Path Selection of China’s Medical Reform Based on Empirical Measurement of Asymmetric Information

Based on the micro individual data from China Health and Nutrition Survey (CHNS) database and the development of China’s medical and health system reform since the reform and opening-up in 1978, this paper applies a two-tier stochastic frontier model to measure the degree of asymmetric information between physicians and patients in China’s medical service market and estimate the impact of bargaining on the final price of health care both holistically and annually. The empirical results show that: 1) The information mastered by physicians and patients have significant but opposing effects on the final medical service price. Physicians hold more information relative to patients and thus have stronger bargaining power. 2) Almost all patients will be forced to accept a medical price higher than the benchmark price by 16.52% on average by receiving excessive and expensive medical services, despite the differences in bargaining power between physicians and patients are highly heterogeneous. 3) The series of medical reforms in China had different impacts on medical care pricing efficiency. Especially, the early medical reforms could impose a positive impact on the efficiency of the medical market in the initial stage. I further propose that the path choice of China’s new round of medical reform should be as follows: It is necessary to insist on government-led efforts to protect the public welfare of medical services, and to introduce market mechanisms to mobilize social resources and improve the efficiency of the medical market. In addition, I point out that the establishment and improvement of the general practitioner system may reduce the degree of information asymmetry between physicians and patients, thus alleviating China's deteriorating physician-patient relationship.


