The Political Economy of Noncompliance in China: the case of industrial energy policy

Abstract

One of the greatest challenges facing China today is the central government's ability to ensure that policies are implemented effectively at the local level, particularly policies that seek to make China's economic growth model more sustainable. These policies face resistance from local authorities and enterprises that benefit from the status quo. This raises a key research question: why do some provinces more fully implement these central policies? We argue the extent of local implementation is best conceptualized as a rational balance between economic and political incentives: localities with regulatory autonomy, low regulatory capacity and alternative interests will not fully implement policies that are at odds with local economic imperatives. By examining a critical case of central policy implementation—industrial energy intensity reduction in the eleventh five-year plan—this article demonstrates that, regardless of industrial makeup or economic development, provinces that have greater regulatory autonomy for noncompliance coupled with alternative economic interests do not, on average, perform as well. Using a nested analysis approach this study illustrates this argument with both quantitative analysis and original case study evidence from fieldwork interviews.

Introduction

One of the greatest challenges facing China today is the central government's ability to ensure that policies are implemented effectively at the local level. This is particularly true of policies that seek to shift China's economy from investment-led growth to a more environmentally sustainable and socially equitable model based on domestic consumption and high value-added production.1 Such policy shifts often face stiff resistance from local authorities and enterprises that benefit from the status quo.2 As the government prepares new rounds of economic reforms, this raises a key research question: why do some provinces more fully implement central government policies? This article argues that the extent of local implementation is best conceptualized as a function of three political and economic factors: regulatory autonomy, regulatory capacity and economic interests. By examining a critical case of central policy implementation—energy intensity (EI) reduction in the eleventh five-year plan (FYP)—this article demonstrates that, regardless of industrial makeup or economic development, provinces that have greater regulatory autonomy coupled with low regulatory capacity or alternative economic interests do not, on average, perform as well.

Empirically, this study demonstrates this argument using a nested analysis approach—combining a quantitative analysis of provincial success achieving industrial EI goals and case studies from three provinces: Shaanxi, Chongqing and Guangdong. This combines the advantages of each type of analysis; large-N statistical analyses have ‘the ability to simultaneously estimate the effects of rival explanations and/or control variables on an outcome of interest’, while small-N case studies can ‘answer those questions left open by the large-N analysis’.3 The case studies serve as a robustness check on the quantitative analysis, illuminate causality and highlight factors not operationalized quantitatively. This mixed methods approach is important in China research, as China specialists have endeavored to meet the social scientific benchmarks of other subfields.4

Autonomy, capacity, and interests for noncompliance

Policy implementation in China involves the interaction of actors at multiple levels with variation in both individuals' interests and regulatory autonomy. Conceptually, we explain policy implementation using a rationalist model of institutional change put forward by Douglass North and other new institutionalists. North explains institutional change based on the interaction of organizations and the political institutions that structure behavior.5 In this view, institutions are not determinative but simply incentivize organizations who themselves are important agents in shaping, transforming or ignoring rules.6 In China, the central government promulgates institutional rules, but implementation is often delegated to local levels, who themselves interact with firms and industry. Understanding the interaction between rule-makers and rule-takers thus becomes a central element in understanding policy implementation and institutional change.

While there has been a move towards greater fiscal centralization and formalization of policy making, China's political system remains decentralized and informal institutions remain important.7 Beijing can set the agenda, determine policy and decide fiscal allocations, but the provinces control policy implementation. Thus, the fundamental challenge is not the promulgation of policies and laws, but the extent to which they are implemented and enforced at the local level. Within this system, provinces vary in relative power, motivations and regulatory opportunities to comply with central policies. The result is the central problem in Chinese policy making—that localities often ignore or avoid compliance with central government regulations.

China's unique development has resulted in a fragmented policy structure, providing the institutional space, or regulatory autonomy, for noncompliance. During the reform era, China made effective use of existing, ‘second best’ institutions rather than pursuing Eastern European, ‘big bang’-style institution building.8 This ensured system-wide stability by minimizing big breaks and compensating ‘losers’.9 Pierson argues, due to path dependence, that as ‘the number of decisions made and the number of actors involved grow, relations of interdependence—among actors, organizations, and institutions—expand geometrically’.10 Thus, more actors have gained greater autonomy to influence policy outcomes, particularly in the economic realm.11 Under ‘fragmented authoritarianism’, central policies ‘become increasingly malleable to the parochial organizational and political goals of various vertical agencies and regions charged with enforcing that policy’.12 Sheng argues that this local fragmentation helps explain variation in the ability of the center to secure policy compliance at the provincial level.13

Additionally, institutionalized, decentralized policy experimentation or ‘proceeding from point to surface’ (youdian daomian), has been a central mechanism of policy implementation. In order to implement policies ‘in accordance with local conditions’ (yindi zhiyi), local officials set up ‘experimental points’ (shidian). Successful policies become ‘model experiences’ (dianxing jingyan) and are spread to wider geographic regions.14 Local governments' proximity to experiments means that they have more information and can coordinate more effectively than the central government.15 Experimentation under conditions of institutional fragmentation provides space for noncompliance.16 Although Beijing can intervene in provincial affairs and exerts control through the nomenklatura personnel system,17 there remains structural space for noncompliance with central policy. Even China's audit system, which checks that local agencies and firms are following central directives, allows for significant leeway in implementation, enforcement strategies and outcomes.18

Within this structure, provinces' importance in the policy process, particularly in economic matters, provides them with the resources to determine the regulatory capacity for compliance. Bo Zhiyue argues, ‘China is a country of provinces, and governing China is essentially governing China's provinces’.19 While tax reform in 1994 recentralized fiscal and political authority, it also more clearly delineated central–local relations, granting provinces significant control of policy implementation.20 This, in turn, drove interprovincial economic competition, further empowering provincial economic institutions.21 Provinces' importance in building regulatory capacity, based on their authority over local taxes and resource allocation, enables them to enforce compliance. This capacity varies depending on a province's ability and willingness to extract rents from local industries and the extent to which central regulatory structures supersede local institutional authority. While provinces vary in their institutional autonomy, high state capacity can also make the difference in better regulation. However, this is also contingent on a province's interests.

Economic and career incentives account for provincial interests in noncompliance. The phrase ‘shang you zhengce, xia you duice’, or ‘the higher-ups have policies, the lower levels have countermeasures’, captures the friction between central and local governments. Lampton calls this implementation bias: ‘every central initiative will be distorted in favor of the organization or locality responsible for implementation’.22 For example, local governments ‘take social, political, and economic considerations into account’ when determining industrial regulatory policy.23 As a result, local agencies often have to please local officials rather than carry out their institutional mandate.24

The most immediate concern for local officials is how they are scored in the cadre evaluation (ganbu pingjia) system. Typically, during the 11th FYP, economic indicators accounted for 60–65 points out of 100, skewing the processes in favor of bolstering economic performance.25 Lynette Ong explains:

performance targets (kaohe zhibiao) in the evaluation criteria are ranked in terms of their significance, from ‘priority targets with veto power’ (yipiao foujue) and ‘hard targets’ (ying zhibiao) to ‘soft or ordinary targets’ (yiban zhibiao) …. Scoring high on ‘hard targets’ … is what really makes or breaks local cadres' careers.

