The Influence of Sustainability Disclosure on Financial Performance: A Study of Indonesian Firms

: This study examines the correlation between the disclosure of sustainability measures and the financial success of companies in Indonesia. The increasing importance of sustainability disclosure, which includes environmental, social, and governance factors, for firms to demonstrate their dedication to sustainable practices, has generated significant debate on its influence on financial results. This study investigates the impact of sustainability disclosures on the financial performance of companies in Indonesia, thus adding to the existing body of knowledge on this topic. The study utilizes a mixed-method approach, incorporating qualitative content analysis of data extracted from annual reports, as well as quantitative analysis derived from financial statements of publicly traded corporations. The sample consists of companies from three major industry sectors, each demonstrating different levels of quality in disclosing their sustainability practices. Accounting-based indicators like return on assets (ROA) and return on equity (ROE) are used to evaluate financial performance. The findings demonstrate a direct and favorable relationship between the caliber of sustainability disclosures and financial performance, specifically in sectors that are highly responsive to environmental concerns. Companies that have more comprehensive and transparent sustainability reporting processes in these industries generally achieve better performance compared to those with less comprehensive reporting. These conclusions have substantial ramifications for firms, investors, and policymakers. Enhancing sustainability disclosure can enhance a company's financial performance and act as a significant factor for investment choices, providing information about a company's dedication to sustainability and related risks. Policymakers can utilize these observations to support the implementation of improved sustainability reporting regulations, thereby fostering sustainable economic growth in Indonesia. Ultimately, the research confirms that Indonesian companies who provide detailed and reliable information on their sustainability efforts have a positive correlation with their financial performance. This emphasizes the significance of improving these practices to achieve both economic prosperity and sustainable development objectives.


INTRODUCTION
In contemporary corporate dynamics, sustainability has emerged as a principal consideration, embodying an enterprise's commitment to sustainable economic progression.This commitment, as defined by the World Business Council for Sustainable Development, requires collaborative efforts to improve the living standards of employees, their families, and the wider community.The shift towards sustainability is evident in corporate mission statements, which have evolved from focusing solely on immediate financial gains to a broader perspective of societal value creation.This evolution has redefined corporations as key contributors to societal welfare and environmental protection, going beyond their traditional fiduciary roles.The paradigm shift towards sustainability has elevated the importance of Sustainability Reporting.According to the Global Reporting Initiative, this type of reporting involves a systematic process of evaluating, disclosing, and being accountable for sustainable development contributions.The widespread adoption of sustainability reporting is evident with a substantial percentage of the top global enterprises integrating corporate responsibility reporting into their operations.This global trend is mirrored regionally, with many corporations in the Asia-Pacific and Europe actively engaging in sustainability reporting, marking a significant shift in corporate transparency and accountability.
The introduction of Environmental, Social, and Governance (ESG) metrics marks a significant development in measuring sustainability, blurring the lines between short-term benefits and long-term corporate value.These metrics have become essential in assessing firms' stewardship in environmental and societal domains, correlating with enhanced returns to stakeholders.However, traditional financial ledgers, while crucial, have been shown to be insufficient in capturing a company's comprehensive health, as  The scoring system outlined below represents an adaptation from the methodology developed by (Gunawan, 2021) focusing on qualitative measurement.This system is employed to evaluate the level and depth of corporate disclosures in support of the Sustainable Development Goals (SDGs).It distinguishes the extent to which a company communicates its SDG-related initiatives, ranging from basic narrative description to comprehensive disclosure including quantitative data, goals, and strategies.The assessment employed a scoring method developed based on understanding the significance of strategic information, targets, and achievements (Gunawan, 2007).The following is the description of the scoring system for companies' disclosure in support of the Sustainable Development Goals (SDGs): a) A score of "0" is given if the company does not disclose any support for the SDGs.b) A score of "1" is awarded if the company discloses its SDG support solely in narrative form.c) A score of "2" is assigned when the company presents its SDG support narratively along with its achievements.d) A score of "3" is given if the company's narrative disclosure of SDG support narratively and includes targets.e) A score of "4" is designated when the company's narrative disclosure of SDG support is accompanied by strategies.f) A score of "5" is given if the company provides a narrative of SDG support, including both achievements and targets.g) A score of "6" is awarded if the company's narrative disclosure of SDG support encompasses both achievements and strategies.h) A score of "7" is provided when the company's narrative disclosure of SDG support is paired with both targets and strategies.i) Finally, a score of "8" is given when the company narratively discloses its SDG support inclusive of achievements, targets, and strategies.
According to (Hassan, et al., 2018), "descriptive statistics are used to explain, provide an overview of the characteristics of a series of data without drawing general conclusions".Descriptive Statistical Analysis is also expected to provide explanations that can be easily understood by data users regarding the problem being analyzed (Cai, et al., 2020).In this research, descriptive statistical analysis will help convey the amount of disclosure of each company's targets, strategies, and achievements to describe the results of calculating scores.
The targeted population for this study encompasses the financial reports (annual reports) of food & beverages, coal & oil refining, and chemical & pharmaceutical industries listed on the Indonesia Stock Exchange from 2018-2022.The study utilizes a nonprobability sampling technique, specifically the purposive sampling method.This involves selecting samples based on specific considerations or criteria ensuring their relevance for the research.

