Does privatization decrease the structural efficiency in the Chinese hospital sector?

Abstract

This paper examines the resource mismatch in China’s healthcare sector from 2010 to 2019. We use structural efficiency to measure the resource mismatch in the healthcare industry (0.2–0.7). We find that an increase in the proportion of private hospitals decreases the healthcare resource mismatch across provincial regions. This ameliorative effect is evident in non-coastal as well as central provinces. In addition, we also find that an increase in the proportion of private hospitals leads to a greater supply of healthcare resources and the number of healthcare services, which in turn better meets the healthcare needs of local residents and decreases the healthcare resource mismatch between regions in each province.

Introduction

With the rapid growth of the world’s population, the mobility of citizens from different regions deepens the mismatch of healthcare resources between regions [44, 71], such as the mismatch of healthcare resources between regions in the United States, Denmark, and China [25, 52, 69, 72]. At the same time, the mismatch between hospital inputs and outputs [1, 3, 7] and the imbalance between supply and demand in hospitals [65] further deepen the mismatch of hospital resources between regions of the world, which negatively affects hospital efficiency [65]. Therefore, there is an urgent need for the Chinese government to reduce the mismatch of healthcare resources between regions. Several studies have found that privatisation helps reduce resource mismatches in the banking, manufacturing and real estate sectors and improves sectoral efficiency [17, 36]. Therefore, it is necessary to explore whether privatisation in the hospital sector can reduce hospital resource mismatch between regions.

Medical inequality has always existed in China and has gradually evolved into a more serious problem of hospital resource misallocation among regions. As of 2022, the eastern provinces have a higher percentage of tertiary hospitals, and these provinces have 13,000 hospitals, 2.2 million doctors, which is more than the central and western provinces. However, the number of health technicians per 1,000 people, hospital bed use, and hospital beds per capita are lower in the eastern provinces than in the central and western provinces. Relevant literature has found that market-based procedures can reduce the mismatch of inter-regional medical resources [50, 66, 71], reduce the problem of long-distance access to medical care, and lower the cost of medical care for residents. Therefore, this paper explores the path of marketization to reduce the mismatch of inter-regional medical resources.

China’s market-oriented reform of the healthcare sector can be broadly divided into four stages: firstly, 1985–1992, the initial stage of healthcare sector reform, when the government began to marketise the healthcare sector. Next was 1992–2000, when the market-oriented reform of the healthcare sector emerged. in 1998, the State Council issued the Decision on Establishing a Basic Medical Insurance System for Urban Workers, marking the official shift of China’s healthcare system reform to a market-oriented model; and then 2000–2005, when healthcare organisations were divided into two major parts, for-profit and non-profit, with non-profit public hospitals In 2005, the government strongly encouraged social capital to enter the healthcare sector, which led to a steady growth in the number of private hospitals and a clarification of the property rights of healthcare institutions. In the final phase, from 2006 to the present, China launched a new healthcare reform aimed at resolving the “difficult and expensive access to healthcare” problem that had been widely reported by the public, and in April 2012, the Ministry of Health (MOH) allowed social capital to independently apply for the establishment of either for-profit or not-for-profit healthcare organisations. Subsequently, On 11 June 2015, the General Office of the State Council issued a number of policies and measures to promote the accelerated development of social medical services, which further eased the threshold of entry for private institutions. In May 2017, the Opinions on Supporting Social Forces to Provide Multi-level and Diversified Medical Services” further opened up the medical service market to social capital, lowered the access threshold, and stimulated the enthusiasm of social capital to organise medical institutions. In 2019, the “Opinions on Promoting the Continuously Healthy and Standardised Development of Socially-run Medical Care” was issued, which strictly controlled the number and scale of public hospitals, leaving enough space for the development of socially-run medical care. At present, China’s private medical and healthcare institutions have great potential for development, and their influence is gradually increasing.

In existing studies, there are various means of marketisation in the healthcare sector, mainly through the establishment of private hospitals and the introduction of social capital shareholding in public healthcare institutions. Brekke et al. [10] showed that competition between public and private hospitals reduces the mortality rate from diseases [8] and improves the quality of healthcare services [48]. Camilleri and Callaghan’s [11] study found that private hospitals tended to provide better quality services compared to public hospitals [18, 37], thereby attracting more patients, a result that had a significant impact on both public and private hospitals [61]. White and Collyer [68] conducted a large-scale study of hospitals in Australia, involving changes such as privatisation, deregulation and the entry of foreign capital into the hospital sector, and found that private hospitals were more dynamic than public hospitals [17, 63].

Non-parametric estimation methods, mainly data envelopment analysis (DEA), are the most commonly used tools in current research on healthcare efficiency measurement. This method was first introduced by Sherma and Fried [57] for healthcare efficiency assessment [2, 23], followed by O’Neill et al. [47], Hofmarcher et al. [30] and Steinmann et al. [60] for healthcare systems in Austria, Germany and Switzerland, respectively. In the US, Nayar and Ozcan [45] focused on healthcare efficiency in Virginia and found that technically efficient healthcare organisations tend to provide high quality healthcare services. Clement et al. [12] further developed the methodology for assessing the efficiency of hospitals, and incorporated non-desired outputs, which injected new dimensions into the assessment system. Hu et al. [31] conducted an in-depth study on the technical efficiency of healthcare in various provinces in China and explored the main factors affecting efficiency, while Li and He [41] analysed the key factors affecting healthcare efficiency from the perspectives of both government subsidies and market structure [9, 13].

There is a paucity of research related to the mismatch of healthcare resources between regions, while research related to structural efficiency is also in its infancy. The concept of structural efficiency was first introduced by Farrell [20], and Färe et al. [21] further developed the theory by measuring structural efficiency through a linear model. Li and Ng [42] applied structural efficiency to firm studies to assess the overall efficiency of a group of firms. Measuring structural efficiency usually requires the introduction of price information, which can be achieved through shadow price modelling [29, 40]. On the other hand, Zheng and Li [76] take a macro perspective and decompose structural efficiency into individual technical efficiency and inter-provincial allocative efficiency, which provides a new perspective for a comprehensive understanding of healthcare efficiency.

