Energy productivity and Chinese local officials’ promotions: Evidence from provincial governors

Energy Policy 95 (2016) 103–112

Contents lists available at ScienceDirect

Energy Policy journal homepage: www.elsevier.com/locate/enpol

Energy productivity and Chinese local officials’ promotions: Evidence from provincial governors Xiude Chen a, Quande Qin b,c,d,n, Y.-M. Wei c,d a

School of Management, Guangdong University of Technology, Guangzhou 510520, China Department of Management Science, Shenzhen University, Shenzhen 518060, China c Center for Energy and Environmental Policy Research, Beijing Institute of Technology, Beijing 100081, China d School of Management and Economics, Beijing Institute of Technology, Beijing 100081, China b

H I G H L I G H T S







The data of position changes for China’s provincial governors during 1978–2012 are utilized. Energy productivity has a positive impact on provincial governors’ promotion in China. Political incentive is an important driver of the improvement in China’s energy productivity. The correlation between energy productivity and local officials’ promotions was evolved.

art ic l e i nf o

a b s t r a c t

Article history: Received 6 September 2015 Received in revised form 17 March 2016 Accepted 25 April 2016

Improving energy productivity is one of the most cost-effective ways to achieve a sustainable development target. The existing literature has shown some factors that have driven the improvement in China’s energy productivity. However, these studies do little to tackle the role of Chinese local officials. Political promotions can be seen as the most important career incentive for Chinese local officials. Hence, we intend to study whether energy productivity affects Chinese local officials’ promotions in this paper. The data of position changes for the 31 provincial governors during 1978‐2012 are utilized. We adopted probit models to empirically examine the correlation between provincial governors’ political promotions and energy productivity. The empirical results demonstrate that (1) energy productivity has a significantly positive impact on provincial governors’ political promotions in China, meaning that the provincial governors have the momentum to improve energy productivity; and (2) the effect of energy productivity on provincial governors’ political promotions has evolved, dynamically changing along with the transformation of the economic growth mode and the adjustment of the local officials’ promotion mechanism. The results are helpful in understanding the drivers of the improvement in China’s energy productivity and provide insightful implications for conducting energy policy in China. & 2016 Elsevier Ltd. All rights reserved.

Keywords: Energy productivity Energy efficiency Local officials’ promotions Probit model

1. Introduction Since the implementation of the reform and opening-up policies, China’s economic reforms have resulted in spectacular growth. The gross domestic product (GDP) has increased from 364.52 billion RMB in 1978 to 51.93 trillion RMB in 2012, with an average annual growth of 9.97%.1 Meanwhile, China’s energy consumption is also exhibiting a rapid growth trend. According to n Corresponding author at: Department of Management Science, Shenzhen University, Shenzhen 518060, China. E-mail addresses: [email protected] (X. Chen), [email protected] (Q. Qin), [email protected] (Y.-M. Wei). 1 Data is from the “China Statistical Yearbook (1981–2013)”.

http://dx.doi.org/10.1016/j.enpol.2016.04.041 0301-4215/& 2016 Elsevier Ltd. All rights reserved.

the “China Statistical Yearbook 1981–2013”, China’s total energy consumption was 57,144 ten-thousand tons of standard coal in 1978 but reached to 361,732 ten-thousand tons of standard coal in 2012. Based on the data in BP's “Statistical Review of World Energy 2011″, China consumed 20.3% of the world's energy, exceeding the 19% rate of the US, and became the world's largest consumer of energy (BP, 2011). It is expected that China’s energy demand will increase in the future, putting enormous strain on China’s economy and the environment. Under these circumstances, it is imperative for China to improve its energy efficiency. Energy efficiency growth has at least two beneficial effects on social and economic development. First, it contributes to reducing carbon dioxide emissions without necessarily increasing energy consumption. Second, it provides national energy security benefits

104

X. Chen et al. / Energy Policy 95 (2016) 103–112

during the process of economic growth. Energy productivity, the inverse of energy intensity, is defined as the ratio of output divided by energy consumption (Patterson, 1996; Han et al., 2007; Miketa and Mulder, 2005; Dimitropoulos, 2007). Generally, energy productivity can be seen as a useful indicator for understanding the energy efficiency of an industry or an economy (Patterson, 1996; Hu and Wang, 2006; Chang and Hu, 2010). For all governments, improving energy productivity is an import way to maximize the economic welfare extracted from the energy system. Perhaps, nowhere else in the world is the issue more salient than in China. China has experienced a steady increase in energy productivity from the onset of economic reform in late 1970 s (Han et al., 2007; Chang and Hu, 2010). Data from the China Center for Economic Research (CCER) database demonstrates that the average energy consumption elasticity during 1978–2012 was approximately 0.56, which was less than 1 and further decreased to approximately 0.5 in the most recent ten years. What is driving the increase in China’s energy productivity? In previous studies, the decomposition approach was widely employed. Some factors that have affected China’s energy productivity have been found. These factors include sectoral adjustments, energy prices, technological progress, energy efficiency, structural shifts, the energy consumption structure, foreign direct investment and energy use technology (Sinton and Levine, 1994; Patterson, 1996; Garbaccio et al., 1999; Fisher-Vanden et al., 2004; Fisher-Vanden et al., 2006; Hang and Tu, 2007; Liao et al., 2007; Fan et al., 2007; Ma and Stern, 2008; Chai et al., 2009; Feng et al., 2009; Chang and Hu, 2010; Wang, 2011; Elliott et al., 2013; Lin and Du, 2014). The abovementioned literature has important implications for our understanding of the dynamic evolution of China’s energy productivity. Many empirical results demonstrated that the political incentives for local officials constitute a major key to understanding economic performance in China (Bo, 1996; Zhou, 2004; Li and Zhou, 2005; Zhang and Gao, 2008). Like politicians elsewhere, Chinese local officials care most about their political future. Zhuravskaya (2000) and Bardhan (2006) suggested that local official governance has brought about differences in economic development among Russia, India and China since the 1990s. Su et al., (2012) and Zhang and Gao (2008) found that China’s success in economic development was due to local officials competing for economic growth. It had long been assumed that local GDP growth was the main yardstick for the central government for appraising local officials’ performance (Chen et al., 2005; Li and Zhou, 2005). China has a regionally decentralized authoritarian system, which is characterized as a combination of political centralization and economic regional decentralization (Xu, 2011; Wan et al., 2015). Actually, local officials hold the ultimate right to choose the mode of economic growth (Li and Zhou, 2005). In this paper, we want to examine whether local officials have the incentive to improve energy productivity. It is assumed that local officials’ promotion incentives have been playing a vital role in Chinese economic development. To encourage Chinese local officials to focus on energy productivity, the central government should take it as an assessment indicator. In fact, the Chinese government has enacted or amended a number of laws and regulations focused on energy saving and emission reduction since the reform and opening-up, such as the “Provisional Regulations on the Control of Energy Conservation (1986)” and “Comprehensive Work Program of Energy Saving and Emission Reduction (2007)”. From a policy point of view, energy conservation work conducted by local government officials should be integrated into the assessment of the local officials’ political performance alongside output growth. If energy productivity became an important criterion for official promotion, career-minded officials would try to excel in tournament-like competitions. The

incentive for local officials is crucial for improving energy productivity. Does energy performance really affect the political promotions of Chinese local officials? To the best of our knowledge, there exist few studies in the literature that examined the question. In this paper, we collected data of position changes for the 31 provincial governors during 1978–2012 and utilized a probit model to study whether energy productivity does matter for provincial governors’ political promotions. This paper focuses on the effects of energy productivity on the likelihood of the promotion of local officials. This paper argues that energy productivity has a stronger impact on the likelihood of promotion of provincial governors. Performance in energy productivity correlated positively with the probability of promotion during local officials’ tenure, and the correlation has gradually evolved. The remainder of this paper is organized as follows. Section 2 explains the probit method. Section 3 introduces the variables and data. In Section 4, we present the empirical results and analysis. The conclusions are drawn in Section 5.