Panel Data Regression
Panel data regression analysis stands as a robust statistical method that merges the intricacies of cross-sectional data (data across multiple subjects at a single point in time) with those of time-series data (data across multiple time periods for the same subject), offering a multifaceted perspective that captures both individual-specific variations and temporal changes.The quintessence of this approach lies in its ability to accommodate variability and control for potential heterogeneity inherent in the subjects, which singledimensional analyses might overlook.
Indicators pivotal to this analysis include time-invariant individual characteristics, which are controlled for in fixed effects models by allowing each entity to have its own intercept.This technique accounts for any unobserved differences among subjects that do not change over time.Alternatively, random effects models, which assume that these individual-specific effects are random and uncorrelated with the explanatory variables, are utilized when such an assumption is plausible, thus providing a more efficient estimation under the right conditions.Hybrid models, however, can be employed to harness the strengths of both fixed and random effects approaches, mitigating the limitations inherent in each.
The indicators or variables of interest in panel data regression typically involve those that capture the evolution of phenomena over time and across different entities.These could range from economic indicators like GDP growth, investment levels, and employment rates to social indicators such as educational attainment or health outcomes.By leveraging panel data, researchers gain insights into the dynamics of change, allowing them to make more informed conclusions about causality and the impacts of policy changes or market shifts, which single time point (cross-sectional) or single subject (time-series) data would fail to reveal adequately.
The research employs the panel regression test method, as described by (Pesaran, 2021) in (Hassan, et al., 2018), which integrates cross-sectional and time series data types.This approach offers several advantages over standard cross-sectional and time series methods: a) The combination of cross-sectional and time series data in panel data results in more informative and diverse data, reduced collinearity between variables, increased degrees of freedom, and enhanced efficiency.b) Analyzing cross-sectional and time series data over multiple periods makes panel data suitable for investigating dynamic changes.c) Panel data has the capability to identify and measure negative influences that may not be apparent in pure cross-sectional or pure time series data.d) Panel data allows for the examination of more intricate models of behavior, such as economies of scale and technological changes, offering a better understanding compared to pure cross-sectional or time series data.e) Since panel data encompasses individuals, companies, cities, countries, etc., over time, it inherently involves heterogeneity within these units.Techniques for estimating panel data can explicitly incorporate this heterogeneity for each specific individual variable.
Following the hypothesis that has been developed, this study proposed two mathematical model:  Direct relationship between financial performance (ROE, ROA) and the quality of ESG reporting: The proxy for financial performance variables can be divided into two distinct approaches, each reflecting different dimensions of financial performance.The first approach involves profitability and is reflected in two key ratios: Return on Equity (ROE), which measures a company's efficiency in generating profits for shareholders, and secondly Return on Assets (ROA), indicating the company's ability to leverage its assets to generate earnings.Thus, these approaches offer a holistic view of the financial health of an entity through different lenses, encompassing operational efficiency, asset utilization, and investor perceptions of the company's value in the market.
The panel regression test is applied in this study to assess the impact of sustainability reporting quality on financial performance across three sectors: the food & beverage industry, the coal & oil refinery industry, and the pharmaceutical industry.Based on the Table 0-3 of regression results fromSDG on financial performance in the food and beverage sector, it shows that there is not a influence of theSDG variables on ROE.The numbers in parentheses are the t-statistics, which are used to determine the statistical significance of the coefficients.The t-statistics for SDG are 1.81 for ROE and 0.92 for ROA.Normally, a t-statistic greater than 2 (or less than -2) is considered statistically significant at the 5% level, indicating that the relationship between the variable and the outcome is unlikely to be due to chance.Here, neither t-statistic reaches that threshold, suggesting that the SDG variable is not significantly related to either ROE or ROA at the 5% level.The following are the results of panel data regression analysis in selecting the best model in the Food & Beverage sector.

