Determinants Of Out-Of-Pocket Health Payments With Policy Effect In Zhejiang Province——Based On Big Data

Expensive to see a doctor has been one of the most prominent problems in health reform in our country. At present, the health investment of government is inadequate, the proportion of personal medical expenses is large, and personal medical burden is too heavy. Besides, the public hospital is currently lack of public welfare, excessively seeking economical efficiency, and feeding the hospital with medical charges, which result in the phenomenon that it is difficult and expensive to see a doctor intensified. Since the new medical reform in2009, with the large increasing of health insurance coverage in China, as well as drug price addition phasing out, the problem that medical treatment is expensive eased. However, compared with developed countries, the proportion of out-of-pocket health payments (OOP) is still too high in China, therefore, how to reduce the proportion of OOP becomes a hot topic in health reform.From research direction, the existing literature studied the impact of OOP from the perspective of economics, by economic theory-oriented, and the result is not uniquely determined. This paper will establish non-parametric path convergence model using the data of Zhejiang province from the perspective of big data, guided by statistical data driven, promoting the linear path identification method, and then embed the non-parametric methods of automatically eliminating variables, which could eliminate the defect of principal component method, remove relatively independent variables, and identify the determinants of OOP and policy effect. The empirical results show that:First, among the determinants of oop, the factors of linear mechanism include the first-order lag item of oop, the proportion of medical income in the hospital total revenue, the proportion of drug income, and the proportion of financial aid, while the factors of nonlinear mechanism include the number of hospital staff, actual using bed rate, return on assets, health care coverage, economic development level, as well as per capita disposable income. But average education attainment is not the determinant of oop.Second, the mechanism of the determinants has become more complicated, which means linear factors play a main role before2008and the nonlinear factors play a main role after2008. The first-order lag item of oop plays a key role in linear part. Return on assets plays a main role and per capita disposable income takes second place in nonlinear part before2008, while per capita disposable income plays a main role, health care coverage and economic development level take second place after2008.Third, compared with the benchmark model, the variables of linear mechanism have the scale effect of Crowding-in, and the variables of nonlinear mechanism have the negative allocation effect. The negative allocation effect of return on assets is the biggest before2008, followed by per capita disposable income, while the effect of per capita disposable income is the biggest after2008, followed by health care coverage and economic development level. It means that making consideration for return on assets, health care coverage, economic development level, as well as per capita disposable income, can effectively reduce the oop.Fourth, based on the expansion effect of a single macro factor, the effect of implementing the expansion policy of health care coverage and economic development level is best in the beginning of health care reform. In recent years, the moderate expansion policy of per capita disposable income might make the oop decline, which could alleviate the phenomenon that it is expensive to see a doctor to some extent. In addition, the continued expansion policy of all factors increasing by50%did not achieve the result of policy effect increased significantly. Hence we can think that the expansion effect of health care coverage, economic development level and per capita disposable income need to be moderate, too much of a good thing.Fifth, together with these three factors, the comprehensive policy effect is not the simple addition of the effect of three factors. It is thus clear that the relationship between these factors and oop can’t be linear, which also confirmed the superiority of the semi-parametric model adopted by this paper on the other hand.