Kinsale Insurance

Kinsale provides a useful carrier-side benchmark for the MGA market because it has built many of the capabilities delegated underwriters claim as advantages—specialized teams, fast response, granular data and a willingness to write difficult risks—while retaining underwriting, claims, reserving and capital responsibility inside one insurer. Kinsale does not grant underwriting authority to brokers, agents or program managers. Every submission is controlled by its own underwriters and every claim is managed in-house. That structure costs more fixed infrastructure than renting a balance sheet, but it removes one of delegated insurance’s hardest governance problems: the separation between the party selecting risk and the party absorbing loss. The 2025 results show the economic power of the model. Audited gross written premium reached $1.98 billion, net earned premium was $1.58 billion and the combined ratio was 75.9%, including a 55.1% loss ratio and 20.8% expense ratio. Net income was $503.6 million and operating return on equity, a company-defined non-GAAP measure, was 26.4%. These are unusually strong results, but they should be read with the reserve detail rather than as a single headline. The loss ratio benefited by 3.9 points from favorable prior-year development. Current-accident-year losses were 57.1%, plus 1.9 points of catastrophe loss. Older construction-liability years developed adversely even as more recent years and property lines released reserves, a reminder that long-tail casualty success remains provisional for years. Kinsale’s underwriting strategy concentrates on small and midsized accounts where competition is lower and policy terms can be tailored. Average premium per policy was $13,400 in 2025. The company processed roughly 751,000 requested new-business submissions after excluding unsolicited cross-sell offers, quoted 75% and bound 9.6%. That funnel illustrates why workflow efficiency matters: a carrier can examine hundreds of thousands of opportunities while binding only a small minority. Speed is commercially valuable only if the system preserves risk selection, coverage precision and price. Kinsale’s proprietary platform centralizes submission, policy, claims and reserve data and supported about 340 underwriting employees, 90 claims professionals and 140 technology employees and contractors at year-end. In 2025, it expanded AI-enabled tools across the company while warning in its regulatory filing that generative systems can fabricate information, introduce bias and lack source traceability. The governance test is whether AI reduces administrative work without silently changing appetite, pricing or claim decisions. Distribution creates a different concentration. RSG Specialty produced 18.8% of 2025 premium, AmWINS 17.1% and CRC 11.7%; together the three accounted for 47.6%. Kinsale paid average broker commission of 14.8% and argues that fast service and a direct underwriting model let it compete without program-manager economics. Large wholesalers can deliver broad access and high submission flow, but they also hold meaningful negotiating leverage. A weakening relationship, changing broker incentive or shift toward another market could affect growth. Geography is similarly concentrated: California, Florida and Texas represented 49% of premium. Those states mix catastrophe exposure with litigious casualty venues, making portfolio aggregation and terms as important as account-level selection. The company’s book was 70.7% casualty and 29.3% property in 2025. Commercial property premium fell 17.9% as rates declined and admitted competition returned, while excess casualty, general casualty and small-business property grew. This is a significant signal of cycle response: Kinsale allowed property volume to contract rather than chase softened pricing. At the same time, net retention rose to 81.7%, increasing the share of underwriting result held on its own balance sheet. Higher retention can improve economics when pricing and reserves are right, but it increases sensitivity to adverse development and catastrophe loss. For MGA and capacity executives, Kinsale’s lesson is not that delegation is inferior. It is that specialized risk selection becomes more durable when authority, data, claims feedback, expense discipline and capital consequence sit close together. A delegated model must recreate that alignment contractually and operationally. The indicators worth watching are current-accident-year loss ratio excluding catastrophe, reserve development by vintage and line, casualty rate change versus loss trend, quote and bind conversion by cohort, policy wording and endorsement error, claims caseload and closure, broker concentration, commission movement, premium and PML by state, net retention, reinsurance cost, underwriter productivity and whether AI-supported workflows improve consistency without weakening human accountability.