Chain-Ladder Method

Chain-ladder is the classic development method: use how claims historically grew from one maturity to the next, then apply those growth patterns to newer accident or report years that are not fully developed yet.

It is popular because the mechanics are simple and the method turns a claims triangle into a projection engine. Its weakness is that it leans heavily on the idea that historical development patterns remain useful for the unfinished years.

Suppose an accident year has cumulative reported losses of 800 at the latest observed age. If the remaining age-to-age development factors multiply to 1.25, then the projected ultimate is 800 x 1.25 = 1,000 and the indicated reserve is 1,000 - 800 = 200.

That is the chain-ladder logic in miniature: observed cumulative amount times remaining development equals projected ultimate, and reserve is whatever has not developed yet.

Exam 5 and ASTAM both treat development-triangle logic as foundational because it gives a direct way to organize loss data, spot changing development behavior, and project unpaid claims.

In real work, chain-ladder is often a baseline method. Even when another reserving method is preferred for the final estimate, chain-ladder is still useful as a diagnostic reference point.

A common mistake is using the method mechanically without checking whether the triangle has shifted because of changes in case reserving, claims handling, large losses, mix of business, or reporting speed. The method assumes the historical development pattern is still informative.

Another mistake is treating chain-ladder as a truth machine instead of as one view of the reserve problem. It works best when paired with diagnostics and alternative methods like Bornhuetter-Ferguson.

One useful way to think about chain-ladder is as structured forecasting under delayed observation. You are not predicting a raw future from scratch. You are forecasting how incomplete observations are likely to mature based on past development behavior.