Carbon disclosure, carbon performance, and cost of capital

Yu He; Qingliang Tang; Kaitian Wang

doi:10.1080/21697221.2014.855976

Carbon disclosure, carbon performance, and cost of capital

He, Yu; Tang, Qingliang; Wang, Kaitian 2013-12-01 00:00:00 China Journal of Accounting Studies, 2013 Vol. 1, Nos. 3–4, 190–220, http://dx.doi.org/10.1080/21697221.2014.855976 a b a Yu He *, Qingliang Tang and Kaitian Wang a b School of Accounting, Nanjing University of Finance and Economics, China; School of Business, University of Western Sydney, Australia More and more ﬁrms are voluntarily disclosing carbon information as a response to the challenge of climate change. This research investigated the interactions among carbon disclosure, carbon performance, and the cost of capital. Because unobservable overall strategic decisions by management affect each of these outcomes and phenomena, we used a simultaneous equations model to analyse our data. We used data from S&P 500 ﬁrms that participated in the Carbon Disclosure Project (CDP) in 2010. We found that the cost of capital is negatively associated with carbon disclosure, which is consistent with voluntary disclosure theory. This relationship is weaker for ﬁrms with good carbon performance. In addition, there is an inverse relationship between carbon disclosure and carbon performance, which is consistent with legitimacy theory. Our results suggest that voluntary carbon disclosure is a rational choice that ﬁrms make to reduce the pressure exerted by legitimacy threats and to lower the cost of capital. Keywords: carbon disclosure; carbon performance; cost of capital; simultaneous equations 1. Introduction Corporate responses to climate change have shifted dramatically over the past two decades (Kolk, Levy, & Pinkse, 2008). Until the mid-1990s, North American ﬁrms seldom talked about the topic. Nevertheless, a few ﬁrms in sectors related to fossil fuels perceived the prospect of regulations of greenhouse gas (GHG) emissions as a substan- tial threat. New industry groups were created, such as the Global Climate Coalition and the Climate Council, which spared no effort in preventing the international community from imposing caps on GHG emissions and played a major role in preventing the United States from joining the Kyoto Protocol (Kolk et al., 2008; Levy & Egan, 2003). However, more recently, business has tended to converge on a more constructive stance that views climate change as an opportunity rather than a burden (Margolick & Russell, 2004). Financial markets have started to reward companies that are moving ahead on climate change, while those lagging behind are assigned more risk (Cogan, 2006; Kolk et al., 2008). Investors and environmental non-governmental organizations (NGOs) are pushing companies to disclose information related to their GHG emissions, since carbon disclosure provides information that is crucial to an accurate valuation of assets. NGOs can also use the information to pressure ﬁrms into improving carbon performance (O’Dwyer, 2005). Several initiatives have emerged that attempt to leverage the inﬂuence of institutional investors to create demand for carbon disclosure as an *Corresponding author. Email: yu.he@njue.edu.cn Paper accepted by Kangtao Ye. © 2013 Accounting Society of China China Journal of Accounting Studies 191 adjunct to conventional ﬁnancial systems. One of the most prominent is the Carbon Disclosure Project (CDP). CDP is an independent and not-for-proﬁt organization based in the United King- dom that addresses the climate change concerns of institutional investors (Tran, Oka- for, & Herremans, 2011). The CDP represents 534 institutional investors with more than US$64 trillion in assets under management, and it can be seen as a secondary stakeholder that has facilitated collaborative engagement to increase corporate account- ability in relation to climate change (Arenas, Lozano, & Albareda, 2009). By means of a standard questionnaire, the CDP collects climate change data on GHG emissions, carbon risks and opportunities, and the actions that companies are taking to reduce emissions. In 2010, CDP sent this questionnaire to more than 4700 of the world’s larg- est corporations, including S&P 500 ﬁrms. Some 70% (350) of S&P 500 ﬁrms partici- pated in the CDP in 2010, an increase from 66% (332) in 2009, 63% (314) in 2008, and 56% (280) in 2007 (PwC & CDP, 2010). Overall, this activity sends an important message to investors from companies that climate change is an important business concern. However, the rapid increase in participation in the CDP naturally raises ques- tions among researchers: what are the rationales behind this type of voluntary carbon disclosure? What beneﬁts do ﬁrms gain by spending resources on compiling and pub- lishing these standalone reports? A number of factors potentially provide answers to these questions, such as the pressure on businesses to establish and comply with environmental and social norms and standards (Cormier, Magnan, & Van Velthoven, 2005). In this research, we examined whether carbon disclosure is associated with a reduction in the cost of capital. The cost of capital is a critical issue in a ﬁrm’s ﬁnancing and general oper- ating decisions (Dhaliwal, Li, Tsang, & Yang, 2011). Corporate executives appear to believe that voluntary disclosure can reduce the cost of capital (Graham, harvey, & Rajgopal, 2005), and there is a long-standing interest among academics in the rela- tionship between disclosure and the cost of capital (Botosan, 1997; Botosan & Plumlee, 2002; Dhaliwal et al., 2011; Diamond & Verrecchia, 1991; Francis, Nanda, & Olsson, 2008; Leuz & Verrecchia, 2000; Richardson & Welker, 2001). While these investigations have advanced our knowledge, no study has yet attempted to examine carbon disclosure, carbon performance, and the cost of capital within a sin- gle inclusive model. Ullmann (1985) posited that management implements policies and decisions that simultaneously affect the ﬁrm’s environmental disclosure, environ- mental performance, and economic performance. Al-Tuwaijri, Christensen, and Hughes (2004) argued that if these corporate functions are endogenously determined, then piecemeal ordinary least squares (OLS) estimation of pairwise relationships among these three functions will produce biased and inconsistent results. They per- formed a Hausman (1978) test and rejected their null hypothesis of no endogeneity among these three variables. Thus, we investigated collectively the relationships among carbon disclosure, car- bon performance, and the cost of capital. We employed a sample of US S&P 500 cor- porations that present their CDP reports on the CDP website. Our analyses showed that the cost of capital is signiﬁcantly and negatively associated with carbon disclosure. However, the negative relationship resides largely in ﬁrms with poor carbon perfor- mance. In addition, we documented a negative relationship between prior disclosure and contemporaneous emission reduction, suggesting that poor carbon performers tend to present (rather than withhold) carbon information in advance in attempts to mitigate any future negative impacts on the market. Overall, we observed that a potential reduc- 192 He et al. tion in the cost of capital motivates ﬁrms to present CDP reports, and this is related to a ﬁrm’s long-term development strategies and performance sustainability. This study is the ﬁrst to use an innovative approach to examine the relationships among carbon disclosure, carbon performance, and the cost of capital. Here, we extend the traditional research on voluntary disclosure from the narrow focus of ﬁnancial disclosure to nonﬁnancial disclosure, i.e. carbon disclosure. This research has made the following speciﬁc contributions. First, we focused on carbon disclo- sure and performance rather than social responsibility (Dhaliwal et al., 2011)or environmental reporting (Plumlee, Brown, & Marshall, 2008). Carbon study is a new area of research, and our ﬁndings on carbon disclosure and performance are different from ﬁndings on environmental disclosure. For example, we showed that carbon disclosure is broader than environmental disclosure. Carbon risks are more pervasive; ﬁrms must adopt a more comprehensive strategy and actions to deal with climate change. All ﬁrms emit carbon, while many ﬁrms do not have environmental issues. In addition, most environmental disclosure is nondiscretionary, for example, toxic emissions and chemical waste, whereas carbon disclosure is largely voluntary. For all these reasons, the nature of the relationship between carbon information and market reactions may be different from that documented in prior studies. Second, the measure of carbon performance used in our study was innovative compared with previous studies. Our measurement was more comprehensive than other environmen- tal performance measures, since we used sector- and ﬁrm-speciﬁc data such as waste, toxic chemical emissions, etc. Third, our data were taken from the CDP reports, rather than from environmental information included in annual reports, which are often biased by self-selection (Tang & Luo, 2013). CDP reports are designed by an NGO, and its format and contents are accepted and adopted by more than 4000 large global ﬁrms. Such information is relatively more consistent (Luo, Lan, & Tang, 2012). Fourth, this research introduced an empirical proxy for carbon disclosure. In contrast to prior studies, which often used self-constructed indices (Plumlee et al., 2008) or indicator variables (e.g. whether or not a report was published) (Dhaliwal et al., 2011), we used the carbon disclosure score index from the CDP, which has been widely presented on many ﬁnancial websites (e.g., http://www.google.com/ﬁnance#). This index is more comprehensive than others and covers many aspects of relevant information, such as carbon governance mecha- nisms, carbon risks and opportunities, carbon strategy and targets, carbon actions and processes, carbon emissions and reporting, carbon emission trading and offset- ting, carbon communications and engagement, and more (Tang & Luo, 2013; Luo et al., 2012). Finally, our methodologies differed from previous carbon accounting research. We conducted a joint estimation using three-stage least squares (3SLS) simultaneous equations models to deal with endogeneity. In addition, carbon disclo- sure and performance are the focus of legislation that is either proposed or has been implemented by many governments and rule makers. For example, carbon account- ing standards, external assurance/veriﬁcation, carbon disclosure, and carbon reporting are all subject to future regulations. Thus, our results are potentially useful for capi- tal market participants and government bodies who are concerned about climate change-related issues and activities. The remainder of this paper proceeds as follows: Section 2 develops our hypothe- ses, Section 3 describes our sample and methodology, Section 4 presents the empirical results, and the ﬁnal section summarizes and concludes. China Journal of Accounting Studies 193 2. Literature review and hypotheses development The existing literature in environmental accounting research can be categorized into three broad groups (Clarkson, Li, Richardson, & Vasvari, 2008). The ﬁrst group found that environmental disclosure is relevant to ﬁrm valuation, the second investigated the determinants of discretionary environmental disclosure, and the third line of studies observed mixed results regarding the relationship between environmental disclosure and environmental performance. 2.1. The relationship between carbon disclosure and the cost of capital The consensus in the literature appears to be that a negative relationship exists between the quality of ﬁnancial disclosure and the cost of capital (Botosan, 1997; Core, 2001; Diamond & Verrecchia, 1991; Healy & Palepu, 2001; Leuz & Wysocki, 2008). Greater ﬁnancial disclosure increases investors’ awareness of a ﬁrm’s exis- tence and enlarges its investor base, which improves risk-sharing and reduces the cost of capital (Merton, 1987). In addition, expanded disclosure can narrow informa- tion asymmetry among investors or between managers and investors. Some informa- tion-disadvantaged investors are less willing to trade if disclosure is inadequate. The resultant illiquidity increases the bid-ask spread and transaction costs (Verrecchia, 2001), leading to a higher required rate of return or cost of capital (Amihud & Mendelson, 1986). In a traditional capital market setting, Lambert, Leuz, and Ver- recchia (2007) investigated the direct and indirect effects of disclosure quality on the cost of capital. They found that the direct effect occurs because higher-quality disclosures affect the ﬁrm’s assessed covariance with other ﬁrms’ cash ﬂows, which are non-diversiﬁable. The direct effect on the cost of capital can be attributed to a reduction in the estimation of information risk (Lambert et al., 2007; Diamond & Verrecchia, 1991; Leuz & Verrecchia, 2000). The indirect effect occurs because higher-quality disclosures affect a ﬁrm’s real decisions, which likely changes the ﬁrm’s ratio of expected future cash ﬂows to covariance of these cash ﬂows with the sum of all the cash ﬂows in the market. The sign of this indirect effect on the cost of capital is uncertain. Lambert et al. (2007) sought to determine the conditions under which an increase in information quality leads to an unambiguous decline in the cost of capital. They found that the net effect of increased disclosures on the cost of capital is a function of the strength of the relationship and whether the indi- rect effect is positive or negative (as the direct effect is unambiguously negative). These mechanisms can be extended to non-ﬁnancial disclosure, as long as the information concerned is value relevant (Sinkin, Wright, & Burnett, 2008). Richardson, Welker, and Hutchinson (1999) presented a model for the inﬂuence of environmental behaviours and the related disclosures on a ﬁrm’s value through net present value assessments of projects, including expected future regulatory costs and market effects. Richardson and Welker (2001) argued that there may be a direct inﬂuence of environmen- tal disclosure on the cost of equity capital, either through affecting investor preferences or through reduced information asymmetry or estimation risk. Investor preference effects arise if investors are willing to accept a lower rate of return on investments from an orga- nization that supports an environmental cause for which some investors have an afﬁnity. In addition, carbon disclosures may be seen as a ﬁrm’s credible commitment to environ- mental issues in their long-term strategic and production systems (Plumlee et al., 2008). Finally, higher-quality environmental disclosures, which have been linked to increased 194 He et al. environmental activities, may affect regulators’ decisions, thus inﬂuencing costs for ﬁrms and for their competitors (Decker, 2002; Decker & Pope, 2005; Salop & Scheffman, 1983). Clarkson, Fang, and Li (2011a) investigated the relevance of environmental disclosures and found that voluntary environmental disclosures enhanced ﬁrm value. Based on this literature, we tested the following hypothesis (stated in alternative form). H1: There is an inverse relationship between the level of carbon disclosure and the cost of capital. According to legitimacy theory, carbon disclosure is a function of social and political pressures. Firms with poor carbon performance face greater pressures and have greater incentives to disclose environmental information in an attempt to change public perception. Patten (1992) noted that, whereas economic legitimacy is monitored by the market, social legitimacy is monitored by the public policy process. Disclosure is one method available to ﬁrms to enhance their legitimacy, and it is often easier to manage the ﬁrm’s image than to make actual changes to performance. Therefore, higher polluting ﬁrms tend to disclose a greater quantity of information (Clarkson, Overell, & Chapple, 2011c), whereas the level of carbon disclosure by better performing ﬁrms is probably lower. This is because these ﬁrms face a smaller legitimacy problem, and stakeholders presumably pay less attention to these ﬁrms. Thus, investors are probably less sensitive to the carbon disclosure of good performers. Hence, the inverse relation- ship between the level of carbon disclosure and the cost of capital is predicted to be weaker for ﬁrms with higher carbon performance. H1a: The inverse relationship between the level of carbon disclosure and the cost of capital is weaker for ﬁrms with good carbon performance. 2.2. The relationship between carbon disclosure and carbon performance There are two theories regarding the relationship between carbon disclosure and carbon performance in the literature. The ﬁrst one is voluntary disclosure theory, which suggests that companies have incentives to disclose ‘good news’ to differentiate themselves from companies with ‘bad news’ to avoid the adverse selection problem (Dye, 1985; Verrecchia, 1983). Bewley and Li (2000) and Li, Richardson, and Thornton (1997) have argued that true environmental performance is not directly observable to investors; thus, companies with superior performance tend to make direct voluntary disclosures that cannot be easily matched by poor performers (Clarkson et al., 2008). Hence, this theory predicts a positive association between environmental performance and the level of discretionary environmental disclosure. The other group of theories is known as legitimacy theories, and this group argues that companies with threatened legitimacy are likely to make self-serving disclosures, referred to as ‘legitimization’ (Adams, 2004; Gray, Kouhy, & Lavers, 1995; Hughes, Anderson, & Golden, 2001). Researchers argue that ﬁrms vulnerable to certain types of criticism will make voluntary disclosures to deﬂect or nullify suspicion or doubt with regard to that area of potential criticism. Those organizations whose social legitimacy is threatened tend to increase disclosures to (1) educate and inform relevant members of the public about (actual) changes in their performance; (2) change perceptions about their performance; (3) deﬂect attention from the issue of concern by highlighting other accomplishments; and (4) attempt to change public expectations of their performance (Gray et al., 1995; China Journal of Accounting Studies 195 Lindblom, 1994). Therefore, legitimacy theories predict a negative association, and the empirical evidence in existing literature is largely mixed (Tang & Luo, 2011). For example, Wiseman (1982) examined a sample of 26 ﬁrms in environmentally sensitive industries using an indexing procedure to measure the extent of disclosure of 18 environmental items and found that no relationship existed between environmental disclosure and performance. Freedman and Wasley (1990) used the same indexing procedure for 50 US companies and conducted Spearman rank order correlation tests. They found that neither annual reports nor environmental disclosures (as communicated via the Securities and Exchange Commission’s 10-K report) were signiﬁcantly associ- ated with actual pollution control performance. Ingram and Frazier (1980) found that environmental disclosures did not relate strongly to environmental performance, as measured by CEP indices. However, Bewley and Li (2000), who adopted the same rat- ing scheme used by Wiseman (1982), found a negative association. They used a cross- sectional sample of Canadian manufacturing ﬁrms and documented that ﬁrms with more news media coverage of their environmental exposure, higher pollution propen- sity, and more political exposure were more likely to disclose general environmental information. Hughes et al. (2001) investigated whether disclosures differed between ﬁrms who were rated good, mixed, or poor in their environmental activities by the CEP and whether these differences in disclosure could be used to determine actual environmental performance levels. The study focused on 51 US manufacturing ﬁrms for 1992 and 1993 and found that the poor performers made the most disclosures. Patten (2002) identiﬁed three issues with previous studies: (1) failure to control for other factors; (2) inadequate sample selection; and (3) inadequate measures of environ- mental performance using CEP indices. Thus, Patten used Toxics Release Inventory data, normalized by sales, as a proxy for environmental performance and examined 131 US companies. The results indicated that there was a signiﬁcant negative relationship. Campbell (2003) examined environmental disclosures from the annual reports of a sample of ten UK FTSE 100 Index companies in ﬁve sectors between 1974 and 2000. The study provided evidence for legitimacy theory in the United Kingdom as an expli- cator of variability in environmental disclosure, adding to the most notable previous studies from Australia (Deegan & Gordon, 1996; Deegan & Rankin, 1996;O’Donovan, 2002) and North America (Buhr, 1998; Patten, 1992). Al-Tuwaijri et al. (2004) explored the relationships among environmental disclo- sure, environmental performance, and economic performance using a simultaneous equations approach. The study adopted a similar disclosure-scoring methodology based on content analysis that incorporated disclosures of four key environmental indicators: (1) the total amount of toxic waste generated and transferred or recycled; (2) ﬁnancial penalties resulting from violations of ten federal environmental laws; (3) Potential Responsible Party designation for the clean-up responsibility of hazardous waste sites; and (4) the occurrence of reported oil and chemical spills. These disclosures were based on information reported on Form 10-K and are largely non-discretionary. The study found a positive association between environmental performance and disclosure. More recently, Clarkson et al. (2008) focused on purely voluntary disclosure media, such as corporate Internet websites and stand-alone environmental reports. In addition, using a content analysis index, the study found that environmental performance was positively related to the level of discretionary disclosures in environmental and social reports or related web disclosures. However, the authors showed that ﬁrms whose envi- ronmental legitimacy is threatened may make more ‘soft claims’ regarding their com- 196 He et al. mitment to the environment. This result is predicted by legitimacy theory but cannot be explained by voluntary (or economic) disclosure theories. In a more recent study, Clarkson et al. (2011c) examined how both the level and the nature of environmental information disclosed by Australian ﬁrms related to their underlying environmental per- formance and found consistently that ﬁrms with a higher propensity toward pollution disclosed more environmental information. In summary, existing studies have shown mixed results. One reason for the inconclusive ﬁndings is the use of different indices developed by the authors to measure the chosen factors for environmental disclosure, and these different indices may be inconsistent with each other since they often include different kinds of environ- mental information. Another possibility is that these studies did not adequately address endogeneity issues. Thus, we used the CDP carbon disclosure index (CDI) and a joint- estimation approach to mitigate these concerns. Since voluntary disclosure theories and legitimacy theories have provided opposing predictions, we tested the following two competing hypotheses (stated in the alternative form). H2a: Carbon performance is positively associated with the level of discretionary carbon disclosure, which is consistent with voluntary disclosure theories. H2b: Carbon performance is negatively related to the level of discretionary carbon disclosure, which is consistent with legitimacy theories. 3. Sample and methodology 3.1. Sample description This study used a cross-sectional research design. Sample ﬁrms must meet all the fol- lowing criteria. That is, the ﬁrm: (1) was in the list of both 2009 and 2010 CDP reports and in the S&P 500 2009 and 2010 reports; (2) had a carbon disclosure score and carbon emission data for 2010; (3) did not experience takeovers, mergers, and/or acqui- sitions in 2010 and 2009; (4) had a cost of capital that could be computed; (5) and had complete ﬁnancial data reported in S&P’s Compustat database and the Center for Research in Security Prices (CRSP) database. Selection processes were conducted step by step. Of the 2010 S&P 500, 350 ﬁrms completed the CDP questionnaire, and only 240 ﬁrms met selection criteria (1) through (3). The cost of capital of 28 ﬁrms could not be computed, and 21 ﬁrms do not have complete Compustat and CRSP data. The ﬁnal sample therefore included 181 disclosing ﬁrms that met all of the selection criteria. Table 1 shows the industry distribution of all S&P 500 and sample ﬁrms. 3.2. Empirical models and variable deﬁnitions Ullmann (1985) conducted a meta-analysis of prior empirical studies that investigated the relationships among social disclosure, social performance, and economic perfor- mance. He noted that the mixed results in prior empirical studies might be caused by incomplete speciﬁcation of the empirical models measuring the statistical signiﬁcance of pairwise association. Al-Tuwaijri et al. (2004) argued that this body of empirical environmental research may have produced ambiguous results by failing to recognize the potential for endogenous relationships among environmental disclosure, environ- mental performance, and economic performance. Thus, they sought to detect the pres- ence of endogenous relationships among these three constructs by using a Hausman China Journal of Accounting Studies 197 Table 1. Sample distribution by industry. Industries All ﬁrms (no.) % Firms responding (no.) % Sample ﬁrms (no.) % Consumer Discretionary 80 16 49 14 18 10 Consumer staples 41 8 37 11 26 14 Energy 39 8 23 7 15 8 Financials 78 16 51 15 10 6 Health care 52 10 37 11 19 10 Industrials 58 12 37 11 24 13 Information Technology 77 15 56 16 33 18 Materials 32 6 25 7 18 10 Telecommunications 9 2 6 2 3 2 Utilities 34 7 29 8 15 8 Total 500 100 350 100 181 100 (1978) test. Similarly, we estimated the relationships among carbon disclosure, carbon performance, and the cost of capital by employing a system of simultaneous equations, deﬁned in the following structural form: COST ¼ CD þ BETA þ FD þ SIZE þ MB þ INTENSITY þ IND þ e i;t i;t i;t i;t i;t i;t i;t i;t i;t (1) CD ¼ CP þ FD þ FIN þ TOBINQ þ LEV þ SIZE þ LIQUIDITY i;t i;t i;t i;t i;t i;t i;t i;t (2) þ ROA þ INTENSITY þ LITIGATION þ IND þ e i;t i;t i;t i;t i;t CP ¼ CD þ FIN þ TOBINQ þ LEV þ SIZE þ MB þ ROA i;t i;t1 i;t i;t i;t i;t i;t i;t (3) þ INTENSITY þ LITIGATION þ IND þ e i;t i;t i;t i;t where COST , CD , and CP are endogenous variables and the other variables are i,t i,t i,t predetermined variables. In equation (1), COST is the implied cost of equity capital in year t estimated by the price/earnings to growth (PEG) formula in Easton (2004), which was recommended by Botosan and Plumlee (2005). CD is an empirical proxy for voluntary carbon disclo- sure of ﬁrm i in year t and is calculated using data from the CDP report. A CDP report presents a ﬁrm’s carbon-related activities and information. Technically, a company is assigned a score (which ranges from 0 to 5) based on the content of the information provided in the answer to a particular question in the carbon questionnaire. The ﬁnal score, CD, is a percentage that is equal to the total score earned, divided by the total score available. Thus, CD increases with the level of carbon disclosure, i.e. a higher CD value for a ﬁrm suggests that its carbon emissions and carbon-related strategies and actions are more transparent and visible than ﬁrms with lower CD. A negative coefﬁ- cient with CD is expected, which would support H1 (that a high level of carbon disclo- sure decreases the cost of capital). Following Botosan (1997), we tested the validity of the CD score by calculating correlations between the extent of disclosure and various ﬁrm-speciﬁc variables that have been shown in prior research to be signiﬁcant drivers of a ﬁrm’s level of disclo- sure. We expected that the extent of carbon disclosure would be positively associated with ﬁrm size (Botosan, 1997), ﬁnancial leverage (Botosan, 1997), and capital raising (Luo et al., 2012). Table 2 presents the correlation coefﬁcients between the extent of 198 He et al. Table 2. Correlation analysis of carbon disclosure scores and ﬁrm characteristics. Panel A: Univariate analysis SIZE LEV FIN *** ** Pearson correlation with CD 0.19 0.12 0.15 ** ** Spearman’s rho correlation with CD 0.19 0.12 0.16 Panel B: Multivariate analysis Intercept SIZE LEV FIN ** *** * ** coefﬁcient 25.9635 3.0106 12.1731 21.6269 t-statistic 2.41 3.12 1.90 2.22 p-value 0.017 0.002 0.059 0.028 adj. R 6.89% Prob. > F 0.001 Notes: This table contains Pearson correlation and Spearman’s rho correlation coefﬁcients between CD *** ** * (carbon disclosure score) and ﬁrm characteristics. indicate statistical signiﬁcance at the 1%, 5%, and 10% levels, respectively. SIZE is the natural logarithm of the market value of equity at the end of year t. LEV is the leverage ratio, which is deﬁned as the ratio of total debt divided by total assets. FIN is the amount of debt or equity capital raised by the ﬁrm, scaled by total assets at the end of year t. carbon disclosure and these ﬁrm characteristics. The ﬁndings in Table 2 reveal that ﬁrm size, ﬁnancial leverage, and ﬁrms’ raising of capital ownership are positively associated with carbon disclosure level. Overall, these ﬁndings conﬁrm the validity and reliability of the CDI. The other control variables in equation (1) are derived from prior research. Follow- ing Botosan (1997), the market model BETA, which is estimated using CRSP monthly data with a minimum of 30 out of 60 months’ worth of returns (Botosan and Plumlee, 2005), was included to control for systematic risk. Francis et al. (2008) found that earn- ings quality inﬂuenced the relationship between voluntary disclosure and the cost of capital. Hence, we included earnings quality (FD) in equation (1). FD is estimated as the absolute value of abnormal accruals, estimated based on the modiﬁed Jones model. Fama and French (1992) found that expected returns were negatively associated with ﬁrm size and positively associated with the book-to-market ratio. Thus, we included ﬁrm size (SIZE) and the market-to-book ratio (MB) in equation (1). SIZE is the natural logarithm of the market value of common equity at the end of year t. MB is the mar- ket-to-book ratio, deﬁned as the market value of equity, divided by the book value of equity at the end of year t. Firms in GHG-intensive sectors (such as utilities, energy, and materials ) have more carbon emissions and a higher propensity toward pollution (Bewley and Li, 2000; Clarkson et al., 2008) as a result of their inherent operation pro- cesses and are likely to be the targets of a wide variety of climate change regulations on the national, regional, and industry levels. Such environmental legislation may have a ﬁnancial effect and signiﬁcantly increase GHG-related liabilities and costs (Stanny and Ely, 2008). Therefore, we also included the variable INTENSITY, which is an indicator variable that equals 1 if the ﬁrm operates in a GHG-intensive industry and 0 otherwise. In all three equations, we included industry indicators to control for poten- tial industry effects. In equation (2), CD has the same deﬁnition as in equation (1). Following previ- ous literature (Clarkson, Li, Richardson, & Vasvari, 2011b), we used CP as a proxy for carbon performance, calculated as the inverse of total carbon emission per million dollars of sales turnover (net). Thus, CP is a measure of carbon efﬁciency and increases in carbon performance. A positive coefﬁcient on CP would support H2a, China Journal of Accounting Studies 199 and a negative coefﬁcient on CP would support H2b. Francis et al. (2008) found that ﬁrms with good earnings quality provide more expansive voluntary disclosures than ﬁrms with poor earnings quality. Thus, we included earnings quality (FD) in equation (2). Frankel, McNichols, and Wilson (1995) argued that ﬁrms raising capital in the public market have a greater propensity to make voluntary disclosures. We controlled for a ﬁrm’s ﬁnancing activities (FIN) by assessing the amount of debt or equity capi- tal raised by the ﬁrm during year t, scaled by total assets at the end of year t. Fol- lowing Dhaliwal et al. (2011), we also controlled for growth opportunities (TOBINQ), because ﬁrms in an expansionary period are more ﬁnancially constrained and have fewer resources for improving carbon performance and disclosure. On the other hand, growth ﬁrms also tend to have higher levels of information asymmetry, which may prompt managers to provide additional disclosures to attract potential investors (Dhal- iwal et al., 2011). Thus, the net effect of TOBINQ on carbon disclosure is unknown ex ante. TOBINQ is deﬁned as the market value of common equity plus the book value of preferred stock, book value of long-term debt, and current liabilities, scaled by the book value of total assets. Since debt holders demand greater disclosure so that they may monitor the ﬁrm’s ﬁnancial and operation activities (Leftwich, Watts, & Zimmerman, 1981), we included the debt ratio (LEV) in equation (2). LEV is the ratio of total debt divided by total assets. We also controlled for ﬁrm size (SIZE), because size captures various factors motivating ﬁrms to disclose carbon information, such as public pressure or ﬁnancial resources (Lang & Lundholm, 1993). In addition, there are some incentives for