Analysis for Food & Beverage Subsector
Based on the outcomes of both the Chow and Hausman tests, the most suitable model for the data is the random effects model.The Chow test indicated that there are significant differences across groups, which would typically suggest the use of a fixed effects model.However, the Hausman test, which more directly compares the fixed and random effects models, did not find a systematic difference between the two models' coefficients.The p-value of the Hausman test was not below the conventional threshold of 0.05, suggesting that the unique errors are not correlated with the regressors, which validates the use of the random effects model.

a. Hypothesis Testing (the influence of Sustainability Disclosure on ROE) Table 0-4 REM Regression Result for RQ#1 Food & Beverage Subsector
The output provided shows the results of a random-effects GLS regression with robust standard errors.The coefficient forSDG (x1) is 0.001568.This suggests that there is a positive but small association betweenSDG and ROE; for every one-unit increase inSDG, ROE increases by 0.001568 units.The constant (_cons) is 0.277022, which is the expected value of ROE whenSDG is zero, accounting for the random effects.The robust standard error forSDG (x1) is 0.008683, which is larger than the standard error from the non-robust model.This adjustment for robustness often leads to larger standard errors if heteroskedasticity is present.Although the model includes corrections for heteroskedasticity within clusters, it does not sufficiently explain the variation in ROE, as indicated by the low R-squared values.The marginal significance of the Wald test suggests that while the model might have some explanatory power, it is not strong, and additional variables or model modifications may be necessary for a more accurate analysis.The Wald chi-squared statistic is 3.The between R-squared is 0.0657, which is somewhat higher, suggesting that the model explains some of the variation in ROE between groups.The overall R-squared is 0.0109, which is still very low, indicating that the model, as a whole, explains very little of the variance in ROE.

