Moderating Effect of Earnings Management in the Relationship between Sustainability Reporting Initiatives and Value Relevance

The purpose of this study is to investigate whether sustainability disclosures are associated with value relevance in Bangladesh. The moderating effect of earnings management (EM) is also examined to observe the right direction in this relationship. Based on prior studies on sustainability disclosure and global reporting initiatives guidelines, this research uses the content analysis approach to assess the magnitude of sustainability initiatives of 30 Bangladeshi banking companies over the period 2009–2017. The Ohlson price model and discretionary accruals are also employed as measures of value relevant of sustainability disclosure and EM, respectively. The findings state that sustainability reports positively affect the equity value, whereas EM negatively moderates the direction of this association. The results also confirm that management should be responsive of the impending capital market effects of voluntary disclosures regarding sustainability issues. These findings could have several implications for banks, investors, and policymakers.


INTRODUCTION
The concept of value relevance is not new in the area of research. If the accounting value has anticipated association with the market value of equity, then it is defined as value relevant, and most presentations of this are based on accounting variables (Byun & Oh, 2017). However, in general, previous studies stated that accounting information only is not enough to elucidate an organisation's market value of equity and its deviations (Shan, 2015). Subsequently, many academics showed interest in investigating the value relevance of non-financial information to close the rising gap between the book value and market value of corporate stocks (Mostafa, 2017;Rahman et al., 2020). Besides, the worldwide movement for socially responsible investment reveals that the approach organizations integrate social, environmental, and economic concerns into their stakeholders, such as employees and shareholders that support the long-term perspective hypothesis. Thus, socially responsible firms also act in an ethical manner when disclosing their financial information. From this point of view, empirical findings reveal that organisations that are more dedicated to CSR reports or manners deliver more general monetary disclosures and are less involved in earnings manipulation (Gras-Gil et al., 2016). On the other hand, prior Bangladeshi studies document that firms that provide more CSR disclosures overstate their earnings through income increasing discretionary accruals. Furthermore, prior studies also advocate that to camouflage professional opportunistic motives (under opportunism hypothesis) managers may use CSR as a strategic weapon. Managers may escape inspection from stakeholders through CSR undertakings that protect their job. Moreover, a healthy relationship with stakeholders can be used as a defensive tool against aggressive buyouts. Hence, managers who are involved in earnings manipulations may be motivated by extensive CSR activities to shed their entrenchment mechanisms. Thus, empirical results concerning the positive or negative stimulus of CSRR on EM remain indecisive. By reason of this indecisive relationship between CSRR and EM, EM was used as a moderating variable rather than mediating variable to find out the clear direction between the relationship of CSRR and VR.
Multiple sources inspired this study's decision to center its examination in the context of Bangladeshi banking sector. First, banks make superior CSR commitments every year (Ullah & Rahman, 2015). Prior studies indicated that banks perform excellent sustainability disclosure compared to non-financial organisations, and it is also apparent that market participants value non-financial information. However, a few authors contend that financial firms (banks) are more inclined to manipulate their earnings than non-financial-related associations (Chahine et al., 2019;Grougiou et al., 2014). Nevertheless, this study investigated an emerging economy, Bangladesh, where the poor enforcement of corporate regulations and monitoring system, lack of established corporate governance issues and capital market is less developed . On the other hand, prior studies document that Bangladesh is still lagging behind with regard to CSR disclosures Sobhani et al., 2012). However, the findings of the previous Bangladeshi studies indicated that banking sector disclose more CSR information than non-financial organizations (Ullah & Rahman, 2015). Therefore, it is crucial to investigate whether non-financial information, specifically CSR information, that is disclosed by companies in their annual report or other standalone reports (such as sustainability or CSR reports) is value relevant or not of the Bangladeshi listed banking sector. Besides, to the best of the author's knowledge, no researchers have explored the relationship between CSRR, EM, and VR in the overall banking sector, both conventional and Islamic. Thus, it is essential to investigate the opportunistic motives of bank managers that ultimately impact on share price. This study also aimed to mitigate this research gap. Thus, the following research questions are addressed: RQ1: What is the relationship between sustainability reporting and value relevance in the banking sector of Bangladesh? RQ2: Does EM moderate the relationship between sustainability reporting and value relevance? disclosures in a separate section, standalone CSR reports, executive reports, corporate governance revelations, CEO statements as well as notes to the financial statements confined in annual reports. Annual reports will also use to gather financial data related to measuring EM, and stock price-related information will be collected from the Dhaka Stock Exchange (DSE) website and the annual reports. The sample consists of the following financial sectors: Conventional Banks (23), Islamic Banks (7). This study uses content analysis to assess the extent of sustainability reporting. Table 1 shows the final sample size of this sector, which selected as fact-finding of this research. The list of the banks was found from the DSE website (www.dsebd.org) in December 2018.