managers to enhance the liquidity of their ﬁrm’s stock via improved disclosure (Dhaliwal et al., 2011). Therefore, we included the effect of liquidity (LIQUIDITY) in equation (2). LIQUIDITY is the ratio of the number of shares traded in year t to the weighted total shares outstanding at the end of every month of year t. Prior empirical studies have explored the relationship between ﬁrms’ proﬁtability and their environmental disclosure (Aerts, Cormier, & Magnan, 2008; Bewley & Li, 2000; Clarkson et al., 2008; Magness, 2006). Thus, we controlled for return on assets (ROA) in equation (2). ROA is measured as the ratio of income before extraordinary items over total assets at the end of year t.A ﬁrm that operates in a GHG-intensive industry may have incentives to voluntarily disclose carbon infor- mation to prepare for possible future legislation to avoid the retroactive costs of regu- latory compliance (Al-Tuwaijri et al., 2004). Thus, we also included INTENSITY in equation (2). Skinner (1997) contended that ﬁrms facing a higher level of litigation risk (LITIGATION) have incentives to make voluntary disclosure to pre-empt potential lawsuits. Following Dhaliwal et al. (2011), we controlled for LITIGATION, which is an indicator variable that equals 1 if the ﬁrm operates in a high-litigation industry (Standard Industrial Classiﬁcation [SIC] codes of 2833-2836, 3570-3577, 3600-3674, 5200-5961, and 7370) and 0 otherwise (Francis, Philbrick, & Schipper, 1994; Mat- sumoto, 2002). Following Al-Tuwaijri et al. (2004), we included previous disclosure (CD )in i,t–1 equation (3). A ﬁrm’s prior carbon disclosures may represent a lower bound for current carbon performance. And investors’ expectations of carbon performance are condi- tioned on information provided by prior carbon disclosures (Al-Tuwaijri et al., 2004). If a ﬁrm expects poor carbon performance in the future, it may present full carbon information in advance to avoid ‘punishment’ in the future because it had withheld car- bon information. We expect the relationship between CD and CP to be signiﬁ- i,t–1 i,t cantly negative. Following Clarkson et al. (2011b) and Luo et al. (2012), we identiﬁed proxies for both a ﬁrm’s ﬁnancial resources and its management capabilities to examine 200 He et al. the determinants of carbon performance. We captured ﬁnancial resources using proﬁt- ability (ROA), ﬁnancing activities (FIN), and debt ratio (LEV) and used growth oppor- tunities (TOBINQ), ﬁrm size (SIZE), and market-to-book ratio (MB) to serve as proxies for unobservable management talent or capability. Following Dhaliwal et al. (2011)and Luo et al. (2012), we also included the variables INTENSITY and LITIGATION to control for the effect of potential regulatory risks, physical risks, and other risks. All other variables are as deﬁned earlier. Next, we considered the impact of carbon performance on the association between carbon disclosure and the cost of capital. Although ﬁrms may be motivated by a possi- ble reduction in the cost of equity capital when making decisions regarding carbon dis- closure, from the perspective of investors, carbon disclosure per se may not necessarily warrant a lower cost of equity capital (Dhaliwal et al., 2011). This is because corporate managers in poor-performing ﬁrms could also attempt to disclose more to achieve, restore, or maintain legitimacy (Laine, 2009; Suchman, 1995), and such disclosure obviously would not indicate good carbon performance (Bebbington, Larrinaga, & Moneva, 2008; Lindblom, 1993). A ﬁrm is a citizen of society and bound by social contracts. Thus, a ﬁrm is expected to carry out various activities considered desirable by the community in return for approval of its objectives and other rewards, and this ultimately guarantees its continued existence (Milne and Patten, 2002; Suchman, 1995). If a ﬁrm’s true social or environmental performance is below expectations, it would face a legitimacy threat. In such a case, the ﬁrm might disclose some information, instead of improving its performance, in a bid to alter the perception of the public. Patten (1992) concluded that it appears that, at least for environmental disclosures, threats to a ﬁrm’s legitimacy usually entice ﬁrms to include more information on social responsibility in their annual reports. Deegan and Rankin (1996) found a positive corre- lation between prosecution by Australian state environmental protection authorities and an increase in the level of environmental disclosure. Archel, Husillos, Larrinaga, and Spence (2009) found that ﬁrms in their study used social and environmental disclosures strategically to legitimize new production processes through the manipulation of social perceptions. Thus, we augmented equation (1) by adding a measure of a ﬁrm’s relative carbon performance (HICP) as well as the interaction between CD and HICP to equa- tion (1). HICP is an indicator variable that equals 1 if the ﬁrm’s carbon performance in year t is better than the sample median and 0 otherwise. The purpose of the interaction term is to determine whether carbon performance plays a role in the relationship between carbon disclosure and the cost of capital. All the other variables in the new system of equations are as deﬁned earlier. The following is the revised system of equa- tions: COST ¼ CD þ HICP þ CD HICP þ BETA þ FD þ SIZE þ MB i;t i;t i;t i;t i;t i;t i;t i;t i;t þ INTENSITY þ IND þ e ð1’Þ i;t i;t i;t CD ¼ CP þ FD þ FIN þ TOBINQ þ LEV þ SIZE þ LIQUIDITY i;t i;t i;t i;t i;t i;t i;t i;t þ ROA þ INTENSITY þ LITIGATION þ IND þ e ð2’Þ i;t i;t i;t i;t i;t CP ¼ CD þ FIN þ TOBINQ þ LEV þ SIZE þ MB þ ROA i;t i;t1 i;t i;t i;t i;t i;t i;t (3’) þ INTENSITY þ LITIGATION þ IND þ e i;t i;t i;t i;t China Journal of Accounting Studies 201 4. Empirical results 4.1. Descriptive statistics Table 3 provides descriptive statistics and results of t-tests for comparison of mean values of cost of capital. Table 4 depicts both the parametric and non-parametric, pairwise correlation coefﬁcients for the variables used in our tests. Table 3 shows that the mean cost of capital of all sample ﬁrms was 9.9%. When classiﬁed by the level of carbon disclosure, the cost of capital of the top 50% of ﬁrms (in terms of disclosure) (9.6%) was lower, although insigniﬁcantly so (t=1.03), than the bottom 50% of ﬁrms (10.2%), and that of the top 25% of ﬁrms (9%) was signiﬁcantly lower (t=2.12; p<0.05) than the lowest 25% of ﬁrms (10.4%) (Table 3, Line 1). This difference was also reﬂected in a signiﬁcantly negative correlation coefﬁcient for COST versus CD of –0.16 (Table 4), providing initial support for H1. The mean value for CD was 64.16%, and there was a signiﬁcant difference in CD between the top 50% and the bottom 50% of the sample ﬁrms (t=–18.05; p<0.000) and between the top 25% and bottom 25% of ﬁrms (t=–25.65, p<0.000; Table 3, line 2). This result suggests that the ﬁrms in the sample had different incentives and adopted different strategies for carbon disclosure. The mean value for CP was 0.07 (Table 3, line 3); this indicates that, on average, every 1 tonne of CO generated 0.07 million dollars of sales in the sample ﬁrms. However, there were signiﬁcant differences in carbon performance in different industries, with the lowest mean CP (0.0009) seen in the utilities industry and the highest mean CP (0.32) seen in the ﬁnancial industry. Because CP is a measure of carbon efﬁciency, the results (Table 3, line 3) show that ﬁrms in the top 50% in terms of disclosure tended to be less car- bon efﬁcient (CP=0.055) than the bottom 50% of ﬁrms (CP=0.084) (top 25% CP=0.44, bottom 25% CP=0.099). Table 4 (column 3 and 5) also presents a nega- tive relationship between carbon disclosure and carbon performance. This evidence suggests that ﬁrms with poor carbon performance disclosed more, which is consis- tent with the legitimacy theory. However, the negative correlations in both Tables 3 and 4 were not statistically signiﬁcant (p<0.15). The t-test results presented in Table 3 show that the relationships of CP, BETA, FD, TOBINQ, LEV, MB, and ROA with CD were not signiﬁcant, and FIN (t=−2.24; p<0.05) and SIZE (t=−2.84; p<0.05) were signiﬁcantly related to CD. This suggests that larger ﬁrms tended to disclose more, and ﬁrms raising capital also had a greater propensity to make voluntary carbon disclosures, which would reduce the cost of capi- tal. Note that LIQUIDITY was negatively associated with disclosure (t=1.42; and t=2.19; p<0.05), suggesting that managers with low-liquidity ﬁrms had some incentive to improve disclosure to enhance liquidity. This is consistent the with the signiﬁcant negative Spearman and Pearson correlation coefﬁcients between CD and LIQUIDITY shown in Table 4 ( p<0.10). The other results in Table 4 were generally consistent with our predictions or prior studies. For example, the cost of capital (COST) was signiﬁcantly and positively associated with systematic risk (BETA) (Botosan and Plumlee, 2005; Francis et al., 2008) and signiﬁcantly and negatively related to ﬁrm size (SIZE), earnings quality (FD), and market-to-book ratio (MB) (e.g. Fama and French, 1992; Francis et al., 2008). The negative coefﬁcient of carbon performance (CP) and INTENSITY (p<0.01) suggests that carbon efﬁciency of ﬁrms in a GHG-intensive industry was, on average, poorer than that of ﬁrms in a low carbon-intensive industry. 202 He et al. Table 3. Summary statistics and results of T-test for mean comparisons (classiﬁed by the level of carbon disclosure). CD (top 50%=1, bottom 50%=0) CD (top 25%=1, bottom 25%=0) t t CD =1 CD =0 t-value CD =1 CD =0 t-value t t t t Full Sample (n=88) (n=93) (difference) (n=48) (n=46) (difference) COST 0.099 0.096 0.102 1.03 0.090 0.104 2.12 CD 64.16 76.77 52.23 –18.05 82.94 44.58 –25.65 CD 57.21 67.56 47.42 –9.00 71.83 40.26 –10.32 t–1 CP 0.070 0.055 0.084 1.21 0.044 0.099 1.51 BETA 1.098 1.100 1.097 –0.04 0.967 1.078 1.02 FD –0.158 –0.168 –0.148 0.65 –0.157 –0.138 0.46 FIN 0.072 0.091 0.054 –2.24 0.086 0.046 –1.76 TOBINQ 1.591 1.636 1.549 –0.77 1.710 1.527 –1.12 LEV 0.597 0.610 0.586 –0.92 0.614 0.560 –1.52 SIZE 9.755 9.887 9.631 –1.49 10.125 9.435 –2.84 MB 3.300 3.570 3.043 –1.10 3.789 3.090 –0.99 LIQUIDITY 2.470 2.305 2.626 1.42 2.061 2.677 2.19 ROA 0.066 0.070 0.062 –1.04 0.075 0.063 –1.06 INTENSITY The number of low (high) INTENSITY ﬁrms is 133 (48). t t LITIGATION The number of low (high) LITIGATION ﬁrms is 134 (47). t t All continuous variables are winsorized at the 1st and 99th percentiles. Variable deﬁnitions: COST is the implied cost of equity capital in year t estimated by the PEG formula in Easton (2004), which was recommended by Botosan and Plumlee (2005). CD is a measure of the ﬁrm’s voluntary carbon disclosure in year t presented by CDP, which is an inter- national collaboration of institutional investors concerned about the business implications of climate change. CP is the ﬁrm’s carbon performance, measured as the inverse of total carbon emission per million dollars of sales turnover (net); it was also used in Clarkson et al. (2008). BETA is market-model beta calculated from the monthly CRSP stock returns during year t−5 to year t. FD is a proxy for earnings quality, measured as the absolute value of abnormal accruals, estimated using the modiﬁed Jones model. FIN is the amount t t of debt or equity capital raised by the ﬁrm, scaled by total assets at the end of year t. TOBINQ is the market value of common equity plus the book value of preferred stock, book value of long-term debt, and current liability, scaled by the book value of total assets. LEV is the leverage ratio, which is deﬁned as the ratio of total debt divided by total assets. SIZE is the natural logarithm of the market value of equity at the end of year tMB is the market-to-book ratio, deﬁned as the market value of equity divided by the book value t t of equity at the end of year t. LIQUIDITY is the ratio of the number of shares traded in year t to the weighted total shares outstanding at the end of every month of year t. ROA t t is the total return on assets, measured as the ratio of income before extraordinary items over total assets at the end of year t. INTENSITY is an indicator variable that equals 1 if the ﬁrm operates in a GHG-intensive industry and 0 otherwise; it was also used in Matsumura et al. (2010). LITIGATION is an indicator variable that equals 1 if the ﬁrm operates in a high-litigation industry (SIC codes of 2833-2836, 3570-3577, 3600-3674, 5200-5961, and 7370) and 0 otherwise. China Journal of Accounting Studies 203 Table 4. Spearman/Pearson correlation coefﬁcients. COST CD CD CP BETA FD FIN TOBINQ t t t–1 t t t t t ** ** *** COST 1.00 –0.16 –0.17 0.03 0.42 –0.08 0.01 –0.12 ** *** ** CD –0.16 1.00 0.70 –0.12 –0.06 –0.05 0.16 0.01 ** *** ** CD –0.18 0.70 1.00 –0.01 –0.05 –0.01 0.17 0.10 t–1 ** *** *** CP 0.03 –0.10 –0.07 1.00 0.18 0.70 –0.03 0.22 *** ** BETA 0.42 –0.07 –0.05 0.06 1.00 0.02 –0.04 –0.15 *** FD –0.10 –0.01 0.00 0.31 0.02 1.00 –0.05 0.05 ** ** *** FIN 0.01 0.15 0.15 –0.04 –0.04 0.02 1.00 0.27 ** *** ** TOBINQ –0.17 0.05 0.10 –0.08 –0.20 0.10 0.18 1.00 *** LEV 0.03 0.12 0.07 0.09 0.08 0.03 0.09 –0.43 *** *** *** *** *** SIZE –0.20 0.19 0.25 0.05 –0.25 –0.03 –0.09 0.21 ** * ** *** MB –0.16 0.10 0.14 –0.06 –0.17 0.04 0.08 0.50 *** ** LIQUIDITY 0.42 –0.15 –0.09 0.01 0.56 –0.02 0.11 –0.07 *** *** *** ROA –0.24 0.08 0.11 0.00 –0.28 0.01 0.06 0.68 *** *** *** INTENSITY –0.01 0.07 0.03 –0.25 –0.12 –0.50 –0.03 –0.19 *** LITIGATION –0.01 –0.03 0.08 0.02 0.09 0.11 0.04 0.30 LEV SIZE MB LIQUIDITY ROA INTENSITY LITIGATION t t t t t t t * ** *** *** COST –0.02 –0.14 –0.16 0.34 –0.23 –0.02 0.01 ** * CD 0.12 0.19 0.06 –0.14 0.03 0.07 0.00 *** * * CD 0.07 0.21 0.13 –0.14 0.08 0.03 0.08 t–1 * ** ** *** *** CP –0.14 0.09 0.19 0.11 0.19 –0.67 0.31 ** * *** *** * BETA –0.01 –0.18 –0.14 0.53 –0.24 –0.12 0.08 *** FD –0.01 0.02 0.08 0.01 0.01 –0.56 0.12 *** FIN 0.09 0.05 0.28 0.05 0.11 0.06 –0.01 *** *** *** *** *** *** TOBINQ –0.42 0.22 0.86 –0.07 0.72 –0.21 0.29 ** *** *** LEV 1.00 –0.17 0.00 –0.07 –0.44 0.03 –0.35 ** *** *** *** SIZE –0.15 1.00 0.21 –0.43 0.32 –0.06 0.08 *** ** *** *** * MB 0.28 0.07 1.00 –0.15 0.57 –0.22 0.13 *** * *** *** LIQUIDITY –0.07 –0.39 –0.13 1.00 –0.20 –0.03 0.20 *** *** *** * ** ** ROA –0.44 0.33 0.31 –0.13 1.00 –0.18 0.16 ** *** *** INTENSITY 0.03 –0.06 –0.19 –0.04 –0.20 1.00 –0.33 *** ** ** *** LITIGATION –0.38 0.09 –0.03 0.19 0.19 –0.33 1.00 t 204 He et al. Table 4 (Continued) * ** *** , , Indicate that the estimated coefﬁcient is statistically signiﬁcant at the 10%, 5%, and 1% levels, respectively. n=181. All continuous variables are winsorized at the 1st and 99th percentiles. The Spearman (Pearson) correlations are above (below) the diagonal. Variable deﬁnitions: COST is the implied cost of equity capital in year t estimated by the PEG formula in Easton (2004), which was recommended by Botosan and Plumlee (2005). CD is a measure of the ﬁrm’s voluntary carbon disclosure in year t presented by CDP, which is an international collaboration of institutional investors concerned about the business implications of climate change. CP is the ﬁrm’s carbon performance, measured as the inverse of total carbon emissions per million dollars of sales turnover (net); it was also used in Clarkson et al. (2008). BETA is market-model beta calculated from the monthly CRSP stock returns during year t−5 to year t. FD is a proxy for earnings quality, measured as the absolute value of abnormal accruals, estimated using the modiﬁed Jones model. FIN is the amount of debt or equity capital raised by the ﬁrm, scaled by total assets at the end of year t. TOBINQ is the market value of common equity plus the book value of t t preferred stock, book value of long-term debt, and current liability, scaled by the book value of total assets. LEV is the leverage ratio, which is deﬁned as the ratio of total debt divided by total assets; SIZE is the natural logarithm of the market value of equity at the end of year t. MB is the market-to-book ratio, deﬁned as the market value of equity t t divided by book value of equity at the end of year t. LIQUIDITY is the ratio of the number of shares traded in year t to the weighted total shares outstanding at the end of every month of year t. ROA is the total return on assets, measured as the ratio of income before extraordinary items over total assets at the end of year t. INTENSITY is an indicator t t variable that equals 1 if the ﬁrm operates in a GHG-intensive industry and 0 otherwise; it was also used in Matsumura et al. (2010). LITIGATION is an indicator variable that equals 1 if the ﬁrm operates in a high-litigation industry (SIC codes of 2833-2836, 3570-3577, 3600-3674, 5200-5961, and 7370) and 0 otherwise. China Journal of Accounting Studies 205 4.2. The endogeneity problem Al-Tuwaijri et al. (2004) argued that managers’ overall strategies likely affect carbon disclosure, carbon performance, and the cost of capital simultaneously and thus an OLS regression analysis is inappropriate. Hence, we examined the endogenous relationships among the three dependent variables (COST, CD, and CP) by employ- ing a Hausman (1978) test, and we rejected the null hypothesis of no endogeneity with respect to CP in equation (2) (t=–6.96, p<0.000). This result suggests that OLS estimators are potentially biased and inconsistent. We continued our analysis using two-stage least squares (2SLS) and 3SLS simultaneous equation models to control for endogeneity to obtain asymptotically unbiased results. We used the Haus- man (1978) speciﬁcation test to compare 2SLS estimates with 3SLS estimates; the results were largely similar, but 3SLS was more efﬁcient than 2SLS. Thus, here we report only 3SLS results. Our simultaneous estimation of the parameters using the 3SLS model incorporated the available information from all the equations. After sys- tem estimation with 3SLS, we used the procedures of the Hansen-Sargan test to report an over-identiﬁcation statistic. As indicated by Davidson and MacKinnon (2004, p. 532), a Hansen-Sargan test of the over-identifying restrictions is based on the 3SLS criterion function evaluated at the 3SLS point and interval parameter esti- mates. The result shows that the Hansen-Sargan over-identiﬁcation statistic was 17.899 and insigniﬁcant at the 5% level. This result conﬁrms the validity of the instruments. 4.3. Regression analysis (three-stage least squares) Table 5 reports the results of our 3SLS simultaneous equation model. Following the example of Dhaliwal et al. (2011), we created a new variable, HICP, and an interaction term, CD*HICP, to capture the impact of carbon disclosure on the cost of capital conditional on carbon performance. We report the results from both sys- tems of equations, and each system comprises three equations: one system of equations excludes (left side of Table 5) and the other includes the variable of HICP and the interaction term CD*HICP (right side of Table 5) in equation (1). Table 5 shows, without the interaction term CD*HICP in equation (1), that the coefﬁcient of CD was −0.0004 at the 10% level of signiﬁcance (t=–1.839), sug- gesting a negative relationship between carbon disclosure and the cost of capital (COST), which supports hypothesis H1. This result is also consistent with the ﬁnd- ings of Dhaliwal et al. (2011). Table 5 also shows a negative coefﬁcient for HICP, suggesting an inverse association between carbon performance and the cost of cap- ital. Next, we examined the impact of the interaction between CD and HICP; the coefﬁcient of CD was –0.001 at a higher level of signiﬁcance (p<0.05) (t=–2.485). However, the interaction term CD*HICP was signiﬁcantly positive (p<0.05). This result suggests that the negative relationship between the cost of capital and car- bon disclosure is weaker in ﬁrms with good carbon performance. This is probably because these ﬁrms disclose less information than poor performers. The result sup- ports hypothesis H1a and highlights the importance of carbon disclosure for inves- tors. It is also consistent with the legitimacy theory that ﬁrms with poor carbon performance have stronger incentives to disclose more, because these ﬁrms are more likely to have the problem of threatened legitimacy. The results of t-tests in Table 3 and regression analysis in equation (2) of Table 5 conﬁrm this interpreta- 206 He et al. Table 5. Three-stage least squares (3SLS) regression results (R – total emission) and PEG coefﬁcients (t-statistics). Dependent variables Dependent variables Independent variables COST [Eq.(1)] CD [Eq.(2)] CP [Eq.(3)] COST [Eq.(1’)] CD [Eq.(2’)] CP [Eq.(3’)] t t t t t t * ** CD –0.0004 –0.001 (–1.839) (–2.485) *** *** CD –0.001 –0.002 t–1 (–5.215) (–7.234) *** *** CP –369.860 –350.704 (–2.886) (–9.970) ** HICP –0.064 (–2.427) * ** CD HICP 0.001 t t (2.436) *** *** BETA 0.024 0.021 (4.548) (4.185) FD –0.019 –0.009 –0.017 9.266 (–1.345) (–0.001) (–1.116) (1.419) FIN 3.103 –0.019 3.111 0.001 (0.084) (–0.209) (0.145) (0.008) TOBINQ –1.581 –0.001 –1.992 –0.008 (–0.201) (–0.050) (–0.430) (–0.399) ** ** *** ** LEV 72.754 0.186 68.800 0.162 (1.973) (2.382) (3.675) (2.061) ** ** SIZE –0.005 2.155 0.005 –0.006 2.542 0.008 (–2.002) (0.549) (0.537) (–2.485) (1.094) (0.798) MB –0.001 –0.000 –0.001 0.002 (–1.327) (–0.003) (–1.360) (1.099) LIQUIDITY –1.455 –0.466 (–1.292) (–0.575) ROA 77.532 0.182 89.030 0.171 (0.662) (0.663) (1.328) (0.621) *** INTENSITY –0.030 –2.526 –0.010 0.204 –4.932 0.015 (–1.487) (–0.056) (–0.117) (6.384) (–0.246) (0.187) LITIGATION –14.309 –0.034 –13.130 –0.033 (–1.159) (–1.248) (–1.928) (–1.204) *** _cons 0.153 0.000 –0.085 0.000 2.088 –0.073 Industry indicators Yes Yes Yes Yes Yes Yes Adjusted system R 0.773 0.763 * ** *** , , Indicate that the estimated coefﬁcient is statistically signiﬁcant at the 10%, 5%, and 1% levels, respec- tively. n=181. All continuous variables are winsorized at the 1st and 99th percentiles. Variable deﬁnitions: COST is the implied cost of equity capital in year t estimated by the PEG formula in Easton (2004), which was recommended by Botosan and Plumlee (2005). CD is a measure of the ﬁrm’s voluntary carbon disclo- sure in year t presented by CDP, which is an international collaboration of institutional investors concerned about the business implications of climate change. CP is the ﬁrm’s carbon performance, measured as the inverse of total carbon emissions per million dollars of sales turnover (net); it was also used in Clarkson et al. (2008). HICP is an indicator variable that equals 1 if the ﬁrm’s CP in year t is higher than the sample median and 0 otherwise. BETA is market-model beta calculated from the monthly CRSP stock returns during year t − 5 to year t. FD is a proxy for earnings quality, measured as the absolute value of abnormal accruals, esti- mated based on the modiﬁed Jones model. FIN is the amount of debt or equity capital raised by the ﬁrm, scaled by total assets at the end of year t. TOBINQ is the market value of common equity plus the book value of preferred stock, book value of long-term debt, and current liability, scaled by the book value of total assets. LEV is the leverage ratio, which is deﬁned as the ratio of total debt divided by total assets. SIZE is t t the natural logarithm of the market value of equity at the end of year t. MB is the market-to-book ratio, deﬁned as the market value of equity divided by book value of equity at the end of year t. LIQUIDITY is the ratio of the number of shares traded in year t to the weighted total shares outstanding at the end of every China Journal of Accounting Studies 207 month of year t. ROA is the total return on assets, measured as the ratio of income before extraordinary items over total assets at the end of year t. INTENSITY is an indicator variable that equals 1 if the ﬁrm operates in a GHG-intensive industry and 0 otherwise; it was also used in Matsumura et al. (2010). LITIGATION is an indicator variable that equals 1 if the ﬁrm operates in a high-litigation industry (SIC codes of 2833-2836, 3570-3577, 3600-3674, 5200-5961, and 7370) and 0 otherwise. tion. On the other hand, it is also possible that the current format and content of the CDP report is not adequate for users to distinguish between good and poor performers. Thus, further efforts should be made to standardize the format and contents of carbon disclosure reports. The coefﬁcient estimates of the control vari- ables in equation (1) were generally consistent with the univariate analysis in Tables 3 and 4. As predicted, systematic risk (BETA) and carbon intensity of an industry (INTENSITY) were positively associated and ﬁrm size (SIZE) and earnings quality (FD) were negatively associated with the cost of capital. However, only the coefﬁcients of BETA and SIZE were signiﬁcant in both systems of equations. MB was not signiﬁcant in either system, and INTENSITY was signiﬁcant only in the system with HICP and CD*HICP. Equation (2) in Table 5 examines the determinants of carbon disclosure and shows that the coefﬁcient of CP was signiﬁcantly negative (–369.860 and –350.704; p<0.01) in both systems of equations. The result is consistent with legitimacy theory and sup- ports hypothesis H2b, suggesting that poorer performers disclose more. This evidence provides an explanation for the result obtained from equation (1) that the negative rela- tionship between carbon disclosure and cost of capital is more pronounced in ﬁrms with poorer performance. For the control variables in equation (2), the debt ratio (LEV) was signiﬁcantly positively associated with CD in both systems of equations. This is consistent with Leftwich et al. (1981), who argued that debt servicing plays a monitor- ing role and that debt holders demand greater disclosure. Litigation risk (LITIGATION) was negatively associated with CD (p<0.10), but signiﬁcantly only when HICP and CD*HICP were included in equation (1). This result is consistent with Dhaliwal et al. (2011) but inconsistent with Skinner (1997). All other control variables were insigniﬁ- cant. Equation (3) speciﬁes the determinants of carbon performance. The signiﬁcantly negative relationship between CP and past carbon disclosure (CD )(p<0.01 in both i,t–1 systems) suggests that poor carbon performers may present full carbon information in advance to avoid negative surprises and future market punishment caused by withhold- ing carbon information (Matsumoto, 2002). All other control variables in equation (3) were insigniﬁcant, except for debt ratio (LEV). In sum, the 3SLS results suggest two signiﬁcant relationships among dependent variables. First, carbon disclosure was signiﬁcantly and negatively associated with the cost of capital, suggesting that the market rewards ﬁrms with higher levels of carbon transparency. However, the negative relationship between carbon disclosure and the cost of capital was more pronounced in poor carbon performers, suggesting that without sufﬁcient disclosure, market participants may not fully understand a ﬁrm’s good carbon performance. Second, there is an inverse relationship between carbon disclosure and performance, supporting the proposition that poor carbon per- formers have more incentive to disclose carbon information as a result of the pres- sure of legitimacy. This may explain why disclosure is less effective in ﬁrms with good performance. 208 He et al. 4.4. Sensitivity analysis 4.4.1. Alternative cost of capital measure We used the PEG ratios developed by Easton (2004) to measure the implied cost of capital because extant tests show that this proxy has better construct validity than other measures (Botosan & Plumlee, 2005). Notwithstanding this evidence, we investigated the sensitivity of our main results to another ex ante cost of capital metrics: debt rating (an ex ante cost of debt measure). We used debt rating because this allowed us to side- step the debate about which is the ‘best’ ex ante cost-of-equity proxy (Francis et al., 2008), and because Sengupta (1998) concluded that disclosure quality (as measured by the scores from the Association of Investment Management and Research) is negatively associated with cost of debt. For tests of debt rating, we used S&P credit ratings (which range from AAA [highest quality] to D [default]), as available on Compustat. We employed the numeric transformation rules of S&P credit ratings in Francis et al. (2008) and excluded those ﬁrms that had no debt ratings. The results are shown in Table 6. Panel A of Table 6 shows that the mean of cost of debt capital of all sample ﬁrms was 9.062, which is close to the cost of equity capital (9.90) (Table 3). Also, there was an insigniﬁcant (signiﬁcant) difference in the cost of debt between the top 50% (25%) (i.e. high-disclosure) of ﬁrms and the bottom 50% (25%) (i.e. low- disclosure) of ﬁrms. This result is identical to the result in our main analysis in Table 3. Panel B of Table 6 shows the results of the 3SLS simultaneous equation model. In equation (1), both with and without HICP, CD was always signiﬁcantly and neg- atively associated with the cost of capital (p<0.05), which is consistent with the observations using PEG ratios as a proxy for the cost of capital (Table 5). How- ever, HICP and CD*HICP were no longer signiﬁcant, suggesting that carbon disclo- sure had a similar impact on the cost of debt in high- and low-performance ﬁrms. The results of all other control variables were also consistent with those observed using PEG ratios, with two exceptions: the indicator variable of a carbon-intensive industry (INTENSITY) was no longer statistically signiﬁcant, and market-to-book ratio (MB) was signiﬁcantly negative, consistent with the results of Dhaliwal et al. (2011). The results of equation (2) were generally consistent with those presented in our main analysis, except that the coefﬁcient of CP was statistically signiﬁcant (t=−1.864, p<0.10) only in the system with HICP, suggesting that carbon disclosure is less sensi- tive to carbon performance in the case of cost of debt. This provides evidence of the importance for controlling for endogeneity in this research setting (Al-Tuwaijri et al., 2004) and validates our adoption of a joint-estimation research design. In equation (3), we observed results identical to those presented in our main analysis, except that the coefﬁcient of proﬁtability (ROA) was statistically signiﬁcantly positive (p<0.01). Taken as a whole, our main inferences held when we replaced cost of equity with cost of debt. 4.4.2. Alternative carbon performance proxy In the prior analysis, we used the inverse of total carbon emissions per million dol- lars of sales turnover (net) to measure carbon performance. Total carbon emissions consist of three deﬁned ‘scopes’ of GHG emissions: scopes 1, 2, and 3. Scope 1 China Journal of Accounting Studies 209 Table 6. Three-stage least squares (3SLS) regression results (credit ratings –total emission) and coefﬁcients (t-statistics). Panel A: Mean comparison CD (top 50%=1, low 50%=0) CD (top 25%=1, low 25%=0) t t Full sample CD =1 CD =0 t-value CD =1 CD =0 t-value t t t t (n=33) (n=32) (difference) (n=15) (n=15) (difference) COST 9.062 8.545 9.594 1.57 8.333 10 2.14 Panel B: Three-stage least squares (3SLS) results Independent variables Dependent variables Dependent variables COST [Eq.(1)] CD [Eq.(2)] CP [Eq.(3)] COST [Eq.(1’)] CD [Eq.(2’)] CP [Eq.(3’)] t t t t t t ** ** CD –0.047 –0.057 (–2.389) (–2.307) *** *** CD –0.002 –0.002 t–1 (–2.729) (–2.871) CP –62.639 –73.711 (–1.571) (–1.864) HICP –2.620 (–1.103) CD HICP 0.012 t t (0.327) *** *** BETA 1.690 1.781 (3.639) (4.049) FD –1.940 –15.010 –1.365 –14.818 (–1.372) (–1.204) (–1.022) (–1.193) FIN 38.012 0.194 39.591 0.168 (1.186) (0.890) (1.226) (0.767) TOBINQ –1.052 –1.844 –0.045 (–0.184) (–0.323) (–0.906) *** *** LEV 26.281 0.460 29.265 0.476 (1.290) (2.781) (1.432) (2.878) *** *** SIZE –0.834 1.288 –0.008 –0.803 1.220 –0.008 (–3.080) (0.539) (–0.565) (–3.156) (0.507) (–0.564) ** ** MB –0.230 –0.007 –0.257 –0.008 (–2.098) (–0.598) (–2.485) (–0.673) (Continued) 210 He et al. Table 6. (Continued). Panel B: Three-stage least squares (3SLS) results Independent variables Dependent variables Dependent variables COST [Eq.(1)] CD [Eq.(2)] CP [Eq.(3)] COST [Eq.(1’)] CD [Eq.(2’)] CP [Eq.(3’)] t t t t t t LIQUIDITY –0.092 –0.085 (–0.707) (–0.656) *** *** ROA –30.349 1.348 –15.158 1.384 (–0.316) (2.991) (–0.158) (3.051) INTENSITY 0.386 32.622 –0.038 0.375 31.242 –0.129 (0.226) (1.334) (–0.444) (0.219) (1.269) (–0.704) LITIGATION –5.326 (–0.985) –5.252 0.018 (–0.938) 0.014 (–0.919) (0.509) *** *** _cons 18.197 0.000 –0.083 19.438 0.000 0.000 Industry indicators Yes Yes Yes Yes Yes Yes Adjusted system R 0.867 0.886 * ** *** , , Indicate that the estimated coefﬁcient is statistically signiﬁcant at the 10%, 5%, and 1% levels, respectively. n=65. All continuous variables are winsorized at the 2nd and 98th percentiles. Variable deﬁnitions: COST is the implied cost of capital in year t derived from the numeric transformation rules of S&P credit ratings in Francis et al. (2008). CD is a measure of the ﬁrm’s voluntary carbon disclosure in year t presented by CDP, which is an international collaboration of institutional investors concerned about the busi- ness implications of climate change; CP is the ﬁrm’s carbon performance, measured as the inverse of total carbon emissions per million dollars of sales turnover (net); this was also used in Clarkson et al. (2008). HICP is indicator variable that equals 1 if the ﬁrm’s CP in year t is higher than the sample median and 0 otherwise. BETA is market-model t t beta calculated from the month CRSP stock returns during year t−5 to year t. FD is a proxy for earnings quality, measured as the absolute value of abnormal accruals, estimated based on the modiﬁed Jones model. FIN is the amount of debt or equity capital raised by the ﬁrm, scaled by total assets at the end of year t. TOBINQ is the market value of t t common equity plus the book value of preferred stock, book value of long-term debt, and current liability, scaled by the book value of total assets. LEV is the leverage ratio, which is deﬁned as the ratio of total debt divided by total assets. SIZE is the natural logarithm of the market value of equity at the end of year t. MB is the market-to-book ratio, t t deﬁned as the market value of equity divided by book value of equity at the end of year t. LIQUIDITY is the ratio of the number of shares traded in year t to the weighted total shares outstanding at the end of every month of year t. ROA is the total return on assets, measured as the ratio of income before extraordinary items over total assets at the end of year t. INTENSITY is an indicator variable that equals 1 if the ﬁrm operates in a GHG-intensive industry and 0 otherwise; it was also used in Matsumura et al. (2010). LITIGA- TION is and indicator variable that equals 1 if the ﬁrm operates in a high-litigation industry (SIC codes of 2833-2836, 3570-3577, 3600-3674, 5200-5961, and 7370) and 0 other- wise. China Journal of Accounting Studies 211 Table 7. Three-stage least squares (3SLS) results (alternative carbon performance proxy) and coefﬁcients (t-statistics). Dependent variables Dependent variables Independent variables COST [Eq.