Investigating the Determinants of the Growth of the New Energy Industry: Using Quantile Regression Approach.

• Author: Xu, Bin

1. INTRODUCTION

   China is now the largest coal producer and consumer all over the world (Davidson and Perez-Arriaga, 2020). Statistics show that in 2019, China's coal production and consumption are 3.85 billion tons and 4.86 billion tons, respectively. The use of a large amount of coal will inevitably lead to a rapid increase in C[O.sub.2] emissions (Coglianese et al., 2020). Since 2006, China has become the world's largest C[O.sub.2] emitter.

   Also, sustained economic growth has promoted the increase of residents' income, and more residents are buying cars. China is currently the world's largest automobile producer and consumer (Sueyoshi et al., 2021). Large-scale use of motor vehicles must consume lots of fossil fuel. China is an oil-poor country, and its domestic oil production cannot meet the market demand. This has prompted China to continuously expand oil importation, causing a rapid increase in foreign oil dependence (Kyritsis and Serletis, 2019). Statistics show that China's foreign oil dependence rose to 70.8% in 2019.

   China's level of C[O.sub.2] emissions has made it the world's focus and the excessive dependence on foreign oil does not only exposes China to huge energy security risks but also makes China's economy vulnerable to the fluctuations in international oil prices. New energy resources mainly include solar, wind, biomass, tidal, hydrogen and nuclear energies. Expanding the supply of new energy resources can help reduce C[O.sub.2] emissions and environmental pollution. According to the definition of the United Nations Conference on New and Renewable Energy (1980), the new energy industry is a series of processes performed by enterprises and units that develop and produce new energy. China has a vast territory, and there are significant differences between provinces in terms of economic growth, industrial level, population size, energy endowment, and technological talents (Zhang et al., 2019). This has led to significant regional differences in the new energy industry.

   The important role of the new energy industry in China has prompted a good number of researches. Analyzing existing related researches, it is found that there are two shortcomings; (1) Most literature assumes that economic variables obey normal distributions and use traditional mean models (e.g. panel data model and vector autoregressive model) for regression estimations (Cai and Menegaki, 2019; Li et al., 2020). The changes in the economic phenomena are complex, which leads to the distribution of economic variables not being normal. Moreover, the extreme values (i.e., maximum and minimum values) often imply important information, and the mean models cannot estimate their important effects. The scales of new energy in some provinces such as Beijing, Shanghai, and Tianjin, are too small. Other provinces such as Yunnan, Hubei, and Sichuan provinces, are very large. The impact of the same factor on the new energy industry in different provinces is different. Hence, the traditional mean model cannot estimate this heterogeneous impact. This heterogeneous impact is essential for local governments to formulate effective policies for new energy development. One of the main motivations of this article is to estimate the heterogeneous impact of influencing factors in different provinces. (2) Assuming that economic sequences satisfy the classical econometric hypotheses (e.g., normal distribution and non-autocorrelation), most existing studies use the ordinary least squares (OLS) method to estimate parameters. These econometric assumptions will not be satisfied. In this case, the ordinary least squares estimates are often biased and inconsistent. Another motivation of this paper is to obtain a robust estimation result which requires a more suitable parameter estimation method.

   This article improves on the following two aspects: (1) this paper uses the quantile regression method to examine the new energy industry. The quantile regression can estimate the impacts of the explanatory variables on the dependent variable at different quantiles. The mean models can only give the average effects of the explanatory variables on the explained variable, which is equivalent to giving the effects of the 50th quantiles in the explained variable. (2) This paper uses the bootstrap method to estimate the parameters of the model. The bootstrap method uses repeated sampling techniques to obtain an approximate distribution of the estimated parameters. This is not restricted by the number of original samples. Therefore, the parameter estimates obtained by the bootstrap method are significantly better than the ordinary least squares estimates (OLSE). The main purpose of this article is to comprehensively study the heterogeneous influence of each influencing factor on the new energy industry in different quantiles of the provinces, and analyze the reasons for these heterogeneous effects. Finally, this paper proposes corresponding policy implications, based on the results of the empirical analysis which can provide references for local governments to formulate policies for the development of new energy industries in the region.

