On the valuation date (e.g., Dec 31, 2025), the actuary looks at the of the triangle. For Accident Year 2023, we have 36 months of development, but not 48. The reserving actuary projects the "tail" (48 months to ultimate) to calculate the Total Reserve = (Projected Ultimate Loss) - (Cumulative Paid to Date) - (Case Reserves already held).
: Costs to settle claims, divided into allocated (specific to a claim) and unallocated (general department costs). On the valuation date (e
Traditional property and casualty (P&C) ratemaking and loss reserving have historically operated as two distinct disciplines—one forward-looking for pricing, the other backward-looking for liability valuation. However, emerging systemic risks (e.g., climate change, cyber liability, social inflation) and regulatory shifts (e.g., IFRS 17, LDTI) demand a unified, stochastic approach. This paper develops a Dynamic Actuarial Feedback Model (DAFM) that integrates ratemaking and loss reserving into a coherent, real-time framework. We demonstrate that ignoring the interdependence between pricing assumptions and reserve development leads to systematic undercapitalization. Using a combination of Markov chain Monte Carlo (MCMC) simulations for reserve variability and generalized linear mixed models (GLMMs) for rate indication, we prove that a joint calibration reduces the mean squared error of ultimate loss forecasts by 18–27% compared to siloed methods. The paper concludes with a case study on commercial auto liability, showing how the DAFM captures tail dependencies often missed by chain-ladder or Cape Cod approaches. : Costs to settle claims, divided into allocated
Ratemaking is often visualized through the "Fundamental Insurance Equation": This paper develops a Dynamic Actuarial Feedback Model