Hard targets are consistent across regions and include revenue, industrial production and investment targets. Cadres that perform well on quantifiable, hard targets get promoted faster and receive higher year-end bonuses, providing strong career incentives to not comply with directives—such as those related to environmental protection—that compromise economic performance.26

Additionally, local governments rely on enterprises to fund government programs and provide stability and favor local enterprises in order to bolster tax revenue, extra-budgetary funds and employment. Larger firms have greater influence because they provide jobs and inputs for other sectors of the economy.27 Firms and the state are enmeshed in a mutually dependent relationship, even after privatization.28 Moreover, there are significant advantages for enterprises to cultivate relationships with government officials.29 These vested economic interests, accustomed to ‘living under the bureaucratic umbrella’, resist change.30 A survey of 200 Chinese officials and scholars in 2005 found that 50% believe China's economic and political reform is constrained by ‘some groups with vested economic interests’.31

Interregional competition also drives measures to support business in pursuit of extensive investment-led growth. Provincial authorities use a number of strategies, particularly in ‘sectors that are perceived as occupying the “commanding heights” of the economy’, including ‘appointments, approvals (pizhun), tax holidays, and interest-rate forgiveness’.32 Local governments even ‘contribute generously to [industrial] groundwork—grading, infrastructure buildout, and clustering related sectors of the value chain’.33 Naughton argues, ‘proinvestment and progrowth policies are built into the Chinese system in an extremely profound way’.34 Local governments and industry enjoy a symbiotic relationship, and this shapes the practical, economic and career interests of local governments, subverting policies that challenge low-cost economic performance.

China's decentralized policy process means that provinces have both significant authority over economic policy and the regulatory autonomy to influence policy outcomes. Additionally, local economic interests check central-level mandates for balanced growth and non-economic goals. Conversely, many regions comply due to limited autonomy vis-à-vis the central government, greater local regulatory capacity or a convergence of interests. As many scholars have argued, provincial leadership and agency matters,35 but the task for scholars of central–local relations is to parse out precisely how it impacts outcomes. Thus an important empirical puzzle remains: what explains variation in provincial compliance with policies that potentially threaten economic performance? Secondly, from a policy standpoint, how can institutional incentives be manipulated to ensure broader compliance of policies in the ‘public good’ such as environmental regulations?

A critical case: industrial energy intensity

Empirically, this article examines a critical policy to better understand the factors influencing local noncompliance: the surprise boom in industrial EI in 200236 and the central government's response. EI measures the energy efficiency of production; it is the amount of energy used per unit output GDP.37 Industrial EI reduction is a critical component of China's energy conservation and emissions reduction (ECER) policies. This is a critical test case for the central government's ability to shift to balanced growth. First, due to environmental protests,38 worries about energy security and the need for a stable energy supply, ECER policies are a priority; noncompliance is not simply the result of weak central emphasis. Second, strict EI reduction goals impose real, short-term adjustment costs on industry and the local governments that rely on them.39 In a 2011 survey by Johnson Controls, Chinese industrial enterprises cited insufficient return on investment, particularly in the short term, as the top barrier to investing in energy efficiency.40 Local preferences for low-cost production form an important impediment to implementation of ECER policies.41 Finally, ECER implementation is typical of Chinese policy implementation; locals are primarily responsible, face central government oversight, and have space for experimentation and adaptation.42

The EI boom: an externality of extensive growth

From 2002 to 2005, EI rose 5% per year, reversing a long-term downward trend.43 The simplest explanation for the boom is local government investment in infrastructure, which fueled growth in heavy industry via high demand for steel, gravel, cement and other energy intensive outputs.44 Industry is also a safe investment for local governments; it is unlikely to default, inputs for industry are relatively cheap, and local governments often do not have many other investment opportunities.45 Liao et al. find that the ‘excessive expansion of high-energy consuming sub-sectors and the high investment ratio were foremost sources of the increasing energy intensity’.46 The connection between local government interests and industrial EI growth is clear.

Beijing's response: upgrading experiments

In response, the central government embarked on a series of programs aimed at reducing EI, setting a goal of a 20% reduction in aggregate national EI by 2010. The National Development and Reform Commission's 2004 ‘Medium and Long-Term Plan for Energy Conservation’ set targets for industrial energy conservation.47 In 2005, the State Council proposed putting environmental and economic goals on the same level.48

Though top-level policy making structures action, local innovation serves as a source of policy ideas and change,49 and local implementation is the key to a policy's effectiveness.50 In response to the industrial EI boom, Beijing selected a successful 2003 energy efficiency experiment from Shandong Province for national implementation; the Top-1,000 Enterprises Program targets energy intensive enterprises for specific, planned reductions in energy consumption.51 This plan was accompanied by the Small Plant Closure Program, which aimed to shutter inefficient factories, among other programs.

The Top-1,000 Program highlights the importance of active local governments to policy success. According to Lawrence Berkeley National Laboratory,

The energy saving authorities of the province, district, or city are directed to collaborate with related organizations to lead and implement the Top-1,000 program, including the tracking, supervision, and management of the energy-saving activities of the enterprises.52

Managers of state-owned enterprises (SOEs) and local officials who did not meet targets via regulation would not be promoted, an important shift in the incentives for compliance.53 Acting on information from lower levels of government, Beijing selected successful policies for national implementation, left room for local adjustment and stipulated sanctions for noncompliance—seemingly a recipe for success.

Results of 11th FYP: mixed effectiveness

Overall, China reduced national EI 19.1% by the end of the 11th FYP in 2010 (although there may have been data manipulation).54 Reports on the success of industrial ECER programs have been generally positive at the national level.55 Secondary industry excluding construction has seen a dramatic 25% drop in EI across all provinces. Secondary industry includes ‘mining and quarrying, manufacturing, production and supply of electricity, water, and gas, and construction’.56 These changes are illustrated in Table 1.

Table 1 Secondary industry (minus construction) percentage change in EI 2005–2009

	China overall average: − 0.258
Inner Mongolia	− 0.495	Shanxi	− 0.288
Yunnan	− 0.455	Tianjin	− 0.286
Anhui	− 0.439	Gansu	− 0.283
Jilin	− 0.431	Hubei	− 0.275
Hunan	− 0.393	Chongqing	− 0.26
Shaanxi	− 0.386	Hebei	− 0.244
Guangxi	− 0.366	Shandong	− 0.242
Hainan	− 0.366	Jiangsu	− 0.206
Liaoning	− 0.364	Guangdong	− 0.18
Fujian	− 0.36	Heilongjiang	− 0.15
Guizhou	− 0.352	Xinjiang	− 0.128
Qinghai	− 0.35	Sichuan	− 0.125
Jiangxi	− 0.349	Shanghai
Ningxia	− 0.349	Zhejiang
Beijing	− 0.336	Tibet
Henan	− 0.311

Notes: Zhejiang, Shanghai and Tibet did not report the necessary data.

Source: Provincial Statistical Yearbooks, various years.

However, Table 1 also shows wide variation in the relative reduction in industrial EI across provinces, ranging from a drop of 49.5% in Inner Mongolia to 12.5% in Sichuan. Also, reported reductions may not have entirely been the result of competent policy implementation. There are reports of officials distorting data, not reporting consumption and using temporary factory shutdowns to influence outcomes.57 Some factories used ‘off-grid’ diesel generators to reduce measured energy consumption.58 Also, monitoring programs were weak59 and much of the reduction in EI is attributable to the global economic downturn.60 Thus, though there may have been a national reduction in industrial EI, there are many loopholes in the regulatory system and policy implementation across provinces appears to be varied.

Explaining provincial variation: testing the political economy model

Using econometric analysis, this section illustrates how variation in regulatory autonomy, local regulatory capacity and local interests at the provincial level explain variation in provincial EI reduction. Regardless of a province's economic development or industrial makeup, provinces with greater autonomy from the central government, less regulatory capacity and more incentives to aid inefficient local enterprises performed, on average, more poorly in reducing industrial EI over this period.

Methodology

We first conducted large-N regression analysis of newly aggregated data for the period 2005–2009. Using provincial data allows one to control for factors common across China. Two conditions must be met for sub-national analysis to yield clear results, both of which were illustrated above. First, the central government must exercise varying control over the provincial units' policy choices. Second, the central government must show a credible commitment to meeting its purported goals. Regulating industrial EI is a central goal for the CCP, making cross-provincial analysis the appropriate methodology for understanding variation in implementation.61

Since the government has an incentive to use statistics to bolster its legitimacy, Chinese data raise questions of reliability.62 For this article, statistics were pulled from the China Statistical Yearbook and Provincial Statistical Yearbooks published by China's National Bureau of Statistics. Each statistical yearbook follows the same data collection procedures and methodologies. This study uses data available across provinces and, to account for any unreported variation across provinces, most dependent variables are within-province ratios, for example, percentage of gross regional product (GRP) derived from industry. This methodology standardizes numbers across provinces. Overall, so long as interpretation of absolute levels takes into account potential biases, and the assumption that biases can be considered randomly distributed is valid, then analyzing government-generated data in relative terms remains useful for explaining variation across provinces.

