Footnote:
In: The International Journal of Business and Finance Research, v. 8 (5) p. 95-114, 2014
Nach Informationen von SSRN wurde die ursprüngliche Fassung des Dokuments 2014 erstellt
Description:
This study predicts the medical expenditure of national health insurance by a Back-Propagation Neural Network (BPN). Monte Carlo Simulation and Multiple Regression Analysis are used to compare the results of the BPN. Empirical results show the performance indicator modeled on BPN is the best, followed by those modeled on Multiple Regression Analysis and Monte Carlo Simulation. In estimating the opportunity cost that will be lost when the forecasting model overestimates the expenditure, and the resource cost that will occur when the model underestimates the expenditure, the Monte Carlo Simulation and Multiple Regression Analysis are likely to be better forecasting methods. Finally, the three key factors affecting medical expenditure are the aging population index, the inflation rate and the number of insured population. This study makes a contribution extant literature by using a BPN to predict the medical expenditure performance indicator error rate. A BPN is better than other models in terms of TIC, but may not be the best forecasting method in a variety of cost conditions