Research on marketisation in the healthcare sector has focused on the impact of the degree of market competition, hospital ownership and hospital class on the efficiency of the healthcare sector. Dalmau et al. [14] found that the technical efficiency of healthcare increases when market concentration is low, i.e., when the market is competitive, which suggests that the market structure has a significant impact on the technical efficiency of healthcare. Sloan et al. [59] analysed the effect of different healthcare ownership on healthcare efficiency and they found significant differences in efficiency between public and private hospitals. Sheikhzadeh et al. [56] developed an efficiency model for public and private hospitals in Iran and the results showed that public hospitals are less efficient than private hospitals. They also suggested that inefficient hospitals could improve their efficiency by transferring, renting and selling idle resources and transferring surplus doctors and nurses to efficient hospitals [54]. Jehu et al. [34] assessed the efficiency of 128 hospitals and found that public hospitals were generally more efficient than private hospitals, but in terms of technical efficiency, public hospitals were lower than private hospitals. Feng and Gravelle [22] stated that the care determines the quality of care, therefore, an increase in the care contributes to the efficiency of healthcare. Kelly and Stoye [37] showed that an increase in private hospitals leads to more healthcare services being received by the patients, which contributes to the overall efficiency.

Based on the summary of the above literature, the following conclusions can be drawn:

Firstly, based on the above related literature, it can be found that there are great differences in resource allocation efficiency and mismatch between supply and demand of healthcare services among Chinese provinces. And in reality, the problems of cross-location medical care and the difficulty and high cost of medical care are prominent and difficult to solve. The problem of cross-location medical care leads to the misallocation of medical resources [43, 46], which leads to the problem of medical resource mismatch between two regions; the input-output structure of medical resources between provinces generates losses, which further raises the inter-regional medical resource mismatch. Based on relevant research and real-life problems, there is a serious inter-regional mismatch of medical resources among China’s provinces.

Secondly, research on the degree of marketisation in the healthcare sector has focused on the introduction of social capital to organise hospitals, thereby contributing to changes in market structure [10]. Private hospitals can compete with public hospitals [14], thus improving the quality of healthcare services [61], increasing the number of healthcare services and lowering the cost of healthcare services in order to meet people’s healthcare needs. However, few studies have been conducted specifically on the impact of changes in the degree of marketisation on the mismatch of healthcare resources between regions. This paper explores the impact of marketisation on the healthcare resource mismatch across China’s regions from this perspective.

Finally, the current research on the impact of market-oriented behaviours on the efficiency of the healthcare industry mostly focuses on the analysis of the impact of market competition [6, 37] and privatisation of the healthcare industry on the efficiency of the healthcare industry, but there are fewer studies related to the degree of marketization of the healthcare industry affecting the mismatch of healthcare resources between regions. In recent years, national policies have strongly supported the development of private hospitals, and the study of the degree of marketisation of the healthcare industry affecting the mismatch of healthcare resources between regions is of great research significance.

Based on the above relevant studies, this paper finds that the main means of market-based reform in the healthcare industry is to introduce social capital into the healthcare industry. Social capital will participate in the competition of the industry [61] and change the degree of marketisation of the healthcare industry [10], which will have a certain impact on the industry, resulting in the healthcare industry in each region will redistribute the healthcare resources, which in turn will reduce the resource mismatch between regions. Therefore, the hypothesis of the benchmark regression of this paper is proposed:

Hypothesis 1

An increase in the degree of marketisation of the healthcare industry can decreasethe healthcare resource mismatch between regions in China.

Regarding the analysis of the mechanism of the degree of marketisation of the healthcare industry to decrease the inter-regional mismatch of healthcare resources, this paper refers to a large amount of literature and selects a few mechanism paths to be analysed. The first is the mechanism in terms of medical service quality. Social capital enters the medical and health care industry, mostly new private medical and health care institutions [5], and then private medical and health care institutions participate in the competition of the medical and health care industry [10]. Due to the first-mover advantage of public healthcare institutions, private hospitals can only compete with public healthcare institutions by improving the quality of healthcare services. With high-quality healthcare services in the region, residents will take local healthcare services due to the inconvenience and high cost of travelling to other places to seek medical treatment, reducing the number of travelling to other places to seek healthcare services, and thus easing the mismatch of healthcare resources between regions [37, 75]. As a result of the entry of social capital, the quality of medical services in the region will be improved, and the mismatch of resources between regions will be alleviated.

Lastly, there is the mechanism in respect of the quantity of medical services. When social capital enters the medical and health care industry, it will inevitably adopt certain strategies in order to compete. As a result of the entry of private medical and health institutions, the number of services in the medical and health industry will inevitably increase, thus better meeting the medical needs of local residents and reducing medical costs, thus reducing the need to travel to other places for medical treatment and turning to local private medical and health institutions for better distribution of medical resources, thus decreasing the mismatch of medical resources between regions.

Based on the above analysis of the relevant mechanisms, the following relevant hypotheses are proposed:

Hypothesis 2

Increased marketisation of the healthcare industry can improve the quality of healthcare services and thus decrease the inter-regional mismatch of healthcare resources in China.

Hypothesis 3

Increased marketisation of the healthcare industry can increase the supply of regional healthcare services in terms of quantity, which in turn decreases the healthcare resource mismatch between regions in China.

The rest of the paper is structured as follows. Methodology section discusses the methodology and results of structural efficiency. Empirical results section presents the dataset and the baseline regression model, and describes the results of the baseline regression, endogeneity test, and robustness test. In Discussions section, further analysis identifies two mechanisms and discusses regional heterogeneity. The section also summarises the policy implications and conclusions.