b. Hypothesis Testing (the influence of Sustainability Disclosure on ROA) Table 0-5 REM Regression Result for RQ#1 Food & Beverage Subsector
The output from the random-effects Generalized Least Squares (GLS) regression analysis provided indicates that the model is attempting to estimate the effect of SDG (x1) on ROA (y3), with 'kode' as the group variable for the panel data set.Coefficient Analysis: The coefficient for SDG (x1) is estimated to be 0.002144, which implies that for every one-unit increase in SDG, there is an expected increase of approximately 0.002144 units in ROA.However, the associated p-value of 0.404 indicates that this relationship is not statistically significant at the conventional alpha level of 0.05.This means we cannot confidently assert that changes in SDG have a predictable effect on ROA based on this dataset.Significance of the Constant: The constant (intercept) of the model is significant (p < 0.001), with an estimated value of 0.0611586.This suggests that if SDG were zero, the average ROA would be approximately 6.12%.The significance of the constant term indicates that factors not included in the model may have a baseline effect on ROA.R-squared Values: The overall R-squared is 0.0252, which means that overall, the model explains only about 2.52% of the variance in ROA across all observations.Wald Chi-Squared Test: The Wald chi-squared statistic is low (0.70), with a p-value of 0.4040, signaling that SDG is not a significant predictor of ROA in the context of this random-effects model.
In conclusion, the random-effects model does not provide strong evidence to suggest that SDG is a significant predictor of ROA.While there is some variation in ROA between groups, the model's overall explanatory power is quite limited, indicating that other variables not included in the model might better account for the observed variation in ROA.Additionally, the non-significant pvalue for SDG and the low R-squared values suggest that further research is needed, possibly including more variables, or exploring different model specifications, to better understand the determinants of ROA.The table above is presenting the outcomes of a regression analysis that examines the impact of ESG (Environmental, Social, Governance) reporting and company size on financial performance, specifically on Return on Equity (ROE) and Return on Assets (ROA).The coefficient for SDG is 0.001 for ROE and 0.0001 for ROA.These are very small effect sizes and the t-statistics (0.63 for ROE and 0.42 for ROA) indicate these are not statistically significant.The Size coefficient is 0.015 for ROE and 0.008 for ROA.For ROA, the size coefficient is significant, and the t-statistic is 2.81.The main takeaway from this table is that while SDG does not have a statistically significant impact on financial performance in terms of ROE and ROA.

a. Hypothesis Testing (the influence of Sustainability Disclosure on ROE moderated by size) Table 0-7 REM Regression Result for RQ#3 (SDG and ROE moderated by Firm Size) for Food & Beverage Subsector
The provided output details a random-effects GLS regression analysis with Return on Equity (ROE, y2) as the dependent variable, and Sustainable Development Goals alignment (SDG, x1) and size (x3) as the independent variables.The data is structured as panel data with 'kode' representing the panel groups.Coefficients and Significance: The coefficient for SDG (x1) is 0.001303 with a standard error of 0.0020626.The p-value is 0.528, which indicates that the effect of SDG on ROE is not statistically significant at conventional levels.The coefficient for size (x3) is 0.0152861 with a standard error of 0.031678.The p-value is 0.629, also indicating a lack of statistical significance.The constant (intercept) is -0.0774181 with a p-value of 0.741, suggesting that when SDG and size are at zero, the average ROE would be negative, but this result is not statistically significant.Wald Chi-Squared Test: The Wald chisquared statistic is 0.83 with a p-value of 0.6597, indicating that the independent variables as a group are not statistically significant in explaining the variability in ROE.