(1)] CD [Eq.(2)] CP [Eq.(3)] COST [Eq.(1’)] CD [Eq.(2’)] CP [Eq.(3’)] t t t t t t *** *** CD –0.001 –0.001 (–2.962) (–3.411) *** *** CD –0.093 –0.083 t-1 (–8.166) (–7.922) *** *** CP –5.359 –8.106 (–3.186) (–5.995) *** HICP –0.077 (–3.064) * *** CD HICP 0.001 t t (3.054) *** *** BETA 0.020 0.018 (3.909) (3.522) FD –0.016 1.188 –0.013 –3.868 (–1.164) (0.159) (–0.877) (–0.542) FIN 14.525 0.841 13.186 1.020 (0.645) (0.215) (0.554) (0.263) TOBINQ 1.834 0.459 3.877 0.350 (0.359) (0.514) (0.726) (0.396) LEV 13.224 1.605 16.966 0.965 (0.686) (0.459) (0.833) (0.280) * ** SIZE –0.004 1.145 0.125 –0.006 0.449 0.068 (–1.873) (0.446) (0.287) (–2.528) (0.165) (0.155) MB –0.001 0.023 –0.001 0.065 (–1.313) (0.254) (–1.362) (0.769) LIQUIDITY –1.403 –1.067 (–1.656) (–1.278) ROA –0.696 –2.503 –16.097 –2.462 (–0.010) (–0.207) (–0.216) (–0.205) *** ** *** *** *** INTENSITY 0.139 –73.508 –14.416 0.184 –116.960 3.271 (5.903) (–2.424) (–5.900) (6.245) (–4.440) (0.699) LITIGATION –1.742 0.034 –0.907 0.257 (–0.245) (0.028) (–0.120) (0.211) *** *** *** _cons 0.000 121.505 17.321 0.000 165.478 0.000 Industry indicators Yes Yes Yes Yes Yes Yes Adjusted system R 0.701 0.705 * ** *** , , Indicate that the estimated coefﬁcient is statistically signiﬁcant at the 10%, 5%, and 1% levels, respec- tively. n=181. All continuous variables are winsorized at the 1st and 99th percentiles. Variable Deﬁnitions: COST is the implied cost of equity capital in year t estimated by the PEG formula in Easton (2004), which was recommended by Botosan and Plumlee (2005). CD is a measure of the ﬁrm’s voluntary carbon disclo- sure in year t presented by CDP, which is an international collaboration of institutional investors concerned about the business implications of climate change. CP is the ﬁrm’s carbon performance, measured as the inverse of scope 1 emissions per million dollars of sales turnover (net). HICP is indicator variable that equals 1 if the ﬁrm’s CP in year t is higher than the sample median, and 0 otherwise. BETA is market-model beta calculated from the month CRSP stock returns during year t−5 to year t. FD is a proxy for earnings quality, measured as the absolute value of abnormal accruals, estimated based on the modiﬁed Jones model. FIN is the amount of debt or equity capital raised by the ﬁrm, scaled by total assets at the end of year t. TOBINQ is the market value of common equity plus the book value of preferred stock, book value of long- term debt, and current liability, scaled by the book value of total assets. LEV is the leverage ratio, which is deﬁned as the ratio of total debt divided by total assets. SIZE is the natural logarithm of the market value of equity at the end of year t. MB is the market-to-book ratio, deﬁned as the market value of equity divided by book value of equity at the end of year t. LIQUIDITY is the ratio of the number of shares traded in year t to the weighted total shares outstanding at the end of every month of year t. ROA is the total return on assets, measured as the ratio of income before extraordinary items over total assets at the end of year t. INTENSITY t 212 He et al. is an indicator variable that equals 1 if the ﬁrm operates in a GHG-intensive industry and 0 otherwise; it was also used in Matsumura et al. (2010). LITIGATION is an indicator variable that equals 1 if the ﬁrm operates in a high-litigation industry (SIC codes of 2833-2836, 3570-3577, 3600-3674, 5200-5961, and 7370) and 0 otherwise. consists of direct emissions that come from sources that a ﬁrm owns or controls. This generally includes emissions from fossil fuel combustion for electricity, heat or steam generation, production processes for cement and steel manufacturing, transpor- tation of company-owned vehicles or aeroplanes, and fugitive emissions, such as refrigerants and methane (Kolk et al., 2008; Phillips, 2004). Scope 2 emissions are referred to as indirect emissions, which are generated from activities necessary to support production activities such as those purchased from an off-site facility, for example, electricity or steam. Scope 3 emissions are harder to deﬁne as they have a broader reach and are more all-encompassing. They are brought about by activi- ties that are part of the life cycle of the product and are created by employees, sup- pliers, customers, and contractors; for example, employees driving to work, activities embedded in purchased supplies or inventory, or transporting and disposal of prod- ucts. Scope 1 emissions are the most widely reported, because a ﬁrm can be held directly accountable for such emissions. In 2010, direct emissions (scope 1) repre- sented 1.54 billion tonnes of CO -equivalent emissions, or 84% of total emissions reported (PwC & CDP, 2010). On the other hand, indirect emissions (scope 2) come from sources where the point of release is not within the ﬁrm itself, but either upstream or downstream in the supply chain; scope 3 emissions are still not well-deﬁned, so many ﬁrms do not disclose this information. Thus, we also used the inverse of scope 1 emissions per million dollars of sales turnover (net) to mea- sure carbon performance, because there is a concern that scopes 2 and 3 emissions may be, to some extent, incomparable between ﬁrms in different sectors. Other model speciﬁcations were the same. The results are presented in Table 7 and are virtually the same as those reported in Table 5. 4.4.3. Past carbon disclosure proxy Dhaliwal et al. (2011) argued that endogeneity and self-selection issues can arise if a study examines a contemporaneous relationship between social responsibility dis- closure and the cost of capital. On the one hand, if the disclosure is motivated by a ﬁrm’s desire to reduce its high cost of capital, then researchers should observe a positive relationship between disclosure and the cost of capital. On the other hand, if disclosure leads to a lower cost of equity capital, there should be a negative rela- tionship between these factors. Therefore, the contemporaneous relationship between disclosure and the cost of capital could be ambiguous. To address the potential for endogeneity and self-selection issues, we employed a lead-lag approach in this sen- sitivity analysis, i.e. we replaced carbon disclosure (CD ) with prior carbon disclo- i,t sure (CD ) in equation (1) in our system of equations and kept the other i,t–1 speciﬁcations unchanged. The results are shown in Table 8 and are generally the same as those in Table 5, except that the coefﬁcient of carbon disclosure (CD ) i,t–1 was signiﬁcantly negative (p<0.05) only in the system that included HICP and CD*HICP. This result is consistent with Dhaliwal et al. (2011) and validates the interaction effect between carbon disclosure and carbon performance on the cost of capital in our empirical analysis. In sum, the use of a lead-lag approach did not alter the main inferences obtained from previous analyses. China Journal of Accounting Studies 213 Table 8. Three-stage least squares (3SLS) regression results (prior carbon disclosure proxy) and coefﬁcients (t-statistics). Dependent variables Dependent variables Independent variables COST [Eq.(1)] CD [Eq.(2)] CP [Eq.(3)] COST [Eq.(1’)] CD [Eq.(2’)] CP [Eq.(3’)] t t t t t t CD *** ** *** CD –0.0002 –0.001 –0.0004 –0.003 t–1 (–1.365) (–5.039) (–2.083) (–7.333) *** *** CP –396.544 –165.751 (–3.093) (–4.797) HICP –0.024 (–1.466) * * CD HICP 0.0004 t t (1.667) *** *** BETA 0.024 0.022 (4.568) (4.259) FD –0.020 1.601 –0.021 4.009 (–1.423) (0.166) (–1.371) (0.597) FIN 1.081 –0.020 13.944 0.017 (0.029) (–0.221) (0.962) (0.184) TOBINQ –1.485 –0.002 –1.494 –0.001 (–0.189) (–0.088) (-0.477) (–0.058) ** ** *** ** LEV 77.330 0.181 38.584 0.203 (2.098) (2.322) (2.915) (2.427) ** ** * SIZE –0.005 2.335 0.005 –0.006 2.841 0.012 (–2.271) (0.596) (0.516) (–2.567) (1.790) (1.217) MB –0.001 0.000 –0.001 –0.000 (–1.372) (0.153) (–1.426) (–0.020) LIQUIDITY –0.993 -0.729 (–0.880) (-0.872) ROA 82.322 0.179 47.520 0.197 (0.703) (0.652) (1.038) (0.713) *** INTENSITY 0.110 –0.240 –0.093 –0.030 13.691 0.015 (4.958) (–0.005) (–0.894) (–1.508) (0.713) (0.179) LITIGATION –15.213 –0.035 –5.915 –0.029 (–1.234) (–1.254) (–1.259) (–1.051) *** _cons 0.000 0.000 0.000 0.160 0.000 –0.099 Industry indicators Yes Yes Yes Yes Yes Yes Adjusted system R 0.773 0.754 * ** *** , , Indicate that the estimated coefﬁcient is statistically signiﬁcant at the 10%, 5%, and 1% levels, st th respectively. n=181. All continuous variables are winsorized at the 1 and 99 percentiles. Variable deﬁni- tions: COST is the implied cost of equity capital in year t estimated by the PEG formula in Easton (2004), which was recommended by Botosan and Plumlee (2005). CD is a measure of the ﬁrm’s volun- tary carbon disclosure in year t presented by CDP, which is an international collaboration of institutional investors concerned about the business implications of climate change. CP is the ﬁrm’s carbon perfor- mance, measured as the inverse of total carbon emissions per million dollars of sales turnover (net); it was also used in Clarkson et al. (2008). HICP is indicator variable that equals 1 if the ﬁrm’sCPin year t is higher than the sample median, and 0 otherwise. BETA is market-model beta calculated from the month CRSP stock returns during year t−5 to year t. FD is a proxy for earnings quality, measured as the absolute value of abnormal accruals, estimated based on the modiﬁed Jones model. FIN is the amount of debt or equity capital raised by the ﬁrm, scaled by total assets at the end of year t. TOBINQ is the market value of common equity plus the book value of preferred stock, book value of long-term debt, and current liability, scaled by the book value of total assets. LEV is the leverage ratio, which is deﬁned as the ratio of total debt divided by total assets. SIZE is the natural logarithm of the market value of equity at the end of year t. MB is the market-to-book ratio, deﬁned as the market value of equity divided by book value of equity at the end of year t. LIQUIDITY is the ratio of the number of shares traded in year t to the weighted total shares outstanding at the end of every month of year t. 214 He et al. ROA is the total return on assets, measured as the ratio of income before extraordinary items over total assets at the end of year t. INTENSITY is an indicator variable that equals 1 if the ﬁrm operates in a GHG-intensive industry and 0 otherwise; it was also used in Matsumura et al. (2010). LITIGATION is an indicator variable that equals 1 if the ﬁrm operates in a high-litigation industry (SIC codes of 2833- 2836, 3570-3577, 3600-3674, 5200-5961, and 7370) and 0 otherwise. 4.4.4. Additional analysis Al-Tuwaijri et al. (2004) used an OLS procedure and found that the results differed notably from those of 3SLS. They argued that this was evidence of the importance of controlling for endogeneity and that this validated the appropriateness of a joint- estimation research design. Thus, we also employed the OLS approach; the results are reported in Table 9. The results using OLS estimation were signiﬁcantly differ- ent from those observed using 3SLS. The adjusted system R was reduced from 0.763 (0.773) in Table 5 (3SLS estimation) to 0.610 (0.599) in Table 9 (OLS esti- mation), and the correlations between carbon performance (CP) and carbon disclo- sure (CD or CD ) were no longer signiﬁcant. Moreover, the correlation between i,t i,t–1 the cost of capital and carbon disclosure (CD ) was not signiﬁcant in the system without HICP and CD*HICP, and the levels of signiﬁcance of some control vari- ables, such as FIN, LEV, SIZE, and LITIGATION in equation (2) were changed. These results suggest that a joint-estimation research design has provided more robust and consistent ﬁndings. 5. Conclusions and limitations This study investigated the relationships between carbon disclosure, carbon perfor- mance, and the cost of capital. Our research model speciﬁcation is consistent with the argument of Ullmann (1985) that the execution of corporate responsibility is determined by management’s (unobservable) overall strategy. Thus, these constructs might be endogenous. We used a simultaneous equation research design to signiﬁ- cantly improve our analyses of the relationships among carbon performance, carbon disclosure, and the cost of capital versus that obtained by independently estimated OLS models (Al-Tuwaijri et al., 2004). We found that carbon disclosure was nega- tively associated with the cost of capital and that this negative relationship largely took place in ﬁrms with poorer carbon performance. We conjecture that this may be owing to the inadequate disclosure of good carbon performers. Consistent with this explanation, we found that disclosure of good carbon performers was signiﬁcantly lower than that of poor performers. In addition, following the example of Dhaliwal et al. (2011), we also employed a lead-lag approach and documented a negative relationship between past carbon disclosure and carbon performance. This association is consistent with the notion that a ﬁrm may present fuller carbon information in advance to avoid future punishment caused by withholding carbon information, if poor carbon performance is expected in the future. The negative relationship in dis- closure between good and poor performers is consistent with legitimacy theory. Because they are bound by social contracts, poor carbon performers are willing to increase disclosure in return for approval of their objectives and other rewards, such as a lower cost of capital, and this ultimately helps assure the continued existence of these ﬁrms. China Journal of Accounting Studies 215 Table 9. OLS regression results coefﬁcients (t-statistics) Dependent variables Dependent variables Independent variables COST [Eq.(1)] CD [Eq.(2)] CP [Eq.(3)] COST [Eq.(1’)] CD [Eq.(2’)] CP [Eq.(3’)] t t t t t t ** CD –0.0002 –0.001 (–1.115) (–2.350) CD –0.001 –0.001 t–1 (–1.291) (–1.291) CP –10.112 –10.112 (–1.151) (–1.151) ** HICP –0.047 (–2.183) * ** CD HICP 0.001 t t (2.309) *** *** BETA 0.023 0.022 (4.211) (4.024) FD –0.020 1.348 –0.019 1.348 (–1.333) (0.186) (–1.173) (0.186) ** ** FIN 22.949 –0.056 22.949 –0.056 (2.172) (–0.578) (2.172) (–0.578) TOBINQ –1.257 0.014 –1.257 0.014 (–0.538) (0.554) (–0.538) (0.554) ** ** LEV 12.724 0.238 12.724 0.238 (1.429) (2.385) (1.429) (2.385) ** ** *** ** SIZE –0.006 3.079 0.001 –0.006 3.079 0.001 (–2.347) (2.544) (0.076) (–2.638) (2.544) (0.076) MB –0.001 –0.005 –0.001 –0.005 (–1.372) (–1.011) (–1.427) (–1.011) LIQUIDITY –0.858 –0.858 (–0.939) (–0.939) ROA 17.633 0.229 17.633 0.229 (0.524) (0.779) (0.524) (0.779) *** INTENSITY 0.013 –3.349 –0.139 0.188 –3.349 –0.139 (0.625) (–0.337) (–1.194) (6.190) (–0.337) (–1.194) LITIGATION 0.283 –0.038 0.283 –0.038 (0.084) (–1.288) (0.084) (–1.288) *** _cons 0.147 26.125 0.000 0.000 26.125 0.000 Industry indicators Yes Yes Yes Yes Yes Yes Adjusted system R 0.599 0.610 * ** *** , , Indicate that the estimated coefﬁcient is statistically signiﬁcant at the 10%, 5%, and 1% levels, respectively. n=181. All continuous variables are winsorized at the 1st and 99th percentiles. Variable deﬁ- nitions: COST is the implied cost of equity capital in year t estimated by the PEG formula in Easton (2004), which was recommended by Botosan and Plumlee (2005). CD is a measure of the ﬁrm’s volun- tary carbon disclosure in year t presented by CDP, which is an international collaboration of institutional investors concerned about the business implications of climate change. CP is the ﬁrm’s carbon perfor- mance, measured as the inverse of total carbon emissions per million dollars of sales turnover (net); it was also used in Clarkson et al. (2008). HICP is indicator variable that equals 1 if the ﬁrm’sCPin year t is higher than the sample median, and 0 otherwise. BETA is market-model beta calculated from the month CRSP stock returns during year t−5 to year t. FD is a proxy for earnings quality, measured as the absolute value of abnormal accruals, estimated based on the modiﬁed Jones model. FIN is the amount of debt or equity capital raised by the ﬁrm, scaled by total assets at the end of year t. TOBINQ is the market value of common equity plus the book value of preferred stock, book value of long-term debt, and current liability, scaled by the book value of total assets. LEV is the leverage ratio, which is deﬁned as the ratio of total debt divided by total assets. SIZE is the natural logarithm of the market value of equity at the end of year t. MB is the market-to-book ratio, deﬁned as the market value of equity divided by book value of equity at the end of year t. LIQUIDITY is the ratio of the number of shares traded in year t to the weighted total shares outstanding at the end of every month of year t. ROA is the total return on assets, measured as the ratio of income before extraordinary items over total t 216 He et al. assets at the end of year t. INTENSITY is an indicator variable that equals 1 if the ﬁrm operates in a GHG-intensive industry and 0 otherwise; it was also used in Matsumura et al. (2010). LITIGATION is an indicator variable that equals 1 if the ﬁrm operates in a high-litigation industry (SIC codes of 2833- 2836, 3570-3577, 3600-3674, 5200-5961, and 7370) and 0 otherwise. These results highlight the importance of carbon disclosure for market participants and suggest that carbon disclosure is an important part of the process of fulﬁlling stake- holders’ demand for environmental accountability. This applies to all ﬁrms regardless of their carbon performance records. Like all cross-sectional studies, limitations apply to the interpretation of our results regarding whether the time period examined (2009–2010) is representative and whether the observed relationships among the variables of interest are relatively stable. For example, our sample was drawn from S&P 500 ﬁrms, which might introduce a size bias. However, because at that time CDP questionnaires were sent only to S&P 500 ﬁrms in the United States, we cannot address this issue. Whereas our results may be generalized for large ﬁrms, it would be overreaching to infer that small ﬁrms may behave similarly. In addition, this study used the PEG model of Easton (2004) to estimate the cost of capital. This model is applied only to those ﬁrms with positive expected earnings growth, which may lead to some self-selection bias, although Botosan and Plumlee (2005) found that PEG estimates were consis- tently and predictably related to risk. Finally, because our ﬁnding of a signiﬁcant association between carbon disclosure and the cost of capital implies value relevance of carbon disclosure to this measure, further research might consider using other proxies for economic performance, such as market return, to determine the perceived usefulness of carbon information. Acknowledgements We acknowledge ﬁnancial support from the National Social Science Fund (Research grant number: 13BGL057), the Humanity and Social Science Foundation of Ministry of Education (Research grant number: 09YJC790137), the Social Science Foundation of Jiang Su Prov- ince, People’s Republic of China (Research grant number: 09SHD012), the Jiangsu Provin- cial Department of Education Fund of Philosophy and Social Science (2013SJB6300036), as well as project funding from the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD) and the constructive suggestions of two anonymous reviewers. Note 1. Utilities, energy, and materials are classiﬁed as carbon-intensive sectors (Tang & Luo, 2011). References Adams, C. A. (2004). The ethical, social and environmental reporting-performance portrayal gap. Accounting, Auditing and Accountability Journal, 17, 731–757. Aerts, W., Cormier, D., & Magnan, M. (2008). Corporate environmental disclosure, ﬁnancial markets and the media: An international perspective. Ecological Economics, 64, 643– Al-Tuwaijri, S. A., Christensen, T. E., & Hughes, K. E. (2004). The relations among environmen- tal disclosure, environmental performance, and economic performance: A simultaneous equa- tions approach. Accounting, Organizations and Society, 29, 447–471. China Journal of Accounting Studies 217 Amihud, Y., & Mendelson, H. (1986). Asset pricing and the bid-ask spread. Journal of Financial Economics, 17, 223–249. Archel, P., Husillos, J., Larrinaga, C., & Spence, C. (2009). Social disclosure, legitimacy the- ory and the role of the state. Accounting, Auditing & Accountability Journal, 22, 1284– Arenas, D., Lozano, J. M., & Albareda, L. (2009). The role of NGOs in CSR: Mutual percep- tions among stakeholders. Journal of Business Ethics, 88, 175–197. Bebbington, J., Larrinaga, C., & Moneva, J. M. (2008). Corporate social reporting and reputation risk management. Accounting, Auditing & Accountability Journal, 21, 337–361. Bewley, K., & Li, Y. (2000). Disclosure of environmental information by Canadian manufactur- ing companies: A voluntary disclosure perspective. Advances in Environmental Accounting and Management, 1, 201–226. Botosan, C. A. (1997). Disclosure level and the cost of equity capital. The Accounting Review, 72, 323–349. Botosan, C. A., & Plumlee, M. A. (2002). A re-examination of disclosure level and the expected cost of equity capital. Journal of Accounting Research, 40,21–40. Botosan, C. A., & Plumlee, M. A. (2005). Assessing alternative proxies for the expected risk pre- mium. The Accounting Review, 80,21–53. Buhr, N. (1998). Environmental performance, legislation and annual report disclosure: The case of acid rain and Falconbridge. Accounting, Auditing & Accountability Journal, 11, 163–190. Campbell, D. (2003). Intra- and intersectoral effects in environmental disclosures: Evidence for legitimacy theory? Business Strategy and the Environment, 12, 357–371. Clarkson, P. M., Li, Y., Richardson, G. D., & Vasvari, F. P. (2008). Revisiting the relation between environmental performance and environmental disclosure: An empirical analysis. Accounting Organizations and Society, 33, 303–327. Clarkson, P. M., Fang, X. H., & Li, Y. (2011a). The relevance of environmental disclosures for investors and other stakeholder groups: are such disclosures incrementally informative? Working paper. Clarkson, P. M., Li, Y., Richardson, G. D., & Vasvari, F. P. (2011b). Does it really pay to be green? Determinants and consequences of proactive environmental strategies. Journal of Accounting and Public Policy, 30, 122–144. Clarkson, P. M., Overell, M. B., & Chapple, L. (2011c). Environmental reporting and its relation to corporate environmental performance. Abacus, 47,27–60. Cogan, D. C. (2006). Corporate Governance and Climate Change: Making the Connection. Boston, MA: Ceres. Core, J. E. (2001). A review of the empirical disclosure literature: Discussion. Journal of Accounting and Economics, 31, 441–456. Cormier, D., Magnan, M., & Van Velthoven, B. (2005). Environmental disclosure quality in large German companies: Economic incentives, public pressures or institutional conditions? Euro- pean Accounting Review, 14,3–39. Davidson, R., & mackinnon, J. (2004). Econometric theory and methods. New York: Oxford Uni- versity Press. Decker, C. S. (2002). Shifting the burden: Regulatory response to pollution control. Journal of Economic Research, 7,41–76. Decker, C. S., & Pope, C. R. (2005). Adherence to environmental law: The strategic complementarities of compliance decisions. The Quarterly Review of Economics and Finance, 45, 641–661. Deegan, C., & Gordon, B. (1996). A study of environmental disclosure practices of Australian corporations. Accounting and Business Research, 26, 187–199. Deegan, C., & Rankin, M. (1996). Do Australian companies report environmental news objec- tively? Accounting, Auditing and Accountability Journal, 9,50–67. Dhaliwal, D. S., Li, O. Z., Tsang, A., & Yang, Y. G. (2011). Voluntary nonﬁnancial disclosure and the cost of equity capital: The initiation of corporate social responsibility reporting. The Accounting Review, 86,59–100. Diamond, D. W., & Verrecchia, R. E. (1991). Disclosure, liquidity, and the cost of capital. The Journal of Finance, 46, 1325–1359. 218 He et al. Dye, R. A. (1985). Disclosure of nonproprietary information. Journal of Accounting Research, 23, 123–145. Easton, P. D. (2004). PE Ratios, PEG Ratios, and Estimating the Implied Expected Rate of Return on Equity Capital. The Accounting Review, 79,73–95. Fama, E. F., & French, K. R. (1992). The cross-section of expected stock returns. The Journal of Finance, 47, 427–465. Francis, J., Nanda, D., & Olsson, P. (2008). Voluntary disclosure, earnings quality, and cost of capital. Journal of Accounting Research, 46,53–99. Francis, J., Philbrick, D., & Schipper, K. (1994). Shareholder litigation and corporate disclosures. Journal of Accounting Research, 32, 137–164. Frankel, R., mcnichols, M., & Wilson, G. P. (1995). Discretionary disclosure and external ﬁnanc- ing. The Accounting Review, 70, 135–150. Freedman, M., & Wasley, C. (1990). The association between environmental performance and environmental disclosure in annual reports and 10 Ks. Advances in Public Interest Account- ing, 3, 183–193. Graham, J. R., Harvey, C. R., & Rajgopal, S. (2005). The economic implications of corporate ﬁnancial reporting. Journal of Accounting and Economics, 40,3–73. Gray, R., Kouhy, R., & Lavers, S. (1995). Corporate social and environmental reporting, A review of the literature and a longitudinal study of UK disclosure. Accounting, Auditing & Accountability Journal, 8,47–77. Hausman, J. A. (1978). Speciﬁcation tests in econometrics. Econometrica, 46, 1251–1271. Healy, P. M., & Palepu, K. G. (2001). Information asymmetry, corporate disclosure, and the capital markets: A review of the empirical disclosure literature. Journal of Accounting and Economics, 31, 405–440. Hughes, S. B., Anderson, A., & Golden, S. (2001). Corporate environmental disclosures: Are they useful in determining environmental performance? Journal of Accounting and Public Policy, 20, 217–240. Ingram, R. W., & Frazier, K. B. (1980). Environmental performance and corporate disclosure. Journal of Accounting Research, 18, 614–622. Kolk, A., Levy, D., & Pinkse, J. (2008). Corporate responses in an emerging climate regime: The institutionalization and commensuration of carbon disclosure. European Accounting Review, 17, 719–745. Laine, M. (2009). Ensuring legitimacy through rhetorical changes? A longitudinal interpretation of the environmental disclosures of a leading Finnish chemical company. Accounting, Audit- ing & Accountability Journal, 22, 1029–1054. Lambert, R., Leuz, C., & Verrecchia, R. E. (2007). Accounting information, disclosure, and the cost of capital. Journal of Accounting Research, 45, 385–420. Lang, M., & Lundholm, R. (1993). Cross-sectional determinants of analyst ratings of corporate disclosures. Journal of Accounting Research, 31, 246–271. Leftwich, R. W., Watts, R. L., & Zimmerman, J. L. (1981). Voluntary corporate disclosure: The case of interim reporting. Journal of Accounting Research, 19,50–77. Leuz, C., & Verrecchia, R. E. (2000). The economic consequences of increased disclosure. Journal of Accounting Research, 38,91–124. Leuz, C., & Wysocki, P. (2008). Economic consequences of ﬁnancial reporting and disclosure regulation: A review and suggestions for future research. University of Chicago. Levy, D. L., & Egan, D. (2003). A neo-Gramscian approach to corporate political strategy: Conﬂict and accommodation in the climate change negotiations. Journal of Management Studies, 40, 803–830. Li, Y., Richardson, G. D., & Thornton, D. B. (1997). Corporate disclosure of environmental liability information: Theory and evidence. Contemporary Accounting Research, 14, 435– Lindblom, C. (1993). The implications of organizational legitimacy for corporate social performance and disclosure. Paper presented at the Critical Perspectives on Accounting Conference, New York. Lindblom, C. (1994). The implications of organizational legitimacy for corporate social performance disclosure. Paper presented at the Critical Perspectives on Accounting Confer- ence. New York. China Journal of Accounting Studies 219 Luo, L., Lan, Y. C., & Tang, Q. (2012). Corporate incentives to disclose carbon information: Evidence from the CDP Global 500 Report. Journal of International Financial Management & Accounting, 23,93–120. Magness, V. (2006). Strategic posture, ﬁnancial performance and environmental disclosure: An empirical test of legitimacy theory. Accounting, Auditing & Accountability Journal, 19, 540–563. Margolick, M., & Russell, D. (2004). Corporate greenhouse gas reduction targets. Arlington VA: Pew Center on Global Climate Change/Global Change Strategies International. Matsumoto, D. A. (2002). Management’s incentives to avoid negative earnings surprises. The Accounting Review, 77, 483–514. Matsumura, E., Prakash, R., & Vera-Munoz, S. (2010). Carbon emissions and ﬁrm value. Avail- able at SSRN 1688738. Merton, R. C. (1987). A simple model of capital market equilibrium with incomplete information. The Journal of Finance, 42, 483–510. Milne, M. J., & Patten, D. M. (2002). Securing organizational legitimacy: An experimental decision case examining the impact of environmental disclosures. Accounting, Auditing & Accountability Journal, 15, 372–405. O’Donovan, G. (2002). Environmental disclosures in the annual report: Extending the applicabil- ity and predictive power of legitimacy theory. Accounting, Auditing & Accountability Journal, 15, 344–371. O’Dwyer, B. (2005). Stakeholder democracy: Challenges and contributions from social account- ing. Business Ethics: A European Review, 14,28–41. Patten, D. M. (1992). Intra-industry environmental disclosures in response to the Alaskan oil spill: A note on legitimacy theory. Accounting, Organizations and Society, 17, 471–475. Patten, D. M. (2002). The relation between environmental performance and environmental disclo- sure: A research note. Accounting, Organizations and Society, 27, 763–773. Phillips, G. (Ed.) (2004). Monitoring, veriﬁcation and accounting of emissions. London: Risk Books. Plumlee, M., Brown, D., & Marshall., S. (2008). The impact of voluntary environmental disclo- sure quality on ﬁrm value. University of Utah. PwC, & CDP. (2010). Carbon Disclosure Project 2010 S&P 500 Report. Richardson, A. J., & Welker, M. (2001). Social disclosure, ﬁnancial disclosure and the cost of equity capital. Accounting, Organizations and Society, 26, 597–616. Richardson, A. J., Welker, M., & Hutchinson, I. R. (1999). Managing capital market reactions to corporate social responsibility. International Journal of Management Reviews, 1,17–43. Salop, S. C., & Scheffman, D. T. (1983). Raising rivals’ costs. American Economic Review, 73, 267–271. Sengupta, P. (1998). Corporate disclosure quality and the cost of debt. The Accounting Review, 73, 459–474. Sinkin, C., Wright, C. J., & Burnett, R. D. (2008). Eco-efﬁciency and ﬁrm value. Journal of Accounting and Public Policy, 27, 167–176. Skinner, D. J. (1997). Earnings disclosures and stockholder lawsuits. Journal of Accounting and Economics, 23, 249–282. Stanny, E., & Ely, K. (2008). Corporate environmental disclosures about the effects of climate change. Corporate Social Responsibility and Environmental Management, 15, 338–348. Suchman, M. C. (1995). Managing legitimacy: Strategic and institutional approaches. The Acad- emy of Management Review, 20, 571–610. Tang, Q., & Luo, L. (2011). Transparency of corporate carbon disclosure: International evi- dence. School of Business, University of Western Sydney. Tang, Q., & Luo, L. (2013). Carbon management systems and carbon mitigation, Australian Accounting Review, 23. Tran, J., Okafor, O., & Herremans, I. M. (2011). GAAP for GHG emissions accounting: How consistent are GHG emissions reported among different channels? Presented at the CAAA Annual Conference. Ullmann, A. A. (1985). Data in search of a theory: A critical examination of the relationships among social performance, social disclosure, and economic performance of U.S. ﬁrms. The Academy of Management Review, 10, 540–557. 220 He et al. Verrecchia, R. E. (1983). Discretionary disclosure. Journal of Accounting and Economics, 5, 179–194. Verrecchia, R. E. (2001). Essays on disclosure. Journal of Accounting and Economics, 32, 97–180. Wiseman, J. (1982). An evaluation of environmental disclosures made in corporate annual reports. Accounting, Organizations and Society, 7,53–63. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png China Journal of Accounting Studies Taylor & Francis http://www.deepdyve.com/lp/taylor-francis/carbon-disclosure-carbon-performance-and-cost-of-capital-29k00D6zjI

Loading next page...

References (85)

R. Merton (1987)
Presidential Address: A simple model of capital market equilibrium with incomplete information
R. Leftwich, R. Watts, J. Zimmerman (1981)
Voluntary Corporate Disclosure: The Case of Interim Reporting
Journal of Accounting Research, 19
J. Graham, Campbell Harvey, Shivaram Rajgopal (2004)
The Economic Implications of Corporate Financial Reporting
Corporate Finance: Valuation
C. Deegan, Michaela Rankin (1996)
Do Australian companies report environmental news objectively?: An analysis of environmental disclosures by firms prosecuted successfully by the Environmental Protection Authority
Accounting, Auditing & Accountability Journal, 9
Pablo Archel, Javier Husillos, Carlos Larrinaga, Crawford Spence (2009)
Social disclosure, legitimacy theory and the role of the state
Accounting, Auditing & Accountability Journal, 22
Douglas Skinner (1997)
Earnings Disclosures and Stockholder Lawsuits
Chicago Booth ARC: Financial Accounting (Topic)
Christine Botosan, Marlene Plumlee (2005)
Assessing Alternative Proxies for the Expected Risk Premium
Accounting review: A quarterly journal of the American Accounting Association, 80
(2010)
Carbon emissions and firm value. Available at SSRN 1688738
(2001)
Essays on disclosure
J. Core (2001)
A Review of the Empirical Disclosure Literature: Discussion
Journal of Accounting Abstracts
(1983)
Discretionary disclosure
Brendan O’Dwyer (2005)
Stakeholder Democracy: Challenges and Contributions from Social Accounting
Business Ethics: A European Review, 14
M. Milne, Dennis Patten (2002)
Securing organizational legitimacy
Accounting, Auditing & Accountability Journal, 15
Christine Botosan, Marlene Plumlee (2002)
A Re‐examination of Disclosure Level and the Expected Cost of Equity Capital
Journal of Accounting Research, 40
D. Campbell (2003)
Intra- and intersectoral effects in environmental disclosures: evidence for legitimacy theory?
Business Strategy and The Environment, 12
J. Francis, D. Nanda, Per Olsson (2008)
Voluntary Disclosure, Earnings Quality, and Cost of Capital
Journal of Accounting Research, 46
Yue Li, Peter Clarkson, G. Richardson, Florin Vasvari (2006)
Revisiting the Relation between Environmental Performance and Environmental Disclosure: An Empirical Analysis
Econometrics eJournal
Yue Li, G. Richardson, Daniel Thornton (1997)
Corporate Disclosure of Environmental Liability Information: Theory and Evidence*
Contemporary Accounting Research, 14
Peter Clarkson, Yue Li, G. Richardson, Florin Vasvari (2006)
Does it Really Pay to Be Green? Determinants and Consequences of Proactive Environmental Strategies
SEIN Environmental Impacts of Business eJournal
A. Richardson, Michael Welker (2001)
Social disclosure, financial disclosure and the cost of equity capital
Accounting Organizations and Society, 26
(2010)
LITIGATION t is an indicator variable that equals 1 if the firm operates in a high-litigation industry (SIC codes of 2833-2836
Richard Frankel, M. McNichols (1995)
Discretionary Disclosure and External Financing
Accounting review: A quarterly journal of the American Accounting Association, 70
R. Ingram, K. Frazier (1980)
ENVIRONMENTAL PERFORMANCE AND CORPORATE DISCLOSURE
Journal of Accounting Research, 18
Elizabeth Stanny, K. Ely (2008)
Corporate environmental disclosures about the effects of climate change
Corporate Social Responsibility and Environmental Management, 15
C. Leuz, Peter Wysocki (2008)
Economic Consequences of Financial Reporting and Disclosure Regulation: A Review and Suggestions for Future Research
Behavioral & Experimental Accounting
D. Cormier, M. Magnan, Barbara Velthoven (2005)
Environmental disclosure quality in large German companies: Economic incentives, public pressures or institutional conditions?
European Accounting Review, 14
Douglas Diamond, Robert Verrecchia (1991)
Disclosure, Liquidity, and the Cost of Capital
Journal of Finance, 46
Janice Tran, O. Okafor, I. Herremans (2010)
GAAP for GHG Emissions Accounting: How Consistent are GHG Emissions Reported Among Different Channels?
S. Hughes, A. Anderson, S. Golden (2001)
Corporate environmental disclosures: are they useful in determining environmental performance?
Journal of Accounting and Public Policy, 20
Dennis Patten (1992)
INTRA-INDUSTRY ENVIRONMENTAL DISCLOSURES IN RESPONSE TO THE ALASKAN OIL SPILL: A NOTE ON...
, 17
Dennis Patten (2002)
The relation between environmental performance and environmental disclosure: a research note
Accounting Organizations and Society, 27
J. Hausman (1978)
Specification tests in econometrics
Applied Econometrics, 38
(2004)
Monitoring, verification and accounting of emissions
C. Adams (2004)
The ethical, social and environmental reporting‐performance portrayal gap
Accounting, Auditing & Accountability Journal, 17
Qingliang Tang, L. Luo (2011)
Transparency of Corporate Carbon Disclosure: International Evidence
E. Matsumura, Rachna Prakash, Sandra Vera-Muñoz (2010)
Carbon Emissions and Firm Value
Financial Accounting eJournal
Peter Easton (2003)
Pe Ratios, Peg Ratios, and Estimating the Implied Expected Rate of Return on Equity Capital
Capital Markets: Asset Pricing & Valuation eJournal
J. Bebbington, Carlos Larrinaga, J. Moneva (2008)
Corporate social reporting and reputation risk management
Accounting, Auditing & Accountability Journal, 21
(1994)
The implications of organizational legitimacy for corporate social performance disclosure. Paper presented at the Critical Perspectives on Accounting Conference
R. Davidson, J. MacKinnon (2003)
Econometric Theory and Methods
M. Laine (2009)
Ensuring legitimacy through rhetorical changes?: A longitudinal interpretation of the environmental disclosures of a leading Finnish chemical company
Accounting, Auditing & Accountability Journal, 22
P. Sengupta (1998)
Corporate Disclosure Quality and the Cost of Debt
Accounting review: A quarterly journal of the American Accounting Association, 73
Nola Buhr (1998)
Environmental performance, legislation and annual report disclosure: the case of acid rain and Falconbridge
Accounting, Auditing & Accountability Journal, 11
C. Lindblom (1994)
The implications of Organizational Legitimacy for Corporate Social Performance and Disclosure
J. Lewellen (2014)
The Cross Section of Expected Stock Returns
Household Finance eJournal
Dawn Matsumoto (2002)
Management's Incentives to Avoid Negative Earnings Surprises
Accounting review: A quarterly journal of the American Accounting Association, 77
Mark Suchman (1995)
Managing Legitimacy: Strategic and Institutional Approaches
Academy of Management Review, 20
M. Overell (2007)
An examination of environmental reporting and its relation to corporate environmental performance
David Levy, D. Egan (2003)
A Neo‐Gramscian Approach to Corporate Political Strategy: Conflict and Accommodation in the Climate Change Negotiations*
Journal of Management Studies, 40
(2004)
Corporate greenhouse gas reduction targets. Arlington VA: Pew Center on Global Climate Change/Global Change Strategies International
M. Milne (2001)
SECURING ORGANIZATIONAL LEGITIMACY: AN EXPERIMENTAL DECISION CASE EXAMINING THE IMPACT OF ENVIRONMENTAL DISCLOSURES*
Arieh Ullmann (1985)
Data in Search of a Theory: A Critical Examination of the Relationships Among Social Performance, Social Disclosure, and Economic Performance of U.S. Firms
Academy of Management Review, 10
Mark Lang, Russell Lundholm (1993)
CROSS- SECTIONAL DETERMINANTS OF ANALYST RATINGS OF CORPORATE DISCLOSURES
Journal of Accounting Research, 31
C. Leuz, Robert Verrecchia (1999)
The Economic Consequences of Increased Disclosure
Journal of Accounting Abstracts
R. Dye (1985)
DISCLOSURE OF NONPROPRIETARY INFORMATION
Journal of Accounting Research, 23
Christopher Decker (2002)
Shifting the Burden : Regulatory Responses to Pollution Control
Journal of economic research, 7
J. Francis, Donna Philbrick, K. Schipper (1994)
SHAREHOLDER LITIGATION AND CORPORATE DISCLOSURES
Journal of Accounting Research, 32
Christine Botosan (1997)
Disclosure level and the cost of equity capital
Accounting review: A quarterly journal of the American Accounting Association, 72
Daniel Arenas, J. Lozano, Laura Albareda (2009)
The Role of NGOs in CSR: Mutual Perceptions Among Stakeholders
Journal of Business Ethics, 88
Gady Jacoby, David Fowler (2001)
On Asset Pricing and the Bid-Ask Spread
L. Luo, Y. Lan, Qingliang Tang (2012)
Corporate Incentives to Disclose Carbon Information: Evidence from the CDP Global 500 Report
Corporate Finance: Governance
W. Aerts, D. Cormier, M. Magnan (2008)
Corporate environmental disclosure, financial markets and the media: An international perspective
Ecological Economics, 64
Richard Lambert, C. Leuz, Robert Verrecchia (2006)
Accounting Information, Disclosure, and the Cost of Capital
Financial Accounting
C.G.F. Sinkin, C. Wright, Royce Burnett (2008)
Eco-efficiency and firm value
Journal of Accounting and Public Policy, 27
G. O’Donovan (2002)
Environmental disclosures in the annual report: Extending the applicability and predictive power of legitimacy theory
Accounting, Auditing & Accountability Journal, 15
Sulaiman Al-Tuwaijri, Theodore Christensen, K. Hughes (2003)
The Relations Among Environmental Disclosure, Environmental Performance, and Economic Performance: A Simultaneous Equations Approach
Financial Accounting
A. Kolk, David Levy, J. Pinkse (2008)
Corporate Responses in an Emerging Climate Regime: The Institutionalization and Commensuration of Carbon Disclosure
European Accounting Review, 17
D. Scheffman, R. Higgins (2014)
Raising Rivals’ Costs
The American Economic Review, 73
D. Dhaliwal, Oliver Li, Albert Tsang, Yong Yang (2011)
Voluntary Nonfinancial Disclosure and the Cost of Equity Capital: The Initiation of Corporate Social Responsibility Reporting
The Accounting Review, 86
(2004)
Corporate greenhouse gas reduction targets
R. Gray, Reza Kouhy, Simon Lavers (1995)
Corporate social and environmental reporting
Accounting, Auditing & Accountability Journal, 8
Kathryn Bewley, Yue Li (2000)
Disclosure of environmental information by Canadian manufacturing companies: A voluntary disclosure perspective
, 1
D. Cogan (2003)
Corporate Governance and Climate Change: Making the Connection
S. Marshall, Darrell Brown, Marlene Plumlee (2009)
The Impact of Voluntary Environmental Disclosure Quality on Firm Value
, 2009
Christopher Decker, Christopher Pope (2005)
Adherence to environmental law: the strategic complementarities of compliance decisions
The Quarterly Review of Economics and Finance, 45
A. Richardson, Michael Welker, Ian Hutchinson (1999)
Managing Capital Market Reactions to Corporate Social Resposibility
International Journal of Management Reviews, 1
J. Wiseman (1982)
An evaluation of environmental disclosures made in corporate annual reports
Accounting Organizations and Society, 7
(1995)
Corporate social and environmental reporting, A review of the literature and a longitudinal study of UK disclosure
V. Magness (2006)
Strategic posture, financial performance and environmental disclosure: An empirical test of legitimacy theory
Accounting, Auditing & Accountability Journal, 19
C. Deegan, Benedict Gordon (1996)
A Study of the Environmental Disclosure Practices of Australian Corporations
Accounting and Business Research, 26
Peter Clarkson, Xiaohua Fang, Yue Li, G. Richardson (2010)
The Relevance of Environmental Disclosures for Investors and Other Stakeholder Groups: Are Such Disclosures Incrementally Informative?
Financial Accounting eJournal
(2004)
Monitoring, verification and accounting of emissions. London: Risk Books
(1990)
The association between environmental performance and environmental disclosure in annual reports and 10 Ks
Qingliang Tang, L. Luo (2014)
Carbon Management Systems and Carbon Mitigation
Australian Accounting Review, 24
P. Healy, K. Palepu (2000)
Information Asymmetry, Corporate Disclosure and the Capital Markets: A Review of the Empirical Disclosure Literature
Management Practice

Publisher: Taylor & Francis
Copyright: © 2013 Accounting Society of China
ISSN: 2169-7221
eISSN: 2169-7213
DOI: 10.1080/21697221.2014.855976
Publisher site: See Article on Publisher Site

Abstract

China Journal of Accounting Studies, 2013 Vol. 1, Nos. 3–4, 190–220, http://dx.doi.org/10.1080/21697221.2014.855976 a b a Yu He *, Qingliang Tang and Kaitian Wang a b School of Accounting, Nanjing University of Finance and Economics, China; School of Business, University of Western Sydney, Australia More and more ﬁrms are voluntarily disclosing carbon information as a response to the challenge of climate change. This research investigated the interactions among carbon disclosure, carbon performance, and the cost of capital. Because unobservable overall strategic decisions by management affect each of these outcomes and phenomena, we used a simultaneous equations model to analyse our data. We used data from S&P 500 ﬁrms that participated in the Carbon Disclosure Project (CDP) in 2010. We found that the cost of capital is negatively associated with carbon disclosure, which is consistent with voluntary disclosure theory. This relationship is weaker for ﬁrms with good carbon performance. In addition, there is an inverse relationship between carbon disclosure and carbon performance, which is consistent with legitimacy theory. Our results suggest that voluntary carbon disclosure is a rational choice that ﬁrms make to reduce the pressure exerted by legitimacy threats and to lower the cost of capital. Keywords: carbon disclosure; carbon performance; cost of capital; simultaneous equations 1. Introduction Corporate responses to climate change have shifted dramatically over the past two decades (Kolk, Levy, & Pinkse, 2008). Until the mid-1990s, North American ﬁrms seldom talked about the topic. Nevertheless, a few ﬁrms in sectors related to fossil fuels perceived the prospect of regulations of greenhouse gas (GHG) emissions as a substan- tial threat. New industry groups were created, such as the Global Climate Coalition and the Climate Council, which spared no effort in preventing the international community from imposing caps on GHG emissions and played a major role in preventing the United States from joining the Kyoto Protocol (Kolk et al., 2008; Levy & Egan, 2003). However, more recently, business has tended to converge on a more constructive stance that views climate change as an opportunity rather than a burden (Margolick & Russell, 2004). Financial markets have started to reward companies that are moving ahead on climate change, while those lagging behind are assigned more risk (Cogan, 2006; Kolk et al., 2008). Investors and environmental non-governmental organizations (NGOs) are pushing companies to disclose information related to their GHG emissions, since carbon disclosure provides information that is crucial to an accurate valuation of assets. NGOs can also use the information to pressure ﬁrms into improving carbon performance (O’Dwyer, 2005). Several initiatives have emerged that attempt to leverage the inﬂuence of institutional investors to create demand for carbon disclosure as an *Corresponding author. Email: yu.he@njue.edu.cn Paper accepted by Kangtao Ye. © 2013 Accounting Society of China China Journal of Accounting Studies 191 adjunct to conventional ﬁnancial systems. One of the most prominent is the Carbon Disclosure Project (CDP). CDP is an independent and not-for-proﬁt organization based in the United King- dom that addresses the climate change concerns of institutional investors (Tran, Oka- for, & Herremans, 2011). The CDP represents 534 institutional investors with more than US$64 trillion in assets under management, and it can be seen as a secondary stakeholder that has facilitated collaborative engagement to increase corporate account- ability in relation to climate change (Arenas, Lozano, & Albareda, 2009). By means of a standard questionnaire, the CDP collects climate change data on GHG emissions, carbon risks and opportunities, and the actions that companies are taking to reduce emissions. In 2010, CDP sent this questionnaire to more than 4700 of the world’s larg- est corporations, including S&P 500 ﬁrms. Some 70% (350) of S&P 500 ﬁrms partici- pated in the CDP in 2010, an increase from 66% (332) in 2009, 63% (314) in 2008, and 56% (280) in 2007 (PwC & CDP, 2010). Overall, this activity sends an important message to investors from companies that climate change is an important business concern. However, the rapid increase in participation in the CDP naturally raises ques- tions among researchers: what are the rationales behind this type of voluntary carbon disclosure? What beneﬁts do ﬁrms gain by spending resources on compiling and pub- lishing these standalone reports? A number of factors potentially provide answers to these questions, such as the pressure on businesses to establish and comply with environmental and social norms and standards (Cormier, Magnan, & Van Velthoven, 2005). In this research, we examined whether carbon disclosure is associated with a reduction in the cost of capital. The cost of capital is a critical issue in a ﬁrm’s ﬁnancing and general oper- ating decisions (Dhaliwal, Li, Tsang, & Yang, 2011). Corporate executives appear to believe that voluntary disclosure can reduce the cost of capital (Graham, harvey, & Rajgopal, 2005), and there is a long-standing interest among academics in the rela- tionship between disclosure and the cost of capital (Botosan, 1997; Botosan & Plumlee, 2002; Dhaliwal et al., 2011; Diamond & Verrecchia, 1991; Francis, Nanda, & Olsson, 2008; Leuz & Verrecchia, 2000; Richardson & Welker, 2001). While these investigations have advanced our knowledge, no study has yet attempted to examine carbon disclosure, carbon performance, and the cost of capital within a sin- gle inclusive model. Ullmann (1985) posited that management implements policies and decisions that simultaneously affect the ﬁrm’s environmental disclosure, environ- mental performance, and economic performance. Al-Tuwaijri, Christensen, and Hughes (2004) argued that if these corporate functions are endogenously determined, then piecemeal ordinary least squares (OLS) estimation of pairwise relationships among these three functions will produce biased and inconsistent results. They per- formed a Hausman (1978) test and rejected their null hypothesis of no endogeneity among these three variables. Thus, we investigated collectively the relationships among carbon disclosure, car- bon performance, and the cost of capital. We employed a sample of US S&P 500 cor- porations that present their CDP reports on the CDP website. Our analyses showed that the cost of capital is signiﬁcantly and negatively associated with carbon disclosure. However, the negative relationship resides largely in ﬁrms with poor carbon perfor- mance. In addition, we documented a negative relationship between prior disclosure and contemporaneous emission reduction, suggesting that poor carbon performers tend to present (rather than withhold) carbon information in advance in attempts to mitigate any future negative impacts on the market. Overall, we observed that a potential reduc- 192 He et al. tion in the cost of capital motivates ﬁrms to present CDP reports, and this is related to a ﬁrm’s long-term development strategies and performance sustainability. This study is the ﬁrst to use an innovative approach to examine the relationships among carbon disclosure, carbon performance, and the cost of capital. Here, we extend the traditional research on voluntary disclosure from the narrow focus of ﬁnancial disclosure to nonﬁnancial disclosure, i.e. carbon disclosure. This research has made the following speciﬁc contributions. First, we focused on carbon disclo- sure and performance rather than social responsibility (Dhaliwal et al., 2011)or environmental reporting (Plumlee, Brown, & Marshall, 2008). Carbon study is a new area of research, and our ﬁndings on carbon disclosure and performance are different from ﬁndings on environmental disclosure. For example, we showed that carbon disclosure is broader than environmental disclosure. Carbon risks are more pervasive; ﬁrms must adopt a more comprehensive strategy and actions to deal with climate change. All ﬁrms emit carbon, while many ﬁrms do not have environmental issues. In addition, most environmental disclosure is nondiscretionary, for example, toxic emissions and chemical waste, whereas carbon disclosure is largely voluntary. For all these reasons, the nature of the relationship between carbon information and market reactions may be different from that documented in prior studies. Second, the measure of carbon performance used in our study was innovative compared with previous studies. Our measurement was more comprehensive than other environmen- tal performance measures, since we used sector- and ﬁrm-speciﬁc data such as waste, toxic chemical emissions, etc. Third, our data were taken from the CDP reports, rather than from environmental information included in annual reports, which are often biased by self-selection (Tang & Luo, 2013). CDP reports are designed by an NGO, and its format and contents are accepted and adopted by more than 4000 large global ﬁrms. Such information is relatively more consistent (Luo, Lan, & Tang, 2012). Fourth, this research introduced an empirical proxy for carbon disclosure. In contrast to prior studies, which often used self-constructed indices (Plumlee et al., 2008) or indicator variables (e.g. whether or not a report was published) (Dhaliwal et al., 2011), we used the carbon disclosure score index from the CDP, which has been widely presented on many ﬁnancial websites (e.g., http://www.google.com/ﬁnance#). This index is more comprehensive than others and covers many aspects of relevant information, such as carbon governance mecha- nisms, carbon risks and opportunities, carbon strategy and targets, carbon actions and processes, carbon emissions and reporting, carbon emission trading and offset- ting, carbon communications and engagement, and more (Tang & Luo, 2013; Luo et al., 2012). Finally, our methodologies differed from previous carbon accounting research. We conducted a joint estimation using three-stage least squares (3SLS) simultaneous equations models to deal with endogeneity. In addition, carbon disclo- sure and performance are the focus of legislation that is either proposed or has been implemented by many governments and rule makers. For example, carbon account- ing standards, external assurance/veriﬁcation, carbon disclosure, and carbon reporting are all subject to future regulations. Thus, our results are potentially useful for capi- tal market participants and government bodies who are concerned about climate change-related issues and activities. The remainder of this paper proceeds as follows: Section 2 develops our hypothe- ses, Section 3 describes our sample and methodology, Section 4 presents the empirical results, and the ﬁnal section summarizes and concludes. China Journal of Accounting Studies 193 2. Literature review and hypotheses development The existing literature in environmental accounting research can be categorized into three broad groups (Clarkson, Li, Richardson, & Vasvari, 2008). The ﬁrst group found that environmental disclosure is relevant to ﬁrm valuation, the second investigated the determinants of discretionary environmental disclosure, and the third line of studies observed mixed results regarding the relationship between environmental disclosure and environmental performance. 2.1. The relationship between carbon disclosure and the cost of capital The consensus in the literature appears to be that a negative relationship exists between the quality of ﬁnancial disclosure and the cost of capital (Botosan, 1997; Core, 2001; Diamond & Verrecchia, 1991; Healy & Palepu, 2001; Leuz & Wysocki, 2008). Greater ﬁnancial disclosure increases investors’ awareness of a ﬁrm’s exis- tence and enlarges its investor base, which improves risk-sharing and reduces the cost of capital (Merton, 1987). In addition, expanded disclosure can narrow informa- tion asymmetry among investors or between managers and investors. Some informa- tion-disadvantaged investors are less willing to trade if disclosure is inadequate. The resultant illiquidity increases the bid-ask spread and transaction costs (Verrecchia, 2001), leading to a higher required rate of return or cost of capital (Amihud & Mendelson, 1986). In a traditional capital market setting, Lambert, Leuz, and Ver- recchia (2007) investigated the direct and indirect effects of disclosure quality on the cost of capital. They found that the direct effect occurs because higher-quality disclosures affect the ﬁrm’s assessed covariance with other ﬁrms’ cash ﬂows, which are non-diversiﬁable. The direct effect on the cost of capital can be attributed to a reduction in the estimation of information risk (Lambert et al., 2007; Diamond & Verrecchia, 1991; Leuz & Verrecchia, 2000). The indirect effect occurs because higher-quality disclosures affect a ﬁrm’s real decisions, which likely changes the ﬁrm’s ratio of expected future cash ﬂows to covariance of these cash ﬂows with the sum of all the cash ﬂows in the market. The sign of this indirect effect on the cost of capital is uncertain. Lambert et al. (2007) sought to determine the conditions under which an increase in information quality leads to an unambiguous decline in the cost of capital. They found that the net effect of increased disclosures on the cost of capital is a function of the strength of the relationship and whether the indi- rect effect is positive or negative (as the direct effect is unambiguously negative). These mechanisms can be extended to non-ﬁnancial disclosure, as long as the information concerned is value relevant (Sinkin, Wright, & Burnett, 2008). Richardson, Welker, and Hutchinson (1999) presented a model for the inﬂuence of environmental behaviours and the related disclosures on a ﬁrm’s value through net present value assessments of projects, including expected future regulatory costs and market effects. Richardson and Welker (2001) argued that there may be a direct inﬂuence of environmen- tal disclosure on the cost of equity capital, either through affecting investor preferences or through reduced information asymmetry or estimation risk. Investor preference effects arise if investors are willing to accept a lower rate of return on investments from an orga- nization that supports an environmental cause for which some investors have an afﬁnity. In addition, carbon disclosures may be seen as a ﬁrm’s credible commitment to environ- mental issues in their long-term strategic and production systems (Plumlee et al., 2008). Finally, higher-quality environmental disclosures, which have been linked to increased 194 He et al. environmental activities, may affect regulators’ decisions, thus inﬂuencing costs for ﬁrms and for their competitors (Decker, 2002; Decker & Pope, 2005; Salop & Scheffman, 1983). Clarkson, Fang, and Li (2011a) investigated the relevance of environmental disclosures and found that voluntary environmental disclosures enhanced ﬁrm value. Based on this literature, we tested the following hypothesis (stated in alternative form). H1: There is an inverse relationship between the level of carbon disclosure and the cost of capital. According to legitimacy theory, carbon disclosure is a function of social and political pressures. Firms with poor carbon performance face greater pressures and have greater incentives to disclose environmental information in an attempt to change public perception. Patten (1992) noted that, whereas economic legitimacy is monitored by the market, social legitimacy is monitored by the public policy process. Disclosure is one method available to ﬁrms to enhance their legitimacy, and it is often easier to manage the ﬁrm’s image than to make actual changes to performance. Therefore, higher polluting ﬁrms tend to disclose a greater quantity of information (Clarkson, Overell, & Chapple, 2011c), whereas the level of carbon disclosure by better performing ﬁrms is probably lower. This is because these ﬁrms face a smaller legitimacy problem, and stakeholders presumably pay less attention to these ﬁrms. Thus, investors are probably less sensitive to the carbon disclosure of good performers. Hence, the inverse relation- ship between the level of carbon disclosure and the cost of capital is predicted to be weaker for ﬁrms with higher carbon performance. H1a: The inverse relationship between the level of carbon disclosure and the cost of capital is weaker for ﬁrms with good carbon performance. 2.2. The relationship between carbon disclosure and carbon performance There are two theories regarding the relationship between carbon disclosure and carbon performance in the literature. The ﬁrst one is voluntary disclosure theory, which suggests that companies have incentives to disclose ‘good news’ to differentiate themselves from companies with ‘bad news’ to avoid the adverse selection problem (Dye, 1985; Verrecchia, 1983). Bewley and Li (2000) and Li, Richardson, and Thornton (1997) have argued that true environmental performance is not directly observable to investors; thus, companies with superior performance tend to make direct voluntary disclosures that cannot be easily matched by poor performers (Clarkson et al., 2008). Hence, this theory predicts a positive association between environmental performance and the level of discretionary environmental disclosure. The other group of theories is known as legitimacy theories, and this group argues that companies with threatened legitimacy are likely to make self-serving disclosures, referred to as ‘legitimization’ (Adams, 2004; Gray, Kouhy, & Lavers, 1995; Hughes, Anderson, & Golden, 2001). Researchers argue that ﬁrms vulnerable to certain types of criticism will make voluntary disclosures to deﬂect or nullify suspicion or doubt with regard to that area of potential criticism. Those organizations whose social legitimacy is threatened tend to increase disclosures to (1) educate and inform relevant members of the public about (actual) changes in their performance; (2) change perceptions about their performance; (3) deﬂect attention from the issue of concern by highlighting other accomplishments; and (4) attempt to change public expectations of their performance (Gray et al., 1995; China Journal of Accounting Studies 195 Lindblom, 1994). Therefore, legitimacy theories predict a negative association, and the empirical evidence in existing literature is largely mixed (Tang & Luo, 2011). For example, Wiseman (1982) examined a sample of 26 ﬁrms in environmentally sensitive industries using an indexing procedure to measure the extent of disclosure of 18 environmental items and found that no relationship existed between environmental disclosure and performance. Freedman and Wasley (1990) used the same indexing procedure for 50 US companies and conducted Spearman rank order correlation tests. They found that neither annual reports nor environmental disclosures (as communicated via the Securities and Exchange Commission’s 10-K report) were signiﬁcantly associ- ated with actual pollution control performance. Ingram and Frazier (1980) found that environmental disclosures did not relate strongly to environmental performance, as measured by CEP indices. However, Bewley and Li (2000), who adopted the same rat- ing scheme used by Wiseman (1982), found a negative association. They used a cross- sectional sample of Canadian manufacturing ﬁrms and documented that ﬁrms with more news media coverage of their environmental exposure, higher pollution propen- sity, and more political exposure were more likely to disclose general environmental information. Hughes et al. (2001) investigated whether disclosures differed between ﬁrms who were rated good, mixed, or poor in their environmental activities by the CEP and whether these differences in disclosure could be used to determine actual environmental performance levels. The study focused on 51 US manufacturing ﬁrms for 1992 and 1993 and found that the poor performers made the most disclosures. Patten (2002) identiﬁed three issues with previous studies: (1) failure to control for other factors; (2) inadequate sample selection; and (3) inadequate measures of environ- mental performance using CEP indices. Thus, Patten used Toxics Release Inventory data, normalized by sales, as a proxy for environmental performance and examined 131 US companies. The results indicated that there was a signiﬁcant negative relationship. Campbell (2003) examined environmental disclosures from the annual reports of a sample of ten UK FTSE 100 Index companies in ﬁve sectors between 1974 and 2000. The study provided evidence for legitimacy theory in the United Kingdom as an expli- cator of variability in environmental disclosure, adding to the most notable previous studies from Australia (Deegan & Gordon, 1996; Deegan & Rankin, 1996;O’Donovan, 2002) and North America (Buhr, 1998; Patten, 1992). Al-Tuwaijri et al. (2004) explored the relationships among environmental disclo- sure, environmental performance, and economic performance using a simultaneous equations approach. The study adopted a similar disclosure-scoring methodology based on content analysis that incorporated disclosures of four key environmental indicators: (1) the total amount of toxic waste generated and transferred or recycled; (2) ﬁnancial penalties resulting from violations of ten federal environmental laws; (3) Potential Responsible Party designation for the clean-up responsibility of hazardous waste sites; and (4) the occurrence of reported oil and chemical spills. These disclosures were based on information reported on Form 10-K and are largely non-discretionary. The study found a positive association between environmental performance and disclosure. More recently, Clarkson et al. (2008) focused on purely voluntary disclosure media, such as corporate Internet websites and stand-alone environmental reports. In addition, using a content analysis index, the study found that environmental performance was positively related to the level of discretionary disclosures in environmental and social reports or related web disclosures. However, the authors showed that ﬁrms whose envi- ronmental legitimacy is threatened may make more ‘soft claims’ regarding their com- 196 He et al. mitment to the environment. This result is predicted by legitimacy theory but cannot be explained by voluntary (or economic) disclosure theories. In a more recent study, Clarkson et al. (2011c) examined how both the level and the nature of environmental information disclosed by Australian ﬁrms related to their underlying environmental per- formance and found consistently that ﬁrms with a higher propensity toward pollution disclosed more environmental information. In summary, existing studies have shown mixed results. One reason for the inconclusive ﬁndings is the use of different indices developed by the authors to measure the chosen factors for environmental disclosure, and these different indices may be inconsistent with each other since they often include different kinds of environ- mental information. Another possibility is that these studies did not adequately address endogeneity issues. Thus, we used the CDP carbon disclosure index (CDI) and a joint- estimation approach to mitigate these concerns. Since voluntary disclosure theories and legitimacy theories have provided opposing predictions, we tested the following two competing hypotheses (stated in the alternative form). H2a: Carbon performance is positively associated with the level of discretionary carbon disclosure, which is consistent with voluntary disclosure theories. H2b: Carbon performance is negatively related to the level of discretionary carbon disclosure, which is consistent with legitimacy theories. 