Faculty of Business and Economics Chair of Business Management, esp. Management Accounting and Management Control

Transcription

1 Faculty of Business and Economics Chair of Business Management, esp. Management Accounting and Management Control Interrelations among Corporate Environmental Performance, Corporate Environmental Disclosure and Corporate Financial Performance: Addressing the endogeneity problem Christoph Trumpp Thomas W. Guenther, TU Dresden, Germany

2 Agenda 1. Introduction 2. Literature Review and underlying theories 3. Research Design 4. Empirical Results 5. Conclusion Folie 2 von 19

3 1 Introduction Motivation research on the relationship between Corporate Environmental Performance (CEP), Corporate Environmental Disclosure (CED) and Corporate Financial Performance (CFP) characterized as data in search of a theory (Ullmann, 1985) CEP increasingly interesting to customers, investors, and other stakeholders (de Villiers and van Staden, 2011) every company decision generates an impact on the natural environment, even though the organization may be unaware of what these impacts are (Etzion, 2007) increase of social responsible investments all around the world which is reflected in a growth of CED (Clarkson et al., 2011b; KPMG, 2011) Folie 3 von 19

4 1 Introduction Endogeneity huge amount of research on the pairwise relationships between CEP, CED and CFP with contradictory results CEP-CFP: e.g., Bush & Hoffmann (2011) vs. Cordeiro & Sarkis (1997) CEP-CED: e.g., Clarkson et al. (2008) vs. Clarkson et al. (2011b) CED-CFP: e.g., Jones et al. (2007) vs. Wu et al. (2010) separate dimensions of CFP: profitabilty (PRO) and stock market performance (SMA) (Hamann et al., 2013) omitted variable influencing CEP and CFP (Ambec and Lanoie, 2008) interrelations between the three constructs, because of empirical tested pairwise relationships simultaneity omitted variables and simultaneity may cause endogeneity (Chenhall and Moers, 2007) We want to clarify contradictory results by analyzing the relationships between the three constructs CEP, CED and CFP simultaneously. Folie 4 von 19

5 1 Introduction Previous literature Al-Tuwaijri et al. (2004) refers to Ullmann (1985): idea of a corporate strategy sample of 198 US firms, minimum of environmental exposure, 1994 data CEP: TRI data; CED: scoring of 4 indicators; CFP: stock market performance address the problem of endogeneity with three-stage least squares (3SLS) simultaneous equation approach identify endogeneity for CEP relationships not fully specified in structural equations dependent variable CEP CED CFP CEP + + CED t-1 + O CFP O Folie 5 von 19

6 1 Introduction Previous literature Clarkson et al. (2011a), robustness check concept of Al-Tuwaijri et al. (2004) as robustness analysis in their study of CEP-CFP relationship sample of 191 US firms from 5 polluting industries, 2003 data CEP: TRI data; CED: scoring similar to Clarkson et al. (2008); CFP: stock market performance results indicate no endogeneity problem with respect to CEP and CFP different results to Al-Tuwaijri et al. (2004) dependent variable CEP CED CFP CEP + + CED + + CFP + O Folie 6 von 19

7 2 Literature Review and underlying theories Relation between CEP and CFP synergetic model of the CEP-CFP relationship: slack resource theory (Preston & O Bannon, 1997), natural-resource-based view (NRBV) (Hart, 1995; Chan, 2005) and instrumental stakeholder theory (Jones, 1995) indicate that CEP and CFP connected reciprocally (Preston and O Bannon, 1997; Orlitzky et al., 2003; Surroca et al., 2010) meta-analyses summarize prior empirical studies CEP-Focus: Dixon-Fowler et al. (2013) (+) CSP-Focus with CEP subgroup: Orlitzky et al. (2003) (+); Allouche and Laroche (2005) (+); Margolis et al. (2007) (+) H1a: Corporate environmental performance is positively associated with profitability. H1b: Corporate environmental performance is positively associated with stock market performance. Folie 7 von 19

8 2 Literature Review and underlying theories Relation between CEP and CED voluntary disclosure theory: good environmental performers will communicate their position by pointing out an extensive environmental report (Dye, 1985; Verrecchia, 1983; Clarkson et al., 2008) socio-political theories: CED result of political and social pressure facing the company (Gray et al. 1995; Patten, 2002; Clarkson et al., 2008) CED measured by content analysis: Patten (2002) (-); Cho and Patten (2007) (-); Clarkson et al. (2008) (+); Clarkson et al. (2011b) (-) CED measured by quantitative score: de Villiers and van Staden (2011) (-) CED measured by disclosure 0/1: Cho et al. (2012) (-); Silva-Gao (2012) (+) addressing climate change disclosure: Dawkins and Fraas (2011a) (+) H2: Corporate environmental performance is associated with corporate environmental disclosure. Folie 8 von 19

