Has the Wind Power Price Policy Promoted the High-Quality Development of China’s Wind Power Industry?—Analysis Based on Total Factor Productivity

Abstract

The wind power price policy has promoted the rapid development of the wind power industry in China. However, China’s wind power industry is facing high-quality development problems such as wind curtailment and blind investment. Exploring the relationship between the change in wind power price policy and China’s high-quality development of wind power is of great significance for the energy system to achieve carbon neutrality. This paper constructs an SBM-GML global covariance model, calculates the total factor productivity of wind power in 30 provinces in China from 2015 to 2019 and conducts index decomposition, and selects provincial panel data from 2015 to 2019 to empirically test the impact of the wind power price policy on the total factor productivity of wind power in China. The results show that the wind power price policy can significantly improve the total factor productivity of wind power, and the results of the endogenous test and robustness test show the effectiveness of the model. The wind power price policy is helpful to promote the technological progress of wind power, affect the reduction of the price difference between wind power benchmark price and coal power benchmark price, and then promote the improvement of wind power total factor productivity so as to meet the policy requirements of wind power full grid parity. In addition, the impact of wind power price policy on the total factor productivity of wind power has obvious regional heterogeneity. Future price policy formulation should clarify industry development expectations, consider regional differences, and establish a sound, market-oriented electricity pricing mechanism.

1. Introduction

The wind power price policy mainly includes a fixed feed-in tariff mechanism and a renewable energy quota system. The fixed feed-in tariff mechanism is essentially an electricity price policy with subsidies [1], which is the key to the early development of renewable energy. In the long run, market-oriented policies will only take effect when renewable energy technology is mature and fully integrated into the electricity market [2]. The fixed feed-in tariff mechanism will encourage renewable energy manufacturers to invest as much as possible until the marginal cost of renewable energy electricity is equal to the feed-in tariff of renewable energy power implemented by the government [3]. At the same time, the small differences in fixed feed-in tariff subsidies between regions will have a significant impact on the development of regional renewable energy. Therefore, by increasing the feed-in tariff level in resource-poor areas, it can promote an increase in renewable energy installed capacity [4]. At present, China’s fixed electricity price mechanism model for wind power is a variable electricity price subsidy. This model is divided into different resource areas according to the endowment of wind energy resources, and the corresponding feed-in tariff is approved based on the average cost of different resource areas. The portion exceeding the benchmark electricity price of coal power is subsidized through the renewable energy development fund [5]. From 2012 to 2021, China’s wind power added 28.43 million kW of installed capacity annually, with a cumulative installed capacity exceeding 340 million kW (National Energy Administration, 2022).

The price support policy has supported the large-scale development of China’s wind power industry, formed a relatively complete industrial technology system, and achieved the original intention of policy formulation. However, the wind power industry is also facing increasingly prominent high-quality development problems, such as the serious phenomenon of “wind curtailment and power rationing”, wind power enterprises falling into a loss trap, a huge gap in subsidy funds, and blind expansion of wind power installed capacity. As shown in Figure 1, the development of the wind power industry fell into a trough around 2012. In 2012, the total amount of wind curtailment and power rationing in China exceeded 10 billion kilowatt hours, and the average utilization hours decreased significantly (National Energy Administration, 2013).

At this time, the government should intervene appropriately to regulate the behavior of the market participants and safeguard the development of high-quality [6]. Subsequently, the government gradually lowered the benchmark electricity price in various resource areas in 2014, 2016, and 2019. From 2021, the newly approved onshore wind power projects will be fully implemented at parity and will not be subsidized. High-quality development mainly solves the problem of structural imbalance in the development process. The unbalanced supply and demand of power products lead to overcapacity [7], which requires the role of the price mechanism. As shown in Figure 2, from a policy design perspective, the changes in wind power benchmark electricity prices are mainly concentrated from 2014 to 2019, and it is a downward trend. From a practical perspective, by 2021, the two conditions that need to be met for full grid parity have been realized: decrease in investment costs and solve the problem of “wind curtailment and power rationing”. On the one hand, the wind turbines have been enlarged, and the average bidding price of wind turbines has been reduced by more than 30%. On the other hand, the national average utilization hours of wind power are 2097 h, and the national average wind curtailment rate is 3%, down one percentage point year-on-year, and the situation of “wind curtailment and power rationing” has improved. Effective institutional arrangements are a prerequisite for high-quality development [8]. Does this high-quality development benefit from the wind power price policy, and what are the effects of high-quality development? Both require us to further analyze the changes in the total factor productivity of wind power.

Total factor productivity (TFP), a productivity indicator that measures the quality of high growth in an industry, is a concentrated expression of efficiency change and power change [9]. The high-quality development of the wind power industry relies on the continuous improvement of total factor productivity [10]. In addition, the high-quality development of the wind power industry is increasingly constrained by the wind curtailment rate [11], and there are few global calculations of the TFP of the wind power industry that consider unexpected outputs at home and abroad. Therefore, the TFP considering the wind curtailment rate can better measure the high-quality development level of the wind power industry. This article aims to answer the following questions: firstly, how does the TFP of wind power change when considering unexpected output, and is technological progress the main source driving the change in the TFP of wind power? Secondly, to measure the effectiveness of the government’s electricity price policy and explore the relationship between the wind power price policy and TFP of wind power. Finally, if the wind power price policy promotes the increase of TFP, what is the mechanism behind it?

Based on the above problems, this article attempts to introduce the SBM-based GML index model [12] to evaluate the TFP of wind power in various provinces in China and tries to innovate the existing literature from the following aspects: (1) Based on the SBM-DEA model, considering wind curtailment rate as an unexpected output, calculating the TFP of wind power in each province, and studying the dynamic changes of wind power TFP by using the GML index. (2) Due to the wind power benchmark electricity price being adjusted for the first time in 2014 and changed to a guide price in 2019, this article takes the guide price as the highest limit for wind power bidding, uses provincial panel data to empirically analyze the internal impact of wind power price policy on the TFP of wind power in various provinces in China from 2015 to 2019, and analyzes the theoretical basis for achieving full grid parity. (3) Empirical analysis of the heterogeneous impact of wind power price policy on TFP, and then a reasonable reference for the formulation of wind power price policy.

