CHEMICAL ENGINEERING TRANSACTIONS VOL. 62, 2017 A publication of The Italian Association of Chemical Engineering Online at www.aidic.it/cet Guest Editors: Fei Song, Haibo Wang, Fang He Copyright © 2017, AIDIC Servizi S.r.l. ISBN 978-88-95608- 60-0; ISSN 2283-9216 Empirical Study for Influencing Factors on Environmental Accounting Information Disclosure in Chemical Industry Nian Yanga, Zhen Lib,Yuan Fengb a Economics and Trade Department, Hebei Finance University, Baoding 071000, China b Xingtai Polytechnic College, Xingtai 054035, China yangshirly@aliyun.com In the cyclic and low-carbon economic environment, it is of great significance for the sustainable development of the whole society to strengthen the study for environmental accounting (EA) information disclosure. Now, the number of enterprises for EA information disclosure among Chinese chemical enterprises presents the continuously increasing trend, But, as a whole, the enterprises have lower EA information disclosure level, and the large gap still exists between these enterprises in terms of EA information disclosure level. By the empirical analysis of multiple linear regression (MLR), it can be found that the enterprise size, Rate of Return on Common Stockholders’ Equity, debt level, state-owned holding nature of the enterprise have positive correlation with EC information disclosure level. The proportion of independent directors in enterprise and EC information disclosure have positive correlation, but without passing the test of significance, mainly because of the incomplete system of independent directors in China. 1. Introduction Nowadays, with the continuous economic development, the environmental pollution has been intensified, and more attention of the public has been paid to the relationship between the enterprise business activity and ecology environment. To make systematic and fully disclosure of EA information has become the important means for the enterprise to emphasize on the environmental protection (Indjejikian, 2007). Carry on the EA information disclosure can fully reflect the resource utilization and environmental pollution management of the enterprise (Rodrigue, 2014). Following the continuous development of scientific information, the enterprise stakeholder shall need comprehensive EA information for decision making; in order to improve its social image and realize sustainable development, it has become the trend for the enterprise to disclose the comprehensive and systematic EA information (Ana and Jesús, 2010). In the cyclic and low-carbon economic environment, by continuously promoting the enterprise level of EA information disclosure, it has positive effect on the sound and sustainable development of the whole economic society (Wang et al., 2014) About EA study, many scholars and experts at home and abroad have carried on lots of analysis, and developed lots of research results. The existing literatures are mainly focused on the different aspects of EA information disclosure in terms of disclosure content (Trueman, 1987; Lenter et al., 2003), disclosure method (Maines et al., 2003), evaluation method of disclosure quality (Arnold, 1998; Mauldin and Richtermeyer, 2004) and disclosure situation and influencing factor (Ro, 1980; Iatridis, 2008), mainly for normative study, but lack of the empirical study for EA information disclosure, esp. there was no study for one certain industry. Therefore, based on the chemical industry, this paper makes empirical analysis of the influencing factors on EA information disclosure, which has certain practical and innovative meaning. 2. Multiple linear regression (MLR) analysis method Regression analysis is the mathematical statistics method treating the statistical correlation of the variables, with the basic thought to find the mathematical expression form representing the relationship between independent variable and dependent variable (Wessel and Jurs, 1994). The regression analysis with two or more independent variables is called multiple regression (Bersten, 1998). DOI: 10.3303/CET1762266 Please cite this article as: Nian Yang, Zhen Li, Yuan Feng, 2017, Empirical study for influencing factors on environmental accounting information disclosure in chemical industry, Chemical Engineering Transactions, 62, 1591-1596 DOI:10.3303/CET1762266 1591 Given the linear regression model of random y and general variables X1, X2，…, Xk: Y = + + + ⋯ + (1) Where, y is dependent variable, Xi independent variable, βi regression parameter, and ε random error. The linear regression model aims to divide Y into two parts: certainty part and uncertainty part (Clouser and Jurs, 1996). In actual analysis, multi-observations were made to obtain n groups of sample data (yi; xi1，xi2, …, xip). The MLR model can be shown as: = + + + ⋯ += + + + ⋯ +⋯= + + + ⋯ + (2) Formula (2) can be simplified as: = + + + ⋯ + (3) In MLR analysis, it was necessary to test the fitting degree of the model, and also determine the significance level of every variable parameter in the regression equation (Lin and Wu, 1999). Normally the multi-coefficient of determination should be mainly adopted testing the fitting degree of MLR equation; refer to the formula as follows: = = 1 − = 1 − ∑( )∑( ) (4) Where, SSR means the regression sum of square; SSE the residual sum of square; SST the sum of square for total. The value range R2 of was [0, 1]; the lower R2 meant the less fitting degree of regression equation, while the higher R2 meant the more fitting degree (Shimada et al., 2000). Considering that R2 was easily influenced by the number of independent variables, it needed to be adjusted normally, specifically speaking, multiplying the SSE and SST by its degree of freedom respectively, to effectively reduce the effect of number of independent variables on fitting degree (And and Jurs, 1995). Refer to the formula as follows: = 1 − = 1 − (1 − ) (5) The formula (5) shows, the higher , the better. F is generally used to test the significance of multiple regression equation. The formula is as follows: F = //( ) (6) The higher F means that the dependent variable (DV) change caused by the independent variable (IV) change is greater than that of independent variables caused by random variable. Also, F-statistics can reflect the fitting degree of regression equation. Make certain changes to formula (4) and (6), to obtain: F = /( )/( ) (7) Formula (7) shows, with higher fitting degree, F-statistics shall be more significant; with more significant F- statistics, the fitting degree shall be higher (Bakker et al., 2004). 3. Empirical analysis for environmental accounting information disclosure 3.1 General situation of EA information disclosure in chemical industry The chemical industry mainly includes the enterprises for chemical production and development, which easily leads to a mass of “Three Wastes” (industrial wastewater, waste gases and residues), and pollutes the environment greatly, because of the heavy use of raw chemical materials. Until now, there have been 293 listed enterprises in chemical industry altogether, mainly composed of three major types of enterprises: petrochemical enterprise, basic chemical enterprise, and chemical fiber enterprise; refer to Figure 1 for the proportion of these three types. 1592 Figure 1: Classification of enterprises in chemical industry With the environmental requirements promoted continuously, the number of enterprises for EA information disclosure in chemical enterprise has been increasing. Refer to Figure 2 for details. Figure 2: Comparison chart of environmental accounting information disclosure carrier in chemical enterprises The EA information disclosure in chemical industry is mainly shown in text description and digital description, where the text description includes the overall situation of environmental protection, environmental risks, research and development project of environmental protection, and environmental achievements of the enterprise etc.; the digital description includes inputs, expenditure and research & development fee of the environmental protection in the enterprises. But now, as a whole, the enterprises have lower EA information disclosure level, and the large gap still exists between these enterprises in terms of EA information disclosure level. 3.2 Model construction By selecting the chemical fiber enterprises, resulting in more pollution in chemical industry, as the study samples in the time range 2013-2015 (92 enterprises in total), this paper makes empirical analysis of the influencing factors on the EA information disclosure, and construct the multiple regression model as follows: EDI = + SIZE + OE + EBT + SH + INDPR + (8) Where, EDI is the environmental accounting information index, as dependent variable; β1 and β9 mean regression coefficient, β0 constant term, and ε random error. See Table 1 for other independent variables. 1593 Table 1: Independent variables definition table Variable symbol Variable name SIZE Enterprise total assets scale ROE Net asset yield DEBT Asset liability ratio NSH The nature of the actual controlling shareholders RINDPR Proportion of independent directors The descriptive statistics is made for every variable in the model. See table 2 for detail. Table 2: Descriptive statistics of variables Variable N Minimum Maximum Mean Std.Deviation EDI 276 0.00000 0.60000 0.20945 0.14493 SIZE 276 18.73265 23.99843 20.67359 0.899627 ROE 276 -0.45572 0.49827 0.06994 0.127438 DEBT 276 0.17993 0.81125 0.48732 0.16843 NSH 276 0 1 0.75 0.436 RINDPR 276 0.34452 0.89975 0.54438 0.10341 3.3 Correlation analysis Considering that many variables are used in the model, and independent variables and dependent variables have positive correlation, the