Empirical Models and Measurement of the Variables
In line with previous studies (de Klerk & de Villiers, 2012;Lourenço et al., 2014;Reverte, 2014;Verbeeten et al., 2016), this thesis assesses the previous Ohlson (1995) valuation model that operationalized the concept of value relevance. Definitions of variables are given in Table 2.
Besides, this research applies the indicator of EM intrinsically, to examine the existence of EM practice in the listed banking sector of Bangladesh. The second objective of this study was to assess the moderating effects of EM on the relationship between value relevance (as dependent variable) and sustainability reporting (as an independent variable) which pertained to hypothesis H 2 .

Measurement of the Independent Variable: Construction of Sustainability Reporting Index
The SR Index is developed to measure the extent of SR disclosure via the annual reports by banking companies listed in the DSE in Bangladesh. An extensive review of the past literature on SR disclosure indices conducted to identify potential items of information. Besides the reporting guidelines provided by the Global Reporting Initiatives (GRI, 2015) regarding sustainability reporting and the financial sector, specific disclosure is also adopted to construct the revelation index. Furthermore, the following five steps are taken into consideration when developing the index.
Step 1: Content Analysis Based on prior research studies, this study uses content analysis to measure the magnitude of SR disclosures. Content analysis is a technique that converts written text into numerical code (Ullah & Rahman, 2015) and creates several groups based on designated criteria. In line with the previous studies, the annual report's content analysis will be used to gather the data related to this study. The content analysis method has been utilized to analyze the narrative disclosures in annual reports, sustainability reports, standalone reports, and other reports . A key element of content analysis research design is the document to be analyzed . The Appendix shows the aspects of content analysis.
Step 2: Coding of Data and Selecting Index Calculation Approach Based on the review of the extant literature, there are two approaches for index construction; a weighted index approach and an un-weighted index approach. Weighted index methodologies have been criticized for the absence of a precise measure for weights and lack of solid theoretical basis for assigning weights . To avoid the disadvantages and complexity of weighting, this study adopts an un-weighted reporting index procedure. Moreover, the unweighted index is a suitable research instrument in disclosure studies when the research is aimed at all relevant stakeholders of business reports instead of a specific user group.
Step 3: Calculation of Index Score Based on this approach, each of the three components is assigned a maximum raw score and a minimum raw score based on disclosure or not. Zero is assigned if the company does not disclose that information, whereas a rating of 1 is assigned if the company reveals that particular information. This method is similar to other researchers' binary method Ullah & Rahman, 2015).
Step 4: Measuring the SR Index The total score of the SR Index is calculated as the ratio of the actual score awarded to the company to the maximum possible score for that company (based on the number of applicable SR Index items). This issue has been addressed by previous corporate SR reporting studies using similar measurements Rahman et al., 2020). Hence, the overall disclosure index that represents the total SR score is calculated as follows: SR Index Score ðSRISÞ ¼ P actual score awarded to the company P maximum possible score for the company Step 5: Testing of Internal Consistency Reliability This study will apply the Cronbach's coefficient alpha (Cronbach, 1951) to measure the SR index's internal consistency and reliability. The coefficient alpha for the three board classifications of the disclosure index is more than 0.70. This measurement delivers good support in the condition where the set of selected items in the disclosure index captures the same fundamental construct . Following Rahman et al. (2020) and Muttakin et al. (2015), the sustainability reporting (SR) index is calculated in Equation 2 as follows: where d i = 1 if the item di is reported; d i = 0 if the item is not reported; and n is the number of items.
This SR index is shown in the Appendix.