2. Literature Review

Quantitative research on wind power price policy at home and abroad has focused on the past decade, but there are few empirical studies on the relationship between wind power price policy and TFP of wind power, and much of the literature mainly focuses on the effectiveness and issues of wind power price policy [13,14]. At present, quantitative research on the TFP of wind power in China mainly focuses on the DEA model without considering the wind curtailment rate as an unexpected output [11,15]. Therefore, this article will measure the TFP of wind power by constructing an SBM-DEA model that includes unexpected outputs and then explore the relationship between wind power price policy and TFP, as well as the underlying logic, to fill the gap in the literature.

Domestic and foreign literature related to the article mainly focuses on the following three aspects: the effectiveness of the wind power price policy, issues with the wind power price policy, and the calculation of wind power TFP.

First, the effectiveness of the wind power price policy. Wind power benchmark electricity price, as a feed-in tariff policy, is an important means to ultimately achieve marketization of electricity price. The primary purpose of the wind power feed-in tariff policy is to promote the development of the wind power industry and realize the effectiveness of the policy. Guo [14] used three policies, including benchmark electricity price, as explanatory variables to study the effectiveness of renewable energy policies. The results showed that the benchmark electricity price had a significant promoting effect on renewable energy power generation. Jamil, M.N. [16] found that energy consumption and electricity are important factors affecting economic development. With the impact of carbon reduction, the impact of new energy policies is becoming increasingly prominent. Ritzenhofen [17] found that the feed-in tariff policy can effectively improve the power supply of renewable energy through a dynamic long-term model simulation. Huang [5] found that the subsidy policy did not play a significant role in promoting the wind power industry through empirical analysis of the performance of the wind power feed-in tariff policy on the wind power industry, and other supporting regulatory policies should be introduced. Zhao [18] analyzed the industrial effect of the feed-in tariff policy by establishing a multi-level model. The results showed that the wind power feed-in tariff policy has a significant industrial effect, and industrial development was related to the time, quantity, and similarity of policy implementation.

Second, issues with the wind power price policy. Compared with the industrial perfection of developed countries, there are still some problems in China’s wind power industry, such as policy mechanisms and industrial fluctuation. Lin [19] constructed a model by using a stochastic dynamic recursive method to quantitatively evaluate China’s wind power benchmark electricity price policy. They found that an appropriate increase in wind power benchmark electricity price can maximize policy benefits, but wind power development still requires other policy support. Guo [20] analyzed the data of the wind power industry from 2006 to 2016 and found that the wind power benchmark electricity price has promoted the rapid development of the wind power industry, but it also exposed problems such as wind curtailment and blind investment. It is necessary to adjust wind power policies to promote the healthy development of the wind power industry. Devine [21] found that the feed-in tariff system needs to consider the risk preferences of market participants, and the policy is optimal when the risk allocation is in the middle. Yuan [22] researched the evolution of China’s wind power policy in the past 30 years and believed that policy support for wind power should continue to be increased, and the feed-in tariff system should be improved by matching technological progress. Xia [23] argue that while the feed-in tariff system brings investment returns, it can also lead to excessive investment and an increase in wind curtailment rate, which can be reduced through policy design. Liu [24] explored the adaptability of the fixed feed-in tariff system to electricity marketization and considered that the fixed feed-in tariff system raises the wind curtailment rate, and the consumption quota ratio should be set to promote the integration of them.

Third, the calculation of wind power TFP. The calculation methods of TFP are mainly divided into parametric methods and non-parametric methods. At present, the mainstream method is data envelopment analysis (DEA), which is one of the non-parametric methods. It has the advantage of allowing the existence of inefficient behavior and decomposing the changes in TFP [24]. Rolf [25], Coelli [26], and Goyo [27] used DEA and SFA methods earlier to evaluate the TFP of the domestic power generation sector. Feriol [28] revealed the technological development laws of new energy. Wu [29] used China as an example to calculate the key factors affecting wind power generation by using the DEA model. Wang [11] used the DEA model to evaluate the utilization efficiency of wind energy resources in various provinces in China. Under the constraints of environmental regulations, the output of TFP in the power generation sector consists of two parts: expected and unexpected. In terms of unexpected output, Qu [30] examined the inter-provincial efficiency of China’s thermal power industry by considering carbon dioxide as an unexpected output and thinking that the power policy measures affect the efficiency value of the thermal power industry. In addition, there is also literature that treats solid waste and wastewater as unexpected outputs [31]. In the selection of the DEA method, the SBM model can effectively solve the problem of input–output redundancy in efficiency estimation, which is more advantageous than the CCR model and BCC model [32].

Reviewing the above research, it can be found that the existing literature has the following shortcomings: firstly, when analyzing the impact of wind power price policy on the wind power industry, the existing literature mainly selects conventional indicators such as wind power installed capacity and wind power generation for the output of the wind power industry [5,11,15]. These indicators can only reflect the increase in the overall scale of the industry but cannot accurately reflect the improvement of the internal quality and efficiency of the industry. With the change in China’s economic growth from quantity to quality, we need to improve the quality and efficiency of production factors in the new energy industry to achieve industrial economic growth. Secondly, there is not much research on the impact of wind power price policy on the internal mechanism of the industry in the existing literature. As an important policy tool to ensure the sustainable development of the wind power industry, the benchmark electricity price mechanism has profound practical significance in terms of its inherent impact on industrial economic growth, as well as the policy requirements for achieving full grid parity. Finally, the existing literature rarely considers the “wind curtailment rate” as a constraint. This is obviously incomplete to analyze the TFP of wind power without considering the “negative output”. At the same time, it lacks further decomposition of TFP, especially changes in technological progress, which is of great significance for studying the impact of wind power price policy on the high-quality development of the wind power industry.