The model

Based on the previous analysis, this section posits a stylized model:

(1)

where autonomy, capacity and interests are vectors for either one or a number of variables that captures the relationships of interest. The theoretical microeconomic foundations explained earlier guide variable selection and help ensure against simple endogeneity resulting from reverse causality and omitted variable bias.

Dependent variable: policy success (reduction of industrial EI)

This study examines the percent change in provincial industrial EI 2005–2009 highlighted in Table 1 as the dependent variable. The greater the reduction, the greater success a province had implementing central policy. Since industry is at the center of central–local policy friction and is the main focus of government regulation, industrial EI is a better test of the framework than aggregate EI, which includes energy usage and GRP from primary industry (farming and forestry), construction, tertiary industry (services) and society. Industrial energy use is also the main component of aggregate EI, as illustrated in Figure 1. Based on provincial statistical yearbook data, this study calculates industrial EI as the total standard coal equivalent (SCE) used by secondary industry (excluding construction) divided by the value of industrial output.

Figure 1 Energy consumption in China, 2000–2010. Source: Romankiewicz et al., Addressing the Effectiveness of Industrial Energy Efficiency Incentives in Overcoming Investment Barriers in China.

Independent variables: the autonomy, capacity and interests for noncompliance

The study includes explanatory variables that most closely approximate our model of EI policy implementation. Based on fieldwork interviews and the available data, we settled on the explanatory variables listed in Table 2. The study measures the independent variables as the average value between 2005 and 2009 and specifies a model that uses the base year 2005 as a robustness check.

Table 2 Explanatory variables

Variable	Average value	Standard deviation	Minimum value (province)	Maximum value (province)
SOE presence	0.460	0.179	0.131 (Jiangsu)	0.784 (Gansu)
Regulatory capacity	0.080	0.019	0.054 (Henan)	0.150 (Beijing)
LME presence	0.682	0.090	0.518 (Hunan)	0.820 (Xinjiang)
Provincial CC index	7.768	1.680	6 (Hebei, Jilin, Fujian, Hunan, Hainan, Qinghai and Ningxia)	13 (Xinjiang)
Industrial share of GRP	0.398	0.077	0.178 (Hainan)	0.508 (Shandong)

The study's measurements of regulatory autonomy are twofold. First, the study measures the ability of the center to directly regulate industry based on the proportion of SOEs in a province. Due to central ownership and control of management, there is less provincial government latitude in policy implementation.63 SOE managers are given numerical goals and face direct pressure from the central government.64 This has the outward effect of improving local regulatory capacity by reducing the local regulatory burden, but also means that local institutional authority is superseded by the central government. Thus, a higher proportion of SOEs should correlate with greater EI reduction, consistent with central preferences. The study measures this as the ratio of output derived from SOEs divided by the total industrial output of the province, and averages the ratio from 2005 and 2009.

Second, the study measures regulatory autonomy for noncompliance using Bo Zhiyue's Central Committee Index, which serves as a powerful proxy for varying provincial autonomy in a decentralized system.65 Bo uses provincial representation on the Central Committee (CC) to show relative provincial power, weighing officials by their standing in the central government.66 His index also weights Politburo membership more heavily, an approach other authors have emphasized.67 Overall, Sheng argues that the center increased its control of the provinces from 1978–2002,68 although the share of provincial seats increased significantly again during the period covered in this study.69 Previous studies of representation have focused on the ability of provinces to extract resources from the center, with inconclusive results.70 This article differs from these studies in conceptualizing power not as a mechanism for resource extraction, but as autonomy from central control, thus distinguishing between recentralization of fiscal authority and decentralization of decision-making power.71 Relative provincial bargaining power vis-à-vis the central government provides indirect benefits in sheltering provinces from additional scrutiny. Provinces with greater autonomy will likely perform more poorly than those that do not have the opportunity to shirk responsibilities.

The study measures provincial regulatory capacity as tax revenue as a percentage of GRP, a commonly used approach.72 This captures both the local government's fiscal resources and its ability to extract from and administer firms, thereby controlling firm behavior.73 Since one method local governments use to support extensive industrial growth is reducing tax burdens, more tax revenue indicates both greater capacity to finance regulatory institutions and a willingness to control industry.74 Those areas better able to collect taxes should thus be able to force greater compliance.

There are two possible paths to quantify provincial interests: intuitively, industry may present a vested interest prompting provincial authorities to shield it from increased regulation. The model includes industrial GRP, the percentage of provincial GRP that comes from secondary industry (excluding construction), to measure the relative importance of industry in provincial economies. If the existence of industry in a region stymies policy implementation, then this measure would be significant and negative.

However, secondary industry does not necessarily equate to the presence of strong or obvious vested interests. Rather, the proportion of large and medium industrial enterprises (LMEs), which constitute influential vested interests, may be key in provincial interest formation.75 Larger enterprises can directly influence official decision making due to their central role in local economies and connections to local leaders.76 Though large enterprises face Party control, their overwhelming economic might grants them greater influence than diffuse industrial associations.77 Fieldwork supports this interpretation; government officials often argue that supporting the few largest enterprises in their administrative region is in the best economic interest of the region as a whole.78 Additionally, provinces that have more LMEs tend to have more energy intensive industrial sectors, and thus an interest in limiting the impact of energy efficiency regulation.79 The study measures the presence of LMEs in a local economy as the ratio of LME industrial output to total industrial output.

Control variables

In order to guard against omitted variable bias, the study includes specifications with additional control variables. Per capita GRP measures economic development. It may be that regions with a higher level of economic development have more resources to deploy in implementing policies, or there may be advantages to backwardness, with less developed regions having more potential for improvement. The controls also include GRP from the base year as a measure of the economic starting point, and economic growth as a measure of growing resources.

The study also controls for differing starting levels of EI across provinces. One model includes a measure of industrial EI at the beginning of the period. Another control measures the relative energy intensiveness of industry in a province as the ratio of output from energy intensive industry to total industrial output. This is different from industrial GRP, as a high level of industrialization does not mean that that industry is energy intensive.80

Finally, the study measures foreign ownership as the ratio of foreign industrial output to total industrial output. During field research, government officials indicated that regulating foreign enterprises was generally simple, as target firms did not wish to face penalties.

Regression results

Building on Equation (1) this study estimates the following model:

(2)

where it examines the percentage change in provincial industrial EI. This way one can determine what mix of autonomy, capacity and interests leads to optimal policy performance in implementing industrial EI regulations. Because the study is modeling change over a set period of time, it employs standard OLS regression with robust standard errors. The results are reported in Table 3.