b. Hypothesis Testing (the influence of Sustainability Disclosure on ROA moderated by size) Table 0-8 REM Regression Result for RQ#3 (SDG and ROA moderated by Firm Size) for Food & Beverage Subsector
The statement discusses a scenario in which the random effects model initially did not show significance in the estimated coefficients.As a response to this, robust standard errors were used to address the issue.When robust standard errors are employed, it typically means that the estimation has been adjusted to account for potential heteroskedasticity, which is a problem where the variance of the error terms is not constant across observations.Heteroskedasticity can lead to inefficiencies in the ordinary least squares (OLS) estimates and can result in standard errors that are biased, leading to unreliable hypothesis tests.By using robust standard errors, the estimation becomes more reliable as it corrects for this inconsistency in variance, and hence, it improves the robustness of the model against violations of the homoscedasticity assumption.This adjustment is crucial, especially in panel data analysis, as it can lead to more accurate inferences about the significance of the model's coefficients.The information provided is from the output of a random-effects GLS regression analysis where the dependent variable is ROA (y3), and there are two independent variables: SDG (x1) and size (x3).This model is accounting for random effects across different groups represented by 'kode'.Coefficients: The coefficient for SDG (x1) is 0.0001043, with a robust standard error of 0.0002462.The coefficient is not statistically significant, as indicated by a p-value of 0.672.This suggests that, after accounting for group random effects and using robust standard errors, there is no clear evidence of an impact of SDG on ROA.The coefficient for size (x3) is significantly different from zero (p = 0.005) with a value of 0.0089422, suggesting that size has a positive impact on ROA.For every one-unit increase in size, ROA increases by 0.0089422 units, all else being equal.The R-squared values are as follows: within = 0.0110, between = 0.0866, overall = 0.0526.These values indicate that the model explains 1.10% of the variance within groups, 8.66% of the variance between groups, and 5.26% of the total variance in ROA.These are relatively low values, suggesting that the model has limited explanatory power.Wald Chi-Squared Test: The Wald chi-squared test statistic is 8.58 with a p-value of 0.0137, indicating that the independent variables, when considered together, do have a statistically significant relationship with ROA at the 5% significance level.

Conclusion:
This random-effects model with robust standard errors suggests that while SDG does not have a significant impact on ROA, the size of the entity (or some proxy for size) does.The model's overall explanatory power is relatively low, but there is a significant portion of variance attributed to differences between groups.

Result Summary on Food and Beverage Sector
In this study of the Food & Beverage (F&B) sector, the impact of SDG on financial performance indicators such as Return on Equity (ROE) and Return on Assets (ROA) was examined.The findings revealed a minimal and statistically insignificant impact of SDG on these financial metrics, consistent with previous research.Two tests, the Chow Test and the Hausman Test were conducted to determine the most appropriate model for the analysis.The results led to the selection of the Random Effect Model (REM) for its efficiency, highlighting the common dilemma in econometric modeling where different tests may suggest different approaches.
In the REM analysis, SDG showed a small and non-significant positive association with both ROE and ROA.The introduction of firm size as a moderating factor demonstrated a significant positive effect on ROA, but not on ROE, aligning with other research indicating that larger firms often derive more tangible benefits from SDG activities.The application of robust standard errors to address potential heteroskedasticity did not significantly alter the results, underscoring the robustness of these findings.Overall, the study contributes to the ongoing discourse about the tangible financial benefits of SDG practices, suggesting that while SDG may not directly enhance financial performance indicators in the short term, its role in long-term value creation, risk mitigation, and stakeholder engagement remains a vital aspect of corporate strategy.
The F&B sector's unique challenges and opportunities were also highlighted, with the significance of firm size in positively affecting ROA, but not ROE, indicating that larger firms are more likely to harness the benefits of SDG.The contrasting impacts ofSDG across sectors such as coal mining and minerals versus the F&B industry were attributed to the unique operational, regulatory, and market dynamics inherent to each sector.In the F&B sector, the impact of SDG initiatives is more subtle and often manifests in areas like brand reputation and customer loyalty rather than direct financial performance, reflecting the sector-specific nature of SDG impact.
The study's findings underscore the need for a nuanced understanding of the relationship between SDG practices and financial performance, considering the specific characteristics of each industry.The F&B sector may realize the financial benefits of SDG initiatives over a longer term through sustained customer engagement and market positioning, rather than immediate improvements in financial metrics like ROE and ROA.The Table 0-9 above is a statistical analysis within the context of the Coal & Oil Refinery Subsector, specifically examining the impact of SDG (Environmental, Social, and Governance) Reporting on the financial performance of companies.For ROE (t+1), the coefficient for SDG is 0.033 with a t-statistic of 2.16, which is statistically significant at the 5% level (p < 0.05), denoted by two asterisks (**).For ROA (t+1), the coefficient is 0.001 with a t-statistic of 3.52, which is statistically significant at the 1% level (p < 0.01), denoted by three asterisks (***).The analysis suggests that in the Coal & Oil Refinery Subsector, SDG has a positive and statistically significant impact on both ROE and ROA.This could mean that for companies in this sector, a greater emphasis on SDG practices and reporting may be associated with better financial performance.The negative constant for ROE suggests that without the positive influence of SDG, the expected ROE might be negative, which could reflect challenges in the subsector, such as regulatory pressures or market conditions.Conversely, the positive constant for ROA indicates baseline profitability in terms of asset utilization.Given the significant coefficients and the context of the industry, this could imply that investors and stakeholders may reward companies that are actively engaging in SDG with higher valuations, potentially due to perceived lower risks or better management practices.However, the relatively low R-squared values suggest that other factors not included in the model also play a significant role in determining financial performance.