2. Probit method A probit model is employed to handle a regression problem where the dependent variable can take only two values (Ai and Norton, 2003). The purpose of the model is to estimate the probability that an observation with particular characteristics will fall into a specific category. In this paper, the probit model is utilized to examine whether the performance of energy productivity during an official's tenure affects the official's promotion. Assume the central government assesses local officials based on a series of factors x i and obtains a score yi*. Whether a local official should be promoted depends entirely on the relative size of yi*. Further assume there is a linear functional relationship between the yi* and x i as formula (1) describes:

y*i = xi β + εi

(1)

where εi represents an independent and identically distributed random error term. β represents the coefficient of assessment factors x i . yi* could not be directly observed, but we could use the appointment and removal of local officials as an indirect reflection indicator. Assume yi represents the appointment and removal results of local officials, which is a dummy variable. When a local official obtains a promotion, yi is equal to 1; otherwise, yi is equal to 0. Let us further assume there is assessment score threshold γ that determines whether a local official should obtain a promotion; then,

⎧ 1 ify* > γ yi = ⎨ ⎩ 0 ify* ≤ γ ⎪

⎪

(2)

Thus, we can employ the probit model to analyze the relationship between the assessment indicator x i and local officials’ promotions yi . In formula (2), if x i includes a constant component, γ could be arbitrary. The specific probit model can be set as follows:

yi = I(xi β > γ )

(3)

where I (∙) is the indicator function. We can determine the relationship between various assessment factors and local officials’ promotions according to estimation of the parameter β and its significance. For example, if the estimated parameter of energy productivity is significantly positive, that indicates that the energy productivity has a significant positive impact on the probability of local officials’ promotions. However, the estimated parameter β of the probit model can

X. Chen et al. / Energy Policy 95 (2016) 103–112

explain only the impact direction and cannot be employed to reflect the marginal effects of specific explanatory variables directly. Therefore, we would further estimate the marginal effects of the explanatory variables according to formula (4): Marginal effect of variables and data

∂P (yi = 1) ∂xj

= ϕ(xi β )βj

(4)

where ϕ(∙) is the standard normal probability density function.

3. Variables and data 3.1. Variables

(1) Dependent variable The dependent variable is a governor promotion dummy variable, which is represented by promotion. When a governor is promoted, promotion is set to 1; otherwise, promotion is 0. It is noted that the local officials are a group, including the secretary of a provincial party committee (SPPC), deputy secretary of a provincial party committee, governor and vice governor, etc. It is unrealistic to measure the promotion of all the local officials directly. We choose the governor, chairman of the autonomous region and mayor of the municipality as the representative of local officials (hereafter referred to as governor). The reason the governor is chosen as the research object is because a governor is directly responsible for the administration and economic development of a region and has absolute decision-making power in their areas of responsibility under China’s political and economic system. If a governor was promoted to the SPPC or transferred to the central government with a higher administrative level, we believe that this governor received a promotion; in other cases, they are deemed not to have been promoted. (2) Independent variables This is the first time that energy productivity is introduced into the influence factors of local officials’ promotions. In fact, the Chinese central government does not necessarily conduct an annual assessment of local officials or place more emphasis on energy performance within a relatively long term (such as the five-year term of office). Therefore, we considered the influence of two energy productivity indicators, which are the current energy productivity of the year in which the local official’ position changed (represented by Prodenergy ) and the average energy productivity of local officials’ entire tenure (represented by AV _Prodenergy ). The annual energy productivity of jurisdictions within the mandate of local officials is calculated by formula (5):

Prodenergy =

RGDP Energy

(5)

where RGDP represents the real GDP of all the provinces in China, which is calculated in constant 1978 prices. The RGDP excludes the impact of price changes in different accounting periods and truly reflects the output of energy input. Energy represents the total energy consumption converted to tenthousand tons of standard coal according to the energy consumption structure including coal, oil, hydro and wind power, etc. Energy consumption in terms of liter oil equivalent of an economy forms the Energy Balance Table. (3) Control variables In addition to energy productivity, the promotion of Chinese

105

local officials may also be affected by other factors. According to the existing studies, regional economic growth, education, nationality, age and tenure are likely to be influencing factors of Chinese local officials’ promotions (Zhang and Gao, 2008). To ensure the accuracy of the conclusions of this study, all the factors above are included as control variables in our empirical models; in addition, study abroad experience, engineer or economist experience and gender are also considered. As for measuring the economic performance, we considered the growth of real GDP by region. Similar to energy productivity, we also considered the influence of two economic performance indicators, which are the growth of real GDP by region in the year of local officials’ position changes (represented by CGGDP ) and the average of the growth of real GDP by region during local officials’ entire tenure (represented by AGGDP ). Education is divided into five levels: secondary and below, junior college, bachelor’s degree, master’s degree and doctor’s degree. We designed five dummy variables ( Dum _edu1, Dum _edu2, Dum _edu3, Dum _edu4 and Dum _edu5) to represent the corresponding level, and four dummy variables ( Dum _edu2, Dum _edu3, Dum _edu4 and Dum _edu5) are included in our empirical models. Nationality is measured by a dummy variable, which is represented by Nationality . When the governor is of Han ethnicity, Nationality is set to 1; otherwise, Nationality is 0. Because some studies suggest that the effect of age on local officials’ promotions may be nonlinear, we introduce the actual age variable Age and the over 65 years old dummy variable Dum _65 to control the influence of age. We utilize the difference between outgoing year and the year of taking office to measure the tenure of local officials. Finally, we also introduce four dummies to control for the influence of study abroad experience, engineer or economist experience and gender, which are represented by Abroad , Engineer , Economist and Gender respectively. According to the characteristics of China's current political system, Feng and Wu (2013) proposed that the social network of the local officials affects local officials’ promotions. In this paper, if the provincial governor has worked in a Communist Youth League office, the dummy variable, Youth, is set to 1; otherwise, the variable is 0. If the provincial governor has working experience in a Central level party committee or in government agencies (primarily, the State Council and various ministries and CPC Central Committee Party committee subsidiaries), the dummy variable, Central , is set to 1; otherwise, the variable is 0. 3.2. Data Before the reform and opening-up in 1978, the Chinese central government decided the appointment and removal of local government officials based primarily on purely political indicators; however, the standard has gradually evolved from considering purely political indicators to economic performance indicators since 1978 (Zhou, 2004). Therefore, energy performance might have been an influencing factor for Chinese local officials’ promotions since only 1978. In this study, we select the provincial governors who had the experience of political turnover in the 31 provinces in mainland China during 1978–2012 as our research objects. As the energy consumption data are seriously missing in Tibet, it is excluded in the final empirical analysis. Chongqing City and Hainan Province were formally established in 1997 and 1988, respectively, and they are analyzed from the year of their establishment. The data of total energy consumption were collected from the China Energy Statistical Yearbook 1987–2013, the China Statistical Yearbook 1981–2013 and the statistical yearbooks of various provinces, and we employed interpolation to complement missing data. The regional GDP and its deflator data were collected from the China Statistical Yearbook 1981–2013, the China Compendium

