Non-Financial Information Disclosures and Firm Risk

Lincoln University Digital Thesis

Copyright Statement

The digital copy of this thesis is protected by the Copyright Act 1994 (New Zealand).

This thesis may be consulted by you, provided you comply with the provisions of the Act and the following conditions of use:

you will use the copy only for the purposes of research or private study you will recognise the author's right to be identified as the author of the thesis and

due acknowledgement will be made to the author where appropriate you will obtain the author's permission before publishing any material from the

thesis.

Non-Financial Information Disclosures and Firm Risk

A thesis

submitted in partial fulfilment

of the requirements for the Degree of

Doctor of Philosophy

at

Lincoln University

by

Muhammad Arif

Lincoln University

2020

I dedicate this work to my daughter Amsah, parents, wife, siblings,

friends, colleagues and all the teachers who supported and

encouraged me to reach this point in my life

i

Abstract of a thesis submitted in partial fulfilment of the requirements for the Degree of Doctor of Philosophy

Non-Financial Information Disclosures and Firm Risk

By

Muhammad Arif

Abstract

Recently, world leaders and organisations have realised the seriousness of unsustainable

production practices, ethical misconduct and governance failures by businesses that have

caused severe environmental hazards, employees’ and customers’ rights violations and

financial fraud. As a result, there have been increasing calls by regulators and stakeholders,

including firms’ shareholders, to improve the quantity and quality of information disclosures

vis-à-vis a firm’s business operations, financial transactions, governance structures, work

ethics, environmental and social ramifications. Besides, organisational theories such as the

agency, stakeholder and legitimacy theories highlight the importance of a transparent

information environment in mitigating concerns such as agency conflicts and legitimacy

issues.

Generally, firms use conventional financial statements and reports that offer limited

information about the firm’s activities. However, more recently, firms started to supplement

conventional financial statements with non-financial information (NFI) that encompasses

information related to their policies and activities for a sustainable environment, societal

well-being, work ethics and governance. This additional disclosure provides opportunities to

explore the impact of additional such information disclosure on a firm’s information

environment, future performance and risk.

Recognizing this opportunity and the knowledge gap in the literature, this study uses the S&P

1200 index firms’ environmental, social and governance (ESG) disclosures for 2008-2018 to

investigate their contribution to reducing the uncertainty surrounding firms’ future earnings.

The results show that ESG disclosures have a significant negative impact on earnings risk,

ii

signifying that the additional disclosures relating to ESG issues reduce the uncertainty

surrounding firms’ future earnings risk. The disaggregated analysis of ESG disclosure scores

shows a more pronounced effect of the social and environmental dimensions on earnings risk.

The results are robust to various sub-sample analyses and endogeneity controls.

This study also explores how sensitive industry status augments the extent of a firm’s NFI

disclosures as well as the association between NFI disclosures and earnings risk. The results

show that sensitive industry firms make higher NFI disclosures and, as a result, have a lower

earnings risk than non-sensitive industry firms. This study provides novel evidence concerning

the causal impact of the European Union’s NFI reporting regulations on the level of ESG

disclosures of regulated firms and the changes in NFI disclosures and earnings risk association

after the mandating NFI reporting. The results show that the mandatory NFI reporting regime

has not only increased the extent of NFI disclosures for European firms, but has also improved

the efficiency of NFI disclosures in alleviating the uncertainty surrounding the future earnings

of the sample firms.

Our findings provide important implications for corporate managers, stakeholders and

regulators. Specifically, the findings suggest that corporate managers can alleviate agency

conflicts and legitimacy issues by reporting information regarding firms’ policies and efforts

to manage ESG issues. Our finding regarding the efficiency of the EU directive to increase the

quantity of NFI disclosures further suggests that regulators from other parts of the world

should shift towards mandatory NFI reporting. We propose that mandatory NFI regimes

would improve the adoption of sustainable business practices that are necessary to achieve

long-term targets such as sustainable development and the Paris Agreement’s targets.

Keywords: Non-financial information disclosures, earnings risk, sensitive industries, non-

financial reporting regulations, directive 2014/95/EU

iii

Table of Contents

Abstract ……………………………………………………………………………………………………………………………….i

Table of Contents ...................................................................................................................... iii

List of Tables ............................................................................................................................. vi

List of Figure ............................................................................................................................. vii

Acknowledgements ................................................................................................................. viii

Chapter 1 Introduction………………………………………………………………………………………………………..1

1.1. Introduction ............................................................................................................................ 1

1.2. Knowledge Gaps...................................................................................................................... 5

1.3. Research Questions ................................................................................................................ 6

1.4. Purpose of the Research ......................................................................................................... 7

1.5. Significance of the Research ................................................................................................... 8

1.6. Research Findings ................................................................................................................. 10

1.7. Organisation of the Thesis .................................................................................................... 11

Chapter 2 Literature Review .................................................................................................. 13

2.1. Introduction .......................................................................................................................... 13

2.2. Theoretical Foundations ....................................................................................................... 14

2.2.1. Agency Theory ............................................................................................................... 15

2.2.2. Legitimacy Theory ......................................................................................................... 16

2.2.3. Stakeholder Theory ....................................................................................................... 18

2.3. Definition Evolution of Non-financial Information ............................................................... 19

2.3.1. Institutional contribution to the definition evolution of Non-financial Information ..... 20

2.4. Evolution of Non-financial Information Reporting Research ................................................ 22

2.4.1. Initial studies on the measurement of NFI information disclosures .............................. 22

2.4.2. The development of improved NFI disclosure databases .............................................. 23

2.5. Empirical Studies on the Impact of NFI Disclosures .............................................................. 24

2.5.1. NFI disclosures and the firm performance link .............................................................. 25

2.5.2. The NFI and information asymmetry link ...................................................................... 25

2.5.3. NFI disclosures and the firm risk link ............................................................................. 27

2.5.4. The linkage between NFI disclosures and firms’ earnings risk (the case of sensitive

industry firms) ............................................................................................................................... 30

2.6. NFI Disclosures Regulations .................................................................................................. 32

iv

2.6.1. NFI disclosures regulations in Europe Directive 2014/95/EU ........................................ 33

2.6.2. The consequences of the NFI disclosure regulations ..................................................... 34

2.7. Chapter Summary ................................................................................................................. 36

Chapter 3 Sample, Data and Methodology ............................................................................ 38

3.1. Introduction ......................................................................................................................... 38

3.2. Sample .................................................................................................................................. 38

3.3. Dependent and Independent Variables ................................................................................ 40

3.3.1. Analysts’ earnings forecasts dispersion – earnings risk relationship ............................ 40

3.3.2. Non-Financial Information disclosures .......................................................................... 41

3.3.3. Sensitive industries ........................................................................................................ 42

3.3.4. Non-Financial Information disclosure regulations ........................................................ 42

3.4. Control Variables ................................................................................................................... 43

3.4.1. Firm size ........................................................................................................................ 43

3.4.2. The number of analysts following a firm ...................................................................... 44

3.4.3. Financial leverage ......................................................................................................... 44

3.4.4. Loss ................................................................................................................................ 45

3.4.5. Earnings volatility .......................................................................................................... 45

3.4.6. Financial opaqueness .................................................................................................... 45

3.4.7. Firm growth ................................................................................................................... 46

3.4.8. Firm profitability ........................................................................................................... 46

3.4.9. Industry and index dummies ......................................................................................... 46

3.5. Basic Diagnostic Test ............................................................................................................. 47

3.5.1. Pairwise correlation ...................................................................................................... 48

3.6. Regression models ................................................................................................................ 48

3.7. Robustness and Endogeneity ................................................................................................ 52

3.8. Chapter Summary ................................................................................................................. 53

Chapter 4 Results and Discussion ........................................................................................... 54

4.1. Introduction .......................................................................................................................... 54

4.2. Descriptive Statistics ............................................................................................................. 54

4.3. Diagnostics Test .................................................................................................................... 57

4.3.1. Pairwise correlation analysis ............................................................................................... 57

4.4. The Impact of NFI Disclosures on Earnings Risk .................................................................... 61

4.4.1. Baseline results ................................................................................................................... 61

4.4.2. NFI quartile results .............................................................................................................. 64

4.4.3. Subsample analysis ............................................................................................................. 67

v

4.4.4. Robustness tests .................................................................................................................. 70

4.5. Sensitive Industries Analysis ................................................................................................ 77

4.5.1. Level of NFI disclosure: Sensitive v/s non-sensitive industry firms ...................................... 78

4.5.2 NFI disclosures and future earnings risk: Sensitive versus non-sensitive industry firms ...... 81

4.5.3 NFI disclosures and future earnings risk: Interaction approach........................................... 82

4.6. Chapter Summary ................................................................................................................. 85

Chapter 5 Mandatory NFI Disclosures and Firm Risk ......................................................... 87

5.1. Introduction ......................................................................................................................... 87

5.2. The European Union’s NFI Directive 2014/95/EU ................................................................ 87

5.3. The Impact of Directive 2014/95/EU on NFI Disclosures ...................................................... 88

5.4. The Impact of the EU Directive on Future Earnings Risk ...................................................... 93

5.5. Chapter Summary ................................................................................................................ 97

Chapter 6 Summary and Conclusions ................................................................................ 99

6.1. Introduction ......................................................................................................................... 99

6.2. Research Background and Objectives ................................................................................... 99

6.3. Study Sample, Data and Empirical Models ........................................................................ 100

6.4. Summary of Findings ........................................................................................................... 100

6.4.1. NFI disclosures and earnings risk ...................................................................................... 101

6.4.2. Sensitive industries and NFI disclosures ............................................................................ 102

6.4.3. NFI disclosures and earnings risk (sensitive vs non-sensitive industry firms) ................... 102

6.4.4. The impact of mandatory NFI regulations ........................................................................ 103

6.4.5. The impact of mandatory NFI regulations on the NFI and earnings risk link .................... 103

6.5. Practical and Policy implications ......................................................................................... 104

6.5.1. Implications for corporate managers................................................................................ 104

6.5.2. Implications for firm stakeholders .................................................................................... 104

6.5.3. Implications for regulators ................................................................................................ 105

6.6. Research Contribution ........................................................................................................ 106

6.7. Research Limitations ........................................................................................................... 107

6.8. Future Research ................................................................................................................. 108

Appendix A ............................................................................................................................. 109

Appendix B ............................................................................................................................. 115

Appendix C ............................................................................................................................. 117

References ………………………………………………………………………………………………………………………139

vi

List of Tables

Table 3. 1. Index and industry-wide companies that comprise the study sample ............................... 39

Table 3. 2. A summary of the study variables and expected signs ....................................................... 47

Table 4. 1. The descriptive statistics of the study variables ................................................................. 56

Table 4. 2. Pairwise correlation matrix of the study variables ............................................................. 60

Table 4. 3. The impact of NFI disclosures on firms’ earnings risk ......................................................... 62

Table 4. 4. The impact of NFI disclosure quartiles on firms’ earnings risk ........................................... 66

Table 4. 5. The impact of NFI disclosures on stock return volatility ..................................................... 70

Table 4. 6. The impact of NFI disclosures on firms’ earnings risk (Alternative proxies) ....................... 71

Table 4. 7. The impact of NFI disclosures on firms’ earnings risk (Alternative proxies) ....................... 72

Table 4. 8. The impact of NFI disclosure on firms’ earnings risk (Lagged variables) ............................ 74

Table 4. 9. The impact of NFI disclosure on firms’ earnings risk (System GMM estimator) ................. 76

Table 4. 10. Balance table of ES statistic comparison ........................................................................... 80

Table 4. 11. The effect of sensitive industry membership on NFI disclosure levels ............................. 81

Table 4. 12. The impact of NFI disclosures on firms’ earnings risk – Split sample approach ............... 83

Table 4. 13. The impact of NFI disclosures on firms’ earnings risk – Interaction approach ................. 84

Table 5. 1. The balance table of ES statistics ........................................................................................ 90

Table 5. 2. The impact of the EU directive on NFI disclosures .............................................................. 92

Table 5. 3. The impact of NFI disclosures on firms’ earnings risk (pre and post EU directive) ............. 95

Table 5. 4. The impact of NFI disclosures on firms’ earnings risk (Australian sample) ......................... 96

Table A.1. The Hausman test estimates of the study variables .......................................................... 109

Table A.2. The impact of NFI disclosures on firms’ earnings risk ....................................................... 110


Table A. 4. The impact of NFI disclosures on firms’ earnings risk ....................................................... 112


Table A.6. The impact of NFI disclosures on firms’ earnings risk (non-financial firms) ..................... 114

vii

List of Figure

Figure 4.1. Evolution of NFI disclosures ................................................................................................ 57

Figure B.1. Overlap plot of propensity scores between sensitive and non-sensitive firms ................ 115

Figure B.2. Overlap plot of propensity scores between regulated and unregulated firms ................ 116

viii

Acknowledgements

All the praises and thanks be to Allah Almighty, Who guided me and gave me the strength to

complete my PhD studies. May the peace and blessing of Allah Almighty be upon Muhammad

(peace be upon him), who is the perfect mentor.

Firstl, I express my utmost gratitude to Prof. Christopher Gan for his guidance, support and

motivation throughout this journey. He did not only guide and support me to complete the

PhD thesis but also provided many opportunities to learn and grow in the academic field. He

has always been a source of motivation and inspiration throughout my time at Lincoln

University. I would also sincerely thank my asscociate supervisor, Dr Muhammad Nadeem

(Senior Lecture at Otago Business School, Otago University), for his valuable guidance and

contribution to my PhD research thesis.

I thank my employer Shaheed Benazir Bhutto University (SBBU), Shaheed Benazirabad,

Pakistan, and the Higher Education Commission, Pakistan, for the provision of funds for my

PhD studies without which this journey would not have been possible. Notably, I thank Mr

Najam Ud din Sohu, Akhtar Hussain Mangi and Qurban Farooqi for their continuous support

during my PhD tenure at Lincoln University. I would also like to thank Nicos Tescos who

ensured my employment at Lincoln University.

The completion of my PhD studies and my stay in New Zealand would not have been possible

without continuous support from my family, especially my parents, wife, siblings and in-laws.

They have always encouraged and supported me during this time. I pay special thanks to my

father and mentor, Muhammad Ramzan. He has always been a source of inspiration and

motivation to achieve higher standards of education throughout my life.

I also thank all my friends who supported me to this point in my life. Notably, my sincere

gratitude goes to Mudassar Hasan, Abu Bakr Naeem, Aon Waqas, Javed Ali, Rashid Waheed,

Sanaullah, Raheel Khan, Tauseef Khan Babar and Mabruk Billah for their support and guidance

throughout my life. Lastly but most importantly, I extend sincere gratitude to all my teachers

and alma maters. They nurtured my skills and abilities to achieve the ultimate goal in my

academic career, the completion of my PhD studies.

1

Chapter 1

Introduction

1.1. Introduction

Today’s world is overloaded with an abundance of information and there has never been

more demand for data. Similarly, businesses are increasingly required to report information

related to all aspects of their operations and activities. Requests for such information not only

come from the firms’ shareholders but also other stakeholders such as suppliers, regulators,

NGOs and creditors, who demand information about firms’ business processes, products,

policies and impacts on society (Camilleri, 2015).

Stakeholders, including the firms’ shareholders, seek information to make informed

decisions. For example, shareholders and creditors make investment decisions; suppliers

make decisions to conduct business with a firm and customers make purchase decisions. All

these stakeholders need information to make the best decisions and narrow the information

gap between themselves and corporate managers, better known as the information

asymmetry problem in the finance literature (Bhattacharya & Singh, 2018; Cui, Jo, & Na, 2018;

Easley & O'hara, 2004).

To satisfy their stakholders’ information needs, historically, firms have provided information

mainly on their financial matters using conventional financial statements such as the

statement of financial position, income statement and statement of stakeholders’ equity.

However, these statements mainly cater to the information needs of stakeholders with a

financial interest in the firm, i.e., shareholders and creditors. The information needs of

stakeholders seeking data about a firm’s policies and activities for a sustainable environment,

society well-being, work ethics and governance issues, remain unnswered (Bozzolan, Favotto,

& Ricceri, 2003; Manes-Rossi, Tiron-Tudor, Nicolò, & Zanellato, 2018).

As a result, corporate managers started to report more information in their business

communications to cater for the needs of the broad range of stakeholders. This information

is mainly non-financial and includes details of a firm’s policies and practices concerning

governance, environmental sustainability, society, and employees (Burgman, Roos, Boldt-

Christmas, & Pike, 2007; Cortesi & Vena, 2019; Felber, Campos, & Sanchis, 2019). Firms use

2

several reports and statements to make non-financial information (NFI) disclosures. The most

common statements are: corporate social responsibility (CSR) reports, statements of the code

of ethics and conduct, statements from the senior decision-maker (CEO, President),

statements of corporate intent, corporate governance reports, and, more recently, integrated

reports. Some of the NFI disclosure reports such as CSR and sustainability reports are used

interchangeably in this study.

Historically, NFI disclosures have had a long journey, starting with the concept of social

responsibility (Bowen, 1953) and stakeholder concerns (Davis, 1960). Sethi (1975)

emphasised the need to link social responsibility with corporate strategy and actions.

Progressively, Carroll (1979) provided a detailed framework of corporate social performance

consisting of four stages of customer social responsibility (CSR) development: economic, legal,

ethical, and philanthropic obligations. These ground-breaking studies laid the foundations for

issues that business managers should address in their corporate communications. There has

been a steady increase in NFI disclosures and many third party organisations have started to

develop NFI databases1 that measure the extent of corporate disclosures and performance in

the areas of environmental, social, economic and governance issues. Given the availability of

comprehensive NFI databases, NFI disclosures are also available in disaggregated components

as environmental, social and governance (ESG) disclosures.

Although there has been a recent noteworthy increase in the quantity and quality of NFI

reporting (Arvidsson, 2011; Coluccia, Fontana, & Solimene, 2016; Jain, Keneley, & Thomson,

2015; Sethi, Martell, & Demir, 2017) around the world, these disclosures have been largely

voluntary (Hoffmann, Dietsche, & Hobelsberger, 2018; Muslu, Mutlu, Radhakrishnan, &

Tsang, 2019). Because of their voluntary nature and a lack of accounting standards and

frameworks, the format and content of NFI disclosures have generally been unstandardized

and inconsistent (Demir & Min, 2019; Muslu et al., 2019; Sethi et al., 2017). Additionally, rapid

growth and changes in the content and methodology of sustainability reporting tools

complicate the use of NFI disclosures for organisational stakeholders (Siew, 2015).

Consequently, there has been a lack of interest from academics, analysts and practitioners

1 Kinder, Lydenberg and Domini (KLD), Thomson Reuters ESG, Refinitiv ESG, Bloomberg ESG database and others.

3

about using the information content of these statements and reports in their efforts to

comprehend organisations’ NFI disclosures and their quality.

Nevertheless, after the establishment of international standards and organisations such as

ISO 26000, the Global Reporting Initiative (GRI), Sustainability Accounting Standards Board

(SASB) and the International Integrated Reporting Council (IIRC), firms now have more

knowledge and better guidance available to develop their NFI disclosures. Therefore,

organisations adopting the improved reporting frameworks produce standardised, coherent

NFI disclosures that are useful to measure the quantity and quality of NFI reporting across a

large variety of organisations (Bhattacharyya & Yang, 2019; Crisóstomo, de Azevedo

Prudêncio, & Forte, 2017; Demir & Min, 2019). Additionally, countries and economic regions2

are enacting NFI disclosure regulations and developing reporting frameworks to provide

consistent, mandatory guidelines for firms’ NFI disclosures. These regulations3 mainly target

achieving sustainable, accountable, responsible and transparent business practice targets set

by relevant governments, the United Nations’ sustainable development goals (SDGs), and the

United Nations Paris Climate Change Agreement (Nations, 2015).

In this vein, (Directives, 2014) enacted by the European Parliament, provides a detailed

framework for mandatory NFI reporting by large4 organisations and required all member

states to adopt the framework by the end of 2016. This directive amends the non-mandatory

Directive 2013/34/EU and Regulation (EC) No 562/2006 that were in place in the EU for

disclosure of NFI. This recent directive requires firms to report information on six broad

topics: 1) environmental, 2) social, 3) employee matters, 4) respect for human rights, 5) anti-

corruption, and 6) bribery matters, as the minimum mandatory requirement of NFI

disclosures. Firms are also required to report how their business model, policies, policy

outcomes, potential risks and key performance indicators (KPIs) synchronise with the six

broad topics in NFI disclosures.

2 An economic region is a group of countries that have geographical proximity and share similar economic and

financial patterns and regulations e,g,. ASEAN, the EU, South Asian Association for Regional Cooperation

(SAARC). 3 NFI reporting is a stock exchange listing requirement in Malaysia and South Africa. China and Denmark made ESG

reporting mandatory for large organisations in 2008 and 2009, respectively. 4 Companies with a total balance sheet, net turnover and average number of employees over EUR 400,000, EUR

800,000 and 500, respectively.

4

Significant developments in the reporting and regulatory frameworks provide opportunities

to explore the impact of NFI disclosure regulations on the quantity and quality of corporate

NFI disclosures. Notably, the recent NFI reporting regulation by the EU is vital for two reasons.

First, it has to be adopted by all EU member countries, so it has considerable implications in

its scope. Secondly, it affects some of the largest corporations in the world.

The literature also reports the impact of NFI reporting regulations on the extent of NFI

disclosures and how such regulations affect firm risk and value. For example, Ioannou and

Serafeim (2017) report increased ESG disclosures following the enforcement of mandatory

NFI disclosure regulations in China, Denmark, Malaysia, and South Africa. They also report an

increase in firm value for reporting firms. Similarly, using South Africa’s integrated reporting

regulation, Bernardi and Stark (2018) report an increased level of ESG disclosures that, in turn,

results in increased analysts’ forecast accuracy. These findings show NFI regulations not only

increase the level of CSR disclosures, but also moderate the link between CSR disclosures and

information asymmetry and firm value. Given the recent NFI regulation enacted by the

European parliament in 2014, it is pertinent to investigate how the enforcement of this

regulation affects the disclosure practices of the regulated organisations.

Potential improvements in the quality of NFI disclosures under a mandatory reporting regime

can serve as a channel to reduce information asymmetry among a range of stakeholders. The

improved quality of disclosed information serves as a legitimacy enhancing tool and offers

potential benefits. For example, using South Africa’s mandated integrated reporting case, Lee

and Yeo (2016) suggest that integrated reporting mitigates the information asymmetry

among managers and creditors. It serves as a tool for enhancing firm value. Similarly, Bernardi

and Stark (2018) report an increase in analysts’ earnings forecast accuracy after the

mandating of integrated reporting in South Africa. Their results suggest that the mandated

disclosure regime prompted higher quality of information disclosures, resulting in reduced

information asymmetry among financial analysts. Therefore, it is important to explore

whether improved disclosure helps regulated organisations to reduce their risk.

In a broader context, the literature reports a significant negative relationship between NFI

disclosures and firm risk. Prior studies use standard measures of firm risk such as cost of

capital, bid-ask spread, idiosyncratic risk, systematic risk and analysts’ forecasts. Studies on

5

the relationship between NFI disclosure and analysts’ forecasts dispersion provide evidence

of a significant negative relationship (Bouslah, Kryzanowski, & M’Zali, 2013; Chien & Lu, 2015;

Dhaliwal, Li, Tsang, & Yang, 2011; Jo & Na, 2012; Kothari, Li, & Short, 2009). Some studies use

analysts’ forecasts dispersion as a measure of information asymmetry (Dhaliwal,

Radhakrishnan, Tsang, & Yang, 2012; Schulz, 2017) and others as a measure of risk (Chien &

Lu, 2015; Kothari et al., 2009) because of different underlying assumptions. Nevertheless,

Barron, Stanford, and Yu (2009) reconcile the contradictory evidence of analysts forecasts

dispersion as a measure of information asymmetry or uncertainty. They conclude that the

dispersion level of analysts’ forecasts represents uncertainty, which is negatively associated

with firms’ future stock returns. Consequently, we apply this explanation to use analysts’

forecasts dispersion as a measure a firm’s future earnings risk. We maintain that analysts’

forecasts dispersion is a better measure for analysing the complexity of a firm’s information

environment than traditional risk proxies such as systematic, idiosyncratic and total risk.

1.2. Knowledge Gaps

The availability of comprehensive NFI disclosure datasets and their division into

environmental, social and governance components provides promising avenues to

understand which dimensions of NFI disclosures are most valued by the market participants

such as financial analysts, investors and regulators. Therefore, it is worthwhile to investigate

how the total and different dimensions of NFI disclosures affect an organisation’s information

environment and influence its future earnings’ risk. This enquiry becomes more important for

firms operating in sensitive industries5. Given the environmental and social sensitivity of their

operations, firms in sensitive industries face stringent scrutiny from stakeholders to keep their

communications transparent and provide all the relevant information to stakeholders (Emma

& Jennifer, 2021; Sethi, Martell, & Demir, 2016; Yusoff, Yusoff, Abd Rahman, & Darus, 2019).

Previous empirical studies on NFI disclosure and the firm risk relationship report economically

valuable, statistically significant findings for a diverse collection of samples and periods. For

example, (see Benlemlih et al., 2018; Oikonomou, Brooks and Pavelin, 2012; Sassen, Hinze

and Hardeck, 2016). However, to the best of our knowledge, no previous study explores the

5 Following Baron et al. (2011) and Garcia et al. (2017), we treat a firm as a sensitive industry firm if it belongs to

the alcohol, tobacco, gambling, weapon production, adult entertainment, oil, gas, and consumable fuels, metals

and mining, paper and forest products, chemicals, construction materials, or energy businesses.

6

disaggregated6 effect of NFI disclosures on earnings risk. The literature is also silent on the

impact of NFI disclosures on earnings risk for sensitive industries, i.e., industries that need to

make higher NFI disclosures to maintain the legitimacy of their operations.

Because of increasing pressure from stakeholders and regulators, NFI disclosures are shifting

from voluntary to mandatory. This change in the status of NFI disclosures has significant

implications for corporate disclosure policies. Prior studies report the effect of NFI regulation

on the extent of a firm’s disclosures and its mediating impact on firm value (Ioannou &

Serafeim, 2017) and analysts’ forecasts accuracy (Bernardi & Stark, 2018) at country level.

Similarly, recent NFI regulations in the EU provide a unique opportunity to examine the effect

of mandatory NFI regulations on the magnitude of firms’ disclosures at a regional level. It is

also worthwhile investigating the mediating impact of mandatory NFI reporting regulations

on firms’ earnings risk and the NFI disclosure relationship.

1.3. Research Questions

Based on the research gaps identified from the literature review, this study aims to address

the following research questions:

i. What is the impact of cumulative and disaggregated NFI disclosure scores on the

earnings risk for S&P1200 index firms?

ii. What is the impact of sensitive industry status on the level of overall and

disaggregated NFI disclosures for S&P1200 index firms?

iii. Does the impact of total and disaggregated NFI disclosure scores on earning risk vary

between sensitive and non-sensitive industry firms included in the S&P 1200 index?

iv. How did the enforcement of mandatory NFI disclosure requirements change the

degree of NFI disclosures at the total and disaggregated levels for S&P 350 EU index

firms?

v. Did the enforcement of mandatory NFI disclosure requirements strengthen the

relationship between NFI disclosure and earnings risk for S&P 350 EU index firms?

6 NFI disclosures are commonly disaggregated into environmental, social and governance (ESG) components.

7

1.4. Purpose of the Research

As discussed earlier, organisations increasingly make NFI disclosures to provide much-needed

information to market participants so they can make informed decisions. Using NFI

disclosures, organisations aim to send positive signals to stakeholders and market participants

to enhance organisational legitimacy (Ochi, 2018). The purpose of these signals is two-fold;

first, these signals improve firms’ legitimacy. Secondly, they improve firms’ information

environment, which results in better forecasts of their future earnings.

To test this idea, this study first explores the nature and magnitude of NFI disclosures’ impact

on future earnings risk for S&P1200 index companies. The proxy for NFI disclosure is ESG

disclosure scores reported in the Bloomberg ESG database and the proposed proxy for

earnings risk is analysts’ earnings forecasts dispersion. We adopt Barron et al. (2009)

definition to measure the dispersion of analysts’ earnings forecast for a firm, which represents

a firm’s earnings risk. This study also explores the impact of individual components of ESG,

environmental, social and governance disclosures, on earnings risk. The disaggregated

analysis aims to ascertain whether different categories of ESG disclosure have varying

importance in improving a firm’s information environment.

Secondly, this study aims to measure the impact of sensitive industry status on firms NFI

disclosures and the resulting impact of NFI disclosure scores on firm risk for sensitive industry

firms. We aim to confirm and extend the evidence in the literature that sensitive industry

firms make higher NFI disclosures because of higher environmental and social risk exposure

(Du & Vieira, 2012; Garcia, Mendes-Da-Silva, & Orsato, 2017; Kilian & Hennigs, 2014; Vollero,

Conte, Siano, & Covucci, 2019). Following (Baron, Harjoto, & Jo, 2011; Garcia et al., 2017), we

treat a firm as a sensitive firm if it belongs to the alcohol, tobacco, gambling, weapon

production, adult entertainment, oil, gas, and consumable fuels, metal and mining, paper and

forest products, chemicals, construction materials, or energy businesses.

Thirdly, this study explores the impact of NFI reporting regulations on the quantity of firms’

NFI disclosures. To test this relationship, we use the recent NFI disclosure regulations enacted

by the European parliament as an exogenous shock for the disclosure practices of

organisations domiciled in the EU. Lastly, this study investigates the changes in the

relationship between earnings risk and NFI disclosures after the enforcement of mandatory

8

disclosure regulations. To that end, this study uses difference in differences (DID) analysis to

investigate the enforcement impact of NFI reporting regulations on the relationship between

NFI disclosures and earnings risk for European firms. For the DID method, the treatment group

includes constituents of the S&P Europe 350 Index, which is a sub-index of the S&P Global

1200 Index. We include S&P Europe 350 index companies in the treatment group as the

(Directives, 2014) applies only to companies domiciled in the EU. The other companies in the

S&P Global 1200 Index constitute the control group.

1.5. Significance of the Research

Recently, business organisations have not only faced increased pressure from legislators and

international standards organisations to include NFI disclosures in their corporate

communications, but investors and professional fund managers also assign a higher weight to

NFI disclosures in their investment decisions. A 2018 report7 documents professionally

managed investments of US$20 trillion are tied exclusively to socially responsible investment

(SRI) funds, which have increased from $7.6 trillion in 2010, a growth of 163% in just eight

years. Similarly, signatories to the United Nations-backed Principles of Responsible

Investments (UNPRI) have increased from 850 to 1600. McKinsey and Company (2002) reveal

that investors are willing to pay an average of a 25% premium for rightly governed company

shares. Additionally, consumers are willing to pay higher prices for eco-friendly products

(Fanasch & Frick, 2020; Mostafa, 2016). These findings suggest that socially responsible

investments are gaining momentum so investors and other stakeholders will assign a higher

weight to firms’ NFI disclosures in making investment decisions and assessing financial

performance and risk. Consequently, firms will improve their NFI disclosures to attract more

socially responsible investors and also avoid information the information asymmetry problem

between different stakeholder groups.

However, given the nature of the information and the lack of homogeneous reporting

frameworks, the use of broader NFI disclosure proxies such as stand-alone sustainability/CSR

reports is not a suitable approach. Thus, comprehensive, well-developed NFI disclosure

proxies are required to capture the real effect of NFI disclosures on firms’ earnings risk.

7 Report on US Sustainable, Responsible and Impact Investing Trends

https://www.ussif.org/files/Trends/Trends%202018%20executive%20summary%20FINAL.pdf

9

Consequently, this study uses the Bloomberg ESG disclosure score as a proxy of a firm’s NFI

disclosures. The Bloomberg ESG database provides data for over 11,500 companies across 70

countries. Bloomberg analysts construct ESG disclosure scores by using an over 900 data field

across a variety of NFI indicators, such as air quality, water and energy management, human

capital, waste management, compensation, diversity, shareholders rights. Bloomberg uses

company reported data from CSR reports, annual reports, company websites and other

company sources to maintain accuracy and consistency in their database8.

Given the depth and breadth of data used, Bloomberg ESG disclosure scores serve as a better

proxy than broader proxies of NFI disclosures such as stand-alone CSR reports, and inclusion

in sustainability indices. Additionally, the availability of individual scores for the ESG

dimensions is useful to measure the disaggregated impact of each component of ESG

disclosures on a firm’s risk. The disaggregated analysis will help to identify which aspects of

ESG disclosures drive the association between firm risk and NFI disclosures.

This study uses analysts’ earnings forecasts dispersion as a proxy for a firm’s earnings risk. We

use this proxy for two reasons. First, the accuracy of analysts’ earnings forecasts for a firm

mainly depends on the availability of public information. Secondly, this proxy better captures

the risk inherent in the information environment of a firm compared with conventional risk

proxies such as stock return volatility, systemic risk and idiosyncratic risk.

The arguments presented above suggest the use of precise, well developed NFI disclosures

and firm risk proxies to capture true nature and magnitude of the relationship between firm

risk and NFI disclosures. To the best of our knowledge, no prior study provides categorical

evidence of a relationship between firm risk and NFI disclosures at the international level.

Hence, this study’s results promise to add new insights to existing NFI disclosure literature.

Further, the use of analysts’ forecasts dispersion as an earnings risk proxy is a novel, valuable

contribution to the current knowledge concerning the effects of NFI disclosures on firms’ risk

profile. This study also presents evidence about sensitive industry firms that face stringent

scrutiny because of the nature of their business.

8 https://data.bloomberglp.com/professional/sites/10/1148330431.pdf

10

This study’s findings provide useful suggestions for corporate managers and decision makers

about reducing the uncertainty surrounding future earnings of a firm through the information

channel. This study’s results can potentially be used to mitigate agency problems and

legitimacy issues that are critical for modern firms, especially sensitive industry ones.

Addressing the agency and legitimacy issues have become essential because of ever-

increasing governance and sustainability issues, stakeholder pressures and shifting investors’

preferences towards socially responsible investment. Thus, this study is a timely enquiry to

answer pressing questions that corporate managers face regarding their efforts to minimise

the adverse environmental and social effects of business operations and practices.

Prior studies, e.g., (Schulz, 2017; Siew, Balatbat, & Carmichael, 2016), investigating the

relationship between information asymmetry and firm risk indicate the need for the

enactment and enforcement of mandatory NFI disclosure regulations to improve timeliness

and standardisation of NFI disclosures. Those studies also suggest standardisation and

timeliness of NFI disclosures would enhance analysts’ information processing efficiency.

Given the importance of NFI regulations, this study investigates the impact of mandatory NFI

regulations on the extent of NFI disclosures and determines the changes in the relationship

between firms’ earnings risk and NFI disclosures after the enforcement of such regulations.

Importantly, this study investigates the impact of the NFI regulations enacted by the EU. This

regulation provides the opportunity to investigate the impact of NFI reporting regulations at

a regional level. Exploring the impact of mandatory NFI regulations at a regional level is a

valuable contribution to the literature since current evidence is limited to country-level

(Bernardi & Stark, 2018; Ioannou & Serafeim, 2017) regulations. To the best of our knowledge,

this is the first study that attempts to ascertain the impact of a legislative change in this

research area at an international level.

1.6. Research Findings

This study’s results reveal important insights regarding the relationship between NFI

disclosures and firms’ earnings risk. First, this study shows that cumulative and disaggregated

NFI disclosures exhibit a strong negative impact on firms’ earnings risk. Notably, the social

and environmental dimensions of NFI disclosures exhibit negative impact on firms’ earnings

risk. However, governance disclosures do not reduce earnings risk for our sample firms. The

11

quartile-based analysis reveals that higher quartiles of overall and disaggregated NFI

disclosures have a more substantial negative impact than lower quartiles. These results are

robust for alternative measures of firm risk and firm characteristics, sub-sample analyses, and

endogeneity problems.

Second, using a propensity score weighted sample, this study shows that sensitive industry

firms report higher overall and disaggregated NFI disclosures than non-sensitive firms. This

difference is more pronounced for the environmental and social dimensions. Using the

interaction term and split-sample approaches, this study confirms that the disclosure of

environmental information provides more benefit to sensitive industry firms in reducing

earnings risk than for non-sensitive industry firms. Governance disclosures benefit non-

sensitive industry firms only with regard to reducing uncertainty surrounding future earnings.

This study offers significant evidence that sensitive industry firms may increase the legitimacy

of their operations and reduce earnings risk by improving their NFI disclosures, especially the

disclosure of environmental information.

Finally, this study uses a recently enacted EU NFI reporting directive to quantify its impact on

the NFI disclosures level and the resulting impact on the earnings risk of European firms in

the study sample. Using a propensity weighted sample, this study finds a significant positive

impact of the EU directive on the level of cumulative and disaggregated disclosure scores of

the treated firms compared with the control group firms. The difference-in-differences

analysis results confirm that the post-directive cumulative and disaggregated NFI disclosures

have a more pronounced impact on earnings risk than pre-directive disclosures. These

findings provide decisive evidence to conclude that mandating NFI disclosures results in

higher disclosures that in turn, increase the efficiency of reported information in reducing

uncertainty surrounding a firm’s future earnings.

1.7. Organisation of the Thesis

The rest of the thesis is as follows: Chapter 2 presents a literature review related to NFI

disclosures, the theoretical foundations of NFI disclosures, the relationship between NFI

disclosures and firm risk. The literature on the impact of NFI disclosure regulations on the

degree of disclosures and firm risk is discussed. Chapter 3 discusses the data collection,

measurement of data variables and the study’s research methodology. Chapter 4 separately

12

presents the descriptive and empirical results on the impact of NFI disclosures on firm risk for

the S&P 1200 companies and sensitive industries firms. Chapter 5 presents the evidence on

the impact of NFI disclosure regulations on the degree of NFI disclosures and the mediating

impact on the firm earnings risk and NFI disclosures relationship. Chapter 6 discusses the

policy and market implications of the findings and concludes the thesis. Finally, the study’s

limitations and future directions for NFI disclosures research are discussed.

13

Chapter 2

Literature Review

2.1. Introduction

Business organisations have long been recognised for their influential role in economic, social

and legal systems around the world. As a result, the activities and operations of business

entities come under scrutiny from different stakeholders such as shareholders, creditors,

government agencies and non-governmental organizations (NGOs). Additionally, because of

increasing stakeholder demands, media pressure and consumer awareness, firms now face

immense pressures to be more transparent, sustainable and ethical in their communications

regarding their business practices and operations. Therefore, to legitimise their actions and

meet the information needs of the stakeholders, businesses publish information related to

their policies, operations and business outcomes. Traditionally, the information content of

corporate communications has been investor and creditor centric; mostly financial. However,

because of increased awareness and concerns about social, environmental and governance

issues among stakeholders, firms have started to integrate NFI disclosures in their

communications. NFI disclosures contain information about business policies, processes,

work ethics, concern for employees, and contribution to environment and society, i.e., mostly

non-financial information.