correlation test for the variables should be conducted before empirical analysis shown in Table 3. Generally, between the independent variables, the high correlation coefficient means there exists the multicollinearity, which shall influence the results of empirical analysis; the data in Table 3 shows that basically all correlation coefficients are rather small, therefore, the multicollinearity doesn’t exist in the independent variables of the model. Table 3: Correlation analysis of variables Insize Roe Debt Nsh Rindpr SIZE 1.0000 ROE 0.0652 1.0000 DEBT 0.5458 -0.1528 1.0000 NSH 0.2149 -0.0051 0.2664 1.0000 RINDPR -0.0347 -0.0163 0.0408 -0.2152 1.0000 3.4 Multiple regression analysis In this paper, the regression analysis was made for the model by SPSS17.0. See Table 4 for details. Table 4: Results of linear regression analysis Variable Coefficient Std.Error T-statistic Prob. β0 -0.75528 0.191123 -4.19427 0.0000 SIZE 0.08326 0.019263 4.39753 0.0000 ROE 0.18525 0.073256 2.40153 0.0169 DEBT 0.13983 0.045824 3.10034 0.0023 NSH 0.09284 0.016542 6.59984 0.0000 RINDPR 0.09834 0.142475 0.72384 0.4698 R-squared 0.49923 Mean dependent var 0.2695 Adjusted R-squared 0.48932 S.D. dependent var 0.14253 Log Likelihood 182.5823 F-statistic 29.7326 Dubin-Watson state 1.984245 Prob(F- statistic) 0.0000 In Table 4, at F=30.8208, well above F0.05 (5.324), and P=0, lower than 0.05, it means the significance of the model. But P value of the proportion of independent director (RINDPR) was 0.4698, greater than 0.05, indicating the failure in test of significance. Deducting the variable RINDPR, re-make regression analysis of the model to obtain Table 5. Table 5 shows R2=0.500428, and amended determination coefficient =0.496283, indicating that the independent variable has greater effect on the dependent variable. At DW=1.988652, very close to 2, it means 1594 that the residual errors are mutual independent, without auto-correlation. Above all, the enterprise size (SIZE) and EA information disclosure level have positive correlation, presenting high significance level at 1% (P=0.0000), because with larger enterprise size, it can more easily draw the attention of the society and the public, and the pressure by all parties for information disclosure shall be greater. Besides, in order to get the support of the government and public, the enterprise has been motivated to disclose more EA information to some extent. Table 5: Results of linear regression analysis Variable Coefficient Std.Error T-statistic Prob. β0 -0.724563 0.171634 -4.192553 0.0000 SIZE 0.086463 0.019623 4.392742 0.0000 ROE 0.193243 0.077345 2.400531 0.0169 DEBT 0.149882 0.047821 3.187492 0.0017 NSH 0.112673 0.019234 2.155628 0.0341 R-squared 0.500428 Mean dependent var 0.271963 Adjusted R-squared 0.496283 S.D. dependent var 0.133274 Log Likelihood 182.3754 F-statistic 37.92734 Dubin-Watson state 1.988652 Prob(F- statistic) 0.000000 Table 6: Multiple collinearity test statistics Dependent variable coefficient beta R-squared Tolerance VIF SIZE 0.08642 0.2854 0.302943 0.674338 1.448238 ROE 0.18548 0.1239 0.058372 0.958332 1.048723 DEBT 0.14376 0.1936 0.337259 0.649985 1.554294 NSH 0.10729 0.3221 0.167233 0.834472 1.198352 Based on the data in Table 5 and 6, the final regression equation is calculated: EDI = −0.724563 + 0.08642SIZE + 0.18548ROE + 0.14376DEBT + 0.10729NS (9) Rate of Return on Common Stockholders’ Equity (ROE) and EA information disclosure level have positive correlation, presenting high significance level at 5% (P=0.0169). The profitable enterprise, with more resources available, can make better use of resources to reduce the cost and gain the competitive advantage. The enterprise debt (DEBT) and EA information disclosure level have positive correlation, presenting high significance level at 1% (P=0.0017), which means the enterprise starts to focus on the investment and financing risks in the EA information disclosure process. Also, the creditors attach great importance to the EA information when financing the enterprise. State-owned holding nature (NSH) and EA information disclosure level have positive correlation, presenting high significance level at 1% (P=0.0341). Compared with the private holding enterprises, the state-owned holding enterprises have stronger awareness of social responsibility, being more willing to disclose high-level EA information to the society. The proportion of independent directors (RINDPR) and EA information disclosure level have positive correlation, but without passing the test of significance (P=0.4698), mainly because of the incomplete system of independent director: most directors, nominated by the Chairman of Board and Board of supervisors, and then elected by general meeting of stakeholders, haven’t played their real roles. 