Introduction and Literature Review
In recent years, the contradictions in the field of medical and health care in China have been aggravating day by day [1], which not only affect national health but also bring about a series of social problems such as poverty and tension between physicians and patients. Two kinds of medical reform ideas of "government-led" and "market competition" have been causing widespread controversy in the civil and academic world. The government-led reform idea is that the government mainly organizes most of medical and health institutions and provides low-cost medical and health services to all the people [2] [3]. The idea of market competition reform is that market competition can play an important role in improving the efficiency of medical and healthcare markets and reducing medical expenditures [4] [5]. Later, some scholars also believed that the roles of the government and the market should both be played in medical reform [6] [7]. Previous studies mainly carried out normative analysis from the theoretical level. However, due to the lack of relevant empirical evidence, it is very difficult to make a convincing conclusion as to which kind of medical reform idea will produce the expected effect.
Since the reform and opening-up in 1978, the Chinese government has implemented a series of medical reform measures based on the above two different medical reform ideas. It is particularly urgent and has practical significance for the provision of China's new round of medical reform to test whether these previous reform measures have really improved the efficiency of China's medical market and then to clarify the responsibility of the government and the market in the entire medical reform. Yet, there are many factors that determine the efficiency of the medical market. Therefore, it is difficult to estimate the efficiency of the Chinese medical market directly from the overall perspective. Among the many factors that affect patients' demand for medical services, medical expenses are the most concerned [8] [9] [10] [11] [12]. Patients will also refer to medical expenses when purchasing medical services. Bernard Gauthier and Waly Wane (2011) found that low-income patients chose to bypass the high-quality medical institutions they could not afford, while high-income patients preferred institutions that offered better quality medical services. And the significant difference in the choice of medical services for patients exists in both urban and rural areas [13].
In the highly competitive Chinese medical service market [14], the actual total cost of medical services reached varies greatly [15] [16]. Even if the medical services provided are fairly homogeneous and standard, the same physician may charge different patients different fees, and different physicians may charge different fees for the same patient. The price dispersion cannot be explained en-  [20]. Previous studies have also shown that the medical services market has a very significant asymmetry of information [21] [ 22]. The professionalism of medical services makes physicians have the advantage of information asymmetry, which gives them the incentive to induce demand in their own interests (e.g., overprescribing, prescribing expensive drugs or conducting expensive tests, etc.) [23]. Feldstein's (1970) study shows that the asymmetric information advantage can induce moral hazard of healthcare providers-demand inducement, which increases the patient's healthcare expenditures [24]. Rice (1983) found the amount of medical services would increase after the decline in physician service fees [25]. Yip (1998) concluded that based on demand inducement hypothesis, when a physician's job is affected by the income effect, this physician will make up for the decrease in income by increasing the volume of services [26]. Hengpeng Zhu (2007) pointed out that serious asymmetric information in the medical service market, coupled with the physician-patient inconsistent and even conflicting economic interests provide the possibility and space for physicians to implement moral hazard behaviors: physicians will pursue their own economic interests by using information superiority to induce patients to consume too much medical services and medicines regardless of the interests of patients [27]. And incentives that link physicians' salaries to their income from treatment will allow physicians to take full advantage of asymmetric information to induce more medical needs and then increase health expenditures for patients or the health care system [28]. Although the price of medical services in China is not formed spontaneously through the regulation of market supply and demand, but determined by the relevant government departments, and the price of medicines is also limited by the guidance of the government [29] [30], in the environment of asymmetric information, however, the decision-making power of medical services and the control of total medical expenses are still in the hands of physicians [11], and physicians have absolute discretion over the consumption of medical services. Therefore, the information advantage allows physicians to have bargaining power over the patient's final medical expenditures.
Under China's current medical service system, it is reasonable to assume that patients also have some bargaining power. Because, on the one hand, people who need health care services can have different options. Mommaerts (2018) examined the degree of substitution between nursing home residency and coresidence with adult children [31]. And patients can choose other medical services instead of medical treatment, for example, patients can choose to self-medication, or go directly to the pharmacy to buy medicine. Although such alternative therapies currently occur in common diseases such as colds, the scope of alternative therapies will grow with the development of medical technology [32]. On the other hand, the establishment of the medical information disclosure system [33],  [37].
In the United States, health insurance is primarily provided by commercial insurance, so it is uncommon in health care markets for patients to directly negotiate with providers for health care services rendered, particularly if patients have access to health insurance. Insurers typically bargain on behalf of their enrollees. Unlike the United States, China's medical insurance is mainly provided by the government. At present, China's medical insurance can be divided into social medical insurance projects which are compulsory implemented by the government and commercial health insurance projects which are voluntarily purchased by consumers. Insufficient coverage, inadequate protection, incompetent cost control, inefficient operation and unfair security system are the five major problems facing China's medical insurance [38]. Due to the lack of risk-sharing mechanism, and the current medical insurance payment method is mainly the after-the-fact payment paid by the third-party, a large amount of medical expenses also requires personal direct payment. In addition, China's current medical insurance institutions and medical institutions are separated from each other [39], and there is no negotiation mechanism with medical institutions. Therefore, the trend of rising medical service costs has not been effectively suppressed [40] [41]. From 1978 to 2005, personal cash expenditure on health increased from 20.43% to 52.2% in total health expenditure [38]. Because of the lack of fairness in self-funded medical care and the weak anti-risk ability, the economic ability of patients directly determines the timeliness and effectiveness of the medical services they receive, so the inequality of people's income is transformed into the inequality of access to medical services.
In summary, the final expenditures of medical services for patients to go to a hospital are determined by the bargaining process between the physician and the patient. The final medical expenditure reflects the bargaining power of both physicians and patients. And the bargaining power reflects the degree of information asymmetry between the two sides. Therefore, from the perspective of the final medical expenditures paid by patients, I measure the surplus extracted by the physician and the patient in the process of determining the final medical expenses, thus verifying the effectiveness of China's medical reform. Obviously, if the reform of the medical service market really works, we should observe that over time, the surplus acquired by physicians should gradually decrease, and the surplus of patients should gradually increase. results show that the information held by both physicians and patients has an important impact on the final price of medical services, and physicians have stronger bargaining power than patients. Therefore, on average, the price of medical services reached is 26.61% higher than the fair benchmark price. In addition, the net surplus between physicians and patients hardly changed with time. Specifically, the price of medical services for each year is roughly higher than the fair benchmark price of about 26% from 1989 to 2006. Therefore, they concluded that since the reform and opening-up, China's medical service system reform has not effectively solved the problem of "expensive medical bills and limited access to quality medical services" [18] [19]. I think the conclusions of these two articles are inconsistent with the reality of China's health care reform.
Thus, I use the same methodology and similar data from the same database to conduct an overall and annual empirical measure of the degree of asymmetric information between physicians and patients in China's medical service market and the impact of bargaining on the final price of health care.
The contribution of this paper can be summarized into the following four aspects: 1) I estimate the bargaining power of both physicians and patients during the pricing process of medical services both holistically and annually based on China's micro data using the two-tier stochastic frontier approach, and I find that the information mastered by physicians and patients have significant but opposing effects on the final medical service price, and physicians hold more information relative to patients and thus have stronger bargaining power. The remainder structure of the paper is organized as follows: Section 2 briefly reviews the development of China's medical and health system reform. Section 3 describes the intuition and structure of the two-tier stochastic frontier estimation procedure. Section 4 introduces the sources of data and the selection of variables, and Section 5 presents the empirical results. Section 6 provides in-depth discussions on the path selection of China's new round of medical reform, while my conclusions are included in Section 7.
dominated the investment and has constructed the overall medical insurance system from urban to rural areas. In general, China's health care decision-making at this stage has achieved fairness and a significant improvement in people's health. Life expectancy per capita rose from a very low level of 35 years before 1949 to 68 years in 1978 [42]. However, as consumers in the public health care system barely bear the medical expenses, the incentive mechanism encourages the demand side to make excessive use of medical resources, and thus the medical expenses are expanding rapidly. Since the establishment of the public health care system in 1952, efforts to remedy system deficiencies have not been interrupted. 2) Market-oriented reform stage , at which the medical reform policy tended to encourage self-financing medical institutions to replace fiscal expenditure in the health sector, due to the large-scale fiscal pressure caused by the previous public health care and labor insurance system. Otherwise, insufficient medical investment before and after 1979 could hamper the operation of medical system. At this stage, the quantity and quality of medical and health services have been greatly improved, and to some extent, the problems of inefficient hospitals and insufficient supply of medical services have been solved.
Meanwhile, the burden of government medical expenditure has also been signif-