106

X. Chen et al. / Energy Policy 95 (2016) 103–112

of Statistics 1949–2004 (National Bureau of Statistics of China, 2005), the Gross Domestic Product of China 1952–2004 (National Bureau of Statistics of China, 2007) and the statistical yearbooks of various provinces. All the data related to Chinese local officials were collected from the Official Records of the People’s Republic of China (He, 2003), Xinhuanet.com (http://www.xinhuanet.com/) and People.com.cn (http://www.people.com.cn/) by manual retrieval. To guarantee the accuracy of the data from the network, we independently retrieved the information of all the local officials through the three sites listed above and conducted some necessary verification.

4. Empirical results 4.1. Descriptive statistics Table 1 presents descriptive statistics of the provincial governors’ turnovers in China during 1978–2012. A total of 290 persons were appointed as governors in the 31 provinces of mainland China. As of December 31, 2012, 257 provincial governors experienced political turnovers; among them, 81 governors obtained promotions. After excluding Tibet, these three data points were 281, 248 and 81, respectively. Due to the lack of energy consumption data for Tibet, we included the 248 provincial governors who had experience of political turnovers as our basic research object. Table 2 reports the descriptive statistics results of the main variables. We found that (1) The mean of the variable promotion was 0.3266, and this demonstrates that approximately one-third of the provincial governors received political promotions. (2) The means of Prodenergy and AV _Prodenergy are 0.1406 and 0.1334, respectively. It could be seen that in the year when a governor experienced his political turnover, the province under his jurisdiction consumed, on average, approximately 10,000 tons of standard coal energy, and GDP reached approximately 0.01425 billion RMB; and from the point of view of the province under his jurisdiction over the entire term, it consumed, on average, approximately 10,000 tons of standard coal energy, which could lead to GDP of approximately 0.01334 billion RMB. Therefore, the current energy productivity is higher than the average energy productivity of their tenure, indicating that China’s energy productivity has improved yearly. Meanwhile, we analyzed the change trend of Chinese annual energy productivity during 1978–2012, and the results are presented in Fig. 1. Fig. 1 shows that Chinese annual energy productivity substantially increased, particularly since 1990; the average annual growth trends in energy productivity are apparent. (3) The mean of CGGDP and AGGDP are 0.1058 and 0.1018, respectively. There was an overwhelming male-dominated phenomenon in the appointment of governors; among the 248 governors, only 3 governors are female; accounting for approximately 1.2%. This ratio demonstrates that the probability of women being appointed as a local chief executive Table 1 The appointment and removal of governor in China during 1978–2012.

Appointment of governor (person) Removal of governor (person) Promotion (person)

Includes Tibet

Does not include Tibet

290 257 81

281 248 81

Table 2 Descriptive statistics of the main variables. Variables

N

Mean

Median

Std. dev.

Promotion Prodenergy

248 248

0.3266 0.1406

0.0000 0.1139

AV _Prodenergy

248

0.1334

0.1077

0.0811

0.0257

0.5012

CGGDP AGGDP Gender Nationality Dum_edu1 Dum_edu2 Dum_edu3 Dum_edu4 Dum_edu5 Age Dum_65 Tenure Abroad Engineer Economist Youth Central

248 248 248 248 248 248 248 248 248 248 248 248 248 248 248 248 248

0.1058 0.1018 0.9879 0.9073 0.3226 0.1008 0.4274 0.1169 0.0323 60.5927 0.2460 3.9879 0.0685 0.2056 0.0282 0.2621 0.1653

0.1055 0.1023 1.0000 1.0000 0.0000 0.0000 0.0000 0.0000 0.0000 61.0000 0.0000 3.0000 0.0000 0.0000 0.0000 0.0000 0.0000

0.0425 0.0347 0.1095 0.2907 0.4684 0.3017 0.4957 0.3220 0.1770 5.3265 0.4315 2.1624 0.2532 0.4050 0.1660 0.4407 0.3722

-0.0910 -0.0910 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 46.0000 0.0000 1.0000 0.0000 0.0000 0.0000 0.0000 0.0000

0.4020 0.2237 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 75.0000 1.0000 12.0000 1.0000 1.0000 1.0000 1.0000 1.0000

0.4699 0.0888

Min 0.0000 0.0265

Max 1.0000 0.6201

Fig. 1. Change trend in Chinese annual energy productivity (1978–2012).

was very low. More than 90% of the governors were Han, and less than 10% of the governors were Minorities; most of the Minority governors were serving in Xinjiang, Inner Mongolia and Guangxi, and this is highly consistent with the Chinese ethnic regional autonomy policy. The mean values of Dum _edu1, Dum _edu2, Dum _edu3, Dum _edu4 and Dum _edu5 are 0.3226, 0.1008, 0.4272, 0.1169 and 0.0323, respectively. The average tenure of governors was approximately 4 years, which is slightly shorter than the 5-year terms stipulated by the Organic Law of local People’s Congress and local governments of the People’s Republic of China. The average age at the time of the departure of the governors was approximately 60 years old, and approximately 25% of the governors reached or exceeded the mandatory retirement age of 65. Approximately 7% of the governors had the experience of studying abroad, 3% of the governors were economists, and 20% of the governors were engineers; this indicates that Chinese local officials favor technology. The percentage of governors having working experience in the Communist Youth League office and Central level party was 26.21 and 16.53, respectively. Furthermore, we distinguished the governor samples for the promotion group and the non-promotion group and conducted a group mean comparison to examine whether there are significant

X. Chen et al. / Energy Policy 95 (2016) 103–112

average economic growth is also faster in the promotion group. Therefore, we concluded preliminarily that the economic growth performance and energy productivity are likely to affect the promotion of Chinese local officials.