Although businesses have started to make more NFI disclosures, these disclosures have been

mostly voluntary and unstandardized. Given the voluntary nature of non-financial

information and the lack of reporting standards, the content and presentation of such

information have been very diverse and inconsistent (Demir & Min, 2019; Muslu et al., 2019;

Sethi et al., 2017). However, there have been enormous developments such as establishment

NFI reporting frameworks, e.g., the global reporting initiative (GRI), the International

Integrated Reporting Council (IIRC), and the Sustainability Accounting Standards Board (SASB)

frameworks. The evolution of mandatory NFI reporting regulations in different countries has

changed and improved the quantity and quality of NFI disclosures over the years.

Given the broad nature of information covered under NFI disclosures, the literature on this

topic uses a variety of terms to characterize seemingly similar information. Therefore, terms

like corporate social responsibility (CSR), environmental, social and governance (ESG), and

14

sustainability disclosures are used interchangeably. Similarly, there are various reports that

firms use to report NFI disclosures such as CSR reports, sustainability reports, statement of

the code of ethics and conduct, the statement from senior decision-maker (CEO, President),

statement of corporate intent, corporate governance reports, and, more recently, integrated

reports. As NFI disclosures contain diverse information, more recently, data providers9,

practitioners and academics have started to divide NFI disclosures into three broad

categories: environmental, social and governance disclosures to empirically measure the

effect of NFI disclosures on the various factors of firm profitability and risk.

Although studies on NFI disclosures and their implications entail a very diverse, detailed

knowledge base in the finance and accounting literature, the following sections review the

literature that is most relevant to this study’s scope. The review starts from the theoretical

foundations that conceptualize the needs and consequences of NFI disclosures and follows

the historical development of definitional constructs and NFI reporting. The subsequent

sections present the literature regarding the impact of NFI disclosures on firm performance,

information asymmetry and firm risk. The last section discusses the literature on the

developments of NFI disclosure regulations and their impact and mediating effect on the

extent of disclosures, firm value and information asymmetry.

2.2. Theoretical Foundations

Given the far-reaching effect of NFI disclosures, there are several motivations for a firm to

disclose NFI in its corporate communications. Prior studies use several theories to address the

issue of the social responsibilities of firms and identify sources that require firms to be more

transparent in their business conduct and information disclosures. For example, using the

well-known agency cost framework (Borghei, Leung, & Guthrie, 2018; Garcia et al., 2017)

provides evidence of reduced information asymmetry for firms with improved ESG related

disclosures. These studies show NFI disclosures help firms resolve the information asymmetry

problem that has multiple adverse effects on the organization-stakeholder relationship. For

example, information asymmetry could result in lower investor confidence, mispricing of firm

value and higher cost of capital.

9 Bloomberg ESG database, MSCI ESG research, Refinitiv ESG score and others.

15

Studies also provide evidence of comparatively higher ESG related disclosures from

sensitive/controversial industry firms (Gamerschlag, Möller, & Verbeeten, 2011; Garcia et al.,

2017). These authors maintain that sensitive industry firms go through strict scrutiny by

stakeholders so they make higher ESG disclosures to maintain the legitimacy of their business

operation. The literature on NFI disclosures identifies three main theories that fundamentally

identify the motives and grounds for managers to disclose NFI. These theories are: 1) the

Agency Theory; 2) the Legitimacy Theory; and 3) the Stakeholder Theory.

2.2.1. Agency Theory

Jensen and Meckling (1976) define the agency relationship as “a contract under which one or

more persons (the principal(s)) engage another person (the agent) to perform some service

on their behalf which involves delegating some decision making authority to the agent. If both

parties to the relationship are utility maximisers there is a good reason to believe that the

agent will not always act in the best interests of the principal” (p. 5). They also introduce the

concept of agency cost as a combination of three elements: “1) the monitoring expenditures

by the principal; 2) the bonding expenditures by the agent; and 3) the residual loss” (p.5).

They define “residual loss” as the decrease in the market value of a firm arising from a poor

agency relationship between principals and agents. On the other hand, the conflict of interest

between principals and agents generates an information gap, because the agents, being in-

charge of the business, possess first-hand information about the firm and principals depend

on agents to get access to this information. Principals not only pay incentives to the agents to

align their objectives with them but also provide more incentives to keep the information gap

as small as possible (Shapiro, 2005). These foundational works on agency theory provide the

rationale for NFI disclosures by corporate managers to align principal-agent objectives and

reduce agency costs via a reduction in information asymmetry.

Many researchers have used agency theory to conceptualize the relationship between NFI

disclosures and information asymmetry, firm performance and risk. For example, Borghei et

al. (2018) maintain that corporate managers make extra efforts to disclose more NFI to

resolve agency problems arising from information asymmetry. They report a decrease in

information asymmetry following carbon disclosures for non-greenhouse gas registered

companies in Australia. Similarly, (Garcia et al., 2017; Ness & Mirza, 1991) find a reduction in

16

information asymmetry because of the adoption of integrated reporting. Jo and Harjoto

(2012) reject the “CSR over-investment” argument of agency theory since they find a positive

relationship between firms’ CSR and financial performance. They argue that disclosure of CSR

information plays an important role in increasing a firm’s financial performance. Further, ESG

disclosure transparency increases firm value through the information asymmetry and agency

cost reduction channel (Yu, Guo, & Luu, 2018). Looking from the lense of agency theory, Velte

(2020) report a two-way association between institutional shareholders and corporate ESG

disclosures.

Studies also report that NFI disclosures are more prevalent in economies that promote higher

levels of stakeholder protection. For example, Martínez‐Ferrero, Ruiz‐Cano, and García‐

Sánchez (2016) report that additional, voluntary NFI disclosures can help reduce agency

problems. They also suggest that environments with a higher social responsibility

commitment reduce agency conflict by increased voluntary disclosures. Conversely,

Kartadjumena and Rodgers (2019) report a negative impact of climate and environmental

reporting by managers on financial performance and market value of Indonesian banks. They

argue that banking firms in Indonesia are practising sustainability reporting as an altruistic

rather than a strategic motive. Recognizing the importance of NFI disclosures in risk

optimization for shareholders, this study uses agency theory concepts to ascertain a reduction

in the agency problem as a result of lower future earnings risk.

2.2.2. Legitimacy Theory

Legitimacy theory considers business organizations as part of a larger social system, hence

firms are required to abide by the social and ethical norms of the society where they operate.

A comprehensive, straightforward definition of legitimacy theory is given by (Kaplan &

Ruland, 1991) as “organizational legitimacy is a process, legitimation, by which an

organization seeks approval (or avoidance of sanction) from groups in society” (p. 370).

Legitimacy theory comprehends organizational efforts to be more transparent and provide

additional information as a way to legitimize firms’ existence in society and avoid adverse

outcomes of their operations and actions. Additionally, organizations pursue business

legitimacy objectives to abide by the social contract that they have with society. These social

contracts can be macro- or micro-social contracts as explained by the integrative social

17

contract theory. According to (Donaldson & Dunfee, 1999), macrosocial contracts are hyper

norms that are acceptable at the level of society as a whole like anti-bribery policies,

employees’ rights protection, and sustainable use of natural resources. Microsocial contracts

are specific to a particular society or culture like religious and cultural sensitivities. For

example, multinational organizations operating in Muslim countries cannot use an obscene

advertisement to promote their products. With macrosocial contracts, organizations make

more NFI disclosures as a result of regulations introduced by governments and international

standard organizations to maintain and enhance their legitimacy (Arena, Liong, & Vourvachis,

2018; Bernardi & Stark, 2018).

Similarly, firms operating in countries that promote higher levels of stakeholder protection,

make more environmental and social disclosures to meet the socially acceptable standards of

business conduct (Martínez‐Ferrero et al., 2016). Additionally, to establish legitimacy and

maintain a balance between corporate objectives, firms carefully trade-off between voluntary

and mandatory corporate disclosures. For instance, Fallan and Fallan (2019) maintain that

firms make a trade-off between corporate tax behaviour and environmental performance

disclosures. Although both are essential elements of NFI disclosures, they cater to the needs

of different stakeholders and have different reporting requirements. Hence, firms must

decide an optimum focus level on such elements to maintain legitimacy from the perspectives

of different stakeholders. These social contracts are influenced by organizational

characteristics. For example, large firms are considered more liable to practice socially

acceptable practices because of their more significant impact on society. Similarly, firms

working in the natural resources sector are considered more accountable to practice

sustainable business practices than other firms.

Prior studies report the influence of firms’ characteristics on their legitimacy seeking

behaviour. For example, Reverte (2009) reports firms’ media exposure, size and industry to

be the main determinants of their CSR disclosures. Hence, the author claims disclosures of

NFI as legitimacy seeking behaviour. Similarly, Khan, Muttakin, and Siddiqui (2013) find

corporate governance indicators, such as foreign ownership, board independence, and

presence of an audit committee, as major determinants of CSR disclosures that strengthen

the view of the corporate legitimacy theory. Firms operating in environmentally sensitive

18

and/or socially controversial industries face more scrutiny from society. Hence, they need to

make more significant NFI disclosures to maintain their legitimacy. Consequently, several

studies report evidence of higher NFI disclosures for sensitive industry firms compared with

other industry firms, e.g., (Gamerschlag et al., 2011; Garcia et al., 2017).

Overall, firms have to make careful choices when deciding the level of NFI disclosures because

the nature and importance of NFI disclosure elements differ significantly and proper

understanding of such differences is required to establish legitimacy. This study uses the

legitimacy concept to understand how firms use different dimensions of NFI disclosures to

address changing stakeholder information needs and regulatory pressures, which can

enhance organizational legitimacy measured by the transparency of the information

environment.

2.2.3. Stakeholder Theory

The seminal work Strategic Management: A Stakeholder Approach (Freeman, 1984) lays the

foundations of stakeholder theory. In more recent work, Freeman (2010) defines

stakeholders as: “any group or individual that can affect or be affected by the realization of

the organization’s purpose” (p. 26). According to his definition, stakeholders consist of a

variety of individuals and groups that might be internal to a firm like employees and

shareholders or external like suppliers, customers, competitors and governments. These

individuals and groups seek information related to firms’ processes and performance to make

informed decisions. Hence, firms disclose additional information to meet the informational

needs of their stakeholders and so reduce any potential information asymmetry between

different stakeholders (Garcia et al., 2017).

Though large amounts of NFI disclosures are voluntary; stakeholder theory provides the

perspecptive for firms’ voluntary disclosures. Mukherjee and Nuñez (2019) maintain that

firms make voluntary disclosures alongside mandatory disclosures to address the concerns of

the diverse group of stakeholders. Firms also make voluntary disclosures to meet the

demands of different pressure groups such as environmental and climate change

organizations. Stakeholder-oriented countries and regions increase stakeholder bargaining

power and force firms to make voluntary disclosures. As a result, studies report a higher level

19

of NFI in stakeholder-oriented countries than in their counterparts (García-Sánchez, Suárez-

Fernández, & Martínez-Ferrero, 2019; Landi & Sciarelli, 2019).

The stakeholder theory justifies increased information disclosures by corporation managers

(Ullmann, 1985). Ratnatunga & Jones (2012) describe the practicality of stakeholder theory

to justify the motivations of corporate managers to disclose additional information. Branco

and Rodrigues (2007) state that firms’ NFI (CSR) disclosures should be capable of addressing

firms’ contributions to satisfy the long-run needs of their stakeholders and convey the efforts

undertaken by a firm to minimise their social and environmental concerns. In other words, to

satisfy stakeholders’ information needs and address concerns about environmental and social

impacts of a firm, NFI disclosures are the most suitable approach that managers can use

(Huang & Kung, 2010).

A financial analyst is an important stakeholder of firms. Financial analysts use company

reported information with market information to forecast firms’ future earnings and market

price. Studies show NFI disclosures could improve (reduce) the accuracy (dispersion) of

analysts earnings forecasts (Bernardi & Stark, 2018; Dhaliwal, Radhakrishnan, Tsang, & Yang,

2012). Other studies also show analysts forecast accuracy is value relevant (Wei & Xue, 2015)

and cost-effective (Park & Park, 2019) for investors. This study uses stakeholder theory to test

whether higher NFI disclosures improve the accuracy of analysts’ earnings forecasts that

ultimately relate to lower earnings risk for firms. The precision of analysts earnings forecasts

for a firm may influence many other stakeholders such as investors, creditors and suppliers

since these stakeholders use analysts’ forecasts to make decisions about firms.

2.3. Definition Evolution of Non-financial Information

The idea of businesses being responsible towards society and the demands to include non-

financial information in corporate disclosures dates back more than a half-century when H. R.

Bowen (1953) introduced the concept of social responsibility (SR) in his book Social

Responsibilities of the Businessman. He proposed an initial definition of social responsibility

for businesses as: “It refers to the obligations of businessmen to pursue those policies, to

make those decisions, or to follow those lines of action which are desirable in terms of the

objectives and values of our society” (p. 6). Davis (1960) presents the managerial perspective

of socially responsible behaviour and argues that a socially responsible stance by a firm could

20

bear fruit in the long-run. Davis also argues that because of the possession of resources,

organizations have social power and, to avoid gradual erosion of this power, it should be

coupled with social responsibility (pp. 70-73).

Though the initial definition of CSR pointed towards the responsibility of businesses towards

society as a whole, Johnson (1971), for the first time, presented CSR from the stakeholder’s

perspective. He states that organizations should not only be concerned about the wealth

maximisation of their stockholders but also consider the interests of employees, suppliers,

dealers and the local community (p. 50). Sethi (1975) introduced the progressive concept of

corporate social performance (CSP) by linking CSR with a company’s strategy and actions. The

author argues that organizations must be socially responsive rather than just being merely

socially obliging. Taking things further, Carroll (1979) defines CSR as a process consisting of

four stages of CSR development: economic, legal, ethical, and philanthropic obligations.

In the early 1980s, Peter Drucker gave a new dimension to the CSR concept by emphasizing

the idea of compatibility between firms’ responsibilities and business opportunities (Drucker,

1984). Following Drucker’s work, several studies have focussed on the marketing aspect of

CSR, such as cause-related marketing, social sponsorship, environmental marketing,

communicating with consumers concerning CSR issues and corporate reputation (Barone,

Miyazaki, & Taylor, 2000; Brown & Dacin, 1997; Caruana & Crane, 2008; Crouch, 2006;

Handelman & Arnold, 1999; Varadarajan & Menon, 1988; Wagner, Lutz, & Weitz, 2009).

From the early 1980s onwards, there has been less focus on the development of a CSR

definition. Instead, researchers and practitioners started to focus more on the measurement

of CSR disclosures and how such disclosures can affect different aspects of firms’ business.

Wang (2015) describes the period 1980-1989 as the exploration stage of CSR development,

where researchers started to explore the content of CSR information reported in firms’

communications and ascertain the impact of such information on the financial performance

of the reporting firms.

2.3.1. Institutional contribution to the definition evolution of Non-financial

Information

Apart from individual researchers who contributed to the evolution of CSR definitions and

frameworks, various institutions have played a vital role in developing CSR frameworks for

21

the firms to understand their social responsibilities and understand how to incorporate such

information in their communications. In 1971, the Committee for Economic Development

(CED) presented a broader view of CSR based on the reasoning that the social contract

between business and society is changing very substantially. The CED framework for CSR

includes three circles of responsibility for business entities: the inner circle of responsibility

asks the firms to be responsible for their economics functions such as production, economic

growth, and shareholder wealth maximisation. The intermediate circle relates to social

responsibility towards the environment, employees’ relationships and the provision of

accurate information about products and processes to customers. Finally, the outer circle

covers the broader role of the firms, such as the well-being of society and improving the social

environment (CED, 1971). Manne and Wallich (1972) in a debate organized by the American

Enterprise Institute expressed the view that firms’ acts of responsibility should be considered

socially responsible only when it is done purely for the benefit of society or any stakeholder

group, and it is not bound by any legal obligations.

In the aftermath of Exxon Valdez oil spill10, environmental protection groups and socially

responsible investment (SRI) funds pushed for more NFI disclosures. Several social investment

professionals formed the Coalition for Environmentally Responsible Economies (CERES). In

1989, CERES put forth its “Valdez principles” for better management of environmental issues

and disclosure of information related to firms’ environmental and social impact. Continuing

their work, CERES and the Tellus Institute, with the support of United Nations Environment

Programme (UNEP), launched the Global Reporting Initiative (GRI) in 1997 to establish a

universal reporting framework that incorporates economic, governance and CSR information

in a single report framework (Gilbert, 2002). Major world organizations such as the United

Nations (UN) and the Organization for Economic Co-operation and Development (OECD)

extended their efforts to develop NFI disclosure frameworks to promote sustainable business

practices. The then UN president, Kofi Annan, launched the UN global compact initiative on

July 26, 2000, based on 10 principles divided into four major components: 1) Human rights,

2) Labour, 3) Environment, and 4) Corruption to promote responsible business practices. This

10 The Exxon Valdez oil spill was the worst oil spill in the history of U.S. It resulted in an oil spill of over 11

million gallons of crude oil when an Exxon shipping company owned oil tanker faced an accident near Tatitlek,

Alaska, on March 24, 1989. See https://www.history.com/topics/1980s/exxon-valdez-oil-spill

22

initiative is the largest global effort, with over 13000 corporate participants and other

stakeholders from around 170 countries. The OCED issued guidelines for multinational

enterprises that provided recommendations and standards for the promotion of responsible

business practices among multinational corporations based on a few basic principles11. More

recently, in 2011, the OECD guidelines were published and endorsed by both OCED and non-

OECD countries for firms operating in their legal territories.

In 2009, participants in Prince’s accounting for sustainability forum discussed the formation

of the International Integrated Reporting Council (IIRC), which was subsequently established

on 2 August 2010. After its formation, IIRC formulated a framework for reporting a wide range

of corporate information. According to IIRC’s reporting framework, integrated reporting (IR)

includes: i) strategic focus; ii) connectivity of information; iii) future orientation; iv)

responsiveness and stakeholder inclusiveness; and v) conciseness, reliability, and materiality

(IIRC, 2011). In 2013, IIRC presented its international IR framework, which is considered the

most detailed NFI reporting framework. Similarly, the Sustainability Accounting Standard

Board (SASB) founded in 2011 developed and circulated sustainability accounting standards.

Contrary to the other frameworks, SASB’s objective is to incorporate its standards into Form

10-K that is a compulsory filling requirement for companies registered with the U.S. Securities

and Exchange Commission.

2.4. Evolution of Non-financial Information Reporting Research

After the initial development of NFI disclosure concepts and definitions, interest in measuring

the extent of NFI disclosures practised by organizations started to develop in academic circles.

This led to the development of NFI reporting research using content analysis because this was

the most appropriate, unbiased method to measure the extent of NFI disclosures practised

by firms then. Later, with the development of NFI databases, the focus shifted to measuring

the impact of NFI disclosures on firms’ financial performance, market value, information

asymmetry and risk measures.

2.4.1. Initial studies on the measurement of NFI information disclosures

11 “Employment and industrial relations, human rights, environment, information disclosures, combating

bribery, consumer interests, science and technology, competition, and taxation”.

23

Preliminary study on NFI reporting in the 1970s by (Bowman & Haire, 1975) tried to explore

the extent of CSR related NFI reported in annual reports of food processing firms in the United

States (U.S.). They used search criteria such as “social responsibility”, “corporate citizenship”,

“social action” and “social responsiveness”. Similarly, using a list of six NFI categories:

community involvement, personnel, environment, equal opportunity, products and others,

developed by Ernst & Ernst (1978), Abbott and Monsen (1979) developed a social

involvement disclosures (SID) index for a sample of Fortune 500 companies to measure the

extent of NFI disclosed in their annual reports. (Cochran & Wood, 1984) used a reputation

index develop by (Moskowitz, 1972) as a proxy for NFI disclosures. It categorizes firms into

“outstanding”, “honourable mention” and “worst” categories depending upon their CSR

related disclosures. Although initial studies tried to measure the extent of NFI disclosures,

they lack a common framework for defining the scope and meaning of CSR activities. Sethi

(1975) suggests focussing on “stable classification” and “stable meaning” of CSR activities and

proposes a framework to operationalise the extent of CSR activities by introducing the

concept of CSP. Preston (1982), using Fortune 500 firms’ annual reports, proposes an

analytical format to summarize a variety of CSR disclosures into a more concise, meaningful

report. These studies made operationalisation of CSR information easier and motivated

researchers to pursue empirical research in the domain.

However, the preliminary studies on the relationship between firms’ CSR disclosures and

financial performance did not find a consistently significant association because of poor

operationalisation of CSR disclosures and deficiencies in available datasets (Ullmann, 1985),

model misspecifications (McGuire, Sundgren, & Schneeweis, 1988), behavioural and

perceptual measures used (Wokutch & McKinney, 1991), and the use of unidimensional

indices of CSR (Chen & Metcalf, 1980; McGuire et al., 1988; O'Bannon & Preston, 1993; Shane

& Spicer, 1983). These issues led to the development of NFI disclosure databases that provide

standardised, reliable measures of ESG related activities practised by the firms.

2.4.2. The development of improved NFI disclosure databases

24

In 1991, Kinder, Lydenberg and Domini (KLD) introduced their database that rates companies

on eight attributes12 of NFI disclosures. Based on these attributes, analysts assign

performance scores to each company. In 2010, KLD was acquired by MSCI and re-named it as

MSCI ESG research. According to Waddock and Graves (1997), the KLD database provided an

improved, detailed source of CSP. The Thomson Reuters ESG database provides data for over

5000 publicly listed companies using over 400 data points provided in corporate

communications. Another important database that provides ESG related data is the Refinitiv

ESG Scores that gathers data on ESG disclosures and ESG controversies13 to make combined

ESG scores for companies. This database covers over 6500 companies around the globe. Some

indices are developed by grouping companies into a benchmark index based on their NFI

disclosures. Some commonly used indices are MSCI KLD 400 Social Index (1990), Dow Jones

Sustainability index (1999), Calvert social index (2000) and FTSE4GOOD series.

These databases provide in-depth knowledge about ESG performance and disclosure scores

of sample companies; this study uses recent, more comprehensive databases that specifically

report ESG disclosure scores rather than performance scores (Eccles, Serafeim, & Krzus, 2011;

Tamimi & Sebastianelli, 2017). The Bloomberg ESG database provides data for over 11,500

companies across 70 countries. Bloomberg analysts construct ESG disclosures scores using

over 900 data field across a variety of NFI indicators, such as air quality, water and energy

management, human capital, waste management, compensation, diversity, and shareholders

rights. Bloomberg uses company reported data from CSR reports, annual reports, company

websites and other company sources to maintain accuracy and consistency in their

database14.

2.5. Empirical Studies on the Impact of NFI Disclosures

Given the availability of better measures of NFI disclosures, recent studies have started to

focus on the empirical domains in NFI disclosures research. Various strands of the NFI

disclosures literature provide evidence of the impact of NFI disclosures and CSP on firms’

12 The five attributes directly related to stakeholders are community relations, employee relations, performance

with respect to the environment, product characteristics, and treatment of women and minorities. Three

attributes related to the external environment are military contracting, participation in nuclear power and

involvement in South Africa. 13 There are 23 ESG controversies. Anti-competition, business ethics, and Intellectual property are examples of

ESG controversies. 14 https://data.bloomberglp.com/professional/sites/10/1148330431.pdf

25

financial performance, market value, cost of equity, information asymmetry and firm risk. Of

these strands, the most studied topic is the impact of a firm’s NFI disclosures on financial

performance, followed by information asymmetry and firm risk. Even though this study

focuses on NFI disclosures and the risk firm risk link, a brief review of other strands highlights

the importance of NFI disclosures.

2.5.1. NFI disclosures and the firm performance link

Studies on firms’ NFI disclosures or CSP and the financial performance link concentrate on the

causal relationship between a firm’s performance on ESG issues and its impact on financial

performance. Various researchers provide evidence of a significant positive relationship

between a firm’s CSP and accounting-based financial performance measures (Chopra & Wu,

2016; Greening & Turban, 2000; Kassinis & Soteriou, 2003; Russo & Fouts, 1997) and market

based financial performance (Bauer & Hann, 2010; Graham & Maher, 2006; Graham, Maher,

& Northcut, 2001; Konar & Cohen, 2001; Thomas, 2001). Rettab, Brik, & Mellahi (2009) report

CSR to be positively associated with three elements of firm performance: financial

performance, corporate reputation and employee commitment. Frank and Obloj (2014) argue

that a firm’s commitment to employee-related issues like a safe working environment, health

facilities, employee empowerment and equal opportunities, result in enhanced firm

productivity. Similarly, CSR engagement and its disclosures support attaining increased

employee commitment and a higher level of legitimacy from the community and government

(Nguyen & Nguyen, 2015). Yoon and Chung (2018) report external CSR practices are positively

related to market share and internal CSR practices help to improve the operating profits of

firms. Thus, employment of CSR in organizational processes improves sustainability in

organizational performance that enhances firms’ effectiveness.

2.5.2. The NFI and information asymmetry link

There is a literature strand that focuses on the role of NFI disclosures as a tool that may reduce

the opaqueness of a firm’s information environment. Kim and Verrecchia (1994) define

information asymmetry as “a trading situation where some market participants have better

access to relevant information than others”. According to Diamond and Verrecchia (1991) and

Lambert, Leuz, and Verrecchia (2007), a lack of publicly available information creates an

incentive for some investors to obtain privately held information, which results in information

26

asymmetry among investors. Thus, transparency in information disclosures is a mechanism

that reduces the information gap between market participants and between managers and

stakeholders. As a result, investors can make accurate investment decisions and stakeholders

are protected from exploitation by managers’ conflicting incentives (Martínez-Ferrero,

Rodríguez-Ariza, García-Sánchez, & Cuadrado-Ballesteros, 2017). Moreover, corporate

managers may disclose more information to investors to increase favourable assessment of

their firms when the overall information environment suffers from asymmetric information

(Elliott & Jacobson, 1994). Studies exploring the link between NFI disclosures and information

asymmetry use different proxies for information asymmetry, such as bid-ask spread (Cho, Lee,

& Pfeiffer Jr, 2013; Cui et al., 2018; Michaels & Grüning, 2017), analysts forecasts dispersion

(Martínez-Ferrero et al., 2017; Michaels & Grüning, 2017), share price volatility and liquidity

(Cormier, Ledoux, & Magnan, 2011; Xu & Liu, 2018).

Most studies in this strand provide evidence of a significant negative relationship between

firms’ NFI disclosures and proxies of information asymmetry. For example, using KLD

performance scores, Cho et al. (2013) provide evidence of a significant reduction in

information asymmetry through a company’s KLD performance scores. They find both

positive and negative CSR performance scores reduce the bid-ask spread and the greater

presence of institutional investors attenuates this link. Nguyen, Agbola, and Choi (2016) use

ASSET415 CSR performance scores to proxy for Australian firms’ CSR performance and report

a negative relationship between CSR performance scores and the bid-ask spread. Their

findings suggest that firm size and market power strengthen the CSR-information asymmetry

link. Similarly, but using analyst forecast dispersion and transaction cost as a proxy for

information asymmetry, Cui et al. (2018) report a negative association between a firm’s CSR

performance scores and information asymmetry. They also report the information

asymmetry and CSR negative relationship is strengthened in high-risk industries. (Xu & Liu,

2018) report a negative relationship between social and environmental disclosures and a

firm’s information asymmetry. Benjamin, Regasa, Wellalage, and M Marathamuthu (2020)

and Lu, Shailer, and Yu (2017) report positive influence of voluntary environment disclosure

15 ASSET4 is a Thomson Reuters ESG database that provide CSR information in four broad pillars: economic,

environmental, social and corporate governance performance. It uses over 750 data points to calculate

performance scores.

27

on corporate cash holding in firms opearating in sensitive industries and opaque

environment. They argue that environmental disclosures mitigate negative effects of

corporate cash holdings, which lowers information asymmetry problem. Differing from above

mentioned findings, Martínez‐Ferrero et al. (2016) emphasize a causal link between

information asymmetry and NFI reporting because there is evidence of reverse causality

(Cormier et al., 2011).

2.5.3. NFI disclosures and the firm risk link

In addition to the information asymmetry domain of CSP research, studies have explored the

link between a firm’s CSP performance and risk. Essentially, the reasoning behind the CSP,

firm risk link relates to NFI disclosures that strengthen a firm’s moral capital and goodwill that,

in return, provide an “insurance-like” shelter to preserve financial performance (Godfrey,

2005; Godfrey, Merrill, & Hansen, 2009). Similarly, Benlemlih, Shaukat, Qiu, and Trojanowski

(2018) suggest firms promote corporate transparency by making broad, objective social and

environmental disclosures that, in turn, reduce their idiosyncratic and total risk. Firms that

actively invest in CSR initiatives build moral and social capital that results in increased trust in

stakeholders and shareholders. Such firms perform better in terms of market returns and

growth in times of market downturn when the corporate environment generally lacks the

trust of market participants (Lins, Servaes, & Tamayo, 2017).

Firm risk is a broad concept that entails different concepts; therefore, different proxies are

used in the literature to measure firm risk. Previous studies used various measures of firm risk

such as idiosyncratic risk, systematic risk, total risk and default risk, to explore NFI disclosures

and firm risk association. The most used proxies for systematic risk are CAPM beta and Fama-

French four-factor model beta. The commonly used proxies for total risk and idiosyncratic risk

are the standard deviation of daily stock returns and standard deviation of residuals from the

Fama-French four-factor model, respectively. Using a meta-analysis of previous studies,

Orlitzky and Benjamin (2001) report evidence of a significant negative relationship between

a firm’s CSP performance and risk. Because of the use of meta-analysis, their study provides

measurable, generalised and reliable proof of the CSP-firm risk link that was missing in the

literature then. According to their findings, the negative relationship between a firm’s CSP

and market risk is more robust than accounting risk. They subdivided market risk into total

risk (return volatility of stock returns – diversifiable risk) and systematic risk (market beta –

28

non-diversifiable risk) and report that CSP has a substantial effect on diversifiable risk and a

moderate effect on non-diversifiable risk.

Oikonomou, Brooks, and Pavelin (2012) explore the CSP – systematic risk relationship using a

longitudinal study of S&P 500 firms. According to their findings, the CSP-firm risk link is

stronger for firms that better address social and environmental concerns than firms that

possess strengths related to social and environmental factors. Their findings suggest market

participants assign more weight to the management of social and environmental concerns

than possession of strengths related to social and environmental factors. Similarly, using a

longitudinal study of S&P500 firms, Chang, Kim, and Li (2014) report a heterogeneous impact

of institutional CSR activities that target secondary stakeholders and technical CSR activities

that target primary stakeholders. According to their findings, only CSR strengths that target

secondary stakeholders are negatively related with firm risk; strengths that target primary

stakeholders are positively associated only with a firm’s financial performance. Another study

on S&P500 firms reveals that the impact of CSR concerns on firm risk is stronger than CSR

strengths. They also maintain that CSR engagement may not affect the current performance

of the firm, but it may adversely affect future performance (Nguyen & A. Nguyen, 2015).

These findings show firms’ engagement in CSR activities is not just window-dressing; it helps

reduce their risk. The results of such a relationship are more pronounced for firms in

environmentally sensitive industries (Jo & Na, 2012).

More recently, academics and practitioners have started using the concept of ESG

performance/disclosures in place of CSP because of the extensive information included in

firms’ NFI disclosures. Moreover, the availability of improved, detailed databases, such as

ASSET4 and Bloomberg ESG disclosures scores, have made access to NFI disclosures easier.

These databases use numerous data points16 to form firms’ ESG scores and rank each ESG

factor on a scale ranging from 0 to 100. These databases are superior to other databases

because they provide a quantitative score that ranges between 0-100, which makes

quantification easier, whereas other databases such as KLD use a limited score range (-2 to

2). These databases provide a way to explore the disaggregated impact of each ESG factor,

16 This score is based on 100 of 219 raw data points that Bloomberg collects, and is weighted to emphasize the

most commonly disclosed data fields.

29

namely, environmental, social and governance. For instance, using ASSET4 ESG performance

scores, Sassen, Hinze, and Hardeck (2016) report the disaggregated effect of ESG factors on

total, idiosyncratic and systematic risks. According to their results, the social component of

ESG performance has a significant negative effect on all three proxies of risk, but only the

environmental component negatively affects idiosyncratic risk. They did not find any effect of

the governance factor on firm risk. Similarly, using environmental and social disclosures

scores from Bloomberg ESG database, Benlemlih et al. (2018) report a significant negative

effect of social and environmental disclosures on a firm’s total and idiosyncratic risk.

However, these disclosure scores do not show any effect on a firm’s systematic risk, a finding

that is consistent with prior studies. These studies show that disaggregated disclosures show

varying effects on firms’ risk measures. Thus, it is essential to use a data source that provides

the flexibility of using the individual dimensions of ESG disclosures.

Using website traffic data as a measure of adequate information disclosures and investor

recognition, Chien and Lu (2015) suggest higher website traffic reduces information

asymmetry and, as a result, reduces firm risk. They report a significant negative relationship

between website traffic and firm risk measured by analysts’ forecast dispersion, cost of equity

and return volatility. Similarly, Kothari et al. (2009) report a significant negative relationship

between firm risk measures (analysts’ forecast dispersion, stock return volatility, and cost of

capital) and favourable disclosures reported by a corporation, analysts and the business press.

They also report an increase in firm risk following unfavourable disclosures.

In terms of risk measures, the literature on the relationship of NFI and firm risk commonly

focusses on traditional firm risk proxies such as total risk (measured as the volatility of stock

returns), idiosyncratic risk and systemic risk. However, another risk measure that is more

relevant to a firm’s information environment and captures risk associated with information

disclosures more profoundly than traditional risk proxies, is less investigated in the existing

literature. This measure is commonly referred to as analysts’ forecast dispersion. It is used

interchangeably as a proxy for information asymmetry and earnings uncertainty. However,

Barron et al. (2009) reconcile the contradictory evidence of (Diether, Malloy, & Scherbina,

2002; Johnson, 2004) of analysts’ forecast dispersion being a measure of uncertainty or

30

information asymmetry. They maintain that the dispersion in analysts’ forecast represents

uncertainty, which is negatively associate with firms’ future stock returns.

Based on these explanations and supporting evidence from the literature, we use analysts’

earnings forecast dispersion as a measure of firms’ future earning uncertainty and propose it

as a better risk measure than traditional risk proxies in the context of NFI disclosures and firm

risk link. We hypothesize the association between NFI disclosures and firms’ earnings risk as:

H1: NFI disclosure scores are negatively associated with firms’ future earnings risk.

Disaggregated NFI disclosure factors can affect financial analysts’ earnings forecasts

differently. To account for the impact of individual ESG components, we propose three

supplementary testable hypotheses:

Hypothesis 1a. Environmental disclosure scores are negatively associated with firms’ future

earnings risk.

Hypothesis 1b. Social disclosure scores are negatively associated with firms’ future earnings

risk.

Hypothesis 1c. Governance disclosure scores are negatively associated with firms’ future

earnings risk.

2.5.4. The linkage between NFI disclosures and firms’ earnings risk (the case of

sensitive industry firms)

The corporate finance literature, in general, and the NFI disclosure literature, in particular,

pay special attention to firms operating in environmentally or socially sensitive industries.

Firms that operate in the extractive and energy sectors are categorised as environmentally

sensitive and firms that operate in alcohol, tobacco, gambling, weapon production, and adult

entertainment are classified as socially sensitive firms. These firms are inherently prone to

higher stakeholder pressures because of the sensitive nature of their business operations. As

a result, these firms are required to report higher NFI disclosures than firms operating in non-

sensitive sectors. The literature provides some evidence regarding the impact of being in a

sensitive industry on a firms’ disclosure practices. For example, using a sample of sensitive

industry firms from BRICS countries, Garcia et al. (2017) report higher environmental

performance of sensitive industry firms than non-sensitive counterparts.

31

Previous studies also show a relationship between sensitive industry firms’ NFI disclosures

and their risk profile. For instance, using a sample of European firms, Sassen et al. (2016)

report a negative relationship between environmental performance and firm risk factors such

as idiosyncratic, systemic and total risks for environmentally sensitive firms. Similarly, using a

sample of U.S. based socially sensitive firms, Jo and Na (2012) report that CSR engagement

reduces firm risk more profoundly for socially sensitive firms than for non-sensitive

counterparts. These studies show that the sensitive nature of a firm’s operation could impact

its NFI disclosures and affect the relationship between NFI disclosures and earnings risk. As a

result, we hypothesise that the NFI disclosures of sensitive industry firms differ from non-

sensitive firms because sensitive industry firms face greater scrutiny from stakeholders

because of the nature of their operation. Thus, we hypothesise the following relationship:

H2: Sensitive industry firms report higher ESG disclosures than non-sensitive firms.

Similarly, we hypothesise that the disaggregated disclosures of sensitive industry firms are

higher than for non-sensitive firms. Thus we hypothesise the following relationships:

Hypothesis 2a. Sensitive industry firms report higher environmental disclosures than non-

sensitive firms.

Hypothesis 2b. Sensitive industry firms report higher social disclosures than non-sensitive

firms.

Hypothesis 2c. Sensitive industry firms report higher governance disclosures than non-

sensitive firms.

Moreover, based on the evidence reported by (Jo & Na, 2012) and (Sassen et al., 2016), we

hypothesise that the negative relationship between NFI disclosures and earnings risk will be

more pronounced for sensitive industry firms than for other industries. To test this

relationship, we hypothesise that:

H3: NFI disclosure scores are negatively related to firms’ future earnings risk for sensitive

industry firms.

Disaggregated NFI disclosures can affect firms’ earnings risk differently. To account for the

individual impact of ESG components, we propose three supplementary testable hypotheses.

32

Hypothesis 3a. Environmental disclosure scores are negatively associated with firms’ future

earnings risk for sensitive industry firms.

Hypothesis 3b. Social disclosure scores are negatively associated with firms’ future earnings

risk for sensitive industry firms.

Hypothesis 3c. Governance disclosure scores are negatively associated with firms’ future

earnings risk for sensitive industry firms.

2.6. NFI Disclosures Regulations

NFI reporting has been mainly voluntary, but some countries/regions have developed NFI

reporting frameworks or enforced international NFI reporting frameworks for firms operating

within their legal system. Many countries mandated CSR reporting in the past, but the most

recent regulations on NFI reporting come from Malaysia, China, Denmark, South Africa and

the European Union (EU). In Malaysia, CSR reporting has been made a listing requirement by

the stock exchange Bursa Malaysia for all firms from December 2017. Under the

requirements, companies are required to disclose their CSR activities and explain any failure

to comply with CSR reporting requirements. The Shanghai stock Exchange (SHSE) and the

Shenzhen Stock Exchange (SZSE) directed particular17 companies to disclose NFI by the end of

the financial year 2008. In the same year, state council commissions issued the opinion

guidelines on CSR reporting for state-owned corporations after the revision of Article 5 of

company law.

In Denmark, large companies18 were mandated to disclose their CSR activities in a

supplementary report or explain the absence of any CSR activity in the management review

report. This regulation was enacted by an amendment to the Danish Financial Statements Act

in October, 2008, and came into force on 1 January, 2009. The Johannesburg Stock Exchange

(JSE) mandated CSR reporting for listed companies following the issuance of the King III report

on corporate governance. JSE mandated integrated reporting from the start of the 2011

financial year. Companies in both those countries were mandated to disclose their ESG

17 SHSE mandated CSR reporting for companies from the financial sector, companies with cross country listing

and companies included in SHSE corporate governance index. SHSE mandated a disclosures requirement for

companies that are part of Shenzhen 100 index. 18 In the Danish Financial Statements Act, a company is considered large if it satisfies any two of: 1) total assets

above 143 Million Danish marks; 2) net revenue over 286 Danish marks; or 3) average 250 or more full-time

employees.

33

policies and report on the achievement of the objectives set in the policies. Although the

regulations mandated CSR reporting in China, Malaysia, South Africa and Denmark, they lack

a comprehensive framework of metrics for CSR reporting (Ioannou & Serafeim, 2017).

2.6.1. NFI disclosures regulations in Europe Directive 2014/95/EU

On 22 October, 2014, the European Parliament enacted its directive 2014/95/EU that

mandates disclosure of non-financial and diversity information for large19 entities and

business groups. This directive amended Directive 2013/34/EU and Regulation (EC) No

562/2006 that were in place for NFI disclosures in the EU. This new directive requires firms to

report information on six broad topics: 1) environmental, 2) social, 3) employee matters, 4)

respect for human rights, 5) anti-corruption, and 6) bribery matters as minimum

requirements of NFI disclosures. Firms are also required to report how their business model,

policies, policy outcomes, potential risks and key performance indicators (KPIs) synchronise

with the six broad topics of NFI disclosures. Like other NFI reporting frameworks, this

framework does not provide any NFI performance metrics. However, the six major topics

cover almost all aspects of ESG related activities. Moreover, recital 720 of directive 2014/EU/95

provides details of potential information to be included under each topic. This makes NFI

reporting under the directive more standardized and comparable than previous NFI reporting

regulations. Thus, this EU directive has the most comprehensive regulations for NFI

disclosures and is expected to bring major changes in policies and disclosures related to the

CSR activities of firms operating in the EU. Given the comprehensive nature of directive

2014/95/EU, this study uses the directive as a case study of NFI disclosures regulations.

Recent studies that investigate the implementation of the EU directive in different member

states indicate that different European countries have transposed the EU directive into their

reporting regulations. For example, Testarmata, Ciaburri, Fortuna, and Sergiacomi (2020)

report that the leading European countries, France, Germany, Italy and Spain, have

transposed the EU directive into their reporting regulations. They also maintain that NFI

reporting regulations have increased the uniformity, transparency and comparability of

information disclosed by companies (p. 67). Matuszak and Różańska (2017) report that a

19 Companies with total balance sheet, net turnover and average number of employees above EUR 400,000, EUR

800,000 and 50, respectively. 20 See recital 7 of directive 2014/95/EU.

34

relatively smaller country, Poland, has also transposed the EU directive into its reporting

regulations. They suggest that the transposition of the EU directive into Poland’s reporting

regulations will increase the quantity and quality of NFI disclosures. These studies show that

the recently enacted EU directive is receiving widespread acceptance from EU member states

and it is expected to have substantial effects on the NFI disclosure practices of the firms

operating in the EU. Moreover, Korca and Costa (2021) conducting a review of the research

on the the EU directive suggest further investigation should explore the impact of the

directive on non-financial disclorsures and firm performance. Therefore, investigating the

impact of mandatory NFI disclosures in the context of the EU directive could provide

important insights regarding the acceptability of mandatory NFI disclosure requirements and

its impact of NFI disclosures.

2.6.2. The consequences of the NFI disclosure regulations

Empirical evidence on the impact of mandatory CSR reporting on the quantity of NFI

disclosures is sparse. Using a worldwide sample of 58 countries, (Ioannou & Serafeim, 2011)

report a significant positive effect of CSR reporting regulations on the amount of NFI

information reporting measured by different country-specific measurements such as social

responsibility, sustainable development, employee training, corporate board and ethical

practices. According to their findings, regulation effects are more substantial for countries

with better law enforcement systems, better mechanisms of CSR reporting assurance and

higher per capita GDP. Continuing their work using a better source of CSR measurement,

Ioannou & Serafeim (2017) report increased ESG disclosures following the enforcement of

mandatory NFI disclosure regulations in China, Denmark, Malaysia and South Africa.

Given the importance of NFI disclosure legislation (directive 2014/95/EU), we propose using

the regulation as an exogenous shock for NFI disclosures of European firms. We expect NFI

disclosures to be higher after the implementation of the directive. To test the effect of

legislative change on NFI disclosures, we propose the following hypotheses.

Hypothesis 4. Ceteris paribus, there is a significant difference between NFI disclosure scores

of European firms before and after the enforcement of directive 2014/95/EU.

To test the impact of the legislative change on the different components of NFI disclosure

scores, we propose the following auxiliary hypotheses.

35

Hypothesis 4a. Ceteris paribus, there is a significant difference between environmental

disclosure scores of the European firms before and after the enforcement of directive

2014/95/EU.

Hypothesis 4b. Ceteris paribus, there is a significant difference between governance disclosure

scores of the European firms before and after the enforcement of directive

2014/95/EU.

Hypothesis 4c. Ceteris paribus, there is a significant difference between social disclosure

scores of the European firms before and after the enforcement of directive

2014/95/EU.

Studies also report a favourable impact of NFI disclosures in a mandatory NFI reporting regime

on firm value and information asymmetry. For example, Ioannou and Serafeim (2017) report

increased firm value for firms that increase their disclosures post-regulation. Similarly, under

South Africa’s Integrated Reporting Regulation, Bernardi and Stark (2018) report an increased

level of ESG disclosures that, in turn, results in increased analysts’ forecast accuracy. These

studies show NFI regulations not only increase the level of CSR disclosures, but they also

moderate the CSR disclosures and information asymmetry, and CSR performance and firm

value links. These findings further strengthen the views of those who support regulations for

NFI reporting.

Given the empirical evidence concerning the favourable impact of NFI disclosures after

enforcement of mandatory regulations, this study proposes to measure the difference

between the pre- and post-regulation association between NFI disclosures and firm risk. We

propose the following hypotheses:

Hypothesis 5. There is a significant difference between the association of NFI disclosures and

firms’ future earnings risk before and after the enforcement of directive 2014/95/EU.

Hypothesis 5a. There is a significant difference between the association of environment

disclosure scores and firms’ future earnings risk before and after the enforcement of

directive 2014/95/EU.

Hypothesis 5b. There is a significant difference between the association of governance

disclosure scores and firms’ future earnings risk before and after the enforcement of

directive 2014/95/EU.

36

Hypothesis 5c. There is a significant difference between social disclosure scores and firms’

future earnings risk before and after the enforcement of directive 2014/95/EU.

2.7. Chapter Summary

This chapter briefly reviews the different strands in NFI disclosures literature that are most

relevant to this study. Starting with the theoretical foundations of NFI disclosures research,

we briefly review the theories that provide a conceptual understanding and reasoning for NFI

disclosures. Agency theory that focuses on the investor-manager relationship as agent and

principal maintains that managers should provide all relevant information to investors so they

can make informed investment decisions. Studies on the association between NFI disclosure

and information asymmetry report contradictory evidence since some researchers maintain

increased NFI disclosures reduce agency problems (Garcia et al., 2017; Martínez‐Ferrero et

al., 2016; Ness & Mirza, 1991) and others report the converse (Kartadjumena & Rodgers,

2019). Studies on the application of stakeholder theory in NFI literature provide the reasoning

for voluntary, improved NFI disclosures by corporate managers. According to these studies,

corporate managers disclose NFI to address investor concerns (Branco & Rodrigues, 2007;

Huang & Kung, 2010; Mukherjee & Nuñez, 2019), reduce information asymmetry (Garcia et

al., 2017), and maintain legitimacy in a high stakeholder protection environment (García-

Sánchez et al., 2019; Landi & Sciarelli, 2019). Studies use legitimacy theory to explain the

motivations of a firm’s NFI disclosures from the perspective of legitimizing business activities

(Dhaliwal, Li, Tsang, & Yang, 2014; Gamerschlag et al., 2011; Khan et al., 2013), and comply

with disclosures regulations (Arena et al., 2018).

The chapter also presents the pioneering work of Bowen (1953), who presented the first

definition of CSR, followed by Davis (1960) and Johnson (1971). These academics presented

two contrasting CSR views from the managerial and stakeholder perspectives, respectively.

Later, Drucker (1984) presented the compatibility concept between corporate objectives and

CSR activities that was subsequently applied by many researchers to provide evidence of such

compatibility. Together with academics, different institutions contributed to the evolution of

the CSR definition and framework. At first, the CED presented three-tier CSR framework that

covers the economic, social and environmental aspects of organizational responsibility. Later,

in the aftermath of Exxon Valdez oil spill, CERES (a group of investment professionals) put

37

forth the “Valdez Principles” that require better management and disclosures of

environmental issues by organizations. Subsequently, CERES together with UNEP, launched

GRI to develop a unanimous NFI reporting framework. In 2013, IIRC presented its

international integrated reporting framework that provides organizations the guidelines to

report NFI together with financial information in an integrated manner.

This chapter also provides a review of empirical studies. The empirical studies on the impact

of NFI disclosures on firm performance report a positive impact on accounting-based (Chopra

& Wu, 2016; Frank & Obloj, 2014; Greening & Turban, 2000; Kassinis & Soteriou, 2003; Russo

& Fouts, 1997) and market-based (Bauer & Hann, 2010; Graham & Maher, 2006; Graham et

al., 2001; Konar & Cohen, 2001; Thomas, 2001; Yoon & Chung, 2018) financial performance

measures. The empirical studies report a significant negative impact of NFI disclosures on

different measures of information asymmetry like bid-ask spread (Cho et al., 2013; Cui et al.,

2018; Michaels & Grüning, 2017), analysts’ forecast dispersion (Martínez-Ferrero et al., 2017;

Michaels & Grüning, 2017), and share price volatility and liquidity (Cormier et al., 2011; Xu &

Liu, 2018). Evidence of a negative relationship between NFI disclosures and firm risk

measured by stock return volatility (Benlemlih et al., 2018; Sassen et al., 2016), systematic

risk (Oikonomou et al., 2012; Orlitzky & Benjamin, 2001) and analyst forecast dispersion

(Chien & Lu, 2015; Kothari et al., 2009) was reviewed.

Finally, this chapter reviews developments in NFI reporting regulations and their impact on

the level of NFI disclosures in different countries. In the recent past, stock markets in South

Africa, China, Malaysia and Denmark have mandated NFI for listed companies. More recently,

the EU enacted directive 2014/95/EU that mandates NFI disclosures for companies operating

in member states. The reviewed studies show the evidence of an increase in the quantity of

NFI disclosures (Ioannou & Serafeim, 2011, 2017) that leads to an increase (decrease) in firm

value (information asymmetry) (Bernardi & Stark, 2018; Ioannou & Serafeim, 2017).

38

Chapter 3

Sample, Data and Methodology

3.1. Introduction

This chapter presents the sample, data and methodology of the study. The chapter first

introduces the sample and data variables used in the study. Next, we present the basic

diagnostics tests and empirical models used in the study to check the validity of the data and

test the research hypotheses, respectively. The chapter concludes with tests on the

robustness of estimates, potential robustness and endogeneity control measures.

3.2. Sample

The study sample comprises firms from the S&P Global 1200 index. That index (hereafter the

global index) encompasses seven dominant regional indices that cover companies from all

over the world21. The sub-indices are: S&P 500, S&P 350, S&P TOPIX 150, S&P/TSX 60, S&P

Asia 50, and S&P Latin America 40, which include companies from the US, Europe, Japan,

Canada, Australia, four Asian countries (Hong Kong, Korea, Singapore and Taiwan) and five

Latin American countries (Brazil, Chile, Colombia, Mexico and Peru), respectively. The sample

period for this study is 2008-2018. The length of the study’s sample period is limited by the

fact that Bloomberg started publishing ESG disclosure scores from 2008 after it acquired the

U.K. based New Energy Finance, which reported news and data on carbon and clean energy

markets, and the fact that, at the time of data collection, ESG data were not available beyond

2018.

We select the global index companies as our sample firms for various reasons. First, the

companies included in the global index genuinely embody a global sample that represents

70% of the total global market capitalisation. It is often used as a proxy of world stock market

performance (Nanayakkara & Colombage, 2019; Shamsuddin & Kim, 2010) and is an

internationally diversified portfolio of firms (Cosset, Somé, & Valéry, 2016; Hassan &

Giorgioni, 2019; Hassan, 2018). The global index offers sufficient representation of each world

region relative to its size in the international equity market (Del Bosco & Misani, 2016). Hence,

the results of this study can be generalised more confidently than other studies that focus on

21 See https://us.spindices.com/indices/equity/sp-global-1200

39

companies incorporated in a specific country (Becchetti, Ciciretti, & Hasan, 2015; Bernardi &

Stark, 2018; Cai, Cui, & Jo, 2016; Chang et al., 2014), a group of developed countries

(Benlemlih & Girerd‐Potin, 2017), or a geographical region (Sassen et al., 2016). Second,

global index companies represent a random sample because we do not select firms that

already make a higher or lower NFI disclosures, such as the Dow Jones Sustainability Index

(DJSI) World22 companies.

Table 3. 1. Index and industry-wide companies that comprise the study sample

Note: Table 1 presents the index and industry-wise details of companies included in the final sample.

Table 3.1 presents the details of the final sample companies in each index and industry in the

final sample. Financial and industrial sector companies are the two largest groups in our

sample, followed by consumer discretionary and materials. Since this study collects data from

two different databases, several global index companies were not included in the final sample

because of non-availability of data for those companies in both databases. This study uses the

Institutional Brokers Estimate System (IBES) to extract analysts’ earnings per share (EPS)

forecasts and the standard deviation of the EPS forecasts. IBES is a commonly used database

in NFI disclosure literature, e.g., see (Bernardi & Stark, 2018; Dhaliwal et al., 2012; Li & Wu,

2014). This study collects NFI disclosures data from the Bloomberg ESG database, which is an

established source of NFI disclosures in the, e.g., see (Buchanan, Cao, & Chen, 2018; Nollet,

Filis, & Mitrokostas, 2016).

22 The DJSI World is composed of world leading sustainability companies created by RobecoSAM. See

https://eu.spindices.com/indices/equity/dow-jones-sustainability-world-index

Industry AS 50 AU 50 CA 60 EU 350 JP 150 LA 40 US 500 Grand Total

Communication Services 4 3 5 30 7 2 11 62

Consumer Discretionary 4 3 4 45 18 2 37 113

Consumer Staples 2 1 3 27 12 6 26 77

Energy 3 4 13 12 1 3 18 54

Financials 12 10 9 66 11 4 50 162

Health Care 0 1 2 16 6 0 36 61

Industrials 6 6 5 56 39 4 44 160

Information Technology 8 0 1 13 16 0 40 78

Materials 5 11 10 35 18 7 14 100

MISC 0 5 6 21 12 6 1 51

Real Estate 2 5 0 5 4 0 13 29

Utilities 3 1 2 17 6 6 22 57

Grand Total 49 50 60 343 150 40 312 1004

40

3.3. Dependent and Independent Variables

This study uses analysts earnings forecasts dispersion as proxy for earnings risk as the primary

dependent variable. We use total and disaggregated ESG disclosure scores as dependent

variables for supplementary analysis. However, the overall and disaggregated ESG disclosure

scores are also the leading independent variables in this study. NFI disclosure regulations and

the sensitive industry status of a company are supplementary independent variables. Based

on previous studies, we control for several firm-specific variables. The next sections present

detailed explanations of all data variables used in the study.

3.3.1. Analysts’ earnings forecasts dispersion – earnings risk relationship

Analysts’ earnings forecasts dispersion is the key dependent variable used in this study to

proxy for a firm's future earnings risk. Barron et al. (2009) and Li and Wu (2014) define

analysts’ earnings forecast dispersion as the standard deviation of analysts’ earnings forecast

scaled by (consensus analysts forecast/(actual earnings per share). Although researchers

traditionally use analysts earnings forecasts error and dispersion as proxies for information

asymmetry (Cui et al., 2018; Martínez-Ferrero, Rodríguez-Ariza, García-Sánchez, & Cuadrado-

Ballesteros, 2018; Yang, Cheng, Sun, & Lu, 2019), many researchers also use it as a measure

of firm risk. For example, Gregoriou, Ioannidis, and Skerratt (2005) report analysts’ forecast

dispersion to be a significant factor in determining a firm's bid-ask spread alongside other

measures of firm risk. Similarly, Chien and Lu (2015) report a significant relationship between

firms' NFI disclosures and analysts forecasts dispersion alongside other proxies of risk such as

cost of equity and the volatility of stock returns. Notably, Barron et al. (2009) reconcile the

contradictory evidence of (Diether et al., 2002) and (Johnson, 2004) on analysts’ forecasts

dispersion as a measure of uncertainty or information asymmetry and maintain that the level

of dispersion in analysts forecasts represents uncertainty on future earnings, which is

negatively associated with a firm's future stock returns. They further argue that changes in

analysts’ forecasts dispersion reflect information asymmetry not the level of analysts’

forecasts dispersion.

This study adopts Barron et al. (2009) view that the level of dispersion in analysts’ earnings

forecasts represents uncertainty in future earnings hence serves as a proxy of a firm's future

earnings risk. Following Barron et al. (2009), a firm's earnings risk is calculated by:

41

𝐹𝑖𝑟𝑚′𝑠 𝑒𝑎𝑟𝑛𝑖𝑛𝑔𝑠 𝑟𝑖𝑠𝑘 = (1 − 1

𝑛) ∗ 𝐷 + 𝑆𝐸 (3.1)

where: a firm's earning risk equals the uncertainty part of analysts’ earnings forecasts

dispersion; D = forecast dispersion — i.e., the sample variance of the individual forecasts (𝐹𝐶𝑖)

around the mean forecast (𝐹𝐶̅̅̅̅ ); SE = squared error in the mean forecast measured as the

difference between earnings per share and the mean forecast, (𝐸𝑆𝑃𝑖 − 𝐹𝐶̅̅̅̅ )2 ; and N = the

number of individual forecasts.

Notably, our firm’s earnings risk variable captures the overall uncertainty in analysts’ earnings

forecasts that includes analyst-specific uncertainty (D in equation 3.1) plus the common

uncertainty among analysts (SE in equation 3.1). In doing so, we capture the overall

uncertainty surrounding analysts’ earnings forecasts for a particular firm thus it represents

the future earnings risk for a firm.

3.3.2. Non-Financial Information disclosures

As discussed earlier, the Bloomberg ESG disclosure scores are used as the leading

independent variable and as an auxiliary dependent variable in this study. We use ESG

disclosure scores reported in the Bloomberg ESG database to proxy for a firm’s NFI

disclosures. Bloomberg analysts construct ESG disclosure scores using over 900 data fields

across a variety of NFI indicators, such as air quality, water, and energy management, human

capital, waste management, compensation, diversity, and shareholders' rights. To construct

the disclosure scores, Bloomberg uses company reported data from CSR and annual reports,

company websites and other company sources to maintain the accuracy and consistency of

their database23.

Overall, ESG disclosure scores present a bigger picture of a firm’s efforts to inform its

stakeholders about the policies and actions implemented by the firm for a better society and

a safer environment. Given the depth and breadth of data used to calculate the scores,

Bloomberg ESG disclosure scores serve as a better proxy than the broader proxies of NFI

disclosures such as stand-alone CSR reports and self-created disclosure indices. Moreover,

Bloomberg analysts also construct sub-scores that individually consider environmental, social

23 See https://data.bloomberglp.com/professional/sites/10/1148330431.pdf

42

and governance disclosures separately. These sub-scores are the environmental disclosure

score (EDS), social disclosure score (SDS), and governance disclosure score (GDS). These

subdivisions of ESG scores serve the needs of stakeholders who are interested in a particular

area of sustainability, such as the environment or governance. We apply these sub-scores to

measure the disaggregated impact of ESG disclosures on a company's future earnings risk.

The Bloomberg ESG database provides data for over 11,500 companies across 70 countries.

The comprehensive coverage of this database lines up with our sample companies that are

spread around 30 different countries across seven regions. Several researchers such as

(Buchanan et al., 2018; Han, Kim, & Yu, 2016; Nollet et al., 2016; Wang & Sarkis, 2017), have

used Bloomberg ESG scores as a measure of a firm's NFI disclosures which confirms the

reliability and acceptability of these scores as a sound indicator of firm's NFI disclosures.

3.3.3. Sensitive industries

Prior studies on NFI disclosures maintain that because of the harmful effects of their business

activities on society and the environment, sensitive industry firms are subject to higher

reputation risk than their counterparts in other industries. Reporting more information about

business activities will increase the transparency of business operations that, as a resultant,

provides higher benefits for sensitive industry firms (Garcia et al., 2017). Following (Baron et

al., 2011; Garcia et al., 2017)and Garcia et al. (2017), we treat a firm as a sensitive industry

firm if it belongs to the alcohol, tobacco, gambling, weapon production, adult entertainment,

oil, gas, and consumable fuels, metals and mining, paper and forest products, chemicals,

construction materials, or energy businesses. This study uses a dummy variable to measure

the impact of this variable; the sensitive industry variable equals 1 if a firm belongs to a

sensitive industry and 0 otherwise.

3.3.4. Non-Financial Information disclosure regulations

Recently, the demand for mandating NFI disclosures has received unprecedented support

from different groups of stakeholders, forcing countries such as South Africa, China, Malaysia,

and Denmark, to regulate NFI disclosures that have been mostly voluntary in the past. More

recently, the European Parliament enacted NFI disclosure regulation in its non-financial

reporting directive (Directives, 2014), that provides a detailed framework for NFI reporting by

43

large organisations operating in the EU. The regulation applied from 1 January, 2017, for

target firms operating in all member states.

The regulation of NFI disclosures prompted researchers such as Bernardi and Stark (2018) and

Ioannou and Serafeim (2017) to measure the impact of regulations on the level of NFI

disclosures by target firms and how such regulations can mediate the association of NFI

disclosures and firm value or risk. Recognising the potential impact of NFI disclosure

regulations on improving NFI disclosures and so reducing the information opacity of a

reporting firm's earnings, this study uses the recent EU regulation as an exogenous shock to

measure the direct and mediating impact of regulations on the level of NFI disclosures by

target firms and the reduction in their earnings risk, respectively. Hence, the regulation

variable equals 0 before 2017 and 1 afterwards.

3.4. Control Variables

Prior studies show that a firm’s information environment is affected by several firm-specific

factors such as size, financial distress, the visibility, earnings volatility, profitability, and firm

growth, among others. Based on the prior literature on NFI disclosures and their association

with firm risk (Bernardi & Stark, 2018; Dhaliwal et al., 2012; Jo & Na, 2012; Sassen et al., 2016),

we control for firm size, debt ratio, the number of analysts following a firm, earnings per share

(EPS) volatility, loss, and book to market ratio to achieve reliable, accurate estimates.

3.4.1. Firm size

Firm size is a critical variable in corporate finance literature because it affects a firm's

profitability, returns, and value-related performance indicators. Likewise, prior studies show

that larger firms have a higher impact on society and the environment because of their scale

of operations. They are also highly visible and extensively followed by the media and

stakeholders. Consequently, larger firms make higher NFI disclosures (Ali, Frynas, &

Mahmood, 2017; Branco & Rodrigues, 2008; Chiu & Wang, 2015; Tagesson, Blank, Broberg,

& Collin, 2009) and have lower dispersion in analysts’ earnings forecasts (Bernardi & Stark,

2018; Flores, Fasan, Mendes‐da‐Silva, & Sampaio, 2019). Thus, controlling for firm size is

essential to measure the actual impact of NFI disclosures on a company’s earnings risk. This

44

study uses the natural log of total assets at the end of a year to proxy for firm size and we

expect a negative (positive) effect of firm size on earnings risk (NFI disclosures).

3.4.2. The number of analysts following a firm

The number of analysts following a firm indicates a higher visibility and information demand

for that firm among investors and other stakeholders. Accordingly, Baldini, Dal Maso,

Liberatore, Mazzi, and Terzani (2018) and (Ioannou & Serafeim, 2012) report a significant

positive relationship between the number of analysts following and NFI disclosures. However,

some studies show a significant negative relationship between the number of analysts

following and analysts’ forecasts absolute error (Garrido‐Miralles, Zorio‐Grima, & García‐

Benau, 2016; Martínez‐Ferrero et al., 2016) whereas others report this relationship to be

insignificant (Bernardi & Stark, 2018; D. Lee, 2017). According to Dhaliwal et al. (2012), the

number of analysts following a firm suggests competition among analysts. Thus, the greater

the number of analysts, the better the accuracy of analyst forecasts. This study uses the

number of analysts following the firm as a proxy for a firm's NFI disclosures demand and

analysts’ competition. The expected sign of this variable is negative when we use earnings

risk as the dependent variable and otherwise in the case of NFI disclosures.

3.4.3. Financial leverage

In general, the leverage ratio provides ideas about the capital structure of a company and the

extent to which creditors can influence decision-making (X. Liu & Anbumozhi, 2009). Further,

in the context of agency theory, Jensen and Meckling (1979) state that highly leveraged firms

are more likely to make NFI disclosures to reduce the agency cost. Similarly, Andrikopoulos

and Kriklani (2013) maintain that companies with a lower leverage face less pressure from

creditors to disclose NFI. Nevertheless, leverage could provide additional funding to initiate

new investments and sustainable operations so highly leveraged firms may lack the ability to

invest intensively in CSR activities because of their high cost (Andrikopoulos & Kriklani, 2013;

Stanny & Ely, 2008). Accordingly, Purushothaman, Tower, Hancock, and Taplin (2000) and

Branco and Rodrigues (2008) report a negative relationship between the degree of financial

leverage and CSR disclosures by Singaporean and Portuguese companies, respectively.

Additionally, financially distressed firms are likely to have an opaque information

environment, which impairs analysts' ability to accurately forecast their future income

45

(Espahbodi, Dugar, & Tehranian, 2001; Iatridis, 2016). This study uses the debt to total asset

ratio as a proxy for financial leverage and we hypothesise a positive relationship with earnings

risk and a negative relationship with NFI disclosures.

3.4.4. Loss

As profitability and stability in earnings are critical factors in projecting the future earnings of

a firm. Conversely, it is challenging to forecast earnings for organisations that report losses

(Hope, 2003). Accordingly, prior studies show a significant positive relationship between a

loss in the current year and analysts’ earnings forecasts error and dispersion for the following

year (Cormier & Magnan, 2014; Dhaliwal et al., 2012; Martínez‐Ferrero et al., 2016).

Therefore, it is essential to control for a firm’s current year reported earnings when its future

earnings forecasts are being made. Therefore, this study uses loss as a dummy variable equal

to 1 if a firm reports negative earnings in the current year, 0 otherwise. We expect a positive

relationship between loss and earnings risk.

3.4.5. Earnings volatility

Dichev and Tang (2009) report that uncertain earnings increase the complexity of future

earnings forecasts, thus we expect higher earnings uncertainty to be positively associated

with analysts’ earnings forecasts dispersion. Following (Dhaliwal et al., 2012), this study uses

earnings per share volatility as a proxy for firm earnings uncertainty measured as the natural

logarithm of time-series standard deviation for a firm's EPS using the last 10 years rolling

window. To avoid higher dependence on recent earnings, we use a 10-year rolling window

that considers substantial amount of time for earnings volatility estimation.

3.4.6. Financial opaqueness

Future earnings of firms with a higher level of financial opaqueness are challenging to predict

so there exists a negative relationship between analysts’ forecast accuracy and financial

opaqueness (Hope, 2003). Dhaliwal et al. (2012) report that information asymmetry and NFI

disclosure are more robust for firms with a higher level of financial opaqueness. To control

for this variable, we follow (Bhattacharya, Daouk, & Welker, 2003) and (Dhaliwal et al., 2012)

to compute firm-level financial transparency using an average of the past three years scaled

accruals. The calculation of accruals is as follows:

46

𝐴𝑐𝑐𝑟𝑢𝑎𝑙 =∆𝐶𝐴− ∆𝐶𝐿− ∆𝐶𝐴𝑆𝐻+ ∆𝑆𝑇𝐷−𝐷𝐸𝑃+ ∆𝑇𝑃

𝑙𝑎𝑔 (𝑇𝐴) (3.2)

where: ∆𝐶𝐴, ∆𝐶𝐿, ∆𝐶𝐴𝑆𝐻, ∆𝑆𝑇𝐷 and ∆𝑇𝑃 equal the change in the total current assets,

current liabilities, cash, and the short-term portion of debt and income tax payable,

respectively. 𝐷𝐸𝑃 measures the depreciation and amortisation for the year and 𝑙𝑎𝑔 (𝑇𝐴) is

the total assets of a firm in the previous year. If a firm has higher than the industry-year mean

of accruals for a particular year, Accrual equals 1 and 0 otherwise.

3.4.7. Firm growth

A growing firm would attract more investors, which, in turn, increases the information

demand for such firms. In their classic work, Fama and French (1998) argue that growth stocks

are riskier than values stocks because of their underlying characteristics such as small size and

higher growth. Consequently, information processing of growth firms is relatively complex for

the analysts, which leads to greater disagreement among analysts, causing forecast errors

and greater dispersion (Maaloul, Ben Amar, & Zeghal, 2016). This study uses the market-to-

book ratio as a proxy of growth opportunities available to a company and we expect it to be

negatively related to a firm’s earnings risk.

3.4.8. Firm profitability

A firm’s financial performance is also considered an important variable determining the level

of a firm’s NFI disclosures (Cormier & Magnan, 1999; Gamerschlag et al., 2011; Tagesson et

al., 2009) and has a significant influence on the accuracy of analysts’ earnings forecasts

(Bernardi & Stark, 2018; Cormier & Magnan, 2014; Dhaliwal et al., 2012). Therefore, to control

the for financial performance effect, we use return on assets (ROA) as a proxy for a company’s

financial performance because it embodies the performance of total investments made by an

organisation. Based on the literature, the financial performance variable carries a positive and

negative sign against level NFI disclosures and earnings risk, respectively.

3.4.9. Industry and index dummies

We use industry and index fixed effects as control variables in our empirical analysis.

Industries included in the sample are as follows: communication services, consumer staples,

consumer discretionary, energy, finacials, health care, industrials, information technology,

47

materials, real estate, and utilities. The indexes included in the sample are: S&P 500, S&P

Europe 350, S&P/TOPIX 150, S&P/ASX All Australian 50, S&P Asia 50, S&P Latin America 40

and S&P/TSX 60.

Table 3. 2. A summary of the study variables and expected signs Variable Variable Type Proxy Literature Expected

Sign

NFI disclosures Independent ESG disclosure scores

(Buchanan, Cao, & Chen, 2018; Han et al., 2016; Nollet, Filis, & Mitrokostas, 2016; Wang & Sarkis, 2017)

-ve

Sensitive industry Firms

Independent Dummy variable (Baron et al., 2011; Garcia et al., 2017) -ve

NFI disclosure Regulations

Independent Dummy variable (Ioannou & Serafeim, 2011; Ioannou & Serafeim, 2017; Bernardi and Stark, 2018)

-ve / +ve

Firm Size Control Logarithm of total market

capitalization

(Ali, Frynas, & Mahmood, 2017; Branco & Rodrigues, 2008; Chiu & Wang, 2015; Tagesson, Blank, Broberg, & Collin, 2009)

-ve

Information demand for the firm

Control Number of analysts following

(Garrido‐Miralles et al., 2016; Martínez‐Ferrero et al., 2016; Bernardi & Stark, 2018; Lee, 2017; Dhaliwal et al. 2012)

-ve / +ve

Financial leverage

Control Debt ratio (Stanny & Ely, 2008) +ve

Firm Growth Control Market-to-book ratio

Maaloul et al., 2016 +ve

Loss Control Dummy variable (Cormier & Magnan, 2014; Dhaliwal et al., 2012; Martínez‐Ferrero et al., 2016)

+ve

Earnings Volatility

Control Standard deviation of EPS – 10 years rolling window

(Dichev and Tang ,2009; Dhaliwal et al., 2012)

+ve

Financial Opaqueness

Control Dummy variable (Hope, 2003; Dhaliwal et al. 2012 ; Choi & Wong, 2007; Gul, Kim, & Qiu, 2010)

+ve

Financial performance

Control Return on assets (Cormier & Magnan, 1999; Gamerschlag et al., 2011; Tagesson et al., 2009)

+ve

Note: Table 3.2 provides a summary of variables and expected signs used in the empirical analysis.

3.5. Basic Diagnostic Test

This study uses regression analysis to test the hypotheses because this analysis method

provides a simple, but dominant way to determine the relationship between two variables

while controlling for other factors. However, before applying the regression analysis, some

diagnostic checks need to be performed to examine the appropriateness of the data for the

48

regression approach. This study performs the following diagnostic checks before regression

analysis.

3.5.1. Pairwise correlation

One of the underlying assumptions for regression analysis is to check the presence of

multicollinearity in the data. Explicitly, there should be no perfect collinear variables among

the independent variables as perfectly collinear explanatory variables undermine the

accuracy of regression results and also result in the violation of one of the underlying classic

linear regression model (CLRM) assumptions (Baltagi, 2008; Gujarati, 2009). Accordingly, we

uses pairwise Pearson correlation analysis to test the multicollinearity assumption of the

regression analysis. This test also provides a basic idea of the association between two

variables and whether the association is worth further investigation. Further, in the post-

estimation analysis, this study also applies variance inflation factor (VIF) test to quantify the

severity of correlation among the variables. As a rule of thumb, a VIF score of less than four

indicates the absence of multicollinearity.

3.6. Regression models

We analyse the impact of NFI disclosures on a company's earnings risk for global index

companies. In doing so, we use a multivariate regression framework because this method

provides the flexibility of testing the impact of an independent variable on a dependent

variable while controlling for several factors. Additionally, the regression framework allows

the inclusion of interaction terms and indicator variables that are essential variables in this

study.

𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠 𝑟𝑖𝑠𝑘 = 𝛽0 + 𝛽1𝐸𝑆𝐺𝐷𝑆𝑖𝑡 + 𝛽2 𝑙𝑛𝑠𝑖𝑧𝑒𝑖𝑡 + 𝛽3𝐿𝐸𝑉𝑖𝑡 + 𝛽4𝐿𝑁𝐴𝐹𝑖𝑡 + 𝛽5𝑀𝑇𝐵𝑖𝑡 +

𝛽6 𝐸𝑉𝑂𝐿𝑖𝑡 + 𝛽7𝐿𝑜𝑠𝑠𝑖𝑡 + 𝛽8𝐴𝑐𝑐𝑟𝑢𝑎𝑙𝑠𝑖𝑡 + 𝛽9𝑅𝑂𝐴𝑖𝑡 + 𝛽9 𝐼𝑁𝐷 + 𝛾𝑡 + 𝛿𝑖 + ∈𝑖𝑡 (3.3)

In equation (3.3), we test the impact of NFI disclosures on earnings risk for global index

companies. 𝐸𝑆𝐺𝐷𝑆 measures the ESG disclosure scores, a proxy for NFI disclosures that

captures the impact of NFI disclosures on a firm’s earning risk. Studies show that individual

ESG dimensions have varying effects on the risk profile of an organisation. For example, Chang

et al. (2014) show the varying impact of CSR dimensions that target institutional and primary

stakeholders. Similarly, Benlemlih et al. (2018) report the social dimension as having a larger

49

negative impact on a firm’s total and idiosyncratic risk than the environmental dimension. To

ascertain these varying effects, we use the disaggregated ESG factors, i.e., environmental,

social and governance disclosure scores, to measure their impact on earnings risk.

Lins et al. (2017) report that a higher intensity of CSR disclosures has a more significant impact

on stock returns for the reporting companies. We use quartiles of ESG disclosure scores to

measure the impact of overall and disaggregated NFI disclosures on firms’ earnings risk. We

split ESG disclosure scores into four quartiles with dummies for second, third and fourth

quartile; the intercept captures the effect of the first quartile.

𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠 𝑟𝑖𝑠𝑘 = 𝛽0 + 𝛽2𝑄2𝐸𝑆𝐺𝐷𝑆𝑖𝑡 + 𝛽3𝑄3𝐸𝑆𝐺𝐷𝑆𝑖𝑡 + 𝛽4𝑄4𝐸𝑆𝐺𝐷𝑆𝑖𝑡 +

𝛽2𝑙𝑛𝑠𝑖𝑧𝑒𝑖𝑡 + 𝛽3𝐿𝐸𝑉𝑖𝑡 + 𝛽4𝐿𝑁𝐴𝐹𝑖𝑡 + 𝛽5𝑀𝑇𝐵𝑖𝑡 + 𝛽6 𝐸𝑉𝑂𝐿𝑖𝑡 + 𝛽7𝐿𝑜𝑠𝑠𝑖𝑡 +

𝛽9𝑅𝑂𝐴𝑖𝑡 + 𝛽8𝐴𝑐𝑐𝑟𝑢𝑎𝑙𝑠𝑖𝑡 + 𝛾𝑡 + 𝛿𝑖 + ∈𝑖𝑡 (3.4)

In equation (3.4), 2𝐸𝑆𝐺𝐷𝑆 , 𝑄3𝐸𝑆𝐺𝐷𝑆,and 𝑄4𝐸𝑆𝐺𝐷𝑆 capture the impact of the second, third

and fourth quartiles of the NFI disclosure on earnings risk and the constant term captures the

coefficient for firms in the first quartile of NFI disclosures.

Previous studies report that sensitive industry firms make higher NFI disclosures than non-

sensitive firms. For example, using a sample from BRICS countries, Garcia et al. (2017) report

that sensitive industry firms produce a greater ESG performance than their non-sensitive

counterparts. Thus, this study uses a propensity weighted sample to ascertain the causal

effect of sensitive industry on the cumulative and disaggregated disclosures of sensitive

industry firms in the sample.

𝐸𝑆𝐺𝐷𝑆𝑖𝑡 = 𝛽0 + 𝛽1𝑆𝑒𝑛𝑠𝑖𝑡𝑖𝑣𝑒 + 𝛽4𝑙𝑛𝑠𝑖𝑧𝑒𝑖𝑡 + 𝛽5𝐿𝐸𝑉𝑖𝑡 + 𝛽6𝐿𝑁𝐴𝐹𝑖𝑡 + 𝛽7𝑀𝑇𝐵𝑖𝑡 +


In equation (3.5), 𝑆𝑒𝑛𝑠𝑖𝑡𝑖𝑣𝑒 is an indicator variable equal to 1 if a firm belongs to sensitive

industry, 0 otherwise.

We split the sample into sensitive and non-sensitive firms to ascertain whether sensitive firms

benefit more in terms of reduced future earnings risk by reporting NFI disclosures. We expect

the effect of NFI disclosures to be both economically and statistically higher for sensitive

industry firms since these firms go through strict scrutiny because of their sensitive status.

Thus, disclosing more NFI will reduce information asymmetry for these firms more than for

50

non-sensitive industry firms, resulting in reduced earnings risk. Jo and Na (2012) report a

relatively lower market risk for sensitive industry firms that have more CSR disclosures in both

the short and the long run.

For the control variables, 𝑙𝑛𝑠𝑖𝑧𝑒 is the natural logarithm of a firm's market capitalisation at

year-end and 𝐿𝐸𝑉 represents financial distress calculated as total debt scaled by total assets.

𝐿𝑁𝐴𝐹 is the variable that proxies for the number of analysts reporting the earnings forecasts

for a company during the current year, and 𝑀𝑇𝐵 measures the market growth opportunities

of a company. 𝐸𝑉𝑂𝐿 is a measure of a firm's earnings volatility based on the standard

deviation of earnings per share. 𝐿𝑜𝑠𝑠, 𝐴𝑐𝑐𝑟𝑢𝑎𝑙𝑠 and 𝐼𝑁𝐷 are dummy variables that equal 1

if a firm reports negative earnings, shows financial opacity or belongs to sensitive industry

and 0 otherwise, respectively.

To test the effect of mandatory NFI disclosure regulations, we first measure the impact of

regulations on the quantity of NFI disclosures using the following model:

𝐸𝑆𝐺𝐷𝑆𝑖𝑡 = 𝛽0 + 𝛽1𝑅𝑒𝑔 ∗ 𝑡𝑟𝑒𝑎𝑡𝑒𝑑 + 𝛽2𝑙𝑛𝑠𝑖𝑧𝑒𝑖𝑡 + 𝛽3𝐿𝐸𝑉𝑖𝑡 + 𝛽4𝐿𝑁𝐴𝐹𝑖𝑡 + 𝛽5𝑀𝑇𝐵𝑖𝑡 +


In equation (3.6), 𝐸𝑆𝐺𝐷𝑆 is a measure of the NFI disclosures. 𝑅𝑒𝑔 ∗ 𝑡𝑟𝑒𝑎𝑡𝑒𝑑 is an interaction

term that exclusively captures the effect of regulations on the treatment group firms’ NFI

disclosures. The interaction variable is constructed by interacting the treated dummy (S&P

350 index) with a regulation period (post-2017) dummy. The pre- and post-regulation periods

are 2008-2016 and 2017-2018, respectively. Therefore, the post-regulation interaction term

captures the effect of the regulations on NFI disclosures for the treated firms whereas the

constant term captures the disclosure scores of the control group of firms. We use a

propensity score weighting method to mitigate the problem of incomparability of sample

firms in the treatment and the control groups. Equation (3.6) also includes other time-varying

firm-level variables to control for their impact. We also apply industry and year fixed effects

to control for industry and time effects.

Finally, to test the impact of regulation-driven changes in ESG disclosure scores on a firm’s

earnings risk, we use difference-in-differences (DID) analysis. We decompose the ESG

51

disclosure scores into 𝑃𝑟𝑒𝑅𝑒𝑔𝑡𝑟𝑒𝑎𝑡𝑒𝑑𝐸𝑆𝐺𝐷𝑆 and 𝑃𝑜𝑠𝑡𝑅𝑒𝑔𝑡𝑟𝑒𝑎𝑡𝑒𝑑𝐸𝑆𝐺𝐷𝑆 components as

defined in equation (3.7):

𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠 𝑟𝑖𝑠𝑘𝑖𝑡 = 𝛽0 + 𝛽1𝑃𝑟𝑒𝑅𝑒𝑔 ∗ 𝑡𝑟𝑒𝑎𝑡𝑒𝑑 ∗ 𝐸𝑆𝐺𝐷𝑆𝑖𝑡 + 𝛽2𝑃𝑜𝑠𝑡𝑅𝑒𝑔 ∗ 𝑡𝑟𝑒𝑎𝑡𝑒𝑑 ∗

𝐸𝑆𝐺𝐷𝑆𝑖𝑡 + 𝛽3 𝑙𝑛𝑠𝑖𝑧𝑒𝑖𝑡 + 𝛽5𝐿𝐸𝑉𝑖𝑡 + 𝛽6𝐿𝑁𝐴𝐹𝑖𝑡 + 𝛽7 𝐸𝑉𝑂𝐿𝑖𝑡 + 𝛽8𝑀𝑇𝐵𝑖𝑡 + 𝛽9𝐿𝑜𝑠𝑠𝑖𝑡 +

𝛽9𝑅𝑂𝐴𝑖𝑡 + 𝛽10𝐴𝑐𝑐𝑟𝑢𝑎𝑙𝑠𝑖𝑡 + 𝛾𝑡 + 𝛿𝑖 + ∈𝑖𝑡 (3.7)

𝑃𝑟𝑒𝑅𝑒𝑔𝑡𝑟𝑒𝑎𝑡𝑒𝑑𝐸𝑆𝐺𝐷𝑆 = (1-Reg) * treated * ESGDS

𝑃𝑜𝑠𝑡𝑅𝑒𝑔𝑡𝑟𝑒𝑎𝑡𝑒𝑑𝐸𝑆𝐺𝐷𝑆 = Reg * treated * ESGDS

𝑃𝑟𝑒𝑅𝑒𝑔𝑡𝑟𝑒𝑎𝑡𝑒𝑑𝐸𝑆𝐺𝐷𝑆 captures the ESG scores for S&P 350 Europe index firms before the

enactment of the regulations and zero otherwise and 𝑃𝑜𝑠𝑡𝑅𝑒𝑔𝑡𝑟𝑒𝑎𝑡𝑒𝑑𝐸𝑆𝐺𝐷𝑆 represents

the ESG scores for treated firms in the post-enactment period and zero otherwise. Reg = 1 for

any firm-year after the enforcement (Directives, 2014), 0 otherwise. This method is beneficial

to compare the impact of ESG disclosure scores on a firm’s earnings risk for S&P 350 Europe

index firms both in the pre- and post-regulation period with the control group (firms in

S&P1200 index other than S&P350) firms.

Given that our sample is restricted to S&P global index companies, we follow Baltagi (2008)

and use fixed effects estimations for all the models above. According to Baltagi (2008) “The

fixed effects model is an appropriate specification if we are focussing on a specific set of N

firms . . . and our inference is restricted to the behaviour of this set of firms” (p. 14). The

Hausman and joint significance Chow (F-test) tests also support the use of fixed-effects

estimation. Alternatively, pooled ordinary least square (OLS) estimation is another potential

method to estimate the impact of NFI disclosure scores on firm earnings risk. However, we

opted against applying this technique since it delivers inconsistent estimates when the fixed

effects technique is appropriate (Cameron & Trivedi, 2005).

Nevertheless, Roberts and Whited (2013) recommend caution in the use of fixed effects

estimation in situations where: (1) the dependent variable is a first differenced variable, i.e.,

a change in investments or market prices; (2) if the research question relates to understanding

cross-sectional variation among the cross-sections, i.e., understanding cross-sectional

variation in analysts’ forecasts error or NFI disclosures in the global index companies in the

context of this study; and (3) where the lagged dependent variable is an explanatory variable,

52

i.e., last year’s earnings per share, when the dependent variable is the current year’s earnings

per share. Notably, our study variables and research questions do not suffer from any of these

limitations; hence the use of fixed effects is a suitable estimation method for the study.

3.7. Robustness and Endogeneity

Panel data models potentially suffer from the presence of non-constant error of variance,

which could be a problem in our data because of the presence of firms with varying

characteristics such as firm size. Consequently, the presence of non-constant variance errors

would result in heteroscedasticity, producing inefficient regression estimates (Baltagi, 2008).

To overcome this problem, we report cluster robust standard errors that correct for the

possible presence of heteroscedasticity. Additionally, studies show panel data studies suffer

from cross-sectional dependence that results in imprecise regression estimates. To correct

for this error, we use (Driscoll & Kraay, 1998) standard error estimation method that controls

for cross-sectional dependence in samples with a large number of cross-sections and a small

number of time-series observations.

We replace firm fixed effects with sector fixed effects along with time fixed effects to allow

more cross-sectional variation. Additionally, we use different proxies for firm characteristics,

i.e., market capitalisation for firm size and volatility of stock returns for firm risk, to test the

robustness of our control variables measures. Prior studies show the realtionship between a

firm's NFI disclosures and risk suffers from omitted variable bias and simultaneity bias (Gangi,

Meles, Monferrà, & Mustilli, 2018; Jo & Na, 2012; D. Lee, 2017). The use of the fixed effects

model mitigates the omitted variable bias, but simultaneity bias is potentially present in our

models. Following (Chang et al., 2014; Oikonomou et al., 2012), we use lagged NFI disclosures

and control variables to resolve the potential simultaneity bias. Further, because of the

endogenous nature of the NFI disclosure variables, we also perform the generalised methods

of momentum (GMM) estimation that provides the flexibility of including both past levels of

a firm’s earnings risk and fixed effects to account for the dynamic aspect of the NFI disclosure

and firm earnings risk relationship and time-invariant unobservable heterogeneity (Cai et al.,

2016). In addition, the GMM estimation method controls for reverse causality that is also a

potential source of endogeneity in the NFI disclosures and firm risk link (Lueg, Krastev, & Lueg,

2019).

53


This chapter outlines the study sample, the dependent, independent and control variables,

the empirical models and the robustness measures used in this study. The study sample

comprises data from S&P 1200 global index companies from 2008-2018. The primary

independent variables are the amount of NFI disclosures characterised by ESG disclosure

scores and NFI disclosure regulations in the EU directive 2014. We use a fixed effects

regression model to ascertain the impact of NFI disclosures on a firm’s earnings risk.

Additionally, this study uses the difference-in-differences approach to capture the mediating

impact of NFI regulations on the relationship between NFI disclosures and earnings risk.

Additionally, this study generates a propensity score-matched sample of regulated and non-

regulated firms to capture the effect of NFI disclosure regulations on the extent of disclosures.

Based on evidence from literature, we control for firm-specific characteristics such as size,

financial performance, growth, earnings volatility, financial opaqueness, information

demand, earnings volatility and financial leverage, to disentangle their effect from the NFI

disclosures and firm risk relationship. To achieve robust regression estimates, we use

standard errors that account for potential sources of bias such as heteroscedasticity,

autocorrelation and cross-sectional dependence (Driscoll & Kraay, 1998). Additionally, to

address endogeneity concerns, we perform a system GMM estimation that uses the dynamic

aspect of the NFI disclosures and firm earnings risk relationship and controls for time-

invariant unobservable heterogeneity.

54

Chapter 4

Results and Discussion

4.1. Introduction

This chapter presents the descriptive statistics, data diagnostics tests and the empirical

analyses to test the research hypotheses concerning the impact of NFI disclosures on earnings

risk, the impact of industry-type on the level of NFI disclosures and the mediating impact of

industry-type on the relationship between NFI disclosures and earnings risk. The descriptive

analysis covers the descriptive statistics of the study variables and the evolution of NFI

disclosures during the sample period. Next, the basic diagnostic test of pairwise correlation

are presented is check that our data meet conditions of regression analysis. Next, the chapter

tests the first hypothesis by exploring the linear and quartile-based impact of NFI disclosures

on the sample firms’ earnings risk. Additional subsample and robustness tests are performed

to ascertain the robustness of our baseline results.

To test the second hypothesis, we present evidence on how the sensitive nature of different

industries determines the NFI disclosure practices of sensitive industry firms and explores

whether sensitive industry membership has varying effects on the different dimensions of NFI

disclosures. The resulting impact of the different levels of NFI disclosures by sensitive and

non-sensitive industry firms on the association between NFI disclosures and future earnings

risk is explored using a split-sample and interaction term approach. The spilt-sample and

interaction term analysis assists in testing the study’s third hypothesis. The chapter ends with

a summary of the findings regarding the impact of NFI disclosures on the firm earnings risk

and the role that sensitive industry membership plays in determining the NFI disclosures of

the sample firms.

4.2. Descriptive Statistics

Table 4.1 presents the descriptive statistics of the study variables. To account for potential

outliers, 1% values of all study variables are winsorised at each end. The primary dependent

variable represented by the earnings risk variable captures the overall uncertainty in an

analyst’s earnings forecast that includes analyst specific uncertainty plus the common

uncertainty among analysts. Following, Barron et al. (1998), we define analyst specific

55

uncertainty as the difference between individual analyst’s forecasts and the mean of analysts’

forecast EPS. Common uncertainty measures the difference between the actual earnings per

share and analysts’ mean forecast EPS. In this sphere, we capture the overall uncertainty

surrounding the analysts’ earnings forecasts for a particular firm; thus, it represents the future

earnings risk for a firm. The mean value for earnings risk shows that, on average, the

uncertainty in analysts’ forecasts dispersion is positive (2.201), which implies that analysts

tend to make dispersed EPS forecasts for the sample organisations compared with their actual

and mean forecasted EPS. Moreover, because of the higher standard deviation (5.656) and

large positive skewness (3.026), this study uses log-transformed values of the earnings risk in

the empirical analyses. The use of log-transformed earnings risk is consistent with the

literature (Barron et al., 2009; Bernardi & Stark, 2018) and does not change the primary

inferences of the results if log-transformation is not applied.

The study’s primary independent variables are the NFI disclosures measured as the

environmental, social and governance (ESG) disclosure scores. The mean of the aggregate ESG

disclosure variable that denotes the overall NFI disclosures of a firm shows that, on average,

sample firms make a moderate disclosures (39.68). The higher standard deviation (14.39)

shows that NFI disclosure practices vary substantially among the sample firms, which is an

expected outcome, considering the diversity of firms in our sample. Moreover, the

disaggregated NFI disclosures show that sample firms make relatively more NFI disclosures

concerning the governance dimension (52.27) than for the environmental (35.18) and social

(37.70) domains. The regulatory requirements to disclose governance-related information is

one possible explanation for the relatively higher levels of governance disclosure scores. The

sample firms report identically on environmental and social disclosures in terms of their mean

and standard deviations measures, indicating that those NFI dimensions receive equal

consideration from corporate managers. Our findings regarding moderate aggregate

disclosures and relatively higher disclosure scores for the governance dimension supports

other studies that used S&P1200 firms as their sample (Del Bosco & Misani, 2016; Hassan &

Giorgioni, 2019).

Table 4.1 also includes the descriptive statistics for the control variables used in the study.

Based on the prior literature, the control variables are firm size, number of analysts following

56

a firm, the debt-to-asset ratio, EPS volatility, profitability, and book to market ratio (Bernardi

& Stark, 2018; Dhaliwal et al., 2012; Jo & Na, 2012; Sassen et al., 2016). Firm size and the

analysts following variables serve as controls for a firm’s standing because of its size and/or

sensitivity concerning ESG issues. This study uses log values of firm size and analysts following

variables with an average score of 10.157 and 2.75, respectively. The debt ratio variable,

which serves as a control for capital structure, shows an average ratio of 25.71 and standard

deviation of 16, indicating considerable variation in the capital structure of the sample firms.

Table 4. 1. The descriptive statistics of the study variables Variable Obs Mean Std Dev. Min. Max. p25 p50 p75

Earnings risk 9,887 2.201 5.656 -.006 23.282 0.006 0.062 0.763

Log Earnings risk 9,885 -4.329 3.369 -11.912 5.883 -6.586 -4.668 -2.309

ESGDS 9,813 39.683 14.396 11.842 69.709 28.926 41.322 50.718

EDS 9,129 35.185 17.058 1.785 71.900 21.379 37.209 48.062

GDS 9,817 57.274 9.087 28.571 76.785 51.786 57.143 62.500

SDS 9,757 37.705 17.679 3.333 77.193 24.561 38.597 50.877

Firm Size 10,055 10.157 1.515 7.267 14.443 9.079 9.951 11.017

NAF 10,346 2.750 .575 .6931 3.637 2.485 2.833 3.135

Leverage 10,052 25.708 16.005 0 70.769 14.003 24.199 35.908

Volatility: EPS 10,511 0.836 2.241 -5.531 10.700 -0.556 0.314 1.586

Growth 9,802 2.642 3.087 0.270 21.871 1.060 1.716 2.928

ROA 10,235 4.376 6.126 -18.22 24.534 0.972 3.672 7.281

Note: The table presents the descriptive statistics for the study variables. Earnings risk is represented by the log of average analysts’ forecasts dispersion; ESGDS, EDS, GDS and SDS are proxies for overall ESG, environmental, social and governance disclosures, respectively. Firm size is the log of total assets, leverage is debt ratio, growth equal market-to-book ratio, RoA is net income divided by total assets, and NAF is the number of analysts following a firm.

Table 4.1 also shows the descriptive statistics for EPS volatility that serves as a control variable

for a firm’s earnings risk. The standard deviation of this variable shows significant dispersion

in the earnings risk of the sample firms; it ranges from -5.53 to 10.70. Finally, the return on

asset (ROA) and market to book ratio variables, that proxy for profitability and growth of

sample firms, show average scores of 4.37 and 2.64, respectively. Both variables show

respective high standard deviations, indicating that the sample firms have substantially

different profitability and growth profiles. Overall, the descriptive analysis shows higher

variation for NFI disclosure and firm characteristic variables, which is an expected outcome

given the diversity and inherent differences present in our sample firms.

The descriptive statistics of overall and disaggregated NFI variables in the descriptive analysis

show the full sample means; the literature shows levels of NFI disclosures have improved

57

significantly over the years (Cahaya, Porter, Tower, & Brown, 2015; Hassan, Adhikariparajuli,

Fletcher, & Elamer, 2019). To present a broad picture of how NFI disclosures have changed

during the sample period, Figure 4.1 graphically presents the annual aggregate and sub-

dimension average disclosure scores. The figure shows that aggregate disclosure scores have

steadily increased during the sample period from the low 30s to mid-40s. Relative to other

dimensions, a greater increase (29 to 42) is seen in the social disclosure scores during the

sample period, which suggests that, relatively, social indicators are receiving more attention

from corporate managers.

Figure 4.1. Evolution of NFI disclosures

Note: Figure 4.1 presents the annual evolution of NFI disclosures.

4.3. Diagnostics Test

Before the empirical analysis, we perform the pairwise correlation analysis to examine the

appropriateness of data variables for the regression framework. Gujarati (2009) and Baltagi

(2008) maintain that the presence of perfect collinear variables in a regression equation

undermines the estimates; hence a quick check of the pairwise correlation helps identify any

such collinearity among the variables.

4.3.1. Pairwise correlation analysis

Table 4.3 presents the pairwise correlation coefficients of all variable against each other. A

detailed look at the correlation coefficients shows that there are no perfectly collinear

variables among the selected variables. However, there is a very high pairwise correlation

58

among the aggregate NFI disclosure variable and its sub-dimensions. For example, the social

disclosure scores variable has a correlation coefficient of 0.82 with the overall NFI disclosure

scores variable. Similarly, the NFI sub-dimensions have high correlations amongst each other.

For instance, the environmental dimension exhibits a correlation coefficient of 0.45 and 0.62

with the governance and social dimensions, respectively. Therefore, one should be cautious

in using the different dimensions of NFI disclosures together in a regression equation when

measuring their impact on a firm performance/risk indicator. Accordingly, we use the

disaggregated dimensions of NFI disclosures independently to avoid potential collinearity

problems in the regression analyses.

As hypothesised, the earnings risk variable shows a negative relationship with all NFI

disclosure variables except the environmental disclosures. A thorough sub-index correlation

analysis reveals that the constituents of specific indices such as S&P AS50 drive the positive

relationship between the environmental disclosures and earnings risk. Further, the variables

concerning the risk profile of a firm such as leverage, earnings volatility and negative income,

positively correlate with our earnings risk proxy. Similarly, sensitive industry firms also show

a significant positive correlation with earnings risk. Conversely, the growth and ROA variables

have a negative correlation with earnings risk, signifying that the profitability and growth of

a firm reduce the risks surrounding future earnings. Likewise, higher analyst following also

improves the earnings forecast for a firm, which indicates the analyst following serves as a

tool of firm transparency.

The results also reveal that all the aggregate and disaggregated NFI disclosure variables

display a positive correlation with the firm size, analyst following and sensitive industry

variables. However, the growth and profitability of a firm are positively related to only the

governance domain of the NFI of the study sample firms. At the same time, these variables

show a negative relationship with the other dimensions of NFI disclosures. The variables

concerning a firm’s financial reporting opaqueness (Accrual), negative income and earnings

volatility generally maintain a negative correlation with NFI disclosures. Earnings volatility

maintains a positive correlation with overall and environmental disclosures.

Overall, the correlation analysis shows that the investigated variables maintain anticipated

correlations and there are no perfectly collinear variables that may reduce the efficiency of

59

regression estimates. Nevertheless, some variables show mixed results in terms of their

expected correlation direction. This inconsistency may arise from the diversity of our sample

since it contains firms from a variety of industries and geographical locations that treat NFI

disclosures differently.

60

Table 4. 2. Pairwise correlation matrix of the study variables

Note: The Table presents the pairwise correlation coefficients of the study variables. For variable definitions, refer to Table 4.1. *** = p<0.01; ** = p<0.05; and * = p<0.1.

Variable 1 2 3 4 5 6 7 8 9 10 11 12 13 14

1 Earnings Risk 1 -0.025** 0.062*** -0.227*** -0.091*** -0.005 0.001 -0.336*** 0.517*** -0.203*** -0.131*** 0.100*** 0.046*** 0.066***

2 ESGDS -0.018* 1 0.943*** 0.629*** 0.805*** 0.291*** 0.004 -0.063*** 0.036*** -0.028** 0.195*** -0.024** -0.108*** 0.091***

3 EDS 0.061*** 0.950*** 1 0.449*** 0.618*** 0.261*** -0.007 -0.100*** 0.153*** -0.041*** 0.131*** -0.019* -0.080*** 0.097***

4 GDS -0.210*** 0.600*** 0.431*** 1 0.514*** 0.278*** -0.004 0.059*** -0.201*** 0.037*** 0.216*** -0.050*** -0.122*** 0.021*

5 SDS -0.089*** 0.819*** 0.640*** 0.464*** 1 0.196*** 0.052*** -0.009 -0.134*** -0.015 0.221*** -0.014 -0.098*** 0.093***

6 Firm Size 0.017 0.272*** 0.241*** 0.265*** 0.185*** 1 0.054*** -0.332*** 0.124*** -0.353*** 0.204*** -0.024** -0.204*** -0.037***

7 Leverage 0.024** -0.007 -0.016 -0.014 0.035*** 0.029*** 1 0.010 -0.049*** -0.159*** -0.044*** 0.082*** 0.010 0.074***

8 Growth -0.215*** -0.041*** -0.063*** 0.056*** -0.017 -0.259*** 0.151*** 1 -0.359*** 0.583*** 0.085*** -0.224*** 0.048*** -0.134***

9 Volatility: EPS 0.653*** 0.013 0.130*** -0.211*** -0.144*** 0.100*** -0.007 -0.233*** 1 -0.193*** -0.123*** 0.085*** -0.001 0.063***

10 ROA -0.178*** -0.036*** -0.044*** 0.033*** -0.031*** -0.261*** -0.172*** 0.435*** -0.170*** 1 0.062*** -0.520*** 0.076*** -0.022**

11 NAF -0.115*** 0.194*** 0.137*** 0.197*** 0.215*** 0.222*** -0.057*** 0.057*** -0.124*** 0.060*** 1 -0.048*** -0.063*** -0.085***

12 Loss 0.111*** -0.026** -0.021* -0.053*** -0.015 -0.022** 0.080*** -0.111*** 0.075*** -0.522*** -0.046*** 1 0.029*** 0.050***

13 Accrual 0.043*** -0.101*** -0.074*** -0.121*** -0.094*** -0.194*** 0.012 0.045*** 0.004 0.063*** -0.060*** 0.029*** 1 0.038***

14 Sensitive 0.059*** 0.097*** 0.099*** 0.018 0.094*** -0.059*** 0.057*** -0.110*** 0.059*** -0.041*** -0.086*** 0.050*** 0.038*** 1

61

4.4. The Impact of NFI Disclosures on Earnings Risk

4.4.1. Baseline results

The diagnostics test performed in the previous section provides sufficient support for the

nonexistence of any perfectly collinear variable and unit-root problem in our data, indicating

that the data are suitable for regression analysis. Accordingly, we use panel data regression

techniques to capture the impact of NFI disclosures on a firm’s earnings risk. Specifically, we

apply the panel fixed-effects regression method to test the first hypothesis about the impact

of NFI disclosures on earnings risk. As explained in Chapter 3.6, the fixed-effects regression

method is the most suitable technique given the nature of the study sample. Additionally, the

Hausman test (Table A1) and joint significance Chow F-test also support the fixed-effects

estimation method. We use (Driscoll & Kraay, 1998) standard error estimation method that

controls for cross-sectional dependence in samples with a large number of cross-sections and

a small number of time-series observations. In addition to controlling cross-sectional

dependence, the standard errors estimation method also controls for heteroskedasticity and

auto-correlation problems. Thus, this method is superior to conventional standard and

cluster-robust standard error estimation methods.

Table 4.3, Column 1, presents the results of the impact of aggregate NFI disclosures on

earnings risk. The overall NFI disclosure scores have a significant negative impact on the

future earnings risk, indicating that firms that make more NFI disclosures have more

predictable future earnings than firms that make lower NFI disclosures. From an economic

relevance view point, NFI disclosures reduce the future earnings risk by 1.6%, signifying that

the inclusion of ESG information could provide financial analysts more certainty about the

organisations’ future earnings. The increased certainty translates into lower information

asymmetry among stakeholders, providing the organisation the needed legitimacy regarding

their business operations. This finding agrees with existing studies that report a negative

relationship between NFI disclosures and different firm risk proxies such as total risk

(Benlemlih et al., 2018), systematic risk (Lueg et al., 2019; Oikonomou et al., 2012; Orlitzky &

Benjamin, 2001), firm-specific risk (Benlemlih et al., 2018; Sassen et al., 2016), and analysts

forecasts dispersion (Martínez-Ferrero et al., 2018; Xu & Liu, 2018). Accordingly, this result

62

provides a strong basis to reject the null hypothesis regarding the neutral association between

NFI disclosures and firm earning risk.

The results in Table 4.3 show that the total NFI disclosure score has a significant negative

impact on a firm’s future earnings risk. However, the impact of the individual dimensions of

NFI disclosures, environmental, social and governance disclosures, could have a dissimilar

impact on a firm’s future earnings risk given their unique nature and varying importance for

Table 4. 3. The impact of NFI disclosures on firms’ earnings risk Dependent Variable: Earnings Risk

VARIABLE 1 2 3 4

ESGDS -0.016*** (0.003)

EDS -0.008*** (0.002)

GDS -0.006 (0.006)

SDS -0.013***

(0.003) Leverage 0.018*** 0.019*** 0.018*** 0.017***

(0.005) (0.004) (0.005) (0.004) Firm Size -0.246*** -0.201** -0.308*** -0.257***

(0.070) (0.086) (0.073) (0.071) Growth -0.097*** -0.115*** -0.101*** -0.093***

(0.011) (0.011) (0.010) (0.011) Volatility: EPS 0.370*** 0.421*** 0.360*** 0.362***

(0.037) (0.033) (0.038) (0.037) Loss 0.330*** 0.298*** 0.337*** 0.328***

(0.104) (0.109) (0.105) (0.109) Accrual 0.177*** 0.191*** 0.179*** 0.183***

(0.043) (0.050) (0.044) (0.045) ROA -0.011 -0.009 -0.011 -0.010

(0.012) (0.011) (0.011) (0.012) LNAF -0.130* -0.167** -0.160** -0.144**

(0.071) (0.077) (0.080) (0.066) Constant -1.395 -2.086** -0.972 -1.366

(0.876) (0.982) (0.902) (0.893) FE time, industry and index Yes Yes Yes Yes R-Squared 0.0546 0.0533 0.0511 0.0543 Observations 8,885 8,305 8,891 8,850 Number of groups 915 887 914 913

Note: The Table presents the impact of overall and disaggregated NFI disclosures on a firm’s earnings risk. See

Table 4.1 for variable definitions; standard errors are in parentheses. *** = p<0.01, ** = p<0.05, * = p<0.1.

different stakeholders. Therefore, Table 4.3 presents the isolated impact of each NFI

dimension on a firm’s earnings risk. Table 4.3, Column 2, shows that environmental

disclosures have a significant negative impact on future earnings risk. This signifies that the

reporting of environmental initiatives such as lower greenhouse gas emissions, waste

63

reduction, and recycling of resources by a firm improve its information environment, which

leads to reduced dispersion in forecast and actual earnings. Similarly, Bernardi and Stark

(2018) and Cormier and Magnan (2015) report a significant contribution of environmental

disclosures to improving the information environment of a firm, which leads to a precise

forecast of future earnings. Rezaee, Alipour, Faraji, Ghanbari, and Jamshidinavid (2020)and

Zeng, Zhang, Zhou, Zhao, and Chen (2020) report a negative relationship between

environmental disclosures and proxies of firm risk such as market and total risk, respectively.

The results on social disclosure scores also exhibit a significant negative association with

future earnings risk. This signifies that disclosure of organisational initiatives concerning the

diversity of the workforce, sustainability governance, a better relationship with community

and customers and human capital improves the information landscape and results in better

estimates of future earnings. Our results agree with Cormier and Magnan (2014) and García‐

Sánchez, Gómez‐Miranda, David, and Rodríguez‐Ariza (2019) findings on the impact of social

dimension of NFI disclosures in improving earnings forecasts. Arayssi, Dah, and Jizi (2016)

report a negative relationship between social engagement and risk for firms that have women

directors on their board.

On the other hand, governance disclosure scores do not show any impact on a firm’s future

earnings risk, meaning that the disclosure of governance issues does not improve forecasts of

the future earnings of a firm. Similarly, Bernardi and Stark (2018) report that governance

disclosures do not add any value in the information environment of analysts in South Africa.

Using the KLD database, Erragragui (2018) report that concerns related to the governance

dimension do not have any effect on the cost of debt for U.S. based firms whereas, at the

same time, environmental concerns increase the cost of debt. However, he also reports that

environmental and governance strengths reduce the cost of equity for U.S. based firms, which

indicates that governance strengths and concerns play conflicting roles in reducing firm risk.

Similarly, using specific corporate governance disclosures related to ownership structure,

investor rights, financial transparency and board structure, Yu (2010) reports evidence of a

negative relationship between corporate governance disclosures and analysts’ forecasts

dispersion for an international sample of firms. These contradictory results show that an in-

64

depth, detailed analysis of governance disclosures may reveal more information about their

impact on future earnings risk and other risk indicators.

All the control variables, except ROA, show the expected direction and magnitude. In

particular, firm size, growth and analyst following show a negative relationship with a firm’s

future earnings risk, indicating that larger and growing firms have stable earnings streams

that are forecast more accurately than their smaller counterparts. Similarly, analyst following

reduces the dispersion in forecast and actual earnings, which indicates the monitoring effect

of analyst following. Our results corroborate the literature that shows a positive impact of

firm size, growth and analyst following in reducing a firm’s risk. Conversely, the variables

related to the risk profile of a firm such as leverage, negative earnings, and volatility of EPS

show a positive relationship with a firm’s earnings risk. The positive association between the

risk-related factors and future earnings risk of a firm not only indicates that the risk profile of

a firm increases the dispersion of forecast and actual earnings, but also validates the choice

of risk proxy. The opaqueness of financial disclosures proxied by Accrual also shows a positive

relationship with a firm’s future earnings risk, indicating that financial opacity makes it

difficult to forecast the correct future earnings of a firm. Lastly, this study controls for time,

industry and sub-index effects that account for potential aggregate time-trend effects,

differences in firms belonging to diverse industries and geographies.

4.4.2. NFI quartile results

The results discussed above focus on the average impact of aggregate and disaggregated NFI

disclosures on the future earnings risk of a firm. However, Lins et al. (2017) show that the

higher quartiles of corporate social responsibility (CSR) disclosures have a greater impact on

stock market performance than lower quartiles. Accordingly, we use the quartiles of NFI

disclosures to ascertain their varying impact on the future earnings risk of the sample firms.

We include four dummy variables in our analysis to capture the four quartiles of aggregate

and disaggregated NFI disclosures so that 1st (4th) quartile captures all the firms that make the

lowest (highest) disclosures compared with other firms.

Table. 4.4 presents the results of the NFI disclosure quartiles at the aggregate and sub-

dimension levels. The aggregate disclosure results presented in Table 4.4, Column 1, show an

increasingly negative coefficient for NFI disclosure quartiles. For example, the negative impact

65

of the overall NFI disclosures on future earnings risk of a firm with disclosure scores in the 3rd

quartile of ESGDS disclosures is 0.463 percentage points higher than for a firm with disclosure

scores in the 1st quartile. Further, Table 4.4, Column 2 to 4, show a similar pattern of

increasing impact for all sub-dimensions of NFI disclosures. This finding indicates that firms

making higher disclosures of ESG issues would have a better information environment than

lower disclosure counterparts, which reduces the gap between forecast and actual earnings

estimates.

Table. 4.4 presents the results of the NFI disclosure quartiles at the aggregate and sub-

dimension levels. The aggregate disclosure results in Table 4.4, Column 1, show an

increasingly negative coefficient of NFI disclosure quartiles. For example, the negative impact

of the overall NFI disclosures on future earnings risk of a firm with disclosure scores in the 3rd

quartile of ESGDS disclosures is 0.463 percentage points higher than a firm with disclosure

scores in the 1st quartile. Table 4.4, Columns 2 to 4, show a similar pattern of increasing impact

for all sub-dimensions of NFI disclosures. This indicates that firms making higher disclosures

on ESG issues have better information environments than lower disclosure level counterparts,

consequently, reducing the gap between forecaste and actual earnings estimates.

Interestingly, the higher quartiles of governance disclosures also show a negative impact on

future earnings risk, which was not present under the linear measure of governance

disclosures. This observation reinforces the notion that higher disclosures in all categories of

NFI disclosures improve analysts’ information processing efficiency and lead to more precise

estimates. These findings contradict the view that higher disclosure of information increases

the information load for analysts, leading to imprecise forecasts. Lee and Lee (2019) and Lins

et al. (2017) report an enhanced effect of higher quantiles/quartiles of CSR disclosures on

firms’ value and stock returns, respectively. Based on the quartile regression results, it is

evident that firms disclosing a higher level of NFI disclosures will benefit more in terms of

lower future earnings risk than firms making lower NFI disclosures.

Overall, the baseline results show that NFI disclosures have a significant negative impact on a

firm’s future earnings risk. Additionally, the individual dimensions of NFI disclosures show

varying impacts that signify the disparate importance of each dimension for different

stakeholders.

66

Table 4. 4. The impact of NFI disclosure quartiles on firms’ earnings risk Dependent Variable: Earnings Risk

VARIABLE 1 2 3 4

QESGDS2 -0.236*** (0.053)

QESGDS3 -0.340*** (0.091)

QESGDS4 -0.408*** (0.125)

QEDS2 -0.199*** (0.046)

QEDS3 -0.157** (0.066)

QEDS4 -0.279*** (0.063)

QGDS2 -0.095* (0.056)

QGDS3 -0.306*** (0.090)

QGDS4 -0.257** (0.127)

QSDS2 -0.262***

(0.067) QSDS3 -0.437***

(0.093) QSDS4 -0.453***

(0.092) Leverage 0.018*** 0.019*** 0.018*** 0.017***

(0.005) (0.004) (0.005) (0.004) Firm Size -0.273*** -0.210** -0.282*** -0.270***

(0.072) (0.087) (0.075) (0.072) Growth -0.099*** -0.116*** -0.100*** -0.096***

(0.011) (0.011) (0.011) (0.010) Volatility: EPS 0.367*** 0.417*** 0.366*** 0.368***

(0.038) (0.032) (0.036) (0.038) Loss 0.337*** 0.296*** 0.339*** 0.328***

(0.104) (0.112) (0.103) (0.107) Accrual 0.184*** 0.198*** 0.178*** 0.184***

(0.045) (0.051) (0.044) (0.045) ROA -0.011 -0.008 -0.010 -0.010

(0.012) (0.011) (0.011) (0.012) LNAF -0.144* -0.170** -0.145* -0.162**

(0.074) (0.079) (0.078) (0.069) Constant -1.471 -2.117** -1.469 -1.408

(0.897) (1.000) (0.914) (0.910) FE time, industry and index Yes Yes Yes Yes R-Squared 0.0535 0.0531 0.0528 0.0533 Observations 8,885 8,305 8,891 8,850 Number of groups 915 887 914 913

Note: The Table presents the impact of overall and disaggregated NFI disclosure quartiles on a firm’s earnings

risk; Q2, Q3 and Q4 represent second, third and fourth quartiles of overall ESG and sub-dimension disclosure

scores. Fixed effects mean industry and index fixed effects, see Table 4.1 for variable definitins; standard

errors are are in parentheses. *** = p<0.01, ** = p<0.05, * = p<0.1.

67

Notably, the impact of the social dimension is economically more pronounced than the

environmental dimension of NFI disclosures, indicating that the disclosure of social issues

carries more value than the other dimensions of NFI disclosures in minimising the dispersion

of future earnings risk of a firm. These findings suggest that a more transparent information

environment could serve as a tool to minimise agency problems arising from information

asymmetry between corporate managers and shareholders.

Because of the breadth of information covered under its sub-dimensions, ESG disclosures

cater to a large group of stakeholders, including financial analysts. Accordingly, higher

disclosures concerning different dimensions can mitigate different stakeholders’ concerns

related to the social and environmental impact of a firm. Thus, addressing the information

needs of a diverse group of stakeholders would result in a lower chance of customer

complaints, lawsuits by consumer and environmental groups and penalties by regulating

agencies. The resulting improved stakeholder connections and lower concerns about

unexpected stakeholder actions also serve as an instrument for analysts to make more precise

forecasts about future earnings for the NFI reporting firm.

4.4.3. Subsample analysis

The baseline analysis provides sufficient evidence on the significant negative impact of overall

and disaggregated NFI disclosures on a firm’s earnings risk except for the governance

dimension of NFI disclosures that does not exhibit a significant relationship. To further analyse

this relationship, we perform subsample analysis to account for differing firm characteristics

such as size, debt level, income status (negative or positive) and volatility of stock returns

(high vs low volatility). For example, subsequent analysis is undertaken to ascertain whether

the impact of NFI disclosures varies between large and small firms or between high leverage

and low leverage firms24. These analyses are vital given that existing studies report that firm

characteristics such as size, levergage and stock return volatility influence the association

between NFI disclosures and firm risk. For example, Oikonomou et al. (2012) report higher

association between CSR concerns and firms’ financial risk during periods of high stock returns

24 All subsample analysis tables are presented in Appendix A.

68

volatility. Moreover, Sassen et al. (2016) report that highly visible firms show higher

association between ESG disclosures and firms’ financial risk than their low visibility

counterparts. Thus, the proposed subsample analyses could provide additional information

about the association between NFI disclosures and firms’ earnings risk.

Table. A.2 compares the results for large and small firms. We define a firm as large if the value

of its total assets is more than the median value of total assets of the sample firms. According

to the results, larger firms benefit more in terms of reducing their future earnings risk by

reporting NFI disclosures than smaller firms. This effect is more pronounced for

environmental-related reporting and is non-existent for governance disclosures. These

findings complement the literature that shows larger firms have stable earnings (Chung &

Kim, 1994) and they exhibit a higher level of disclosures (Lang & Lundholm, 1996), resulting

in a better information environment. It is essential to understand that, because of their scale

of operations and impact on society, larger firms come under stricter scrutiny from

stakeholders than smaller firms. Hence, this finding provides a useful insight for corporate

managers and stakeholders of large firms to improve the information environment of their

firm through ESG reporting, especially efforts undertaken to mitigate the impact of their

operations on the environment.

We apply a sub-sample analysis to high and low leverage firms. Any firm having a debt ratio

higher than the median debt ratio of the sample firms is classified as a high leverage firm and

otherwise. Table A.3 shows the differences between high and low leverage firms regarding

the impact of NFI disclosures on future earnings risk. The table shows that the overall NFI

disclosures reduce the gap between forecast and actual earnings slightly more for high

leverage firms than for low leverage counterparts. Similarly, this difference is more

pronounced when disaggregared disclosures such as environmental and governance are

considered. This observation is particularly relevant as the literature shows that highly

leverage firms face higher monitoring and, as a result, make better investments in ESG issues

(Iatridis, 2015). Therefore, additional monitoring exercised by creditors could enhance the

quality of disclosures and reduce the future earnings risk of high leverage firms.

Next, we apply another sub-sample analysis to high and low stock return volatility samples.

Firms with stock returns higher than the median return of the sample firms are categorised

69

as high-risk firms and otherwise. Table. A.4 shows that the overall impact of NFI disclosures is

more robust for high stock return volatility firms than for lower volatility returns firm. This

observation is also true for the social dimension of NFI disclosures. These findings agree with

Oikonomou et al. (2012) as they report a higher association between firm CSR disclosures and

firm risk during time of high stock market volatility. Thus, our findings complement earlier

findings on the higher impact of NFI disclosures for financially distressed firms. This finding is

particularly important for financial managers operating in high risk stock markets and in

general for corporate decision makers during market-wide contagion such as global financial

crisis and the current covid-19 pandemic crisis.

Lastly, we perform sub-sample analysis for firms reporting a negative income in the current

year against firms reporting a profit. This analysis is important given that the negative income

alleviates the information asymmetry problem for stakeholders, requiring firms to disclose

more information that can explain the reason for negative income. Table A.5 presents the

results for loss-making and profitable firms. It shows the overall impact of NFI disclosures is

significantly higher for loss reporting firms than for firms reporting a positive income. All the

sub-dimensions of NFI disclosures show a significantly higher coefficient for firms

experiencing negative income than for profit-making firms. This difference is more

pronounced for environmental disclosure than for the other dimensions. This finding suggests

that information asymmetry arising from negative earnings can be reduced by reporting more

information regarding a firm’s efforts on ESG issues.

Overall, the subsample analyses show that the impact of NFI disclosures varies based on firm

characteristics such as size, risk and profitability. However, the study’s results suggest that,

regardless of differing firm characteristics, NFI disclosures improve the information

environment for a firm, resulting in lower future earnings risk. These results provide beneficial

understandings about the impact of NFI disclosures and show that they are useful whether a

firm is experiencing information asymmetry because of financial distress or a large firm is

experiencing higher stakeholder scrutiny because of their operational scale. All firms can

address stakeholders’ concerns by improving their information environment through NFI

reporting.

70

4.4.4. Robustness tests

We perform several robustness tests to check the robustness of the baseline results. First, to

control for country level institutional factors, we rerun the baseline results including country

dummies. The untabulated results show that our results remain qualitatively similar to the

baseline results.

Table 4. 5. The impact of NFI disclosures on stock return volatility Dependent Variable: Stock Return Volatility

VARIABLE 1 2 3 4

ESGDS -0.102*** (0.030)

EDS -0.062*** (0.023)

GDS -0.021

(0.019) SDS -0.066**

(0.026)

Leverage 0.191*** 0.186*** 0.193*** 0.194***

(0.022) (0.023) (0.023) (0.024)

Firm Size -2.334* -2.056* -2.566** -2.247*

(1.222) (1.195) (1.194) (1.190)

Growth -0.366*** -0.358*** -0.378*** -0.356***

(0.082) (0.078) (0.087) (0.083)

Volatility : EPS 1.592*** 1.597*** 1.560*** 1.516***

(0.366) (0.337) (0.359) (0.373)

Loss Dummy 6.874*** 6.014*** 6.880*** 6.793***

(1.174) (0.902) (1.182) (1.159)

Accrual Dummy 0.125 0.245 0.115 0.171

(0.256) (0.233) (0.262) (0.252)

RoA -0.029 -0.031 -0.028 -0.025

(0.078) (0.063) (0.078) (0.076)

LNAF -1.578** -1.054** -1.742** -1.625**

(0.629) (0.471) (0.713) (0.664)

Constant 54.802*** 48.741*** 54.984*** 52.458***

(13.309) (12.948) (13.131) (12.865)

FE time, industry and index Yes Yes Yes Yes

R-squared 0.6009 0.5933 0.6008 0.6017

Observations 8,888 8,312 8,894 8,853

Number of groups 916 888 915 914

Note: The Table presents the impact of overall and disaggregated NFI disclosures on a firm’s stock returns volatility. See Table 4.1 for the variable definitions; standard errors are in parentheses. *** - p<0.01, ** = p<0.05, * - p<0.1.

Second, we use annualised volatility of stocks returns as an alternative proxy for firm risk. This

proxy is widely used as a proxy of firm risk along with the analysts’ forecasts dispersion in the

71

firm information environment literature, e.g., see (Chien & Lu, 2015) and (Kothari et al., 2009).

Hence, evidence of a significant negative relationship between an alternative firm risk proxy

and NFI disclosures will enhance our understanding of the hypothesised relationship between

firm risk and NFI disclosures.

Table 4. 6. The impact of NFI disclosures on firms’ earnings risk (Alternative proxies) Dependent Variable: Earnings Risk

VARIABLE 1 2 3 4

ESGDS -0.009** (0.003)

EDS -0.004* (0.002)

GDS -0.002 (0.005)

SDS -0.009***

(0.003)

Debt to equity 0.042*** 0.041*** 0.042*** 0.043***

(0.007) (0.007) (0.006) (0.007)

Log of Market cap. -0.739*** -0.787*** -0.766*** -0.732***

(0.067) (0.066) (0.063) (0.069)

Sales Growth -0.215** -0.194 -0.214** -0.197*

(0.107) (0.141) (0.105) (0.110)

Volatility: EPS 0.306*** 0.349*** 0.298*** 0.304***

(0.040) (0.032) (0.042) (0.039)

Loss 0.206** 0.181* 0.210** 0.203*

(0.102) (0.106) (0.102) (0.104)

Accrual 0.176*** 0.182*** 0.176*** 0.182***

(0.043) (0.049) (0.044) (0.044)

ROA -0.001 0.002 0.001 -0.000

(0.011) (0.010) (0.011) (0.011) LNAF -0.082 -0.081 -0.101 -0.072

(0.067) (0.080) (0.072) (0.069) Constant 2.767*** 3.044*** 2.865*** 2.661***

(0.631) (0.632) (0.578) (0.657) FE time, industry and index Yes Yes Yes Yes

R-Squared 0.0757 0.0748 0.0744 0.0762

Observations 8,808 8,237 8,814 8,774 Number of groups 912 884 911 910

Note: The Table presents the impact of overall and disaggregated NFI disclosures on a firm’s earnings risk. Debt to equity, log of market capitalisation and sales growth are alternative proxies for leverage, firm size and growth. See Table 4.1 for variable definitions; standard errors in parentheses. *** = p<0.01, ** = p<0.05, * = p<0.1

Table. 4.6 presents the results of the impact of the NFI disclosure proxies used in this study

on stock return volatility. The results show that the overall disclosures along with the

environmental and social dimensions exhibit a significant negative relationship with stock

returns volatility, signifying that firms reporting higher NFI disclosures experience less risk in

72

terms of stock returns volatility. In line with the baseline results, governance disclosures do

not show any significant impact on earnings risk. These findings complement and support our

baseline results regarding the role of NFI disclosures in reducing uncertainty surrounding a

firm’s stock returns.

Table 4. 7. The impact of NFI disclosures on firms’ earnings risk (Alternative proxies) Dependent Variable: Earnings Risk

VARIABLE 1 2 3 4

ESGDS -0.015*** (0.003)

EDS -0.008*** (0.002)

GDS -0.006 (0.006)

SDS -0.013***

(0.003) Firm Size -0.267*** -0.226** -0.326*** -0.278***

(0.073) (0.090) (0.076) (0.075) Volatility: EPS 0.364*** 0.415*** 0.353*** 0.356***

(0.035) (0.030) (0.036) (0.034) Loss dummy 0.306*** 0.274*** 0.311*** 0.302***

(0.100) (0.102) (0.102) (0.105) Accrual dummy 0.188*** 0.206*** 0.189*** 0.195***

(0.046) (0.053) (0.047) (0.048) Log NAF -0.129* -0.169** -0.157* -0.142**

(0.071) (0.077) (0.080) (0.065) Growth_NFS -0.071*** -0.089*** -0.074*** -0.066***

(0.010) (0.011) (0.010) (0.009) RoA_ NFS -0.014 -0.012 -0.013 -0.012

(0.011) (0.010) (0.011) (0.011) Leverage_ NFS -0.023** -0.037** -0.023** -0.026**

(0.011) (0.016) (0.011) (0.012) Growth_FS -0.407*** -0.409*** -0.417*** -0.410***

(0.058) (0.064) (0.058) (0.060) RoA_FS 0.002 0.014 0.002 0.003

(0.019) (0.016) (0.019) (0.019) Leverage_FS 0.000 0.000 0.000 0.000

(0.000) (0.000) (0.000) (0.000) Constant -1.184 -1.795* -0.758 -1.138

(0.888) (1.001) (0.923) (0.903) FE time, industry and index Yes Yes Yes Yes

R-squared 0.0618 0.0602 0.0586 0.0619

Observations 8,885 8,305 8,891 8,850 Number of groups 915 887 914 913

Note: Growth_NFS (FS), RoA_NFS (FS) and Leverage_FS (NFS) represent market-to-book ratio, return on assets and debt ratio of non-financial firms (financial firms), respectively. See Table 4.1 for other variable definitions; standard errors are in parentheses *** = p<0.01, ** = p<0.05, * = p<0.1.

After testing for an alternative proxy for the dependent variable, we substitute some of the

control variables with alternative proxies. First, the proxies for firm size, leverage and growth

73

are changed to the log of market capitalisation, debt to equity, and sales growth,

respectively25. Table. 4.6 shows that the impact of NFI disclosures under the alternative

proxies have similar coefficients to the baseline results, signifying that the results are robust

for different measures of firm size, leverage and growth. In addition, alternative proxies

except for the sales growth variable also show a significant relationship with earnings risk.

Since the full sample analysis also contains financial service sector firms, this study follows

(Bernardi & Stark, 2018) to spilt the leverage, profitability and growth variables into two

components. For example, the debt ratio is split into two components so that each variable

contains only the debt ratios for either financial or non-financial firms. This way, the study

model captures the effect of each component separately and does not affect the primary

relationship between NFI disclosure and earnings risk. Table. 4.7 present the results, where

the main conclusion vis-à-vis the impact of overall NFI disclosures and environmental and

social dimension remains the same. We also perform a spilt sample analysis only on non-

financial firms. Table A.6 results confirm that the main conclusion about the significant

negative impact of NFI disclosures on future earning risk stays the same.

To address the endogeneity concern arising from the bidirectional contemporaneous

association between NFI disclosures and earnings risk, we follow (Oikonomou et al., 2012) to

include one-period lagged independent variables of NFI disclosures and firm characteristics

used in our baseline model. Table. 4.8 show that the main conclusions of the baseline results

do not change, indicating that simultaneity and bidirectional effects do not alter our main

conclusions. However, as opposed to baseline results, the dimension of governance

disclosure scores also show a significant impact on earnings risk alongside environmental and

social disclosures.

The results in Table 4.8 indicate that governance-related disclosures have a lagged effect on

earnings risk. This finding agrees with Godfrey et al. (2009) argument that including lagged

values of NFI disclosures ensures that all the considered information is public knowledge. The

finding of the lagged effect for governance disclosures also aligns with the literature that

25 The baseline results use log of total assets, debt ratio, and market to book ratio as proxies of firm size,

leverage and growth, respectively.

74

shows governance information shows a lagged effect on firm value performance and risk (Y.

Liu, Wei, & Xie, 2014; Oikonomou et al., 2012).

Table 4. 8. The impact of NFI disclosure on firms’ earnings risk (Lagged variables) Dependent Variable: Earnings Risk

VARIABLE 1 2 3 4

Lagged ESGDS -0.018*** (0.004)

Lagged EDS -0.010*** (0.003)

Lagged GDS -0.011**

(0.005) Lagged SDS -0.011***

(0.004)

Lagged leverage 0.020*** 0.021*** 0.020*** 0.019***

(0.002) (0.002) (0.002) (0.002)

Lagged firm size 0.006 0.035 -0.053 -0.026

(0.094) (0.106) (0.090) (0.104)

Lagged Growth -0.059*** -0.066*** -0.064*** -0.057***

(0.007) (0.011) (0.008) (0.007)

Lagged Volatility: EPS 0.064* 0.114** 0.047 0.050

(0.037) (0.049) (0.037) (0.035)

Lagged Loss dummy 0.683*** 0.629*** 0.688*** 0.676***

(0.073) (0.074) (0.074) (0.071)

Lagged Accrual dummy 0.147*** 0.108*** 0.148*** 0.155***

(0.027) (0.030) (0.028) (0.029)

Lagged RoA 0.003 0.002 0.004 0.003

(0.009) (0.009) (0.009) (0.009)

Lagged NAF -0.014** -0.015*** -0.016*** -0.015***

(0.005) (0.005) (0.005) (0.005)

Constant -4.021*** -4.634*** -3.439*** -3.977***

(1.013) (1.114) (1.062) (1.100)

FE time, industry and index Yes Yes Yes Yes

R-squared 0.0515 0.0468 0.0484 0.0488

Observations 8,148 7,587 8,154 8,113

Number of groups 914 882 913 912 Note: The Table presents the impact of overall and disaggregated one-period lagged NFI disclosures on the current level of earnings risk. All the control variables are also one-period lagged. Standard errors are in parentheses *** = p<0.01, ** = p<0.05, * = p<0.1.

In a different disposition, following Bernardi and Stark (2018), we include the lagged

dependent variable as an independent variable to address the lead-lag effect of previous

earnings risk that may affect the contemporaneous association between NFI disclosures and

75

earnings risk. The results in Table A.7 show that adding the lagged dependent variable

produces significant coefficients; however, the baseline results do not change.

The significant lagged effects of both the independent and dependent variable raise concern

of an unobserved heterogeneity problem in our analysis. Studies such as Cai et al. (2016),

Ioannou and Serafeim (2015) and Jo and Harjoto (2011) report a similar concern regarding

NFI disclosures as an endogenous variable resulting in the simultaneous determination of the

negative relationship between NFI disclosure and firm risk. Additionally, we consider the

potential of reverse causality between NFI disclosures and firm risk. For example, the current

or past realisation of firm risk may also influence the future NFI disclosures of a firm to reduce

information asymmetry. Given these concerns, this study uses Blundell and Bond (1998)

system generalised methods of the moment (GMM) estimator to calculate robust estimates

concerning the dynamic relationship between NFI disclosures and a firm’s earnings risk.

According to Blundell and Bond (1998), system GMM is more robust than other GMM

estimators like the first-difference GMM estimator of (Arellano & Bover, 1995) and the non-

linear GMM estimator of (Ahn & Schmidt, 1995). Additionally, the system GMM, like other

GMM estimators, provides the flexibility of using lagged values of the independent and

dependent variables as instruments. Thus, the system GMM becomes a prime instrument of

choice to address endogeneity concerns in the absence of strictly exogenous instruments

(Mohamed, Masih, & Bacha, 2015; Nadeem, 2016).

To that end, we use the STATA’s xtabond2 command developed by (Roodman, 2009) to

explore the dynamic relationship between a firm’s NFI disclosures and earnings risk. Given

the significant lagged effect of the independent variables, this study assumes lagged firm risk

and all the independent variables as endogenous variables. Following the arguments

presented by (Y. Liu et al., 2014) and (Wintoki, Linck, & Netter, 2012) distant historical values

of the dependent variable and firm characteristics are assumed to provide an exogenous

source of variation for the current level of a firm’s earnings risk. Accordingly, lagged values of

t-5 to t-9 are used as instruments in the system GMM estimation. Robust standard errors are

used to address heteroscedasticity concerns.

Table 4.9 presents the results of the system GMM estimates that show the significant

negative impact of the overall and sub-dimension NFI disclosures on earnings risk, confirming

76

that firms reporting more NFI disclosures are better at reducing the gap between their

forecast and actual earnings. The results of the aggregate NFI disclosures presented in Table

4.9 are identical to our baseline results. However, the results of the environmental (social)

dimension exhibits a moderately higher (lower) coefficient. The governance dimension also

has a significant adverse effect as was present in the one-period lagged specification, but the

contemporaneous baseline model was unable to show such a relationship. These results

reinforce our earlier argument that reporting information related to ESG issues significantly

reduces information asymmetry among analysts, resulting in more precise forecasts that

ultimately reduce a firm’s earnings risk.

Table 4. 9. The impact of NFI disclosure on firms’ earnings risk (System GMM estimator) VARIABLE Dependent Variable: Earnings Risk

1 2 3 4

Lagged earnings risk 0.426*** (0.110)

0.384*** (0.109)

0.372*** (0.118)

0.430*** (0.107)

ESGDS -0.021***

(0.008)

EDS

-0.018*** (0.006)

GDS

-0.029** (0.014)

SDS

-0.011** (0.005)

Control variables Yes Yes Yes Yes

FE Year and Industry Yes Yes Yes Yes

AR (1) p- value 0.000 0.000 0.000 0.000

AR (2) p- value 0.186 0.266 0.309 0.124

Hansen test of overid p- value 0.225 0.631 0.054 0.210

Difference-in-Hansen Test for exogeneity 0.230 0.289 0.226 0.490

Observations 6,072 5,600 6,081 6,046

Number of id 859 817 859 855

Note: Lagged earnings risk is a one-period lagged dependent variable. AR(1) and AR(2) represent the first and second-order autocorrelation tests for the first-difference residuals. The Hansen test and difference-in-Hansen test show the test statistics for over-identification and exogeneity of instruments, respectively. Standard are errors in parentheses *** = p<0.01, ** = p<0.05, * = p<0.1

Table 4.9 also contains the results of specification tests of autocorrelation AR(2), Hansen test

of over-identification and difference-in-Hansen test of exogeneity. The p-values of AR(2)

range from 0.124 to 0.309, suggesting that the null hypothesis concerning second-order

autocorrelation is not rejected for all models. Except for governance disclosures model with

a p-value of 0.054, the highly insignificant p-value for the Hansen tests signifies the validity of

our instruments for the models. We also test another GMM estimator assumption that the

77

correlation between the endogenous variables and unobserved fixed-effects is constant over

time. This system GMM estimator assumption provides the flexibility of adding lagged

differences of endogenous variables as instruments in the system GMM estimation process

and can be conveniently tested using the difference-in-Hansen test of exogeneity

(Eichenbaum, Hansen, & Singleton, 1988; Wintoki et al., 2012). The highly insignificant p-

values of difference-in-Hansen test for all models suggest strong non-rejection of the null

hypothesis concerning the exogeneity of the additional subset of instruments. Overall, the

system GMM specification tests support our earlier conclusion about the significant negative

relationship between NFI disclosures and earnings risk.

4.5. Sensitive Industries Analysis

The above analysis shows the impact of NFI disclosures on a firm’s future earnings risk without

considering the inherent negative impact of its business operations on the society and

environment. However, prior studies on NFI disclosures maintain that, because of the harmful

effects of their business activities on the society and environment, sensitive industry firms are

subject to higher reputation risk than their counterparts. Therefore, to increase the legitimacy

of their business operations, firms in sensitive industries make more NFI disclosures than their

counterparts (Garcia et al., 2017; Grougiou, Dedoulis, & Leventis, 2016; Kilian & Hennigs,

2014). Additionally, previous studies show that reporting more information about business

activities provides more benefits for sensitive industry firms in terms of higher market value

(De Klerk, De Villiers, & Van Staden, 2015; Reverte, 2016), lower cost of capital and debt

(Gerwanski; Reverte, 2012), and a firm’s market risk (Jo & Na, 2012; Zeng et al., 2020). We

separately analyse the impact of NFI disclosures on future earnings risk for sensitive and non-

sensitive industry firms to ascertain the differences that may arise because of a company’s

involvement in ESG-sensitive industries.

Before analysing the differences between the impact of NFI disclosures on a firm’s future

earnings risk between sensitive and non-sensitive industry firms, we first test the second

hypothesis on differences in the level of NFI disclosures between the two firm types of firms.

This analysis extends the existing evidence of higher NFI disclosures by sensitive industry firms

for an international sample, resulting in far-reaching understanding about the relationship

between involvement in an ESG sensitive business and NFI disclosure practices. To the best

78

of our knowledge, such evidence is non-existent in the literature. Following (Baron et al.,

2011; Garcia et al., 2017), we treat a firm as a sensitive firm if it belongs to the alcohol,

tobacco, gambling, weapon production, adult entertainment, oil, gas and consumable fuels,

metal and mining, paper and forest products, chemical, construction materials, or energy

businesses.

4.5.1. Level of NFI disclosure: Sensitive v/s non-sensitive industry firms

To explore the differences in the level of NFI disclosures between sensitive and non-sensitive

industry firms, we use the propensity score estimation technique to draw a matched sample

of non-sensitive firms that are comparable to sensitive industry firms. There are several

methods to perform propensity score estimation such as nearest neighbour matching, Kernel

matching, radius matching and stratification matching. However, these matching techniques

may suffer from several drawbacks. For example, Guo and Fraser (2014) maintain that

propensity matching techniques may require a large number of control group participants to

provide reliable matching estimates. Additionally, Randolph and Falbe (2014) report that,

depending on the matching technique (one-to-one or one to many matches, exact, nearest

neighbour or optimal matching; many participants in the control group may not be used for

matching because they do not match a participant in the treatment group. Consequently,

there is a potential risk of losing many observations from the treatment group, resulting in

the loss of valuable information and generalisability of the treatment effect (Austin, 2017).

We use the propensity score weighting (PSW) technique introduced by (McCaffrey, Ridgeway,

& Morral, 2004) to overcome the problems mentioned above with propensity score matching

techniques. This method of propensity score estimation is more efficient than conventional

matching techniques and provides better results when the treated and control groups have

quantitatively different characteristics. For example, in our sample, sensitive industry firms

show a higher financial opaqueness than non-sensitive firms. Firm size for both types of firms

also differs significantly26. Hence, accounting for these divergences is necessary because

existing studies show that, along with other firm characteristics, earnings management and

26 In untabulated results, we performed the mean comparison test of firm size, volatility of EPS and return on

assets. All the mean comparison tests show both types of firm are significantly different from each other.

79

firm size are dominant determinants of NFI disclosures (Abdi, Kacem, & Omri, 2018; Fahad &

Nidheesh, 2020; Ismail & El‐Shaib, 2012).

Using generalised boosted models, the PSW technique eliminates the differences between

the covariates (firm characteristics here) of control and treatment groups, resulting in the

formation of a control group that is comparable to the treatment group. Therefore, the

influence on NFI disclosures because of being a sensitive industry firm relative to non-

sensitive industry firm can be captured in a propensity-matched sample under the PSW

method. McCaffrey et al. (2013) suggest that, if there is imbalance and insufficient overlap

between the two groups, the casual estimation may produce inefficient results. To check

these conditions, we present overlap plot and balance tables of PSW estimation in Appendix

B. The overlap boxplot in Figure B.1 shows a sufficient overlap between the sensitive and non-

sensitive industry groups to indicate that the treatment and control groups have firms that

are identical to each other in terms of firm-specific characteristics. Additionally, we use the

standardised effect size differences (ES) statistic to summarise the differences in the means

of covariates for treatment and control groups. The ES statistic is calculated as:

ES Statistic = 𝐷𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 𝑖𝑛 𝑡ℎ𝑒 𝑚𝑒𝑎𝑛𝑠 𝑜𝑓 𝑎 𝑐𝑜𝑣𝑎𝑟𝑖𝑡𝑒𝑠 𝑓𝑜𝑟 𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡 𝑎𝑛𝑑 𝑐𝑜𝑛𝑡𝑟𝑜𝑙 𝑔𝑟𝑜𝑢𝑝𝑠

𝑆𝑡𝑎𝑛𝑑𝑎𝑟𝑑 𝑑𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛 𝑜𝑓 𝑡ℎ𝑒 𝑐𝑜𝑣𝑎𝑟𝑖𝑎𝑡𝑒

Table 4.10 shows the ES statistic before and after the PSW, which indicates the statistically

significant differences in firm size and EPS volatility between the control and treatment

groups in the unweighted sample. For example, the unweighted sample scores for firm size

are statistically significant difference of 0.075 between two types of firm. However, the

differences do not exist in the weighted samples scores, indicating that PSW successfully

eliminated them. Based on these results, it is suggested that there is sufficient overlap and

balance between the weighted samples of sensitive and non-sensitive industry firms,

indicating that PSW successfully creates a control sample of non-sensitive firms that is

analogous to the treatment (sensitive firms) sample.

After the estimation of the PSW matched samples, we use the average treatment effect (ATE)

method to estimate the treatment effects of sensitive industry status on the NFI disclosure

practices of the treatment group firms. The results in Table 4.11, Panel A, show that sensitive

industry firms make higher aggregate NFI disclosures than their non-sensitive industry

80

counterparts. For example, a typical sensitive industry firm, as defined earlier, would make

2.78 points more ESG disclosures than a typical non-sensitive firm. Similarly, sensitive industry

firms outperform their counterpart in the environmental and social dimesion of NFI

disclosure.

Table 4. 10. Balance table of ES statistic comparison Unweighted Score Weighted Score

Covariate ES Statistics T-statistics p-value ES Statistics T-statistics p-value

Accrual dummy 0.023 0.970 0.332 0.007 0.293 0.770

Leverage -0.026 -1.075 0.283 -0.003 -0.125 0.900

Growth 0.027 1.056 0.291 -0.005 -0.223 0.824

Loss dummy 0.071 2.842 0.004 0.013 0.565 0.572

Firm Size -0.075 -3.045 0.002 -0.017 -0.676 0.499

NAF 0.027 1.128 0.259 0.004 0.165 0.869

RoA -0.037 -1.299 0.194 -0.001 -0.032 0.975

Standard dev: EPS -0.062 -2.593 0.010 -0.013 -0.530 0.596

Note: Table presents the ES statistics for unweighted and weighted samples. Values highlighted in red colour

present significant imbalance between weighted and unweighted samples.

However, no difference exists between the two types of firm for governance disclosures. We

perform a comparative analysis to ascertain the efficiency of ATE estimation using a matched

sample approach compare with the unmatched sample. Table 4.11, Panel B, shows the effect

of the sensitive industry type on NFI disclosures. It is observable that the unmatched sample

captures a relatively higher effect of sensitive industry status on all dimension of NFI

disclosures than the matched sample. For example, the effect of sensitive industry type on

social disclosure scores and the unmatched sample is 1.85 times higher than similarly

matched sample estimate (5.456 vs 2.947). This finding supports the fact that the matched

sample approach provides more precise estimates than to the unmatched sample.

To check the robustness of the ATE estimates, we perform a doubly robust analysis. According

to Cuong (2013) and McCaffrey et al. (2013), including the pre-treatment covariate (firm

characteristics, in this case) in the weighted ATE estimation produces “doubly robust”

estimates that are more efficient than the simple ATE estimator. Table 4.11, Panel C, shows

that ATE is more pronounced when firm characteristics and time-effects are added to the

regression model. The study’s findings on the impact of sensitive industry status on NFI

disclosures supports the literature (Garcia et al., 2017; Grougiou et al., 2016; Kilian & Hennigs,

2014) that report relatively more disclosures by sensitive industry firms than their non-

sensitive counterparts.

81

Table 4. 11. The effect of sensitive industry membership on NFI disclosure levels DEPENDENT VARIABLE ESGDS EDS GDS SDS

1 2 3 4

Panel A – ATE Sensitive industry dummy 2.776** 3.250*** 0.792 2.947**

(1.113) (1.177) (0.640) (1.411) Constant 38.979*** 34.451*** 57.172*** 36.831***

(0.444) (0.551) (0.247) (0.499) Observations 9,813 9,129 9,817 9,757 R-squared 0.009 0.009 0.002 0.007 Panel B – Linear regression without PSW

Sensitive industry dummy 4.581*** 4.851*** 1.560*** 5.456*** (0.342) (0.436) (0.213) (0.427) Panel C – Doubly robust Sensitive industry dummy 3.138*** 3.968*** 1.268** 3.390** (1.152) (1.251) (0.586) (1.490) Firm Size 2.305*** 2.292*** 1.759*** 1.620*** (0.378) (0.443) (0.176) (0.567) Growth -0.232 0.114 0.123 -0.551** (0.208) (0.169) (0.082) (0.242) Volatility: EPS 0.063 0.746*** -0.778*** -0.969*** (0.186) (0.255) (0.098) (0.218) Loss dummy 0.330 -0.028 0.211 0.614

(0.886) (1.045) (0.536) (1.164) Accrual dummy -0.566 -0.492 -0.825** -1.128 (0.749) (0.796) (0.369) (0.924) ROA 0.052 0.016 0.012 0.049

(0.066) (0.077) (0.042) (0.085) LNAF 3.570*** 2.448** 2.118*** 4.919*** (0.839) (0.969) (0.487) (1.015) Constant -1.534 -2.948 29.113*** -1.795 (3.724) (4.294) (1.988) (5.356) FE time, industry and index Yes Yes Yes Yes Observations 8,911 8,330 8,917 8,876 R-squared 0.164 0.107 0.213 0.140

Note: The Table presents the ATE results for PSW and unweighted samples in Panels A and B, respectively. Doubly robust estimates are presented in Panel C. The sensitive industry dummy equals 1 for a sensitive firm, 0 otherwise. See Table 4.1 for other variable definitions. Standard errors are in parentheses *** = p<0.01, ** = p<0.05, * = p<0.1.

4.5.2 NFI disclosures and future earnings risk: Sensitive versus non-sensitive industry firms

To test the third hypothesis vis-à-vis the impact of NFI disclosures of sensitive industry firms

compared with non-sensitive counterparts, we use a split-sample approach that serves a dual

purpose. First, the impact of NFI disclosures is observed exclusively for each sub-sample and,

second, a qualitative comparison of results may reveal if any difference exists concerning the

impact of NFI disclosures on future earnings risk for the two types of firm.

82

Table 4.12 presents the results of the fixed-effects regression of the non-sensitive and

sensitive industry firms in Panels A and B, respectively. For the non-sensitive industry results,

the aggregate and disaggregated components of NFI disclosures exhibit a significant negative

impact on future earnings risk. These results agree with our full sample results and,

expectedly, show a quantitatively lower impact of environmental disclosures on future

earnings risk. Interestingly, governance disclosures show a significant negative impact on

future earnings risk that is not present in the full sample analysis. This observation indicates

that governance disclosures increase the transparency of the information environment for

non-sensitive industry firms. Moreover, adding time fixed effects reduces the statistical

significance of NFI disclosure estimates but the governance and social dimensions remain

significant at the 5% significance level.

Table 4.12, Panel B, presents the results for sensitive industry firms that exhibit a slightly

higher impact of NFI disclosures on future earnings risk when overall disclosures are

considered. The analysis of disaggregated disclosures reveals interesting results. First, the

environmental dimension shows a quantitatively higher coefficient than the same dimension

for non-sensitive industry firms. This observation indicates that reporting information related

to environmental impacts benefits sensitive industry firms more than non-sensitive

counterparts regarding the reduction of dispersion between forecast and actual earnings.

Secondly, governance disclosures do not exhibit any significant relationship with future

earnings risk for sensitive industry firms, signifying that the impact of governance disclosures

are limited to non-sensitive industry firms. Lastly, the social dimension shows no difference

between the two types of industry, indicating that social issues carry equal weight regardless

of a firm’s industry affiliation. These results are sensitive to time fixed effects except for the

social disclosure scores.

4.5.3 NFI disclosures and future earnings risk: Interaction approach

The split sample method provides important insights into the impact of NFI disclosures for

sensitive and non-sensitive industries. However, the method has two limitations. First, under

fixed-effects model estimation, it is problematic to test whether the coefficients of NFI

disclosures from the two samples are statistically different. Second, the group residual

variance may differ between the sensitive and non-sensitive industry groups. For example,

83

the relationship between firm size and earnings risk may differ for the two types of firm,

causing the group variance to differ. To this end, we apply the interaction term technique to

capture the impact of sensitive and non-sensitive industry firms in a single estimation model.

Accordingly, we create four interaction terms by multiplying the sensitive industry dummy

with the overall and disaggregated NFI disclosure scores. Next, the F-test is conducted to test

the equality between the coefficients of NFI disclosures and interacted NFI disclosure

variables. Thus this method accounts for both limitations present in the spilt sample method.

Table 4. 12. The impact of NFI disclosures on firms’ earnings risk – Split sample approach Dependent Variable: Earnings Risk

VARIABLE 1 2 3 4 1 2 3 4 Panel A - Non-sensitive firms Panel B- Sensitive firms ESGDS -0.011*** -0.019***

(0.003) (0.005) EDS -0.003 -0.012**

(0.002) (0.005) GDS -0.015*** -0.012

(0.005) (0.008) SDS -0.018*** -0.021***

(0.002) (0.003)

Leverage 0.018*** 0.018*** 0.018*** 0.018*** 0.035*** 0.035*** 0.034*** 0.035***

(0.004) (0.004) (0.004) (0.004) (0.006) (0.006) (0.006) (0.005)

Firm Size -0.215*** -0.247*** -0.239*** -0.190*** -0.153 -0.177 -0.187 -0.141

(0.071) (0.069) (0.071) (0.064) (0.254) (0.256) (0.261) (0.248)

Growth -0.095*** -0.097*** -0.098*** -0.094*** -0.213*** -0.212*** -0.216*** -0.214***

(0.011) (0.011) (0.011) (0.012) (0.041) (0.043) (0.043) (0.042)

Volatility: EPS 0.145*** 0.134*** 0.197*** 0.228*** 0.006 -0.029 0.062* 0.089**

(0.023) (0.027) (0.025) (0.019) (0.038) (0.040) (0.032) (0.042)

Loss Dummy 0.359*** 0.364*** 0.361*** 0.358*** 0.113 0.109 0.118 0.129

(0.120) (0.122) (0.118) (0.119) (0.161) (0.163) (0.163) (0.158)

Accrual dummy 0.187*** 0.189*** 0.187*** 0.183*** 0.217** 0.220*** 0.234*** 0.231***

(0.053) (0.053) (0.053) (0.052) (0.084) (0.083) (0.084) (0.084)

ROA -0.011 -0.010 -0.011 -0.011 0.005 0.005 0.005 0.007

(0.009) (0.009) (0.009) (0.009) (0.021) (0.021) (0.021) (0.020)

Log NAF -0.232*** -0.244*** -0.234*** -0.216*** 0.058 0.047 0.016 0.090

(0.061) (0.064) (0.065) (0.057) (0.144) (0.144) (0.153) (0.150)

Constant -1.266 -1.280 -0.605 -1.314 -2.373 -2.498 -2.047 -2.587

(0.915) (0.925) (0.856) (0.862) (2.376) (2.391) (2.335) (2.359) FE Year and industry

Yes Yes Yes Yes Yes Yes Yes Yes

R-Squared 0.0483 0.047 0.0484 0.0532 0.0559 0.0539 0.0523 0.0595

Observations 6,512 6,512 6,512 6,512 1,775 1,775 1,775 1,775

Number of groups 704 704 704 704 182 182 182 182

Note: The Table presents the results of NFI disclosures on earnings risk for non-sensitive (sensitive) firms in

Panel A (B). See Table 4.1 for other variable definitions. Standard are errors in parentheses *** = p<0.01, ** =

p<0.05, * = p<0.1.

Table 4.13 separately shows the coefficients of the overall and sub-dimension NFI disclosures

for the total sample and sensitive industry firms. In the first column, the overall NFI

disclosures coefficient of sensitive industry firms is statistically and economically more

pronounced than the non-sensitive firms (-.015 vs -0.06). However, the F-statistic shows that

84

these coefficients are not statistically different from each other (p-value 0.154), contradicting

the notion that the impact of the overall NFI disclosures on future earnings risk is higher for

sensitive industry firms than for non-sensitive counterparts.

Table 4. 13. The impact of NFI disclosures on firms’ earnings risk – Interaction approach Variable Dependent Variable: Earnings Risk

ESGDS -0.006** (0.002)

EDS

0.004** (0.002)

GDS

-0.037***

(0.004)

SDS

-0.020***

(0.002)

SenESGDS -0.015***

(0.005)

SenEDS

-0.011** (0.004)

SenGDS

-0.005 (0.008)

SenSDS

-0.008** (0.004)

Sensitive industry dummy 0.655*** (0.243)

0.361** (0.178)

0.313 (0.473)

0.392** (0.185)

Leverage 0.005** (0.002)

0.005** (0.002)

0.004** (0.002)

0.006*** (0.002)

Firm Size -0.108***

(0.024) -0.134***

(0.024) -0.071***

(0.023) -0.099***

(0.023) Growth -0.060*** -0.062*** -0.053*** -0.058*** (0.012) (0.012) (0.012) (0.012)

Volatility: EPS 0.660*** (0.021)

0.663*** (0.021)

0.691*** (0.021)

0.686*** (0.021)

Loss dummy 0.729*** (0.121)

0.728*** (0.122)

0.683*** (0.121)

0.691*** (0.120)

Accrual dummy 0.178*** (0.060)

0.193*** (0.060)

0.155*** (0.060)

0.164*** (0.060)

ROA -0.011 (0.007)

-0.012 (0.007)

-0.010 (0.007)

-0.012 (0.007)

LNAF -0.222***

(0.068) -0.253***

(0.068) -0.221***

(0.068) -0.164** (0.068)

Constant -2.141***

(0.330) -2.204***

(0.325) -0.601 (0.377)

-1.765*** (0.328)

R-squared 0.469 0.468 0.475 0.476 FE time and Industry Yes Yes Yes Yes Observations 8,223 8,223 8,223 8,223 F- STATISTIC 2.03 8.39 9.53 4.40 p-value 0.1541 0.0038 0.0020 0.036

Note: Table presents the results of NFI disclosures and for sensitive and non-sensitive industry firms. SenESGDS, SenEDS, SenGDS and SenSDS represent overall and disaggregated disclosures for sensitive industry firms. Sensitive is a dummy variable for sensitive industry firms. Refer to table 4.1 for other variable description. Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

85

Nevertheless, the environmental dimension results in the second column show that sensitive

industry firms disclosing environment-related information benefit more than a non-sensitive

counterpart (-.011 vs 0.004) in terms of reducing future earnings risk. The significant value of

the F-statistic further confirms that the impact of environmental disclosures for both types of

firm is statistically different concerning the future earnings risk. Last two columns in Table

4.13 present the results for the governance and social dimensions. The results show that non-

sensitive industry firms benefit more than sensitive counterpart in terms of reducing the gap

between forecast and actual earnings by reporting information related to governance and

social issues. The significant value of the F-statistic confirms that these differences are

statistically significant at conventional levels of significance. Lastly, the inclusion of time and

industry fixed effects does not change our earlier findings, indicating that these results are

robust to these fixed effects.

Overall, the split sample and interaction terms approaches present similar results on the

relationship between NFI disclosures and earnings risk. These findings indicate that

environmental disclosures bring higher benefits to sensitive industry firms, which agrees with

the literature. On the other hand, the governance disclosure impact is limited to only non-

sensitive firms; this signifies that these firms can further improve their information

environment by disclosing governance-related information.


This chapter presents empirical evidence on the impact of NFI disclosures on a firm’s earnings

risk. The results show that overall ESG disclosures show a significant negative relationship

with future earnings risk. The disaggregated dimensions of social and environmental

disclosures also exhibit a negative impact on earnings risk. The relationship between NFI

disclosures is quantitatively more pronounced for firms reporting higher NFI disclosures. Sub-

sample analysis shows that the impact of NFI disclosures varies based on firm characteristics

such as size, risk and profitability.

The robustness analysis shows that the study’s results are robust for alternative proxies of

firm risk and control variables such as firm size, growth and leverage. This study explores the

dynamic nature of the relationship between NFI disclosures and earnings risk to address

concerns such as omitted variable and reverse causality bias. The dynamic analysis using

86

system GMM estimation shows that the baseline results do not suffer from omitted variable

or reverse causality biases.

This chapter presents evidence of sensitive industry membership on NFI disclosures practices

and shows that sensitive industry firms reporting on ESG issues outperform non-sensitive

counterparts in overall and disaggregated disclosures. However, the resulting impact of the

higher NFI disclosures by sensitive industry firms does not show a greater reduction in

earnings risk except for environmental disclosures. The findings for non-sensitive firms’

disclosures on the governance and social dimensions produce relatively better outcomes in

reducing the earnings risk than their sensitive counterparts.

87

Chapter 5

Mandatory NFI Disclosures and Firm Risk

5.1. Introduction

This chapter presents the empirical analysis of the impact of mandatory non-financial

information (NFI) reporting regulations on the level of NFI disclosures of the regulated firms.

The chapter also presents a difference-in-difference analysis to ascertain the change in the

impact of NFI disclosures on a firm’s earnings risk before and after the mandating of NFI

reporting. To that end, the chapter uses the recent EU directive 2014/95/EU as an example

of mandated NFI reporting regulations. To test the study’s fourth hypothesis that assumes a

positive impact of NFI reporting regulations on the level of NFI disclosures, we use a

propensity-matched sample approach to estimate the impact of the EU directive on the level

of aggregate and disaggregated NFI disclosures for S&P350 EU index firms. The chapter

estimates the changes in the impact of NFI disclosures on earnings risk by using the

difference-in-differences technique that helps capture the impact before and after the

enactment of the NFI regulations. Different robustness tests are performed to evaluate the

validity of the results.

The chapter is structured as follows: Section 5.1 introduces NFI directive 2014/95/EU and

discusses the recent evidence of its acceptance in EU member states. Section 5.2 explores the

causal impact of the EU directive on NFI disclosures of European firms. Section 5.3 provides

evidence regarding increased NFI disclosures impact on earnings risk in the post-regulation

period. Section 5.4 summarises the chapter.

5.2. The European Union’s NFI Directive 2014/95/EU

The European Parliament enacted directive 2014/95/EU on 22 October 2014. It mandates the

disclosure of non-financial and diversity information for large27 entities and business groups

operating in the EU’s jurisdiction. The European Parliament allowed member states to

transpose the new directive into their laws by December 2016 to ensure compliance from

calendar year 2017 or during calendar year 2017. Recent studies show widespread

27 Companies with total balance sheet, net turnover and average number of employees above EUR 400,000, EUR

800,000 and 500, respectively.

88

compliance with the new EU directive. For example, using a sample from 10 European

countries and seven different industries, Manes-Rossi et al. (2018) report a higher level of

compliance in terms of content coverage by the 50 largest companies. They also report that

the content requirements of the new EU directive are exhaustive compared with the Global

Reporting Initiative (GRI) and International Integrated Reporting Framework, making the EU

directive qualitatively more robust. Testarmata et al. (2020) show that harmonisation of NFI

reporting in the EU has improved comparability, consistency and transparency among

European companies.

For content disclosures, directive 2014/95/EU requires firms to report information on six

broad topics: 1) environmental; 2) social; 3) employee matters; 4) respect for human rights;

5) anti-corruption; and 6) bribery matters, as a minimum requirement of NFI disclosures. Like

other NFI reporting frameworks, this framework does not provide any NFI performance

metrics. However, the six major topics cover almost all facets of ESG related activities.

Therefore, the EU directive is expected to bring significant changes in the quantity and quality

of disclosures for large firms operating in the EU. Recent studies also suggest a likely increase

in the quality of NFI reporting after the implementation of directive 2014/95/EU, e.g.e, see

(Arvidsson, 2019; Dumitru, Dyduch, Gușe, & Krasodomska, 2017).

Mandating by the EU directive offers a unique opportunity to analyse the impact of

mandatory NFI reporting on the level of NFI disclosures at a regional level that has not been

studied so far. Additionally, any subsequent increase in the quality of NFI disclosures in the

post-directive period could be further explored by analysing the change in the effect of NFI

disclosures on a firm’s information environment. As a result, this study treats directive

2014/95/EU as an exogenous shock for large European firms and measures its impact on the

level of aggregate and sub-dimension level of ESG disclosures. Moreover, this study attempts

to ascertain the difference between the magnitude of NFI disclosures impact on a firm’s

earnings risk before and after the EU directive.

5.3. The Impact of Directive 2014/95/EU on NFI Disclosures

Section 4.1 shows an increasing trend of NFI disclosures during the sample period. This

observation raises an empirical challenge concerning the measurement of real impact of

disclosure regulations on the level of disclosures for the targeted firms. Different firm

89

characteristics and industry type can influence the level of disclosures before and after the

mandating regulation. It is very important to control for factors that may induce a change in

disclosure level. We use a propensity matched sample approach to draw a control group of

firms from the sample firms. This approach offers the benefit of creating an analogous control

sample using propensity score matching based on the observable characteristics of a sampling

unit – firms, in this case. Accordingly, we use firm characteristics, such as firm size, growth,

profitability, earnings volatility, leverage, pre-existing ESG disclosures, financial opaqueness,

the number of analysts following a firm, and industry type, as covariates to construct a

comparable control group from the sample firms. Since all S&P350 index firms in the sample

fall under the definition of large firms, all S&P350 index firms qualify to be part of the

treatment group. All non-S&P350 index companies are identified as the control group.

We use the propensity score weighting (PSW) technique to generate a group of control firms.

We use the PSW method because it provides a better matching estimator than other

matching techniques. As discussed in section 4.5.1, the efficiency of a control group based on

the PSW technique can be measured by checking the balance and overlap conditions. It is

important to understand that, traditionally, a statistically significant standardised effect size

differences (ES) value of at least 0.25 is assumed to be a significant difference, i.e., any ES

statistic value less than 0.25 does not affect the balancing property of PSW. Table 5.1 shows

the balance scores between treatment and control groups before and after the weighting

process in Panels A and B, respectively.

The unweighted sample (see Table 5.1 Panel A) shows significant differences between the

treated and control firms for several variables. For example, the ESG disclosure scores,

earnings volatility and the number of analysts following a firm, exhibit substantial differences

(absolute ES > 0.25) between the unweighted treatment and control samples. This finding

confirms that the use of an unweighted sample can produce biased estimates as sample firms

in the control and treatment groups are inherently different based on several characteristics.

However, the weighted result presented in Table 5.1, Panel B, exhibits no significant

difference between the treatment and control samples. Thus, the PSW estimation process

successfully eliminates the inherent differences in the sample firms, resulting in the formation

of an analogous control sample that can be used for causal effect estimation in the next step.

90

Figure B.2 (Appendix B) shows that there is a sufficient overlap between the treatment and

control groups, indicating that the overlap condition of the PSW is also satisfied. These

observations show that PSW successfully generated a comparable group of control firms and

provides the perfect environment to measure the actual effect of an external shock on the

treated companies’ NFI disclosure practices.

Table 5. 1. The balance table of ES statistics Covariates Panel A - Unweighted Score Panel B- Weighted Score

ES Statistics T-stat p-value ES Statistics T-stat p-value ESGDS 0.5720 29.04 0.000 0.180 6.826 0.000 NAF 0.6780 34.16 0.000 0.203 7.796 0.000 Accrual dummy -0.1180 -6.003 0.000 -0.046 -1.882 0.060 Leverage 0.0420 2.001 0.045 -0.002 -0.086 0.931 Growth 0.0120 0.575 0.565 -0.039 -1.772 0.076 Loss dummy 0.0880 4.222 0.000 0.041 1.647 0.100 Firm Size 0.2170 9.919 0.000 0.055 2.276 0.023 RoA -0.0250 -1.106 0.269 -0.006 -0.258 0.797 Volatility: EPS -0.5320 -28.98 0.000 -0.216 -7.968 0.000

Note: Table 5.1 presents the ES statistics for unweighted and weighted samples. ES statistics values greater than

0.25 present a significant imbalance between weighted and unweighted samples. ESGDS is the overall ESG

disclosures score; NAF is the number of analysts following a firm; the Accrual dummy is indicator variable for financial reporting quality; Leverage is debt to total asset ratio; the Loss dummy is the indicator variable for negative income; Firm size is the log of total assets; RoA is net income divided by total assets; and Volatility: EPS is the volatility of earnings.

Post PSW estimation, we estimate the average treatment effect (ATE) of the EU directive on

the NFI disclosures of the treated firms. For this, the treatment variable is constructed by

interacting the treated dummy (the S&P 350 index) with a regulation period (post-2017)

dummy. The pre- and post-regulation periods are defined as 2008-2016 and 2017-2018,

respectively. Therefore, the post-regulation interaction term captures the effect of regulation

on NFI disclosures for the treated firms. The estimates of the interaction terms allow us to

test the study’s fourth hypothesis that assumes a significant difference between treated and

non-treated firms. Table 5.2, Panel A, shows the ATE results concerning the regulations’

impact on the overall and disaggregated NFI disclosures when the reference group is the pre-

directive disclosure score of S&P 350 index firms. The ATE results show a statistically

significant positive regulation effect on the overall and sub-dimension NFI disclosures with

social disclosures displaying the greatest increase followed by the environment and

governance disclosures. In the economic context, the cumulative effect of regulations on the

overall ESG disclosures during the post-regulation period was an increase of 12.05%. The ATE

ranges from 6.51 % to 14.84% for the disaggregated dimensions governance and social

disclosures, respectively. These findings show that the mandating directive has substantially

91

increased the magnitude of NFI disclosures of large firms in the EU, which signifies the

importance of regulations that mandate NFI disclosures.

Next, we expand the reference group from the pre-directive S&P350 index companies to all

non-S&P350 index group companies that are part of the study sample. This expansion of the

reference group serves two objectives. First, the inclusion of the non-S&P 350 index in the

reference group facilitates a comparison with a diverse set of firms from all over the world.

Second, non-S&P 350 companies provide disclosure observations in the post-directive period

that serve as a comparison point for the treated firms during that period. The latter objective

also controls for a potential time-trend effect that we observe in the disclosure practices of

the sample firms.

Table 5.2, Panel B, presents the results of the expanded reference group setting which also

show a significant positive impact of the EU directives on NFI disclosures of the treated firms

compared with the non-treated group. This comparison shows that the impact of the EU

directive on treated firms is not sensitive to the composition of the reference group and is

independent of the current momentum of NFI disclosures around the world. Additionally, the

impact of the EU directive is more pronounced in economic terms when the reference group

includes geographically diverse firms, indicating that some of the sample firms are lagging in

NFI reporting despite the heightened importance of NFI disclosures. Furthermore, we

perform a doubly robust analysis by including the firm characteristics, time and industry

effects, in the basic model. Table 5.2, Panel C, presents the doubly robust ATE estimates that

confirm the robustness of our results regarding the impact of the EU directive on NFI

disclosures in the presence of additional controls such as firm characteristics and industry

effects.

Thus, the results signify the importance of NFI disclosure regulations and indicate that such

regulations help increase the socially responsible activities of the target firms. The study’s

results complement and advance the literature that reports the effectiveness of mandatory

disclosure regulations in increasing NFI disclosures in different parts of the world. For

example, using a worldwide sample of 58 countries, Ioannou and Serafeim (2011) report a

significant positive effect of CSR reporting regulations on the amount of NFI information

92

reporting measured by different country-specific measurements such as social responsibility

sustainable development, employee training and corporate board ethical practices.

Table 5. 2. The impact of the EU directive on NFI disclosures 1 2 3 4 VARIABLES ESGDS EDS GDS SDS Panel A Interaction 4.965*** 3.452*** 3.687*** 6.301***

(0.530) (0.909) (0.444) (0.601) Constant 41.189*** 35.258*** 56.629*** 42.451***

(0.851) (0.969) (0.581) (0.940) Observations 3,408 3,335 3,406 3,395 R-squared 0.019 0.006 0.019 0.023 Panel B Interaction 6.642*** 3.899*** 3.258*** 10.961***

(0.664) (0.97) (0.456) (0.751) Constant 39.512*** 34.811*** 57.058*** 37.790***

(0.447) (0.535) (0.282) (0.483) Observations 9,813 9,129 9,817 9,757 R-squared 0.015 0.004 0.009 0.029 Panel C Interaction 6.336*** 4.082*** 2.801*** 10.743***

(0.629) (0.936) (0.468) (0.698) Firm size 3.049*** 3.219*** 1.706*** 2.526***

(0.295) (0.376) (0.183) (0.357) Leverage -0.004 -0.02 0.006 0.032

(0.027) (0.032) (0.017) (0.032) Volatility: EPS -0.04 0.692*** -0.796*** -0.856***

(0.167) (0.237) (0.124) (0.186) RoA 0.014 0.011 0.051 -0.043

(0.046) (0.061) (0.033) (0.054) LNAF 0.268*** 0.213*** 0.126*** 0.331***

(0.05) (0.061) (0.035) (0.059) Loss dummy -1.480* -1.611 -0.447 -1.846*

(0.831) (1.088) (0.537) (1.011) Accrual dummy -1.372** -1.091 -1.686*** -1.807**

(0.667) (0.863) (0.466) (0.718) Constant 3.631 -3.656 36.599*** 8.988**

(3.073) (3.909) (1.954) (3.924) Year and Industry effects

Yes Yes Yes Yes Observations 9,057 8,441 9,063 9,019 R-squared 0.19 0.137 0.155 0.167

Note: Table presents the impact of the EU directive on NFI disclosures. The Interaction term captures the

effect of the EU directive of NFI disclosures; ESGDS, EDS, GDS and SDS are proxies for overall ESG,

environmental, social and governance disclosures, respectively. Standard errors are in parentheses ***

= p<0.01, ** = p<0.05, * = p<0.1.

Continuing their work using Bloomberg ESG disclosure scores, Ioannou and Serafeim (2017)

report increased ESG disclosures following the enforcement of mandatory NFI disclosure

regulations in China, Denmark, Malaysia and South Africa. Using the instance of mandated

integrated reporting in South Africa, Bernardi and Stark (2018) reveal enhanced levels of ESG

disclosures.

93

For EU directive 2014/95/EU, different country-level studies show that, following the

mandate of the directive, large companies have increased their NFI disclosures in Italy

(Leopizzi, Iazzi, Venturelli, & Principale, 2020; Venturelli, Caputo, Cosma, Leopizzi, & Pizzi,

2017), Romania (Dumitru et al., 2017) and France, Germany and Italy (Boyer-Allirol & Barbu,

2017). However, using an advanced technique of propensity score weighting, we for the first

time provide a comparative analysis of the impact of the EU directive on large European firms

(treated) compared with non-European (untreated) firms. Additionally, the use of

disaggregated disclosure scores helps measure the impact of the EU directive on

environmental, social and governance disclosures. As a result, we find the EU directive is more

effective in increasing social disclosures compared with environmental and social disclosures.

Thus, it is important to incorporate more dimensions of environmental disclosures in the

recent directive to achieve the sustainability targets set under the United Nation’s Paris

Climate Change Agreement and sustainable development goals.

5.4. The Impact of the EU Directive on Future Earnings Risk

Article 2 of the EU directive 2014/95/EU required the European Commission to publish

guidelines that provide useful, relevant and comparable standards for NFI reporting. It states:

“The Commission shall prepare non-binding guidelines on methodology for reporting non-

financial information, including non-financial key performance indicators, general and

sectoral, with a view to facilitating relevant, useful and comparable disclosure of non-financial

information by undertakings. In doing so, the Commission shall consult relevant stakeholders”.

Consequently, the European Commission published its NFI reporting guidelines (2017/C

215/01) on 26 June 201728. According to the European Commission, “The aim of these

guidelines is to help companies disclose high quality, relevant, useful, consistent and more

comparable non-financial (environmental, social and governance-related) information in a

way that fosters resilient and sustainable growth and employment, and provides transparency

to stakeholders. These non-binding guidelines are proposed within the remit of the reporting

requirements provided for under the Directive. They are intended to help companies draw up

relevant, useful, concise non-financial statements according to the requirements of the

28 For details see : https://ec.europa.eu/commission/presscorner/detail/en/IP_17_1702

https://ec.europa.eu/commission/presscorner/detail/en/IP_17_1702

94

Directive. Significant efforts have been made to avoid undue administrative burden,

boilerplate disclosures, or a mere box-ticking exercise”.

Therefore, the mandating of NFI reporting and the subsequent guidelines would not only

increase the quantity of NFI disclosures but may also improve the quality of disclosed

information. Therefore, we hypothesise NFI disclosures have enhanced effects on earnings

risk after the mandating of the directive compared with the earlier period. We use the

difference-in-differences (DiD) method to ascertain the differences between pre and post

directive effects of NFI disclosures on earnings risk. We also use the disaggregated NFI

disclosure scores in the analysis to test whether the different dimensions show different

effects from the overall NFI disclosures.

Table 5.3 shows that the post directive impact (-0.041) of the overall NFI disclosures on

earnings risk is more pronounced than the impact before (-0.031) the directive. The

cumulative impact of the overall NFI disclosures in the post-directive period show an increase

of 32.56% compared with pre-directive period. This indicates that mandating the directive

and subsequent guidelines for NFI reporting make NFI disclosures more information-rich,

resulting in reduced earnings risk. This outcome supports (Bernardi & Stark, 2018) findings

regarding the enhanced negative association between ESG reporting and analysts’ forecasts

accuracy after the mandating of integrated reporting in South Africa.

All the disaggregated NFI disclosure dimensions show an enhanced negative impact on

earnings risk in the post-directive period. More importantly, the governance disclosures that

did not show any significant impact on earnings risk in the pre-directive period now exhibit a

significant (at the 5% level) negative impact in the post-directive period. This observation

indicates that the quality, consistency and relevance of NFI disclosures induced by the

directive have enhanced the efficiency of the information disclosed for governance

disclosures along with the other NFI disclosure dimensions. The results for the control

variables show the expected positive (negative) relationship of leverage, earnings volatility,

loss dummy and financial opaqueness (analyst following) with earnings volatility.

Table 5.3 provides statistical evidence of the differences between the pre- and post-directive

coefficients of NFI disclosures, i.e., the main objective for the fifth hypothesis of this study.

We use the F-test to test the equality of pre- and post-directive coefficients. The F-statistic

95

results show that the pre- and post-directive coefficients of the overall and disaggregated

dimensions of NFI are statistically different at conventional significance levels. This

observation reinforces the earlier conclusion that the post-directive impact of NFI disclosures

is statistically and economically more pronounced than pre-directive effects.

Table 5. 3. The impact of NFI disclosures on firms’ earnings risk (pre and post EU directive) VARIABLES Dependent variable: Earnings risk

1 2 3 4 PreESGDS -0.031***

(0.008) PostESGDS -0.041***

(0.006) PreEDS -0.018***

(0.004) PostEDS -0.029***

(0.004) PreGDS -0.000

(0.006) PostGDS -0.011**

(0.005) PreSDS -0.023***

(0.006) PostSDS -0.032***

(0.005) Firm size 0.284* 0.364*** 0.222 0.274**

(0.145) (0.133) (0.148) (0.126) Leverage 0.028*** 0.025*** 0.031*** 0.022***

(0.006) (0.004) (0.007) (0.005) Volatility: EPS 0.447*** 0.442*** 0.458*** 0.439***

(0.027) (0.026) (0.025) (0.023) RoA -0.006 0.001 -0.004 -0.003

(0.009) (0.009) (0.010) (0.010) LNAF -0.028*** -0.029*** -0.033*** -0.030***

(0.006) (0.005) (0.006) (0.006) Loss dummy 0.426*** 0.474*** 0.451*** 0.463***

(0.150) (0.153) (0.148) (0.156) Accrual dummy 0.190*** 0.211*** 0.201*** 0.188***

(0.060) (0.065) (0.065) (0.058) Constant -6.336*** -7.839*** -7.015*** -6.379***

(1.529) (1.441) (1.548) (1.383) Year and Industry effects Yes Yes Yes Yes F-Statistic 5.13 6.33 6.90 5.72 F-test, p-value 0.0243 0.0124 0.0091 0.0174 R-squared 0.0383 0.0333 0.0330 0.0365 Observations 3,087 3,027 3,085 3,082 Number of groups 308 306 307 307 Note: Pre and Post represent the impact of overall and disaggregated NFI disclosures before and after mandating of the EU directive. Earnings risk represents the log of average analyst forecast dispersion. For other variables definitions see table 5.1. Standard errors are in parentheses. ***= p<0.01, ** = p<0.05, * = p<0.1.

Overall, the results in Table 5.3 provide substantial evidence to reject the null hypothesis of

equality of pre- and post-directive NFI disclosures’ impact on earnings risk. Moreover, these

results are robust for year and industry effects. To ascertain that some unstipulated global

96

factors such as the rising trend of NFI disclosures or the increasing importance of NFI

disclosures do not influence the above results regarding the impact of NFI disclosures, we

perform an equivalent analysis for an Australian sample (S&P/ASX All Australian 50) firms.

Table 5. 4. The impact of NFI disclosures on firms’ earnings risk (Australian sample) VARIABLES Dependent variable: Earnings risk

1 2 3 4

PreESGDS -0.040*** (0.012)

PostESGDS -0.042*** (0.009)

PreEDS -0.024*** (0.007)

PostEDS -0.028*** (0.006)

PreGDS -0.033*

(0.019) PostGDS -0.039**

(0.017) PreSDS -0.024***

(0.008)

PostSGDS -0.029***

(0.006)

Firm size 0.565*** 0.490** 0.596*** 0.470**

(0.208) (0.186) (0.187) (0.199)

Leverage 0.004 -0.004 0.005 0.015***

(0.004) (0.008) (0.004) (0.005)

Volatility: EPS 0.088 0.173 0.140 0.092

(0.180) (0.181) (0.171) (0.172)

RoA -0.040*** -0.035*** -0.037*** -0.037***

(0.009) (0.008) (0.008) (0.008)

LNAF -0.027 -0.025 -0.043 -0.012

(0.026) (0.027) (0.028) (0.028)

Loss dummy -0.280 -0.340* -0.243 -0.365*

(0.188) (0.169) (0.194) (0.207)

Accrual dummy 0.067 0.009 0.099 0.217

(0.185) (0.185) (0.176) (0.179)

Constant -10.116*** -10.010*** -9.837*** -10.306***

(1.570) (1.501) (1.337) (1.634)

Year and Industry effects Yes Yes Yes Yes

F-statistic 0.2600 0.8300 2.2700 2.2100

F-test, p-value 0.6107 0.3677 0.1394 0.1442

Observations 412 402 411 407


Note: Pre and Post represent the impact of overall and disaggregated NFI disclosures in the corresponding period before and after mandating of the EU directive. See Table 5.1 for other variable description. Standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1

Table 5.4 shows that the overall and disaggregated NFI disclosures exhibit a significant

negative impact on the earnings risk for the corresponding pre- and post-directive periods.

For example, in Table 5.4, Column 1, the PreESGDS variable captures the impact of overall NFI

97

disclosures on future earnings risk for 2008-2016 that corresponds to the pre-directive

periods and PostESGDS captures the similar impact for 2017-2018 that corresponds to post-

directive period. Although both PreESGDS and PostESGDS coefficients are statistically

significant at 1% level, the F-test shows that there is no statistically significant difference

between the two variables as is the case for the S&P350 index firms. This finding further

reinforces our conclusion that mandatory NFI reporting regulations improve the quantity and

quality of information disclosures that results in better estimates in analysts’ earnings

forecasts and consequently lowers the uncertainty around the future earnings of firms.

Overall, the DiD analysis indicates that the EU directive has increased the informativeness of

NFI disclosures by large European firms. Thus, financial analysts should pay closer attention

to incorporating such information in their analytical forecasts to achieve robust estimates. On

the other hand, our results also shows that corporate managers can reduce uncertainty

around future earnings by reporting more information regarding ESG issues. Additionally,

corporate managers can address the agency problem and legitimacy issues that might arise

because of information asymmetry between different groups of stakeholders. Moreover, the

DiD analysis also shows that mandating NFI disclosures not only increases the quantity of

disclosures, which could be an aim of the regulatory authorities, but also improves the quality

of disclosures, which is equally beneficial for firms and analysts.


This chapter explores the impact of mandatory NFI reporting regulations (the EU directive) on

the level of NFI disclosures of the EU S&P350 index firms by estimating the average treatment

effects (ATE) for a propensity score matched sample. The ATE results show that the EU

directive had a strong positive impact on the overall and disaggregated NFI disclosures with

the social and governance dimensions exhibiting significantly higher disclosures after the

directive compared with prior disclosure levels. This finding provides evidence to conclude

that mandatory NFI reporting regulations improve the level of disclosures for the large

European firms. The results are robust to alternative reference groups, time and industry

effects.

This chapter uses DiD analysis to ascertain the differences in the impact of NFI disclosures on

earnings risk before and after the enactment of the EU directive. The results show that the

98

post-directive impact of NFI disclosures on earnings risk is substantially greater than in the

pre-directive period. This indicates that the introduction of mandatory NFI reporting not only

increased the level of disclosures, but also enhances the quality of the reported information

that leads to better estimates of future earnings. The results are robust for different firm,

industry and time effects.

This chapter’s findings signify the importance of NFI disclosure regulations in increasing the

amount and transparency of corporate NFI reporting and providing much-needed certainty

surrounding the firm income that is the primary concern of corporate managers and firm

stakeholders. Mandatory NFI regulations could also help countries achieve sustainable and

ethical business practice targets set out in accordance with different global initiatives such as

the Paris agreement and the United Nations’ sustainable development goals.

99

Chapter 6

Summary and Conclusions

6.1. Introduction

This chapter summarises the study’s background, objectives, sample and methodological

approaches used to answer the study’s objectives. The chapter then summarises the key

findings and provides concluding remarks regarding the practical and policy implications of

study’s findings. The chapter also highlights the contributions of the study to the current

literature on ESG disclosure research and discusses the research limitations. Finally, the

chapter proposes potential future research ideas in the ESG disclosures domain.

6.2. Research Background and Objectives

There are increasing concerns about unsustainable production practices, ethical dilemmas

and complex governance structures of modern firms. Therefore, shareholders, along with

other stakeholders, increasingly require firms to enhance the quantity and quality of their

financial and non-financial information disclosures. As a result, firms supplement

conventional financial statements with NFI disclosures to meet stakeholders’ information

needs and address the concerns relating to the sustainability of the business’s operations and

impact on the environment and society. Firms operating in sensitive industries face strict

scrutiny because of the potentially harmful impact of their activities on the environment and

society. As a result, there have been collective efforts such as the Paris Climate Change

Agreement, the United Nations sustainable development goals and the European green deal

to develop sustainable ways of doing business. Similarly, efforts are underway to mandate

the disclosure of business information that has been mostly voluntary, such as issues related

to the environment, society and corporate governance. One such disclosure regulation is the

European directive on non-financial reporting that mandates the reporting of NFI information

for large firms operating in the EU.

Against this backdrop, this study explores the impact of NFI disclosures on the information

environment of the reporting firms and attempts to ascertain the reduction in earnings risk

in the context of environmental, social and governance (ESG) reporting. This study aimed to

quantify the impact of sensitive industry status on the level of NFI disclosures and the

100

resulting decrease in earnings risk for sensitive industry firms. The study also measured the

impact of the EU regulation that mandates NFI disclosure reporting for large EU firms. Finally,

the study attempted to explore the changes in the impact of NFI disclosures on earnings risk

before and after the EU directive. Specifically, we answer the following research questions:

i. What is the impact of cumulative and disaggregated NFI disclosure scores on the

earnings risk for S&P1200 index firms?

ii. What is the impact of sensitive industry position on the level of overall and

disaggregated NFI disclosures for S&P1200 index firms?

iii. Does the impact of total and disaggregated NFI disclosure scores on earning risk vary

between sensitive and non-sensitive industry firms in the S&P 1200 index?

iv. How did mandatory NFI disclosure requirements change the degree of NFI disclosures

at the total and disaggregated levels for the S&P 350 EU index firms?

v. Did the enforcement of mandatory NFI disclosure requirements strengthen the

relationship between NFI disclosures and earnings risk for the S&P 350 EU index firms?

6.3. Study Sample, Data and Empirical Models

This study uses the S&P1200 global index constituents as sample firms i.e., an international,

diversified, representative sample. The study used the Bloomberg database to collect ESG

disclosure scores and financial data for the sample firms and the IBES database to collect

analysts’ earnings forecasts and analyst following data. The data period is from 2008 to 2018.

To answer the research questions, we used panel fixed-effect regression, propensity score

weighting approach and a DiD model. We also used various subsamples, alternative variables

and system GMM estimators to check the robustness of our regression estimates.

6.4. Summary of Findings

This section summarises the study’s key findings. Before the formal empirical analysis, we

performed pairwise correlation and unit-root tests to determine the suitability of the data for

regression analysis. The diagnostic test’s results show that the data are free from

multicollinearity and unit-root problems. The descriptive statistics show that the data

101

variables do not suffer from outlier bias. The next section gives the key findings for each

research objective.

6.4.1. NFI disclosures and earnings risk

The first objective of this study was to determine the impact of cumulative and disaggregated

NFI disclosures on the earnings risk of the sample firms. To this end, we used different

estimation methods to estimate the relationship between NFI disclosures and earnings risk.

Table 6.1 summarises the main results for these estimates. First, we used a panel fixed-effect

regression method that shows that the cumulative NFI disclosures exhibit a strong negative

impact on earnings risk. The environmental and social dimensions also show a negative

impact on earnings risk. However, governance disclosures do not contribute to reduced

earnings risk for our sample firms. We used quartile-based analysis to explore how different

quartiles of NFI disclosures affect the earnings risk.

Table 6. 1. The impact of NFI disclosures on firms’ earnings risk Dependent Variable Earnings Risk

NFI Disclosure variable Total

sample

Large

firms

High-Lev

firms

Neg-Income

firms

High-stock

volatility

GMM

estimation

ESG disclosures (-)*** (-)*** (-)*** (-)*** (-)* (-)***

Environment disclosures (-)*** (-)*** (-)*** (-)*** (-) (-)***

Governance disclosures (-) (-) (-)** (-)** (-) (-)***

Social disclosures (-)*** (-)*** (-)*** (-)*** (-)** (-)***

Note: (-) represents the direction of the relationship between NFI disclosures and risk. *, **, *** denote the statistical significance of the estimates at 1%, 5% and 10% significance, respectively.

The results show that the higher quartiles of overall and disaggregated NFI disclosures exhibit

a more substantial negative impact than the lower quartiles. This finding indicates that high

NFI reporting firms are better at reducing earnings risk compared with low reporting firms.

We used sub-sample analyses based on firm size, profitability, leverage and stock return

volatility that support the main conclusion of the baseline models disclosure effects do not

change because of varying firm characteristics. We used alternative proxies of the dependent

and control variables that produce similar results to the baseline models. Finally, the system

GMM estimators also confirmed that the study’s results do not suffer from endogeneity or

102

simultaneity biases. Overall, this study found strong evidence to conclude that NFI disclosures

improve the information environment for firms, which results in lower earnings risk.

6.4.2. Sensitive industries and NFI disclosures

The literature shows that sensitive industry firms face strict scrutiny because of the sensitive

nature of their operations. Therefore, using a propensity score weighting approach, we

showed that sensitive industry firms report more overall and disaggregated NFI disclosures

than non-sensitive firms (Table 6.2). The difference is more pronounced for the

environmental and social dimensions. This finding indicates that sensitive industry firms

realise the need to make more NFI disclosures to maintain the legitimacy of their operations.

Table 6. 2. The impact of sensitive industry status on NFI disclosures

Note: (+) represents the direction of the relationship between sensitive industry status and NFI disclosures. *, **, *** denote the statistical significance of the estimates at 1%, 5% and 10% significance, respectively.

6.4.3. NFI disclosures and earnings risk (sensitive vs non-sensitive industry firms)

Using interaction terms and split-sample approaches, we confirmed that the disclosure of

environmental information provides more benefit to sensitive industry firms in terms of

reducing earnings risk than for non-sensitive industry firms (Table 6.3). Moreover, the

governance disclosures benefit only non-sensitive industry firms with regard to reducing

uncertainty of future earnings. Thus, based on our empirical results, we present substantial

evidence that sensitive industry firms may increase the legitimacy of their operations and

reduce earnings risk by improving their NFI disclosures, especially environmental information.

Table 6. 3. The impact of NFI disclosures and firms’ earnings risk (sensitive vs non-sensitive industry firms)

Dependent variable Split sample Interaction term approach

Earnings risk Sensitive firms Non-sensitive firms Sensitive firms Non- sensitive firms

ESG disclosures (-)*** (-)*** (-)*** (-)***

Environment Disclosures (-)** (-) (-)** (-)**

Governance Disclosures (-) (-)*** (-) (-)***

Social Disclosures (-)*** (-)*** (-)*** (-)***

Note: (-) represents the direction of the association between NFI disclosures and earnings risk. *, **, *** denote the statistical significance of the estimates at 1%, 5% and 10% significance, respectively.

Dependent Variable - NFI disclosures

ESG

disclosure

Environment

Disclosures

Governance

Disclosures

Social

Disclosures Sensitive

industry

(+)*** (+)*** (+) (+)***

103

6.4.4. The impact of mandatory NFI regulations

Next, we used the case of the recently enacted EU NFI reporting directive to quantify its

impact on the level of NFI disclosures for European firms in the study sample. Using a

propensity-weighted sample, we found a significant positive impact of the EU directive on the

level of cumulative as well as disaggregated disclosure scores of the treated firms compared

with the control group of firms that comprised: 1) pre-directive European firms and 2) pre

and post-directive non-European firms (see Table 6.4). These findings show that the EU

directive has not only increased the quantity of NFI disclosures for European firms from the

pre-directive levels, but also shows that the post directive disclosures of European firms are

substantially more than the non-European firms during the post-regulation period.

Table 6. 4. The impact of a mandatory NFI directive on NFI disclosures

Note: (+) represents the direction of the association between mandatory NFI disclosure regulations and NFI disclosures. *, **, *** denote the statistical significance of the estimates at 1%, 5% and 10% significance, respectively.

6.4.5. The impact of mandatory NFI regulations on the NFI and earnings risk link

Lastly, we investigated the impact of the EU directive on the relationship between NFI

disclosures and earnings risk. We hypothesised that the post-directive NFI disclosures would

have an enhanced effect on earnings risk compared with pre-directive disclosures. To this

end, we used the difference-in-differences method to measure the pre- and post-directive

estimates of NFI disclosures. The results (see Table 6.5) confirmed that the post-directive

cumulative NFI disclosures have a more pronounced impact on earnings risk than pre-

directive disclosures. All sub-dimensions of NFI disclosures also show a relatively enhanced

impact on earnings risk in the post-directive period. These findings provide decisive conclusive

evidence that mandating NFI disclosures results in more disclosures that, in return, increase

the efficiency of reported information in reducing uncertainty surrounding a firm’s future

earnings.

Dependent Variable - NFI disclosures

ESG

disclosures

Environment

disclosures

Governance

disclosures

Social

disclosures Mandatory NFI directive (+)*** (+)*** (+)*** (+)***

104

Table 6. 5. The impact of NFI disclosures on firms’ earnings risk pre versus post the EU directive

Dependent variable

Earnings risk

Pre-directive

period

Post-directive

period

F-Statistic P-Value

ESG disclosures (-0.031)*** (-0.041)*** 5.13 0.0243

Environment Disclosures (-0.018) *** (-0.029)*** 6.33 0.0124

Governance Disclosures (-0.000) (-0.011)*** 6.90 0.0091

Social Disclosures (-0.023)*** (-0.032)*** 5.72 0.0365

Note: (-) represents the direction of the relationship between NFI disclosures and earnings risk. *, **, *** denote the statistical significance of the estimates at 1%, 5% and 10% significance, respectively.

6.5. Practical and Policy implications

This section discusses the practical and policy implications of the study’s findings for various

individuals and groups such as corporate managers, stakeholders and regulators.

6.5.1. Implications for corporate managers

According to agency theory, corporate managers work as agents on behalf of the

shareholders. However, they possess first-hand knowledge about the firm’s activities and

decision-making that creates an information gap between them and the shareholders,

sometimes resulting in agency conflict. Our findings suggest that corporate managers can

alleviate agency conflict by improving the transparency of a firm’s information environment

through reporting more information related to ESG issues. The results also identified that

corporate managers operating in environmentally sensitive, large, highly leveraged and loss

reporting firms could benefit more by reporting ESG related information than others. These

findings further suggest corporate managers managing firms that face stricter scrutiny

because of their size or sensitive industry status can mitigate legitimacy concerns by reporting

ESG related information. Finally, our results suggest that corporate managers can use NFI

disclosures to devise reporting policies for their firm regarding ESG issues that are much

sought after and influence the public image of firms and their decision makers.

6.5.2. Implications for firm stakeholders

Freeman (2010) defines stakeholders as “any group or individual that can affect or be affected

by the realisation of the organisation’s purpose” (p. 26). Accordingly, stakeholders consist of

a variety of individuals and groups that might be internal to a firm such as employees and

shareholders or external such as suppliers, customers, competitors, and government. These

stakeholders also require ESG related information along with the financial information to

105

make informed decisions about the firm. One of the stakeholders is a financial analyst who

forecasts estimates of a firm’s future earnings and market value. The study’s results show that

NFI disclosures are beneficial in reducing the difference between actual and the forecast

earnings by financial analysts. Fundamentally, these findings imply that more precise earnings

estimates by financial analysts can influence the perceptions of other stakeholders such as

suppliers, creditors and regulators since they use analysts’ forecasts to make their decisions.

Likewise, our findings suggest that NFI disclosures improve the transparency of a firm’s

information environment, which serves as a tool to minimise information asymmetry

problems arising from information lop-sidedness between corporate managers and

stakeholders. NFI disclosures can help pro-ESG stakeholders to analyse the social and

environmental impacts of a firm. Our findings suggest that such stakeholders can use ESG

disclosures, with financial information, to make informed decisions about a firm’s future

performance and risks.

6.5.3. Implications for regulators

Countries around the world face increasing pressures from the international community to

make businesses more responsible in terms of their conduct and information disclosures.

Global agreements such as the Paris Agreement, the United Nations development goals and

other such agreements, also provide countries specific targets for reducing their impact on

the environment and society. Accordingly, many countries have enacted regulations that

require businesses to report efforts undertaken to minimise impact on the environment and

progress towards an equitable society. In this context, using the recently enacted EU directive

on NFI reporting, we showed that mandating of NFI disclosures results in more reporting of

NFI information in all ESG domains. Further analysis of post-directive disclosures shows an

improved information environment of firms, suggesting that the mandating of NFI disclosures

not only improves the degree of disclosures, it also enhances the disclosure quality. These

findings suggest that regulators in the other parts of the world should move towards

mandatory NFI disclosure regimes such as the EU directive, which presents promising results

to achieve much-needed business sustainability goals that are essential to achieve country-

level targets set in the global agreements. Furthermore, providing evidence regarding

improved disclosure quality in a mandated disclosure regime, our findings supplement the

106

literature that propose state institutions can facilitate business legitimisation strategies

through policy interventions. For example, see (Archel, Husillos, Larrinaga, & Spence, 2009;

La Torre, Sabelfeld, Blomkvist, Tarquinio, & Dumay, 2018). Finally, mandatory NFI regimes

could also improve disclosures on the governance practices of firms, which is a necessary tool

to avoid fraudulent and risky business practices that could result in systemic failure such as

the recent 2008 global financial crisis.

6.6. Research Contribution

This study contributes in several ways to existing NFI disclosure and firm risk literature. First,

this study uses an internationally diversified, representative sample that covers over 70% of

the total global market capitalisation and is often used as a proxy for world stock market

performance. Hence, the study’s results are more generalisable than existing studies that

focus on companies incorporated in a specific country (Becchetti et al., 2015; Bernardi &

Stark, 2018; Cai et al., 2016; Chang et al., 2014), a group of developed countries (Benlemlih &

Girerd‐Potin, 2017), or a geographical region (Sassen et al., 2016). Second, we used a firm risk

proxy that is more relatable to NFI disclosures than other firm risk proxies used in the

literature such as idiosyncratic or market risk, e.g., see (Benlemlih et al., 2018; Oikonomou et

al., 2012). The use of an earnings risk proxy offers new insights into the relationship between

NFI disclosures and firm risk that further improve understanding of the importance of NFI

disclosures for today’s businesses. Our findings provide essential understandings about the

ESG disclosure and firms’ earnings risk association that can help corporate managers devise

NFI disclosure policies for their firm.

Third, using the improved methodology of propensity score weighting, we quantified the

impact of sensitive industry status on the level of overall and disaggregated disclosures. To

the best of our knowledge, this study is the first to quantify the impact of sensitive industry

status on NFI disclosures for an international sample. This study also contributes to the

understanding of the relationship between NFI disclosures and firm risk in sensitive industry

firms for an international sample; current evidence is limited to individual countries such as

the U.S., China, and Spain, e.g., see (Jo & Na, 2012), (Zeng et al., 2020) and (Reverte, 2012).

107

Fourth, this study contributes to the mandatory NFI disclosure literature by measuring

the impact of the recently enacted EU directive on NFI disclosures for a representative sample

of large European firms. The study’s findings provide important understandings about the

impact of a mandated NFI reporting regime and we offer suggestions for corporate managers

and regulators regarding the benefits of moving towards a mandatory disclosure regime

compared with voluntary disclosure practices. Fifth, this study offers novel evidence

regarding the NFI disclosures, firm risk relationship in a mandatory NFI reporting

environment. Using the DID method, we find an enhanced association between ESG

disclosures and firms’ earnings risk in the post-directive period, which indicates an increase

in the quality of disclosed information under mandated NFI disclosure regime. Finally, Our

findings also offer understandings for regulators and accounting standard setters globally to

consider a shift towards mandatory NFI reporting regimes, which is essential to achieve the

country-level sustainability targets set by United Nations’ Sustainable Development Goals

(SDGs), and the United Nations Paris Climate change agreement.

6.7. Research Limitations

This study has some limitations that emerge from sample selection, data sources and

methodologies used. In the sample selection, although the study sample captures 70% of the

total global market capitalisation, it contains a small number of small and medium-sized firms.

Thus, the study results mainly cater for large firms. Moreover, because of the limited

availability of NFI disclosures data, the study sample period spans only from 2008 to 2018.

Thus, this study could not account for the potential effects of a market-wide crisis such as the

2008’s Global financial crisis on NFI disclosure practices and their relationship with the

sampled firms’ earning risk.

Further, the use of ESG disclosures as a proxy for NFI disclosures also has some limitations.

For example, in a broader context, NFI disclosures could include intellectual capital

disclosures that are not fully covered in the Bloomberg ESG scores. Besides, the ESG disclosure

scores capture the overall efforts of a firm towards ESG issues but these scores do not

facilitate analysis of other efforts such as reduction in Green House Gas (GHG) emissions or

equal career opportunities across genders.

108

This study uses quantitative variables to control for firm characteristics that might affect the

relationship between NFI disclosures and earnings risk. However, there might be some

qualitative variables such as company values, philosophy and culture, that may determine the

level of NFI disclosures. Such variables are not considered in this study. Additionally, this study

uses various estimation methods to establish a causal relationship between NFI disclosures

and earnings risk and avoids biases such as endogeneity and simultaneity. However, we did

not apply other estimation methods such as the instrumental variable approach that might

help identify robust instruments.

6.8. Future Research

This study’s limitations help identify several research avenues for future studies into NFI

disclosures and their relationship with firm risk. First, future studies may focus on small and

medium enterprises (SME) to better understand the NFI disclosure practices of these firms

since the literature provides minimal evidence for this group of firms. The SME sector

constitutes a very significant proportion of business establishments especially in developing

countries; hence, studying NFI disclosures and their relationship with firms’ financial

outcomes could reveal novel results. Second, future studies could consider using specific NFI

disclosures such as GHG emissions, workforce diversity, community and customer relations.

In doing so, future studies may focus on individual industries or firms to better understand

the motives behind NFI disclosures and their relationship with firms’ financial outcomes.

This study provides a foundation for future studies to continue investigating the impact of

mandatory NFI disclosure regimes on firm-level disclosures and the resulting consequences

for firms and economies. Future studies may also investigate the level of compliance with

mandatory NFI disclosure requirements, which will give a better idea about the areas that

need improving and reconsideration. Such studies will guide regulators and policy-makers in

developing holistic disclosure regimes that embody all aspects of business activities and

address the concerns of the variety of stakeholders.

109

Appendix A

Table A.1. The Hausman test estimates of the study variables Variable Fixed Random Difference S.E.

ESGDS -0.0205 -0.0177 -0.0028 0.0014

Leverage 0.0183 0.0140 0.0043 0.0019

Firm Size -0.2664 -0.1059 -0.1606 0.0668

Growth -0.1050 -0.1132 0.0082 0.0049

Volatility: EPS 0.3614 0.7571 -0.3957 0.0390

Loss Dummy 0.3181 0.3549 -0.0368 0.0111

Accrual dummy 0.2029 0.2266 -0.0237 0.0102

RoA -0.0082 -0.0103 0.0021 0.0013

LNAF -0.1752 -0.1800 0.0048 0.0268

Test: Ho: difference in coefficients not systematic.

chi2(9) = (b-B)'[(V_b-V_B)^(-1)](b-B)

Chi- square value = 153.15; p-value = 0.00000

Note: The Table presents the Hausman test estimates of the study variables.

110

Table A.2. The impact of NFI disclosures on firms’ earnings risk

Small firms Large firms

VARIABLE Dependent Variable: Earnings Risk

ESGDS -0.014*** -0.020*** (0.003) (0.004)

EDS -0.005** -0.012*** (0.002) (0.002)

GDS -0.007 -0.008 (0.007) (0.007)

SDS -0.014*** -0.014***

(0.003) (0.005)

Firm Size -0.289*** -0.194 -0.345*** -0.301*** -0.037 0.003 -0.129 -0.063

(0.107) (0.155) (0.108) (0.101) (0.194) (0.184) (0.196) (0.211)

Leverage 0.015*** 0.014*** 0.015*** 0.014*** 0.027*** 0.028*** 0.028*** 0.027***

(0.005) (0.004) (0.005) (0.004) (0.007) (0.007) (0.007) (0.008)

Growth -0.087*** -0.111*** -0.090*** -0.080*** -0.122*** -0.130*** -0.129*** -0.123***

(0.014) (0.017) (0.014) (0.013) (0.025) (0.025) (0.025) (0.026)

Volatility: EPS 0.315*** 0.413*** 0.308*** 0.314*** 0.453*** 0.460*** 0.442*** 0.438***

(0.068) (0.064) (0.065) (0.064) (0.036) (0.038) (0.034) (0.033)

Loss Dummy 0.338** 0.218 0.343** 0.319* 0.288*** 0.299*** 0.288*** 0.287***

(0.171) (0.183) (0.173) (0.181) (0.104) (0.099) (0.105) (0.104)

Accrual dummy 0.264*** 0.287*** 0.259*** 0.270*** 0.037 0.048 0.049 0.043

(0.088) (0.101) (0.088) (0.093) (0.064) (0.058) (0.066) (0.065)

RoA -0.015 -0.016 -0.014 -0.013 -0.003 -0.000 -0.003 -0.003

(0.013) (0.013) (0.014) (0.014) (0.011) (0.009) (0.011) (0.011)

Log NAF -0.040 -0.129 -0.071 -0.074 -0.226*** -0.207*** -0.254*** -0.222***

(0.108) (0.117) (0.123) (0.081) (0.056) (0.064) (0.056) (0.058)

Constant -1.602* -2.415* -1.119 -1.413 -3.270 -4.174* -2.569 -3.279

(0.966) (1.300) (0.901) (0.937) (2.195) (2.176) (2.053) (2.313)

FE time and industry Yes Yes Yes Yes Yes Yes Yes Yes

R-squared 0.0605 0.059 0.0573 0.0597 0.0562 0.0546 0.052 0.0559

Observations 4,386 3,991 4,391 4,359 4,499 4,314 4,500 4,491

Number of groups 543 511 542 541 512 502 511 511

Note: The Table presents the sub-sample analysis for small and large firms. Standard errors are in parentheses; *** = p<0.01, ** = p<0.05, * = p<0.1.

111

Table A.3. The impact of NFI disclosures on firms’ earnings risk Low leverage sample High leverage sample

VARIABLE Dependent Variable – Earnings Risk

ESGDS -0.009* -0.019*** (0.005) (0.003)

EDS -0.005 -0.010*** (0.004) (0.002)

GDS 0.000 -0.015** (0.007) (0.006)

SDS -0.011*** -0.014***

(0.003) (0.003) Firm Size -0.120 -0.058 -0.161 -0.083 -0.239*** -0.260** -0.312*** -0.282***

(0.110) (0.133) (0.110) (0.113) (0.078) (0.123) (0.078) (0.077) Leverage 0.009* 0.008 0.010* 0.010* 0.025*** 0.029*** 0.026*** 0.027***

(0.005) (0.006) (0.006) (0.005) (0.004) (0.005) (0.004) (0.004) Growth -0.127*** -0.135*** -0.128*** -0.123*** -0.076*** -0.100*** -0.081*** -0.073***

(0.019) (0.018) (0.019) (0.020) (0.010) (0.013) (0.010) (0.009) Volatility: EPS 0.371*** 0.398*** 0.363*** 0.359*** 0.473*** 0.511*** 0.471*** 0.468***

(0.026) (0.024) (0.027) (0.024) (0.034) (0.032) (0.032) (0.033) Loss Dummy 0.267* 0.275* 0.276* 0.262* 0.212*** 0.134 0.207*** 0.191**

(0.162) (0.148) (0.160) (0.155) (0.068) (0.090) (0.073) (0.078) Accrual dummy -0.012 -0.014 -0.011 -0.010 -0.015 -0.014 -0.015 -0.015

(0.009) (0.010) (0.009) (0.009) (0.012) (0.012) (0.012) (0.013) RoA -0.294*** -0.303*** -0.316*** -0.297*** -0.031 -0.065 -0.058 -0.009

(0.048) (0.052) (0.050) (0.047) (0.062) (0.084) (0.072) (0.056) LNAF -2.099* -2.845* -2.020 -2.433* -2.106*** -2.229* -1.184 -1.992***

(1.200) (1.473) (1.274) (1.270) (0.760) (1.152) (0.772) (0.755) Constant -0.120 -0.058 -0.161 -0.083 -0.239*** -0.260** -0.312*** -0.282***

(0.110) (0.133) (0.110) (0.113) (0.078) (0.123) (0.078) (0.077) FE time and industry Yes Yes Yes Yes Yes Yes Yes Yes R-squared 0.0469 0.0481 0.0456 0.0477 0.0515 0.0516 0.0487 0.0519 Observations 4,512 4,184 4,515 4,491 4,373 4,121 4,376 4,359 Number of groups 620 600 620 619 642 614 642 639

Note: The Table presents the sub-sample analysis for low and high leverage firms. Standard errors are in parentheses; *** = p<0.01, ** = p<0.05, * = p<0.1.

112

Table A. 4. The impact of NFI disclosures on firms’ earnings risk

Low stock volatility firms High stock volatility firms VARIABLE Dependent Variable – Earnings Risk ESGDS -0.002 -0.015*

(0.005) (0.008) EDS 0.002 -0.008

(0.004) (0.005) GDS -0.006 0.007

(0.004) (0.008) SDS -0.005** -0.015**

(0.003) (0.007) Firm Size -0.354*** -0.340*** -0.350*** -0.345*** -0.162 -0.012 -0.228** -0.150

(0.113) (0.122) (0.128) (0.123) (0.103) (0.138) (0.101) (0.109) Leverage -0.059*** -0.067*** -0.059*** -0.055*** -0.105*** -0.130*** -0.106*** -0.103***

(0.018) (0.018) (0.018) (0.018) (0.018) (0.021) (0.019) (0.018) Growth 0.007 0.006 0.007 0.008 0.022*** 0.025*** 0.022*** 0.020***

(0.006) (0.006) (0.006) (0.007) (0.007) (0.006) (0.007) (0.006) Volatility: EPS 0.272*** 0.280*** 0.274*** 0.266*** 0.420*** 0.531*** 0.408*** 0.419***

(0.051) (0.059) (0.051) (0.054) (0.084) (0.075) (0.081) (0.082) Loss Dummy 0.247** 0.226** 0.249** 0.249** 0.153 0.111 0.170 0.162

(0.108) (0.105) (0.108) (0.109) (0.164) (0.176) (0.166) (0.167) Accrual dummy 0.135* 0.132* 0.134* 0.142* 0.179*** 0.199*** 0.180*** 0.176***

(0.071) (0.076) (0.071) (0.076) (0.058) (0.059) (0.059) (0.062) RoA -0.014*** -0.016*** -0.014*** -0.014*** -0.013 -0.009 -0.012 -0.011

(0.005) (0.006) (0.004) (0.005) (0.017) (0.016) (0.017) (0.017) LNAF -0.108 -0.066 -0.107 -0.088 -0.089 -0.188*** -0.134* -0.138***

(0.127) (0.138) (0.126) (0.134) (0.059) (0.061) (0.072) (0.046) Constant -1.175 -1.499 -0.953 -1.223 -1.943 -3.610** -2.129 -1.945

(1.089) (1.116) (1.185) (1.047) (1.326) (1.519) (1.461) (1.364) FE time and industry Yes Yes Yes Yes Yes Yes Yes Yes R-squared 0.037 0.0367 0.037 0.0373 0.0504 0.0585 0.0474 0.0504 Observations 4,569 4,359 4,568 4,550 4,316 3,946 4,323 4,300 Number of groups 798 777 798 796 899 843 899 897 Note: The Table presents the sub-sample analysis for low and high stock volatility firms. Standard errors are in parentheses. *** = p<0.01, ** = p<0.05, * = p<0.1.

113

Table A.5. The impact of NFI disclosures on firms’ earnings risk Positive income firms Negative income firms

VARIABLE Dependent Variable – Earnings Risk

ESGDS -0.014*** -0.047*** (0.003) (0.009)

EDS -0.006*** -0.035*** (0.002) (0.009)

GDS -0.009* 0.027** (0.005) (0.012)

SDS -0.013*** -0.027***

(0.002) (0.010)

Firm Size -0.285*** -0.266*** -0.336*** -0.294*** -0.159 0.051 -0.205 -0.121

(0.062) (0.072) (0.068) (0.059) (0.284) (0.372) (0.276) (0.283)

Leverage 0.013*** 0.012*** 0.013*** 0.013*** 0.067*** 0.069*** 0.067*** 0.061***

(0.004) (0.004) (0.004) (0.005) (0.009) (0.008) (0.011) (0.009)

Growth -0.090*** -0.106*** -0.093*** -0.087*** -0.021 -0.042 -0.020 -0.018

(0.015) (0.015) (0.015) (0.016) (0.026) (0.027) (0.030) (0.029)

Volatility: EPS 0.377*** 0.409*** 0.368*** 0.367*** 0.272 0.341** 0.324* 0.288*

(0.047) (0.048) (0.050) (0.047) (0.169) (0.145) (0.173) (0.160)

Loss Dummy 0.170*** 0.177*** 0.169*** 0.174*** 0.262 0.307* 0.379** 0.336*

(0.038) (0.045) (0.038) (0.039) (0.163) (0.172) (0.181) (0.178)

Accrual dummy -0.005 -0.009 -0.004 -0.004 -0.010 -0.007 -0.010 -0.006

(0.008) (0.009) (0.008) (0.008) (0.022) (0.017) (0.021) (0.019)

RoA -0.116 -0.126 -0.134* -0.095 0.167 -0.002 0.074 0.025

(0.073) (0.083) (0.079) (0.075) (0.165) (0.269) (0.235) (0.183)

LNAF -1.084 -1.501* -0.540 -1.143 -2.716 -5.089 -5.455** -3.389

(0.793) (0.803) (0.885) (0.769) (2.860) (3.288) (2.655) (2.740)

FE time and industry Yes Yes Yes Yes Yes Yes Yes Yes

R-squared 0.0463 0.0467 0.044 0.0471 0.1033 0.093 0.0851 0.0772

Observations 7,942 7,444 7,944 7,909 943 861 947 941

Number of groups 908 879 907 905 412 379 413 413

Note: The Table presents the sub-sample analysis for negative and positive income firms. Standard errors are in parentheses; *** = p<0.01, ** = p<0.05, * = p<0.1

114

Table A.6. The impact of NFI disclosures on firms’ earnings risk (non-financial firms)

Dependent Variable: Earnings Risk

VARIABLE 1 2 3 4

ESGDS -0.009*** (0.003)

EDS -0.005** (0.002)

GDS 0.003

(0.004)

SDS -0.009***

(0.002)

Leverage 0.015*** 0.012*** 0.015*** 0.013***

(0.004) (0.004) (0.004) (0.004)

Firm Size -0.285*** -0.252** -0.325*** -0.292***

(0.087) (0.109) (0.087) (0.087)

Growth -0.075*** -0.092*** -0.077*** -0.069***

(0.011) (0.011) (0.011) (0.009)

Volatility: EPS 0.361*** 0.422*** 0.352*** 0.355***

(0.038) (0.028) (0.039) (0.037)

Loss 0.219* 0.188 0.225* 0.212

(0.130) (0.135) (0.131) (0.133)

Accrual 0.223*** 0.261*** 0.226*** 0.230***

(0.050) (0.056) (0.051) (0.053)

ROA -0.016 -0.014 -0.015 -0.015

(0.012) (0.011) (0.012) (0.012)

LNAF -0.009 -0.050 -0.030 -0.028

(0.073) (0.079) (0.080) (0.065)

Constant -1.688* -2.013** -1.786* -1.577*

(0.889) (1.018) (0.919) (0.899)

0.0432 0.0448 0.0415 0.0425 Observations 7,408 6,981 7,413 7,378


Note: The Table presents the sub-sample analysis for non-financial firms. See Table 4.1 for variable definitions.

Standard errors are in parentheses; *** = p<0.01, ** = p<0.05, * = p<0.1.

115

Appendix B

Figure B.1. Overlap plot of propensity scores between sensitive and non-sensitive firms

Note: Figure B.1 shows the propensity scores overlap plot between sensitive and non-sensitive samples. ES mean is the standardized effect size differences. KS

mean is the Kolmogorov–Smirnov mean differences.

116

Figure B.2. Overlap plot of propensity scores between regulated and unregulated firms

Note: Figure B.2 shows the propensity scores overlap plot between sensitive and non- sensitive samples. ES mean is standardized effect size differences. KS mean is the

Kolmogorov–Smirnov mean differences.