Measurement of the Dependent Variable: EM
In this study, the measure of EM is the scale of discretionary accruals assessed from the Jones (1991) model as adapted by Yasuda et al. (2004) for banking companies. This study will run the subsequent regression in an attempt to get the discretionary part of the bank's total accruals: where DACC t is the total discretionary accruals assessed as the variance between net incomes and operating cash flows; TA t−1 is the lag of total assets; ΔOI t is the variation in bank's income from operation between t − 1 to t; BRE t is the bank's premises and equipment; and ε t is an error term. The correlation coefficients among the regressors are represented in Table 4. It can be observed that the correlation between SR and EM is (correlation = 0.643). EPS and BV are (correlation = −0.046), and variables are statistically significant at 5% level, though, none of the variance inflation factors (VIFs)-not reported-exceed the critical value of 10 (Hair et al., 2010). Therefore, in this study, it can be said that multicollinearity is not a severe problem.

RESULTS AND DISCUSSION
The results suggest that sustainability information provides more relevant information to shareholders and what only financial information offers, which is supported H1. These outcomes are consistent with those found by Verbeeten et al. (2016) in the German perspective and de Klerk and de Villiers (2012) for South African corporations. Overall, these outcomes appear to sustain the view that non-financial information in terms of sustainability information is value relevant.
Moreover, the results from the estimation of models (1)−(2) are presented in Table 5. As expected, the coefficients for EPS, BV, and SR are positively and significantly associated with the share price. The regression  Table 2. t statistics for the regression coefficients are stated within parentheses. *, ** , *** Significance at the 10%, 5%, and 1% levels, respectively 272 Rahman et al. findings indicate that the main independent variable sustainability reporting (SR) has a significant and positive impact on share value (PRICE) across all three models. Thus, higher CSRR indicates a higher share price, suggesting that additional CSR information enhances the market value of equity. Overall, the study results support research Hypothesis (H1). The adjusted R 2 , is 0.827 (see Table 5); thus, it can be said that EPS, BV, and SR explain 82.7% of the variance of the Bangladeshi Banking companies' equity values. These results are similar to the finding of the recent research studies for Germany and Spanish samples, respectively, by Verbeeten et al. (2016) and Reverte (2014).
However, EM is negatively associated with the share prices. The interaction between SR and EM in model 2 is negatively and significantly associated with the equity value. These explain that SR has an increment value in share prices, but when organizations manipulate earnings, that negatively impacts the share price, which supports H2. Thus, manager manipulation of earnings through discretionary accruals has a negative effect on stock price in the Bangladeshi banking sector. The adjusted R 2 of models 1 and 2 are 0.827 and 0.823, respectively, and these indicate that models explain 82.7% and 82.3% of the variance of the Bangladeshi Banking company's equity values.
For the nexus between SR and VR, multiple regression analysis revealed that SR disclosure has a positive and significant relationship with stock prices. This finding supports the research Hypothesis (H1). These findings are similar to Fu et al. (2019), who suggested that SR information delivers the social and environmental position of an organization to indicate apparent litigation risk, future economic and ecological obligations, and adverse information asymmetries. Thus, firms that disclose higher SR activities are likely to have a higher market value of equity paralleled to other concerns with inferior intensities in SR reporting in the banking sector of Bangladesh. Besides, the inference of the Stakeholder-Legitimacy theory is that SR has a positive stimulus on the market value of the firm, and it may be a managerial means that supports the efficient utilization of the resources and the findings of this study are in line with this theoretical background.
Using the Ohlson price model, there was a negative effect on value relevance for companies engaged in EM when SR practices are used to conceal their poor earnings. The adverse moderating impact of EM on the relationship between SR and VR support Hypothesis (H2) was robust in multiple regression estimators. Besides, these results of the study reapprove the claims of previous studies those had reported a negative relationship between EM and value relevance. The findings of the current study also confirm the importance of earnings quality and its positive impact on equity's market value. In short, this study has made an original and novel contribution through its models.
In this study, we have executed numerous robustness tests to confirm the outcomes are robust to different measurements. Foremost, we change share prices corresponding to 3 months after the AGM, which is conducted after the year-end and expected to publish all relevant reports. This study did not find any significant changes in the relationship between SR-VR or moderating effects from EM (discretionary accruals proxy) on this relationship. These outcomes support that the main results in Table 5 were not dissimilar with the price proxy of value relevance. The results are shown in Table 6. Subsequently, this study employed lagged regression as an additional test for the price models (1-2) to ensure that the results were not affected by endogeneity. The lagged regression results shown in Table 7 are consistent with the un-lagged pooled OLS regression and panel fixed effect estimations, indicating that the results did not suffer from causality relationships. Moreover, this study also carried out a Durbin−Wu−Hausman endogeneity test (Table 7) for SR and value relevance with the null hypothesis being accepted, indicating that there were no endogeneity problems.