Table 3 Group mean comparison of energy productivity and economic performance. promotion¼ 0

promotion ¼ 1

N

Mean

N

Mean

Prodenergy

167

0.1267

81

-0.1693

-0.0427***

AV _Prodenergy

167

0.1206

81

0.1599

-0.0393***

CGGDP AGGDP

167 167

0.1031 0.0961

81 81

0.1114 0.1137

-0.0083 -0.0176***

Variables

107

Mean-difference

4.2. Test results based on probit model Table 4 presents the estimation results for the probit models. The estimation between the current energy productivity and Chinese local officials’ promotions is shown in column (1) to column (3), and the estimation between the average energy productivity of tenure and Chinese local officials’ promotions is shown in column (4) to column (6). In column (1), we did not introduce any control variables and found that the regression coefficient for Prodenergy is 3.3180, which is significant at the 1% level. In columns (2) and (3), we considered the influence of control variables and province effect; the regression coefficients for Prodenergy are 2.7433 and 2.2130, respectively, which are all significant at the 5% and 10% levels, respectively. All the results above indicate that current energy productivity has a significant positive impact on the governors’ promotions, and the higher the current energy productivity is, the greater is the probability that the governor is promoted. The estimation in columns (4) to (6) indicates that the regression coefficients for AV _Prodenergy are 3.6453, 2.6416 and 2.1121, respectively, and the first two estimated coefficients are significant

differences in energy productivity and economic growth between the two groups; the group mean comparison results were shown in Table 3. The means of Prodenergy and AV _Prodenergy in the promotion group are 0.1693 and 0.1599, whereas the means in the non-promotion group are 0.1267 and 0.1206, respectively. Compared with the non-promotion group, the promotion group has a higher current energy productivity and average energy productivity of tenure; the mean differences of the two kinds of energy productivity between the two groups are 0.0427 and 0.0393, respectively, which are significant at the 1% level. As for the regional economic growth performance, the mean difference of the current economic growth is 0.0083, but it is not significant; meanwhile, the mean difference of the average economic growth of tenure is 0.0176, which is significant at the 1% level. Thus, the

Table 4 Energy productivity and Chinese local officials’ promotions: test results based on probit model. Variables

(1)

(2)

(3)

(4)

(5)

(6)

Constant

-0.9297*** (  5.7670) 3.3180***

2.1920 (1.5969) 2.7433**

1.6461 (1.1815) 2.2130*

-0.9498*** (  5.7543)

2.1258 (1.5532)

1.5745 (1.1306)

(3.4301)

(2.3534)

(1.8376)

Prodenergy AV _Prodenergy CGGDP

2.6416**

2.1121

(3.4494)

(2.1235) 1.9141 (0.7854)

(1.6280)

1.7366 (0.7035)

AGGDP Nationality Dum_edu2 Dum_edu3 Dum_edu4 Dum_edu5 Abroad Engineer Economist Tenure Age Dum_65 Youth Central Province effect Observations Pseudo R2 Log pseudolikelihood

3.6453***

Yes 248 0.0397 -150.4610

0.9772* (1.8654) 0.1321 (0.4275) 0.0281 (0.1094) 0.4900 (1.2636) 0.2087 (0.4251) 0.0311 (0.0890) -0.0108 (  0.0619) -0.0762 (  0.1353) -0.1222*** (  2.7956) -0.0627*** (  2.6242) -0.4020 (  1.0312) 0.1054 (0.4149) 0.2601 (1.0880) Yes 245 0.1997 -124.4382

7.6985** (2.3336) 0.9468* (1.8267) 0.1550 (0.4885) 0.0362 (0.1330) 0.4616 (1.1560) 0.1854 (0.3519) -0.0680 (  0.1906) 0.0472 (0.2780) -0.0279 (  0.0496) -0.1230*** (  2.7438) -0.0624*** (  2.6177) -0.3730 (  0.9420) 0.1097 (0.4070) 0.2218 (0.8961) Yes 245 0.2181 -121.5699

Yes 248 0.0402 -150.3720

0.9847* (1.8743) 0.1527 (0.4913) 0.0530 (0.2100) 0.5288 (1.3915) 0.2310 (0.4644) 0.0250 (0.0710) -0.0157 (  0.0905) -0.0227 (  0.0408) -0.1126*** (  2.6826) -0.0624*** (  2.6218) -0.3914 (  1.0159) 0.1014 (0.4003) 0.2573 (1.0876) Yes 245 0.1966 -124.9193

7.9292** (2.4060) 0.9525* (1.8344) 0.1726 (0.5413) 0.0578 (0.2159) 0.4949 (1.2628) 0.2068 (0.3881) -0.0728 (  0.2025) 0.0449 (0.2655) 0.0179 (0.0322) -0.1152*** (  2.6658) -0.0619*** (  2.6030) -0.3640 (  0.9280) 0.1063 (0.3955) 0.2197 (0.8932) Yes 245 0.2160 -121.8939

Note: (1) Because of the low proportion of female governors, Gender and the samples in which Gender ¼ 1 are automatically excluded in the calculation process; (2) Robust z-statistics in parentheses; (3)*** denotes p o 0.01,** denotes p o 0.05,* denotes p o 0.1.

108

X. Chen et al. / Energy Policy 95 (2016) 103–112

at the 1% and 5% levels, respectively; the significant degree of the last estimated coefficient, in which the p-value is 0.104, is slightly lower. Thus, it can be seen that there is a significantly positive relationship between the average energy productivity of tenure and Chinese local officials’ promotions; the higher the average energy productivity of tenure is, the greater is the probability that the governor is promoted. We also found that the economic growth performance and some characteristic factors of provincial governors also have a significant impact on political promotions. The estimation coefficients for CGGDP are 1.7366 and 1.9141 but are not significant. However, the estimation coefficients for AGGDP are 7.6985 and 7.9292, and both are significant at the 5% level, which demonstrates that the average economic growth performance of tenure has a significant positive impact on governors’ promotions; however, the current economic growth performance does not affect the promotion significantly. Therefore, we concluded that the Chinese central government primarily references the long-term economic performance indicators to evaluate local officials. In addition, the probability of minority governors receiving promotions is significantly lower; being older and having longer tenure also significantly reduce the probability of promotion. It seems that there exist latent factors may affect both political promotion and energy productivity, which means there may exist endogeneity in our models. Moreover, endogeneity will result in some problems of estimation. In this paper, marketization degree2 is chosen as the instrumental variable (IV) for energy productivity, and an IV probit model is used to evaluate the endogenity problems. The results suggested that there is no sufficient information to reject the null hypothesis that energy productivity is exogeneity. It indicated that a regular probit regression model is equipped to address our research problem. Due to the space limitations, for details please refer to the Appendix. 4.3. Marginal effect of energy productivity on the probability of Chinese local officials’ promotions We believe that energy productivity has become an important assessment criterion for Chinese local officials’ promotions. How much effect does energy productivity have on Chinese local officials’ promotions? This is another issue that has also aroused our interest. To address this question, we estimated the average marginal effect (partial effect) of the energy productivity variables Prodenergy and AV _Prodenergy in all the regressions shown in Table 4. According to the estimation results reported in Table 5, we found that all the marginal effect has passed the significance test. Without considering any control variables, the average marginal effect of Prodenergy and AV _Prodenergy on the probability of Chinese local officials’ promotions was 1.1444 and 1.2565, respectively. When the province effect and other control variables were considered, the average marginal effect of Prodenergy was 0.7828 and 0.6164, respectively, and the average marginal effect of AV _Prodenergy was 0.7570 and 0.5900, respectively. As shown in Table 5, the average marginal effect of energy productivity was only smaller than the average economic growth of tenure, but it is larger than the other factors. 4.4. Expanded analysis: does the effect of energy productivity on the Chinese local officials’ promotions exist throughout the entire period of 1978–2012? In this paper, empirical evidence demonstrates that energy productivity does affect local officials’ promotions in China. 2

We tested other IVs, such as energy price, but the result was not satisfactory.

