Decomposition of Compositional Effects: Explicating Covariance in Demography Analysis

Vladimir Canudas Romo, Pennsylvania State University
James W. Vaupel, Max Planck Institute for Demographic Research

Demographic dynamics can generally be decomposed into direct effects and compositional effects (due to changes in structures of the population). The magnitude of a compositional effect can often be measured by a rather obscure covariance term. We develop alternative formulas that permit decomposition of compositional effects into demographically meaningful components. We apply the new formulas to decompose change in life expectancy, crude birth and death rates, and the average age of a population.