3. Sample and methodology 3.1. Sample description This study used a cross-sectional research design. Sample ﬁrms must meet all the fol- lowing criteria. That is, the ﬁrm: (1) was in the list of both 2009 and 2010 CDP reports and in the S&P 500 2009 and 2010 reports; (2) had a carbon disclosure score and carbon emission data for 2010; (3) did not experience takeovers, mergers, and/or acqui- sitions in 2010 and 2009; (4) had a cost of capital that could be computed; (5) and had complete ﬁnancial data reported in S&P’s Compustat database and the Center for Research in Security Prices (CRSP) database. Selection processes were conducted step by step. Of the 2010 S&P 500, 350 ﬁrms completed the CDP questionnaire, and only 240 ﬁrms met selection criteria (1) through (3). The cost of capital of 28 ﬁrms could not be computed, and 21 ﬁrms do not have complete Compustat and CRSP data. The ﬁnal sample therefore included 181 disclosing ﬁrms that met all of the selection criteria. Table 1 shows the industry distribution of all S&P 500 and sample ﬁrms. 3.2. Empirical models and variable deﬁnitions Ullmann (1985) conducted a meta-analysis of prior empirical studies that investigated the relationships among social disclosure, social performance, and economic perfor- mance. He noted that the mixed results in prior empirical studies might be caused by incomplete speciﬁcation of the empirical models measuring the statistical signiﬁcance of pairwise association. Al-Tuwaijri et al. (2004) argued that this body of empirical environmental research may have produced ambiguous results by failing to recognize the potential for endogenous relationships among environmental disclosure, environ- mental performance, and economic performance. Thus, they sought to detect the pres- ence of endogenous relationships among these three constructs by using a Hausman China Journal of Accounting Studies 197 Table 1. Sample distribution by industry. Industries All ﬁrms (no.) % Firms responding (no.) % Sample ﬁrms (no.) % Consumer Discretionary 80 16 49 14 18 10 Consumer staples 41 8 37 11 26 14 Energy 39 8 23 7 15 8 Financials 78 16 51 15 10 6 Health care 52 10 37 11 19 10 Industrials 58 12 37 11 24 13 Information Technology 77 15 56 16 33 18 Materials 32 6 25 7 18 10 Telecommunications 9 2 6 2 3 2 Utilities 34 7 29 8 15 8 Total 500 100 350 100 181 100 (1978) test. Similarly, we estimated the relationships among carbon disclosure, carbon performance, and the cost of capital by employing a system of simultaneous equations, deﬁned in the following structural form: COST ¼ CD þ BETA þ FD þ SIZE þ MB þ INTENSITY þ IND þ e i;t i;t i;t i;t i;t i;t i;t i;t i;t (1) CD ¼ CP þ FD þ FIN þ TOBINQ þ LEV þ SIZE þ LIQUIDITY i;t i;t i;t i;t i;t i;t i;t i;t (2) þ ROA þ INTENSITY þ LITIGATION þ IND þ e i;t i;t i;t i;t i;t CP ¼ CD þ FIN þ TOBINQ þ LEV þ SIZE þ MB þ ROA i;t i;t1 i;t i;t i;t i;t i;t i;t (3) þ INTENSITY þ LITIGATION þ IND þ e i;t i;t i;t i;t where COST , CD , and CP are endogenous variables and the other variables are i,t i,t i,t predetermined variables. In equation (1), COST is the implied cost of equity capital in year t estimated by the price/earnings to growth (PEG) formula in Easton (2004), which was recommended by Botosan and Plumlee (2005). CD is an empirical proxy for voluntary carbon disclo- sure of ﬁrm i in year t and is calculated using data from the CDP report. A CDP report presents a ﬁrm’s carbon-related activities and information. Technically, a company is assigned a score (which ranges from 0 to 5) based on the content of the information provided in the answer to a particular question in the carbon questionnaire. The ﬁnal score, CD, is a percentage that is equal to the total score earned, divided by the total score available. Thus, CD increases with the level of carbon disclosure, i.e. a higher CD value for a ﬁrm suggests that its carbon emissions and carbon-related strategies and actions are more transparent and visible than ﬁrms with lower CD. A negative coefﬁ- cient with CD is expected, which would support H1 (that a high level of carbon disclo- sure decreases the cost of capital). Following Botosan (1997), we tested the validity of the CD score by calculating correlations between the extent of disclosure and various ﬁrm-speciﬁc variables that have been shown in prior research to be signiﬁcant drivers of a ﬁrm’s level of disclo- sure. We expected that the extent of carbon disclosure would be positively associated with ﬁrm size (Botosan, 1997), ﬁnancial leverage (Botosan, 1997), and capital raising (Luo et al., 2012). Table 2 presents the correlation coefﬁcients between the extent of 198 He et al. Table 2. Correlation analysis of carbon disclosure scores and ﬁrm characteristics. Panel A: Univariate analysis SIZE LEV FIN *** ** Pearson correlation with CD 0.19 0.12 0.15 ** ** Spearman’s rho correlation with CD 0.19 0.12 0.16 Panel B: Multivariate analysis Intercept SIZE LEV FIN ** *** * ** coefﬁcient 25.9635 3.0106 12.1731 21.6269 t-statistic 2.41 3.12 1.90 2.22 p-value 0.017 0.002 0.059 0.028 adj. R 6.89% Prob. > F 0.001 Notes: This table contains Pearson correlation and Spearman’s rho correlation coefﬁcients between CD *** ** * (carbon disclosure score) and ﬁrm characteristics. indicate statistical signiﬁcance at the 1%, 5%, and 10% levels, respectively. SIZE is the natural logarithm of the market value of equity at the end of year t. LEV is the leverage ratio, which is deﬁned as the ratio of total debt divided by total assets. FIN is the amount of debt or equity capital raised by the ﬁrm, scaled by total assets at the end of year t. carbon disclosure and these ﬁrm characteristics. The ﬁndings in Table 2 reveal that ﬁrm size, ﬁnancial leverage, and ﬁrms’ raising of capital ownership are positively associated with carbon disclosure level. Overall, these ﬁndings conﬁrm the validity and reliability of the CDI. The other control variables in equation (1) are derived from prior research. Follow- ing Botosan (1997), the market model BETA, which is estimated using CRSP monthly data with a minimum of 30 out of 60 months’ worth of returns (Botosan and Plumlee, 2005), was included to control for systematic risk. Francis et al. (2008) found that earn- ings quality inﬂuenced the relationship between voluntary disclosure and the cost of capital. Hence, we included earnings quality (FD) in equation (1). FD is estimated as the absolute value of abnormal accruals, estimated based on the modiﬁed Jones model. Fama and French (1992) found that expected returns were negatively associated with ﬁrm size and positively associated with the book-to-market ratio. Thus, we included ﬁrm size (SIZE) and the market-to-book ratio (MB) in equation (1). SIZE is the natural logarithm of the market value of common equity at the end of year t. MB is the mar- ket-to-book ratio, deﬁned as the market value of equity, divided by the book value of equity at the end of year t. Firms in GHG-intensive sectors (such as utilities, energy, and materials ) have more carbon emissions and a higher propensity toward pollution (Bewley and Li, 2000; Clarkson et al., 2008) as a result of their inherent operation pro- cesses and are likely to be the targets of a wide variety of climate change regulations on the national, regional, and industry levels. Such environmental legislation may have a ﬁnancial effect and signiﬁcantly increase GHG-related liabilities and costs (Stanny and Ely, 2008). Therefore, we also included the variable INTENSITY, which is an indicator variable that equals 1 if the ﬁrm operates in a GHG-intensive industry and 0 otherwise. In all three equations, we included industry indicators to control for poten- tial industry effects. In equation (2), CD has the same deﬁnition as in equation (1). Following previ- ous literature (Clarkson, Li, Richardson, & Vasvari, 2011b), we used CP as a proxy for carbon performance, calculated as the inverse of total carbon emission per million dollars of sales turnover (net). Thus, CP is a measure of carbon efﬁciency and increases in carbon performance. A positive coefﬁcient on CP would support H2a, China Journal of Accounting Studies 199 and a negative coefﬁcient on CP would support H2b. Francis et al. (2008) found that ﬁrms with good earnings quality provide more expansive voluntary disclosures than ﬁrms with poor earnings quality. Thus, we included earnings quality (FD) in equation (2). Frankel, McNichols, and Wilson (1995) argued that ﬁrms raising capital in the public market have a greater propensity to make voluntary disclosures. We controlled for a ﬁrm’s ﬁnancing activities (FIN) by assessing the amount of debt or equity capi- tal raised by the ﬁrm during year t, scaled by total assets at the end of year t. Fol- lowing Dhaliwal et al. (2011), we also controlled for growth opportunities (TOBINQ), because ﬁrms in an expansionary period are more ﬁnancially constrained and have fewer resources for improving carbon performance and disclosure. On the other hand, growth ﬁrms also tend to have higher levels of information asymmetry, which may prompt managers to provide additional disclosures to attract potential investors (Dhal- iwal et al., 2011). Thus, the net effect of TOBINQ on carbon disclosure is unknown ex ante. TOBINQ is deﬁned as the market value of common equity plus the book value of preferred stock, book value of long-term debt, and current liabilities, scaled by the book value of total assets. Since debt holders demand greater disclosure so that they may monitor the ﬁrm’s ﬁnancial and operation activities (Leftwich, Watts, & Zimmerman, 1981), we included the debt ratio (LEV) in equation (2). LEV is the ratio of total debt divided by total assets. We also controlled for ﬁrm size (SIZE), because size captures various factors motivating ﬁrms to disclose carbon information, such as public pressure or ﬁnancial resources (Lang & Lundholm, 1993). In addition, there are some incentives for managers to enhance the liquidity of their ﬁrm’s stock via improved disclosure (Dhaliwal et al., 2011). Therefore, we included the effect of liquidity (LIQUIDITY) in equation (2). LIQUIDITY is the ratio of the number of shares traded in year t to the weighted total shares outstanding at the end of every month of year t. Prior empirical studies have explored the relationship between ﬁrms’ proﬁtability and their environmental disclosure (Aerts, Cormier, & Magnan, 2008; Bewley & Li, 2000; Clarkson et al., 2008; Magness, 2006). Thus, we controlled for return on assets (ROA) in equation (2). ROA is measured as the ratio of income before extraordinary items over total assets at the end of year t.A ﬁrm that operates in a GHG-intensive industry may have incentives to voluntarily disclose carbon infor- mation to prepare for possible future legislation to avoid the retroactive costs of regu- latory compliance (Al-Tuwaijri et al., 2004). Thus, we also included INTENSITY in equation (2). Skinner (1997) contended that ﬁrms facing a higher level of litigation risk (LITIGATION) have incentives to make voluntary disclosure to pre-empt potential lawsuits. Following Dhaliwal et al. (2011), we controlled for LITIGATION, which is an indicator variable that equals 1 if the ﬁrm operates in a high-litigation industry (Standard Industrial Classiﬁcation [SIC] codes of 2833-2836, 3570-3577, 3600-3674, 5200-5961, and 7370) and 0 otherwise (Francis, Philbrick, & Schipper, 1994; Mat- sumoto, 2002). Following Al-Tuwaijri et al. (2004), we included previous disclosure (CD )in i,t–1 equation (3). A ﬁrm’s prior carbon disclosures may represent a lower bound for current carbon performance. And investors’ expectations of carbon performance are condi- tioned on information provided by prior carbon disclosures (Al-Tuwaijri et al., 2004). If a ﬁrm expects poor carbon performance in the future, it may present full carbon information in advance to avoid ‘punishment’ in the future because it had withheld car- bon information. We expect the relationship between CD and CP to be signiﬁ- i,t–1 i,t cantly negative. Following Clarkson et al. (2011b) and Luo et al. (2012), we identiﬁed proxies for both a ﬁrm’s ﬁnancial resources and its management capabilities to examine 200 He et al. the determinants of carbon performance. We captured ﬁnancial resources using proﬁt- ability (ROA), ﬁnancing activities (FIN), and debt ratio (LEV) and used growth oppor- tunities (TOBINQ), ﬁrm size (SIZE), and market-to-book ratio (MB) to serve as proxies for unobservable management talent or capability. Following Dhaliwal et al. (2011)and Luo et al. (2012), we also included the variables INTENSITY and LITIGATION to control for the effect of potential regulatory risks, physical risks, and other risks. All other variables are as deﬁned earlier. Next, we considered the impact of carbon performance on the association between carbon disclosure and the cost of capital. Although ﬁrms may be motivated by a possi- ble reduction in the cost of equity capital when making decisions regarding carbon dis- closure, from the perspective of investors, carbon disclosure per se may not necessarily warrant a lower cost of equity capital (Dhaliwal et al., 2011). This is because corporate managers in poor-performing ﬁrms could also attempt to disclose more to achieve, restore, or maintain legitimacy (Laine, 2009; Suchman, 1995), and such disclosure obviously would not indicate good carbon performance (Bebbington, Larrinaga, & Moneva, 2008; Lindblom, 1993). A ﬁrm is a citizen of society and bound by social contracts. Thus, a ﬁrm is expected to carry out various activities considered desirable by the community in return for approval of its objectives and other rewards, and this ultimately guarantees its continued existence (Milne and Patten, 2002; Suchman, 1995). If a ﬁrm’s true social or environmental performance is below expectations, it would face a legitimacy threat. In such a case, the ﬁrm might disclose some information, instead of improving its performance, in a bid to alter the perception of the public. Patten (1992) concluded that it appears that, at least for environmental disclosures, threats to a ﬁrm’s legitimacy usually entice ﬁrms to include more information on social responsibility in their annual reports. Deegan and Rankin (1996) found a positive corre- lation between prosecution by Australian state environmental protection authorities and an increase in the level of environmental disclosure. Archel, Husillos, Larrinaga, and Spence (2009) found that ﬁrms in their study used social and environmental disclosures strategically to legitimize new production processes through the manipulation of social perceptions. Thus, we augmented equation (1) by adding a measure of a ﬁrm’s relative carbon performance (HICP) as well as the interaction between CD and HICP to equa- tion (1). HICP is an indicator variable that equals 1 if the ﬁrm’s carbon performance in year t is better than the sample median and 0 otherwise. The purpose of the interaction term is to determine whether carbon performance plays a role in the relationship between carbon disclosure and the cost of capital. All the other variables in the new system of equations are as deﬁned earlier. The following is the revised system of equa- tions: COST ¼ CD þ HICP þ CD HICP þ BETA þ FD þ SIZE þ MB i;t i;t i;t i;t i;t i;t i;t i;t i;t þ INTENSITY þ IND þ e ð1’Þ i;t i;t i;t CD ¼ CP þ FD þ FIN þ TOBINQ þ LEV þ SIZE þ LIQUIDITY i;t i;t i;t i;t i;t i;t i;t i;t þ ROA þ INTENSITY þ LITIGATION þ IND þ e ð2’Þ i;t i;t i;t i;t i;t CP ¼ CD þ FIN þ TOBINQ þ LEV þ SIZE þ MB þ ROA i;t i;t1 i;t i;t i;t i;t i;t i;t (3’) þ INTENSITY þ LITIGATION þ IND þ e i;t i;t i;t i;t China Journal of Accounting Studies 201 4. Empirical results 4.1. Descriptive statistics Table 3 provides descriptive statistics and results of t-tests for comparison of mean values of cost of capital. Table 4 depicts both the parametric and non-parametric, pairwise correlation coefﬁcients for the variables used in our tests. Table 3 shows that the mean cost of capital of all sample ﬁrms was 9.9%. When classiﬁed by the level of carbon disclosure, the cost of capital of the top 50% of ﬁrms (in terms of disclosure) (9.6%) was lower, although insigniﬁcantly so (t=1.03), than the bottom 50% of ﬁrms (10.2%), and that of the top 25% of ﬁrms (9%) was signiﬁcantly lower (t=2.12; p<0.05) than the lowest 25% of ﬁrms (10.4%) (Table 3, Line 1). This difference was also reﬂected in a signiﬁcantly negative correlation coefﬁcient for COST versus CD of –0.16 (Table 4), providing initial support for H1. The mean value for CD was 64.16%, and there was a signiﬁcant difference in CD between the top 50% and the bottom 50% of the sample ﬁrms (t=–18.05; p<0.000) and between the top 25% and bottom 25% of ﬁrms (t=–25.65, p<0.000; Table 3, line 2). This result suggests that the ﬁrms in the sample had different incentives and adopted different strategies for carbon disclosure. The mean value for CP was 0.07 (Table 3, line 3); this indicates that, on average, every 1 tonne of CO generated 0.07 million dollars of sales in the sample ﬁrms. However, there were signiﬁcant differences in carbon performance in different industries, with the lowest mean CP (0.0009) seen in the utilities industry and the highest mean CP (0.32) seen in the ﬁnancial industry. Because CP is a measure of carbon efﬁciency, the results (Table 3, line 3) show that ﬁrms in the top 50% in terms of disclosure tended to be less car- bon efﬁcient (CP=0.055) than the bottom 50% of ﬁrms (CP=0.084) (top 25% CP=0.44, bottom 25% CP=0.099). Table 4 (column 3 and 5) also presents a nega- tive relationship between carbon disclosure and carbon performance. This evidence suggests that ﬁrms with poor carbon performance disclosed more, which is consis- tent with the legitimacy theory. However, the negative correlations in both Tables 3 and 4 were not statistically signiﬁcant (p<0.15). The t-test results presented in Table 3 show that the relationships of CP, BETA, FD, TOBINQ, LEV, MB, and ROA with CD were not signiﬁcant, and FIN (t=−2.24; p<0.05) and SIZE (t=−2.84; p<0.05) were signiﬁcantly related to CD. This suggests that larger ﬁrms tended to disclose more, and ﬁrms raising capital also had a greater propensity to make voluntary carbon disclosures, which would reduce the cost of capi- tal. Note that LIQUIDITY was negatively associated with disclosure (t=1.42; and t=2.19; p<0.05), suggesting that managers with low-liquidity ﬁrms had some incentive to improve disclosure to enhance liquidity. This is consistent the with the signiﬁcant negative Spearman and Pearson correlation coefﬁcients between CD and LIQUIDITY shown in Table 4 ( p<0.10). The other results in Table 4 were generally consistent with our predictions or prior studies. For example, the cost of capital (COST) was signiﬁcantly and positively associated with systematic risk (BETA) (Botosan and Plumlee, 2005; Francis et al., 2008) and signiﬁcantly and negatively related to ﬁrm size (SIZE), earnings quality (FD), and market-to-book ratio (MB) (e.g. Fama and French, 1992; Francis et al., 2008). The negative coefﬁcient of carbon performance (CP) and INTENSITY (p<0.01) suggests that carbon efﬁciency of ﬁrms in a GHG-intensive industry was, on average, poorer than that of ﬁrms in a low carbon-intensive industry. 202 He et al. Table 3. Summary statistics and results of T-test for mean comparisons (classiﬁed by the level of carbon disclosure). CD (top 50%=1, bottom 50%=0) CD (top 25%=1, bottom 25%=0) t t CD =1 CD =0 t-value CD =1 CD =0 t-value t t t t Full Sample (n=88) (n=93) (difference) (n=48) (n=46) (difference) COST 0.099 0.096 0.102 1.03 0.090 0.104 2.12 CD 64.16 76.77 52.23 –18.05 82.94 44.58 –25.65 CD 57.21 67.56 47.42 –9.00 71.83 40.26 –10.32 t–1 CP 0.070 0.055 0.084 1.21 0.044 0.099 1.51 BETA 1.098 1.100 1.097 –0.04 0.967 1.078 1.02 FD –0.158 –0.168 –0.148 0.65 –0.157 –0.138 0.46 FIN 0.072 0.091 0.054 –2.24 0.086 0.046 –1.76 TOBINQ 1.591 1.636 1.549 –0.77 1.710 1.527 –1.12 LEV 0.597 0.610 0.586 –0.92 0.614 0.560 –1.52 SIZE 9.755 9.887 9.631 –1.49 10.125 9.435 –2.84 MB 3.300 3.570 3.043 –1.10 3.789 3.090 –0.99 LIQUIDITY 2.470 2.305 2.626 1.42 2.061 2.677 2.19 ROA 0.066 0.070 0.062 –1.04 0.075 0.063 –1.06 INTENSITY The number of low (high) INTENSITY ﬁrms is 133 (48). t t LITIGATION The number of low (high) LITIGATION ﬁrms is 134 (47). t t All continuous variables are winsorized at the 1st and 99th percentiles. Variable deﬁnitions: COST is the implied cost of equity capital in year t estimated by the PEG formula in Easton (2004), which was recommended by Botosan and Plumlee (2005). CD is a measure of the ﬁrm’s voluntary carbon disclosure in year t presented by CDP, which is an inter- national collaboration of institutional investors concerned about the business implications of climate change. CP is the ﬁrm’s carbon performance, measured as the inverse of total carbon emission per million dollars of sales turnover (net); it was also used in Clarkson et al. (2008). BETA is market-model beta calculated from the monthly CRSP stock returns during year t−5 to year t. FD is a proxy for earnings quality, measured as the absolute value of abnormal accruals, estimated using the modiﬁed Jones model. FIN is the amount t t of debt or equity capital raised by the ﬁrm, scaled by total assets at the end of year t. TOBINQ is the market value of common equity plus the book value of preferred stock, book value of long-term debt, and current liability, scaled by the book value of total assets. LEV is the leverage ratio, which is deﬁned as the ratio of total debt divided by total assets. SIZE is the natural logarithm of the market value of equity at the end of year tMB is the market-to-book ratio, deﬁned as the market value of equity divided by the book value t t of equity at the end of year t. LIQUIDITY is the ratio of the number of shares traded in year t to the weighted total shares outstanding at the end of every month of year t. ROA t t is the total return on assets, measured as the ratio of income before extraordinary items over total assets at the end of year t. INTENSITY is an indicator variable that equals 1 if the ﬁrm operates in a GHG-intensive industry and 0 otherwise; it was also used in Matsumura et al. (2010). LITIGATION is an indicator variable that equals 1 if the ﬁrm operates in a high-litigation industry (SIC codes of 2833-2836, 3570-3577, 3600-3674, 5200-5961, and 7370) and 0 otherwise. China Journal of Accounting Studies 203 Table 4. Spearman/Pearson correlation coefﬁcients. COST CD CD CP BETA FD FIN TOBINQ t t t–1 t t t t t ** ** *** COST 1.00 –0.16 –0.17 0.03 0.42 –0.08 0.01 –0.12 ** *** ** CD –0.16 1.00 0.70 –0.12 –0.06 –0.05 0.16 0.01 ** *** ** CD –0.18 0.70 1.00 –0.01 –0.05 –0.01 0.17 0.10 t–1 ** *** *** CP 0.03 –0.10 –0.07 1.00 0.18 0.70 –0.03 0.22 *** ** BETA 0.42 –0.07 –0.05 0.06 1.00 0.02 –0.04 –0.15 *** FD –0.10 –0.01 0.00 0.31 0.02 1.00 –0.05 0.05 ** ** *** FIN 0.01 0.15 0.15 –0.04 –0.04 0.02 1.00 0.27 ** *** ** TOBINQ –0.17 0.05 0.10 –0.08 –0.20 0.10 0.18 1.00 *** LEV 0.03 0.12 0.07 0.09 0.08 0.03 0.09 –0.43 *** *** *** *** *** SIZE –0.20 0.19 0.25 0.05 –0.25 –0.03 –0.09 0.21 ** * ** *** MB –0.16 0.10 0.14 –0.06 –0.17 0.04 0.08 0.50 *** ** LIQUIDITY 0.42 –0.15 –0.09 0.01 0.56 –0.02 0.11 –0.07 *** *** *** ROA –0.24 0.08 0.11 0.00 –0.28 0.01 0.06 0.68 *** *** *** INTENSITY –0.01 0.07 0.03 –0.25 –0.12 –0.50 –0.03 –0.19 *** LITIGATION –0.01 –0.03 0.08 0.02 0.09 0.11 0.04 0.30 LEV SIZE MB LIQUIDITY ROA INTENSITY LITIGATION t t t t t t t * ** *** *** COST –0.02 –0.14 –0.16 0.34 –0.23 –0.02 0.01 ** * CD 0.12 0.19 0.06 –0.14 0.03 0.07 0.00 *** * * CD 0.07 0.21 0.13 –0.14 0.08 0.03 0.08 t–1 * ** ** *** *** CP –0.14 0.09 0.19 0.11 0.19 –0.67 0.31 ** * *** *** * BETA –0.01 –0.18 –0.14 0.53 –0.24 –0.12 0.08 *** FD –0.01 0.02 0.08 0.01 0.01 –0.56 0.12 *** FIN 0.09 0.05 0.28 0.05 0.11 0.06 –0.01 *** *** *** *** *** *** TOBINQ –0.42 0.22 0.86 –0.07 0.72 –0.21 0.29 ** *** *** LEV 1.00 –0.17 0.00 –0.07 –0.44 0.03 –0.35 ** *** *** *** SIZE –0.15 1.00 0.21 –0.43 0.32 –0.06 0.08 *** ** *** *** * MB 0.28 0.07 1.00 –0.15 0.57 –0.22 0.13 *** * *** *** LIQUIDITY –0.07 –0.39 –0.13 1.00 –0.20 –0.03 0.20 *** *** *** * ** ** ROA –0.44 0.33 0.31 –0.13 1.00 –0.18 0.16 ** *** *** INTENSITY 0.03 –0.06 –0.19 –0.04 –0.20 1.00 –0.33 *** ** ** *** LITIGATION –0.38 0.09 –0.03 0.19 0.19 –0.33 1.00 t 204 He et al. Table 4 (Continued) * ** *** , , Indicate that the estimated coefﬁcient is statistically signiﬁcant at the 10%, 5%, and 1% levels, respectively. n=181. All continuous variables are winsorized at the 1st and 99th percentiles. The Spearman (Pearson) correlations are above (below) the diagonal. Variable deﬁnitions: COST is the implied cost of equity capital in year t estimated by the PEG formula in Easton (2004), which was recommended by Botosan and Plumlee (2005). CD is a measure of the ﬁrm’s voluntary carbon disclosure in year t presented by CDP, which is an international collaboration of institutional investors concerned about the business implications of climate change. CP is the ﬁrm’s carbon performance, measured as the inverse of total carbon emissions per million dollars of sales turnover (net); it was also used in Clarkson et al. (2008). BETA is market-model beta calculated from the monthly CRSP stock returns during year t−5 to year t. FD is a proxy for earnings quality, measured as the absolute value of abnormal accruals, estimated using the modiﬁed Jones model. FIN is the amount of debt or equity capital raised by the ﬁrm, scaled by total assets at the end of year t. TOBINQ is the market value of common equity plus the book value of t t preferred stock, book value of long-term debt, and current liability, scaled by the book value of total assets. LEV is the leverage ratio, which is deﬁned as the ratio of total debt divided by total assets; SIZE is the natural logarithm of the market value of equity at the end of year t. MB is the market-to-book ratio, deﬁned as the market value of equity t t divided by book value of equity at the end of year t. LIQUIDITY is the ratio of the number of shares traded in year t to the weighted total shares outstanding at the end of every month of year t. ROA is the total return on assets, measured as the ratio of income before extraordinary items over total assets at the end of year t. INTENSITY is an indicator t t variable that equals 1 if the ﬁrm operates in a GHG-intensive industry and 0 otherwise; it was also used in Matsumura et al. (2010). LITIGATION is an indicator variable that equals 1 if the ﬁrm operates in a high-litigation industry (SIC codes of 2833-2836, 3570-3577, 3600-3674, 5200-5961, and 7370) and 0 otherwise. China Journal of Accounting Studies 205 4.2. The endogeneity problem Al-Tuwaijri et al. (2004) argued that managers’ overall strategies likely affect carbon disclosure, carbon performance, and the cost of capital simultaneously and thus an OLS regression analysis is inappropriate. Hence, we examined the endogenous relationships among the three dependent variables (COST, CD, and CP) by employ- ing a Hausman (1978) test, and we rejected the null hypothesis of no endogeneity with respect to CP in equation (2) (t=–6.96, p<0.000). This result suggests that OLS estimators are potentially biased and inconsistent. We continued our analysis using two-stage least squares (2SLS) and 3SLS simultaneous equation models to control for endogeneity to obtain asymptotically unbiased results. We used the Haus- man (1978) speciﬁcation test to compare 2SLS estimates with 3SLS estimates; the results were largely similar, but 3SLS was more efﬁcient than 2SLS. Thus, here we report only 3SLS results. Our simultaneous estimation of the parameters using the 3SLS model incorporated the available information from all the equations. After sys- tem estimation with 3SLS, we used the procedures of the Hansen-Sargan test to report an over-identiﬁcation statistic. As indicated by Davidson and MacKinnon (2004, p. 532), a Hansen-Sargan test of the over-identifying restrictions is based on the 3SLS criterion function evaluated at the 3SLS point and interval parameter esti- mates. The result shows that the Hansen-Sargan over-identiﬁcation statistic was 17.899 and insigniﬁcant at the 5% level. This result conﬁrms the validity of the instruments. 