A Bayesian Age-Period-Cohort approach for modeling fertility in Puerto Rico

Puerto Rico has one of the lowest total fertility rates (TFR) in the world. Combined with a negative net migration and a high proportion of older adults, its unique situation motivates the need for further demographic analysis. Determining whether low fertility rates are mostly due to period or cohort effects is crucial for developing effective public policies that adapt to changes in fertility and population structures. The main objective of this work is to develop an Age-Period-Cohort model, in order to describe fertility data in Puerto Rico, from 1948-1952 and determine the contribution of period and cohort effects to fertility decline. The APC model was developed following a Bayesian framework, with a Poisson likelihood, RW(2) autorregressive priors for the APC parameters, and Scaled Beta2 priors for the precision parameters. Both frequentist and Bayesian methodologies attribute more importance to cohort effects when explaining fertility changes in Puerto Rico. Birth cohorts born in 1963-1967 onward have notably low fertility rates. There is no evidence of postponement of births in Puerto Rico, contrary to other countries with lowest-low fertility. Birth cohorts born in 1963-1967 onward have notably low fertility rates. This is the first application of APC analysis to fertility data in Puerto Rico, which describes fertility changes in a unique scenario in terms of demographic indicators, and the first APC analysis that shows the predominance of cohort effects when explaining fertility.