On mortality modelling using a density-hazard distribution

Abstract This paper analyzes a new parametric distribution from a family of density-hazard distribution for mortality modelling. The performance of the distribution is assessed using real United States mortality data and compared to an existing study demonstrating a three-parameter logistic distribution employing similar data. Parameter estimation is performed using the nonlinear least squares method. The result shows that the density-hazard distribution consistently gives a better fit for the United States male, female and total population data than the three-parameter logistic distribution for a 60-year period of study with the conclusion derived from the mean square error. In addition, the estimated life expectancy with an underlying density-hazard distribution gives proximity to its empirical counterparts for almost all age groups.