CONCLUSION
For many years, the cornerstone of corporate reports has been presenting the financial information in the annual reports to give a real picture of companies' financial performance. However, recent studies indicate that financial information is not well enough to explain the market value of equity. Still, non-financial information in terms of sustainability revelation also has a significant impact on value relevance. But, intense debates about whether such disclosure is useful for stakeholders and the empirical findings are inconclusive. Therefore, to close this research gap, this study examines the observed relationships among SR, EM, and VR jointly because previous literature has looked at pair-wise relations such as SR-EM, SR-VR, and EM-VR separately. Thus, it is crucial to examine these three variables, such as SR, EM, and VR, jointly. Because the organizations may use extensive SR disclosure to disguise EM practices, that may negatively impact the long-term organization's value maximization, which is the market value of a share. The outcome implies that the disclosure of sustainability revelation affects the market's aptitude to predict impending earnings variations (Byun & Oh, 2017), and EM negatively moderated the relationship.
This study provides several policy implications. First, the outcomes of this study will be of specific interest to the shareholders of an organization who may hope to control their managers' efficiency. This research will  Table 2. The numerical figures in parentheses are t-values. *, ** , and *** indicate significance at the 10%, 5%, and 1% levels, respectively also be of concern to financiers and other market participants who may wish to measure the adverse influence of EM to uphold the reputation of their business. Second, despite the enormous benefits of voluntary SR activities, it is also essential to set common standards for reporting such initiatives. One of the best solutions is to follow the guidelines in line with the globally accepted model, such as GRI guidelines. Finally, one of the significant problems to measure the performance of SR disclosure is the lack of standards and different regulations. Therefore, regulatory authorities and standard setters, due to the growing demand for the credibility of the non-financial information to the investors, should encourage firms to seek external assurance of the SR reports to intensifying their integrity and consistency.
The study has some limitations that could be considered as avenues for future research. First, this study concentrates only on stock price as the value relevance measurement method. Focusing on other firm value methods suggested by prior value pertinence studies would be motivating (Byun & Oh, 2017;. Moreover, these results cannot be generalized; meanwhile, the sample is both industry-and country-specific. Hence, future research could examine the link between SR-EM using cross-country data. The probable effects of different ethnic, institutional, legal, and accounting standards could be adequately controlled. Furthermore, future research may also wish to investigate the relationship between subsamples of SR using non-binary measures of SR in an international setting.