However, does the effect of energy productivity on Chinese local officials’ promotions exist throughout the entire period of 1978– 2012? To answer this question, we chose 1990 as the cut-off point and divided the entire research period into two periods: 1978– 1989 and 1990–2012 and employed the periods to re-estimate the probit models. Selecting 1990 as the cut-off point is based primarily on two considerations. First, in approximately 1990, China began to rethink the resource-driven economic development model and began a transformation. Since then, the Chinese government has unveiled a series of policies, regulations, notifications and documentations about energy saving, emission reduction and environmental protection, such as “The decision of the State Council on further strengthening environmental protection work (1990)”,“Chinese power law (1994) ”, “Arrangement for national energy conservation week (1994)”, “Law of the People’s Republic of China on conserving energy (1997)”, and so on. In 1995, in particular, the 5th Plenary Session of the 14th CPC Central Committee formally proposed that the Chinese economic growth mode must be transformed from an extensive to an intensive growth mode. Second, Fig. 1 also demonstrates that the growth trend of China’s energy productivity after 1990 was more pronounced and that 1990 was likely to be a turning point in China’s energy policy. We utilized the data during the first period of 1978–1989 for estimation, and the results are presented in Table 6, column (1) to column (4). The data indicate that: (1) The estimation coefficients for CGGDP are 5.4545 and 5.5174; however, none of them are significant. The estimation coefficients for AGGDP are 13.6249 and 13.4003, which are significant at the 5% level. Since the 3rd Plenary Session of the 14th CPC Central Committee formally proposed in 1978 to give economic construction priority, over the next ten years, longterm economic performance indicators such as the average growth of real GDP by region of local officials’ entire tenure had important reference values for the selection and promotion of local officials. This finding also supports the “Official Promotion Tournament Theory” proposed by Li and Zhou (2005). (2) The estimation coefficients for Prodenergy are 4.9606 and 1.8967; and the estimation coefficients for AV _Prodenergy are 6.0966 and 3.0072, but none of them are significant, indicating that there is no significant correlation between energy productivity and governors’ promotions. Neither the current energy productivity nor the average energy productivity of tenure were influencing factors in Chinese local officials’ promotions, so they played no role on increasing the probability of promotions. It could be seen that at the beginning of Chinese reform and opening up, economic growth performance had become an important assessment indicator that contributed to officials’ promotions. However, under the background of “it does not matter whether the cat is black or white, as long as it catches mice”, the goal of the Chinese central government to pursue rapid economic growth was implemented in the political system from top to bottom, so economic growth naturally became the major event that Chinese local officials were concerned about, and the costs to achieve rapid economic growth were not deemed important. Therefore, we believe that the Chinese central government did not consider energy conservation in the evaluation system of local officials’ promotions during the period 1978–1989. Then, we utilized the data for the 2nd period, 1990–2012, to estimate our empirical models, and the results are presented in Table 6, column (5) to column (8). The estimation in column (5) to column (8) demonstrates that the regression coefficients for Prodenergy and AV _Prodenergy are 2.9271 and 2.6952 and 3.0323 and

X. Chen et al. / Energy Policy 95 (2016) 103–112

109

Table 5 Average marginal effect(partial effect) of the current energy productivity and the average energy productivity of tenure. (1)

(2)

(3)

(4)

(5)

(6)

Variables

dy/dx

dy/dx

dy/dx

dy/dx

dy/dx

dy/dx

Prodenergy

1.1444***

0.7828**

0.6164*

(3.6296)

(2.4032)

(1.8630)

AV _Prodenergy CGGDP

0.7570**

0.5900*

(3.6564)

(2.1655) 0.5485 (0.7861)

(1.6471)

0.4955 (0.7041)

AGGDP Nationality Dum_edu2 Dum_edu3 Dum_edu4 Dum_edu5 Abroad Engineer Economist Tenure Age Dum_65 Youth Central Observations

1.2565***

248

0.2788* (1.9000) 0.0377 (0.4247) 0.0080 (0.1094) 0.1398 (1.2670) 0.0596 (0.4226) 0.0089 (0.0890) -0.0031 (  0.0619) -0.0217 (  0.1351) -0.0349*** (  2.8673) -0.0179*** (  2.8023) -0.1147 (  1.0258) 0.0301 (0.4167) 0.0742 (1.1046) 245

2.1443** (2.3912) 0.2637* (1.8587) 0.0432 (0.4849) 0.0101 (0.1329) 0.1286 (1.1484) 0.0516 (0.3498) -0.0190 (  0.1903) 0.0131 (0.2778) -0.0078 (  0.0496) -0.0343*** (  2.8035) -0.0174*** (  2.7987) -0.1039 (  0.9415) 0.0306 (0.4086) 0.0618 (0.9058) 245

248

0.2822* (1.9162) 0.0438 (0.4875) 0.0152 (0.2103) 0.1515 (1.3967) 0.0662 (0.4613) 0.0072 (0.0710) -0.0045 (  0.0905) -0.0065 (  0.0408) -0.0323*** (  2.7438) -0.0179*** (  2.8044) -0.1122 (  1.0096) 0.0291 (0.4019) 0.0737 (1.1052) 245

2.2148** (2.4641) 0.2660* (1.8719) 0.0482 (0.5368) 0.0162 (0.2158) 0.1382 (1.2543) 0.0578 (0.3856) -0.0203 (  0.2022) 0.0125 (0.2652) 0.0050 (0.0322) -0.0322*** (  2.7173) -0.0173*** (  2.7854) -0.1017 (  0.9269) 0.0297 (0.3969) 0.0614 (0.9032) 245

Note:(1)*** denotes p o 0.01,** denotes p o 0.05,* denotes p o 0.1;(2) z-statistics in parentheses;(3) dy/dx for factor levels is the discrete change from the base level.

