Aderência de tábuas de mortalidade versus ajustes de leis de mortalidade

Abstract

This study assesses the impact of choosing between theoretical life tables—validated through statistical tests—and tables adjusted by parametric mortality laws (Gompertz, Makeham, and

Heligman-Pollard) on the pricing of actuarial annuities, using data from Chilean life tables B- 2006 H and RV-2004. It employs Kolmogorov-Smirnov, chi-square tests, and error metrics

(MAE, RMSE, R2) to evaluate model adherence to observed data. Parametric laws are fitted to sex-segmented death probabilities using maximum likelihood estimation, with optimal models selected based on AIC and BIC criteria. Actuarial annuity values from both approaches are compared using Mann-Whitney U tests and Cohen’s d statistic to assess significance and practical magnitude. Two scenarios varying temporariness (n), deferment (m), and age

differences among insured individuals are analyzed, revealing that, despite statistically non- significant differences according to classical tests, practical effects are substantial, particularly

for products with long deferment periods and multiple lives. The study concludes that the selected actuarial methodology significantly influences pricing, with parametric adjusted models providing enhanced accuracy, fairness, and financial sustainability.