9 2 Literature Review and underlying theories Relation between CED and CFP lack of specifying theories of CED-CFP relation (Gray et al., 2001) influence of CFP on CED: voluntary disclosure theory, socio-political theories influence of CED on CFP: agency theory influence of CFP on CED: Jones et al. (2007) (+); Gamerschlag et al. (2011) (+); Salama et al. (2012) (o); Luo et al. (2012) (o) influence of CED on CFP: Murray et al. (2006) (o); Cormier and Magnan (2007) (o); Clarkson et al. (2011c) (+) H3a: Corporate environmental disclosure is positively associated with profitability. H3b: Corporate environmental disclosure is positively associated with stock market performance. H4: Corporate environmental performance, corporate environmental disclosure and corporate financial performance are jointly interrelated. Folie 9 von 19

10 2 Literature Review and underlying theories H4: joint interrelation H4: Corporate environmental performance, corporate environmental disclosure and corporate financial performance are jointly interrelated. Folie 10 von 19

11 3 Research Design Addressing the endogeneity problem endogeneity problem exists when an endogenous variable is part of a regression model as explanatory variable (Chenhall and Moers, 2007) omitted variables: an important variable which is associated with the explained and the explanatory variable is missing in the analysis simultaneity: one or more of the explanatory variables are interrelated with the explained variable endogeneity test of Hausmann (1978) three-stage least squares (3SLS) simultaneous equations model to address the endogeneity problem (Hamilton and Nickerson, 2003; Chenhall and Moers, 2007) compare results to single OLS-regression of each structural equation Folie 11 von 19

12 3 Research Design Empirical model CEP it = β 1,0 + β 1,1 CFP it + β 1,2 CED it + β 1,3 R&D it + β 1,4 NEW it + β 1,5 FAI it + β 1,6 GRO it + β 1,7 CF it + β 1,8 SIZE it + β 1,9 CCPI it + β 1,10 LEGOR + β 1,11 CSI + β 1,12 20 INDUSTRY Control i + β 1,21 23 YEAR Control t (2.1) + β 1, FIRM Control i + ε 1 CED it = β 2,0 + β 2,1 CEP it + β 2,2 CFP it +β 2,3 CAPIN it + β 2,4 LEV it + β 2,5 FIN it + β 2,6 VOL it + β 2,7 CONT it + β 2,8 SIZE it + β 2,9 CCPI it + β 2,10 LEGOR + β 2,11 CSI + β 2,12 20 INDUSTRY Control i + β 2,21 23 YEAR Control t (2.2) + β 2, FIRM Control i + ε 2 CFP it = β 3,0 + β 3,1 CEP it + β 3,2 CED it + β 3,3 R&D it + β 3,4 CAPIN it + β 3,5 GRO it + β 3,6 CF it + β 3,7 LEV it + β 3,8 SIZE it + β 3,9 LEGOR+β 3,10 CSI + β 3,11 19 INDUSTRY Control i + β 3,20 22 YEAR Control t + β 3, FIRM Control i + ε 3 Folie 12 von 39 (2.3)

13 3 Research Design Sample description CDP Global 500, S&P 500 or FTSE 350 reports during 2008 to 2011 financial data Thomson Reuters Datastream Worldscope Global 500, S&P 500, FTSE 350 years Sample firm-years listwise deletion Winsorising 1% final Sample firm years (643 firms) Folie 13 von 19

14 4 Results Endogeneity test Endogeneity test recommended by Hausman (1978) reject null hypothesis of no endogeneity for CEP (t = , p < 0.000), CED (t = , p < 0.000) and CFP (t = , p < 0.000) H4 is supported Folie 14 von 19

15 Results of Three-stage least squares panel regression (profitability) CEP-equation (2.1) CED-equation (2.2) CFP-equation (2.3) Variable Exp. Exp. Exp. Coef. z-stat. Coef. z-stat. Sign Sign Sign Coef. z-stat. Endogenous variables CEP +/ *** 7.34 CED +/ * CFP (ROA) *** Exogenous variables R&D NEW ** 2.13 FAI *** 2.64 CAPIN ** GRO *** *** 6.66 CF LEV *** FIN VOL CONT SIZE *** *** CCPI *** * LEGOR ** 2.20 CSI N 1,870 1,870 1,870 χ²-statistic 23, *** 7, *** *** Note: The table represents the three-stage least squares regression with a system of three structural equations concerning Corporate Environmental Performance (CEP), Corporate Environmental Disclosure (CED) and Corporate Financial Performance (CFP), i.e. profitability. The table reports χ²-statistic for the overall fit of the model, the coefficients (Coef.) with the resulted sign and z- statistics (z-stat.). The variables are defined in Table 1. * p<0.10 ** p<0.05 *** p<0.01. Folie 15 von 39

16 Results of Three-stage least squares panel regression (stock market performance) CEP-equation (2.1) CED-equation (2.2) CFP-equation (2.3) Variable Exp. Exp. Exp. Coef. z-stat. Coef. z-stat. Sign Sign Sign Coef. z-stat. Endogenous variables CEP +/ ** 2.08 CED +/ *** 4.66 CFP (TSR) ** *** 3.46 Exogenous variables R&D ** NEW FAI CAPIN ** GRO *** 4.85 CF * LEV ** *** FIN ** VOL CONT SIZE *** *** *** CCPI LEGOR CSI N 1,860 1,860 1,860 χ²-statistic 17, *** 167, *** 1, *** Note: The table represents the three-stage least squares regression with a system of three structural equations concerning Corporate Environmental Performance (CEP), Corporate Environmental Disclosure (CED) and Corporate Financial Performance (CFP), i.e. stock market performance. The table reports χ²-statistic for the overall fit of the model, the coefficients (Coef.) with the resulted sign and z- statistics Vienna, (z-stat.). Oktober The 8, variables 2013 are defined in table 1. * p<0.10 ** p<0.05 *** p<0.01. Folie 16 von 39

17 5 Conclusion Folie 17 von 19

18 5 Conclusion Contributions 1. empirical evidence that CEP, CED and CFP (profitability as well as stock market performance) are endogenous variables only simultaneous investigation with all three constructs provides undistorted results 2. empirical evidence that profitability and stock market performance are different dimensions of CFP conscious choice based on theoretical arguments in further research 3. empirical evidence for a virtuous circle between CEP and CFP after controlling for endogeneity, regardless if CFP is represented by profitability or stock market performance 4. carbon disclosure to CDP is more driven by capital market expectations than by CEP validity of voluntary disclosure theory and the agency theory in the context of carbon disclosure Folie 18 von 19

19 Thank you for attention! Folie 19 von 19

20 Back-up Folie 20 von 19

21 Definitions Corporate Environmental Performance (CEP) results of an organization s management of its environmental aspects (ISO, 1999) multidimensional construct comprising separate dimensions (Lefebvre et al., 2003; Clemens and Bakstran, 2010) Corporate Environmental Disclosure (CED) the set of information items that relate to a firm s past, current and future environmental management activities and performance (Berthelot et al., 2003) Corporate Financial Performance (CFP) economic outcomes resulting from the interplay among an organization s attributes, actions, and environment (Combs et al., 2005) consists of four separate dimensions: liquidity, profitability, growth, and stock market performance (Hamann et al., 2013) Folie 21 von 19

22 Variable Measurement in equation Endogenous variables CEP Corporate Environmental Performance is measured as the negative carbon intensity, where carbon intensity is calculated as total carbon emissions divided by sales. dv x x CED Corporate Environmental Disclosure is measured with the CDP s carbon disclosure score which represent the quality and quantity of answering the CDP questionnaire. x dv x CFP Corporate Financial Performance is measured with two variables from different dimensions of CFP. First, Total Shareholder Return (TSR) is used as indicator for stock market performance. Second, Return on Assets (ROA) is used as indicator for profitability. x x dv Exogenous variables R&D Research and development intensity is measured as R&D expenses divided by sales. x x NEW Newness of equipment is measured as net property, plant and equipment divided by gross x property, plant and equipment. FAI The fixed asset intensity is measured as net property, plant and equipment in relation to total assets. x CAPIN Capital intensity is measured as capital expenditures divided by beginning of period total assets. x x GRO Growth is measured as change in total assets divided by beginning of period total assets. x x CF Cash flow return on sales is measured as net cash flow divided by sales. x x LEV Leverage is measured as total debt divided by total assets. x x FIN Financing raised by a firm is measured as annually change of a company s long term x debt, common equity and preferred stock divided by total assets. VOL Volatility of stock prices is measured as standard deviation of market-adjusted monthly x stock returns (Clarkson et al. 2008, p. 314). CONT Controversies are the sum of all the controversies which a company is exposed as x provided by Thomson Reuters Datastream Asset4 database. SIZE The Company size is measured by natural logarithm of total assets. x x x CCPI The Climate Change Performance Index is provided by Germanwatch. The index x x evaluates the efforts of a country to climate protection (Burck and Bals 2011, p. 3). LEGOR Legal origin represents the law system of a country (La Porta et al. 2008, p. 288). The x x x dichotomous variable equals 1 if the country is a common law country and zero otherwise. CSI The carbon-sensitive industry is based on the average carbon intensity of an industry. Based on CDP it is assumed that Energy, Health Care, Industrials, Materials and Utilities are more carbon-intensive than other industries (CDP, 2009, p. 17). x x x Vienna, Oktober The 8, 2013 dichotomous variable equals 1 if the sector is carbon-intensive and cero otherwise. Note: dv: dependent variable. Folie 22 von 19

23 firms per year firm-years Total Continent Africa Asia Australia Europe North America South America Total ,870 Industry Oil& Gas (ICB 0001) Basic Materials (ICB 1000) Industrials (ICB 2000) Consumer Goods (ICB 3000) Health Care (ICB 4000) Consumer Services (ICB 5000) Telecommunications (ICB 6000) Utilities (ICB 7000) Financials (ICB 8000) Technology (ICB 9000) Total ,870 Note: The overall sample consists of 1,870 firm-years in 643 firms. ICB = Industrial Classification Benchmark. 1 Carbon-intensive industry as declared by CDP (CDP, 2009, p. 17). Folie 23 von 19

24 Variable Mean Median Minimum Maximum S.D. Endogenous variables CEP CED CFP (TSR) CFP (ROA) Exogenous variables R&D NEW FAI CAPIN GRO CF LEV FIN VOL CONT SIZE CCPI LEGOR CSI Note: The table reports descriptive statistics for the endogenous variables Corporate Environmental Performance (CEP), Corporate Environmental Disclosure (CED) and Corporate Financial Performance (CFP) and the exogenous variables. The variables are defined in Table 1. Folie 24 von 19

25 Results of one-way clustered OLS panel regression (profitability) CEP-equation (2.1) CED-equation (2.2) CFP-equation (2.3) Variable Exp. Exp. Exp. Coef. t-stat. Coef. t-stat. Sign Sign Sign Coef. t-stat. Endogenous variables CEP +/ * 1.83 CED +/ *** 2.93 CFP (ROA) * 1.70 Exogenous variables R&D *** *** NEW FAI ** CAPIN *** 3.24 GRO *** 7.82 CF *** 6.58 LEV *** *** FIN * VOL ** CONT *** 2.89 SIZE *** *** CCPI ** LEGOR CSI *** N 1,870 1,870 1,870 R-squared F-statistic 9.10 *** *** *** Highest VIF Note: The table represents the ordinary least squares regression as recommended by Petersen (2009, p. 458) concerning Corporate Environmental Performance (CEP), Corporate Environmental Disclosure (CED) and Corporate Financial Performance (CFP), i.e. profitability. The table reports R-squared and F-statistic for the overall fit of the model, the coefficients (Coef.) with the resulted sign and t-statistics (t-stat.). The variables are defined in Table 1. * p<0.10 ** p<0.05 *** p<0.01. Folie 25 von 39

26 Results of one-way clustered OLS panel regression (stock market performance) CEP-equation (2.1) CED-equation (2.2) CFP-equation (2.3) Variable Exp. Exp. Exp. Coef. t-stat. Coef. t-stat. Sign Sign Sign Coef. t-stat. Endogenous variables CEP +/ *** 2.93 CED +/ CFP (TSR) * Exogenous variables R&D *** *** NEW FAI *** CAPIN GRO * *** 5.43 CF *** 4.71 LEV *** FIN VOL *** CONT *** 3.11 SIZE *** *** CCPI ** LEGOR CSI *** N 1,860 1,860 1,860 R-squared F-statistic 9.24 *** *** *** Highest VIF Note: The table represents one-way clustered ordinary least squares regression as recommended by Petersen (2009, p. 458) concerning Corporate Environmental Performance (CEP), Corporate Environmental Disclosure (CED) and Corporate Financial Performance (CFP), i.e. stock market performance. The table reports R-squared and F-statistic for the overall fit of the model, the coefficients (Coef.) with the resulted sign and t-statistics (t-stat.). The variables are defined in Table 1. * p<0.10 ** p<0.05 *** p<0.01. Folie 26 von 39