Study on the factors of industrial electricity consumption behavior in Hunan Province based on multiple regression analysis

. In recent years, Hunan province has gradually accelerated the adjustment of industrial structure, and has achieved a rapid industrial growth, thereupon the total amount of industrial electricity has increased greatly. There is an internal correlation between the industrial growth and electricity consumption. Therefore, based on this background, this paper established multiple linear regression models to study the influencing factors on the electricity consumption behavior of the three industries with the most electricity consumption in Hunan Province. The regression results show that the iron and steel output, raw coal price index and Mylpic mine price index can significantly affect the electricity consumption of ferrous metal industry. Cement output, cement price and real estate development investment can significantly affect the electricity consumption of non-metal industry. Automobile output, integrated circuit output and per capita disposable monthly income can significantly affect the electricity consumption of transportation and electrical and electronic equipment manufacturing industry in Hunan Province.


Introduction
As we all know, electricity consumption, as the "barometer" of economic development, plays a decisive role for the entire national economy (Shiu et al., 2004 [1] ; Tang et al., 2013 [2] ; Zhang et al., 2017 [3] ; Xia et al., 2020 [4] ). As industry is the pioneer of national economic development, the relationship between industrial electricity consumption and national economy is closer (Yuan et al., 2007 [5] ; Luo et al., 2012 [6] ; Hirsh et al., 2015 [7] ). In recent years, Hunan province has gradually accelerated the adjustment of industrial structure, and has achieved a rapid industrial growth, thereupon the total amount of industrial electricity has increased greatly (Jiang et al., 2012 [8] ; Liu et al., 2020 [9] ).
The industrial electricity consumption accounts for a large proportion of the total social electricity consumption, so its variation has also attracted the attention of scholars. He et al. (2006a [10] , 2006b [11] , 2007 [12] ) used Granger causality test and error correction model to analyze the two-way relationship between electricity consumption and gross domestic product (GDP) growth. Guo (2010) [13] studied the economic electricity relationship of the ferrous metal industry in China. Ge et al. (2016) [14] and Wang et al. (2018) [15] investigated the relationship between Anhui's industrial electricity consumption and value added structure. In conclusion, most of previous studies were based on the characteristics of national electricity use, while less was based on the perspective of provincial electricity demand as the literature on Hunan Province is almost blank. As a complement to this, the paper took Hunan province as the research object and established multiple linear regression models to study the influencing factors on the electricity consumption behavior of the three industries with the most electricity consumption.

The ferrous metal smelting and rolling processing industry
As can be seen from Figure 1, the trend of economic development and electricity consumption of ferrous metal smelting and rolling processing industry was not completely consistent in recent years. Specially, the main products of the ferrous metal industry, steel production, fluctuated greatly from the end of 2011 to the 2012, which may be related to the proposal that the steel industry should focus on solving the problem of excess capacity during the 11th Five-Year Plan period. After that, there can be seen a gradual rose in steel production and a relative stability in electricity consumption.   Figure 2 showed the trend of cement production and electricity consumption of non-metal industries in Hunan Province from 2010 to 2018 was generally similar, but there were some differences. From the monthly data of each year, both of them showed a trend of gradual rise. From the annual data, both cement production and electricity consumption in the non-metal industry showed a turning point in 2015, that is, it did not increase but declined, which was mainly affected by the lack of industrial product market demand at that time, leading to the slowdown of both production and sales growth of the industry products.

The transportation and electronic equipment manufacturing industry
As can be seen from Figure 3, the overall electricity consumption of this industry was on the rise. However, the annual electricity consumption fluctuated greatly, possibly because the industry's electricity consumption had a great relationship with the seasons. As a whole, there was a big drop in February every year, as factories are closed during the Lunar New Year holiday. Also, the electricity consumption in October showed a significant decline, because the industrial load of domestic decreases in October every year and there was a slight increase of electricity consumption in July and August every year due to the hot weather.

Model and empirical results
Stepwise regression model, as one of the multiple regression analysis methods, is to establish the optimal regression model by gradually introducing the research variables. In this paper, a stepwise regression model was established to explore the relationship between the electricity consumption and its main influencing factors. The data selected in the model were the monthly data of Hunan Province from January 2010 to December 2018. All explanatory variable data were obtained from the Wind database. This study established the multiple linear regression models between the electricity consumption of three key industries and its main explanatory variables as follows: Among them, 1 Elec , 2 Elec , 3 Elec respectively represent the electricity consumption of ferrous metal smelting and rolling processing industry, non-metallic mineral products industry, and transportation and electrical and electronic equipment manufacturing industry (unit: 10,000 kWh). output represents the iron and steel output (unit: ten thousand tons); pmi represents the Purchasing Managers' Index (PMI Composite Index) of the steel industry; inv represents the investment amount of real estate development (unit: 100 million yuan); index is the raw coal price index; mylpic is the mine price index of Mylpic; ceoutput denotes cement output (unit: ten thousand tons); cep denotes cement price (unit: yuan/ton); sci represents the climate index of non-metallic mineral products industry; auto denotes automobile output (unit: ten thousand tons); cir is integrated circuit output (unit: ten thousand); pcdi represents per capita disposable income (unit: yuan). Based on the above models and research data, the stepwise regression results were shown in Table 1-3.  Table 1 showed the regression results for ferrous metal industry. Explanatory variables were added successively from Model 1 to Model 4, and the goodness of fit R 2 increased gradually, indicating that the model was improving. In Model 4, R 2    The regression results of non-metallic mineral products industry were shown in Table 2  The regression results of transportation and electrical and electronic equipment manufacturing were shown in Table 3. With the gradual inclusion of explanatory variables, the coefficient values of most variables did not change much and were always significant at the significance level of 0.01, and R 2 increased to 0.852. For each explanatory variable, as the main products of this industry, auto and cir were statistically highly significant. When auto and cir increased by 1%, 3 Elec increased by 0.154% and 0.057% respectively. pcdi was also significant at the significance level of 0.01. When pcdi increased by 1%, 3 Elec increased by 0.753%.

Conclusion
This paper used multiple linear regression analysis models to explore the internal and external factors affecting industrial electricity consumption in Hunan Province. The regression results showed the iron and steel output, raw coal price index and Mylpic mine price index can significantly affect the electricity consumption of ferrous metal industry. Cement output, cement price and real estate development investment can significantly affect the electricity consumption of non-metal industry. Automobile output, integrated circuit output and per capita disposable monthly income can significantly affect the electricity consumption of transportation and electrical and electronic equipment manufacturing industry in Hunan Province. In addition, the models also can be used to forecast the short-term direction of electricity consumption in corresponding industries of Hunan province.