A Brief Review of the Two-Tier Stochastic Frontier Model
The two-tiered stochastic frontier model, which is called in the settings where it is believed that asymmetric information on both sides of the market leads to surpluses which buyers and sellers can both extract, has been successful across a range of applications. Some examples of such technique that can be applied are the investment efficiency of Chinese listed companies [43], the stock market [44] [45], the real estate market [46], the technology market [47], and the timber auction market [48] and so on.
This study closely follows Kumbhakar and Parmeter (2009) in using a two-tier stochastic frontier model to measure market efficiency [49]. However, Kumbhakar and Parmeter (2009) use the model to measure matching efficiency in a typical labor market, in which the decision on the wage level is the result of negotiations between employees and employers, and ultimately the equilibrium wage X. X. Yang Health level depends on the bargaining power (mastered information advantage) of employees and employers. The analysis in the introduction of this paper shows that the medical service expenditure that the patient finally pays to the hospital for medical treatment is determined by the bargaining between the physician and the patient. The ultimate equilibrium price (the actual total expenditure of medical services) reflects the bargaining power of both parties (patients and physicians).
And the bargaining power reflects the degree of information asymmetry between the two parties. Therefore, the bargaining mechanism between employers and employees (as in Kumbhakar and Parameter, 2009) is transferable to that between physicians and patients, and thus this particular model is suitable for the medical services market. Scholars who also use the two-tier stochastic frontier technique to study the impact of incomplete information in the medical services market include: Gaynor & Polachek (1994) [50]; Chawla (2002) [20].
Assume that in a typical medical service market, there are numerous providers and demanders of medical services, physicians and patients have certain information, and the final price P of medical services can be determined by the bargaining between physicians and patients. Then the final price of medical services P can be expressed as: where P represents the physician's reservation price, P represents the patient's maximum price offer ( P P In order to capture not only the impact of physician's bargaining, but also the impact of patient's bargaining, I first denote the benchmark price, which means the expected "fair" medical service price given a vector of characteristics x, by , where θ is the unknown but actual "fair" medical service price, and the expected surplus that patients extract in the process of reaching a medical service price, ( ) represents the expected surplus extracted by physicians.
And which side can "grab" more surplus will depend on the degree of information they possess, and their bargaining power based on it [53]. Accordingly, Equation (1) can be rewritten as: Equation (2) indicates that the final price of health care, in fact, consists of three components: the "fair" price of medical services ( ) conditional on a vector of individual characteristics x, which I will hereinafter refer to as the From the practical point of view, the net surplus, , where , 1, , .
vector of covariates that represent observable characteristics that may impact medical price, β is the corresponding parameter vector) and let i ξ denote the composite error term enclosing At the same time, I assume that i u , i w and i v are independent of each other and are both independent of individual features i x . These distribution assumptions are necessary for empirical analysis, considering that the two-tier frontier model is nonparametrically underidentified. Of course, i w and i u can also be assumed to follow other types of one-sided distribution, such as the half-normal distribution, log-normal distribution, and gamma distribution, etc. Yet, since Kumbhakar and Lovell (2000) showed that using different distributional assumptions would not have a substantial impact on the estimation results [54], I therefore assume exponential distribution due to its simplicity. Based on the above assumptions, the probability density function of i ξ can be derived as follows (for details, see Kumbhakar and Parmeter, 2009 [49]): where ( ) Φ ⋅ and ( ) φ ⋅ are the cumulative distribution function and the probability density function of standard normal distribution respectively, and Furthermore, for a sample containing n observations, the log-likelihood func- By maximizing the log-likelihood function, we can obtain the maximum likelihood estimates of all parameters. It should be noted that since w σ and u σ appear separately in the likelihood function (i.e., w σ only appears in i b and i h , and u σ only appears in i a and i c ), identification of these two values can be achieved. Thus, another reason for assuming that surpluses extracted by physicians and patients obey exponential distribution is that likelihood functions can be expressed in a closed form, and the identification of variance parameters is obvious.
In this paper, I focus on obtaining the estimates of surpluses extracted by the physician and the patient.
and ( ) which indicate the percentage deviation of the medical price from the optimal boundary (benchmark price) resulting from the physician-patient bargaining power, when u and w are small, since the dependent variable in this paper is in logarithmic form. In order to obtain the exact percentage deviation between the actual medical price and the benchmark price, Equations (8) and (9) can be interpreted as follows: which describes the net effect of asymmetric information on medical price.