Methodology

Structural efficiency modelling

How measure the hospital resource misallocation between regions is a difficult issue. Traditional measures of hospital resource mismatch mostly focus on misallocation caused by inadequate input factors [16] an imbalance between supply and demand of hospital services [65], while resource misallocation among regions is difficult to measure by previous methods. Because the output structure of each province is affected by the relative prices of different hospital services, so the output structure of each province is not necessarily optimal globally, but the problem can be solved by shadow prices. The structural efficiency used in this paper then takes this factor into account by introducing a shadow price model [40] that considers the impact of different relative prices of health services on the output structure of each province. The impact of the reallocation of factor inputs among provinces on efficiency gains is also taken into account [42]. These two advantages make structural efficiency a good measure of the hospital resource mismatch among regions. The healthcare resource mismatch between regions in this paper measures the resource mismatch in terms of inputs and outputs of the provincial industry as a whole. Unlike the previous traditional endogenous structural efficiency (Ye et al., 2021), which focuses only on the resource mismatch of input factors, such as labour or capital mismatch [16], this paper is based on a macro perspective and measures the exogenous resource mismatch of inputs and outputs of the industry.

We describe the measurement procedure in this section. Farrell [20] first proposes structural efficiency. And Førsund and Hjalmarsson [24] think that the structural efficiency of a group can be determined by comparing it to the technical efficiency of a typical unit. Ylvinger [73] takes one step further and suggests a new linear technique for a weighted average method. Furthermore, Li and Ng [42] use the shadow price model to solve the price problem. These studies lead to the structural efficiency model used in this paper. The structural efficiency in this paper is measured by the DEA method. We assume that there are J provinces (including municipalities and autonomous regions), each producing inside the production frontier with N inputs and M desired and undesirable outputs. A representative province, as depicted in the image, can only produce at point a and cannot travel to its production frontier. The nation as a whole also produces within its production frontier, or point A in the diagram, but is unable to get to point D.

The country as a whole is made up of distinct provinces, and to distinguish it from individual technical efficiency, we refer to the country’s overall technical efficiency as structural efficiency. To push input capacity to the maximum when the nation’s structural efficiency is inadequate, each province must first significantly raise its technical efficiency. Secondly, each province attempts to modify its product structure and allocate its available resources to those with the highest marginal roles. Furthermore, the national output will rise if resources are allocated among provinces to move inputs from those with low marginal output to those with high marginal output. Thus, the imperfect technical efficiency of the provinces for a given input; efficiency losses caused by the unreasonable structure of output in the provinces; and efficiency losses caused by the unreasonable distribution of inputs among provinces are responsible for the nation’s unsatisfactory structural efficiency.

The next issue is to divide it into three elements and utilize the data to calculate the country’s structural efficiency [42]. We employ a framework to examine the structural efficiency, of a group of manufacturing units [20]. We can assume that each province serves as a production unit, that each province performs the same technical functions related to production, and that the nation is made up of all of these individual provinces. The sum of J provinces represents the nation’s overall output and input:$Y=\mathop \sum \limits_{{j=1}}^{J} {y_j}$,$X=\mathop \sum \limits_{{j=1}}^{J} {x_j}$. Aggregating the overall technology set consisting of these J provinces is:

$${T_g}\left( J \right)=\left\{ {\left( {{X_g},{Y_g}} \right):\left( {{X_g},{Y_g}} \right)=\mathop \sum \limits_{{j=1}}^{J} \left( {{x_i},{y_i}} \right),\left( {{x_i},{y_i}} \right) \in {T_p},j=1, \ldots ,J} \right\}=\mathop \sum \limits_{{j=1}}^{J} {T_p}$$

(1)

The national factors of production are given as ${X_g}$, if $\left( {{X_g},{Y_g}} \right) \in {T_g}\left( J \right)$, Then by properly distributing ${X_g}$ in J provinces, we can produce ${Y_g}$. If in reality the inputs and outputs of province j are $\left( {x_{j}^{0},y_{j}^{0}} \right),j=1, \ldots ,J$, the overall input and output are $\left( {X_{j}^{0},Y_{j}^{0}} \right)=\mathop \sum \nolimits_{{j=1}}^{J} \left( {x_{j}^{0},y_{j}^{0}} \right)$ and the structural efficiency of the country can be defined as $H\left( {{X^0},{Y^0}} \right)=mi{n_\theta }\left\{ {\theta :\left( {\theta X,Y} \right) \in {T_g}} \right\}$. Since ${T_p}$ is a convex set, the structural efficiency of the nation is equal to the technical efficiency of a hypothetical province, $\bar {X}={\raise0.7ex\hbox{${{X^0}}$} \!\mathord{\left/ {\vphantom {{{X^0}} J}}\right.\kern-0pt}\!\lower0.7ex\hbox{$J$}}$, $\bar {Y}={\raise0.7ex\hbox{${{Y^0}}$} \!\mathord{\left/ {\vphantom {{{Y^0}} J}}\right.\kern-0pt}\!\lower0.7ex\hbox{$J$}}$. The structural efficiency of the country at this time is $H\left( {{X^0},{Y^0}} \right)=mi{n_\theta }\left\{ {\theta :\left( {\theta \bar {X},\bar {Y}} \right) \in {T_g}} \right\}$.

The three elements that affect structural efficiency have already been mentioned. It is quite simple to compute the H-value when each province only produces one type of good since the second element, the output structure, need not be taken into account; instead, it is sufficient to add up the maximum possible output from each province and divide it by the actual output. When a province produces numerous goods, the situation is different because different goods are priced differently, the optimal production structure is influenced by the relative pricing of different goods, and the current output structure of each province is not ideal in a global sense. We must estimate the product shadow prices if we want to solve this issue. Finding a shadow pricing vector $\:{p}^{\text{*}}$ that maximizes the shadow revenue (output times shadow price) for the entire nation while maintaining the highest technical efficiency is necessary ($H\left( {{X^0},{Y^0}} \right)*{Y^0}$). Each of these three factors is discussed below.