2.8145, respectively, and all of them are significant at the 5% or 10% levels. By contrast, the regression coefficients for GGGDP and AGGDP are all not significant, this indicates that during 1990–2012, energy productivity had a significant positive impact on governors’ promotions whereas economic growth did not. In other words, the effect of economic performance on Chinese local officials’ promotions began to weaken while the effects of energy productivity began to become prominent. For the first decade of reform and opening-up, the Chinese central government had to stress that “development is the last word” and “development is the top priority”. However, with the continual developing of the Chinese economy, energy crisis and environmental pollution become increasingly serious. The Chinese government gradually realized the importance of resource conservation, environmental protection and sustainable development. Therefore, the Chinese central government gradually adjusted the official evaluation mechanism, which overemphasized pure economic growth performance, and introduced the energy productivity indicators, which partly introduce the quality of economic growth in the performance evaluation system of local government officials. In summary, energy productivity has become an important influence factor for Chinese local officials’ promotions. However, the effect of energy productivity on Chinese local officials’ promotions is not intrinsic. It was gradually introduced in the dynamic process of transformation of the economic growth mode and adjustment of local officials’ assessment mechanism.

5. Conclusions and policy implications In this paper, we utilized the data of provincial governors whose position changed in China during 1978–2012 and adopted probit regression models, empirically examined the correlation between energy productivity and local officials’ promotions. The empirical results demonstrate that (1) both the short-term and long-term energy productivity performance had a significantly positive impact on Chinese local officials’ promotions; and (2) the effect of energy productivity on Chinese local officials’ promotions was not intrinsic. It was gradually introduced in the dynamic process of the transformation of the economic growth mode and adjustment of local officials’ assessment mechanism. Chinese energy productivity has been continuously improved in the past three decades. In order to further improve energy productivity and achieve sustainable development, our observations suggested that Chinese central government should further reform its local officials’ governance mechanism, specifically, the performance evaluation system in promotion incentives. In such circumstances, Chinese local officials should pursue the quality of economic growth and attach importance to the green growth indicators such as energy productivity. Some of the literature attempted to explain China's economic growth from the perspective of local officials (Li and Zhou, 2005) and found that Chinese local officials mainly competed for regional economic growth for their political promotions. In this paper, we examined the influence of energy productivity on Chinese local officials’ promotions and found that a better energy productivity performance contributed to provincial

110

X. Chen et al. / Energy Policy 95 (2016) 103–112

Table 6 Energy productivity and Chinese local officials’ promotions: test results based on two periods. Variables

Year:1978–1989

Year:1990–2012

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

2.9895 (0.9402) 4.9606

3.2135 (0.9946) 1.8967

2.7213 (0.8649)

2.9872 (0.9310)

2.4928 (1.0372) 2.9271**

2.2848 (0.9542) 2.6952*

2.3436 (0.9869)

2.1177 (0.8940)

(1.0392)

(0.3936)

(1.9958)

(1.8479)

AV _Prodenergy

6.0966

3.0072

3.0323*

2.8145*

(0.6158)

(1.8817) 0.0643 (0.0244)

(1.7452)

CGGDP

(1.2899) 5.5174 (1.1999)

Constant

Prodenergy

5.4545 (1.1756)

AGGDP

13.6249** (2.5498)

-0.2328 (  0.0861) 13.4003** (2.5234)

Nationality Dum_edu2 Dum_edu3

0.4182 (0.7098) -0.4885 (  1.1191)

0.6261 (1.0105) -0.4281 (  0.8947)

0.4299 (0.7199) -0.5002 (  1.1324)

0.6242 (0.9901) -0.4451 (  0.9264)

-0.2440** (  2.3930) -0.0620 (  1.1303) -0.3567 (  0.5954) 0.2136 (0.6977) -0.2418 (  0.5408) yes 91 0.1744 -37.3633

-0.2664*** (  2.5780) -0.0753 (  1.3208) -0.1332 (  0.2192) 0.2424 (0.7613) -0.3070 (  0.7521) yes 91 0.2368 -34.5417

-0.2465** (  2.4499) -0.0591 (  1.0848) -0.3791 (  0.6300) 0.2229 (0.7228) -0.2352 (  0.5251) yes 91 0.1788 -37.1676

-0.2689*** (  2.6146) -0.0725 (  1.2781) -0.1621 (  0.2663) 0.2498 (0.7850) -0.3009 (  0.7321) yes 91 0.2385 -34.4644

Dum_edu4 Dum_edu5 Abroad Engineer Economist tenure Age Dum_65 Youth Central Province-effect Observations Pseudo R2 Log pseudolikelihood

0.9998* (1.7980) 0.6384 (1.0308) 0.3599 (0.5418) 0.9724 (1.3261) 0.6361 (0.8207) -0.3653 (  1.0068) 0.1048 (0.4403) -0.0938 (  0.1609) -0.0651 (  1.2408) -0.0759* (  1.7558) -0.6582 (  1.1038) 0.0041 (0.0126) 0.6921* (1.7934) yes 142 0.1991 -77.2001

2.4337 (0.4900) 0.9743* (1.7576) 0.6401 (1.0414) 0.3592 (0.5421) 0.9478 (1.2774) 0.6214 (0.7860) -0.4264 (  1.1786) 0.1250 (0.5183) -0.0734 (  0.1284) -0.0645 (  1.2380) -0.0761* (  1.7417) -0.6416 (  1.0358) -0.0061 (  0.0186) 0.6615* (1.7125) yes 142 0.2004 -77.0735

0.9982* (1.7899) 0.6833 (1.1010) 0.3901 (0.5846) 0.9824 (1.3338) 0.6696 (0.8513) -0.3842 (  1.0464) 0.0960 (0.4041) -0.1001 (  0.1767) -0.0624 (  1.1633) -0.0743* (  1.7350) -0.6492 (  1.0866) 0.0143 (0.0442) 0.6882* (1.7703) yes 142 0.1981 -77.2963

2.8605 (0.5831) 0.9720* (1.7501) 0.6809 (1.1011) 0.3848 (0.5756) 0.9535 (1.2765) 0.6476 (0.8047) -0.4412 (  1.2016) 0.1210 (0.5033) -0.0819 (  0.1478) -0.0623 (  1.1717) -0.0744* (  1.7164) -0.6440 (  1.0320) 0.0029 (0.0089) 0.6561* (1.6798) Yes 142 0.2000 -77.1135

Note: (1) Because of the low proportion of female governors, Gender and samples in which Gender ¼ 1 are automatically excluded in the calculation process; (2) during 1978– 1989, almost no governor was of a minority, or was an engineer or economist, and no governor had the experience of studying abroad, so Nationality, Abroad, Engineer and Economist are all automatically excluded in the calculation process during this period; (3) Robustz-statistics in parentheses; (4)*** denotes p o0.01,** denotes p o0.05,* denotes p o 0.1.

governors’ promotions, which helps provide a better understanding of Chinese energy productivity improvements from the perspective of local officials’ governance. Does energy productivity performance affect the promotion of other levels of local officials, such as mayors? This question will be studied further in the future.

Acknowledgements The authors thank two anonymous referees and an editor of this journal for their valuable comments. This work is partially supported by National Natural Science Foundation of China under Grant 71402103 and 71403060; the Foundation for Distinguished Young Talents in Higher Education of Guangdong, China, under grants 2012WYM_0116 and 2013WYM_ 0014; the Philosophy and Social Science Foundation of Guangdong, China, under grant GD13YGL01; the MOE Youth Foundation Project of Humanities and Social Sciences at Universities in China under grants 13YJC630123 and 14YJC630017; China Postdoctoral Science Foundation Funded Project under grant 2015M580053.

