Methodology Study On Prediction And Determinants Of Total Health Expenditures In China

Objectives:To find the main determinants of Total Health Expenditures in China using scientific and reasonable methods, and to discuss the long-term equilibrium and short-term fluctuation relationship between Total Health Expenditures and the determinants, and to forecast the trends of China's total health expenditure in future with three kinds of prediction methods, and as well as to assess synthetically the precision of those methods. At same time, to provide scientific evidence for health economic policy and the efficient use of resources, as well as the total health expenditure forecast study.Methods:Dynamic regression model was established between Total Health Expenditures and independent variables with the methods of co-integration approach, error correction model and granger causality test, the independent variables included demand factors (income, health expenditures, Population aging, urbanization) and supply factors (The number of physicians and hospital beds). Logarithmic transformation and difference were brought to achieve the homogeneity of variance and stationary of sequence. Trough auto-regression integrated moving average, ARIMA model, and auto-regression model and state space model to fitting the time series data of Total Health Expenditures in China from 1978 to 2005. three indicators such as Mean squared error and Mean absolute percentage error and Root mean squared error were introduced to assess synthetically the precision of this three prediction methods based the time series data of China from 2006 to 2008. After that, making prediction to the total health expenditures trends in 2009-2020 in advance. The dataset was developed with the Excel software. All analyses and predictions were performed using SAS version 9.13. All of the statistical tests were two sided, and the 'significance level' wasα=0.05.Results:①The results from regression model for determinants and trends of the total health expenditures showed that there were long-term equilibrium relationship between the total health expenditures and determinants such as income, health expenditures, population aging, urbanization, the number of physicians and hospital beds, and the short-term fluctuation relationship existed in the total health expenditures and income, and population aging, and the number of hospital beds. However, health expenditure, urbanization and the number of physicians have no effect on the total health expenditures in short-term. The findings showed that the long-term elasticity was 2.110, and the short term elasticity was 0.581. The granger causality test illustrated that the total health expenditures were causality factors of the changes in income, and the number of hospital beds. Meanwhile, the incomes and the population aging were causality factors of the changes in total health expenditures. The three kinds of predictive models demonstrated per capita health expenditure would reach 691.536 Yuan,720.130 Yuan, and 944.466 Yuan in the year 2020, respectively. The predictive value of the state space model was the highest. The second was auto-regression model.②The results of comparative Analysis of three predictive methods that the fitting accuracy of the state space model was the best. So was the auto-regression model. However, ARIMA model has the lower value in Mean absolute percentage error. On the other hand, the sate space model and the auto-regression model have an excellent prediction Accuracy. Auto-regression model had the lowest Mean squared error as well as the state space model had the lowest Mean absolute percentage error.Conclusion:①Income, health expenditures, population aging, urbanization, and the number of physicians and hospital beds were the determinants of the total health expenditures in long-term. And the impact of aging population has becoming dominant gradually. But, health expenditures, urbanization, and the number of physicians have no impact on the total health expenditures in short-term.②The total health expenditures were causality factors of the changes in income and the number of hospital beds. Meanwhile, the income and the population aging were causality factors of the changes in total health expenditures too.③Through adding the influencing factors in the study, and to eliminate the external shock, policy changes and other unpredictable factors, State space model can be applied to the time series prediction research of the total health expenditures. It will make the prediction of the total health expenditures has important practical significance.