The first factor is equal to the sum of the shadow income of the highest potential output in each province, according to Li and Ng’s [42] analytical methodology (${R^{TE}}=\mathop \sum \nolimits_{{j=1}}^{J} F\left( {x_{j}^{0},y_{j}^{0}} \right){\text{*}}{p^{\text{*}}}{\text{*}}y_{j}^{0}$), and divided by the shadow income of the actual national production (${R^0}={p^{\text{*}}}{Y^0}$), it is also equal to the weighted average of the technical efficiency index of each province.

$$ATE=\frac{{{R^{TE}}}}{{{R^0}}}\frac{{\mathop \sum \nolimits_{{j=1}}^{J} F\left( {x_{j}^{0},y_{j}^{0}} \right){p^*}y_{j}^{0}}}{{{p^*}{Y^0}}}=\mathop \sum \nolimits_{{j=1}}^{J} {\omega _j}F\left( {x_{j}^{0},y_{j}^{0}} \right)$$

(2)

where ${\omega _j}=\frac{{{p^{\text{*}}}y_{j}^{0}}}{{{p^{\text{*}}}{Y^0}}}$, j = 1,…, J, which we call the intra-provincial technical efficiency.

The second factor can be used as Farell’s [20] indicator of resource allocation efficiency. As an independent production unit, each province’s efficiency can be described as: $AE\left( {{p^{\text{*}}},x_{j}^{0},y_{j}^{0}} \right)=\frac{{t{r_j}}}{{F\left( {x_{j}^{0},y_{j}^{0}} \right){p^{\text{*}}}{y^0}}}$, where $t{r_j}=mi{n_u}\left\{ {{p^{\text{*}}}u:\left( {x_{j}^{0},u} \right) \in {T_p}} \right\}$ and the sum of the highest shadow income of each province is ${R^{AE}}=\mathop \sum \nolimits_{{j=1}}^{J} t{r_j}$, and thus

$$AAE=\frac{{{R^{AE}}}}{{{R^{TE}}}}=\frac{{\mathop \sum \nolimits_{{j=1}}^{J} t{r_j}}}{{{R^{TE}}}}=\mathop \sum \nolimits_{{j=1}}^{J} \frac{{F\left( {x_{j}^{0},y_{j}^{0}} \right){p^*}y_{j}^{0}}}{{{R^{TE}}}}\frac{{t{r_j}}}{{F\left( {x_{j}^{0},y_{j}^{0}} \right){p^*}y_{j}^{0}}}=\mathop \sum \nolimits_{{j=1}}^{J} {\tau _j}AE\left( {{p^*},x_{j}^{0},y_{j}^{0}} \right)$$

(3)

where ${\tau _j}=\frac{{F\left( {x_{j}^{0},y_{j}^{0}} \right){p^{\text{*}}}y_{j}^{0}}}{{{R^{TE}}}}$, and we call AAE the resource allocation efficiency.

The third factor is the efficiency loss caused by the unreasonable interprovincial factor allocation, which is partially equal to

$$RE=\frac{{H\left( {{X^0},{Y^0}} \right)\left( {p{Y^0}} \right)}}{{{R^{AE}}}}~$$

(4)

We call RE the interprovincial factor allocation efficiency. At this point, $H\left( {{X^0},{Y^0}} \right)=ATE{\text{*}}AAE{\text{*}}RE$. The relationship among the above three indicators can be simplified using Fig. 1.

Fig. 1

Provincial and national efficiency. a The Iso-quant frontier of a province. b The Iso-quant frontier of the nation

The national production frontier is on the right, and the provincial production frontier is on the left. The structural efficiency of the nation can be determined as$OD/OA$, with the shadow price represented by the diagonal line pp’, if we observe that the actual output of the province is a and the real output of the nation is A. There are three reasons why national structural efficiency hasn’t reached its full potential. The first is the technical efficiency (ATE) of each province. This efficiency loss can be eliminated if the provinces on the left are able to make the transition from point a to point b production. Suppose that when all provinces reach the highest technical efficiency, total output is at point B on the right. Since points A and B have different output structures, we can only compare the incomes or shadow prices of their production. Drawing a diagonal line parallel to pp’ gives us point B; from this we can derive, $OB/OA$ the precise technical efficiency of each province. At a given shadow price, the province that produces at point b does not have the highest income; it can only obtain the highest income by switching to production at point c. The total structural efficiency of output is $OC/OB$, and the total structural efficiency of output is (AAA). The total output structure efficiency is $OD/OC$ and the remaining percentage is the inter-provincial factor allocation efficiency (RE). Clearly $OD/OA=\left( {OD/OC} \right)\left( {OC/OB} \right)\left( {OB/OA} \right)$. The structural efficiency of the country can also be decomposed into the three factors mentioned above.

Below we define provincial indicators of the resource mismatch in the healthcare sector, $AEARE=RE{\text{*}}AAE$, it encompasses the losses caused by distortions in the input and output structure and also measures the extent of misallocation of resources, with larger values indicating greater losses.

Input and output data in the model

We use data from the Chinese Health Statistics Yearbooks, which covers the years 2011 to 2020 and details the fundamental evolution of hospitals throughout 31 Chinese provinces, the actual yearbook reflects data from 2010 to 2019. During this period, the government strongly encouraged social investors to enter the healthcare industry, the threshold for social capital to enter the healthcare industry was lowered, and it began to acquire or build new healthcare institutions, and private hospitals grew by leaps and bounds [15, 41]. Using data from this time period, it is possible to better study whether state support for marketisation of the healthcare industry can decrease the mismatch of healthcare resources between regions in China’s provinces.