Appendix Discussing the endogeneity problem Given the complex nature of energy markets and the political system in China, it seems reasonable to think that latent factors may affect both political promotion and energy productivity, which may lead to endogeneity in our probit models. Instrumental Variables (IV) is a popular technique of estimation that is widely used when the correlation between the explanatory variables and the regression error term is suspected. In this paper, we try to find a valid instrumental variable (IV) for energy productivity and estimate IV probit models to test whether there is a significant endogenous bias. Existing studies suggest that market-oriented economic reforms contribute to energy efficiency improvement (Fisher-Vanden, 2003; Fan, Liao and Wei, 2007). Hence, we believe that marketization degree is considered to be correlated with energy productivity. Moreover, China adopted a gradual marketoriented reform strategy, and the marketization degree showed significant non-equilibrium in different regions (Fan, Wang and Zhou, 2011). However, this situation is dominated by the central

X. Chen et al. / Energy Policy 95 (2016) 103–112

government, rather than local officials, which leads us to conjecture that there is no direct link between the marketization degree and Chinese local officials’ promotions. To this end, the marketization degree is chosen as the IV for energy productivity. Marketization index proposed by Fan, Wang and Zhou (2011) is widely used to measure marketization degree (Jiang and Huang, 2011; Zhang and Jin et al., 2013). However, the data of marketization index are only reported from 1997 to 2009, and does not fully cover the time period of our study. Generally, a larger sample size is good for enhancing the robustness of the estimation results. A new estimation of marketization index is proposed by Wei, Wang and Wang (2014) and they presented the data of marketization index from 1985 to 2010. In this paper, the method is employed to estimate the marketization index from 1978 to 1984 and from 2011 to 2012. As a result, we constructed the database of marketization degree from 1978 to 2012. The IV probit models are utilized to evaluate the endogenous bias via the ivprobit command in Stata. The maximum likelihood estimation (mle) is adopted to obtain the results. Table 7 reports the estimation results of the effect of energy productivity on Chinese local officials’ promotions based on the IV probit models. The estimation results of first-step regressions of IV probit models demonstrate that the instrument is a significant predictor of energy productivity. This is because the coefficients of instrument are

111

all significantly positive at the 0.001 level. To further examine the reasonableness of the instrument, the models are also estimated using a linear probability framework to take advantage of the tests of the assumptions underlying the IV model that are available in the ivreg2 command in Stata. The results show that (1) underidentification test results show that the Kleibergen-Paap rk LM statistics are significant at the 0.0000 level; and (2) weak identification test results show that the values of the Cragg-Donald Wald F statistics are higher than those of Stock-Yogo weak ID test critical value at 10% significance level (Stock and Yogo, 2005). The test results show that marketization degree affected energy productivity strongly, indicating using marketization degree as the instrument for energy productivity should not be weak identification. Moreover, we estimate a new regression model, in which marketization degree, energy productivity and other control variables are all included. Results show that the coefficients of energy productivity are all significantly positive while the coefficients of instrument (marketization degree) are insignificant, meaning the marketization degree is exogenous. From the analysis above, it is creditable that marketization degree can be utilized as a valid IV for energy productivity. From Table 7, the estimation results indicate that the coefficients of Prodenergy are 3.2088, 1.2275 and 0.7323, and the first two coefficients are significant at 5% and 10% levels, respectively.

Table 7 Energy productivity and Chinese local officials’ promotions: test results based on IV probit models. Variables

(1)

(2)

(3)

(4)

(5)

(6)

Constant

-0.9005*** (  4.5297) 3.2088**

-0.0197 (  0.0091) 1.2275*

-2.5467 (  0.9501) 0.7323

-0.9492*** (  4.4030)

0.0337 (0.0157)

-2.5296 (  0.9510)

(2.5071)

(1.7452)

(1.4469) 3.7231**

1.4692*

0.8834

(2.5120)

(1.7593) 1.3261 (1.0162)

(1.4557)

Prodenergy AV _Prodenergy CGGDP

1.4493 (1.1009)

AGGDP Nationality Dum_edu2 Dum_edu3 Dum_edu4 Dum_edu5 Abroad Engineer Economist Tenure Age Dum_65 Youth Central Province effect Observations Log pseudolikelihood Wald χ2 test of exogeneity Exogeneity test Wald p-value

Yes 248 194.3339 0.0899 0.7643

1.0696** (2.1307) 0.1480 (0.4032) 0.1117 (0.3815) 0.5415 (1.4180) 0.4172 (0.7192) 0.2822 (0.8687) -0.0596 (  0.2480) -0.0899 (  0.1547) -0.1220** (  2.0775) -0.0498** (  1.9840) -0.4401 (  1.2929) 0.2811 (1.2725) 0.2610 (1.0419) Yes 245 227.5724 0.7221 0.3955

3.4116* (1.8690) 1.0741** (2.2339) 0.1183 (0.3237) 0.1351 (0.4580) 0.5654 (1.4705) 0.4999 (0.8384) 0.2179 (0.6670) -0.0325 (  0.1371) -0.0432 (  0.0766) -0.1222** (  2.0734) -0.0447* (  1.7579) -0.4440 (  1.3041) 0.2866 (1.3085) 0.2692 (1.0663) Yes 248 228.8142 1.2751 0.2587

Yes 245 220.8787 0.0125 0.9111

1.0484** (2.0820) 0.1624 (0.4403) 0.1127 (0.3828) 0.5280 (1.3588) 0.3789 (0.6424) 0.2850 (0.8692) -0.0670 (  0.2791) -0.0748 (  0.1325) -0.1120** (  1.9826) -0.0487* (  1.9454) -0.4451 (  1.2936) 0.2392 (1.0657) 0.2624 (1.0596) Yes 248 244.9268 0.8092 0.3684

3.3368* (1.8551) 1.0552** (2.1676) 0.1448 (0.3954) 0.1465 (0.4921) 0.5625 (1.4340) 0.4863 (0.7973) 0.2171 (0.6530) -0.0462 (  0.1943) -0.0384 (  0.0701) -0.1159** (  2.0347) -0.0438* (  1.7189) -0.4463 (  1.2944) 0.2511 (1.1254) 0.2714 (1.0839) Yes 245 246.1912 1.4863 0.2228

Note: (1) Because of the low proportion of female governors, Gender and the samples in which Gender ¼ 1 are automatically excluded in the calculation process; (2) Robust z-statistics or z-statistics in parentheses; (3)*** denotes p o 0.01,** denotes p o 0.05,* denotes p o0.1

112

X. Chen et al. / Energy Policy 95 (2016) 103–112

Meanwhile, the coefficients of AV _Prodenergy are 3.7231, 1.4692, and 0.8834, and the first two of which are significant at the 5% and 10% levels, respectively. More importantly, the Wald χ2 test results of all the IV probit models are not significant. It is suggested that there is no sufficient information to reject the null hypothesis that energy productivity is exogeneity. Thus a regular probit regression may be appropriate.

References Ai, C., Norton, E.C., 2003. Interaction terms in logit and probit models. Econ. Lett. 80 (1), 123–129. Bardhan, P., 2006. Awakening giants, feet of Clay: A Comparative Assessment of the Rise of China and India. J. South Asian Dev. 1 (1), 1–17. Bo, Z., 1996. Economic performance and political mobility: Chinese provincial leaders. J. Contemp. China 5 (12), 135–154. BP. Statistical Review of World Energy, 2011. Chai, J., Guo, J.E., Wang, S.Y., Lai, K.K., 2009. Why Does Energy Intensity Fluctuate in China? Energy Policy 37 (12), 5717–5731. Chang, T.P., Hu, J.L., 2010. Total-factor energy productivity growth, technical progress, and efficiency change: An empirical study of China. Appl. Energy 87 (10), 3262–3270. Chen, Y., Li, H., Zhou, L.A., 2005. Relative performance evaluation and the turnover of provincial leaders in China. Econ. Lett. 88 (3), 421–425. China State Statistical Bureau. China Statistical Yearbook, 1981–2013. China State Statistical Bureau. China Energy Statistical Yearbook, 1987–2013. China State Statistical Bureau. China Compendium of Statistics 1949–2004, 2005. China State Statistical Bureau. Data of Gross Domestic Product of China 1952–2004, 2007,. Dimitropoulos, J., 2007. Energy productivity improvements and the rebound effect: An overview of the state of knowledge. Energy Policy 35 (12), 6354–6363. Elliott, R.J., Sun, P., Chen, S., 2013. Energy intensity and foreign direct investment: A Chinese city-level study. Energy Econ. 40, 484–494. Fan, Y., Liao, H., Wei, Y.M., 2007. Can market oriented economic reforms contribute to energy efficiency improvement? Evidence from China. Energy Policy 35 (4), 2287–2295. G., Fan, X. L., Wang, H. P., Zhou (2011). NERI INDEX of Marketization of China’s Provinces 2011 Report. Beijing, Economic Science Press. Fisher-Vanden, K., 2003. The Effects of Market Reforms on Structural Change: Implications for Energy Use and Carbon Emissions in China. Energy J. 24 (3), 27–62. Fisher-Vanden, K., Jefferson, G.H., Liu, H., Tao, Q., 2004. What Is Driving China's Decline in Energy Intensity? Resour. Energy Econ. 26 (1), 77–97. Fisher-Vanden, K., Jefferson, G.H., Jingkui, M., Jianyi, X., 2006. Technology development and energy productivity in China. Energy Econ. 28 (5), 690–705. Feng, T., Sun, L., Zhang, Y., 2009. The relationship between energy consumption structure, economic structure and energy intensity in China. Energy Policy 37 (12), 5475–5483. Feng, Y., Wu, C.F., 2013. Is economic performance important for Chinese official promotion? J. Manag. Sci. China 16 (11), 55–68. Garbaccio, R.F., Ho, M.S., Jorgenson, D.W., 1999. Why has the Energy-Output Ratio

Fallen in China? Energy J., 63–91. Han, Z.Y., Fan, Y., Jiao, J.L., Yan, J.S., Wei, Y.M., 2007. Energy Structure, Marginal Efficiency and Substitution Rate: An Empirical Study of China. Energy 32 (6), 935–942. Hang, L., Tu, M., 2007. The impacts of energy prices on energy intensity: Evidence from China. Energy Policy 35 (5), 2978–2988. He, H.S., 2003. Official Records of the People’s Republic of China. China Society Press,, Beijing. Hu, J.L., Wang, S.C., 2006. Total-factor energy efficiency of regions in China. Energy Policy 34 (17), 3206–3217. Jiang, F.X., Huang, J.C., 2011. Managerial Incentive, Debt and Firm Value. Econ. Res. J. 5, 46–60. Li, H., Zhou, L.A., 2005. Political turnover and economic performance: the incentive role of personnel control in China. J. public Econ. 89 (9), 1743–1762. Liao, H., Fan, Y., Wei, Y.M., 2007. What induced China’s Energy Intensity to Fluctuate: 1997–2006? Energy Policy 35 (9), 4640–4649. Lin, B., Du, K., 2014. Decomposing energy intensity change: A combination of index decomposition analysis and production-theoretical decomposition analysis. Appl. Energy 129, 158–165. Ma, C., Stern, D.I., 2008. China’s changing energy intensity trend: A decomposition analysis. Energy Econ. 30 (3), 1037–1053. Miketa, A., Mulder, P., 2005. Energy productivity across developed and developing countries in 10 manufacturing sectors: patterns of growth and convergence. Energy Econ. 27 (3), 429–453. Patterson, M.G., 1996. What is energy efficiency?: Concepts, indicators and methodological issues. Energy Policy 24 (5), 377–390. Sinton, J.E., Levine, M.D., 1994. Changing energy intensity in Chinese industry: The relatively importance of structural shift and intensity change. Energy Policy 22 (3), 239–255. Stock, J.H., Yogo, M., 2005. Testing for weak instruments in linear IV regression. In: Andrews, D.W., Stock, J.H. (Eds.), Identification and Inference for econometric Models: Essays in Honor of Thomas Rothenberg. Cambridge University Press, pp. 80–118. Su, F., Tao, R., Xi, L., Li, M., 2012. Local officials’ incentives and China’s economic growth: Tournament thesis reexamined and alternative explanatory framework. China World Econ. 20 (4), 1–18. Wan, X., Ma, Y., Zhang, K., 2015. Political determinants of intergovernmental transfers in a regionally decentralized authoritarian regime: evidence from China. Appl. Econ. 47 (27), 2803–2820. Wang, C., 2011. Sources of energy productivity growth and its distribution dynamics in China. Resour. Energy Econ. 33 (1), 279–292. Wei, Q., Wang, A., Wang, J., 2014. The Rise of China's coastal Areas: Power of Market. Econ. Res. J. 8, 170–183. Xu, C., 2011. The fundamental institutions of China’s reforms and development. J. Econ. Lit., 1076–1151. Zhuravskaya, E.V., 2000. Incentives to provide local public goods: fiscal federalism, Russian style. J. Public Econ. 76 (3), 337–368. Zhou, L.A., 2004. The Incentive and Cooperation of Government Officials in the Political Tournaments: An Interpretation of the Prolonged Local Protectionism and Duplicative Investments in China. Econ. Res. J. 6, 33–40. Zhang, J., Gao, Y., 2008. Term Limits and Rotation of Chinese Governors: Do They Matter to Economic Growth? J. Asia Pac. Econ. 13 (3), 274–297. Zhang, Z., Jin, X., Yang, Q., Zhang, Y., 2013. An empirical study on the institutional factors of energy conservation and emissions reduction: Evidence from listed companies in China. Energy Policy 57, 36–42.