Herding Cats: Firm Non-Compliance in China's Industrial Energy Efficiency Program.

• Author: Karplus, Valerie J.

1. INTRODUCTION

   Initiatives to raise industrial energy efficiency are now common in many nations, but firm compliance behavior remains poorly understood. Globally, the industrial sector is responsible for 54% of energy consumption (Energy Information Agency, 2016) and multiple forms of local pollution. Over the past several decades, the shares of industrial energy use and emissions concentrated in developing and emerging countries has grown substantially (Energy Information Agency, 2016), prompting national governments to establish programs to control them. In these settings, energy efficiency measures are attractive because they are estimated to rank among the lowest-cost abatement opportunities (Intergovernmental Panel on Climate Change, 2014), and do not explicitly restrict firm output. The centrality of energy efficiency programs in the climate pledges of major developing countries such as China and India raises the importance of understanding the determinants of compliance at the firm level.

   Industrial energy efficiency programs in developing countries differ from advanced industrialized countries in several ways. First, governments may perceive large tradeoffs with economic development when implementing policies designed to directly or indirectly limit energy use (Greenstone and Jack, 2015). Second, rapidly-growing or volatile market conditions in developing countries introduce greater uncertainty into firms' investment decisions. Structural changes in the markets in which firms operate may further create large uncertainties in demand and associated energy requirements (Fisher-Vanden, 2003). Energy efficiency upgrades typically require investing in new equipment or processes, the benefits of which are amortized over multi-year horizons, and returns are easier to predict when operating conditions are stable. Third, state implementation capacity and incentives may be underdeveloped or uneven, especially if the policy is being introduced for the first time and firms lack experience (Kostka, 2016). For instance, firms may collude with regulators, manipulate data, or renege on compliance obligations (Duflo et al., 2013; Ghanem and Zhang, 2014). Implementation may be selective depending on the relationship between state regulators and firms (Wang, 2015; Li and Chan, 2016). All of these attributes are present in the Chinese context.

   We analyze the first nationwide initiative to raise energy efficiency in China: the Top 1.000 (2006-2010) / Top 10,000 (2011-2015) Firms Energy Saving Program (hereafter, T1000P and T10000P). Under the program, the government assigned firms roughly equal energy-saving targets proportional to their energy use in the base year. First, we develop a simple model to represent the firm's decision to comply with the program, as well as its propensity to exaggerate or falsify its progress. Second, we use a statistical method to evaluate the plausibility of compliance data submitted by firms, comparing the T1000P and the T10000P. Third, we evaluate the relationship between pre-existing firm characteristics and compliance outcomes, focusing on the T10000P.

   Our results are consistent with the notion that if compliance incentives are strong but enforcement incentives are not, firms will exaggerate performance. From this baseline, our model projects that strengthening enforcement incentives will tend to reduce compliance rates. Based on prior literature, we hypothesize that firms that are not accountable to the state through ownership ties as well as large firms that are economically important locally, benefit less from complying and would thus be more likely to not comply. Consistent with this logic, our findings suggest that firms exaggerated performance during the T1000P, but not during the T10000P, when enforcement incentives were strengthened. Meanwhile, non-compliance increased from near zero to around 6-8%, which we interpret as evidence of more accurate reporting. In the T10000P, larger firms not controlled by the state and firms in cities with relatively slow growth were associated with lower compliance on average, consistent with smaller marginal benefits of compliance for these firms.

2. LITERATURE AND EMPIRICAL SETTING

   2.1 Firm compliance with energy and environmental policy

   Prior literature on energy efficiency programs focuses largely on developed economies (Allcott and Greenstone, 2012). While much of this literature focuses on individual incentives, a subset probes organizational responses to energy efficiency programs. Studies of firms (DeCanio and Watkins, 1998) and schools (Burlig et al., 2017) have shown that energy efficiency decisions depend on characteristics of the organization. Studies have tended to focus on explaining energy efficiency investments rather than compliance with policy per se. Location (including proximity to headquarters and similar firms, see Doshi et al., 2013), size (Bennear and Olmstead, 2008), and stakeholder pressure (Kagan et al., 2003) have been shown to affect firms' willingness to comply with, or even exceed, regulatory targets. In developed countries, studies suggest that compliance rates are generally high and data is broadly trustworthy (Shimshack, 2014), although high profile cases of data manipulation, such as the Volkswagen emissions testing scandal in 2015, do occur. Evidence from developing countries is much more limited, but suggests that regulatory enforcement is often weak and uneven (Duflo et al., 2013; Van Rooij, 2010).