Table 3 Change in provincial industrial EI

Independent variable	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
SOE presence	− 0.229^*^*	− 0.341^*^*	− 0.230^*^*	− 0.179^*	− 0.203^*^*	− 0.184^*^*
	( − 2.70)	( − 2.31)	( − 2.64)	( − 1.74)	( − 2.25)	( − 2.11)
Regulatory capacity	− 2.261^*^*^*	− 2.056^†	− 2.260^*^*	− 2.214^*^*^*	− 2.364^*^*	− 1.800^*^*^*
	( − 2.99)	( − 1.67)	( − 2.60)	( − 3.07)	( − 2.81)	( − 2.90)
LME presence	0.675^*^*^*	0.868^*^*^*	0.675^*^*	0.661^*^*^*	0.705^*^*^*	0.524^*^*^*
	(3.05)	(2.71)	(2.58)	(3.10)	(2.90)	(3.07)
Provincial CC index	0.034^*^*^*^*	0.032^*^*^*^*	0.034^*^*^*^*	0.031^*^*^*^*	0.031^*^*^*^*	0.028^*^*^*^*
	(7.03)	(4.83)	(6.67)	(4.96)	(5.06)	(5.22)
Industrial share of GRP	− 0.021	− 0.142	− 0.021	− 0.064	− 0.042	0.061
	( − 0.15)	( − 0.74)	( − 0.16)	( − 0.43)	( − 0.28)	(0.51)
Per capita GRP		0.000
		(0.16)
Population		0.000
		(0.69)
Foreign presence		− 0.149
		( − 0.90)
Energy intensity of industry			0.001
			(0.00)
2005 GRP				0.000
				(0.73)
2005 industrial EI					− 0.009
					( − 0.71)
Nominal GRP growth 2005 − 2009						− 0.159^*^*
						( − 2.46)
Constant	− 0.743^*^*^*^*	− 0.784^*^*^*^*	− 0.743^*^*^*^*	− 0.735^*^*^*^*	− 0.708^*^*^*^*	− 0.537^*^*^*^*
	( − 5.96)	( − 4.59)	( − 5.98)	( − 5.91)	( − 5.78)	( − 4.17)
R-squared	0.546	0.594	0.546	0.551	0.556	0.630
# of cases	28	28	28	28	28	28

Notes: The number in parentheses is the z statistic.

^****Significant at, or below, 0.1%; ^***significant at, or below, 1%; ^**significant at, or below, 5%; ^*significant at, or below, 10%.

^†Regulatory capacity has a p-value of 11.1, and is significant at, or below, 11.1%.

Interpretation: opportunity, capacity and interests for noncompliance

It is clear from the results above that institutional autonomy, capacity and interests play a significant role in the success of policy implementation. Areas with a higher proportion of SOEs and higher state capacity performed better, while areas with more autonomy and higher levels of LME presence performed more poorly in reducing industrial EI. Interestingly, the relative presence of industry does not appear to have an effect on policy implementation. Importantly, firm ownership and firm size matter irrespective of whether industry is energy intensive. Additionally, political factors matter irrespective of economic organization, and are highly significant in explaining policy outcomes.

First, the presence of SOEs is highly correlated across specifications with higher average levels of industrial EI reduction (Figure 2). On average, provinces where SOEs had a ten-percentage point greater share of economic output reduced industrial EI by 2.29 more percentage points than the national average. Management of SOEs is clearly tied to central control, which ensures SOE compliance. Localities with high levels of SOEs also have more resources freed up to regulate other enterprises, increasing local regulatory capacity. As a robustness check, the study includes a model specification with SOE investment as an explanatory variable, finding that SOE investment was insignificant and did not affect our previous results.81 SOE managers are not using investment to upgrade production efficiency; they are finding efficiency gains within the existing production line. More SOEs mean less autonomy, and thus better compliance.

Figure 2 Relationship between state ownership and EI reduction.

Secondly, higher scores on the provincial CC index correlate strongly with poor performance in reducing industrial EI, as illustrated in Figure 3. While one might assume that greater representation at the central level ties provinces more closely to central policy, these findings indicate that there are in fact considerable principle–agent problems at work in the policy process. A one-point higher score on the provincial CC index is correlated with, on average, a 3.4 percentage point reduction in performance. Given that four points represent the spread between a middling province and a powerful province, the magnitude of provincial autonomy's impact on policy performance is enormous. These results hold even when controlling for direct provincial governor and party secretary factional ties to the top leadership in Beijing, as measured by Victor Shih, indicating that provinces themselves have variable autonomy, regardless of who is in charge.82 In a decentralized system, some provinces are more autonomous than others, giving them more freedom in implementing, or not implementing, central policies.

Figure 3 Impact of CC index (provincial power) on EI reduction.

Third, the analysis shows a mere one percentage point increase in regulatory capacity (tax revenue as a percentage of GRP) is correlated with a 2.26 percentage point greater reduction of industrial EI over the national average. Thus, local regulatory capacity in terms of tax resources translates into more control of industry and is clearly important in reducing industrial EI and implementing central policy.

Fourth, LME presence helps explain provincial interests: on average, provinces with greater LME presence in industry perform worse. Provinces where LMEs had a ten-percentage point greater share of industrial output reduced industrial EI by 6.75 fewer percentage points than the national average, as illustrated in Figure 4. However, a higher level of industrial presence was insignificant, a counter-intuitive result. One would expect provinces that rely more heavily on industry would perform more poorly, as their interests would ostensibly be in protecting that industry. However, fieldwork indicates that local authorities often form economic interests based on what benefits the largest, and thus more visible and influential, industrial enterprises. It is not the mere presence of industry that drives provincial interest formation, but the relative influence of industry. Additionally, while SOE and LME presence is correlated (r = 0.6258) they have opposite effects. This provides evidence for what Brødsgaard calls ‘integrated fragmentation’, where the central government can exercise control through SOEs, but large, influential SOEs can resist regulation.83

Figure 4 Impact of LMEs on EI reduction.

These findings hold when controlling for initial conditions, economic development, and industrial makeup, as illustrated in Models 2–6. Both initial GRP and initial industrial EI are not statistically significant, indicating there does not seem to be an advantage to backwardness. Model 3 shows that the energy intensiveness of provincial industry does not have a measurable effect. In Model 6, the study controls for economic growth in the period since fast economic growth could be responsible for EI improvements. Though this measure is statistically significant, it does not change earlier findings.

Finally, the study includes a robustness check using the base-year measures of our explanatory variables of interest instead of averages, as reported in Table 4. The signs stay the same, demonstrating the robustness of the earlier models, though LME and SOE presence lose statistical significance. However, the arguments for reverse causality are weak. It could be that a reduction in EI is driving an increase in the level of SOE output over the period, since SOEs might have an advantage in adapting to government rule changes. However, SOE's share of output decreased an average of 20% over the period, only increasing in Tianjin. Most likely, SOE consolidation over the period, which the average measure accounts for, has increased the importance of SOEs in improving performance.

Table 4 Change in industrial EI, LME increase dummy

Independent variable	Original specification	LME dummy model	Original 2005 base year model	2005 base year (with dummy)
SOE presence	− 0.224^*^*^*	− 0.213^*^*^*	− 0.101	− 0.190^*
	( − 3.06)	( − 3.47)	( − 1.02)	( − 1.85)
Regulatory capacity	− 2.211^*^*^*	− 2.555^*^*^*^*	− 1.627^*^*	− 2.375^*^*^*
	( − 3.47)	( − 4.58)	( − 2.07)	( − 3.11)
LME presence	0.665^*	0.629^*^*^*	0.248	0.473^†
	(3.05)	(2.85)	(0.92)	(1.69)
Provincial CC index	0.034^*^*^*^*	0.038^*^*^*^*	0.032^*^*^*^*	0.034^*^*^*^*
	(8.05)	(7.88)	(5.17)	(5.20)
LME increase dummy		0.051^*^*		0.074^*^*^*
		(2.49)		(3.34)
Constant	− 0.750^*^*^*^*	− 0.748^*^*^*^*	− 0.563^*^*^*^*	− 0.657^*^*^*^*
	( − 7.42)	( − 2.53)	( − 5.08)	( − 6.41)
R-squared	0.545	0.591	0.490	0.577
# of cases	28	28	28	28

Notes: The number in parentheses is the z statistic.

^****Significant at, or below, 0.1%; ^***significant at, or below, 1%; ^**significant at, or below, 5%; ^*significant at, or below, 10%. ^†LME presence has a p-value of 10.4, and is significant at, or below, 10.4%.

Additionally, one could imagine that a reduction in industrial EI could drive a lower level of LME presence in a province, as they are outcompeted by more efficient enterprises. In fact, LME presence declined 6.7% on average. However, about a third of provinces experienced an increase in LME presence. The study includes another regression with a dummy that took the value of one for provinces that had an increase in LME presence in the period. This improves the statistical significance of both models. Provinces where LME presence increased saw, on average, a 7.4 percentage point lower reduction in industrial EI. Where LMEs have grown, they ostensibly have done so due to their influence and with government assistance, further reducing EI performance.

