Definition:Compound distribution
📊 Compound distribution is an actuarial modeling construct used in insurance to represent the total aggregate loss from a portfolio by combining two separate probability distributions: one for the number of claims (the frequency distribution) and another for the size of each individual claim (the severity distribution). Rather than attempting to model aggregate losses directly — which would require capturing both the randomness of how many events occur and how large each one turns out to be in a single function — the compound distribution decomposes the problem into its two natural components. This approach is foundational in actuarial science and underpins pricing, reserving, reinsurance structuring, and capital modeling across virtually every line of business.
⚙️ The mechanics are straightforward in principle: the actuary selects a frequency model — commonly a Poisson, negative binomial, or binomial distribution — to describe the random number of claims, and a severity model — such as a lognormal, Pareto, or gamma distribution — to describe the random size of each claim. The aggregate loss is then the random sum of a random number of severity draws. Because the resulting compound distribution rarely has a neat closed-form expression, actuaries evaluate it through recursive methods (such as the Panjer recursion), fast Fourier transforms, or Monte Carlo simulation. In practice, this framework drives excess of loss reinsurance pricing — where understanding the tail of the aggregate distribution determines the price of high-layer coverage — and feeds into regulatory capital requirement calculations under frameworks like Solvency II's internal model approach and the NAIC's risk-based capital standards.
💡 The compound distribution's power lies in its modularity: frequency and severity can be estimated, validated, and updated independently as new data emerges, which gives actuaries flexibility when portfolios grow, mix shifts, or external conditions change. It also allows explicit modeling of catastrophe risk scenarios — for example, by introducing a secondary frequency distribution for large-event counts alongside a heavy-tailed severity component. Understanding the properties of compound distributions is essential for anyone involved in pricing, loss reserving, or enterprise risk management in insurance, because misspecifying either component — underestimating claim frequency or using a severity distribution with too thin a tail — can lead to inadequate premiums, deficient reserves, or insufficient capital. As computational power has expanded, compound distributions have become the building blocks of increasingly sophisticated stochastic models that regulators and rating agencies expect insurers to maintain.
Related concepts: