The Effect of China's Regional Economic Competitiveness on CO2 Emissions-Based on Economic Factors

12 Under the pressure of emission-reduction, China's regional economy needs to eliminate the mode of high energy consumption, high pollution, 13 and high emission. However, the economic development of each region has its own characteristics. Therefore, to analyze the differences in the 14 effect of regional economic development on CO 2 emissions, this work made an investigation from the perspective of regional economic 15 competitiveness. First, using the panel data of thirteen economic factors from 2003-2017, this work evaluated the economic competitiveness of 30 16 provinces from the four aspects: competitiveness of residents ’ wealth (WC), competitiveness of opening-up (OC), competitiveness of technology 17 (TC), and competitiveness of industrial structure (IC). Furthermore, the panel principal component regression model (PPCR) was established for 18 the eastern region, the central region, and the western region according to the score of competitiveness in four aspects. The results suggested that 19 the promoting effect of WC on CO 2 emissions presented the rule opposite to WC. In other words, the stronger the competitive advantage of WC, the 20 weaker its effect on promoting CO 2 emissions. The "pollution refuge" hypothesis was verified in all three regions of China. And the promotion of 21 CO 2 emissions by OC was the strongest in the central region, followed by the eastern region, and the weakest in the western region. TC promoted 22 CO 2 emissions in the central region, but inhibited CO 2 emissions in the eastern and western regions. Finally, the inhibiting effect of IC on CO 2 23 emissions showed the same law as IC. Based on the conclusions, some


Introduction 43
Human economic activities consume much fossil energy and play a vital role in CO2 emissions (Fang et al., 2018). 44 Furthermore, as China is concerned, its economic development has made remarkable achievements. Nevertheless, at the 45 same time, China's energy consumption and pollution emissions have expanded sharply. In 2010, China surpassed the 46 United States as the largest energy consumer, accounting for 20.3 percent of global energy consumption (Wang, 2010). As 47 early as 2007, China was the world's largest emitter, with more than 6 billion tons of CO2 emissions, accounting for 21 percent of 48 global CO2 emissions (IEA, 2007). In 2017, China's total energy consumption and CO2 emissions from fossil fuel combustion were 49 4.49 billion tons of standard coal and 10.4 billion tons (Quéré et al., 2018). Therefore, China is facing tremendous pressure on 50 energy-conservation and emission-reduction, which requires China's economic development to get rid of the old norm of high 51 energy consumption, high pollution, and high emissions, and attain low-carbon economy. 52 The low-carbon economy should focus on many aspects. First of all, technological progress is the key to the low-carbon 53 economy(Kang et al., 2018). Production-based technological advances, which can increase output while reducing factor inputs, 54 are conducive to reducing pollution emissions (Santra, 2017). In addition, the advances of environmentally friendly technological 55 such as clean energy technologies, energy-saving technologies, and carbon sequestration technologies can effectively reduce CO2 56 emissions(Y. Chen et al., 2020). Secondly, industrial structure adjustment and optimization are the main ways to develop a 57 low-carbon economy. Industrial restructuring and optimization can led to the flow of productive factors to high-productivity 58 sectors. (Luan et al., 2021). Moreover, the ultimate goal of economic development is to satisfy consumption. Therefore, the 59 development of a low-carbon economy promotes the gradual upgrading of the consumption from high-carbon material 60 consumption to low-carbon service consumption (Schanes et al., 2016). So, on the basis of technological progress, the 61 eight fossil fuels; denotes the CO2 emissions; denotes the th fuel consumption of ℎ province in the ℎ year. 174 denotes CO2 emissions coefficients of fossil fuels. 175  Total imports and exports divided by GDP ％ Export dependence (ED) Total exports divided by GDP ％ Import dependence (ID) Total imports divided by GDP ％ Foreign direct investment dependence (FDI) Total foreign direct investment divided by GDP ％ Technical output amount (TOA) Contract amount of technical output in technology market CNY 100 million Technical introduction amount (TIA) Contract amount of technical input in technology market CNY 100 million R&D intensity(R&D) Total R&D investment divided by GDP ％ Secondary industry (SI) The added value of the secondary industry divided by GDP ％ Tertiary industry (TI) The added value of the tertiary industry divided by GDP ％

Principal component factor analysis 180
Principal component factor analysis (PCFA) is employed to evaluate economic competitiveness. Factor analysis has a vital role 181 in using a few common factors to explain many variables with solid correlations: dimensionality reduction (Ivosev et al., 2008). 182 The common factor analysis model is as Eq. (2). 183 { 1 = 11 1 + 12 2 + ⋯ + 1 + 1 2 = 21 1 + 22 2 + ⋯ + 2 + 2 ⋮ = 1 1 + 2 2 + ⋯ + + (2) Where, 1 , 2 , ⋯ , are twelve economic factors. 1 , 2 , ⋯ , are independent of each other and are common factors, and 184 ( 1 , 2 , ⋯ , ) = . 1 , 2 , ⋯ , are so-called special factors. The ( = 1, ⋯ , ; = 1, ⋯ , )are the factor loading, 185 which indicates the degree of correlation between and . What needs to be explained is that this work uses the principal 186 component method to solve ( = 1, ⋯ , ; = 1, ⋯ , ). The Factor scores of 1 , 2 , ⋯ , can be obtained by the least 187 squares regression method proposed by Thurstone (1934). The comprehensive score is derived from Eq. (3). 188 = ( 1 × 1 + 2 × 2 + ⋯ + × )⁄ (3) 189 Where, ( = 1,2, ⋯ , ) represents the variance contribution rate of factor , represents the total variance contribution 190 rate of all common factors and = 1 + 2 + ⋯ + .  and Bartlett fully illustrated the high degree of correlation between economic factors (Appendix Table a.2). 215 The results of PCFA are shown in tables 2 and 3. The first four common factors ( 1 , 2 , 3 , 4 ) explained the variance of 90.783 216 percent of the thirteen original economic factors. The factor loadings of the first common factor 1 on urban residents' 217 consumption level, rural residents' consumption level, residents' consumption level, and GDP per capita were 0.905, 0.902, 0.870, 218 and 0.851, respectively. Therefore, we called the first common factor 1 the competitiveness of residents' wealth, which reflected 219 the ability of regional residents to create and consume wealth. The factor loadings of the second common factor 2 on foreign 220 trade dependence, export dependence, import dependence, and foreign direct investment dependence were 0.916, 0.909, 0.799, and 221 0.717, respectively. We named the second common factor 2 as the competitiveness of opening-up, reflecting regional economies' 222 degree of external dependence. The factor loadings of the third common factor 3 on technical output amount, technical 223 introduction amount, and R&D intensity were 0.882, 0.769, and 0.662, respectively. We defined the third common factor 3 as the 224 competitiveness of technology, demonstrating the power of regional technology R&D and technology diffusion. Finally, the 225 factor loadings of the fourth common factor 4 on secondary industry and tertiary industry were -0.956 and 0.784. We defined the 226 fourth common factor 4 as the competitiveness of industrial structure, which reflected the reasonable degree of regional 227 industrial structure. It should be noted that of the thirteen economic factors, only the factor loading of the secondary industry was 228 negative. In other words, the increase of the proportion of secondary industry will reduce the competitiveness of the regional 229 industrial structure. On the contrary, the improvement of the remaining economic factors will enhance the corresponding regional 230 economic competitive advantage. Then, we got the economic competitiveness index system (Appendix Table a.

3). 231
Furthermore, the competitiveness scores of WC, OC, TC, and IC were obtained by least squares regression method (Thurstone, 232 1934). They are shown in the Appendix (    Regarding the competitiveness of residents' wealth, the eastern region is potent than the central and western regions. This is 250 because the four factors contained in the competitiveness of residents' wealth in the eastern region are remarkably higher than in 251 the central and western regions. In addition, although there is no apparent difference in the competitiveness of the residents' 252 wealth in the central and western regions in the first half of the research time, the western region showed a slight advantage over 253 the central region in the second half of the study period. Therefore, we believed that the competitiveness of residents' wealth of 254 the eastern region was the most competitive advantage, the west region was second, the central region was the weakest. In other 255 words, the eastern region is the strongest, the western region second, and the central region the weakest for the ability of 256 residents to create and consume wealth. 257 In terms of the competitiveness of opening-up, the eastern region is most potent, the central region is second, and the western 258 region is the weakest. Moreover, the competitiveness of opening-up for the three regions showed a downward trend, of which the 259 decline was most evident in the eastern region. However, the competitiveness of opening-up for the eastern region still has 260 apparent advantages. As a result, the degree of external dependence of the economies of the three regions has decreased year by 261 year, especially in the eastern regions. It is also important to note that FDI in the central region continues to increase, while there 262

265
For the competitiveness of technology, the eastern region has significant advantages. The intensity of independent R&D in the 266 eastern region is much higher than that in other regions, indicating that the technological progress in the eastern region depends 267 mainly on independent R&D. At the same time, the contract amount of technology export is higher than the technology 268 introduction in the eastern region, indicating that the eastern region is the net output region of the technology. In addition, 269 compared with the eastern region, the competitiveness of technology of the central and western regions is at a significant 270 disadvantage. First, the R&D efforts in the central and western regions are weaker than those in the eastern region. Furthermore, 271 the technology introduction turnover in the central and western regions is higher than the technology output turnover, indicating 272 that both regions are net technology introduction areas. Moreover, the central region's independent technology R&D are more 273 potent than that of the western region. So, the western region's technological progress depends more on technology diffusion than 274 the central region. 275 As far as the competitiveness of industrial structure is concerned, it is arranged from highest to lowest according to western western region are about 45 percent and 40 percent, respectively. However, the decline of the secondary industries in the eastern 278 and central regions was higher than 50 percent. Moreover, the rising of the tertiary industry in the central region is significantly 279 lower than that in the western region. Therefore, the nodes of industrial structure adjustment in the western region are better than 280 those in the eastern and central regions, which makes the industrial structure of the western region more reasonable.  Table 4 showed the unit root test 286 results for the three regions. It can be found that all variables accepted the null hypothesis at their levels, that was, the unit root 287 process. Moreover, all variables rejected the null hypothesis that was unit root process in the first-order difference. Therefore, the 288 first-order difference stationarity can be determined. Furthermore, this work employed Kao test (Kao, 1999), Pedroni test 289 (Pedroni, 2004), and Westerlund test (Westerlund, 2005) to check the long-term cointegration relationship among variables. As 290 shown in Table 5, almost all statistics rejected the null hypothesis that there was no cointegration relationship. Thus, we have 291 determined the cointegration of the panel data used in the PPCR model. 292

Estimation results of the PPCR model 301
Whether the PPCR model established by using the competitiveness score can avoid multicollinearity needs to be verified. 302 Variance expansion factor (VIF) was a widely used tool for judging multicollinear relationship (Farrar et al., 1967). In all three 303 regions, the VIF value of all independent variables was not greater than 10(Appendix Table a (Geisser, 1974). Therefore, the fixed effects model was chosen to fit the PPCR model in 307 all three regions. The estimation results of the three regions are shown in Table 6. 308 From the model estimation results shown in Table 7, it can be seen that the effects of WC, OC, TC, and IC on CO2 emissions 309 were statistically significant in all three regions. Among them, the elastic coefficients of WC were positive in all three regions. 310 Moreover, it showed the rule that the central region was the largest, the eastern region was the smallest, and the western region 311 was in the middle. Therefore, the WC has a promoting effect on CO2 emissions in all three regions, and this promotion has the 312 same law to the elasticity coefficients. Regarding OC, its elasticity coefficients were positive in all three regions, and its value 313 increased according to the order of the western region, the eastern region, and the central region. This implied that OC has the 314 effect of promoting CO2 emissions in all three regions. This promoting effect was most potent in the central region, most minor 315 the western region, and centered in the eastern region. As far as TC was concerned, its elasticity coefficients were positive in the 316 central region, negative in the western and eastern regions. Therefore, TC has the effect of promoting CO2 emissions in the 317 central region, and inhibiting CO2 emissions in the western and eastern regions. Finally, the elasticity coefficients of IC were 318 negative in all three regions, which implied that IC has the effect of inhibiting CO2 emissions in all three regions. Moreover, the 319 inhibiting effect of IC was most potent in the western region, most minor in the central region, and centered in the eastern region. 320

Discussion 324
Based on the above empirical conclusions, we have found some valuable phenomena. 325 The promoting effect of residents' wealth competitiveness on CO2 emissions showed the opposite law to residents' wealth 326 competitiveness. Inevitably, the process of creating wealth and consuming it consumes fossil fuels and brings about CO2 emissions. 327 The competitiveness of residents' wealth reflects their ability to create wealth and their ability to consume wealth. And the ability 328 of residents to create wealth and consume wealth promote each other. First, the ability of residents to create wealth is the basis of 329 the ability to consume wealth. On the one hand, the stronger the ability of residents to create wealth, the greater the ability of 330 residents to pay for wealth, which makes residents "able" to consume. On the other hand, the more wealth-generating the 331 residents can provide, the greater the level of social security that can be provided, which makes them "dare" to consume. 332 Moreover, the consumption power of residents' wealth is the driving force of residents' wealth creation. When the ability of 333 residents to create and consume wealth is weak, the consumption demand of residents mainly meets the material needs of basic 334 survival, such as clothing, food, housing, travel, etc. On the contrary, when residents have a solid ability to create and consume 335 wealth, the consumer demand is mainly to meet their development and enjoyment of service needs, such as: culture, health care, 336 art, education, entertainment, tourism. In other words, the improvement of residents' wealth competitiveness can promote the 337 upgrading of residents' consumption structure from "survival" consumption to "development and enjoyment" consumption. 338 Obviously, "survival" material consumption is a high CO2 consumption relative to "development and enjoyment" service 339 consumption. Therefore, it was found that the higher the competitiveness of residents' wealth, the lower the contribution to CO2 340