Case studies: Shaanxi, Chongqing and Guangdong

As highlighted in the introduction, the nested analysis approach integrates quantitative models with three case studies, summarized in Table 5. These cases further test the causal logic of the previous analysis—providing an additional robustness check—and also highlight factors not operationalized quantitatively. Based on fieldwork conducted from January to April 2011, the following case studies largely confirm the above results. The study examines Shaanxi, which performed well, reducing industrial EI by 39%, Chongqing, which was average with a 26% reduction, and Guangdong, which lagged, with only an 18% reduction. These three provinces vary on many of the key independent variables and provide fertile ground for comparison. These cases highlight the importance of economic growth, leadership and experimentation in reducing EI.

Table 5 Comparison of key explanatory variables in three provinces

Variables	Shaanxi	Chongqing	Guangdong
Performance
Aggregate EI change	− 20.8%	− 16.8%	− 13.4%
Industrial EI change	− 38.6%	− 26.0%	− 18.0%
SOE presence	65.1%	45.3%	16.8%
LME presence
Average level	76.9%	68.8%	66.7%
Change in LMEs	− 9.6%	− 8.2%	− 3.4%
CC index (autonomy)	Weak	Middle	Strong
	6.5	8.5	11.5
Regulatory capacity	Average	High	Average
	8.2%	9.2%	8.7%
Regulatory strength	Positive	Positive	Negative
Experimental point	+	+	−
Local leadership initiative	+	+	−
Emphasis on environment	Mixed	Yes	No
Governance model	Centralized	Statist	Open
Dominant player in economy	Central government	Local/central mix	Local government
Industrial upgrading	High	High	Low
Nominal economic growth 2005–2009	122%	113%	77%

Shaanxi: high performance

Due to geographic and industrial disadvantages, Shaanxi previously lagged in economic development. However, in the last 15 years, Shaanxi's economic development has soared due to expanding secondary industry,84 exploitation of natural resources and Beijing's Western Development Strategy.85 Importantly, government—not private investment—drives growth.86

There are three main reasons for Shaanxi's success in reducing industrial EI: first, an unusually high proportion of industry is state owned. Over the period, SOEs account for over 60% of Shaanxi's industrial output and 73% of all energy intensive output. Thus, central authorities have greater control of Shaanxi's industrial enterprises. Local officials with direct responsibility for overseeing industrial energy use said that SOEs were well behaved since SOE managers had clear incentives from the central government to reduce EI.87 In 2010, several enterprise managers were demoted by the central government for failing to comply with environmental targets.88 This highlights Shaanxi's low institutional authority vis-à-vis the central government.

Second, while still heavily industrial, Shaanxi's economic development has resulted from phasing out or upgrading existing state-owned heavy industry by shuttering and consolidating smaller, inefficient facilities and building new manufacturing and high-tech capacity. In Shaanxi, the catchphrase is ‘taotai luohou’ or ‘eliminate the backward’. SOEs have played a pivotal role in implementing this development strategy.89 And while one of China's main goals under the Western Development Strategy is to shift energy intensive industry from coastal to interior regions,90 these new plants are more efficient. Thus, Shaanxi's development strategy is complementary to Beijing's goals.

Finally, since protecting existing ‘backwards’ industrial stock is not central to the local economic development strategy, local energy regulators can do their jobs.91 Shaanxi also hired more regulators. A local official responsible for implementing the Top-1,000 program argued that they had been able to convince enterprises that energy efficiency was in their interest.92 Since managers saw energy reduction as a cost saver, rather than an environmental imperative, many self-selected to upgrade. This highlights the importance of leadership in framing polices in creative ways that allow managers to re-conceptualize their interests. Local experimentation often entails only small adjustments to the implementation approach, but those small changes, coupled with greater regulatory capacity, can make a significant difference.

Chongqing: state leadership

Chongqing is a traditional base of heavy industry and grew rapidly following its elevation to a directly controlled municipality in 1997. In fact, Chongqing has been experimenting with more liberal economic policy since 1983, when it was named the first inland port open to foreign trade.93 Even during the global economic downturn, Chongqing maintained a 14.3% growth rate due to domestic consumption.94

Chongqing's reduction of industrial EI is due largely to its rapidly growing economy and strong investment in upgrading existing heavy, state-owned industry and developing new, high-value industries.95 For example, Chongqing Iron and Steel recently opened a new plant outside the city, drastically improving the steel mill's efficiency.96 Foreign companies manufacturing low EI goods such as cars and computers for the domestic market have also entered Chongqing.

Chongqing's government has a strong regulatory hand in dealing with local industry and legacy enterprises do not receive special treatment. Industrial tax rates in Chongqing are among the highest in China and tax revenue is used, partly, to improve environmental regulation.97 Though Chongqing is relatively small they have over 100 energy auditors, underscoring their regulatory capacity.98

There remains, however, district-level variation and room for noncompliance. According to public data, from 2007 to 2012, when the Chongqing government began publicizing evaluations of district-level governments' energy savings performance, no district has received a failing grade. However, district-level data show that on 17 occasions (about 5% of the time), districts could have been given an incomplete grade. Government officials argue ‘reasonable’ ad hoc grading adjustments to account for challenges such as lack of funding, economic hardship or regional employment concerns, could explain this phenomenon.99 However, even though incentives still exist for noncompliance in some districts in Chongqing, overall the municipality performed admirably due to strong policy and regulation in most cases.

Finally, the Chongqing government has used its leverage over industry to try new policies and the central government exercises oversight, with room for local experimentation, through a number of ‘green city’ initiatives. For example, all major industries were ordered to move outside of the city. Chongqing also favors green industry; Coco Liu reports, ‘the local government cuts land costs for green businesses’.100 Chongqing has also consolidated industry in industrial parks to better regulate behavior and share resources.101

This finding once again underscores the significance of local interests and leadership preferences. If local preferences are consistent with the central government and they lead by utilizing their regulatory capacity toward those objectives, powerful provinces can be key allies in EI reduction. This highlights the need for a better understanding of local preferences. This also implies that the central government needs to both place forward thinking leaders in power locally and provide career incentives consistent with central objectives.

Guangdong: lagging performance and vested interests

Guangdong performed poorly compared to the national average, even taking into account a relatively low starting EI. Guangdong enjoys a high level of development, making upgrading production efficiency more difficult since there are fewer industries that can be rapidly upgraded. However, we account for initial EI and economic development in the model and find no systematic advantage to backwardness. At the same time, limited energy resources mean production remains inefficient due to high energy demand.102 Overall, Guangdong's autonomy from central interference, strong local vested interests and weak regulatory regime are the main factors for Guangdong's poor performance.

First, Guangdong was an early mover in the reform era and Beijing empowered it to independently direct economic development. Guangdong continues to largely manage its own growth, enjoying a very high level of autonomy vis-à-vis the central government as indicated by its representation on the Politburo and its freedom to handle incidents such as the Wukan protests independently. Additionally, since SOEs only make up 16% of Guangdong's industrial output, central regulatory institutions have little influence.

Second, Guangdong's long-standing industrial vested interests and political culture mean that industry has a highly assertive voice compared to other provinces.103 Though the level of LME penetration is at the national average, it has changed little in the last decade, meaning that the LMEs present in Guangdong are established vested interests. Moreover, Guangdong is particularly responsive to industry.104 Thus, established vested interests have a set path to influence the scope and rigor of regulation. An important distinction can be drawn with Shaanxi, where government–business ties led to superior outcomes due to an alignment of interests and strong SOE presence. In Guangdong, factory owners are concerned with ‘more energy, not more efficient energy’105 and often confidently skirt government regulations.106

Finally, despite Guangdong's wealth, its environmental and energy management regime is weak due to leadership preferences. Guangdong was an early adopter of the environmental quality leadership responsibility system that holds officials responsible for environmental performance.107 However, Guangdong's autonomy allows it to act on its clear preference for economic growth, which influences local cadre priorities. Guangdong has also chosen to use its abundant fiscal resources to hire semi-governmental service organizations, which make money from both regulators and the regulated, to regulate industry, creating conflicts of interest.108

In light of the results in Chongqing, this case once again underscores the importance of local agency and preferences for regulation. Whereas both cases have fairly high regulatory autonomy, only Guangdong utilized that opportunity to shield local industry. The lack of strong local preference for regulation coupled with their failure to build strong regulatory institutions despite ostensibly having the financial resources to do so provides the best explanation for why Guangdong performed poorly.