In response to the above issues, this article uses the SBM model containing unexpected outputs and the global ML index model to measure the changes and decomposition changes in TFP, explore the impact mechanism and heterogeneity of wind power price policy on TFP, and provide a reference for the sustainable development of the wind power industry and the formulation of electricity price policy.

3. Research Hypotheses

3.1. Relationship between Wind Power Price Policy and TFP of Wind Power

The existence of externalities makes the market unable to achieve Pareto optimality, and the fixed electricity price policy can correct the external effects of the wind power industry [33]. At this time, the government should actively participate in economic activities to improve market efficiency and reduce market failure. The R&D investment of new energy enterprises has the characteristics of public goods. Due to the technology spillover caused by positive externality, other enterprises are free-riding, and the R&D investment of new energy enterprises is lower than the socially optimal level [34]. The fixed feed-in tariff policy can alleviate the financing constraints of wind power enterprises and encourage them to invest in R&D [35].

From the perspective of wind power pricing policy, on the one hand, by introducing a price mechanism, the cost of wind power generation will be allocated to power products, reducing the power generation cost of wind power enterprises and promoting their large-scale development. On the other hand, the price policy improves the efficiency of enterprises’ capital investment and encourages more enterprises to enter the market, thereby increasing supply. In the long run, competition among wind power enterprises promotes continuous technological transformation and management innovation [34,35,36], reduces enterprise costs, and promotes the healthy development of wind power enterprises.

From the perspective of the input indicators of TFP, the reduction of benchmark electricity price will reduce the increase in installed capacity, guide enterprises to optimize resource allocation, enhance competitiveness, and avoid the trend of blind investment and expansion in the industry [20]. At the same time, it will reduce the “wind curtailment rate” in wind energy resource-rich areas and avoid causing industry fluctuations such as the prominence of “wind curtailment and power rationing” in 2012.

From the perspective of the output indicators of TFP, the reduction of the benchmark electricity price is conducive to the increase of wind power grid capacity, improving wind power generation efficiency, reducing wind power generation costs, promoting industrial wind power technology progress, and reducing the increase in unexpected output wind curtailment rate. Therefore, the electricity price policy is conducive to the improvement of TFP, and the reduction of benchmark electricity price is negatively correlated with TFP.

Based on the above analysis, the hypotheses are put forward as follows:

H1. 

China’s wind power price policy has a significant driving effect on the TFP of wind power.

H2. 

The wind power price policy helps to promote technological progress or management innovation in enterprises, thereby promoting the improvement of TFP.

3.2. Can the Policy Requirements of Wind Power Full Grid Parity Be Realized?

From the mature development history of the wind power industry in developed countries, most countries have gradually lowered or abolished benchmark electricity prices for wind power, promoting the implementation of competitive electricity price mechanisms [3]. With the development of the industry, it is an inevitable trend for the benchmark electricity price of wind power to be lowered to full grid parity [37]. In the proportion of energy power generation, coal power has always maintained a dominant position in the proportion of power generation in China. With the increase in the proportion of new energy power generation, coal power has begun to transform into a regulated power supply, but the role of coal power is still prominent at this stage [38]. The full grid parity of wind power is mainly aligned with coal power. By analyzing the price difference between the benchmark electricity price of wind power and coal power over the years, it is found that wind power’s benchmark price in various areas is significantly higher than coal power’s benchmark electricity price, which is 1.1–1.9 times. This price difference has effectively promoted the development of the wind power industry [5,30]. With the gradual narrowing of the price difference, the cost reduction and technological innovation of the wind power industry will be forced to improve the TFP, achieve healthy development of the industry, and meet the policy requirements of full grid parity.

H3. 

The wind power price policy reduces the price difference between the benchmark electricity price of wind power and coal power, then improves the TFP of wind power.

4. Methods and Data

4.1. Model Setting

Based on the above analysis, this article uses panel data of 30 provinces (excluding Tibet, Hong Kong, Macao, and Taiwan) in China from 2015 to 2019 for empirical analysis to investigate the relationship between wind power price policy and TFP of wind power. The benchmark regression model is [9]:

$$\ln GTF{P}_{it}={\beta }_0+{\beta }_{1}lnwpp{p}_{it}+\gamma X_{it}+{\delta }_i+{\lambda }_t+{\epsilon }_{it}$$

(1)

where $GTF{P}_{it}$ represents the TFP of province i in year t; $wpp{p}_{it}$ is the benchmark electricity price in province i in year t, and the impact of the change in benchmark electricity price on TFP depends on the sign and size of the coefficient value ${\beta }_1$; ${\beta }_0$ is the constant term, ${\delta }_i$ and ${\lambda }_t$ are the fixed effects at the province level and year level, ${\epsilon }_{it}$ is a residual term; $X_{it}$ is the control variable at the province level. This article selects five indicators as the control variables: the amount of wind energy resources, the proportion of coal consumption to total energy, electricity investment, carbon emissions, and urbanization rate.

4.2. Variable Selection

(1)

Explanatory variable: Global Total Factor Productivity ($GTF{P}_{it}$)

In economics, the commonly used method to measure efficiency is the production frontier analysis. According to the specific form of whether the production function is known, it can be divided into parametric method and non-parametric method. In the field of energy economics, the non-parametric method is mainly used to estimate TFP, of which the DEA with multiple inputs and outputs is the most commonly used [39,40]. However, the conventional DEA model has radial and slack problems, which may lead to an overestimation of efficiency. To solve this problem, Tone [41] proposed an SBM-DEA model which considers the radial and slack problems.