Data Sources
The data used in this paper is from the China Health and Nutrition Survey years (1991,1993,1997,2000,2004,2006). Because this survey data comes from a large number of questionnaires, its sample size is limited compared to the data of listed companies and macroscale statistics, and it is quite noisy. In order not to interfere too much with the data, I carefully screened the data before using it.

The Selection of Variables
Since patients may receive very different levels of care, the variation in medical prices could be mostly due to the services patients received, rather than the bargaining power of providers relative to patients. Therefore, ideally, this study would require price data on a specific type of care (e.g., heart bypass surgery).
However, there is no relevant data available in the questionnaire. Thus, in the measurement of the price of medical services (the dependent variable in this paper), I select the total expenditure of the medical service reported by the patient (corresponding to the questionnaire question M39 "money spent on illness or injury"), which includes the medical expenses and the drug fee, instead of the unit price of a particular type of medical service. Because patients can indirectly "select" medical expenses by choosing to go to different hospitals, and "select" X. X. Yang Health drug prices by deciding whether to buy drugs in the hospital (or to go to a nearby pharmacy), both options make patients have certain bargaining power.
Therefore, although the price data is limited by the setting of the questionnaire itself and cannot be subdivided, this indicator can largely measure the price of the patient's consumption of medical services. From the descriptive statistics in Table 2, the distribution of medical expenditures is very discrete, with a maximum of 9000 and a minimum of 1.01. The standard deviation is also very large.
So, in the subsequent empirical analysis, I performed a natural logarithmic transformation of this variable to minimize the impact of potential outliers on the regression results.
To measure the "benchmark price" ( )   X. X. Yang

The Empirical Results
In this section, instead of determining the marginal effect of covariates, based on the model specification, I conduct variance decomposition analysis of the two-tier stochastic frontier model to not only explore the impact of bargaining power over medical service expenditures but also obtain the holistic and annual estimates of extracted surplus due to different bargaining power between physicians and patients.

