A Comparative Study of Parametric and Nonparametric Estimators of Old-Age Mortality in Sweden

A recent study of Swedish old-age mortality used a modified GompertzMakeham model with a linear hazard for the force of mortality. We propose an alternative model using smooth two-dimensional kernel hazard estimators and introduce a new estimator based on the multiplicative bias correction for the multivariate marker dependent hazard. The multiplicative bias correction appears to have great potential for estimating mortality rates at the highest ages. We also observe that mortality continues to increase at an exponential rate even in old-age