In the previous literature on efficiency evaluation of the wind power industry, unexpected outputs in the production process were often not considered. Referring to the method of Tone [41], this article intends to use the SBM-DEA model, including unexpected output and the global ML index model [42], to analyze the change of wind power TFP from both static and dynamic perspectives. The specific calculation methods are as follows:

Assuming there are n decision-making units, and each with m input factors $X^a=\left(x_1,\cdots ,x_n\right)\in R^{m\times n}$, There are $s_1$ kinds of expected outputs $Y^a=\left(y_1^a,\cdots ,y_n^a\right)\in R^{s_1\times n}$, and $s_2$ kinds of unexpected outputs $Y^b=\left(y_1^b,\cdots ,y_n^b\right)\in R^{s_2\times n}$, where $\mathrm{X}>0$, $Y^a>0$, $Y^b>0$. The set of environmental production technologies is

$$\mathrm{P}=\left\{\left(x,y^a,y^b\right)|x\gg X\lambda ,y^a\ll Y^a\lambda ,y^b\gg Y^b\lambda ,\lambda \gg 0\right\}$$

(2)

The global SBM model considering unexpected output is

$$p^{*}=min\frac{1-\frac{1}{m}{{\displaystyle \sum }}_{i=1}^m\frac{S_i^{-}}{x_{i0}}}{1+\frac{1}{s_1+s_2}\left({{\displaystyle \sum }}_{r=1}^{s_1}\frac{S_r^a}{y_{r0}^a}+{{\displaystyle \sum }}_{r=1}^{s_2}\frac{S_r^b}{y_{r0}^b}\right)}$$

(3)

$$\begin{gathered} \mathrm{S}.\mathrm{T}.x_0=X\lambda +S^{-},y_0^a=Y^a\lambda -S^a,y_0^b=Y^b\lambda +S^b \\ \lambda ,S^{-},S^a,S^b\gg 0 \end{gathered}$$

In Equation (3), $p^{*}$ represents the target efficiency value; λ is the weight vector; the subscript “0” of $x_{i0},y_{i0}^a,y_{i0}^b$ represent the evaluated decision-making units; and $S^{-},S^a,S^b$, respectively, represent the slack vectors of input, expected output, and unexpected output.

In this article, the global technology efficiency index (GEC) and global technology progress index (GTC) are solved by using the global Malmquist–Luenberger (GML) index proposed by Oh [43].

$$\begin{array}{c}GM{L}^{t,t+1}(x^t,y^t,b^t,x^{t+1},y^{t+1},b^{t+1})\hfill \\ =\frac{E^{G,t+1}(x^{t+1},y^{t+1},b^{t+1})}{E^{G,t}(x^t,y^t,b^t)}=GE{C}^{t,t+1}\ast GT{C}^{t,t+1}\hfill \\ =\frac{E^{t+1}(x^{t+1},y^{t+1},b^{t+1})}{E^t(x^t,y^t,b^t)}\frac{E^{G,t+1}(x^{t+1},y^{t+1},b^{t+1})/E^{t+1}(x^{t+1},y^{t+1},b^{t+1})}{E^{G,t}(x^t,y^t,b^t)/E^t(x^t,y^t,b^t)}\hfill \end{array}$$

(4)

The input–output indexes of the SBM-DEA model are shown in

Table 1. Description of input–output variables.

Index	Variable	Variable Description
Input indexes	Capital input	Wind power installed capacity (10,000 kilowatts)
	Resource input	The area with an average wind power density of 150 or more in the land 70 m height layer (10,000 square kilometers)
Expected output	Wind power generation	Wind power generation (100 million kWh)
	Wind power utilization hours	Wind power utilization hours (hours)
Unexpected output	Curtailed wind power	Cutailed wind power (100 million kWh)

.

The input indexes are ① Capital input: this article selects the installed capacity of wind power provinces in China as the proxy variable for capital input [11], and ② Resource input: this article selects the area with an average wind power density of 150 or more in the land 70 m height layer of each province of wind power in China as the proxy variable for resource input.

This article selects the wind power generation and wind power utilization hours of each province in China as the expected output variables [11,15,16].

This article selects curtailed wind power of each province in China as the unexpected output variable.

(2)

Core explanatory variables: Wind power price policy.

Due to the uncertainty of wind power generation and the imperfection of energy storage require the policy management of electricity prices, and the development of wind power requires the government to formulate benchmark electricity prices to maintain the electricity price in a reasonable range, this article selects the benchmark electricity price of wind power in each province announced by the China Development and Reform Commission from 2015 to 2019 as the proxy variable of wind power price policy [20].

(3)

Control variables:

① the amount of wind energy resources; select the area with an average wind power density of 150 or more in the land 70 m height layer of each province of wind power in China as the proxy variable of wind energy resources. ② Coal consumption ratio: the proportion of coal consumption to the total energy. ③ Electricity investment: this article uses investment in electricity, steam, hot water production, and supply industries as proxy variables of electricity investment. ④ Carbon emissions: it is calculated by selecting eight kinds of fossil energy, such as coal, coke, crude oil, gasoline, kerosene, diesel, fuel oil, and natural gas, in each province in China. ⑤ Urbanization rate: the ratio of urban population to the permanent year-end population in each province of China, which comprehensively reflects the level of regional economic development and management [5,15,23,44].

4.3. Data Sources

In 2009, the National Development and Reform Commission divided the country into four types of resource areas according to the status of wind energy resources and engineering construction conditions and formulated the corresponding benchmark feed-in tariffs. Therefore, this article uses provincial panel data with reference to the benchmark electricity price of each province. Due to the data acquisition of Hong Kong, Macao, Taiwan, and Tibet, this article excludes the places and selects 30 provinces in China as the research objects, with a sample interval from 2015 to 2019. The specific data sources are as follows:

(1)

The explained variable is calculated by the efficiency values obtained from the SBM-DEA model, which is the TFP of each province.