This paper refers to the literature of Retzlaff et al. [51], Hu et al. [31], and Li and He [41] for the selection of input and output variables. As the study object is the healthcare sector in China, including hospitals and primary healthcare organisations and excluding other health organisations. Currently, the relevant data are only available for 31 provinces, and without generating the curse of dimensionality, this paper selects five main input and output variables for measuring efficiency, with two inputs, two desired outputs, and one undesired output.

Healthcare beds and medical technicians are key input variables that affect the output and efficiency of the industry [27]. Personnel inputs used the number of physicians (DOCTOR) Since provincial data in China lacks input data on medical equipment in the healthcare sector, this paper uses the number of beds (BED) as the equipment input.

Outputs are categorised into desired and non-desired outputs. The number of Outpatient visits (OPV) and the number of surgeries (SUG) in the healthcare industry are desired outputs. The number of Outpatient visits includes the number of consultations and the number of surgeries as the number of medical services.

Whereas the number of surgical deaths (SUD) is a non-desired output and is the main non-desired output of the healthcare industry. The number of surgical deaths can be estimated based on the surgical mortality rate and the total number of surgeries in the data. The input-output variables required for the model are selected in this paper.

In Table 1, we provide descriptive statistics of inputs and outputs. As can be seen from the above table, in terms of inputs, there is a big difference between the maximum and minimum values of the number of physicians in the healthcare industry in each province of China, which is 50 times, while the mean value of the number of physicians in each province is 350,000 personnel; regarding the number of beds in the input variable, the healthcare industry in the province with the largest number of beds has 667,200 beds, while the healthcare industry in the province with the smallest number of beds has 8,352 beds, with a wide range of beds in each province. number varies greatly. In terms of output, the number of medical services provided by the healthcare sector in each province has been increasing year by year. The minimum number of consultations is 10 million, and the maximum number is nearly 900 million. This indicates that the output the healthcare sector varies greatly across provinces, with the mean level at 200 million visits. The number of surgeries is low relative to the number of consultations, with the healthcare sector in the largest province providing more than eighty million and the smallest two hundred and seventy thousand. As for the undesired outputs, the provinces with the highest number of surgical deaths had about 160,000 and the provinces with the lowest had more than 30,000 deaths. The above relevant data illustrate that there are great differences in the input factors and output levels among the regions.

Table 1 Definitions and descriptive statistics of inputs and outputs

From the table of input-output variables alone, it can be seen that there are great differences in the allocation of resources in the healthcare industry among the provinces in China, which may be due to the local economic level, population density and geographic location, etc. At the same time, different levels of inputs and outputs can be calculated as input-output ratios of each region, and the results also indicate that there is a serious mismatch of healthcare resources among the regions of the provinces in China. It is necessary to study whether the market-oriented behaviour of the healthcare industry can decrease the mismatch of healthcare resources between regions.

Model measurement results

Table 2 is the result of annual averages of the levels of cost efficiency, allocative efficiency, technical efficiency and resource mismatch for each province in China, and there is a ranking of the cost efficiency of each province. The cost efficiency is higher in provinces such as Zhejiang, Guangdong and Hainan, and lower in Liaoning, Heilongjiang and Hebei; the allocation efficiency is also higher in provinces such as Jiangsu, Zhejiang and Guangdong, and lower in Qinghai and Hebei; the technical efficiency is higher in provinces such as Shanghai, Guangdong and Beijing, and lower in Tibet and Heilongjiang; and the mismatch of healthcare resources is higher in provinces such as Tibet, Hainan and Shaanxi, and lower in provinces such as Shandong, Shanghai and Hebei. From the perspective of the location of the provinces, the mismatch of medical resources is lower in coastal provinces and higher in non-coastal regions; the mismatch of medical resources is lower in eastern provinces and higher in central and western regions; and it can be found that the mismatch of medical resources is not very low in some of the regions with developed medical standards. The above results show that there are different degrees of inter-regional mismatch of medical resources in different regions of China.

Table 2 The efficiency and the rationalization of hospital resource allocation by province, 2011–2020

Empirical results

Descriptive statistics for regression variables

The main purpose of this paper is to investigate whether an increased degree of marketisation of the healthcare industry can decrease the mismatch of healthcare resources among regions in Chinese provinces, so the degree of marketisation of the healthcare industry is selected as the core explanatory variable. The first relevant indicator of marketisation in the healthcare industry that comes readily to mind is the proportion of private hospitals [39, 41]. The marketisation behaviour of the healthcare industry is mainly to introduce the power of social capital. Social capital has a great advantage in terms of capital, and has a large amount of money to buy or build new private hospitals as a way to participate in the construction of the healthcare industry and have an impact on the healthcare industry, so it can represent the degree of marketisation. Therefore, the indicator of the degree of marketisation of the healthcare industry in this paper is the proportion of private hospitals. This indicator can measure the degree of marketisation of China’s healthcare industry and can be measured based on currently available data. The higher the proportion of private hospitals, the higher the marketisation of the healthcare industry in each Chinese province.

The control variables selected in this paper mainly refer to related studies [41, 74]. The control variables were selected as per capita gross domestic product (RGDP), proportion of tertiary hospitals (PTH), population density (TPD), government health subsidy (GHCP), PM2.5 concentration (PM), and personal hospitalisation expenditure as a proportion of consumption (RSU) [31, 41, 58].

Gross domestic product (RGDP) per capita: expressed as the logarithm of GDP per capita. The healthcare sector in a region is influenced by the the local economy, while GDP per capita is an important control variable when using provincial data.

Proportion of tertiary hospitals (PTH): expressed as the ratio of tertiary hospitals to the total number of hospitals. The proportion of tertiary hospitals represents the medical technology in a region.

Population density (TPD): expressed using total population divided by administrative area. TPD represents the healthcare demand of a region, and has an important impact on the healthcare industry in each province for the layout of healthcare resources.

Government health subsidy (GHCP): Expressed as government health subsidy. GHCP represents the influence of government power on the healthcare industry in each province of China, and other important confounding factors need to be controlled for when analysing the impact of marketisation.