a. Hypothesis Testing using CEM (the influence of Sustainability Disclosure on ROE)
In the context of determining the optimal econometric model to assess the impact of SDG (Sustainable Development Goals) (x1) on the Return on Equity (ROE, y2), a comprehensive analysis was conducted using a series of statistical tests.The Chow test initially suggested no significant variations across groups, indicating that group-specific effects may not be prominent.This was followed by the Hausman test, which did not show any systematic differences between the Fixed Effects Model (FEM) and Random Effects Model (REM), implying that REM could be a viable option.However, the Breusch-Pagan test countered this by demonstrating a lack of significant variance across the groups, which would typically negate the need for REM.Considering the results of these three diagnostic tests, the most fitting model for this analysis emerges as the Common Effects Model (CEM).The within R-squared is 0.1141, which suggests that the model explains 11.41% of the variability in ROA within the groups.The between R-squared is 0.0000, indicating no variability in ROA between the groups is explained by the model.The overall R-squared is 0.0298, signifying that the model explains 2.98% of the total variance in ROA across all groups and observations.Coefficients: The coefficient for SDG reporting (x1) is 0.0016895 with a standard error of 0.00048.This coefficient is statistically significant (pvalue = 0.000), indicating a positive relationship between SDG reporting and ROA.The constant term (cons) is 0.0632672 with a standard error of 0.0295246.This is also statistically significant (p-value = 0.032), representing the average ROA when SDG reporting is zero.The Wald chi-squared statistic of 12.39 with a p-value of 0.0004 indicates that the model's predictors significantly contribute to the explanation of ROA.
The Random Effects Model shows a significant positive relationship between SDG and ROA, suggesting that companies that report on SDG metrics may see better asset returns.This could imply that investors and market participants view SDG reporting as indicative of a company's efficiency, risk management, or long-term sustainability, which could translate into financial performance.
However, the low overall R-squared indicates that while SDG reporting is a significant predictor of ROA, there are many other factors not included in this model that influence ROA.Also, the between R-squared of 0.0000 suggests that the model does not capture any variability between groups, which may be a concern if we expect differences between groups to influence ROA.The The table presents a statistical analysis examining the effect of Sustainable Development Goals (SDG) on financial performance, specifically on Return on Equity (ROE) and Return on Assets (ROA), with firm size as a moderating variable.The coefficient for SDG reporting on ROE is 0.045, with a t-statistic of 1.63.This indicates a positive relationship between SDG reporting and ROE, but it is not statistically significant at conventional levels (p > 0.1).For ROA, the coefficient is 0.0006 with a t-statistic of 1.57, also suggesting a positive relationship but not reaching statistical significance.

