Title
Using Artificial Neural Network Techniques to Improve the Description and Prediction of Household Financial Ratios
Document Type
Article
Publication Date
1-2020
Abstract
The purpose of the study described in this paper was to shed light on the need for alternative methods to improve descriptions and predictions of household financial ratios. Using data from the 2013, 2015, and 2017 Panel Study of Income Dynamics (PSID), this study examined the descriptive and predictive power of an Artificial Neural Network (ANN) model and an Ordinary Least Squares (OLS) model when evaluating household savings-to-income ratios and debt-to-asset ratios cross-sectionally and across time. Results suggest that ANN models provide a better overall model fit when describing and forecasting financial ratios. Findings confirm that machine learning procedures can provide a robust, efficient, and effective analytic method when an educator, researcher, financial service professional, lender, or policy maker needs to describe and/or predict a household’s future financial situation. Suggestions for the implementation of ANN modeling procedures by household finance researchers, practitioners, and policy makers are provided.
DOI of Published Version
10.1016/j.jbef.2020.100273
Recommended Citation
Heo, Wookjae; Lee, Jin; Park, Narang; and Grable, John E., "Using Artificial Neural Network Techniques to Improve the Description and Prediction of Household Financial Ratios" (2020). Consumer Sciences Faculty Publications. 29.
https://openprairie.sdstate.edu/consumer-sci_pubs/29