PM2.5 concentration (PM): The concentration of PM2.5 in a province, which represents the air quality in a region; regions with low air quality will have a higher demand for healthcare resources and therefore need to control for air quality factors.

Personal hospitalisation expenditure as a proportion of consumption (RSU): expressed as the ratio of personal hospitalisation expenditure to personal consumption expenditure. The ratio of personal hospitalisation expenditure to personal consumption expenditure reflects the willingness of the population to spend on medical services.

The variables of descriptive statistics are shown in Table 3.

Table 3 Descriptive statistics of regression variables

Baseline regression model

In this paper, we will investigate the impact of the degree of marketisation of the healthcare industry on the healthcare resource mismatch between regions in Chinese provinces. Following the example of Hu et al. [31] and Li and He [41] the proportion of private hospitals is selected as a measure of the degree of marketisation of the healthcare industry (PPH). The control variables are gross domestic product per capita (RGDP), proportion of tertiary hospitals (PTH), population density (TPD), government health subsidy (GHCP), PM2.5 concentration (PM), and personal hospitalisation expenditure as a share of consumption (RSU). The following are the regression estimation equations for the baseline regression.

$$\begin{array}{*{20}{c}} {AERA{E_{it}}={\beta _1}PP{H_{it}}+{\beta _2}RGD{P_{it}}+{\beta _3}PT{H_{it}}+{\beta _4}TP{D_{it}}+{\beta _5}GHC{P_{it}}+~{\beta _6}P{M_{it}}} \\ {+{\beta _7}URBANR{T_{it}}+C+{\mu _i}+{\theta _t}+{\varepsilon _{it}}} \end{array}$$

(5)

The degree of marketisation of the healthcare industry (PPH) serves as the main explanatory variable. It is expected that β1 < 0, which means that the healthcare resource mismatch between regions in Chinese provinces will decrease with the increase of the proportion of the degree of marketisation of the healthcare industry, i.e., an increase in the degree of marketisation of the healthcare industry can reduce the healthcare resource mismatch between regions in each province. Therefore, Chinese provinces can decrease the mismatch of healthcare resources between regions by increasing the degree of marketisation of the healthcare industry.

Baseline regression results

The regression equation in Eq. (5) is first estimated, in this paper controlling for province and year fixed effects, and the results are displayed in Table 4. Where the results in columns (1)-(3) are obtained by replacing AERAE with RE, AAE, and ATE, it can be seen that the results for RE and AAE are not significant, and the results for ATE are more significant; therefore, this paper focuses the study on the level of healthcare resource mismatch between regions, and focuses on the results in columns (4) and (5). The degree of marketisation in columns (4) and (5) are both the proportion of private hospitals. The relevant control variables are added in column (1), and the coefficient on the degree of marketisation is significantly negative at the 1% level, suggesting that an increase in the degree of marketisation in the healthcare sector can decrease the healthcare resource mismatch between regions in China; time and individual fixed effects are added in column (2), and the results remain consistent. In summary, the increased degree of marketisation in the healthcare industry can decrease the medical resource mismatch between regions in China.

Table 4 Baseline regression results

The previous regression results indicate that an increase in the proportion of private hospitals has a favourable impact on the decrease of resource mismatch in the healthcare sector in all provinces of China. The results of the benchmark regression suggest that Explanation 1 holds. An increase in the number of private hospitals has the potential to increase the amount of hospital resources and services [10], improve the standard of care in hospitals [37], and promote the development of hospital technology, and thus the degree of marketisation in the healthcare industry decreases the resource mismatch in the healthcare industry in each province in China. Since there may be reverse causality between the degree of marketisation of the healthcare sector and the resource mismatch in the healthcare sector across provinces, the next section focuses on this endogeneity.

In this paper, the VIF test was performed on the selected control variables. The results are shown in Table 5, the VIF test values of all control variables are within 10, indicating that there is no strong correlation between the control variables, so the results of the benchmark regression are more accurate and robust.

Table 5 VIF test results

Endogeneity test

A high proportion of private hospitals in a region indicates an abundance of local hospital resources, a high level and quality of hospital services, and a low resource mismatch in the healthcare industry in all provinces. And when the resource mismatch in a region’s healthcare industry is low, the proportion of private hospitals increases. This is because local medical resources are sufficient, demand is high, and there is sufficient competition in the market. These advantages attract social capital, and the proportion of private hospitals increases. Therefore, there may be a reverse causality between private hospitals and the resource mismatch in the healthcare sector.We implemented instrumental variables (IV) regression and two-stage least squares (2SLS). To more effectively address the endogeneity issue, the Bartik instrumental variable [4, 28] is used to build the first instrumental variable on the percentage of private hospitals in the hospital industry.

We assume that at the beginning of the period (e.g., if we set 2011 as the beginning of the period), the proportion of private hospitals is exogenously determined during the window period, that the proportion of private hospitals changes over time, and that the process of changing the mismatch of resources in the hospitals sector is primarily responsible for the proportion of private hospitals at the end of the period (in this paper, 2020). In order to create Bartik instrumental variables, we aim to use the proportion of private hospitals at the beginning of the period and the exogenous expected value of the growth rate of hospitals in the province. Using the projected national growth rate of the health care industry, excluding province c, for both periods, one can calculate the expected value of the growth rate of private health care organisations for each industry j in province I for both periods. This projected value represents the general growth trend of the national healthcare industry (from the province outwards) and is therefore independent of the resource mismatch in the healthcare industry in that province. This is because there is a necessary correlation between the growth rate of output in the national healthcare sector industry and the growth rate of the healthcare sector in the province. Therefore, using the predicted values as an instrumental variable for the proportion of private hospitals in each province better fulfils the two conditions for an instrumental variable.

