P&C Insurance Analytics

French Motor Third-Party Liability — portfolio profitability and risk segmentation

Methodology

Analysis of 678,013 policies and 24,944 claims from the freMTPL2 actuarial benchmark dataset (Charpentier 2014). The dataset contains policy characteristics and claim experience but no priced premium, so I assumed a flat premium rate of €234/exposure-year — equal to 1.4 × portfolio pure premium €167 (a standard 40% loading for expenses, profit, and risk margin). Combined Ratio adds a flat 28% Expense Ratio, the P&C industry average.

This calibration makes the portfolio break even at the global level by construction. The interesting analysis is therefore in relative profitability — which regions, driver-vehicle segments, and bonus-malus bands deviate from the benchmark.
Data source
freMTPL2 (Charpentier 2014)
Records
678,013 policies
Built with
Python · pandas · Plotly

Portfolio Overview

Regional Profitability

6 regions are loss-making (Combined Ratio > 100%), worst at R21 (199%) — claims nearly 2× premium income. These are candidates for premium re-rating or exposure reduction. The 16 profitable regions average 80% CR, suggesting the book underlying the underwriting is solid where it's been priced correctly.

Risk Segmentation

The 18–25 driver row dominates the worst cells, ranging from 115% to 836% Loss Ratio across all vehicle ages — confirming the textbook young-driver risk pattern. Profitability stabilises sharply from age 26+, with the best segment performance (34%) found in older drivers with newer vehicles. The young-driver tier is a structural pricing target, not a fixable operations issue.

BonusMalus Distribution

63% of policy-years sit at BM = 50 (the best-driver tier), pointing to a healthy retention book of claim-free veteran customers. The 1% of exposure above BM 100 (the malus tail, in amber) represents drivers with claim history — a candidate group for either premium uplift or non-renewal review.