4. Conclusion With the environmental requirements promoted continuously, the number of enterprises for EA information disclosure in chemical enterprise has been increasing. But now, as a whole, the enterprises have lower EA information disclosure level, and the large gap still exists between these enterprises in terms of EA information disclosure level. By the empirical analysis of multiple linear regression (MLR), it can be found that the enterprise size (SIZE), Rate of Return on Common Stockholders’ Equity (ROE), debt level (DEBT), state- owned holding nature (NSH) are the key influencing factors on enterprise EA information disclosure, and they have positive correlation with EC information disclosure. The proportion of independent directors (RINDPR) in enterprise and EC information disclosure have positive correlation, but without passing the test of significance, mainly because of the incomplete system of independent directors in China. 1595 References And D.L.C., Jurs P.C., 1995, Simulation of the 13c nuclear magnetic resonance spectra of ribonucleosides using multiple linear regression analysis and neural networks, Carbohydrate Research, 271(1), 65-77, DOI:10.1016/0008-6215(95)00051-t Arnold A.J., 1998, Uk accounting disclosure practices and information asymmetry during the first quarter of the twentieth century: the effects on book returns and dividend cover, Journal of Business Finance & Accounting, 25(7‐8), 775-794, DOI: 10.1111/1468-5957.00212 Bakker D.P., Busscher H.J., Van Z.J., De V.J., Klijnstra J. W., Hc V. D. M., 2004, Multiple linear regression analysis of bacterial deposition to polyurethane coatings after conditioning film formation in the marine environment, Microbiology, 150(6), 1779-1784, DOI: 10.1099/mic.0.26983-0 Bersten A.D., 1998, Measurement of overinflation by multiple linear regression analysis in patients with acute lung injury, European Respiratory Journal, 12(3), 526-532, DOI: 10.1183/09031936.98.12030526 Clouser D.L., Jurs P.C., 1996, The simulation of 13c nuclear magnetic resonance spectra of dibenzofurans using multiple linear regression analysis and neural networks, Analytica Chimica Acta, 321(2), 127-135, DOI: 10.1016/0003-2670(95)00581-1 García A.C., García-García J., 2010, Determinants of online reporting of accounting information by spanish local government authorities, Local Government Studies, 36(5), 679-695, DOI: 10.1080/03003930.2010.506980 Iatridis G., 2008, Accounting disclosure and firms' financial attributes: evidence from the UK stock market, International Review of Financial Analysis, 17(2), 219-241, DOI: 10.1016/j.irfa.2006.05.003 Indjejikian R.J., 2007, Discussion of accounting information, disclosure, and the cost of capital, Journal of Accounting Research, 45(2), 421-426, DOI: 10.1111/j.1475-679x.2007.00239.x Lenter D., Slemrod J., Shackelford D., 2003, Public disclosure of corporate tax return information: accounting, economics, and legal perspectives, National Tax Journal, 56(4), 803-830, DOI:10.17310/ntj.2003.4.06 Lin Z.C., Wu W.J., 1999, Multiple linear regression analysis of the overlay accuracy model, IEEE Transactions on Semiconductor Manufacturing, 12(2), 229-237, DOI: 10.1109/66.762881 Maines L.A., Bartov E., Fairfield P.M., Hirst D. E., Iannaconi T. A., Mallett R., 2003, Implications of accounting research for the fasb's initiatives on disclosure of information about intangible assets, Accounting Horizons, 17(2), 175-185, DOI: 10.2308/acch.2003.17.2.175 Mauldin E.G., Richtermeyer S.B., 2004, An analysis of erp annual report disclosures, International Journal of Accounting Information Systems, 5(4), 395-416, DOI: 10.1016/j.accinf.2004.04.005 Ro B.T., 1980, The adjustment of security returns to the disclosure of replacement cost accounting information, Journal of Accounting & Economics, 2(2), 159-189, DOI: 10.1016/0165-4101(80)90010-5 Rodrigue M., 2014, Contrasting realities: corporate environmental disclosure and stakeholder-released information, Accounting Auditing & Accountability Journal, 27(1), DOI: 10.1108/aaaj-04-2013-1305 Shimada M., Masuda Y., Yamada Y., Itoh M., Takahashi M., Yatagai T., 2000, Explanation of human skin color by multiple linear regression analysis based on the modified lambert-beer law, Optical Review, 7(4), 348-352, DOI: 10.1007/s10043-000-0348-7 Trueman B., 1987, Discussion of operating decisions and the disclosure of management accounting information, Journal of Accounting Research, 25, 190-193, DOI:10.2307/2491086 Wang P.C., Che F., Fan S.S., Gu C., 2014, Ownership governance, institutional pressures and circular economy accounting information disclosure, Chinese Management Studies, 8(3), 487-501, DOI: 10.1108/cms-10-2013-0192 Wessel M.D., Jurs P.C., 1994, Prediction of reduced ion mobility constants from structural information using multiple linear regression analysis and computational neural networks, Analytical Chemistry, 66(15), 2480- 2487, DOI: 10.1021/ac00087a012 1596