Projection of Total Fertility Rate (TFR) In the United States of America Using a Machine Learning Approach

The United states is among the first world countries which are currently reporting fertility rates below replacement level and this has serious negative consequences in the near future with regards to the availability of youthful labor force, increase in health expenditure and social security costs. In this study, the ANN approach was applied to analyze TFR in the USA. The employed annual data covers the period 1960-2018 and the out-of-sample period ranges over the period 2019-2030. The residuals and forecast evaluation criteria (Error, MSE and MAE) of the applied model indicate that the model is stable in forecasting TFR in the USA. The results of the study indicate that annual total fertility rates in the USA are likely to be around 1.9 births per woman throughout the out-of-sample period. Therefore, the US government is encouraged to promote child bearing especially among women in the 18-30 years age group by providing pro-fertility incentives and reducing the cost of raising children.