(2)

The core explanatory variable is the wind power price policy, which is sourced from the benchmark electricity price of wind power in each province announced by the China Development and Reform Commission from 2015 to 2019.

(3)

In the control variables, the amount of wind energy resources comes from the “Annual Bulletin of China’s Wind Energy and Solar Energy Resources” over the years. The coal consumption ratio, electricity investment, and carbon emissions are obtained from the “China Energy Statistical Yearbook”. The urbanization rate comes from the “China Statistical Yearbook” and “China City Statistical Yearbook” [5,11,15,23]. At the same time, logarithmic processing was performed on the above variables to reduce the impact of heteroscedasticity. The descriptive statistics of variables are shown in

Table 2. Descriptive statistics of variables.

Variable	Number of Observations	Mean Value	Variance	Minimum Value	Maximum Value
Global Total Factor Productivity (GTFP)	150	1.059	0.196	0.374	1.823
Global Pure Technical Efficiency Change (GPEC)	150	1.034	0.268	0.392	2.281
Global Technological Progress Change (GTC)	150	1.104	0.303	0.374	2.018
Wind power price policy (yuan/kwh)	150	0.561	0.053	0.365	0.610
Difference in benchmark electricity price (yuan/kwh)	150	0.203	0.060	0.054	0.336
The amount of wind energy resources (10,000 km2)	150	16.791	27.856	0.300	133.4
Coal consumption ratio (%)	150	0.918	0.491	0.025	2.461
Electricity investment (billion yuan)	150	627.785	448.077	82	2435
Carbon emission (10,000 tons)	150	59,182.75	36,574.38	14,126.88	177,489
Urbanization rate (%)	150	59.807	11.067	42.010	88.300

.

5. Empirical Analysis

5.1. Calculation and Decomposition of Inter-Provincial Total Factor Productivity

Referring to the research of Zhang [45], this article uses the decomposition method of RD [40] to decompose the Malmquist index and further decomposes productivity growth into global technological progress change (GTC), global pure technical efficiency change (GPEC) and global scale efficiency change (GSEC). The calculation results of the global total factor productivity (GTFP) of wind power in 30 provinces, cities, and autonomous regions of China from 2015 to 2019 are shown in

Table 3. 2015–2019 GTFP of wind power and its decomposition in China’s provinces (Geometric mean).

Area	GTFP	GPEC	GTC	Area	GTFP	GPEC	GTC
Beijing	0.696	1.000	0.727	Henan	0.819	0.760	1.082
Tianjin	0.708	0.916	0.788	Hubei	0.952	0.922	1.013
Hebei	1.170	1.000	1.337	Hunan	1.559	1.560	0.949
Shanxi	1.255	1.092	1.282	Guangdong	1.171	1.189	0.982
Inner Mongolia	1.180	1.000	1.456	Guangxi	0.905	1.153	0.864
Liaoning	1.258	1.008	1.331	Hainan	1.009	1.146	0.848
Jilin	1.240	1.107	1.116	Chongqing	0.741	0.915	0.824
Heilongjiang	1.223	0.991	1.224	Sichuan	0.433	0.680	0.724
Shanghai	1.219	1.000	1.207	Guizhou	1.816	1.901	0.943
Jiangsu	0.992	0.960	1.031	Yunnan	1.603	1.000	1.659
Zhejiang	1.161	0.755	1.539	Shaanxi	1.036	1.059	0.971
Anhui	0.974	0.903	1.037	Gansu	1.298	1.103	1.227
Fujian	1.006	1.000	1.006	Qinghai	0.963	0.975	0.979
Jiangxi	1.065	1.075	0.967	Ningxia	1.104	0.976	1.163
Shandong	1.233	0.957	1.459	Xinjiang	1.072	0.915	1.374
East	1.057	0.994	1.114	Central	1.136	1.051	1.084
West	1.105	1.062	1.108	All	1.095	1.034	1.104

.

As can be seen from Table 3 and

Table 4. 2015–2019 GTFP of wind power and its decomposition in China (Geometric mean).

Year	GTFP	GPEC	GTC
2015	0.994	0.988	1.020
2016	0.950	1.055	0.937
2017	1.072	0.998	1.113
2018	1.190	1.045	1.192
2019	1.272	1.083	1.257

, the TFP of wind power in each province in China showed an overall upward trend from 2015 to 2019, taking into account unexpected output. The geometric mean value of GTFP and its decomposition over the years is greater than 1, showing positive growth, especially the rapid growth of technological progress, with an average annual growth rate of 10.4%, which continues to promote the growth rate of wind power in all factors. It is speculated that there are two main reasons for this phenomenon. On the one hand, after the industrial fluctuation in the wind power industry in 2012, the government timely guided the development direction of the industry. At the same time, with the reform of power policy and the development of new energy power, the wind power industry ushered in a period of rapid development. On the other hand, the benchmark price policy for wind power plays a significant role. With the introduction of the price mechanism, the reduction of the benchmark electricity price encourages enterprises to reduce costs through technological change and management innovation, which is conducive to improving the efficiency of wind power generation and reducing the increase in unexpected output wind curtailment rate.

From a regional perspective, the TFP of wind power in China is increasing in the east, middle, and west, but there are also regional disparities. The growth rate in the middle area is higher than western area and higher than eastern area. The geometric mean values of technical efficiency are 0.994, 1.051, and 1.062, respectively, in the east, middle, and west areas, which are better in the west and east, and poorer in the east. The geometric average values of technological progress are 1.114, 1.084, and 1.108, respectively, in the east, middle, and west areas, which is better in the east, followed by the west and middle.

