PlumBot: Bot: Creating new article from JSON

2026-03-13T12:30:44Z

Bot: Creating new article from JSON

New page

📉 '''Frequency and severity''' is the paired analytical framework that forms the backbone of [[Definition:Loss | loss]] modeling, [[Definition:Ratemaking | ratemaking]], and [[Definition:Reserving | reserving]] in the insurance industry, decomposing total incurred losses into two components: how often claims occur ([[Definition:Frequency (insurance) | frequency]]) and how much each claim costs ([[Definition:Severity (insurance) | severity]]). Rather than analyzing aggregate losses as a single figure, this decomposition allows [[Definition:Actuary | actuaries]] and [[Definition:Underwriter | underwriters]] to isolate the distinct drivers behind loss trends — a capability that is essential because the factors influencing claim count often differ fundamentally from those affecting claim size. Virtually every [[Definition:Insurance carrier | insurer]] globally, regardless of line of business or regulatory jurisdiction, organizes its loss analysis around this frequency-severity split.

🔧 The mechanics begin with segregating historical [[Definition:Insurance claim | claims]] data into counts and amounts, then fitting statistical models to each component separately. Frequency is typically modeled using count distributions — Poisson, negative binomial, or their zero-inflated variants — while severity is modeled with continuous distributions such as lognormal, gamma, Pareto, or Weibull, depending on the tail behavior of the loss distribution. The expected total loss for a portfolio is then the product of expected frequency and expected severity, a relationship that holds at the portfolio level and can be further segmented by rating class, peril, geography, or policy year. This separation proves especially valuable when trends diverge: for example, in [[Definition:Auto insurance | motor insurance]], improving vehicle safety technology may reduce frequency while rising medical costs and litigation trends push severity higher, producing a net loss trajectory that only the frequency-severity lens can properly diagnose. [[Definition:Solvency II | Solvency II]] in Europe, [[Definition:IFRS 17 | IFRS 17]] globally, and U.S. statutory reserving standards all implicitly or explicitly rely on frequency-severity decomposition in their approaches to calculating [[Definition:Technical provisions | technical provisions]] and [[Definition:Risk margin | risk margins]].

💡 Beyond its actuarial mechanics, the frequency-severity framework underpins strategic decision-making across the insurance value chain. [[Definition:Reinsurance | Reinsurance]] purchasing decisions hinge on whether a portfolio's risk profile is driven by high-frequency, low-severity attritional losses or by low-frequency, high-severity catastrophic events — the former suggesting [[Definition:Quota share reinsurance | quota share]] structures and the latter pointing toward [[Definition:Excess of loss reinsurance | excess of loss]] protection. [[Definition:Claims | Claims]] departments use the framework to allocate investigative resources, focusing fraud detection efforts on lines where frequency anomalies appear and deploying specialized large-loss adjusters where severity is the dominant concern. For [[Definition:Insurtech | insurtech]] companies building [[Definition:Predictive analytics | predictive models]], frequency and severity serve as the target variables around which feature engineering and algorithm selection revolve. In short, this deceptively simple two-part framework remains the most powerful analytical lens the insurance industry possesses for understanding, pricing, and managing risk.

'''Related concepts:'''
{{Div col|colwidth=20em}}
* [[Definition:Frequency (insurance)]]
* [[Definition:Severity (insurance)]]
* [[Definition:Ratemaking]]
* [[Definition:Actuarial science]]
* [[Definition:Loss ratio]]
* [[Definition:Excess of loss reinsurance]]
{{Div col end}}

Definition:Frequency and severity - Revision history

PlumBot: Bot: Creating new article from JSON