive alternatives are the most price-sensitive. They shop at modest increases and leave when they find a better rate. Unprofitable policyholders with adverse loss histories or limited alternatives are less price-sensitive. They accept increases that would drive better risks away. A uniform renewal increase applied to improve overall margin tends to do the opposite: it accelerates the departure of the policyholders the carrier most wants to keep and retains the ones it should be repricing more aggressively or not renewing at all.
Where the renewal pricing decision breaks down
Renewal pricing models typically work at the segment or rate class level. A rate change is filed for a class, applied to all policyholders in that class, and the expected improvement in loss ratio is modelled against the expected attrition. What is missing is individual-level price elasticity — the degree to which each specific policyholder will respond to a rate increase with a lapse decision.
Elasticity varies considerably even within the same rate class. A policyholder who has been with the carrier for twelve years, has a claims-free history, and lives in an area with limited competitive alternatives behaves differently at renewal than a policyholder of similar loss profile who switched carriers twice in the last three years and has three active quotes. Both sit in the same actuarial segment. They have very different price sensitivity.
A model that estimates individual-level price elasticity — drawing on tenure, prior shopping behaviour where observable, competitive market density, policy characteristics, and renewal interaction history — allows the carrier to calibrate the renewal offer to what each policyholder will accept. The profitable inelastic policyholder receives a rate that reflects the margin the carrier needs. The profitable elastic policyholder receives a more competitive offer that retains them. The unprofitable policyholder receives a rate that either improves the margin or triggers a lapse that the carrier can absorb.
The book quality compound effect
The renewal pricing decision is not a one-period optimisation. Its consequences compound over time through the same adverse selection mechanism that pricing precision addresses in new business. A carrier that consistently prices renewals in ways that drive away profitable policyholders while retaining unprofitable ones builds an increasingly adverse book quality over successive renewal cycles. The loss ratio deteriorates gradually. The combination of rate action and elasticity-aware pricing addresses the problem at its source rather than chasing it through successive rounds of inadequate rate change.
Renewal pricing optimisation works best as a complement to the non-renewal scoring described separately. The two models together produce a complete renewal strategy: retain the profitable elastic risks at competitive rates, retain the profitable inelastic risks at adequate rates, reprice the marginally unprofitable risks to an adequate level, and non-renew the genuinely unprofitable risks that cannot be made profitable at a rate the market will sustain.
The technology dimension
Individual-level renewal pricing models require access to the full policy history, claims record, and behavioural signals for each expiring policy. For carriers running their policy administration on IBM Z, deploying the elasticity model via IBM Machine Learning for z/OS makes the individual renewal price recommendation available within the renewal processing workflow, drawn from the complete policy and loss data on the same platform. The recommendation is returned at the point of renewal offer generation, not as a separate offline exercise that has to be manually applied.
What success looks like
The metrics are renewal retention rate by loss ratio decile, renewal book combined ratio, and rate adequacy on the retained book. The retention rate by loss ratio decile is the most direct indicator: the model is working when retention in the lowest loss deciles is high and retention in the highest loss deciles is low. A uniform increase produces similar retention rates across deciles. The divergence of those retention profiles is the evidence that elasticity-aware pricing is changing the composition of the retained book.