The main reasons for this phenomenon are related to the layout of the wind power industry, the endowment of wind energy resources, and the progress of wind power technology: The eastern area is economically developed, but the wind energy resources are poor, relying on thermal power generation for a long time, with less installed wind power capacity and generation. The middle and western areas are rich in wind energy resources and have a large industrial layout. At the same time, with the progress of wind power technology and the significant reduction of wind curtailment rate, technological efficiency has been improved, which makes the TFP of wind power in these areas continuously increase.

5.2. Benchmark Regression Analysis

The benchmark regression results are shown in

Table 5. Benchmark regression results.

	Mixed Regression lnGTFP	Fixed Effect lnGTFP	Bidirectional Fixed Effects lnGTFP
lnwppp	−0.986 *** (−3.91)	−0.499 ** (−2.14)	−0.952 *** (−2.55)
lnwer	−0.012 (−0.52)	−0.129 *** (−3.59)	−0.173 *** (−3.83)
lnccr	0.196 *** (4.42)	0.159 ** (2.08)	0.189 ** (2.39)
lnepi	−0.119 ** (−2.17)	−0.195 *** (−3.93)	−0.176 *** (−3.51)
lnce	0.035 (0.54)	0.096 *** (0.87)	0.062 (0.57)
lnur	−0.829 (−0.40)	2.333 *** (4.01)	1.294 * (1.71)
Constant term	0.489 * (0.56)	−8.625 *** (−3.80)	−4.597 * (−1.47)
Fixed time effect	No	No	Control
Fixed provincial effect	No	Control	Control
N	150	150	150
$\mathrm{adj}.{\mathrm{R}}^2$	0.258	0.576	0.613

Note: t-statistic in parentheses. *, **, and *** represent significance at 10%, 5%, and 1% levels, respectively.

.

In Table 5, column 1 shows the regression results of mixed OLS, column 2 shows the regression results of controlling for fixed provincial effects, and column 3 shows the regression results of controlling for fixed time and provincial effects. The core explanatory variable of wind power benchmark electricity price in the three benchmark regression models is negative and significant at the 1% level, indicating that hypothesis H1 holds. The results show that the reduction of benchmark electricity price is helpful in improving the TFP of wind power. For every 1% reduction in benchmark electricity price, the TFP of wind power will increase by 0.95%.

The regression results of the control variables: ① there is a significant negative correlation between the amount of wind energy resources and the TFP of wind power. This is mainly due to the relatively large subsidies in areas with abundant wind energy resources, which bring certain rent-seeking opportunities and non-economic barriers, leading to increased costs and abandoned air volume and resulting in inefficient development. ② The proportion of coal consumption in total energy has a significant positive impact on the TFP of wind power, indicating that coal and wind power are complementary rather than competitive. ③ There is a significant negative correlation between electricity investment and TFP of wind power, indicating that wind power has the optimal investment scale effect in energy investment. ④ The improvement of the urbanization rate will significantly increase the TFP of wind power, indicating that the improvement of regional economic development and management level will enhance the development of the wind power industry.

5.3. Endogeneity Test

This article adopts the following methods for testing to address the potential endogeneity issues caused by bidirectional causality and missing variables in the model:

(1)

The current total factor productivity is used as the explained variable, and the benchmark electricity price and control variables lagging behind one period are used for regression analysis [9].

(2)

The instrumental variable method is adopted to solve the endogeneity problem. Based on the following two considerations, this article selects the benchmark electricity price of thermal power as the instrumental variable: ① Thermal power accounts for a major proportion of all kinds of energy grid capacity, and the benchmark electricity price of new energy is based on the thermal power, which shows that there is a good correlation between the benchmark electricity price of thermal power and wind power [23,46]. ② The benchmark electricity price of thermal power is only related to the coal resources in the province, and the coal resources are exogenous to the model in the article.

The regression results of the endogeneity test are shown in

Table 6. Endogeneity test results.

	Lagging Explanatory Variable	1-Stage Regression	2-Stage Regression
lntpbt		0.595 *** (25.74)
lnwppp	0.043 *** (0.71)		−1.395 *** (−5.2)
Constant term	1.326 * (0.39)	−0.408 *** (−3.36)	0.304 * (0.36)
Control variable	Control	Control	Control
1-stage regression F value		149.51
N	120	150	150
$\mathrm{adj}.{\mathrm{R}}^2$	0.485	0.863	0.338

Note: t-statistic is in parentheses. * and *** represent significance at 10% and 1% levels, respectively.

.

The regression results: (1) as shown in column 1 of Table 6, the estimated coefficient of wind power benchmark electricity price is 0.043, which is significant at the 1% level, indicating the robustness of the benchmark regression results. (2) Column 2 in Table 6 is 1-stage using the two-stage least squares method. The regression results show that the thermal power benchmark electricity price coefficient is 0.595, which is significant at the 1% level, indicating a positive correlation between the benchmark electricity price for thermal power and wind power. The 2-stage regression results show that the estimated coefficient of wind power benchmark electricity price is −1.395, indicating that the reduction of benchmark electricity price helps to improve the GTFP of wind power, which is consistent with the benchmark regression results.

5.4. Robustness Test

For the benchmark regression model, this article mainly conducts the following two robustness tests:

① Replacing the explained variable: Due to differences in input variables and models, the final conclusion of TFP is also inconsistent. To avoid the aforementioned impact, this article uses the DDF-GML [46] model to recalculate the input–output indicators and then re-estimates after obtaining the explained variable. The regression results are shown in column 1 in

Table 7. Robustness test results.