Model Specification
Since we can only observe the medical expenses paid by patients, the theoretical fair price ( )  Table 3 presents the results of parameter estimations based on the two-tier stochastic frontier model.
In Table 3, Model (1) uses the OLS estimation method, and Models (2) - (7) use the MLE method based on the two-tier stochastic frontier technique. Model The largest log likelihood value and the LR test value of Model (7) indicates that Model (7) is the best specification, so the subsequent analysis in this paper is mainly based on the variables and measurement results under Model (7). Table 4 reports the results of bargaining power effect analysis. I find that bargaining power has a considerable impact on the formation of medical service prices. Specifically, the stronger bargaining power of physicians compared with patients leads to a positive net effect of negotiations over the price of medical services, which can be found from the estimate of ( )

Variance Decomposition Analysis: The Effect of Bargaining Power on Medical Service Prices
Therefore, in general, bargaining will result in a higher medical price relative to the benchmark price. In addition, the unexplained variation in log price ( ) Note: t statistics are reported in parentheses. *P < 0.1, **P < 0.05, ***P < 0.01. 20.53%), the formation of medical service prices depends more on physicians.
To analyze the surplus extracted by physicians and patients and the net surplus in the bargaining process, I further estimate the unilateral effect of both physicians and patients.  corresponding estimated equations are (10) and (11), which mean the percentage change of the surplus extracted by physicians and patients in the bargaining process relative to the benchmark log price estimated from l n i P x β ′ =

Unilateral Effect Estimation: Surplus Extracted by
. Table 5 shows that on average, surplus extracted by physicians can increase the medical price by 51.28%, while surplus extracted by patients can only decrease medical price by 34.76%. These two opposite effects led to an increase in medical prices by 16.52% relative to benchmark price of medical services. In other words, due to the existence of information asymmetry and the difference in bargaining power between physicians and patients, patients need to spend $116.52 for the $100 medical service in a fair market.
The last three columns in Table 5 (Q1-Q3) show that differences in bargaining power between physicians and patients are highly heterogeneous, and almost all patients are at a disadvantage in the bargaining process. More specifically, the first quartile value of net surplus is 8.96%, which means medical prices are only 8.96% above the benchmark price for 25% of the sample. The top 25% of the surplus extractions are more than 34%, which increases about 18% relative to the average price.