The projected value of the proportion of private hospitals in a province is:

$$PPH{1_{it}}=PP{H_{i0}}*(1+{g_{ijt}})$$

(6)

The instrumental variable for the percentage of private hospitals in each province is given by the aforementioned equation. This technique is initially used to create instrumental variables for factors related to income disparity. It is used in this paper to estimate the share of private hospitals in each province. The results of the regression technique employing two-stage least squares (2SLS) and instrumental variables (IV) are displayed in Table 6.

Table 6 Endogeneity regression results

The above table shows the results of regression using Bartik instrumental variables. According to the results in column (1), The F-test value for the first stage is 13.74, which is greater than 10 and is eligible for the instrumental variables, and they are significantly correlated with the explanatory variables and satisfy correlation. Also, the instrumental variables are not correlated with the explanatory variables and satisfy exogeneity, so the above two conditions indicate that the instrumental variables are valid. According to the conclusion in column (2), the increase in the proportion of private hospitals can decrease the resource mismatch in China’s healthcare industry. This result is consistent with the results of the benchmark regression. Even after considering the endogeneity factor, the increase in the proportion of hospitals structures in each province can still decrease the healthcare resource mismatch between regions.

Robust analysis

Through the previous basic analysis of the sample, it can be seen that the increased marketisation of the healthcare industry in each province can decrease the mismatch of healthcare resources between regions; in order to further verify the role of this improvement, this paper will explore the robustness of the baseline regression results from the following two aspects. Firstly, replacing the model, for the related research on efficiency, some studies believe that Bootstrap sampling test method is more suitable for the regression estimation of efficiency, so in the first type of robustness test replacing the model used in the benchmark regression estimation, replacing the Bootstrap model for the estimation of the regression [38, 44, 62]. Secondly, adopting the replacement of the explanatory variables of the metrics, using the study of Ji et al. [35] to measure the resource mismatch in the healthcare industry by the number of changes in the number of employees in the healthcare industry in each province, which is a commonly used measure in domestic research on resource mismatch; in addition, the ratio of the number of inpatient services in private hospitals to the total number of inpatient services was used to replace the ratio of private hospitals [33, 70]. At the same time, due to the fact that there are relevant policies to support the development of public hospitals in 2015 and 2017, this paper is concerned about the interference with the content of the study, so the data of 2015 and 2017 were removed from the regression. The results of the four robustnesses are shown in the following Table 7.

Table 7 Robustness analysis results

As can be seen from the table above, the results in column (1) are those obtained from the benchmark regression using the Bootstrap model. The coefficient of the degree of marketisation is significant at the 5% level and is significantly negative, which is consistent with the results of the benchmark regression. This suggests that an increase in the degree of marketisation of the healthcare sector in China’s provinces reduces the mismatch of healthcare resources between regions, even after taking into account the fact that the regressions use different models. Column (2) is a robustness test by replacing the measure of the healthcare resource mismatch, using the change in the number of employees in the healthcare industry as a measure of healthcare resource mismatch. The supply of medical resources in the region and the inter-regional medical resource mismatch are closely related. Generally speaking, the supply of medical resources in the region is sufficient, the input structure of the industry is reasonable, and the output is more, which indicates that the inter-regional medical resource mismatch is also lower. The results in column (2) are consistent with the results of the benchmark regression.

Since private hospitals may differ significantly from public healthcare facilities in terms of output and looking only at volume may be somewhat one-sided, the proportion of inpatient services in private hospitals was used to replace the proportion of private hospitals in the robustness analyses. From the results in column (3) of the above table, it can be seen that the coefficients of the degree of marketisation are all significant at the 5% level, and the degree of marketisation of the healthcare industry in each province has increased, the number of inpatient services provided by private hospitals has increased, which promotes the output of healthcare services in the healthcare industry, which can better satisfy the healthcare needs of the local residents, and leads to a more rational allocation of healthcare resources, that is, the provinces in China The mismatch of medical resources between regions also declined. Column (4) is to remove the data for 2015 and 2017, and the results are consistent with the benchmark regression, so the relevant policies do not interfere with the findings of this paper. Therefore, the increased marketisation of China’s healthcare industry can decrease the mismatch of healthcare resources between regions.

Discussions

Results discussions

The previous analyses found that increasing the proportion of private hospitals in each province could decrease the resource mismatch in healthcare resources across China’s provinces. However, it is not clear how increasing the proportion of private hospitals affects the resource mismatch in healthcare institutions in each province. This paper examines how increasing the proportion of private hospitals affects the resource mismatch in hospitals in each province from the perspectives of resource supply in the healthcare industry and healthcare service output, in order to open the “black box” for decreasing the resource mismatch in healthcare institutions in each province.

Mechanism analysis of medical resources supply

Seiford and Zhu [55] find that, from the standpoint of hospital resource supply, it is possible to enhance the desired output in the DEA model by increasing the input or decreasing the undesirable output, hence improving efficiency. The impact of increasing the proportion of private hospitals on the supply of resources to healthcare organisations can be assessed to determine whether the mechanism is justified. The number of beds and the number of physicians in a healthcare facility symbolise the total available resources in the sector [27, 49], so the total number of beds (HBN) and the number of physicians (TPT) are used in this study to represent the resource supply in the healthcare sector. Table 8 shows the regression results.

Table 8 Results of the mechanism analysis of medical resources supply

The table above shows that an increase in the proportion of private hospitals in each province can increase the total number of hospital beds (HBN) and the number of physicians (TPT) in the healthcare industry in each province. This also proves that Hypothesis 2 is valid. An increase in the proportion of private hospitals in each province will lead to an increase in the number of beds and the number of physicians in the healthcare sector, which can increase the expected output, meet the hospitalisation needs of more people, and ultimately decrease the mismatch of healthcare resources between regions [64, 67]. Therefore, an increase in the proportion of private hospitals can decrease the mismatch of healthcare resources between regions by providing more healthcare resources.