emissions. 341
It can be determined that the "pollution refuge" hypothesis is established in all three regions of China. This differs from the 342 conclusion of(Jin et al., 2016), who believed that the "pollution refuge" hypothesis only exists in the western region. At the same 343 time, the promoting effect of opening-up competitiveness on CO2 emissions is the strongest in the central region, followed by the 344 eastern region and the weakest in the western region. The competitiveness of opening-up included the dependence of economic 345 development on foreign trade and the foreign capital, which reflected the external dependence of economic development. The 346 economic development of the Eastern region has made remarkable achievements, which has led to the gradual transformation of 347 its economic development from the demand for economic scale to the pursuit of economic efficiency. Moreover, although the 348 eastern region's external trade has maintained a net export model, its dependence on foreign trade has decreased. Besides, its 349 export commodity structure has gradually transitioned from labor-intensive and resource-intensive manufactured goods to 350 technology-and capital-intensive high-value-added commodities. In addition, the dependence on foreign capital of the Eastern 351 region has declined rapidly, and more attention has been paid to guiding foreign investment into high-tech, low-polluting 352 industries. Therefore, the promoting effect of opening-up competitiveness on carbon emissions has been curbed to some degree 353 in the Eastern region. For the central region, its external economic dependence has been low. However, due to its energy 354 endowment, the central region's exports are mainly steel, cement, and other energy-intensive commodities, which undoubtedly 355 increased CO2 emissions. In addition, the central region's demand for economic scale led to an increase in foreign investment in 356 high-energy-consuming and high-emission industries. Therefore, the opening-up competitiveness of the central region has the 357 most potent effect on CO2 emissions. In other words, the the " pollution refuge" hypothesis is most pronounced in the central 358 region. Similar to the central region, the external economic dependence of the western region has been low, and its foreign trade 359 is also reflected in net exports. However, exports from the western region are mainly primary products, such as unprocessed or 360 initially processed agricultural products and extractive industrial products. At the same time, the economic dependence on FDI in 361 the western region has been low, and there is a more apparent downward trend. Moreover, the economic development of the 362 western region started late, drew lessons from the central region, and strictly controlled foreign investment in 363 high-energy-consuming and high-polluting industries. Therefore, although the competitiveness of opening-up in the western 364 region has a statistically beneficial effect on CO2 emissions, it is feeble. 365 The competitiveness of technology inhibits CO2 emissions in the eastern and western regions, but promotes CO2 emissions in 366 the central region. This is broadly in line with the conclusion of(J. Chen et al., 2020). They confirmed that technological 367 advances have the effect of reducing CO2 emissions in the central and western regions and increasing CO2 emissions in the 368 eastern region. The eastern region has the most potent R&D efforts and is the primary source of Chinese technology. Moreover, 369 benefiting from its economic development and environmental regulation, R&D in the eastern region favors environmentally 370 friendly technologies and low-carbon production technologies. In addition, the eastern region is the area of net technology output, 371 which provides the capital base for technological progress, which is conducive to its technological progress, especially to the 372 progress of environment-friendly technology. As a result, it was found that the technological competitiveness of the eastern 373 region inhibited CO2 emissions. For the central region, because of the pursuit of the economic scale, the environmental 374 supervision is relatively broad, which leads to the R&D bias towards industrial production technology. At the same time, the 375 central region's technological progress is less dependent and prefers the introduction of industrial production technology. None of 376 this is conducive to the progress of technology, especially environment-friendly technology. As a result, the technological 377 competitiveness of the central region contributed to energy demand and CO2 emissions. For the western region, although the 378 technological R&D is low, the technological progress is more dependent than that in the central region. So, the introduction of 379 technology in the western region is more efficient than that in the central region, which is more beneficial to the progress of 380 environment-friendly technology. In addition, compared with the central region, the technological R&D in the western region 381 favors environment-friendly technology. Therefore, the technological competitiveness curbs CO2 emissions in the western region. industry. The adjustment and optimization of the industrial structure are beneficial to the allocation of resources and improve the 389 rationalization of the industrial structure, which is undoubtedly conducive to curbing CO2 emissions. Another interesting finding 390 is that the inhibiting effect of the competitiveness of industrial structure on CO2 emissions is most powerful in the western region, 391 mainly because the nodes of industrial structure adjustment and optimization in the western region are better than those in the 392 eastern and central regions. The lower the node where the proportion of the secondary industry decreases, the fewer CO2 393 emissions the secondary industry will cause. On the contrary, the higher the node where the proportion of the tertiary industry 394 increases, the stronger the inhibiting effect on CO2 emissions. Therefore, for CO2 emissions, the lower the node where the 395 proportion of the secondary industry decreases, the higher the node where the proportion of the tertiary industry increases, and 396 the more reasonable the industrial structure is. Therefore, it was found that the industrial structure is more reasonable in the 397 western region. 398 Compared with the three regions, the improvement of the competitiveness of residents' wealth has the effect of curbing the 411 increase of CO2 emissions. Therefore, striving to improve the competitiveness of residents' wealth in various regions is an 412 effective measure to curb the increase of CO2 emissions. Furthermore, wealth creation-ability and consumption-ability are two 413 mutually reinforcing aspects of the competitiveness of residents' wealth. Therefore, the eastern region should take full advantage 414 of its highest wealth creation-ability and consumption-ability to promote the upgrading of its consumption structure. On the 415 contrary, the central and western regions have low wealth creation and consumption power, and their "survival" material 416 consumption patterns are difficult to change in the short term. So, the central and western regions should focus on the ability of 417 create wealth, that is increase GDP per capita, to promote the improvement of their consumption level and the upgrading of 418 consumption structure. 419