The Annual Results of the Sample Estimates
Since the reform and opening-up in 1978, along with the promotion of economic system reform in various fields, China's medical service system reform has also developed in depth. To explore whether the degree of information asymmetry between physicians and patients will decrease with the progress of medical reform, I analyze the impact of bargaining on medical prices across different years. The results are reported in Table 6. show that the net surplus between physicians and patients barely changes over time [18] [19], Table 6 indicates that the net surplus changes corresponding to each reform. Especially, early waves of reforms rendered significantly high efficiency compared to later waves. I think the main reason for the difference in empirical results is that although the survey data we use involves different years, the data is not panel data, but mixed cross-section data. Namely, patients in each X. X. Yang However, I estimate separately in each year. Of course, to make the number of observations larger, we can also use the rolling estimation method after mixing two or three years of data, to better evaluate the degree of asymmetry information between physicians and patients. Table 6 suggests that in 1991 patients' bargaining power was stronger than physicians', thus, patients can lower their medical expenditures by 6.54% relative to benchmark prices. This time corresponds to the second stage of China's medical reform (1985)(1986)(1987)(1988)(1989)(1990)(1991): the official launch of a comprehensive medical reform.
With the gradual reduction of government direct investment, the marketization of medical institutions has further strengthened the supply of medical and health services, thereby making full use of the "vote-by-foot" effects of patients and enhancing their bargaining power.
However, with the continuous progress of the third stage (1992-1999): marketization of medical institutions, government finances have been gradually withdrawn from the supply side, and the contract responsibility system has been introduced to hospitals. As a result, hospitals have become self-financing agencies.
Nevertheless, even though the market-oriented profit-seeking behavior of medical institutions has greatly remedied the problem of insufficient income of medical and health institutions, it has also led to an unreasonable increase in medical prices coupled with the poor attitude of some medical staff. Therefore, to a cer-Health tain extent, the public welfare of medical institutions and the interests of the people have been harmed. Corresponding to Table 6, physicians always have stronger bargaining power than patients in 1993 and 1997, and the difference in bargaining power was increasing from1993 to 1997.
In 2000, the auction of a total of more than 100 public hospitals achieved the exit of state-owned assets and the injection of private capital. Yet, it can be seen from Table 6  To wrap up, since the reform and opening-up, the basic experience of China's medical and health system reform shows the government-run public medical service is not feasible due to its high cost. From another perspective, the competition mechanism can be introduced to the medical service market by opening medical markets, encouraging the development of private medical service institutions and reforming hospital property rights. And, to a certain extent, the introduction of competition mechanism can ease the pressure of government fiscal expenditure and promote the quantity and quality of medical and health services greatly. Moreover, the above empirical analysis shows that in the early stage, the market-oriented reform of medical institutions can indeed reduce the degree of information asymmetry between physicians and patients and has a positive impact on the efficiency of the medical market. However, under the dual pressures of income-generating assessment and money temptation, profit-seeking is a systematic behavior of the medical system. Public hospitals will gradually act like profit-oriented enterprises. Namely, they will take advantage of their asymmetrical information in medical field relative to patients, resulting in rapid and unreasonable growth in patients' medical expenditures.
X. X. Yang

The Respective Roles of the Government and the Market in Medical Reform
From the characteristics of medical service itself and the experience of medical reform in China since the reform and opening-up in 1978, I believe that the direction of China's new round of medical reform not only should insist on government-led efforts, but also should introduce market mechanisms. The scope of China's health reform can be divided into four segments: public health, medical services, pharmaceutical production and circulation, and medical insurance. In different fields, the respective roles of the government and the market should be different. In fact, the duty of the government and the market in the medical reform has been brilliantly exposed in Adam Smith's "The Wealth of Nations"-"The third and last duty of the sovereign or commonwealth is that of erecting and maintaining those public institutions and those public works, which, though they may be in the highest degree advantageous to a great society, are, however, of such a nature that the profit could never repay the expense to any individual or small number of individuals, and which it therefore cannot be expected that any individual or small number of individuals should erect or maintain. The performance of this duty requires, too, very different degrees of expense in the different periods of society" [57]. Consequently, from the viewpoint of modern market economy, the government should focus on weak parts of medical reform and let market itself regulate strong parts. Only in this way, can respective comparative advantages of social forces and governments be fully utilized. Therefore, I suggest that the responsibilities of the government and the market in the medical reform should be divided as follows (Table 7).
Specifically, in the field of public health services (including establishing health records, women's health care, elderly health care, vaccination, and health education, etc.), the government should shoulder its primary responsibility and shift the emphasis from treatment to prevention. In the field of medical services, a mixed system should be adopted. The government is mainly responsible for financing the establishment of non-profit medical institutions, and let the market regulate those for-profit ones. In the field of drug production and distribution, the circulation and pricing of medicines should mainly be decided by the market, while the government mainly supervises the quality standards of medicines.
Medical insurance should be divided into different levels, then respectively shared by the government, market and individuals.