	Replacing Explained Variable	System GMM
L.lnGTFP		0.611 *** (5.82)
lnwppp	−0.867 ** (−2.45)	−0.520 *** (−2.57)
Constant term	0.058 * (0.02)	0.989 * (1.00)
Control variable	Control	Control
Fixed time effect	Control	Control
Fixed province effect	Control	Control
AR(1) test		−2.68 [0.007]
AR(2) test		−0.35 [0.609]
Hansen test		26.03 [0.165]
N	150	120
adj.R2	0.493

Note: *, **, and *** represent significance at 10%, 5%, and 1% levels, respectively. t-statistic is in parentheses. The p-values of the corresponding statistics are in the square brackets.

: It can be found that the negative correlation between the reduction of benchmark electricity price and total factor productivity of wind power is still stable.

② Replacing the estimation method: considering that TFP is not only affected by current factors but also related to past factors, a lag period of TFP is added to the benchmark regression model to construct a dynamic panel regression model:

$$\ln GTF{P}_{it}=L.lnGTF{P}_{it}+{\beta }_0+{\beta }_{1}lnwpp{p}_{it}+\gamma X_{it}+{\delta }_i+{\lambda }_t+{\epsilon }_{it}$$

(5)

Considering that system GMM estimation is less prone to generating finite sample bias and can improve estimation efficiency compared to the differential GMM estimation method, this article uses system GMM estimation to regress Equation (5). It can be seen that AR(1) adjoint probability is less than 0.1, AR(2) adjoint probability is greater than 0.1, and Hansen statistics adjoint probability is greater than 0.1 and less than 0.25, which shows that the selection of instrumental variables and lag period is reliable.

As shown in column 2 of Table 7, the estimated coefficient of wind power benchmark electricity price is still negative and significant at the level of 1%, further illustrating the robustness of the benchmark regression model.

6. Mechanism Analysis and Heterogeneity Test

6.1. Technology Effect

To verify hypothesis H2, this article constructs a dynamic panel regression model (6) based on the benchmark regression model (1), replacing the explained variable with the GTC of wind power.

$$\ln GT{C}_{it}={\beta }_0+{\beta }_{1}lnwpp{p}_{it}+\gamma X_{it}+{\delta }_i+{\lambda }_t+{\epsilon }_{it}$$

(6)

The regression results are shown in column 1 of

Table 8. Regression results of technical effect and policy effect of benchmark electricity price.

	Technical Effect	Policy Effect
lnwppp	−0.957 ** (−2.25)
lndbep		−0.184 * (−1.77)
Constant term	−0.376 * (−0.11)	−4.762 * (−1.50)
Control variable	Control	Control
Fixed time effect	Control	Control
Fixed provincial effect	Control	Control
N	150	150
$\mathrm{adj}.{\mathrm{R}}^2$	0.490	0.601

Note: * and ** represent significance at 10% and 5% levels, respectively.

; there is a significant negative correlation between the reduction of benchmark electricity prices and wind power GTC, indicating that hypothesis H2 is established.

The following conclusions can be drawn: through the price mechanism, the wind power price policy encourages enterprises to reduce costs through technological change and management innovation, guides enterprises to optimize resource allocation, enhance competitiveness, avoid the trend of blind investment and expansion in the industry, and reduce the “wind curtailment rate”, thereby achieving the goal of improving wind power TFP.

6.2. Policy Effect

To verify hypothesis H3, this article constructs a dynamic panel regression model (7) based on the benchmark regression model (1), replacing the core explanatory variable with the price difference between the benchmark electricity price of wind power and thermal power.

$$\ln GTF{P}_{it}={\beta }_0+{\beta }_{1}lndbe{p}_{it}+\gamma X_{it}+{\delta }_i+{\lambda }_t+{\epsilon }_{it}$$

(7)

The regression results are shown in column 2 of Table 8: The price difference between the benchmark electricity price of wind power and thermal power has a significant negative correlation with wind power TFP, indicating that hypothesis H3 is established.

The following conclusions can be drawn: the full grid parity of wind power is mainly aligned with thermal power. The price difference between the benchmark electricity price of wind power and thermal power forces the cost reduction and technological innovation of the wind power industry, which will help promote the implementation of the wind power competitive electricity price mechanism and meet the policy requirements of wind power full grid parity.

6.3. Heterogeneity Test

(1)

Grouping by wind energy resource endowment

Based on the benchmark regression model (1) and wind energy resource endowment [47], this article uses the average proportion of the amount of wind energy resources in each province from 2015 to 2019 for regression analysis (Provinces abundant in wind energy resources: Xinjiang, Inner Mongolia, Qinghai, Heilongjiang, Gansu, Jilin, Ningxia, Liaoning, Hebei, Yunnan. Provinces with relatively abundant wind energy resources: Shandong, Guangxi, Shanxi, Guangdong, Shaanxi, Henan, Anhui, Jiangsu, Hubei, and Jiangxi. Provinces with poor wind energy resources: Hunan, Guizhou, Sichuan, Zhejiang, Fujian, Hainan, Chongqing, Tianjin, Shanghai, and Beijing) to test regional heterogeneity by dividing the 30 provinces in China into three groups. The regression results are shown in

Table 9. Heterogeneity test results 1.

	Abundant	Relatively Abundant	Poor
lnwppp	−1.036 ** (−2.29)	−1.489 * (−1.85)	−1.650 (−0.96)
Constant term	6.027 * (1.10)	−2.679 (−0.46)	−12.86 * (−1.92)
Control variable	Control	Control	Control
Fixed time effect	Control	Control	Control
Fixed province effect	Control	Control	Control
N	50	50	50
$\mathrm{adj}.{\mathrm{R}}^2$	0.835	0.722	0.495

Note: * and ** represent significance at 10% and 5% levels, respectively.

.

The results show that wind power price policy has the most significant impact on wind power TFP in areas with abundant and relatively abundant wind energy resources. This is because the wind power industry in areas with abundant wind energy resources has a large scale and a long period of development, and wind power generation accounts for a relatively high proportion of all energy generation, making the areas most sensitive to fluctuations in wind power benchmark electricity price. On the contrary, in areas with poor wind energy resources, the proportion of other energy generation is generally high, and the scale of the wind power industry is small. Therefore, TFP is the least sensitive to fluctuations in wind power benchmark price.

