ABSTRACT
This paper aims at identifying convergence clubs in suicide rates for a sample of 20 of FSU countries during the 1994–2016 period. A novel machine learning approach revels that while the null hypothesis of overall convergence has been rejected at the 5% significance level, there are four convergence clubs in the region. Within these clubs, different countries are on different transition paths and speeds of convergence, suggesting different policy interventions.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 Undetermined deaths are excluded from the analysis, and suicide rates are based on a standard population since demographic structures vary over time across countries. Also, suicide rates were missing for roughly 11%, and we imputed them using Amelia II in R (Honaker, King, and Blackwell Citation2011). Results available upon request.