Conclusions and policy implications 399
The competitiveness of opening-up promoted CO2 emissions in three regions. In other words, the external dependence of 420 economic development increases CO2 emissions in the three regions. Therefore, efforts should be made to reduce the external 421 dependence of the economic development of the three regions. From the perspective of foreign trade, we must first try to reduce 422 the degree of economic dependence on foreign trade, and promote the development of a virtuous "inner circle" of the economy. 423 Furthermore, efforts should be made to improve the commodity structure of foreign trade, promote the export of trade in services, 424 and reduce the export of energy-intensive and polluting commodities. From the perspective of FDI, the government should 425 optimize the industrial layout of FDI. In high-energy-and high-polluting industries, FDI should be guided to adopt advanced 426 production technology to reduce pollution emissions. In addition, it is more important to guide FDI into the clean industry, 427 environmental protection industry, and actively play the FDI technology spillover effect. These measures should be more 428 meaningful in the central region. 429 As far as technological competitiveness is concerned, the eastern region has apparent advantages. Therefore, the eastern region 430 should fully play its advantages in R&D and technology diffusion. First, the eastern region should continue to increase 431 investment in R&D of production-oriented and environmental-friendly technologies. Furthermore, it is necessary to encourage 432 technology diffusion from the eastern region to the central and western regions. In addition, R&D often has the characteristics of 433 significant investment and slow return (Hall et al., 2010). On the contrary, the introduction of technology is more targeted and 434 effective. Therefore, for the central and western regions, subject to economic development and R&D capital constraints, 435 emphasis should be placed on increasing investment in technology introduction. Especially in the central region, limited funds 436 should be invested in the introduction of production-oriented technology to improve production efficiency and reduce energy 437 consumption and pollution emissions. 438 After a period of industrial restructuring in various regions, the industrial structure gradually tends to be rationalized, which is 439 manifested in that the competitiveness of the industrial structure inhibits CO2 emissions in all regions. However, there are 440 apparent differences in the rationality of the industrial structure in each region. Therefore, differentiated measures should be 441 taken to improve the rationality of the industrial structure in each region. For the eastern and western regions where the industrial 442 structure is relatively reasonable, we should focus on increasing the proportion of the tertiary industry. For example, the 443 government needs to issue policies to encourage the development of high-end service industries such as tourism, finance, cultural 444 industries, and technical services. On the contrary, for the central region where the industrial structure is less reasonable, we 445 should focus on reducing the secondary industry's proportion. The policy should try its best to guide the resources in the 446 secondary industry to resource-saving and environment-friendly industries. 447