(2)

Grouping by wind power capital investment

Based on the benchmark regression model (1) and wind power capital investment [18], this article uses the average proportion of wind power installed capacity in each province from 2015 to 2019 for regression analysis to test regional heterogeneity (Provinces with many wind power capital investment: Inner Mongolia, Xinjiang, Hebei, Gansu, Shandong, Ningxia, Shanxi, Yunnan, Liaoning, and Jiangsu. Provinces with more capital investment in wind power: Heilongjiang, Jilin, Guizhou, Shaanxi, Henan, Guangdong, Hunan, Hubei, Fujian, and Anhui. Provinces with few capital investments in wind power: Qinghai, Sichuan, Jiangxi, Guangxi, Zhejiang, Shanghai, Tianjin, Chongqing, Hainan, and Beijing) by dividing the 30 provinces in China into three groups. The regression results are shown in

Table 10. Heterogeneity test results 2.

	Many	More	Few
lnwppp	−0.920 ** (−2.63)	−6.462 ** (−2.51)	0.347 (0.30)
Constant term	1.053 (0.22)	−11.827 * (−1.81)	−7.353 (−1.01)
Control variable	Control	Control	Control
Fixed time effect	Control	Control	Control
Fixed province effect	Control	Control	Control
N	50	50	50
$\mathrm{adj}.{\mathrm{R}}^2$	0.901	0.709	0.346

Note: * and ** represent significance at 10% and 5% levels, respectively.

.

The results show that provinces with many and more capital investments have a higher proportion of wind power generation compared to other energy sources. The wind power industry has a large scale and mature technology, and the wind power price policy is conducive to stimulating the adjustment of the industry. Therefore, the TFP of these provinces has the highest sensitivity to the reduction of benchmark electricity prices. On the contrary, provinces with few installed capacities have a lower proportion of the wind power industry and dependence on wind power generation and the lowest sensitivity.

7. Conclusions and Policy Recommendations

7.1. Conclusions

Since the implementation of the fixed feed-in tariff mechanism, the wind power industry in China has developed rapidly, but it also faces a series of high-quality development problems [3,7,8]. Has wind power become a mature and cost-effective source of electricity? Further verification is urgently needed. Although there are numerous literature reviews on the effectiveness of China’s wind power price policy and the TFP of wind power [4,5,13,14,15,16], there is little empirical research on the relationship between them.

Therefore, this article has constructed an SBM-GML global reference model and calculated the TFP of wind power in 30 provinces of China from 2015 to 2019. The results show that the TFP of wind power in China decreased first and then increased, showing a positive growth overall, and there are also regional differences. Then, this article empirically tested the impact of the wind power price policy on the provincial TFP of wind power in China by using provincial panel data and further analyzed the technical and policy effect of the wind power price policy. It is useful for countries that are considering adopting a fixed feed-in tariff mechanism.

The results of this article show that the wind power price policy can significantly improve the TFP of wind power, and the results of the endogenous test and robustness test show the effectiveness of the model, which is consistent with the hypothesis.

In addition, in the mechanism analysis, the technical effect estimation results show that wind power price policy has a significant positive impact on enterprise technological change and technological progress is the main driving force affecting wind power TFP. Price policy improves TFP by promoting technological progress. The policy effect estimation results show that the price difference between wind power and thermal power of benchmark electricity price has a significant negative impact on the TFP of wind power, which is conducive to full grid parity on the supply side of wind power. A good policy design is crucial for promoting the marketization of wind power.

Finally, we explored the regional heterogeneity impact of wind power price policy on wind power TFP. The results showed that provinces with abundant wind energy resources and much wind power capital investment are more sensitive, but this may also put the provinces in an advantageous position in a competition or even lead to monopolies, which is not conducive to the balanced development of the wind power industry. This is also a direction for future research.

From the supply side in the future [48], the wind power price policy also includes the electricity price mechanism of the renewable energy quota system, but it is usually applied to the mature stage of the wind power market. Future research on wind power price policy will also need to include the system to explore the impact of market pricing on the wind power TFP. From the demand side, the floating price mechanism and the green price mechanism will have an important impact on consumer purchases of green electricity in the future. The impact of the price transmission mechanism on wind power TFP may be an interesting topic for future research.

7.2. Policy Recommendations

(1)

Gradually implementing the full grid parity policy for wind power and establishing a market-oriented electricity price mechanism. Although the benchmark electricity price policy for wind power has accelerated the development of the wind power industry, a long-term subsidy will reduce the enterprise’s willingness for technological change and management innovation. Only when the market-oriented electricity price plays the role of resource allocation for electricity price can it continuously promote the upgrading of industrial structure and promote the return of electricity to commodity attributes so as to achieve the purpose of on-demand pricing.

(2)

Accelerate the technological innovation of wind power and continuously improve the TFP of wind power. It can be seen that technological progress is the main driving force in promoting the TFP of wind power. There is still enormous technological potential in the wind power industry chain, including technological breakthroughs such as large-scale and intelligent development of wind turbines, high-power gearboxes, and hundred-meter blades, and technological innovation of high-performance alternative materials, which require continuous research by the industry. Every technological innovation will bring cost reduction and efficiency improvement to the wind power industry.

(3)

Based on the development conditions of different areas, the wind power development model should be tailored to local conditions to promote scale expansion in areas with abundant wind energy resources. For example, the “Three North” areas (north, northeast, and northwest China), with abundant wind energy resources and low electricity consumption cost, are suitable for large-scale centralized development, and the industries with demand for clean electricity can be attracted to the areas. On the contrary, the provinces with poor wind energy resources can boost rural revitalization through local development and utilization.