Conclusion

This article finds that provinces with greater authority over the local economy, more autonomy from the central government, and a higher proportion of influential local economic interests will, on average, perform more poorly in implementing central policies that may be at odds with economic growth, even when there is a high degree of central emphasis or considerable local resources. We find that our autonomy–capacity–interests model of Chinese political economy is both statistically and economically significant across provinces, and is often consistent with local realities. Thus, this framework is a useful approach to analyzing the Chinese policy process and can help explain the inconsistency of implementation across China. Relative power is not the sole determinant of localities' compliance with central mandates. Instead, this model highlights that regulatory autonomy simply provides the space for localities to act. Whether they do so is moderated by their capacity and willingness to do so.

Local preferences and leadership lie at the crux of many policy problems in China. In general, powerful provinces perform worse than others because they use their autonomy to disregard central directives. However, cases such as Chongqing and Guangdong highlight how local preferences are not given by institutions but are often independent of them. Understanding how and why these preferences vary and the ways that central leaders can shape and constrain them remains a central puzzle. These findings have important policy implications for China's ability to shift from extensive to intensive economic growth models; if Beijing can adjust provincial capacity, preferences or autonomy for noncompliance, there may be avenues for success.

In fact, Beijing is implementing just such reforms in the 12th FYP by adjusting the evaluation process, which assigns grades based on performance to firms and local governments. For example, in the newly expanded Top-10,000 Enterprises Program, the central government has tightened evaluation procedures and measurements and firm-level failure reflects more heavily on cadre evaluations.109 Combined with clear signals from the top that environmental protection is a priority for career advancement,110 these adjustments could help change province-level incentives.

Another significant change is the increasingly important role of civil society in holding local regulators accountable for implementing central polices. Policies in the 11th and 12th FYPs include provisions to ‘use all types of media’ to announce both specific yearly policy goals and year-end outcomes. Local civil society oversight can enhance top-down enforcement by alerting central authorities to lax policy implementation and ‘enable environmental regulators to do their jobs’.111 Vocal local civil society groups could also cause local governments to re-imagine their interests so as to avoid unrest or enhanced central scrutiny, thus reducing the influence of large enterprises. Finally, Jessica Teets argues that civil society can work with local governments to monitor firm-level behavior and pilot new programs.112

The autonomy–capacity–interests model of Chinese political economy provides a strong foundation for future analysis of Chinese policy implementation. Incorporating the changing role of civil society and central policies are both highly promising paths for future research.

Notes

^1. Barry Naughton, ‘Economic growth: from high-speed to high-quality’, in Joseph Fewsmith, ed., China Today, China Tomorrow: Domestic Politics, Economy, and Society (Lanham, MD: Rowman & Littlefield Publishers, Inc., 2010), pp. 71–90.

^2. C. Fred Bergsten, Charles Freeman, Nicholas Lardy and Derek Mitchell, China's Rise: Challenges and Opportunities (Washington, DC: Peterson Institute for International Economics, 2009), p. 77.

^3. Evan Lieberman, ‘Nested analysis as a mixed-method strategy for comparative research’, The American Political Science Review 99(3), (2005), pp. 435–452.

^4. Allen Carlson, Mary Gallagher and Melonie Manion, ‘Introduction’, in Allen Carlson, Mary Gallagher, Kenneth Lieberthal and Melanie Manion, eds, Contemporary Chinese Politics: New Sources, Methods, and Field Strategies (Cambridge: Cambridge University Press, 2010), pp. 1–14.

^5. Douglass North, Institutions, Institutional Change, and Economic Performance (Cambridge: Cambridge University Press, 1990).

^6. James Mahoney and Kathleen Thelen, Explaining Institutional Change: Ambiguity, Agency, and Power (New York: Cambridge University Press, 2010).

^7. Where Lieberthal, Lampton, Heilmann and Perry see decentralization and informal institutions as key, Yang and Shirk see centralization and formalization as key. Kenneth Lieberthal, ‘Introduction: the “fragmented authoritarianism” model and its limitations’, in Kenneth Lieberthal and David Lampton, eds, Bureaucracy, Politics, and Decision Making in Post-Mao China (Berkeley, CA: University of California Press, 1992); Sebastian Heilmann and Elizabeth Perry, ‘Embracing uncertainty: guerrilla policy style and adaptive governance in China’, in Sebastian Heilmann and Elizabeth Perry, eds, Mao's Invisible Hand: The Political Foundations of Adaptive Governance in China (Cambridge, MA: Harvard University Press, 2011), pp. 1–29; Dali Yang, Remaking the Chinese Leviathan: Market Transition and the Politics of Governance in China (Stanford, CA: Stanford University Press, 2004); Susan Shirk, The Political Logic of Economic Reform in China (Berkeley, CA: The University of California Press, 1993).

^8. Gerard Turley and Peter Luke, Transition Economics: Two Decades On (New York: Routledge, 2011).

^9. Barry Naughton, ‘A political economy of China's economic transition’, in Loren Brandt and Thomas Rawski, eds, China's Great Economic Transformation (Cambridge: Cambridge University Press, 2008), pp. 91–135.

^10. Paul Pierson, ‘The limits of design: explaining institutional origins and change’, Governance: An International Journal of Policy and Administration 13(4), (2000), pp. 475–499.

^11. Andrew Mertha, ‘“Fragmented authoritarianism 2.0”: political pluralization in the Chinese policy process’, The China Quarterly 200, (2009), pp. 995–1012; Jae Ho Chung, ‘Deputy-provincial cities: embedded yet de facto players’, in Jae Ho Chung and Tao-Chiu Lam, eds, China's Local Administration: Traditions and Changes in the Sub-National Hierarchy (New York: Routledge, 2010), pp. 111–126.

^12. Kenneth Lieberthal and Michel Oksenberg, Policy Making in China: Leaders, Structures, and Processes (Princeton, NJ: Princeton University Press, 1988), p. 4.

^13. Yumin Sheng, Economic Openness and Territorial Politics in China (New York: Cambridge University Press, 2010), p. 115.

^14. Sebastian Heilmann, ‘Policy-making through experimentation: the formation of a distinctive policy process’, in Heilmann and Perry, eds, Mao's Invisible Hand, pp. 62–101.

^15. Chenggang Xu, ‘The fundamental institutions of China's reforms and development’, Journal of Economic Literature 49(4), (2011), p. 1079.

^16. Ma Li, Li Huimin and Qi Ye, ‘Analysis on policy-making process of China's energy-saving performance assessment institution’, Journal of Public Management 9(1), (2012), pp. 1–8.

^17. Yongnian Zheng, ‘Center–local relations: the power to dominate’, in Fewsmith, ed., China Today, China Tomorrow, pp. 193–222.

^18. Ting Gong, ‘Institutional learning and adaptation: developing state audit capacity in China’, Public Administration and Development 29(1), (2009), pp. 33–41.

^19. Zhiyue Bo, ‘Governing China in the early 21st century: provincial perspective’, Journal of Chinese Political Science 7(1–2), (2002), pp. 125–170 at p. 165.

^20. Zheng, ‘Center–local relations’, pp. 193–194.

^21. Xu, ‘The fundamental institutions of China's reforms and development’.

^22. David Lampton, ‘A plum for a peach: bargaining, interest, and bureaucratic politics in China’, in Lieberthal and Lampton, eds, Bureaucracy, Politics, and Decision Making in Post-Mao China, p. 38.