a. Hypothesis Testing (the influence of Sustainability Disclosure on ROE moderated by size)
The Table 0-13 CEM Regression Result for RQ#3 (SDG and ROE moderated by Firm Size) for Coal and Oil Sector below provided the results of a Common Effects Model (CEM) regression analysis for the coal and oil sector, examining the impact of Sustainable Development Goals (SDG) reporting (x1) on Return on Equity (ROE, y2) with firm size (x3) as a moderating variable.The coefficient for SDG reporting (x1) is 0.045825, which suggests a positive association with ROE.However, the t-statistic is 1.63, and the p-value is 0.106, indicating that this association is not statistically significant at the conventional 0.05 level.The coefficient for firm size (x3) is 0.4349853, with a t-statistic of 0.91 and a p-value of 0.362, also indicating no statistically significant impact on ROE at the conventional levels.The constant (_cons) coefficient is -3.880304, but it is not statistically significant (p-value = 0.345), suggesting that the baseline ROE when both SDG reporting and firm size are zero is not significantly different from this value.The F-statistic for the model is 2.75 with a p-value of 0.0677, indicating that the model is not statistically significant at the conventional 0.05 level, but it is marginally significant, suggesting that the independent variables as a whole may have some relationship with ROE.
The R-squared value is 0.0415, meaning that the model explains 4.15% of the variance in ROE.The Adjusted R-squared is lower at 0.0264, which accounts for the number of predictors in the model and indicates a relatively small amount of the variance in ROE is explained by the model.The Root Mean Square Error (Root MSE) is 7.6141, which gives an indication of the standard deviation of the residuals or the average distance of the data points from the fitted model.
The results suggest that, in this model for the coal and oil refinery subsector, neither SDG reporting nor firm size is a significant predictor of ROE when analyzed through a CEM.
The low explanatory power of the model (as indicated by the low R-squared values) suggests that other variables not included in the model may have a more significant impact on ROE.The lack of statistical significance for the moderating effect of firm size might indicate that the relationship between SDG reporting and ROE does not vary significantly with the size of the firm in this sector, or that the sample size or variability within the data is insufficient to detect such an effect.Given the results, it could be beneficial to reevaluate the model, considering additional variables that may influence ROE in the coal and oil sector, or to investigate the model further for potential non-linear effects or interaction effects that were not captured.It might also be worth exploring other forms of moderation analysis or different methodological approaches that could yield more insight into the conditions under which SDG reporting may affect ROE.The model assesses the effect of Sustainable Development Goals (SDG) reporting (x1) on Return on Assets (ROA, y3), with firm size (x3) as a moderating variable, within the coal and oil sector.
The coefficient for SDG reporting (x1) is 0.0006462, but it is not statistically significant (p-value = 0.118), indicating that SDG reporting does not have a significant impact on ROA within the entities after accounting for unobserved heterogeneity.The

METHODOLOGYFigure 1
Figure 1.Research Methodology 26 with a p-value of 0.0710.This test assesses the joint significance of all the coefficients in the model.The p-value, being just above 0.05, indicates a marginal level of significance, suggesting that the model's explanatory variables collectively have a borderline significant impact on ROE.The z-value for SDG (x1) is 1.81 with a corresponding p-value of 0.071.This p-value is greater than the conventional threshold of 0.05, indicating that SDG is not statistically significant at the adjusting for robust standard errors.The constant has a z-value of 0.34 with a p-value of 0.735, also indicating a lack of statistical significance.The R-squared is 0.0002, indicating that the model explains virtually none of the variation in ROE within groups.

Table 0- 13 CEM
Regression Result for RQ#3 (SDG and ROE moderated by Firm Size) for Coal and

b.
Hypothesis Testing (the influence of Sustainability Disclosure on ROA moderated by size) Table 0-14 FEM Regression Result for RQ#3 (SDG and ROA moderated by Firm Size) for Coal and Oil Sector