   The results of the model estimation are briefly described as follows: (1) economic growth exerts the greatest effect on the new energy industry in the lower 10th quantile provinces. (2) Foreign energy dependence has a minimal impact on the 25th-50th quantile provinces. (3) The contribution of technological progress to the new energy industries in the upper 90th quantile provinces is the lowest. This is because these provinces have the least number of patents granted. (4) The agricultural sector plays a driving role in the 10th-25th, 25th-50th, 50th-75th, 75th-90th, and upper 90th quantile provinces. (5) The impact of the energy consumption structure has a step-down effect from the lower 10th quantiles to the upper 90th quantiles.

2. LITERATURE REVIEW

   Based on the methodology of existing literature, previous studies can be roughly divided into three categories.

   (1) The input-output method. Applying the input-output method, the results of Garrett-Peltier (2017) denoted that in the short term, government procurement boosted the rapid growth of new energy companies. However, households and production enterprises were fundamental in ensuring the long-term sustainable growth of new energy companies. Nakano et al. (2017) employed an input-output method to investigate the new energy industry in Japan. It was concluded that compressing thermal power production and reducing the carbon intensity of the transportation industry had promoted new energy consumption. Nagashima et al. (2017) investigated Japanese wind power employing the input-output analysis and found that market demands and technical improvements drove the rapid growth of the wind power industry.

   (2) System optimization method. Applying the system optimization method, Al-Falahi et al. (2017) investigated the wind and solar energy industries. The results showed that the economic growth model and new energy technologies were important factors affecting the new energy industry. The same method was applied to investigate the new energy in western China. The findings showed that constantly strict environmental regulations had prompted producers and residents to increase new energy consumption. (Ye et al., 2017). Golari et al. (2017) examined the new energy industry using a stochastic optimization method. They found that the continuous depletion of fossil energy and strict environmental regulations promoted the new energy industry.

   (3) Econometric method. Using the panel causality test method, Destek and Aslan (2017) investigated the relationships between new energy resources and economic growth. Their conclusion showed that the links between the two variables were different across the country. Paramati et al. (2017) used the cointegration method to conduct empirical research. The results indicated that there was a two-way causal nexus between economic growth and the new energy industry. The panel error correction model was used by Kahia et al. (2017) to inspect the relationship between the new energy industry and major macroeconomic variables. The results showed that the new energy industry and economic growth were mutually reinforcing. Furthermore, Isik et al. (2018) explored the new energy used using the panel Granger causality test method. The results showed that there was a causal link among the new energy industry, economic development, and tourism development.

   From the perspective of influencing factors, the existing literature points out that the main factors affecting the new energy industry are the following.

   (1) Government policy. Liu and Kokko (2013) showed that the government played an important role in the development of the new energy industry. Many factors affecting the new energy industry were controlled by the government. Furthermore, Sun and Nie (2015) found that feed-in tariff policies were more effective in promoting the development of new energy industries than renewable energy portfolio standard policies. This result was supported by Kuik et al. (2019). Also, government subsidies and tax reductions could help reduce the burden on new energy enterprises and promote their rapid growth (Chang et al., 2020).

   (2) Technological innovation. Costa-Campi et al. (2014) used a structural model to explore new energy Spanish companies and found that the higher the R&D intensity, the more it promoted the new energy industry. Data research on new energy manufacturing companies in China found that technology diversification and R&D cooperation helped new energy production companies to reduce production costs and promote rapid industrial growth (Yun et al., 2019; Chen et al., 2020). However, some literature found that the relationship between technological progress and the new energy industry was not linear, but a positive U-shaped nonlinear relationship (Xu and Lin, 2018).

   ...