Mechanism analysis of medical services outputs

In terms of the quantity of medical services, the number of inpatient services, the number of surgeries and the number of outpatient emergencies are the main medical services provided by healthcare institutions; therefore, the number of inpatient services (MF), the number of surgeries (CTF) and the number of outpatient emergencies (OF) can represent the medical service output in the healthcare sector [19]. As the proportion of private hospitals increases, in order to compete, they will provide a greater number of healthcare services, including the number of inpatient services, the number of surgeries, and the number of outpatient emergencies, which can better satisfy the demand for healthcare services of the local residents and reduce the behaviour of seeking medical treatment in other places, thus decreasing the mismatch of healthcare resources between regions [31, 41]. The regression results are shown in Table 9.

Table 9 Results of the mechanism analysis of medical services outputs

It can be seen that an increase in the proportion of private hospitals in each province increases the number of inpatient services, the number of surgeries, and the number of outpatient and emergency room visits, which suggests that the mechanism for decreasing the mismatch of healthcare resources in each region is valid. The results of the regression also prove that Hypothesis 3 is valid.

Since this paper focuses on the healthcare resource mismatch between regions in China, regarding the study of heterogeneity, we mainly focus on the analysis of heterogeneity between regions, which includes whether or not it is a coastal region and regional heterogeneity between the East and West regions.

Heterogeneity for coastal and non-coastal regions

Due to the different economic levels, population densities, geographic locations, and provincial policies in each province, which results in inevitable differences in the intensity of privatisation of the healthcare industry in different provinces [31], and significant differences in the resource mismatch in the healthcare industry in each province, it is necessary to analyse the proportion of private hospitals on the mismatch of healthcare resources in each province in terms of regional variability improvement effect. In this paper, we explore the heterogeneity of this effect from the perspectives of whether it is a coastal region or not, and from the east-central and west-central regions. Table 10 shows the results of the heterogeneity in terms of whether it is a coastal region or not.

Table 10 Heterogeneity regression results for coastal and non-coastal regions

From the above table, it can be seen that there is a significant difference in the impact of increasing the proportion of private hospitals on the resource mismatch in the healthcare industry in all provinces of China. There is no significant decrease in the resource mismatch in the healthcare industry in coastal areas, but the effect of decreasing the resource mismatch in the healthcare industry in non-coastal areas is very obvious.

Coastal areas are more economically developed, social capital to set up private hospitals will give priority to coastal areas, so the degree of privatisation of the healthcare industry in coastal areas is higher, and the introduction of social capital at this time does not have a significant effect on decreasing the mismatch of healthcare resources between regions, whereas in non-coastal areas, due to a lack of capital in the region, the introduction of social capital has an effect on decreasing the mismatch of healthcare resources between regions. The introduction of social capital in non-coastal areas, due to the lack of regional capital, will decrease the mismatch of healthcare resources between regions, which is obvious.

Heterogeneity for east central and west coast regions

The regional heterogeneity analyses in this section are analysed from three regions: the East and the West [32]. China’s three regions, East, Middle and West, have large differences in the development of the healthcare industry in each region due to factors such as geographic location, economic development and population, so it is necessary to explore the impact of the degree of privatisation of the healthcare industry on the mismatch of healthcare resources between regions by dividing the three regions according to the three regions, East, Middle and West [53]. The regression results are shown in Table 11.

Table 11 Heterogeneity regression results for east, central and west regions

From the regression results in the table above, it can be seen that the heterogeneity results are significant only in the central region, indicating that an increase in the privatisation of the healthcare sector in the central region will greatly decrease the inter-regional mismatch of healthcare resources; while the results in the eastern and western regions are not significant. This may be due to the fact that the degree of privatisation of the healthcare industry in the eastern region is higher, and the introduction of social capital into it is less effective; while in the western region, due to the backwardness of the economic level and the small population, social capital will not choose to go in to build private hospitals, and therefore will have little effect on the decrease of the eastern region and the western region.

Method discussions

The structural efficiency approach used in this paper decomposes the loss of resource mismatch and constructs a healthcare resource mismatch indicator that more conveniently measures the extent of healthcare resource mismatch between regions. At the same time, various input-output variables are taken into account, and non-desired outputs are used as inputs to control the minimization, with a view to obtaining maximized outputs.

The current methodology has some shortcomings. First, the input-output variables of the model are not comprehensive and rich enough due to data limitations, which can be expanded by future researchers; second, the model focuses on resource mismatch between regions, which is a macroscopic perspective, and pays little attention to resource mismatch on the micro side. Future researchers can conduct research from this perspective.

Policy implications

The paper finds the following two policy implications. By increasing the degree of marketisation in the healthcare sector in Chinese provinces, the supply of healthcare resources and the output of healthcare services in the healthcare sector can be boosted to meet the healthcare needs of the population [26]. At the same time, the government can focus on the non-coastal and central regions and encourage the introduction of social capital to these regions to provide the degree of marketisation, thereby increasing the healthcare in the hospital sector in the provinces. Last, countries can bring in the power of social capital or foreign capital to develop healthcare institutions, and, adjust the allocation of medical resources between districts can better meet the medical needs of local residents.

Conclusions

This paper assesses the resource mismatch in the healthcare sector in China’s provinces using data from 2010 to 2019. We find that the resource mismatch in China’s provincial healthcare sector has some room for improvement, and the resource mismatch is lower in hospitals in China’s coastal areas relative to non-coastal areas. In addition, the benchmark regression finds that increased marketisation of the healthcare industry has a positive effect on decreasing the healthcare resource mismatch between provinces and regions. And finds the mechanism of its impact mainly in terms of the supply of resources in the healthcare industry and the output of healthcare services. The marketisation of the healthcare industry can accommodate more patients and promote the development of hospital technology, thus decreasing the healthcare resource mismatch between regions in each province. And this effect varies according to regional differences, with a more significant impact in non-coastal and central regions.