^23. Elizabeth Economy, The River Runs Black: The Environmental Challenge to China's Future (Ithaca, NY: Cornell University Press, 2004), p. 110.

^24. Bergsten et al., China's Rise, p. 80; Yang, Remaking the Chinese Leviathan, pp. 100–101.

^25. Eunkyong Choi, ‘Informal tax competition among local governments in China since the 1994 tax reforms’, Issues & Studies 45(2), (2009), pp. 159–183.

^26. Lynette Ong, ‘Fiscal federalism and soft budget constraints: the case of China’, International Political Science Review 33(4), (2012), pp. 455–474.

^27. Sheng, Economic Openness and Territorial Politics in China, p. 117; Bergsten et al., China's Rise, p. 149; Bruce J. Dickson, Wealth into Power: The Communist Party's Embrace of China's Private Sector (Cambridge: Cambridge University Press, 2008), p. 58.

^28. Susan Whiting, Power and Wealth in Rural China: The Political Economy of Institutional Change (Cambridge: Cambridge University Press, 2001), p. 177.

^29. Charles Calomiris, ‘Profiting from government stakes in a command economy: evidence from Chinese asset sales’, Journal of Financial Economics 96(3), (2010), pp. 399–412; Hongbin Li, Lingsheng Meng, Qian Wang and Li-An Zhou, ‘Political connections, financing, and firm performance: evidence from Chinese private firms’, Journal of Development Economics 87(2), (2008), pp. 283–299; Hongbin Li, Lingsheng Meng and Junsen Zhang, ‘Why do entrepreneurs enter politics?: evidence from China’, Economic Inquiry 44(3), (2006), pp. 559–578.

^30. Yang, Remaking the Chinese Leviathan, p. 16.

^31. Keping Yu, Democracy is a Good Thing: Essays on Politics, Society, and Culture in Contemporary China (Washington, DC: Brookings Institution Press, 2008), p. xxvii.

^32. Loren Brandt, Thomas Rawski and John Sutton, ‘China's industrial development’, in Brandt and Rawski, eds, China's Great Economic Transformation, pp. 569–633 at p. 623.

^33. Bergsten et al., China's Rise, p. 147.

^34. Naughton, ‘Economic growth’, p. 76.

^35. Sheng, Economic Openness and Territorial Politics in China, p 116.

^36. Bergsten et al., China's Rise, p. 139.

^37. EI = 1 ton standard coal equivalent (SCE)/10,000 RMB.

^38. Environmental protests rose from 51,000 incidents in 2005 to 90,000 in 2011. Christina Larson, ‘The new epicenter of China's discontent’, Foreign Policy, (23 August 2011); Elizabeth Economy, ‘The great leap backward: the costs of China's environmental crisis’, Foreign Affairs 86(5), (2007), pp. 38–59.

^39. Cai Fang and Du Yang, ‘The political economy of emissions reduction in China’, in Ligang Song and Wing Thye Woo, eds, China's Dilemma: Economic Growth, the Environment and Climate Change (Washington, DC: Brookings Institution Press, 2008), pp. 226–242.

^40. John Romankiewicz, Bo Shen, Lynn Price and Hongyou Lu, Addressing the Effectiveness of Industrial Energy Efficiency Incentives in Overcoming Investment Barriers in China (Lawrence Berkeley Laboratory, October 2012).

^41. Ma et al., ‘Analysis on policy-making process of China's energy-saving performance assessment institution’; Naughton, ‘Economic growth’, p. 82.

^42. Mary-Françoise Renard and Hang Xiong, Strategic Interactions in Environmental Regulation Enforcement: Evidence from Chinese Provinces, HAL Working Papers.

^43. Lynn Price, ‘Assessment of China's energy-saving and emission-reduction accomplishments and opportunities during the 11th five year plan’, Energy Policy 39(4), (2011), pp. 2165–2178.

^44. Hua Liao, Ying Fan and Yi-Ming Wei, ‘What induced China's energy intensity to fluctuate: 1997–2006’, Energy Policy 35(9), (2007), pp. 4640–4649.

^45. Bergsten et al., China's Rise, pp. 147–148.

^46. Liao et al., ‘What induced China's energy intensity to fluctuate’.

^47. Jiahai Yuan, Junjie Kang, Cong Yu and Zhaoguang Hu, ‘Energy conservation and emissions reductions in China—progress and prospective’, Renewable and Sustainable Energy Reviews 15(9), (2011), pp. 4334–4347.

^48. Guang Xia, Xiaofei Pei and Xiaoming Yang, ‘Economic growth and environmental protection in the People's Republic of China’, in Tun Lin and Timothy Swanson, eds, Economic Growth and Environmental Regulation: The People's Republic of China's Path to a Brighter Future (New York: Routledge, 2010), pp. 35–65.

^49. Richard Ferris J. and Hongjun Zhang, ‘Environmental law in the People's Republic of China: an overview describing challenges and providing insights for good governance’, in Kristen Day, ed., China's Environment and the Challenge of Sustainable Development (Armonk, NY: M.E. Sharpe, 2005), pp. 66–101 at p. 81.

^50. John Donaldson, ‘Provinces: paradoxical politics, problematic partners’, in Chung and Lam, eds, China's Local Administration, pp. 14–38.

^51. Lynn Price, Xuejun Wang and Jiang Yun, ‘The challenge of reducing energy consumption of the top-1000 largest industrial enterprises in China’, Energy Policy 38(8), (2010), pp. 6485–6498.

^52. Price, ‘Assessment of China's energy-saving and emission-reduction accomplishments’, pp. 49–50.

^53. Price et al., ‘The challenge of reducing energy consumption of the top-1000 largest industrial enterprises in China’, p. 6. However, this is only true if officials scored below 60 out of 100 on their evaluations.

^54. Wang Xin, ‘On China's energy intensity statistics: toward a comprehensive and transparent indicator’, Energy Policy 39(11), (2011), pp. 7284–7289.

^55. Yuan et al., ‘Energy conservation and emissions reductions in China’.

^56. National Bureau of Statistics, China Statistical Yearbook: 2006 (Beijing: China Statistics Press, 2006), p. 93.

^57. Michael Lelyveld, ‘China cuts to save energy’, Radio Free Asia, (18 October 2010), available at: http://www.rfa.org/english/energy_watch/energy-10182010112348.html (accessed 15 January 2011).

^58. ‘Blackouts trigger diesel shortage in China: state media’, AFP, (7 November 2010), available at: http://www.channelnewsasia.com/stories/afp_asiapacific_business/view/1092079/1/.html (accessed 15 January 2011).

^59. Price et al., ‘The challenge of reducing energy consumption of the top-1000 largest industrial enterprises in China’, p. 20.

^60. Joanna Lewis, ‘Environmental challenges: from the local to the global’, in Fewsmith, ed., China Today, China Tomorrow, p. 269.

^61. Yasheng Huang and Yumin Sheng, ‘Political decentralization and inflation: sub-national evidence from China’, British Journal of Political Science 39(2), (2009), pp. 389–412.

^62. Xi Chen, ‘State-generated data and contentious politics in China’, in Carlson et al., eds, Contemporary Chinese Politics, pp. 15–32.

^63. Andrew Szamosszegi and Cole Kyle, An Analysis of State-owned Enterprises and State Capitalism in China (US–China Economic and Security Review Commission, 26 October 2011).

^64.Plan to Implement the Top-10,000 Enterprises Program (National Development and Reform Commission, 7 December 2011), available at: http://www.ndrc.gov.cn/zcfb/zcfbtz/2011tz/t20111229_453569.htm (accessed 14 April 2013).

^65. Many researchers emphasize the importance of relative provincial autonomy in policy success. Sheng, Economic Openness and Territorial Politics in China, p. 3; Naughton, ‘A political economy of China's economic transition’, p. 124; Zheng, ‘Center–local relations’, p. 197; Shirk, The Political Logic of Economic Reform in China, p. 83.

