Back to the Future: Prospective solvency capital allocation using predictive dynamic mixture of copulas-wavelet combinations

In this study, we propose a dynamic and forward-looking model for solvency capital allocation in insurance companies, combining convex dynamic mixtures of dynamic copulas with multiresolution analysis through wavelets. Our model simultaneously captures the temporal variation of both dependence parameters and copula weights, while also enabling forecasts of the Solvency Capital Requirement (SCR) over different time horizons. This approach overcomes the limitations of standard models, which assume static dependence structures, by incorporating complex, time-varying dynamics that are sensitive to market shocks and extreme events. We apply the model to real data from a large Brazilian insurer, using publicly available microdata from SUSEP. The analysis focuses on the “pricing” submodule of the “non-life underwriting” module, accounting for dependence among incurred claims across major business lines. The results show that the standard model tends to overestimate capital requirements by approximately 13%, whereas our dynamic model provides more parsimonious and risk-aligned estimates. Moreover, the proposed framework reacts prospectively to systemic shocks, adjusting required capital during periods of heightened claims severity. The contribution of this study is two-fold. Methodologically, it advances the theory of dynamic copulas by allowing simultaneous variation in parameters and weights. Practically, we offer a quantitative tool that improves capital calibration, strengthens prudential resilience, and supports strategic decision-making by managers, regulators, and insurers, serving as technical basis for insurers to develop internal models.