Credibility Theory and Distribution of Losses - NU

1. Severity & Frequency ModelsClaim Severity: Modeling the monetary size of claims using continuous distributions like Pareto, Gamma, Weibull, and Lognormal.Claim Frequency: Counting the number of claims using discrete distributions like Poisson, Negative Binomial, and Binomial.📉 2. Policy Modifications & Aggregate LossCoverage Adjustments: Calculating how deductibles, policy limits, and co-insurance change the insurer's net payout.Aggregate Losses: Combining frequency and severity into compound models to calculate total expected portfolio losses.⚖️ 3. Classical Credibility TheoryFull Credibility: Finding the statistical threshold needed to base future premiums entirely on a group's own past data.Partial Credibility: Calculating the weight factor (Z) to blend local group experience with broader industry averages.🔮 4. Advanced Credibility ModelsBayesian Credibility: Updating prior risk beliefs with new claim data to find posterior expected losses.Bühlmann & Bühlmann-Straub Models: Estimating variance parameters to find the mathematically most accurate premium weight.