4.3. Regression analysis (three-stage least squares) Table 5 reports the results of our 3SLS simultaneous equation model. Following the example of Dhaliwal et al. (2011), we created a new variable, HICP, and an interaction term, CD*HICP, to capture the impact of carbon disclosure on the cost of capital conditional on carbon performance. We report the results from both sys- tems of equations, and each system comprises three equations: one system of equations excludes (left side of Table 5) and the other includes the variable of HICP and the interaction term CD*HICP (right side of Table 5) in equation (1). Table 5 shows, without the interaction term CD*HICP in equation (1), that the coefﬁcient of CD was −0.0004 at the 10% level of signiﬁcance (t=–1.839), sug- gesting a negative relationship between carbon disclosure and the cost of capital (COST), which supports hypothesis H1. This result is also consistent with the ﬁnd- ings of Dhaliwal et al. (2011). Table 5 also shows a negative coefﬁcient for HICP, suggesting an inverse association between carbon performance and the cost of cap- ital. Next, we examined the impact of the interaction between CD and HICP; the coefﬁcient of CD was –0.001 at a higher level of signiﬁcance (p<0.05) (t=–2.485). However, the interaction term CD*HICP was signiﬁcantly positive (p<0.05). This result suggests that the negative relationship between the cost of capital and car- bon disclosure is weaker in ﬁrms with good carbon performance. This is probably because these ﬁrms disclose less information than poor performers. The result sup- ports hypothesis H1a and highlights the importance of carbon disclosure for inves- tors. It is also consistent with the legitimacy theory that ﬁrms with poor carbon performance have stronger incentives to disclose more, because these ﬁrms are more likely to have the problem of threatened legitimacy. The results of t-tests in Table 3 and regression analysis in equation (2) of Table 5 conﬁrm this interpreta- 206 He et al. Table 5. Three-stage least squares (3SLS) regression results (R – total emission) and PEG coefﬁcients (t-statistics). Dependent variables Dependent variables Independent variables COST [Eq.(1)] CD [Eq.(2)] CP [Eq.(3)] COST [Eq.(1’)] CD [Eq.(2’)] CP [Eq.(3’)] t t t t t t * ** CD –0.0004 –0.001 (–1.839) (–2.485) *** *** CD –0.001 –0.002 t–1 (–5.215) (–7.234) *** *** CP –369.860 –350.704 (–2.886) (–9.970) ** HICP –0.064 (–2.427) * ** CD HICP 0.001 t t (2.436) *** *** BETA 0.024 0.021 (4.548) (4.185) FD –0.019 –0.009 –0.017 9.266 (–1.345) (–0.001) (–1.116) (1.419) FIN 3.103 –0.019 3.111 0.001 (0.084) (–0.209) (0.145) (0.008) TOBINQ –1.581 –0.001 –1.992 –0.008 (–0.201) (–0.050) (–0.430) (–0.399) ** ** *** ** LEV 72.754 0.186 68.800 0.162 (1.973) (2.382) (3.675) (2.061) ** ** SIZE –0.005 2.155 0.005 –0.006 2.542 0.008 (–2.002) (0.549) (0.537) (–2.485) (1.094) (0.798) MB –0.001 –0.000 –0.001 0.002 (–1.327) (–0.003) (–1.360) (1.099) LIQUIDITY –1.455 –0.466 (–1.292) (–0.575) ROA 77.532 0.182 89.030 0.171 (0.662) (0.663) (1.328) (0.621) *** INTENSITY –0.030 –2.526 –0.010 0.204 –4.932 0.015 (–1.487) (–0.056) (–0.117) (6.384) (–0.246) (0.187) LITIGATION –14.309 –0.034 –13.130 –0.033 (–1.159) (–1.248) (–1.928) (–1.204) *** _cons 0.153 0.000 –0.085 0.000 2.088 –0.073 Industry indicators Yes Yes Yes Yes Yes Yes Adjusted system R 0.773 0.763 * ** *** , , Indicate that the estimated coefﬁcient is statistically signiﬁcant at the 10%, 5%, and 1% levels, respec- tively. n=181. All continuous variables are winsorized at the 1st and 99th percentiles. Variable deﬁnitions: COST is the implied cost of equity capital in year t estimated by the PEG formula in Easton (2004), which was recommended by Botosan and Plumlee (2005). CD is a measure of the ﬁrm’s voluntary carbon disclo- sure in year t presented by CDP, which is an international collaboration of institutional investors concerned about the business implications of climate change. CP is the ﬁrm’s carbon performance, measured as the inverse of total carbon emissions per million dollars of sales turnover (net); it was also used in Clarkson et al. (2008). HICP is an indicator variable that equals 1 if the ﬁrm’s CP in year t is higher than the sample median and 0 otherwise. BETA is market-model beta calculated from the monthly CRSP stock returns during year t − 5 to year t. FD is a proxy for earnings quality, measured as the absolute value of abnormal accruals, esti- mated based on the modiﬁed Jones model. FIN is the amount of debt or equity capital raised by the ﬁrm, scaled by total assets at the end of year t. TOBINQ is the market value of common equity plus the book value of preferred stock, book value of long-term debt, and current liability, scaled by the book value of total assets. LEV is the leverage ratio, which is deﬁned as the ratio of total debt divided by total assets. SIZE is t t the natural logarithm of the market value of equity at the end of year t. MB is the market-to-book ratio, deﬁned as the market value of equity divided by book value of equity at the end of year t. LIQUIDITY is the ratio of the number of shares traded in year t to the weighted total shares outstanding at the end of every China Journal of Accounting Studies 207 month of year t. ROA is the total return on assets, measured as the ratio of income before extraordinary items over total assets at the end of year t. INTENSITY is an indicator variable that equals 1 if the ﬁrm operates in a GHG-intensive industry and 0 otherwise; it was also used in Matsumura et al. (2010). LITIGATION is an indicator variable that equals 1 if the ﬁrm operates in a high-litigation industry (SIC codes of 2833-2836, 3570-3577, 3600-3674, 5200-5961, and 7370) and 0 otherwise. tion. On the other hand, it is also possible that the current format and content of the CDP report is not adequate for users to distinguish between good and poor performers. Thus, further efforts should be made to standardize the format and contents of carbon disclosure reports. The coefﬁcient estimates of the control vari- ables in equation (1) were generally consistent with the univariate analysis in Tables 3 and 4. As predicted, systematic risk (BETA) and carbon intensity of an industry (INTENSITY) were positively associated and ﬁrm size (SIZE) and earnings quality (FD) were negatively associated with the cost of capital. However, only the coefﬁcients of BETA and SIZE were signiﬁcant in both systems of equations. MB was not signiﬁcant in either system, and INTENSITY was signiﬁcant only in the system with HICP and CD*HICP. Equation (2) in Table 5 examines the determinants of carbon disclosure and shows that the coefﬁcient of CP was signiﬁcantly negative (–369.860 and –350.704; p<0.01) in both systems of equations. The result is consistent with legitimacy theory and sup- ports hypothesis H2b, suggesting that poorer performers disclose more. This evidence provides an explanation for the result obtained from equation (1) that the negative rela- tionship between carbon disclosure and cost of capital is more pronounced in ﬁrms with poorer performance. For the control variables in equation (2), the debt ratio (LEV) was signiﬁcantly positively associated with CD in both systems of equations. This is consistent with Leftwich et al. (1981), who argued that debt servicing plays a monitor- ing role and that debt holders demand greater disclosure. Litigation risk (LITIGATION) was negatively associated with CD (p<0.10), but signiﬁcantly only when HICP and CD*HICP were included in equation (1). This result is consistent with Dhaliwal et al. (2011) but inconsistent with Skinner (1997). All other control variables were insigniﬁ- cant. Equation (3) speciﬁes the determinants of carbon performance. The signiﬁcantly negative relationship between CP and past carbon disclosure (CD )(p<0.01 in both i,t–1 systems) suggests that poor carbon performers may present full carbon information in advance to avoid negative surprises and future market punishment caused by withhold- ing carbon information (Matsumoto, 2002). All other control variables in equation (3) were insigniﬁcant, except for debt ratio (LEV). In sum, the 3SLS results suggest two signiﬁcant relationships among dependent variables. First, carbon disclosure was signiﬁcantly and negatively associated with the cost of capital, suggesting that the market rewards ﬁrms with higher levels of carbon transparency. However, the negative relationship between carbon disclosure and the cost of capital was more pronounced in poor carbon performers, suggesting that without sufﬁcient disclosure, market participants may not fully understand a ﬁrm’s good carbon performance. Second, there is an inverse relationship between carbon disclosure and performance, supporting the proposition that poor carbon per- formers have more incentive to disclose carbon information as a result of the pres- sure of legitimacy. This may explain why disclosure is less effective in ﬁrms with good performance. 208 He et al. 4.4. Sensitivity analysis 4.4.1. Alternative cost of capital measure We used the PEG ratios developed by Easton (2004) to measure the implied cost of capital because extant tests show that this proxy has better construct validity than other measures (Botosan & Plumlee, 2005). Notwithstanding this evidence, we investigated the sensitivity of our main results to another ex ante cost of capital metrics: debt rating (an ex ante cost of debt measure). We used debt rating because this allowed us to side- step the debate about which is the ‘best’ ex ante cost-of-equity proxy (Francis et al., 2008), and because Sengupta (1998) concluded that disclosure quality (as measured by the scores from the Association of Investment Management and Research) is negatively associated with cost of debt. For tests of debt rating, we used S&P credit ratings (which range from AAA [highest quality] to D [default]), as available on Compustat. We employed the numeric transformation rules of S&P credit ratings in Francis et al. (2008) and excluded those ﬁrms that had no debt ratings. The results are shown in Table 6. Panel A of Table 6 shows that the mean of cost of debt capital of all sample ﬁrms was 9.062, which is close to the cost of equity capital (9.90) (Table 3). Also, there was an insigniﬁcant (signiﬁcant) difference in the cost of debt between the top 50% (25%) (i.e. high-disclosure) of ﬁrms and the bottom 50% (25%) (i.e. low- disclosure) of ﬁrms. This result is identical to the result in our main analysis in Table 3. Panel B of Table 6 shows the results of the 3SLS simultaneous equation model. In equation (1), both with and without HICP, CD was always signiﬁcantly and neg- atively associated with the cost of capital (p<0.05), which is consistent with the observations using PEG ratios as a proxy for the cost of capital (Table 5). How- ever, HICP and CD*HICP were no longer signiﬁcant, suggesting that carbon disclo- sure had a similar impact on the cost of debt in high- and low-performance ﬁrms. The results of all other control variables were also consistent with those observed using PEG ratios, with two exceptions: the indicator variable of a carbon-intensive industry (INTENSITY) was no longer statistically signiﬁcant, and market-to-book ratio (MB) was signiﬁcantly negative, consistent with the results of Dhaliwal et al. (2011). The results of equation (2) were generally consistent with those presented in our main analysis, except that the coefﬁcient of CP was statistically signiﬁcant (t=−1.864, p<0.10) only in the system with HICP, suggesting that carbon disclosure is less sensi- tive to carbon performance in the case of cost of debt. This provides evidence of the importance for controlling for endogeneity in this research setting (Al-Tuwaijri et al., 2004) and validates our adoption of a joint-estimation research design. In equation (3), we observed results identical to those presented in our main analysis, except that the coefﬁcient of proﬁtability (ROA) was statistically signiﬁcantly positive (p<0.01). Taken as a whole, our main inferences held when we replaced cost of equity with cost of debt. 4.4.2. Alternative carbon performance proxy In the prior analysis, we used the inverse of total carbon emissions per million dol- lars of sales turnover (net) to measure carbon performance. Total carbon emissions consist of three deﬁned ‘scopes’ of GHG emissions: scopes 1, 2, and 3. Scope 1 China Journal of Accounting Studies 209 Table 6. Three-stage least squares (3SLS) regression results (credit ratings –total emission) and coefﬁcients (t-statistics). Panel A: Mean comparison CD (top 50%=1, low 50%=0) CD (top 25%=1, low 25%=0) t t Full sample CD =1 CD =0 t-value CD =1 CD =0 t-value t t t t (n=33) (n=32) (difference) (n=15) (n=15) (difference) COST 9.062 8.545 9.594 1.57 8.333 10 2.14 Panel B: Three-stage least squares (3SLS) results Independent variables Dependent variables Dependent variables COST [Eq.(1)] CD [Eq.(2)] CP [Eq.(3)] COST [Eq.(1’)] CD [Eq.(2’)] CP [Eq.(3’)] t t t t t t ** ** CD –0.047 –0.057 (–2.389) (–2.307) *** *** CD –0.002 –0.002 t–1 (–2.729) (–2.871) CP –62.639 –73.711 (–1.571) (–1.864) HICP –2.620 (–1.103) CD HICP 0.012 t t (0.327) *** *** BETA 1.690 1.781 (3.639) (4.049) FD –1.940 –15.010 –1.365 –14.818 (–1.372) (–1.204) (–1.022) (–1.193) FIN 38.012 0.194 39.591 0.168 (1.186) (0.890) (1.226) (0.767) TOBINQ –1.052 –1.844 –0.045 (–0.184) (–0.323) (–0.906) *** *** LEV 26.281 0.460 29.265 0.476 (1.290) (2.781) (1.432) (2.878) *** *** SIZE –0.834 1.288 –0.008 –0.803 1.220 –0.008 (–3.080) (0.539) (–0.565) (–3.156) (0.507) (–0.564) ** ** MB –0.230 –0.007 –0.257 –0.008 (–2.098) (–0.598) (–2.485) (–0.673) (Continued) 210 He et al. Table 6. (Continued). Panel B: Three-stage least squares (3SLS) results Independent variables Dependent variables Dependent variables COST [Eq.(1)] CD [Eq.(2)] CP [Eq.(3)] COST [Eq.(1’)] CD [Eq.(2’)] CP [Eq.(3’)] t t t t t t LIQUIDITY –0.092 –0.085 (–0.707) (–0.656) *** *** ROA –30.349 1.348 –15.158 1.384 (–0.316) (2.991) (–0.158) (3.051) INTENSITY 0.386 32.622 –0.038 0.375 31.242 –0.129 (0.226) (1.334) (–0.444) (0.219) (1.269) (–0.704) LITIGATION –5.326 (–0.985) –5.252 0.018 (–0.938) 0.014 (–0.919) (0.509) *** *** _cons 18.197 0.000 –0.083 19.438 0.000 0.000 Industry indicators Yes Yes Yes Yes Yes Yes Adjusted system R 0.867 0.886 * ** *** , , Indicate that the estimated coefﬁcient is statistically signiﬁcant at the 10%, 5%, and 1% levels, respectively. n=65. All continuous variables are winsorized at the 2nd and 98th percentiles. Variable deﬁnitions: COST is the implied cost of capital in year t derived from the numeric transformation rules of S&P credit ratings in Francis et al. (2008). CD is a measure of the ﬁrm’s voluntary carbon disclosure in year t presented by CDP, which is an international collaboration of institutional investors concerned about the busi- ness implications of climate change; CP is the ﬁrm’s carbon performance, measured as the inverse of total carbon emissions per million dollars of sales turnover (net); this was also used in Clarkson et al. (2008). HICP is indicator variable that equals 1 if the ﬁrm’s CP in year t is higher than the sample median and 0 otherwise. BETA is market-model t t beta calculated from the month CRSP stock returns during year t−5 to year t. FD is a proxy for earnings quality, measured as the absolute value of abnormal accruals, estimated based on the modiﬁed Jones model. FIN is the amount of debt or equity capital raised by the ﬁrm, scaled by total assets at the end of year t. TOBINQ is the market value of t t common equity plus the book value of preferred stock, book value of long-term debt, and current liability, scaled by the book value of total assets. LEV is the leverage ratio, which is deﬁned as the ratio of total debt divided by total assets. SIZE is the natural logarithm of the market value of equity at the end of year t. MB is the market-to-book ratio, t t deﬁned as the market value of equity divided by book value of equity at the end of year t. LIQUIDITY is the ratio of the number of shares traded in year t to the weighted total shares outstanding at the end of every month of year t. ROA is the total return on assets, measured as the ratio of income before extraordinary items over total assets at the end of year t. INTENSITY is an indicator variable that equals 1 if the ﬁrm operates in a GHG-intensive industry and 0 otherwise; it was also used in Matsumura et al. (2010). LITIGA- TION is and indicator variable that equals 1 if the ﬁrm operates in a high-litigation industry (SIC codes of 2833-2836, 3570-3577, 3600-3674, 5200-5961, and 7370) and 0 other- wise. China Journal of Accounting Studies 211 Table 7. Three-stage least squares (3SLS) results (alternative carbon performance proxy) and coefﬁcients (t-statistics). Dependent variables Dependent variables Independent variables COST [Eq.(1)] CD [Eq.(2)] CP [Eq.(3)] COST [Eq.(1’)] CD [Eq.(2’)] CP [Eq.(3’)] t t t t t t *** *** CD –0.001 –0.001 (–2.962) (–3.411) *** *** CD –0.093 –0.083 t-1 (–8.166) (–7.922) *** *** CP –5.359 –8.106 (–3.186) (–5.995) *** HICP –0.077 (–3.064) * *** CD HICP 0.001 t t (3.054) *** *** BETA 0.020 0.018 (3.909) (3.522) FD –0.016 1.188 –0.013 –3.868 (–1.164) (0.159) (–0.877) (–0.542) FIN 14.525 0.841 13.186 1.020 (0.645) (0.215) (0.554) (0.263) TOBINQ 1.834 0.459 3.877 0.350 (0.359) (0.514) (0.726) (0.396) LEV 13.224 1.605 16.966 0.965 (0.686) (0.459) (0.833) (0.280) * ** SIZE –0.004 1.145 0.125 –0.006 0.449 0.068 (–1.873) (0.446) (0.287) (–2.528) (0.165) (0.155) MB –0.001 0.023 –0.001 0.065 (–1.313) (0.254) (–1.362) (0.769) LIQUIDITY –1.403 –1.067 (–1.656) (–1.278) ROA –0.696 –2.503 –16.097 –2.462 (–0.010) (–0.207) (–0.216) (–0.205) *** ** *** *** *** INTENSITY 0.139 –73.508 –14.416 0.184 –116.960 3.271 (5.903) (–2.424) (–5.900) (6.245) (–4.440) (0.699) LITIGATION –1.742 0.034 –0.907 0.257 (–0.245) (0.028) (–0.120) (0.211) *** *** *** _cons 0.000 121.505 17.321 0.000 165.478 0.000 Industry indicators Yes Yes Yes Yes Yes Yes Adjusted system R 0.701 0.705 * ** *** , , Indicate that the estimated coefﬁcient is statistically signiﬁcant at the 10%, 5%, and 1% levels, respec- tively. n=181. All continuous variables are winsorized at the 1st and 99th percentiles. Variable Deﬁnitions: COST is the implied cost of equity capital in year t estimated by the PEG formula in Easton (2004), which was recommended by Botosan and Plumlee (2005). CD is a measure of the ﬁrm’s voluntary carbon disclo- sure in year t presented by CDP, which is an international collaboration of institutional investors concerned about the business implications of climate change. CP is the ﬁrm’s carbon performance, measured as the inverse of scope 1 emissions per million dollars of sales turnover (net). HICP is indicator variable that equals 1 if the ﬁrm’s CP in year t is higher than the sample median, and 0 otherwise. BETA is market-model beta calculated from the month CRSP stock returns during year t−5 to year t. FD is a proxy for earnings quality, measured as the absolute value of abnormal accruals, estimated based on the modiﬁed Jones model. FIN is the amount of debt or equity capital raised by the ﬁrm, scaled by total assets at the end of year t. TOBINQ is the market value of common equity plus the book value of preferred stock, book value of long- term debt, and current liability, scaled by the book value of total assets. LEV is the leverage ratio, which is deﬁned as the ratio of total debt divided by total assets. SIZE is the natural logarithm of the market value of equity at the end of year t. MB is the market-to-book ratio, deﬁned as the market value of equity divided by book value of equity at the end of year t. LIQUIDITY is the ratio of the number of shares traded in year t to the weighted total shares outstanding at the end of every month of year t. ROA is the total return on assets, measured as the ratio of income before extraordinary items over total assets at the end of year t. INTENSITY t 212 He et al. is an indicator variable that equals 1 if the ﬁrm operates in a GHG-intensive industry and 0 otherwise; it was also used in Matsumura et al. (2010). LITIGATION is an indicator variable that equals 1 if the ﬁrm operates in a high-litigation industry (SIC codes of 2833-2836, 3570-3577, 3600-3674, 5200-5961, and 7370) and 0 otherwise. consists of direct emissions that come from sources that a ﬁrm owns or controls. This generally includes emissions from fossil fuel combustion for electricity, heat or steam generation, production processes for cement and steel manufacturing, transpor- tation of company-owned vehicles or aeroplanes, and fugitive emissions, such as refrigerants and methane (Kolk et al., 2008; Phillips, 2004). Scope 2 emissions are referred to as indirect emissions, which are generated from activities necessary to support production activities such as those purchased from an off-site facility, for example, electricity or steam. Scope 3 emissions are harder to deﬁne as they have a broader reach and are more all-encompassing. They are brought about by activi- ties that are part of the life cycle of the product and are created by employees, sup- pliers, customers, and contractors; for example, employees driving to work, activities embedded in purchased supplies or inventory, or transporting and disposal of prod- ucts. Scope 1 emissions are the most widely reported, because a ﬁrm can be held directly accountable for such emissions. In 2010, direct emissions (scope 1) repre- sented 1.54 billion tonnes of CO -equivalent emissions, or 84% of total emissions reported (PwC & CDP, 2010). On the other hand, indirect emissions (scope 2) come from sources where the point of release is not within the ﬁrm itself, but either upstream or downstream in the supply chain; scope 3 emissions are still not well-deﬁned, so many ﬁrms do not disclose this information. Thus, we also used the inverse of scope 1 emissions per million dollars of sales turnover (net) to mea- sure carbon performance, because there is a concern that scopes 2 and 3 emissions may be, to some extent, incomparable between ﬁrms in different sectors. Other model speciﬁcations were the same. The results are presented in Table 7 and are virtually the same as those reported in Table 5. 4.4.3. Past carbon disclosure proxy Dhaliwal et al. (2011) argued that endogeneity and self-selection issues can arise if a study examines a contemporaneous relationship between social responsibility dis- closure and the cost of capital. On the one hand, if the disclosure is motivated by a ﬁrm’s desire to reduce its high cost of capital, then researchers should observe a positive relationship between disclosure and the cost of capital. On the other hand, if disclosure leads to a lower cost of equity capital, there should be a negative rela- tionship between these factors. Therefore, the contemporaneous relationship between disclosure and the cost of capital could be ambiguous. To address the potential for endogeneity and self-selection issues, we employed a lead-lag approach in this sen- sitivity analysis, i.e. we replaced carbon disclosure (CD ) with prior carbon disclo- i,t sure (CD ) in equation (1) in our system of equations and kept the other i,t–1 speciﬁcations unchanged. The results are shown in Table 8 and are generally the same as those in Table 5, except that the coefﬁcient of carbon disclosure (CD ) i,t–1 was signiﬁcantly negative (p<0.05) only in the system that included HICP and CD*HICP. This result is consistent with Dhaliwal et al. (2011) and validates the interaction effect between carbon disclosure and carbon performance on the cost of capital in our empirical analysis. In sum, the use of a lead-lag approach did not alter the main inferences obtained from previous analyses. China Journal of Accounting Studies 213 Table 8. Three-stage least squares (3SLS) regression results (prior carbon disclosure proxy) and coefﬁcients (t-statistics). Dependent variables Dependent variables Independent variables COST [Eq.(1)] CD [Eq.(2)] CP [Eq.(3)] COST [Eq.(1’)] CD [Eq.(2’)] CP [Eq.