Establish and Improve the General Practitioner System to Reduce the Asymmetry of Physician-Patient Information
Given the important role of information asymmetry in the pricing process of medical services, and the fact that physicians have more information and more bargaining power than patients, there is a growing literature that studies how changing provider incentives can help overcome information asymmetry and reduce provider incentives to induce demand. At present, the common payment mechanism can be divided into retrospective payment system and prospective payment system. The retrospective payment system pays after the medical service is completed, and the "fee for service" (FFS, also known as "pay-for-performance") is the most commonly used one. Yet, due to the large amount of verification work required, it is inevitable that the quantity and quality of verification will be reduced, and ultimately this payment mechanism cannot effectively restrain excessive medical treatment. The prospective payment system, as the name suggests, pays before the medical service occurs, and the hospital "pays for overspending and obtains the surplus" [58]. Prospective payment methods include per capita payment (or capitation), global budget, per diem payment, case-based payment, and diagnosis-related groups (DRGs) et al.
Although prospective payment system is a more cost-constrained method than retrospective payment system, driven by cost savings, prospective payment may encourage providers to lower the quality of medical services or reduce the necessary treatments, resulting in under-service to obtain more surplus. Therefore, in reality, the existence of information asymmetry and the inherent needs of medical service providers to maximize their utility, cut off the possible existence of ideal payment method (which can provide appropriate incentives for medical service providers, and allow them to consciously carry out cost control and quality assurance, and avoid the possibility of overdose or shortage of medical services [59]) [60]. Consequently, I believe that the establishment of a general practitioner system (which not only undertakes community primary health care services, but also differentiates the specialist medical and general medical services markets by determining whether patients need referral treatments) can increase the amount of information of patients, thus effectively improve the bargaining power of patients and efficiently lower the degree of information asymmetry between physicians and patients due to the following two reasons. First, general practitioners are the gatekeepers of the health of residents. As they can provide a wide range of services including family nutrition guidance, disease prevention, disease treatment, child and elderly health care, and medical consultation and so on, general practitioners can greatly enhance the public's ability to prevent diseases and shift the focus from new treatments to disease prevention, thereby reducing the incidence of major diseases. In addition, since general practitioners get closer to patients, understand patients and are familiar with the medical characteristics and medical environment of various hospitals, the degree of information asymmetry between physicians and patients can be reduced more thoroughly.
Second, general practitioners are also the gatekeepers of the health care system. General practitioners provide residents with the first medical diagnosis, and then decide whether they need to be referred to higher-level hospitals according to the severity of their illness. And patients who have passed their serious phase in a referral hospital will be referred to the community hospital for further treatment or rehabilitation. Through primary diagnosis in communities and hierarchical referral system, the general practitioner system can improve the efficiency of physicians' diagnosis and treatment. Additionally, by extending the medical resources to the grassroots level, the general practitioner system can also change the present situation that medical resources are currently too concentrated in large cities and major hospitals and then optimize the allocation of medical and health resources to help solve the problem of expensive medical bills and limited access to quality medical services. In 2011, the State Council of China decided to establish a general practitioner system and promulgated "The Guiding Opinions on Establishing a General Practitioner System", which not only has made a prospective and directive design for the training, practicing and incentive systems of general practitioners, but also is a milestone in the development of General Practice in China. However, the general practitioner system in China started late and was not yet mature. Therefore, with reference to the experiences of foreign general practitioner systems, China should seize the opportunity of the new round of medical reform and accelerate the improvement of the system of general practitioners.

Conclusion
In this paper, I use a two-tier stochastic frontier model to analyze the impact of asymmetric information between physicians and patients on the final medical service expenditures both holistically and annually, based on the China Health and Nutrition Survey (CHNS) data and the development of China's medical and health system reform since 1978. The empirical results show that: 1) Despite the strong heterogeneity in the bargaining power between physicians and patients, on average, the net effect leads to an increase in actual medical service expenses by 16.52% relative to benchmark prices. 2) The series of medical reforms in China had different impacts on medical care pricing efficiency. Especially, the early wave of reforms could have a positive impact on the efficiency of the medical market at an early stage. The above empirical analysis indicates that relying solely on government-led or market-oriented medical system reform cannot improve the efficiency of China's medical market in the long run. Therefore, China's medical reform must combine government-led efforts and market economy, so that not only the public welfare of medical services can be guaran-X. X. Yang teed, but medical efficiency can be improved as well. According to the above empirical analysis, this paper further provides recommendations on the division of responsibilities between the government and the market in the new round of China's medical reform and how to reduce the information asymmetry between physicians and patients.

Conflicts of Interest
The author declares no conflicts of interest regarding the publication of this paper.