^66. ‘An alternate member of the Central Committee receives one point, a full member two, an alternate Politburo member one more, a full Politburo member two more, a standing Politburo member three more, and the general secretary five more.’ Bo Zhiyue, China's Elite Politics: Governance and Democratization (Singapore: World Scientific, 2010), p. 99.

^67. See Dali Yang and Fubing Su, ‘Political institutions, provincial interests, and resource allocation in reformist China’, Journal of Contemporary China 9(24), (2000), pp. 215–230; S. Philip Hsu, ‘Balancing developmental needs with vertical and horizontal power competition in China, 1993–2004’, in S. Philip Hsu, Yu-Shan Wu and Suisheng Zhao, eds, In Search of China's Development Model: Beyond the Beijing Consensus (New York: Routledge, 2011), pp. 128–146.

^68. Yumin Sheng, ‘Central-provincial relations at the CCP central committees: institutions, measurement and empirical trends, 1978–2002’, The China Quarterly 182, (2005), pp. 338–355.

^69. Zhiyue Bo, ‘The 16th central committee of the Chinese communist party: formal institutions and factional groups’, Journal of Contemporary China 13(39), (2004), pp. 223–256.

^70. Dali Yang, ‘Governing China's transition to the market: institutional incentives, politicians' choices, and unintended outcomes’, World Politics 48(3), (1996), pp. 424–452; Su and Yang, ‘Political institutions’; Hsu, ‘Balancing developmental needs.’

^71. Hsu makes this distinction, but focuses on resource extraction. S. Philip Hsu, ‘Deconstructing decentralization in China: fiscal incentive versus local autonomy in policy implementation’, Journal of Contemporary China 13(40), (2004), pp. 567–599.

^72. Yi Feng, ‘Sources of political capacity: a case study of China’, International Studies Review 8(4), (2006), pp. 597–606.

^73. Jonathan Schwartz, ‘The impact of state capacity on enforcement of environmental policies: the case of China’, The Journal of Environment & Development 12(1), (2003), pp. 50–81; Jonathan Hanson and Rachel Sigman, ‘Measuring state capacity: assessing and testing the options’, paper presented at the annual meeting of the American Political Science Association, Seattle, 2011.

^74. Taxes are not just used to prop up industry; new regulations include provisions to reduce taxes for companies that comply with EI reduction goals. See the 11th FYP Top 10 Projects Implementation Plan ‘十一五十大重点节能工程实施意见’.

^75. Dickson, Wealth into Power, p. 50; Yang, Remaking the Chinese Leviathan, p. 110; Xiaoying Ma and Leonard Ortolano, Environmental Regulation in China: Institutions, Enforcement, and Compliance (Lanham, Md: Rowman Littlefield, 2000), pp. 51–52.

^76. Raymond Fisman and Yongxiang Wang, ‘The unsafe side of Chinese crony capitalism’, Harvard Business Review, (January 2013).

^77. Kjeld Erik Brødsgaard, ‘Politics and business group formation in China: the party in control?’, The China Quarterly 211, (2012), pp. 624–648.

^78. Government officials responsible for industrial regulation from across western China expressed similar opinions at the ‘Project completion meeting of low-carbon upgrading of industrial parks in western China’, Chongqing, 20 April 2013.

^79. The energy intensiveness of secondary industry is correlated to LME presence (r = 0.4884).

^80. Industrial GRP and energy intensity of industry are not correlated (r = 0.1098).

^81. The study included measures of the proportion of SOE investment in total provincial fixed asset investment and of investment as a percentage of economic output. Investment levels, changes in investment and the origin of investment had no explanatory value.

^82. Shih argues that ties to central elites influence provincial leadership behavior. Specifications including Shih's measure of factional ties were not statistically significant. Victor Shih, ‘Factions matter: personal networks and the distribution of bank loans in China’, Journal of Contemporary China 13(38), (2004), pp. 3–19.

^83. Brødsgaard, ‘Politics and business group formation’.

^84. Yang Hui-xian, Wang Lan and Gu Hua, ‘A relevancy analysis on the relationship between energy consumption and economic growth in Shaanxi’, Journal of Baoji University of Arts and Sciences 31(4), (2011), pp. 104–112.

^85. Chuanyi Lu, Xiliang Zhang and Jiankun He, ‘A CGE analysis to study the impacts of energy investment on economic growth and carbon dioxide emission: a case of Shaanxi Province in western China’, Energy 35(11), (2010), pp. 4319–4327.

^86. Ren Zhi-gang and Guan Sheng-dong, ‘The choice of growth model and the differences of economic performance in the east and west of China: a comparative study between Shaanxi and Zhejiang’, Modern Economic Science 31(2), (2009), pp. 108–116.

^87. Interview with regulatory official, Shaanxi, March 2011.

^88. Interview with former provincial party official, Beijing, February 2011.

^89. Ren and Guan, ‘The choice of growth model and the differences of economic performance in the east and west of China’.

^90. According to the ‘Industrial Transfer Guidance List’ published by the Ministry of Industry and Information Technology, Shaanxi is classified as a receiving province for heavy industry ‘产业转移指导目录 (2012年本)’.

^91. LMEs are concentrated in mining, which faces heavy central oversight.

^92. Interview with regulatory official, Shaanxi, March 2011.

^93. Christina Larson, ‘Chicago on the Yangtze: welcome to Chongqing, the biggest city you've never heard of’, Foreign Policy, (September/October 2011).

^94. Ninety percent of Chongqing's industrial output is sold domestically. Zhiyue Bo and Chen Gang, ‘Bo Xilai and the Chongqing model’, East Asian Policy 465, (2009), pp. 42–49.

^95.Ibid.; Xiao-kun Li and Wei Jie, ‘Decoupling between environmental pressures and economic growth in Chongqing metropolitan area’, Journal of Natural Resources 25(1), (2010), pp. 139–147.

^96. Coco Liu, ‘Built in a dirty boom, China's biggest city tries to go green’, New York Times, (26 September 2011).

^97. Kevin Lu, ‘The Chongqing model worked’, Foreign Policy, (8 August 2012).

^98. Interview with officials at Chongqing Energy Use and Monitoring Center, Chongqing, March 2011.

^99. Tucker Van Aken, ‘Actions speak louder than words: a political economic take on campaign-style enforcement’, Public Administration Review 75(1), (2015).

^100. Liu, ‘Built in a dirty boom, China's biggest city tries to go green’.

^101.Research Report: Low Carbon Upgrading of Industrial Parks in Western China—Take Chongqing for Example (Chinese Academy of Sciences, April 2013).

^102. Yong Lin and Liu Yun, ‘Suggestions on optimized power development in Guangdong’, Guangdong Electric Power 11, (2011), pp. 1–5.

^103. Jieh-min Wu, ‘Strange bedfellows: dynamics of government-business relations between Chinese local authorities and Taiwanese investors’, Journal of Contemporary China 6(15), (1997), pp. 319–346.

^104. Lance Gore, China in Search of a New Development Model: Guangdong or Chongqing?, EAI Background Brief No. 701 (2012).

^105. Interview with Guangdong factory owner, Beijing, January 2011.

^106. Email correspondence with Guangdong-based businessmen, 31 January 2011 and 1 February 2011.

^107. Carlos Wing-Hung Lo and Shui-yan Tang, ‘Institutional reform, economic change, and local environmental management in China: the case of Guangdong Province’, Environmental Politics 15(2), (2006), pp. 190–210.

^108.Ibid.

^109. Tucker Van Aken, ‘Making the grade: performance targets and industrial energy policy’, China Environment Series 12, (2013).

^110. Henry Sanderson, ‘Chinese cadres told going green rivals GDP to rise in Party’, Bloomberg News, (13 March 2014).

^111. Alex Wang, ‘The search for sustainable legitimacy: environmental law and bureaucracy in China’, Harvard Environmental Law Review 365, (2013).

^112. Jessica Teets, Civil Society under Authoritarianism: The China Model (London: Cambridge University Press, 2014).