(3’)] t t t t t t CD *** ** *** CD –0.0002 –0.001 –0.0004 –0.003 t–1 (–1.365) (–5.039) (–2.083) (–7.333) *** *** CP –396.544 –165.751 (–3.093) (–4.797) HICP –0.024 (–1.466) * * CD HICP 0.0004 t t (1.667) *** *** BETA 0.024 0.022 (4.568) (4.259) FD –0.020 1.601 –0.021 4.009 (–1.423) (0.166) (–1.371) (0.597) FIN 1.081 –0.020 13.944 0.017 (0.029) (–0.221) (0.962) (0.184) TOBINQ –1.485 –0.002 –1.494 –0.001 (–0.189) (–0.088) (-0.477) (–0.058) ** ** *** ** LEV 77.330 0.181 38.584 0.203 (2.098) (2.322) (2.915) (2.427) ** ** * SIZE –0.005 2.335 0.005 –0.006 2.841 0.012 (–2.271) (0.596) (0.516) (–2.567) (1.790) (1.217) MB –0.001 0.000 –0.001 –0.000 (–1.372) (0.153) (–1.426) (–0.020) LIQUIDITY –0.993 -0.729 (–0.880) (-0.872) ROA 82.322 0.179 47.520 0.197 (0.703) (0.652) (1.038) (0.713) *** INTENSITY 0.110 –0.240 –0.093 –0.030 13.691 0.015 (4.958) (–0.005) (–0.894) (–1.508) (0.713) (0.179) LITIGATION –15.213 –0.035 –5.915 –0.029 (–1.234) (–1.254) (–1.259) (–1.051) *** _cons 0.000 0.000 0.000 0.160 0.000 –0.099 Industry indicators Yes Yes Yes Yes Yes Yes Adjusted system R 0.773 0.754 * ** *** , , Indicate that the estimated coefﬁcient is statistically signiﬁcant at the 10%, 5%, and 1% levels, st th respectively. n=181. All continuous variables are winsorized at the 1 and 99 percentiles. Variable deﬁni- tions: COST is the implied cost of equity capital in year t estimated by the PEG formula in Easton (2004), which was recommended by Botosan and Plumlee (2005). CD is a measure of the ﬁrm’s volun- tary carbon disclosure in year t presented by CDP, which is an international collaboration of institutional investors concerned about the business implications of climate change. CP is the ﬁrm’s carbon perfor- mance, measured as the inverse of total carbon emissions per million dollars of sales turnover (net); it was also used in Clarkson et al. (2008). HICP is indicator variable that equals 1 if the ﬁrm’sCPin year t is higher than the sample median, and 0 otherwise. BETA is market-model beta calculated from the month CRSP stock returns during year t−5 to year t. FD is a proxy for earnings quality, measured as the absolute value of abnormal accruals, estimated based on the modiﬁed Jones model. FIN is the amount of debt or equity capital raised by the ﬁrm, scaled by total assets at the end of year t. TOBINQ is the market value of common equity plus the book value of preferred stock, book value of long-term debt, and current liability, scaled by the book value of total assets. LEV is the leverage ratio, which is deﬁned as the ratio of total debt divided by total assets. SIZE is the natural logarithm of the market value of equity at the end of year t. MB is the market-to-book ratio, deﬁned as the market value of equity divided by book value of equity at the end of year t. LIQUIDITY is the ratio of the number of shares traded in year t to the weighted total shares outstanding at the end of every month of year t. 214 He et al. ROA is the total return on assets, measured as the ratio of income before extraordinary items over total assets at the end of year t. INTENSITY is an indicator variable that equals 1 if the ﬁrm operates in a GHG-intensive industry and 0 otherwise; it was also used in Matsumura et al. (2010). LITIGATION is an indicator variable that equals 1 if the ﬁrm operates in a high-litigation industry (SIC codes of 2833- 2836, 3570-3577, 3600-3674, 5200-5961, and 7370) and 0 otherwise. 4.4.4. Additional analysis Al-Tuwaijri et al. (2004) used an OLS procedure and found that the results differed notably from those of 3SLS. They argued that this was evidence of the importance of controlling for endogeneity and that this validated the appropriateness of a joint- estimation research design. Thus, we also employed the OLS approach; the results are reported in Table 9. The results using OLS estimation were signiﬁcantly differ- ent from those observed using 3SLS. The adjusted system R was reduced from 0.763 (0.773) in Table 5 (3SLS estimation) to 0.610 (0.599) in Table 9 (OLS esti- mation), and the correlations between carbon performance (CP) and carbon disclo- sure (CD or CD ) were no longer signiﬁcant. Moreover, the correlation between i,t i,t–1 the cost of capital and carbon disclosure (CD ) was not signiﬁcant in the system without HICP and CD*HICP, and the levels of signiﬁcance of some control vari- ables, such as FIN, LEV, SIZE, and LITIGATION in equation (2) were changed. These results suggest that a joint-estimation research design has provided more robust and consistent ﬁndings. 5. Conclusions and limitations This study investigated the relationships between carbon disclosure, carbon perfor- mance, and the cost of capital. Our research model speciﬁcation is consistent with the argument of Ullmann (1985) that the execution of corporate responsibility is determined by management’s (unobservable) overall strategy. Thus, these constructs might be endogenous. We used a simultaneous equation research design to signiﬁ- cantly improve our analyses of the relationships among carbon performance, carbon disclosure, and the cost of capital versus that obtained by independently estimated OLS models (Al-Tuwaijri et al., 2004). We found that carbon disclosure was nega- tively associated with the cost of capital and that this negative relationship largely took place in ﬁrms with poorer carbon performance. We conjecture that this may be owing to the inadequate disclosure of good carbon performers. Consistent with this explanation, we found that disclosure of good carbon performers was signiﬁcantly lower than that of poor performers. In addition, following the example of Dhaliwal et al. (2011), we also employed a lead-lag approach and documented a negative relationship between past carbon disclosure and carbon performance. This association is consistent with the notion that a ﬁrm may present fuller carbon information in advance to avoid future punishment caused by withholding carbon information, if poor carbon performance is expected in the future. The negative relationship in dis- closure between good and poor performers is consistent with legitimacy theory. Because they are bound by social contracts, poor carbon performers are willing to increase disclosure in return for approval of their objectives and other rewards, such as a lower cost of capital, and this ultimately helps assure the continued existence of these ﬁrms. China Journal of Accounting Studies 215 Table 9. OLS regression results coefﬁcients (t-statistics) Dependent variables Dependent variables Independent variables COST [Eq.(1)] CD [Eq.(2)] CP [Eq.(3)] COST [Eq.(1’)] CD [Eq.(2’)] CP [Eq.(3’)] t t t t t t ** CD –0.0002 –0.001 (–1.115) (–2.350) CD –0.001 –0.001 t–1 (–1.291) (–1.291) CP –10.112 –10.112 (–1.151) (–1.151) ** HICP –0.047 (–2.183) * ** CD HICP 0.001 t t (2.309) *** *** BETA 0.023 0.022 (4.211) (4.024) FD –0.020 1.348 –0.019 1.348 (–1.333) (0.186) (–1.173) (0.186) ** ** FIN 22.949 –0.056 22.949 –0.056 (2.172) (–0.578) (2.172) (–0.578) TOBINQ –1.257 0.014 –1.257 0.014 (–0.538) (0.554) (–0.538) (0.554) ** ** LEV 12.724 0.238 12.724 0.238 (1.429) (2.385) (1.429) (2.385) ** ** *** ** SIZE –0.006 3.079 0.001 –0.006 3.079 0.001 (–2.347) (2.544) (0.076) (–2.638) (2.544) (0.076) MB –0.001 –0.005 –0.001 –0.005 (–1.372) (–1.011) (–1.427) (–1.011) LIQUIDITY –0.858 –0.858 (–0.939) (–0.939) ROA 17.633 0.229 17.633 0.229 (0.524) (0.779) (0.524) (0.779) *** INTENSITY 0.013 –3.349 –0.139 0.188 –3.349 –0.139 (0.625) (–0.337) (–1.194) (6.190) (–0.337) (–1.194) LITIGATION 0.283 –0.038 0.283 –0.038 (0.084) (–1.288) (0.084) (–1.288) *** _cons 0.147 26.125 0.000 0.000 26.125 0.000 Industry indicators Yes Yes Yes Yes Yes Yes Adjusted system R 0.599 0.610 * ** *** , , Indicate that the estimated coefﬁcient is statistically signiﬁcant at the 10%, 5%, and 1% levels, respectively. n=181. All continuous variables are winsorized at the 1st and 99th percentiles. Variable deﬁ- nitions: COST is the implied cost of equity capital in year t estimated by the PEG formula in Easton (2004), which was recommended by Botosan and Plumlee (2005). CD is a measure of the ﬁrm’s volun- tary carbon disclosure in year t presented by CDP, which is an international collaboration of institutional investors concerned about the business implications of climate change. CP is the ﬁrm’s carbon perfor- mance, measured as the inverse of total carbon emissions per million dollars of sales turnover (net); it was also used in Clarkson et al. (2008). HICP is indicator variable that equals 1 if the ﬁrm’sCPin year t is higher than the sample median, and 0 otherwise. BETA is market-model beta calculated from the month CRSP stock returns during year t−5 to year t. FD is a proxy for earnings quality, measured as the absolute value of abnormal accruals, estimated based on the modiﬁed Jones model. FIN is the amount of debt or equity capital raised by the ﬁrm, scaled by total assets at the end of year t. TOBINQ is the market value of common equity plus the book value of preferred stock, book value of long-term debt, and current liability, scaled by the book value of total assets. LEV is the leverage ratio, which is deﬁned as the ratio of total debt divided by total assets. SIZE is the natural logarithm of the market value of equity at the end of year t. MB is the market-to-book ratio, deﬁned as the market value of equity divided by book value of equity at the end of year t. LIQUIDITY is the ratio of the number of shares traded in year t to the weighted total shares outstanding at the end of every month of year t. ROA is the total return on assets, measured as the ratio of income before extraordinary items over total t 216 He et al. assets at the end of year t. INTENSITY is an indicator variable that equals 1 if the ﬁrm operates in a GHG-intensive industry and 0 otherwise; it was also used in Matsumura et al. (2010). LITIGATION is an indicator variable that equals 1 if the ﬁrm operates in a high-litigation industry (SIC codes of 2833- 2836, 3570-3577, 3600-3674, 5200-5961, and 7370) and 0 otherwise. These results highlight the importance of carbon disclosure for market participants and suggest that carbon disclosure is an important part of the process of fulﬁlling stake- holders’ demand for environmental accountability. This applies to all ﬁrms regardless of their carbon performance records. Like all cross-sectional studies, limitations apply to the interpretation of our results regarding whether the time period examined (2009–2010) is representative and whether the observed relationships among the variables of interest are relatively stable. For example, our sample was drawn from S&P 500 ﬁrms, which might introduce a size bias. However, because at that time CDP questionnaires were sent only to S&P 500 ﬁrms in the United States, we cannot address this issue. Whereas our results may be generalized for large ﬁrms, it would be overreaching to infer that small ﬁrms may behave similarly. In addition, this study used the PEG model of Easton (2004) to estimate the cost of capital. This model is applied only to those ﬁrms with positive expected earnings growth, which may lead to some self-selection bias, although Botosan and Plumlee (2005) found that PEG estimates were consis- tently and predictably related to risk. Finally, because our ﬁnding of a signiﬁcant association between carbon disclosure and the cost of capital implies value relevance of carbon disclosure to this measure, further research might consider using other proxies for economic performance, such as market return, to determine the perceived usefulness of carbon information. Acknowledgements We acknowledge ﬁnancial support from the National Social Science Fund (Research grant number: 13BGL057), the Humanity and Social Science Foundation of Ministry of Education (Research grant number: 09YJC790137), the Social Science Foundation of Jiang Su Prov- ince, People’s Republic of China (Research grant number: 09SHD012), the Jiangsu Provin- cial Department of Education Fund of Philosophy and Social Science (2013SJB6300036), as well as project funding from the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD) and the constructive suggestions of two anonymous reviewers. Note 1. Utilities, energy, and materials are classiﬁed as carbon-intensive sectors (Tang & Luo, 2011). References Adams, C. A. (2004). The ethical, social and environmental reporting-performance portrayal gap. Accounting, Auditing and Accountability Journal, 17, 731–757. Aerts, W., Cormier, D., & Magnan, M. (2008). Corporate environmental disclosure, ﬁnancial markets and the media: An international perspective. Ecological Economics, 64, 643– Al-Tuwaijri, S. A., Christensen, T. E., & Hughes, K. E. (2004). The relations among environmen- tal disclosure, environmental performance, and economic performance: A simultaneous equa- tions approach. Accounting, Organizations and Society, 29, 447–471. China Journal of Accounting Studies 217 Amihud, Y., & Mendelson, H. (1986). Asset pricing and the bid-ask spread. Journal of Financial Economics, 17, 223–249. Archel, P., Husillos, J., Larrinaga, C., & Spence, C. (2009). Social disclosure, legitimacy the- ory and the role of the state. Accounting, Auditing & Accountability Journal, 22, 1284– Arenas, D., Lozano, J. M., & Albareda, L. (2009). The role of NGOs in CSR: Mutual percep- tions among stakeholders. Journal of Business Ethics, 88, 175–197. Bebbington, J., Larrinaga, C., & Moneva, J. M. (2008). Corporate social reporting and reputation risk management. Accounting, Auditing & Accountability Journal, 21, 337–361. Bewley, K., & Li, Y. (2000). Disclosure of environmental information by Canadian manufactur- ing companies: A voluntary disclosure perspective. Advances in Environmental Accounting and Management, 1, 201–226. Botosan, C. A. (1997). Disclosure level and the cost of equity capital. The Accounting Review, 72, 323–349. Botosan, C. A., & Plumlee, M. A. (2002). A re-examination of disclosure level and the expected cost of equity capital. Journal of Accounting Research, 40,21–40. Botosan, C. A., & Plumlee, M. A. (2005). Assessing alternative proxies for the expected risk pre- mium. The Accounting Review, 80,21–53. Buhr, N. (1998). Environmental performance, legislation and annual report disclosure: The case of acid rain and Falconbridge. Accounting, Auditing & Accountability Journal, 11, 163–190. Campbell, D. (2003). Intra- and intersectoral effects in environmental disclosures: Evidence for legitimacy theory? Business Strategy and the Environment, 12, 357–371. Clarkson, P. M., Li, Y., Richardson, G. D., & Vasvari, F. P. (2008). Revisiting the relation between environmental performance and environmental disclosure: An empirical analysis. Accounting Organizations and Society, 33, 303–327. Clarkson, P. M., Fang, X. H., & Li, Y. (2011a). The relevance of environmental disclosures for investors and other stakeholder groups: are such disclosures incrementally informative? Working paper. Clarkson, P. M., Li, Y., Richardson, G. D., & Vasvari, F. P. (2011b). Does it really pay to be green? Determinants and consequences of proactive environmental strategies. Journal of Accounting and Public Policy, 30, 122–144. Clarkson, P. M., Overell, M. B., & Chapple, L. (2011c). Environmental reporting and its relation to corporate environmental performance. Abacus, 47,27–60. Cogan, D. C. (2006). Corporate Governance and Climate Change: Making the Connection. Boston, MA: Ceres. Core, J. E. (2001). A review of the empirical disclosure literature: Discussion. Journal of Accounting and Economics, 31, 441–456. Cormier, D., Magnan, M., & Van Velthoven, B. (2005). Environmental disclosure quality in large German companies: Economic incentives, public pressures or institutional conditions? Euro- pean Accounting Review, 14,3–39. Davidson, R., & mackinnon, J. (2004). Econometric theory and methods. New York: Oxford Uni- versity Press. Decker, C. S. (2002). Shifting the burden: Regulatory response to pollution control. Journal of Economic Research, 7,41–76. Decker, C. S., & Pope, C. R. (2005). Adherence to environmental law: The strategic complementarities of compliance decisions. The Quarterly Review of Economics and Finance, 45, 641–661. Deegan, C., & Gordon, B. (1996). A study of environmental disclosure practices of Australian corporations. Accounting and Business Research, 26, 187–199. Deegan, C., & Rankin, M. (1996). Do Australian companies report environmental news objec- tively? Accounting, Auditing and Accountability Journal, 9,50–67. Dhaliwal, D. S., Li, O. Z., Tsang, A., & Yang, Y. G. (2011). Voluntary nonﬁnancial disclosure and the cost of equity capital: The initiation of corporate social responsibility reporting. The Accounting Review, 86,59–100. Diamond, D. W., & Verrecchia, R. E. (1991). Disclosure, liquidity, and the cost of capital. The Journal of Finance, 46, 1325–1359. 218 He et al. Dye, R. A. (1985). Disclosure of nonproprietary information. Journal of Accounting Research, 23, 123–145. Easton, P. D. (2004). PE Ratios, PEG Ratios, and Estimating the Implied Expected Rate of Return on Equity Capital. The Accounting Review, 79,73–95. Fama, E. F., & French, K. R. (1992). The cross-section of expected stock returns. The Journal of Finance, 47, 427–465. Francis, J., Nanda, D., & Olsson, P. (2008). Voluntary disclosure, earnings quality, and cost of capital. Journal of Accounting Research, 46,53–99. Francis, J., Philbrick, D., & Schipper, K. (1994). Shareholder litigation and corporate disclosures. Journal of Accounting Research, 32, 137–164. Frankel, R., mcnichols, M., & Wilson, G. P. (1995). Discretionary disclosure and external ﬁnanc- ing. The Accounting Review, 70, 135–150. Freedman, M., & Wasley, C. (1990). The association between environmental performance and environmental disclosure in annual reports and 10 Ks. Advances in Public Interest Account- ing, 3, 183–193. Graham, J. R., Harvey, C. R., & Rajgopal, S. (2005). The economic implications of corporate ﬁnancial reporting. Journal of Accounting and Economics, 40,3–73. Gray, R., Kouhy, R., & Lavers, S. (1995). Corporate social and environmental reporting, A review of the literature and a longitudinal study of UK disclosure. Accounting, Auditing & Accountability Journal, 8,47–77. Hausman, J. A. (1978). Speciﬁcation tests in econometrics. Econometrica, 46, 1251–1271. Healy, P. M., & Palepu, K. G. (2001). Information asymmetry, corporate disclosure, and the capital markets: A review of the empirical disclosure literature. Journal of Accounting and Economics, 31, 405–440. Hughes, S. B., Anderson, A., & Golden, S. (2001). Corporate environmental disclosures: Are they useful in determining environmental performance? Journal of Accounting and Public Policy, 20, 217–240. Ingram, R. W., & Frazier, K. B. (1980). Environmental performance and corporate disclosure. Journal of Accounting Research, 18, 614–622. Kolk, A., Levy, D., & Pinkse, J. (2008). Corporate responses in an emerging climate regime: The institutionalization and commensuration of carbon disclosure. European Accounting Review, 17, 719–745. Laine, M. (2009). Ensuring legitimacy through rhetorical changes? A longitudinal interpretation of the environmental disclosures of a leading Finnish chemical company. Accounting, Audit- ing & Accountability Journal, 22, 1029–1054. Lambert, R., Leuz, C., & Verrecchia, R. E. (2007). Accounting information, disclosure, and the cost of capital. Journal of Accounting Research, 45, 385–420. Lang, M., & Lundholm, R. (1993). Cross-sectional determinants of analyst ratings of corporate disclosures. Journal of Accounting Research, 31, 246–271. Leftwich, R. W., Watts, R. L., & Zimmerman, J. L. (1981). Voluntary corporate disclosure: The case of interim reporting. Journal of Accounting Research, 19,50–77. Leuz, C., & Verrecchia, R. E. (2000). The economic consequences of increased disclosure. Journal of Accounting Research, 38,91–124. Leuz, C., & Wysocki, P. (2008). Economic consequences of ﬁnancial reporting and disclosure regulation: A review and suggestions for future research. University of Chicago. Levy, D. L., & Egan, D. (2003). A neo-Gramscian approach to corporate political strategy: Conﬂict and accommodation in the climate change negotiations. Journal of Management Studies, 40, 803–830. Li, Y., Richardson, G. D., & Thornton, D. B. (1997). Corporate disclosure of environmental liability information: Theory and evidence. Contemporary Accounting Research, 14, 435– Lindblom, C. (1993). The implications of organizational legitimacy for corporate social performance and disclosure. Paper presented at the Critical Perspectives on Accounting Conference, New York. Lindblom, C. (1994). The implications of organizational legitimacy for corporate social performance disclosure. Paper presented at the Critical Perspectives on Accounting Confer- ence. New York. China Journal of Accounting Studies 219 Luo, L., Lan, Y. C., & Tang, Q. (2012). Corporate incentives to disclose carbon information: Evidence from the CDP Global 500 Report. Journal of International Financial Management & Accounting, 23,93–120. Magness, V. (2006). Strategic posture, ﬁnancial performance and environmental disclosure: An empirical test of legitimacy theory. Accounting, Auditing & Accountability Journal, 19, 540–563. Margolick, M., & Russell, D. (2004). Corporate greenhouse gas reduction targets. Arlington VA: Pew Center on Global Climate Change/Global Change Strategies International. Matsumoto, D. A. (2002). Management’s incentives to avoid negative earnings surprises. The Accounting Review, 77, 483–514. Matsumura, E., Prakash, R., & Vera-Munoz, S. (2010). Carbon emissions and ﬁrm value. Avail- able at SSRN 1688738. Merton, R. C. (1987). A simple model of capital market equilibrium with incomplete information. The Journal of Finance, 42, 483–510. Milne, M. J., & Patten, D. M. (2002). Securing organizational legitimacy: An experimental decision case examining the impact of environmental disclosures. Accounting, Auditing & Accountability Journal, 15, 372–405. O’Donovan, G. (2002). Environmental disclosures in the annual report: Extending the applicabil- ity and predictive power of legitimacy theory. Accounting, Auditing & Accountability Journal, 15, 344–371. O’Dwyer, B. (2005). Stakeholder democracy: Challenges and contributions from social account- ing. Business Ethics: A European Review, 14,28–41. Patten, D. M. (1992). Intra-industry environmental disclosures in response to the Alaskan oil spill: A note on legitimacy theory. Accounting, Organizations and Society, 17, 471–475. Patten, D. M. (2002). The relation between environmental performance and environmental disclo- sure: A research note. Accounting, Organizations and Society, 27, 763–773. Phillips, G. (Ed.) (2004). Monitoring, veriﬁcation and accounting of emissions. London: Risk Books. Plumlee, M., Brown, D., & Marshall., S. (2008). The impact of voluntary environmental disclo- sure quality on ﬁrm value. University of Utah. PwC, & CDP. (2010). Carbon Disclosure Project 2010 S&P 500 Report. Richardson, A. J., & Welker, M. (2001). Social disclosure, ﬁnancial disclosure and the cost of equity capital. Accounting, Organizations and Society, 26, 597–616. Richardson, A. J., Welker, M., & Hutchinson, I. R. (1999). Managing capital market reactions to corporate social responsibility. International Journal of Management Reviews, 1,17–43. Salop, S. C., & Scheffman, D. T. (1983). Raising rivals’ costs. American Economic Review, 73, 267–271. Sengupta, P. (1998). Corporate disclosure quality and the cost of debt. The Accounting Review, 73, 459–474. Sinkin, C., Wright, C. J., & Burnett, R. D. (2008). Eco-efﬁciency and ﬁrm value. Journal of Accounting and Public Policy, 27, 167–176. Skinner, D. J. (1997). Earnings disclosures and stockholder lawsuits. Journal of Accounting and Economics, 23, 249–282. Stanny, E., & Ely, K. (2008). Corporate environmental disclosures about the effects of climate change. Corporate Social Responsibility and Environmental Management, 15, 338–348. Suchman, M. C. (1995). Managing legitimacy: Strategic and institutional approaches. The Acad- emy of Management Review, 20, 571–610. Tang, Q., & Luo, L. (2011). Transparency of corporate carbon disclosure: International evi- dence. School of Business, University of Western Sydney. Tang, Q., & Luo, L. (2013). Carbon management systems and carbon mitigation, Australian Accounting Review, 23. Tran, J., Okafor, O., & Herremans, I. M. (2011). GAAP for GHG emissions accounting: How consistent are GHG emissions reported among different channels? Presented at the CAAA Annual Conference. Ullmann, A. A. (1985). Data in search of a theory: A critical examination of the relationships among social performance, social disclosure, and economic performance of U.S. ﬁrms. The Academy of Management Review, 10, 540–557. 220 He et al. Verrecchia, R. E. (1983). Discretionary disclosure. Journal of Accounting and Economics, 5, 179–194. Verrecchia, R. E. (2001). Essays on disclosure. Journal of Accounting and Economics, 32, 97–180. Wiseman, J. (1982). An evaluation of environmental disclosures made in corporate annual reports. Accounting, Organizations and Society, 7,53–63.

Journal

China Journal of Accounting Studies – Taylor & Francis

Published: Dec 1, 2013

Keywords: carbon disclosure; carbon performance; cost of capital; simultaneous equations

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Carbon disclosure, carbon performance, and cost of capital

Carbon disclosure, carbon performance, and cost of capital

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Carbon disclosure, carbon performance, and cost of capital

Carbon disclosure, carbon performance, and cost of capital

References (85)

Abstract

Journal

Recommended Articles

There are no references for this article.

Our policy towards the use of cookies