Mortality Co-Movement at the National Level: A Quasi-Social Network Analysis
Andrew Noymer, University of California, Irvine
Tanya Jukkala, Södertörn University
Christopher S. Marcum, University of California, Irvine
We combine two types of analysis for life expectancy time series. We use Goodman-Grunfeld analysis to determine co-movements. Goodman-Grunfeld is a nonparametric test of co-movement between two time series; it tests whether co-movements are greater than would be expected, accounting for the trends of each time series, and correcting for serial correlation. We performed an all-pairwise-comparisons on the Human Mortality Database (N countries=40). This gives a matrix of countries whose life expectancies are expanding in lock-step, statistically speaking. We then perform clique analysis, a social network technique. We find that many groups of countries whose life expectancy is moving together. The largest is seven: Belgium, Denmark, Finland, Italy, the Netherlands, England, Wales, and Scotland; there are 8 cliques of size 6, 13 of size 5, and 14 of size 4. We conclude that, historically, mortality decline has been in remarkable synchrony worldwide, pointing to shared fortunes of environmental factors.
Presented in Session 126: Historical Mortality Patterns