Motor Pricing: Limited Telematics Evidence and a Commercial Request
A mid-sized motor insurer has completed a re-rating exercise. The current combined ratio is 108%, and the technical pricing model indicates that an average rate increase of 11.5% is needed to reach a 96% target. A telematics-based discount scheme launched 18 months ago covers only 9,000 policies, of which 1,100 have claims experience. The telematics coefficient is based on this small sample and is statistically significant only at the 90% level. The distribution director argues that any rate rise above 6% will cause a 20% loss of broker volume and wants the report to the pricing committee to present only the 6% option. The regulator requires the annual rate filing in six weeks and expects evidence of model validation. The actuarial team has also found that 3% of telematics data is missing trip data because of a vendor outage in the second quarter.
1. How should the actuary handle the distribution director's request to present only the 6% option?
Read the worked answer
Reasoning. The distribution director is asking the actuary to suppress an option that the pricing committee needs to see. A 6% increase leaves the combined ratio above 100% on the technical view and may worsen solvency. Presenting only that option would present a commercial preference as a technical conclusion.
Response. Explain to the director that the report will present a range, for example 6%, 9% and 11.5%, with the volume, combined-ratio and capital consequences of each. Invite the director to provide evidence for the 20% volume loss so that it can be included as a stated assumption. If the director insists, escalate to the pricing committee chair and record the request.
Communication. Frame the 6% option fairly: it preserves volume but accepts underwriting losses and may require a larger increase later if claims inflation continues. The committee can choose it, but must do so with the consequences visible. This protects the actuary's objectivity and gives the committee an auditable decision.
2. Should the telematics discount be included in the rate filing, and what evidence should support it?
Read the worked answer
Should it be included? Yes, but with explicit caveats. The discount is already in the market and affects competitiveness; ignoring it would misstate the required rate. However, 1,100 claims is a thin sample, and a 90% significance level is weak evidence.
Evidence to present. (1) A credibility-weighted estimate blending telematics and non-telematics experience. (2) Out-of-sample or hold-out validation. (3) Sensitivity of the coefficient to the missing 3% of trip data. (4) A monitoring plan with triggers to withdraw or reprice the discount. In the rate filing, disclose the sample size, the confidence level and the data limitation rather than presenting the coefficient as precise.
Recommendation. Keep the discount but phase it in at a reduced level and review after 12 months. That balances competitive pressure against the risk of mispricing based on thin data. The filing should explain the judgement, not just the number.
3. What rate change would you recommend to the pricing committee, and how would you justify it?
Read the worked answer
Recommendation. I would recommend a segmented increase averaging 9-11%, not the full 11.5% across the board, with larger increases in segments where the model is well supported and smaller increases where competition is strongest. Reasons: the combined ratio of 108% means the current book is loss-making; a 6% rise is unlikely to restore profitability; but a single 11.5% rise may cause more volume loss than the model assumes because the volume elasticity is uncertain.
Presentation. Show three scenarios with combined ratio, volume and capital impact. State the assumptions behind the volume response. Flag that the telematics discount is under-evidenced and recommend a review. Ask the committee to approve the principle and delegate final segment rates within a stated range. This gives a decision that is defensible to the regulator and robust to being wrong.