   China is an important context to study the dynamics of compliance with an energy efficiency program for several reasons. First, it has a vast energy-intensive industrial sector responsible for approximately 55% of the nation's energy use (National Bureau of Statistics, 2013), equivalent to more than 10% of the global total (International Energy Agency, 2014). Second, environmental policies have been introduced for years and strengthened by the central government over time, but implementation ultimately depends on local governments and firms. Despite ever stronger policies, there is evidence of a persistent "implementation gap" (Chan et al., 1995), in which policy ambition exceeds achievement. Third, China offers an opportunity to evaluate an energy efficiency program against the backdrop of a rapidly-growing country with unique institutions. Prior work has described how most policies impose rigid targets on firms, limiting spatial and temporal flexibility in compliance responses (Kostka, 2016). The literature has also pointed to a role for local protection (Lorentzen et al., 2014) and state ownership (Hering and Poncet, 2014; Li and Chan, 2016) in shaping firm's responses to policies. In the T1000P and T10000P, for instance, target achievement was included in the government's system for evaluating the annual performance of state-owned firm leaders (Kostka, 2016), while non-state-owned firms only felt pressure indirectly through local leaders who were similarly evaluated on energy efficiency improvements in their jurisdiction. Prior studies have not explored the relationship between these characteristics and compliance in detail.

   Our analysis also builds on prior studies that evaluate the T1000P and T10000P in China. Multiple studies suggest both programs were highly effective (Zhou et al., 2010; Ke et al., 2012); here, we assess the underlying quality of the data used to make these judgments. Prior studies also identify a number of implementation challenges. These include a lack of incentives for activities that were not capital improvements, e.g. capacity building (Lu et al., 2014) during the T1000P, and a lack of both transparency and oversight (Price et al., 2010). The interaction between oversight and China's cadre evaluation system may have played a critical role. In China, an important distinction between state and non-state firms lies in the evaluation of leaders. State-owned firm leaders are evaluated as part of the national cadre evaluation process, which includes a wide range of criteria, including environmental performance. In principle, the cadre evaluation system provides a strong incentive for firm leaders to perform in order to earn accolades or promotion, which affect status, reputation, and resource access. Energy-saving target achievement under the T1000P and T10000P was part of the cadre evaluation for state-owned enterprise leaders. However, strong incentives to show compliance combined with uneven or incomplete oversight of data quality may have encouraged exaggeration or falsification of environmental performance.

   To summarize, all of the above studies of the T1000P/T10000P adopt a level of administrative aggregation above the firm (e.g. provinces and municipalities) or adopt survey and case study approaches that only cover a small sample of firms. Ours is the first study to examine compliance outcomes at the firm level using a new, multifaceted data set.

   2.2 Top 1,000 and Top 10,000 Enterprises Programs: Structure and Compliance

   The T1000P was established during the Eleventh Five-Year Plan (FYP, 2006-2010) and included 1,008 industrial firms with energy consumption higher than 180,000 tons of coal-equivalent in 2004 in nine sectors (coal, textiles, paper, chemicals, petroleum and petrochemicals, building mterials, iron and steel, non-ferrous metal, and power and heat) (1). Together, these firms accounted for about 30% of China's total energy use in 2005.

   The program was formally launched in April 2006 with a notice from the National Development and Reform Commission (NDRC) (National Development and Reform Commission, 2006). Each firm was assigned an energy-saving target for the Eleventh FYP (2006-2010). The detailed target-setting process has not been disclosed to the public, but it is reported that the target was roughly proportional to a firm's energy use in 2004 with no consideration of firm-specific energy saving potential (Zhao et al., 2014). The target setting process was...