Table 0 -1 Summary of Sample Collection
174 2 Number of Annual Reports (AR), Sustainability Reports (SR), or Combined Reports (CR) published by coal and oil refining subsector companies listed on the Indonesia Stock Exchange (BEI) in 2018-2022 178 3 Number of Annual Reports (AR), Sustainability Reports (SR), or Combined Reports (CR) published by chemical and pharmaceutical industry subsector companies listed on the Indonesia Stock Exchange (BEI) in 2018-2022.63 Number of Annual Reports (AR), Sustainability Reports (SR), or Combined Reports (CR) that can be analyzed to express support for the Sustainable Development Goals.415 Content Analysis Content analysis is a method that encompasses categorization, systematic data recording to provide fresh insights about a phenomenon and analyzing specific patterns (Elo & Kyngäs, 2008).Data measurement in this research was achieved by assigning weights or values to each Sustainable Development Goal (SDG) within the Annual Reports (AR), Sustainability Reports (SR), or Combined Reports (CR) spanning from 2018 to 2022.The scoring system was devised considering the importance of the SDGs, which includes having targets and strategies necessary to showcase a company's commitment to supporting the SDGs, and not merely disclosing narrative information (Hsieh & Shannon, 2005).

Table 0 -2 above
summarizes the scoring scenario used in this analysis.

Table 0 -3 Effect of SDG on Financial Performance in Food & Beverage Sector
ISSN:

Table 0 -10 CEM Regression Result for RQ#1 Coal & Oil Refinery SubsectorTable 0 -
10 above shows the results from a regression analysis, where Return on Equity (ROE, y2) is regressed on Sustainable Development Goals reporting(SDG, x1).This analysis appears to be structured as a Common Effects Model (CEM), considering the previous discussions regarding model selection.The model has an F-statistic of 4.67, with a corresponding p-value of 0.0325.This indicates that the model is statistically significant at the 5% level, suggesting that the SDG reporting has a collectively significant effect on ROE.The R-squared value is 0.0352, which means that approximately 3.52% of the variability in ROE can be explained by the SDG reporting.The Adjusted R-squared, which accounts for the number of predictors in the model, is slightly lower at 0.0277, indicating a small but non-negligible explanatory power of the model.The Root Mean Square Error (Root MSE) is 7.6093, which is a measure of the standard deviation of the prediction errors or residuals.It gives us an idea of the typical size of the prediction errors.The coefficient for SDG reporting (x1) is 0.055883, with a standard error of 0.0258586.The t-statistic for this coefficient is 2.16, and the p-value is 0.033, indicating that the coefficient is statistically significant at the 5% level.This suggests that as SDG reporting increases, ROE is expected to increase by approximately 0.055883 units, holding all else constant.The 95% confidence interval for the SDG reporting coefficient ranges from 0.0047174 to 0.1070487, which does not include zero, reinforcing the significance of the result. ISSN:

2581-8341 Volume 07 Issue 03 March 2024 DOI: 10.47191/ijcsrr/V7-i3-48, Impact Factor: 7.943 IJCSRR @ 2024 www.ijcsrr.org 1870 * Corresponding Author: Lutrika Mufti Rachmat Volume 07 Issue 03 March 2024 Available at: www.ijcsrr.org Page No. 1857-1879
Based on this regression analysis within a CEM framework, there is evidence to suggest that SDG reporting is positively associated with ROE.However, given the low R-squared values, the effect size is quite small, and other unmodeled factors likely explain the majority of the variation in ROE.This model provides some insight into the relationship between SDG reporting and financial performance as measured by ROE, but it should be noted that the explanatory power of the model is limited, and further research might be necessary to uncover additional factors that influence ROE.Based on this assumption, the CEM model is stated as follows: ,+ = −. + . , +  , b.

Hypothesis Testing using REM (the influence of Sustainability Disclosure on ROA) Table 0-11 REM Regression Result for RQ#2 Coal and Oil Refinery Subsector The
Table 0-11 below shows results from a Random Effects Generalized Least Squares (GLS) regression that evaluates the impact of Sustainable Development Goals (SDG) reporting (x1) on Return on Assets (ROA, y3) across different groups coded as 'kode'.The regression is structured to account for variations both within and between 26 groups in the dataset.Each group has exactly 5 observations, indicating a balanced panel structure.