117

Appendix C

List of companies contained in the sample

S&P AS 50

Asustek Computer Inc

AU Optronics Corp

BOC Hong Kong Holdings Ltd

Cathay Financial Holding Co Ltd

China Communications Construction Co Ltd

China Construction Bank Corp

China Life Insurance Co Ltd

China Mobile Ltd

China Steel Corp

Chunghwa Telecom Co Ltd

CLP Holdings Ltd

CNOOC Ltd

DBS Group Holdings Ltd

Esprit Holdings Ltd

Formosa Chemicals & Fibre Corp

Formosa Plastics Corp

Hang Seng Bank Ltd

Hon Hai Precision Industry Co Ltd

Hong Kong & China Gas Co Ltd

Hong Kong Exchanges & Clearing Ltd

HTC Corp

Hutchison Whampoa Ltd

Hyundai Motor Co

Industrial & Commercial Bank of China Ltd

KB Financial Group Inc

Keppel Corp Ltd

Korea Electric Power Corp

Korea Shipbuilding & Offshore Engineering Co Ltd

KT&G Corp

LG Electronics Inc

Li & Fung Ltd

MediaTek Inc

Nan Ya Plastics Corp

Oversea-Chinese Banking Corp Ltd

PetroChina Co Ltd

POSCO

Samsung C&T Corp/Old

Samsung Electronics Co Ltd

Shinhan Financial Group Co Ltd

Singapore Airlines Ltd

118

Singapore Telecommunications Ltd

SK Innovation Co Ltd

SK Telecom Co Ltd

Sun Hung Kai Properties Ltd

Swire Pacific Ltd

Taiwan Semiconductor Manufacturing Co Ltd

United Microelectronics Corp

United Overseas Bank Ltd

Wilmar International Ltd

S&P AU 50

ACN 004 410 833 Ltd

AGL Energy Ltd

Alumina Ltd

Amcor PLC

AMP Ltd

ASX Ltd

Australia & New Zealand Banking Group Ltd

BHP Group Ltd

BlueScope Steel Ltd

Brambles Ltd

CIMIC Group Ltd

Commonwealth Bank of Australia

Crown Resorts Ltd

CSL Ltd

Fairfax Media Ltd

Fortescue Metals Group Ltd

Frugl Group Ltd

Goodman Group

GPT Group/The

Incitec Pivot Ltd

Insurance Australia Group Ltd

Intoll Group

Lendlease Corp Ltd

Lynch Group Holdings Ltd

Macquarie Group Ltd

Newcrest Mining Ltd

Orica Ltd

Origin Energy Ltd

OZ Minerals Ltd

Qantas Airways Ltd

QBE Insurance Group Ltd

Resolution Life Aaph Ltd

Rio Tinto Ltd

Santos Ltd

Spark New Zealand Ltd

Stockland

119

Suncorp Group Ltd

Sydney Airport

Tabcorp Holdings Ltd

Telstra Corp Ltd

TFCF Corp

Toll Holdings Ltd

Transurban Group

Wesfarmers Ltd

Westfield Corp

Westpac Banking Corp

Woodside Petroleum Ltd

Woolworths Group Ltd

Worley Ltd

S&P CA 60

Agnico Eagle Mines Ltd

Agrium Inc

Aimia Inc

ARC Resources Ltd

Augusta Gold Corp

Bank of Montreal

Bank of Nova Scotia/The

Barrick Gold Corp

Bausch Health Cos Inc

BCE Inc

BlackBerry Ltd

Blackstone Mortgage Trust Inc

Bombardier Inc

Brookfield Asset Management Inc

Cameco Corp

Canadian Imperial Bank of Commerce

Canadian National Railway Co

Canadian Natural Resources Ltd

Canadian Oil Sands Ltd

Canadian Pacific Railway Ltd

Canadian Tire Corp Ltd

Corp Shoppers Drug Mart

Enbridge Inc

Enerplus Corp

First Quantum Minerals Ltd

Fortis Inc/Canada

George Weston Ltd

Gildan Activewear Inc

Husky Energy Inc

Imperial Oil Ltd

Inmet Mining Corp

Kinross Gold Corp

120

Loblaw Cos Ltd

Lundin Mining Corp

Magna International Inc

Manulife Financial Corp

Metro Inc/CN

Nexen Energy ULC

Nordion Inc

NOVA Chemicals Corp

Obsidian Energy Ltd

Ovintiv Canada ULC

Petro-Canada

Potash Corp of Saskatchewan Inc

Repsol Oil & Gas Canada Inc

Rogers Communications Inc

Royal Bank of Canada

Shaw Communications Inc

SNC-Lavalin Group Inc

Sun Life Financial Inc

Suncor Energy Inc

TC Energy Corp

Teck Resources Ltd

TELUS Corp

Thomson Reuters Corp

Tim Hortons Inc

Toronto-Dominion Bank/The

TransAlta Corp

Yamana Gold Inc

Yellow Pages Digital & Media Solutions Ltd

S&P EU 350

3i Group PLC

ABB Ltd

Abertis Infraestructuras SA

Abi Sab Group Holding Ltd

Accor SA

Acerinox SA

ACS Actividades de Construccion y Servicios SA

Adecco Group AG

adidas AG

Ageas SA/NV

AGFA-Gevaert NV

AIB Group PLC

Air France-KLM

Air Liquide SA

Airbus SE

Alcatel Lucent SAS

Alfa Laval AB

121

Alleanza Toro SpA

Allergan PLC

Allianz SE

Alpha Services and Holdings SA

Alstom SA

Anglo American PLC

Anheuser-Busch InBev SA/NV

AP Moller - Maersk A/S

ArcelorMittal SA

ASML Holding NV

Assa Abloy AB

Assicurazioni Generali SpA

Associated British Foods PLC

AstraZeneca PLC

Atlantia SpA

Atlas Copco AB

Aviva PLC

AXA SA

BAE Systems PLC

Baloise Holding AG

Banca Monte dei Paschi di Siena SpA

Banca Popolare di Milano Scarl

Banco Bilbao Vizcaya Argentaria SA

Banco Comercial Portugues SA

Banco Espirito Santo SA

Banco Popolare SC

Banco Popular Espanol SA

Banco Santander SA

Bank of Ireland Group PLC

Bankinter SA

Barclays PLC

BASF SE

Bayer AG

Bayerische Motoren Werke AG

Beiersdorf AG

BG Group Ltd

BHP Group PLC

BNP Paribas SA

BNY Mellon Strategic Municipal Bond Fund Inc

Boliden AB

Bouygues SA

BP PLC

Brisa Auto-Estradas de Portugal SA

British American Tobacco PLC

British Land Co PLC/The

BT Group PLC

122

Bunzl PLC

Cable & Wireless Communications Ltd/Old

Cadbury Ltd

Capgemini SE

Capita PLC

Carlsberg AS

Carnival PLC

Carrefour SA

Casino Guichard Perrachon SA

CECONOMY AG

Christian Dior SE

Cie de Saint-Gobain

Cie Financiere Richemont SA

Cie Generale des Etablissements Michelin SCA

Clariant AG

Cobham Ltd

Commerzbank AG

Compass Group PLC

Credit Agricole SA

Credit Suisse Group AG

CRH PLC

Daily Mail & General Trust PLC

Daimler AG

Danone SA

Danske Bank A/S

Darty Ltd

Dassault Systemes SE

Delhaize Group SCA

Dentsu International Ltd

Deutsche Bank AG

Deutsche Boerse AG

Deutsche Lufthansa AG

Deutsche Post AG

Deutsche Telekom AG

Dexia SA

Diageo PLC

Dixons Retail Group Ltd

DNB ASA

Drax Group PLC

EDF Energy Nuclear Generation Group Ltd

EDP - Energias de Portugal SA

EI Group Ltd

Electricite de France SA

Electrocomponents PLC

Electrolux AB

Enel SpA

123

Engie SA

Eni SpA

Equinor ASA

Erste Group Bank AG

EssilorLuxottica SA

Eurobank Ergasias Services and Holdings SA

Experian PLC

Ferguson PLC

Fiat SpA

Finastra Group Holdings Ltd

Firstgroup PLC

Fortum Oyj

Fresenius Medical Care AG & Co KGaA

Friends Life FPG Ltd

G4S PLC

GAM Holding AG

Gates Worldwide Ltd

Givaudan SA

GKN Ltd

GlaxoSmithKline PLC

Groupe Bruxelles Lambert SA

Grupo Ferrovial SA

H & M Hennes & Mauritz AB

Hammerson PLC

Hays PLC

HBOS PLC

Hellenic Telecommunications Organization SA

Henkel AG & Co KGaA

Hermes International

Hibu plc

HOCHTIEF AG

Holcim Ltd

Holmen AB

Home Reit PLC

HSBC Holdings PLC

Hypo Real Estate Holding GmbH

Iberdrola SA

Igeamed SpA

IMI PLC

Imperial Brands PLC

Infineon Technologies AG

Integra Inc

Intercement Portugal SA

InterContinental Hotels Group PLC

International Consolidated Airlines Group SA

International Power Ltd

124

Intesa Sanpaolo SpA

Invensys Ltd

Investor AB

Irish Bank Resolution Corp Ltd/Old

Italiaonline SpA

ITV PLC

J Sainsbury PLC

Johnson Matthey PLC

K+S AG

KBC Group NV

Kering SA

Kingfisher PLC

Kone Oyj

Koninklijke KPN NV

Ladbrokes Coral Group Ltd

Lafarge SA

Lagardere SCA

Land Securities Group PLC

Legal & General Group PLC

Leonardo SpA

Lloyds Banking Group PLC

Logica Ltd

L'Oreal SA

Luxottica Group SpA

LVMH Moet Hennessy Louis Vuitton SE

Man Group PLC/Jersey

MAN SE

Marks & Spencer Group PLC

Mediaset SpA

Mediobanca Banca di Credito Finanziario SpA

Merck KGaA

Mitchells & Butlers PLC

Muenchener Rueckversicherungs-Gesellschaft AG in Muenchen

National Bank of Greece SA

National Grid PLC

Naturgy Energy Group SA

Neles Oyj

Nestle SA

Next PLC

Nobel Biocare Holding AG

Nokia Oyj

Nordea Bank Abp

Norsk Hydro ASA

Novartis AG

Novo Nordisk A/S

Novozymes A/S

125

Om Residual UK Ltd

OMV AG

OPAP SA

Orange SA

Orkla ASA

Pearson PLC

Pernod Ricard SA

Perrigo Corp DAC

Persimmon PLC

Peugeot SA

Pharol SGPS SA

Pirelli & C SpA

Porsche Automobil Holding SE

Provident Financial PLC

Proximus SADP

Prudential PLC

Publicis Groupe SA

Puma SE

Punch Taverns Ltd

Rand Capital Corp

REC Silicon ASA

Reckitt Benckiser Group PLC

RELX PLC

Renault SA

Rentokil Initial PLC

Rentrak Corp

Repsol SA

Rexam Ltd

Rio Tinto PLC

Roche Holding AG

Rolls-Royce Holdings PLC

Royal Dutch Shell PLC

RSA Insurance Group LTD

RWE AG

Ryanair Holdings PLC

Safran SA

Sage Group PLC/The

Saipem SpA

Salzgitter AG

Sampo Oyj

Sandvik AB

Sanofi

SAP SE

Scania AB

Schneider Electric SE

Schroders PLC

126

Securitas AB

Segro PLC

SES SA

Severn Trent PLC

Shire PLC

Sibanye UK Ltd

Siemens AG

Skandinaviska Enskilda Banken AB

Skanska AB

SKF AB

Sky Ltd

Smith & Nephew PLC

Smiths Group PLC

Snam SpA

Societe Generale SA

Sodexo SA

Solvay SA

SSAB AB

SSE PLC

Standard Chartered PLC

Standard Life Aberdeen PLC

STMicroelectronics NV

Stora Enso Oyj

Suez SA

Svenska Cellulosa AB SCA

Svenska Handelsbanken AB

SVF Holdco UK Ltd

Swatch Group AG/The

Swedbank AB

Swedish Match AB

Swiss Life Holding AG

Swiss Reinsurance Co Ltd

Swisscom AG

Syngenta AG

Tate & Lyle PLC

Technicolor SA

Technip SA

Tele2 AB

Telecom Italia SpA/Milano

Telefonaktiebolaget LM Ericsson

Telefonica SA

Telekom Austria AG

Telenor ASA

Television Francaise 1

Telia Co AB

Tesco PLC

127

Thales SA

Thomson Reuters UK Ltd

thyssenkrupp AG

TotalEnergies SE

TUI AG

UBM PLC

UBS AG

UCB SA

Umicore SA

Unibail-Rodamco-Westfield SE

UniCredit SpA

Unilever PLC

Union Fenosa SA

Unione di Banche Italiane SpA

United Utilities Group PLC

Unity Software Inc

Universal Display Corp

UPM-Kymmene Oyj

Valassis Direct Mail Inc

Valeo SA

Vallourec SA

Veolia Environnement SA

Vestas Wind Systems A/S

Vesuvius PLC

Vinci SA

Vivendi SE

Vodafone Group PLC

voestalpine AG

Volkswagen AG

Volvo AB

WH Smith PLC

Whitbread PLC

William Hill PLC

Wm Morrison Supermarkets PLC

WPP PLC

Xstrata Ltd

Yara International ASA

Zurich Insurance Group AG

S&P JP 150

Advantest Corp

Aeon Co Ltd

AGC Inc

Ajinomoto Co Inc

ANA Holdings Inc

Asahi Group Holdings Ltd

Asahi Kasei Corp

128

Astellas Pharma Inc

Bridgestone Corp

Canon Inc

Central Japan Railway Co

Chubu Electric Power Co Inc

Credit Saison Co Ltd

Dai Nippon Printing Co Ltd

Daiichi Sankyo Co Ltd

Daikin Industries Ltd

Daiwa House Industry Co Ltd

Daiwa Securities Group Inc

Denso Corp

Dentsu Group Inc

East Japan Railway Co

Eisai Co Ltd

ENEOS Corp

FANUC Corp

FUJIFILM Holdings Corp

Fujikura Ltd

Fujitsu Ltd

Furukawa Electric Co Ltd

Hirose Electric Co Ltd

Hitachi Ltd

Honda Motor Co Ltd

Hoya Corp

ITOCHU Corp

Japan Airlines Corp

Japan Tobacco Inc

JFE Holdings Inc

JSR Corp

Kajima Corp

Kansai Electric Power Co Inc/The

Kao Corp

Kawasaki Heavy Industries Ltd

Keyence Corp

Kintetsu Group Holdings Co Ltd

Kirin Holdings Co Ltd

Kobe Steel Ltd

Komatsu Ltd

Konica Minolta Inc

Kubota Corp

Kuraray Co Ltd

Kyocera Corp

Kyushu Electric Power Co Inc

Lixil Corp

Marubeni Corp

129

Marui Group Co Ltd

Mitsubishi Chemical Holdings Corp

Mitsubishi Corp

Mitsubishi Electric Corp

Mitsubishi Estate Co Ltd

Mitsubishi Heavy Industries Ltd

Mitsubishi Materials Corp

Mitsubishi UFJ Financial Group Inc

Mitsui & Co Ltd

Mitsui Chemicals Inc

Mitsui Fudosan Co Ltd

Mitsui Mining & Smelting Co Ltd

Mitsui OSK Lines Ltd

Mizuho Financial Group Inc

MS&AD Insurance Group Holdings Inc

Murata Manufacturing Co Ltd

NEC Corp

NGK Insulators Ltd

NH Foods Ltd

Nikon Corp

Nintendo Co Ltd

Nippon Express Co Ltd

Nippon Paper Group Inc

Nippon Steel Corp

Nippon Telegraph & Telephone Corp

Nippon Television Holdings Inc

Nippon Yusen KK

Nipponkoa Insurance Co Ltd

Nissan Motor Co Ltd

Nissin Foods Holdings Co Ltd

Nitto Denko Corp

NOK Corp

Nomura Holdings Inc

NSK Ltd

NTT Data Corp

NTT DOCOMO Inc

Obayashi Corp

Odakyu Electric Railway Co Ltd

Oji Holdings Corp

Oriental Land Co Ltd/Japan

ORIX Corp

Osaka Gas Co Ltd

Panasonic Corp

Panasonic Electric Works Co Ltd

Pioneer Corp

Ricoh Co Ltd

130

Rohm Co Ltd

Sanyo Electric Co Ltd

Secom Co Ltd

Sekisui House Ltd

Seven & i Holdings Co Ltd

Sharp Corp/Japan

Shimizu Corp

Shin-Etsu Chemical Co Ltd

Shinsei Bank Ltd

Shiseido Co Ltd

SMBC Consumer Finance Co Ltd

SMC Corp

SoftBank Group Corp

Sompo Japan Insurance Inc/Old

Sony Group Corp

Sumitomo Chemical Co Ltd

Sumitomo Corp

Sumitomo Electric Industries Ltd

Sumitomo Metal Industries Ltd

Sumitomo Metal Mining Co Ltd

Sumitomo Mitsui Financial Group Inc

Sumitomo Mitsui Trust Bank Ltd

Sumitomo Realty & Development Co Ltd

Suzuki Motor Corp

T&D Holdings Inc

Taisei Corp

Taisho Pharmaceutical Co Ltd

Takeda Pharmaceutical Co Ltd

TDK Corp

Teijin Ltd

Terumo Corp

TFK Co Ltd

Tokio Marine Holdings Inc

Tokyo Electric Power Co Holdings Inc

Tokyo Electron Ltd

Tokyo Gas Co Ltd

Tokyu Corp

TonenGeneral Sekiyu KK

Toppan Printing Co Ltd

Toray Industries Inc

Toshiba Corp

TOTO Ltd

Toyo Seikan Group Holdings Ltd

Toyota Industries Corp

Toyota Motor Corp

Unicharm Corp

131

West Japan Railway Co

Yakult Honsha Co Ltd

Yamada Holdings Co Ltd

Yamato Holdings Co Ltd

Z Holdings Corp

S&P LA 40

Alfa SAB de CV

Ambev SA

America Movil SAB de CV

Aracruz Celulose SA

Banco Bradesco SA

Banco de Chile

Banco Santander Chile

Brasil Telecom Participacoes SA

Cemex SAB de CV

Centrais Eletricas Brasileiras SA

Cia Energetica de Minas Gerais

Cia Paranaense de Energia

Cia Siderurgica Nacional SA

Cyrela Brazil Realty SA Empreendimentos e Participacoes

Embraer SA

Empresas CMPC SA

Empresas COPEC SA

Enel Americas SA

Enel Generacion Chile SA

Falabella SA

Fomento Economico Mexicano SAB de CV

Gerdau SA

Grupo Carso SAB de CV

Grupo Modelo SAB de CV

Grupo Televisa SAB

Itau Unibanco Holding SA

Kimberly-Clark de Mexico SAB de CV

Latam Airlines Group SA

Petrobras Energia Participaciones SA

Petroleo Brasileiro SA

Sociedad Quimica y Minera de Chile SA

Tele Norte Leste Participacoes SA

Telefonos de Mexico SAB de CV

Tenaris SA

Unibanco - Uniao de Bancos Brasileiros SA

Vale SA

Wal-Mart de Mexico SAB de CV

Walmart Inc

S&P US 500

3M Co

132

Abbott Laboratories

Abercrombie & Fitch Co

Adobe Inc

Adtalem Global Education Inc

Affiliated Computer Services Inc

Air Products and Chemicals Inc

Akamai Technologies Inc

Alcoa Corp

Allegheny Technologies Inc

Allergan PLC

Allstate Corp/The

Alpha Appalachia Holdings LLC

Amazon.com Inc

Ameren Corp

American Electric Power Co Inc

American Express Co

American International Group Inc

Amgen Inc

Amphenol Corp

Andeavor

Aon PLC

APA Corp

Apartment Investment and Management Co

Apollo Education Group Inc

Apple Inc

Archer-Daniels-Midland Co

Autodesk Inc

Automatic Data Processing Inc

AutoNation Inc

AutoZone Inc

AvalonBay Communities Inc

Avery Dennison Corp

Avon Products Inc

Baker Hughes Holdings LLC

Bank of New York Mellon Corp/The

Baxter International Inc

Beam Therapeutics Inc

Best Buy Co Inc

Big Lots Inc

Biogen Inc

BJ Services Co

Black & Decker Corp/The

BMC Software Inc

Boston Properties Inc

Boston Scientific Corp

Bristol-Myers Squibb Co

133

Burlington Northern Santa Fe LLC

CA Inc

Cabot Oil & Gas Corp

Cameron International Corp

Capital One Financial Corp

Cardinal Health Inc

Caterpillar Inc

Celgene Corp

CenterPoint Energy Inc

Cephalon Inc

Charles Schwab Corp/The

Chesapeake Energy Corp

Chevron Corp

Chubb Corp/The

Cigna Corp

Cintas Corp

Cisco Systems Inc/Delaware

Citigroup Inc

Cleveland-Cliffs Inc

Clorox Co/The

Coca-Cola Co/The

Coca-Cola Europacific Partners PLC

Colgate-Palmolive Co

Comerica Inc

Consolidated Edison Inc

Constellation Brands Inc

Constellation Energy Group Inc

Corning Inc

Coventry Health Care Inc

CR Bard Inc

Cummins Inc

Danaher Corp

Darden Restaurants Inc

Dean Foods Co

Dell Technologies Inc

Devon Energy Corp

Diamond Offshore Drilling Inc

DIRECTV

Dominion Energy Inc

Dover Corp

Dow Inc

DR Horton Inc

DTE Energy Co

Duke Energy Corp

DuPont de Nemours Inc

Eastman Chemical Co

134

Eastman Kodak Co

Ecolab Inc

Edison International

Electronic Arts Inc

Eli Lilly & Co

EMC Corp

EMD Millipore Corp

Entergy Corp

EOG Resources Inc

EQT Corp

Equifax Inc

Equity Residential

Estee Lauder Cos Inc/The

ETC Sunoco Holdings LLC

Evernorth Health Inc

Eversource Energy

Exelon Corp

Expedia Group Inc

Expeditors International of Washington Inc

Exxon Mobil Corp

Fastenal Co

Federated Hermes Inc

FedEx Corp

Fifth Third Bancorp

First Horizon Corp

FirstEnergy Corp

Fiserv Inc

FLIR Systems Inc

Flowserve Corp

FMC Corp

Forest Road Acquisition Corp

Franklin Resources Inc

Freeport-McMoRan Inc

Frontier Communications Corp

Gap Inc/The

General Dynamics Corp

General Electric Co

Genuine Parts Co

Gilead Sciences Inc

Globe Life Inc

Goldman Sachs Group Inc/The

Goodyear Tire & Rubber Co/The

H&R Block Inc

Harman International Industries Inc

Hartford Financial Services Group Inc/The

Healthpeak Properties Inc

135

Hershey Co/The

Hess Corp

Hillshire Brands Co/The

Home Depot Inc/The

Honeywell International Inc

Hospira Inc

Host Hotels & Resorts Inc

HP Inc

Huntington Bancshares Inc/OH

Illinois Tool Works Inc

Intel Corp

Intercontinental Exchange Inc

International Business Machines Corp

International Flavors & Fragrances Inc

Intuitive Surgical Inc

Iron Mountain Inc

J M Smucker Co/The

Jabil Inc

Jacobs Engineering Group Inc

Jefferies Financial Group Inc

JPMorgan Chase & Co

Juniper Networks Inc

Keurig Dr Pepper Inc

Kimberly-Clark Corp

KLA Corp

Kohl's Corp

Kroger Co/The

L Brands Inc

Laboratory Corp of America Holdings

Leidos Holdings Inc

Lexmark International Inc

Lincoln National Corp

Linde Inc/CT

Lockheed Martin Corp

Loews Corp

Lorillard LLC

Lumen Technologies Inc

Macy's Inc

Marathon Oil Corp

Marsh & McLennan Cos Inc

Marshall & Ilsley Corp

Masco Corp

Mattel Inc

McCormick & Co Inc/MD

McKesson Corp

Mead Johnson Nutrition Co

136

Medco Health Solutions Inc

Medtronic PLC

Merck & Co Inc

MetLife Inc

Microsoft Corp

Molson Coors Beverage Co

Mondelez International Inc

Monster Worldwide Inc

Motorola Solutions Inc

Murphy Oil Corp

Mylan NV

Nabors Industries Ltd

National Semiconductor Corp

NextEra Energy Inc

Nicor Inc

NIKE Inc

NiSource Inc

Noble Energy Inc

Nordstrom Inc

Norfolk Southern Corp

Northern Trust Corp

Northrop Grumman Corp

NOV Inc

Nucor Corp

NVIDIA Corp

NYSE Euronext

ODP Corp/The

O-I Glass Inc

Old Copper Co Inc

O'Reilly Automotive Inc

Patterson Cos Inc

Paychex Inc

Peabody Energy Corp

People's United Financial Inc

Pepco Holdings LLC

PerkinElmer Inc

Pfizer Inc

PG&E Corp

Philip Morris International Inc

Piedmont Lithium Inc

Pitney Bowes Inc

Plum Creek Timber Co Inc

PNC Financial Services Group Inc/The

PPG Industries Inc

PPL Corp

Precision Castparts Corp

137

Principal Financial Group Inc

Procter & Gamble Co/The

Progressive Corp/The

Prudential Financial Inc

Public Service Enterprise Group Inc

Public Storage

Quest Diagnostics Inc

Ralph Lauren Corp

Raytheon Co

Raytheon Technologies Corp

Red Hat Inc

Regions Financial Corp

Republic Services Inc

Robert Half International Inc

Rockwell Automation Inc

Roper Technologies Inc

Rowan Cos Ltd

RS Legacy Corp

Safeway Inc

SanDisk LLC

SCANA Corp

Schlumberger NV

Sealed Air Corp

Sigma-Aldrich Corp

Simon Property Group Inc

Southern Co/The

Southwestern Energy Co

Spectra Energy LLC

Sprint Communications Inc

Sprint Corp

Stanley Black & Decker Inc

Staples Inc

Starbucks Corp

State Street Corp

Stericycle Inc

Stryker Corp

SunTrust Banks Inc

SUPERVALU Inc

Sysco Corp

Target Corp

TECO Energy Inc

Teradata Corp

Teradyne Inc

Textron Inc

Tiffany & Co

Time Warner Inc

138

TJX Cos Inc/The

Total System Services Inc

Travel + Leisure Co

Travelers Cos Inc/The

Truist Financial Corp

Tyson Foods Inc

United Parcel Service Inc

United States Steel Corp

UnitedHealth Group Inc

Unum Group

US Bancorp

VeriSign Inc

Verizon Communications Inc

VF Corp

Viacom Inc

ViacomCBS Inc

Viavi Solutions Inc

Vornado Realty Trust

Walgreens Boots Alliance Inc

Walt Disney Co/The

Waste Management Inc

Waters Corp

WEC Energy Group Inc

Wells Fargo & Co

Welltower Inc

Western Union Co/The

Weyerhaeuser Co

Whirlpool Corp

Williams Cos Inc/The

Xcel Energy Inc

Xerox Holdings Corp

XL Fleet Corp

XTO Energy Inc

Zimmer Biomet Holdings Inc

Zions Bancorp NA

139

References

Abbott, W. F., & Monsen, R. J. (1979). On the measurement of corporate social responsibility: Self-reported disclosures as a method of measuring corporate social involvement. Academy of management journal, 22(3), 501-515.

Abdi, H., Kacem, H., & Omri, M. A. B. (2018). Determinants of Web-based disclosure in the Middle East. Journal of Financial Reporting and Accounting.

Ahn, S. C., & Schmidt, P. (1995). Efficient estimation of models for dynamic panel data. Journal of econometrics, 68(1), 5-28.

Ali, W., Frynas, J. G., & Mahmood, Z. (2017). Determinants of corporate social responsibility (CSR) disclosure in developed and developing countries: a literature review. Corporate Social Responsibility and Environmental Management, 24(4), 273-294.

Andrikopoulos, A., & Kriklani, N. (2013). Environmental disclosure and financial characteristics of the firm: The case of Denmark. Corporate Social Responsibility and Environmental Management, 20(1), 55-64.

Arayssi, M., Dah, M., & Jizi, M. (2016). Women on boards, sustainability reporting and firm performance. Sustainability Accounting, Management and Policy Journal.

Archel, P., Husillos, J., Larrinaga, C., & Spence, C. (2009). Social disclosure, legitimacy theory and the role of the state. Accounting, auditing & accountability journal.

Arellano, M., & Bover, O. (1995). Another look at the instrumental variable estimation of error-components models. Journal of econometrics, 68(1), 29-51.

Arena, C., Liong, R., & Vourvachis, P. (2018). Carrot or stick: CSR disclosures by Southeast Asian companies. Sustainability Accounting, Management and Policy Journal, 9(4), 422-454.

Arvidsson, S. (2011). Disclosure of non-financial information in the annual report: A management-team perspective. Journal of intellectual capital, 12(2), 277-300.

Arvidsson, S. (2019). An exposé of the challenging practice development of sustainability reporting: From the first wave to the EU Directive (2014/95/EU). In Challenges in Managing Sustainable Business (pp. 3-24): Springer.

Austin, P. C. (2017). Double propensity-score adjustment: a solution to design bias or bias due to incomplete matching. Statistical methods in medical research, 26(1), 201-222.

Baldini, M., Dal Maso, L., Liberatore, G., Mazzi, F., & Terzani, S. (2018). Role of country-and firm-level determinants in environmental, social, and governance disclosure. Journal of Business Ethics, 150(1), 79-98.

Baltagi, B. (2008). Econometric analysis of panel data: John Wiley & Sons. Baron, D. P., Harjoto, M. A., & Jo, H. (2011). The economics and politics of corporate social

performance. Business and Politics, 13(2), 1-46. Barone, M. J., Miyazaki, A. D., & Taylor, K. A. (2000). The influence of cause-related marketing on

consumer choice: does one good turn deserve another? Journal of the academy of marketing Science, 28(2), 248-262.

Barron, O. E., Stanford, M. H., & Yu, Y. (2009). Further evidence on the relation between analysts' forecast dispersion and stock returns. Contemporary Accounting Research, 26(2), 329-357.

Bauer, R., & Hann, D. (2010). Corporate environmental management and credit risk. Becchetti, L., Ciciretti, R., & Hasan, I. (2015). Corporate social responsibility, stakeholder risk, and

idiosyncratic volatility. Journal of Corporate Finance, 35, 297-309. Benjamin, S. J., Regasa, D. G., Wellalage, N. H., & M Marathamuthu, M. S. (2020). Waste disclosure

and corporate cash holdings. Applied Economics, 52(49), 5399-5412. Benlemlih, M., & Girerd‐Potin, I. (2017). Corporate social responsibility and firm financial risk

reduction: On the moderating role of the legal environment. Journal of Business Finance & Accounting, 44(7-8), 1137-1166.

140

Benlemlih, M., Shaukat, A., Qiu, Y., & Trojanowski, G. (2018). Environmental and social disclosures and firm risk. Journal of Business Ethics, 152(3), 613-626.

Bernardi, C., & Stark, A. W. (2018). Environmental, social and governance disclosure, integrated reporting, and the accuracy of analyst forecasts. The British Accounting Review, 50(1), 16-31.

Bhattacharya, A., & Singh, P. J. (2018). Antecedents of agency problems in service outsourcing. International Journal of Production Research, 1-17.

Bhattacharya, U., Daouk, H., & Welker, M. (2003). The world price of earnings opacity. The accounting review, 78(3), 641-678.

Bhattacharyya, A., & Yang, H. (2019). Biodiversity disclosure in Australia: effect of GRI and institutional factors. Australasian Journal of Environmental Management, 26(4), 347-369.

Blundell, R., & Bond, S. (1998). Initial conditions and moment restrictions in dynamic panel data models. Journal of econometrics, 87(1), 115-143.

Borghei, Z., Leung, P., & Guthrie, J. (2018). Does voluntary greenhouse gas emissions disclosure reduce information asymmetry? Australian evidence. Afro-Asian Journal of Finance and Accounting, 8(2), 123-147.

Bouslah, K., Kryzanowski, L., & M’Zali, B. (2013). The impact of the dimensions of social performance on firm risk. Journal of Banking & Finance, 37(4), 1258-1273.

Bowen, H. R. (1953). Social responsibilities of the businessman: University of Iowa Press. Bowman, E. H., & Haire, M. (1975). A strategic posture toward corporate social responsibility.

California management review, 18(2), 49-58. Boyer-Allirol, B., & Barbu, E. (2017). Should European environmental law be strengthened? Revue

internationale de droit economique, 31(3), 109-124. Bozzolan, S., Favotto, F., & Ricceri, F. (2003). Italian annual intellectual capital disclosure. Journal of

intellectual capital. Branco, M. C., & Rodrigues, L. L. (2007). Positioning stakeholder theory within the debate on

corporate social responsibility. Electronic Journal of Business Ethics and Organization Studies.

Branco, M. C., & Rodrigues, L. L. (2008). Factors influencing social responsibility disclosure by Portuguese companies. Journal of Business Ethics, 83(4), 685-701.

Brown, T. J., & Dacin, P. A. (1997). The company and the product: Corporate associations and consumer product responses. The Journal of Marketing, 68-84.

Buchanan, B., Cao, C. X., & Chen, C. (2018). Corporate social responsibility, firm value, and influential institutional ownership. Journal of Corporate Finance, 52, 73-95.

Burgman, R., Roos, G., Boldt-Christmas, L., & Pike, S. (2007). Information needs of internal and external stakeholders and how to respond: reporting on operations and intellectual capital. International Journal of Accounting, Auditing and Performance Evaluation, 4(4-5), 529-546.

Cahaya, F. R., Porter, S., Tower, G., & Brown, A. (2015). The Indonesian Government’s coercive pressure on labour disclosures. Sustainability Accounting, Management and Policy Journal.

Cai, L., Cui, J., & Jo, H. (2016). Corporate environmental responsibility and firm risk. Journal of Business Ethics, 139(3), 563-594.

Cameron, A. C., & Trivedi, P. K. (2005). Microeconometrics: methods and applications. In (pp. 703): Cambridge university press.

Camilleri. (2015). Environmental, social and governance disclosures in Europe. Sustainability Accounting, Management and Policy Journal, 6(2), 224-242.

Carroll, A. B. (1979). A three-dimensional conceptual model of corporate performance. Academy of management review, 4(4), 497-505.

Caruana, R., & Crane, A. (2008). Constructing consumer responsibility: Exploring the role of corporate communications. Organization Studies, 29(12), 1495-1519.

CED. (1971). Social responsibilities of business corporations: The Committee. Chang, K., Kim, I., & Li, Y. (2014). The heterogeneous impact of corporate social responsibility

activities that target different stakeholders. Journal of Business Ethics, 125(2), 211-234.

141

Chen, K. H., & Metcalf, R. W. (1980). The relationship between pollution control record and financial indicators revisited. The accounting review, 55(1), 168-177.

Chien, Y.-T., & Lu, H.-M. (2015). Firm websites and the risk of firm. Industrial Management & Data Systems, 115(3), 504-520.

Chiu, T.-K., & Wang, Y.-H. (2015). Determinants of social disclosure quality in Taiwan: An application of stakeholder theory. Journal of Business Ethics, 129(2), 379-398.

Cho, S. Y., Lee, C., & Pfeiffer Jr, R. J. (2013). Corporate social responsibility performance and information asymmetry. Journal of Accounting and Public Policy, 32(1), 71-83.

Chopra, S., & Wu, P.-J. (2016). Eco-activities and operating performance in the computer and electronics industry. European Journal of Operational Research, 248(3), 971-981.

Chung, H. Y., & Kim, J. B. (1994). THE USE OF MULTIPLE INSTRUMENTS FOR MEASUREMENT OF EARNINGS FORECAST ERRORS, FIRM SIZE EFFECT AND THE QUALITY OF ANALYSTS’FORECAST ERRORS. Journal of Business Finance & Accounting, 21(5), 707-727.

Cochran, P. L., & Wood, R. A. (1984). Corporate social responsibility and financial performance. Academy of management journal, 27(1), 42-56.

Coluccia, D., Fontana, S., & Solimene, S. (2016). Disclosure of corporate social responsibility: a comparison between traditional and digital reporting. An empirical analysis on Italian listed companies. International Journal of Managerial and Financial Accounting, 8(3-4), 230-246.

Cormier, D., Ledoux, M.-J., & Magnan, M. (2011). The informational contribution of social and environmental disclosures for investors. Management Decision, 49(8), 1276-1304.

Cormier, D., & Magnan, M. (1999). Corporate environmental disclosure strategies: determinants, costs and benefits. Journal of Accounting, Auditing & Finance, 14(4), 429-451.

Cormier, D., & Magnan, M. (2014). The impact of social responsibility disclosure and governance on financial analysts’ information environment. Corporate Governance, 14(4), 467-484.

Cormier, D., & Magnan, M. (2015). The economic relevance of environmental disclosure and its impact on corporate legitimacy: An empirical investigation. Business Strategy and the Environment, 24(6), 431-450.

Cortesi, A., & Vena, L. (2019). Disclosure quality under integrated reporting: A Value Relevance Approach. Journal of cleaner production, 220, 745-755.

Cosset, J.-C., Somé, H. Y., & Valéry, P. (2016). Does competition matter for corporate governance? The role of country characteristics. Journal of Financial and Quantitative analysis, 51(4), 1231-1267.

Crisóstomo, V. L., de Azevedo Prudêncio, P., & Forte, H. C. (2017). An analysis of the adherence to GRI for disclosing information on social action and sustainability concerns. In Advances in Environmental Accounting & Management: Social and Environmental Accounting in Brazil: Emerald Publishing Limited.

Crouch, C. (2006). Modelling the firm in its market and organizational environment: Methodologies for studying corporate social responsibility. Organization Studies, 27(10), 1533-1551.

Cui, J., Jo, H., & Na, H. (2018). Does corporate social responsibility affect information asymmetry? Journal of Business Ethics, 148(3), 549-572.

Cuong. (2013). Which covariates should be controlled in propensity score matching? Evidence from a simulation study. Statistica Neerlandica, 67(2), 169-180.

Davis, K. (1960). Can business afford to ignore social responsibilities? California management review, 2(3), 70-76.

De Klerk, M., De Villiers, C., & Van Staden, C. (2015). The influence of corporate social responsibility disclosure on share prices. Pacific Accounting Review.

Del Bosco, B., & Misani, N. (2016). The effect of cross-listing on the environmental, social, and governance performance of firms. Journal of World Business, 51(6), 977-990.

Demir, M., & Min, M. (2019). Consistencies and discrepancies in corporate social responsibility reporting in the pharmaceutical industry. Sustainability Accounting, Management and Policy Journal.

142

Dhaliwal, Li, Tsang, & Yang. (2011). Voluntary nonfinancial disclosure and the cost of equity capital: The initiation of corporate social responsibility reporting. The accounting review, 86(1), 59-100.

Dhaliwal, Li, O. Z., Tsang, A., & Yang, Y. G. (2014). Corporate social responsibility disclosure and the cost of equity capital: The roles of stakeholder orientation and financial transparency. Journal of Accounting and Public Policy, 33(4), 328-355.

Dhaliwal, Radhakrishnan, Tsang, & Yang. (2012). Nonfinancial disclosure and analyst forecast accuracy: International evidence on corporate social responsibility disclosure. The accounting review, 87(3), 723-759.

Diamond, D. W., & Verrecchia, R. E. (1991). Disclosure, liquidity, and the cost of capital. The journal of finance, 46(4), 1325-1359.

Dichev, I. D., & Tang, V. W. (2009). Earnings volatility and earnings predictability. Journal of accounting and economics, 47(1-2), 160-181.

Diether, K. B., Malloy, C. J., & Scherbina, A. (2002). Differences of opinion and the cross section of stock returns. The journal of finance, 57(5), 2113-2141.

Directives. (2014). Directive 2014/95/eu of the European parliament and of the council. Retrieved from https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:32014L0095&from=EN

Donaldson, T., & Dunfee, T. W. (1999). Ties that bind: A social contracts approach to business ethics. Driscoll, J. C., & Kraay, A. C. (1998). Consistent covariance matrix estimation with spatially dependent

panel data. Review of economics and statistics, 80(4), 549-560. Drucker, P. F. (1984). Converting social problems into business opportunities: The new meaning of

corporate social responsibility. California Management Review (pre-1986), 26(000002), 53. Du, S., & Vieira, E. T. (2012). Striving for legitimacy through corporate social responsibility: Insights

from oil companies. Journal of Business Ethics, 110(4), 413-427. Dumitru, M., Dyduch, J., Gușe, R.-G., & Krasodomska, J. (2017). Corporate reporting practices in

Poland and Romania–an ex-ante study to the new non-financial reporting European Directive. Accounting in Europe, 14(3), 279-304.

Easley, D., & O'hara, M. (2004). Information and the cost of capital. The journal of finance, 59(4), 1553-1583.

Eccles, R. G., Serafeim, G., & Krzus, M. P. (2011). Market interest in nonfinancial information. Journal of Applied Corporate Finance, 23(4), 113-127.

Eichenbaum, M. S., Hansen, L. P., & Singleton, K. J. (1988). A time series analysis of representative agent models of consumption and leisure choice under uncertainty. The Quarterly Journal of Economics, 103(1), 51-78.

Elliott, R. K., & Jacobson, P. D. (1994). Costs and benefits of business information. Accounting horizons, 8(4), 80-96.

Emma, G.-M., & Jennifer, M.-F. (2021). Is SDG reporting substantial or symbolic? An examination of controversial and environmentally sensitive industries. Journal of cleaner production, 298, 126781.

Erragragui, E. (2018). Do creditors price firms’ environmental, social and governance risks? Research in International Business and Finance, 45, 197-207.

Espahbodi, R., Dugar, A., & Tehranian, H. (2001). Further evidence on optimism and underreaction in analysts' forecasts. Review of Financial Economics, 10(1), 1-21.

Fahad, P., & Nidheesh, K. (2020). Determinants of CSR disclosure: an evidence from India. Journal of Indian Business Research.

Fallan, E., & Fallan, L. (2019). Corporate tax behaviour and environmental disclosure: Strategic trade-offs across elements of CSR? Scandinavian Journal of Management, 101042.

Fama, E. F., & French, K. R. (1998). Value versus growth: The international evidence. The journal of finance, 53(6), 1975-1999.

https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:32014L0095&from=EN

https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:32014L0095&from=EN

143

Fanasch, P., & Frick, B. (2020). The value of signals: Do self-declaration and certification generate price premiums for organic and biodynamic wines? Journal of cleaner production, 249, 119415.

Felber, C., Campos, V., & Sanchis, J. R. (2019). The common good balance sheet, an adequate tool to capture non-financials? Sustainability, 11(14), 3791.

Flores, E., Fasan, M., Mendes‐da‐Silva, W., & Sampaio, J. O. (2019). Integrated reporting and capital markets in an international setting: The role of financial analysts. Business Strategy and the Environment, 28(7), 1465-1480.

Frank, D. H., & Obloj, T. (2014). Firm‐specific human capital, organizational incentives, and agency costs: Evidence from retail banking. Strategic management journal, 35(9), 1279-1301.

Freeman, R. E. (2010). Strategic management: A stakeholder approach: Cambridge university press. Gamerschlag, R., Möller, K., & Verbeeten, F. (2011). Determinants of voluntary CSR disclosure:

empirical evidence from Germany. Review of Managerial Science, 5(2-3), 233-262. Gangi, F., Meles, A., Monferrà, S., & Mustilli, M. (2018). Does corporate social responsibility help the

survivorship of SMEs and large firms? Global Finance Journal. García-Sánchez, I.-M., Suárez-Fernández, O., & Martínez-Ferrero, J. (2019). Female directors and

impression management in sustainability reporting. International Business Review, 28(2), 359-374.

García‐Sánchez, I. M., Gómez‐Miranda, M. E., David, F., & Rodríguez‐Ariza, L. (2019). Analyst coverage and forecast accuracy when CSR reports improve stakeholder engagement: The Global Reporting Initiative‐International Finance Corporation disclosure strategy. Corporate Social Responsibility and Environmental Management, 26(6), 1392-1406.

Garcia, A. S., Mendes-Da-Silva, W., & Orsato, R. J. (2017). Sensitive industries produce better ESG performance: Evidence from emerging markets. Journal of cleaner production, 150, 135-147.

Garrido‐Miralles, P., Zorio‐Grima, A., & García‐Benau, M. A. (2016). Sustainable development, stakeholder engagement and analyst forecasts’ accuracy: Positive evidence from the Spanish setting. Sustainable Development, 24(2), 77-88.

Gerwanski, J. Does it pay off? Integrated reporting and cost of debt: European evidence. Corporate Social Responsibility and Environmental Management.

Gilbert, S. (2002). The transparency evolution. Paper presented at the Environmental Forum. Godfrey, P. C. (2005). The relationship between corporate philanthropy and shareholder wealth: A

risk management perspective. Academy of management review, 30(4), 777-798. Godfrey, P. C., Merrill, C. B., & Hansen, J. M. (2009). The relationship between corporate social

responsibility and shareholder value: An empirical test of the risk management hypothesis. Strategic management journal, 30(4), 425-445.

Graham, A., & Maher, J. J. (2006). Environmental liabilities, bond ratings, and bond yields. In Environmental Accounting (pp. 111-142): Emerald Group Publishing Limited.

Graham, A., Maher, J. J., & Northcut, W. D. (2001). Environmental liability information and bond ratings. Journal of Accounting, Auditing & Finance, 16(2), 93-116.

Greening, D. W., & Turban, D. B. (2000). Corporate social performance as a competitive advantage in attracting a quality workforce. Business & Society, 39(3), 254-280.

Gregoriou, A., Ioannidis, C., & Skerratt, L. (2005). Information Asymmetry and the Bid‐Ask Spread: Evidence From the UK. Journal of Business Finance & Accounting, 32(9‐10), 1801-1826.

Grougiou, V., Dedoulis, E., & Leventis, S. (2016). Corporate social responsibility reporting and organizational stigma: The case of “sin” industries. Journal of Business Research, 69(2), 905-914.

Gujarati, D. N. (2009). Basic econometrics: Tata McGraw-Hill Education. Guo, S., & Fraser, M. W. (2014). Propensity score analysis: Statistical methods and applications (Vol.

11): SAGE publications.

144

Han, J.-J., Kim, H. J., & Yu, J. (2016). Empirical study on relationship between corporate social responsibility and financial performance in Korea. Asian Journal of Sustainability and Social Responsibility, 1(1), 61.

Handelman, J. M., & Arnold, S. J. (1999). The role of marketing actions with a social dimension: Appeals to the institutional environment. The Journal of Marketing, 33-48.

Hassan, A., Adhikariparajuli, M., Fletcher, M., & Elamer, A. (2019). Integrated reporting in UK higher education institutions. Sustainability Accounting, Management and Policy Journal.

Hassan, O., & Giorgioni, G. (2019). The impact of corruption on analyst coverage. Managerial Auditing Journal.

Hassan, O. A. (2018). The impact of voluntary environmental disclosure on firm value: Does organizational visibility play a mediation role? Business Strategy and the Environment, 27(8), 1569-1582.

Hoffmann, E., Dietsche, C., & Hobelsberger, C. (2018). Between mandatory and voluntary: non-financial reporting by German companies. Paper presented at the NachhaltigkeitsManagementForum| Sustainability Management Forum.

Hope, O. K. (2003). Disclosure practices, enforcement of accounting standards, and analysts' forecast accuracy: An international study. Journal of accounting research, 41(2), 235-272.

Huang, & Kung, F.-H. (2010). Drivers of environmental disclosure and stakeholder expectation: Evidence from Taiwan. Journal of Business Ethics, 96(3), 435-451.

Iatridis, G. E. (2015). Corporate philanthropy in the US stock market: Evidence on corporate governance, value relevance and earnings manipulation. International Review of Financial Analysis, 39, 113-126.

Iatridis, G. E. (2016). Financial reporting language in financial statements: Does pessimism restrict the potential for managerial opportunism? International Review of Financial Analysis, 45, 1-17.

IIRC. (2011). Towards integrated reporting: Communicating value in the 21st century. Author, London.

Ioannou, I., & Serafeim, G. (2011). The Consequences of Mandatory Corporate Sustainability Reporting.

Ioannou, I., & Serafeim, G. (2012). What drives corporate social performance? The role of nation-level institutions. Journal of International Business Studies, 43(9), 834-864.

Ioannou, I., & Serafeim, G. (2015). The impact of corporate social responsibility on investment recommendations: Analysts' perceptions and shifting institutional logics. Strategic management journal, 36(7), 1053-1081.

Ioannou, I., & Serafeim, G. (2017). The consequences of mandatory corporate sustainability reporting. Harvard Business School research working paper(11-100).

Ismail, T. H., & El‐Shaib, N. M. (2012). Impact of market and organizational determinants on voluntary disclosure in Egyptian companies. Meditari Accountancy Research.

Jain, A., Keneley, M., & Thomson, D. (2015). Voluntary CSR disclosure works! Evidence from Asia-Pacific banks. Social Responsibility Journal, 11(1), 2-18.

Jensen, M. C., & Meckling, W. H. (1976). Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of financial economics, 3(4), 305-360.

Jensen, M. C., & Meckling, W. H. (1979). Theory of the firm: Managerial behavior, agency costs, and ownership structure. In Economics social institutions (pp. 163-231): Springer.

Jo, H., & Harjoto, M. A. (2011). Corporate governance and firm value: The impact of corporate social responsibility. Journal of Business Ethics, 103(3), 351-383.

Jo, H., & Harjoto, M. A. (2012). The causal effect of corporate governance on corporate social responsibility. Journal of Business Ethics, 106(1), 53-72.

Jo, H., & Na, H. (2012). Does CSR reduce firm risk? Evidence from controversial industry sectors. Journal of Business Ethics, 110(4), 441-456.

Johnson. (1971). Business in contemporary society: Framework and issues: Wadsworth Pub. Co.

145

Johnson. (2004). Forecast dispersion and the cross section of expected returns. The journal of finance, 59(5), 1957-1978.

Kaplan, S. E., & Ruland, R. G. (1991). Positive theory, rationality and accounting regulation (Vol. 2). Kartadjumena, E., & Rodgers, W. (2019). Executive compensation, sustainability, climate,

environmental concerns, and company financial performance: Evidence from Indonesian commercial banks. Sustainability, 11(6), 1673.

Kassinis, G. I., & Soteriou, A. C. (2003). Greening the service profit chain: The impact of environmental management practices. Production and operations Management, 12(3), 386-403.

Khan, A., Muttakin, M. B., & Siddiqui, J. (2013). Corporate governance and corporate social responsibility disclosures: Evidence from an emerging economy. Journal of Business Ethics, 114(2), 207-223.

Kilian, T., & Hennigs, N. (2014). Corporate social responsibility and environmental reporting in controversial industries. European Business Review, 26(1), 79-101.

Kim, O., & Verrecchia, R. E. (1994). Market liquidity and volume around earnings announcements. Journal of accounting and economics, 17(1-2), 41-67.

Konar, S., & Cohen, M. A. (2001). Does the market value environmental performance? Review of economics and statistics, 83(2), 281-289.

Korca, B., & Costa, E. (2021). Directive 2014/95/EU: building a research agenda. Journal of Applied Accounting Research.

Kothari, S. P., Li, X., & Short, J. E. (2009). The effect of disclosures by management, analysts, and business press on cost of capital, return volatility, and analyst forecasts: A study using content analysis. The accounting review, 84(5), 1639-1670.

La Torre, M., Sabelfeld, S., Blomkvist, M., Tarquinio, L., & Dumay, J. (2018). Harmonising non-financial reporting regulation in Europe: Practical forces and projections for future research. Meditari Accountancy Research.

Lambert, R., Leuz, C., & Verrecchia, R. E. (2007). Accounting information, disclosure, and the cost of capital. Journal of accounting research, 45(2), 385-420.

Landi, G., & Sciarelli, M. (2019). Towards a more ethical market: the impact of ESG rating on corporate financial performance. Social Responsibility Journal, 15(1), 11-27.

Lang, M. H., & Lundholm, R. J. (1996). Corporate disclosure policy and analyst behavior. Accounting Review, 467-492.

Lee, & Yeo, G. H.-H. (2016). The association between integrated reporting and firm valuation. Review of Quantitative Finance and Accounting, 47(4), 1221-1250.

Lee, D. (2017). Corporate social responsibility and management forecast accuracy. Journal of Business Ethics, 140(2), 353-367.

Lee, K., & Lee, H. (2019). How does CSR activity affect sustainable growth and value of corporations? Evidence from Korea. Sustainability, 11(2), 508.

Leopizzi, R., Iazzi, A., Venturelli, A., & Principale, S. (2020). Nonfinancial risk disclosure: The “state of the art” of Italian companies. Corporate Social Responsibility and Environmental Management, 27(1), 358-368.

Li, M.-Y., & Wu, J.-S. (2014). Analysts’ forecast dispersion and stock returns: A quantile regression approach. Journal of Behavioral Finance, 15(3), 175-183.

Lins, K. V., Servaes, H., & Tamayo, A. (2017). Social capital, trust, and firm performance: The value of corporate social responsibility during the financial crisis. The journal of finance, 72(4), 1785-1824.

Liu, X., & Anbumozhi, V. (2009). Determinant factors of corporate environmental information disclosure: an empirical study of Chinese listed companies. Journal of cleaner production, 17(6), 593-600.

Liu, Y., Wei, Z., & Xie, F. (2014). Do women directors improve firm performance in China? Journal of Corporate Finance, 28, 169-184.

146

Lu, L. Y., Shailer, G., & Yu, Y. (2017). Corporate social responsibility disclosure and the value of cash holdings. European Accounting Review, 26(4), 729-753.

Lueg, K., Krastev, B., & Lueg, R. (2019). Bidirectional effects between organizational sustainability disclosure and risk. Journal of cleaner production, 229, 268-277.

Maaloul, A., Ben Amar, W., & Zeghal, D. (2016). Voluntary disclosure of intangibles and analysts’ earnings forecasts and recommendations. Journal of Applied Accounting Research, 17(4), 421-439.

Manes-Rossi, F., Tiron-Tudor, A., Nicolò, G., & Zanellato, G. (2018). Ensuring more sustainable reporting in Europe using non-financial disclosure—De facto and de jure evidence. Sustainability, 10(4), 1162.

Manne, H. G., & Wallich, H. (1972). The modern corporation and social responsibility. Books. Martínez-Ferrero, J., Rodríguez-Ariza, L., García-Sánchez, I.-M., & Cuadrado-Ballesteros, B. (2017).

Corporate social responsibility disclosure and information asymmetry: the role of family ownership. Review of Managerial Science, 1-32.

Martínez-Ferrero, J., Rodríguez-Ariza, L., García-Sánchez, I.-M., & Cuadrado-Ballesteros, B. (2018). Corporate social responsibility disclosure and information asymmetry: the role of family ownership. Review of Managerial Science, 12(4), 885-916.

Martínez‐Ferrero, J., Ruiz‐Cano, D., & García‐Sánchez, I. M. (2016). The causal link between sustainable disclosure and information asymmetry: The moderating role of the stakeholder protection context. Corporate Social Responsibility and Environmental Management, 23(5), 319-332.

Matuszak, Ł., & Różańska, E. (2017). CSR disclosure in Polish-listed companies in the light of Directive 2014/95/EU requirements: Empirical evidence. Sustainability, 9(12), 2304.

McCaffrey, D. F., Griffin, B. A., Almirall, D., Slaughter, M. E., Ramchand, R., & Burgette, L. F. (2013). A tutorial on propensity score estimation for multiple treatments using generalized boosted models. Statistics in medicine, 32(19), 3388-3414.

McCaffrey, D. F., Ridgeway, G., & Morral, A. R. (2004). Propensity score estimation with boosted regression for evaluating causal effects in observational studies. Psychological methods, 9(4), 403.

McGuire, J. B., Sundgren, A., & Schneeweis, T. (1988). Corporate social responsibility and firm financial performance. Academy of management journal, 31(4), 854-872.

McKinsey, & Company. (2002). Governance is increasingly at the heart of investment decisions, new Mckinsey survey shows. Retrieved from https://www.iasplus.com/en/binary/resource/mckinsey.pdf

Michaels, A., & Grüning, M. (2017). Relationship of corporate social responsibility disclosure on information asymmetry and the cost of capital. Journal of Management Control, 28(3), 251-274.

Mohamed, H. H., Masih, M., & Bacha, O. I. (2015). Why do issuers issue Sukuk or conventional bond? Evidence from Malaysian listed firms using partial adjustment models. Pacific-Basin Finance Journal, 34, 233-252.

Moskowitz, M. (1972). Choosing socially responsible stocks. Business and Society Review, 1(1), 71-75. Mostafa, M. M. (2016). Egyptian consumers' willingness to pay for carbon-labeled products: A

contingent valuation analysis of socio-economic factors. Journal of cleaner production, 135, 821-828.

Mukherjee, A., & Nuñez, R. (2019). Doing well by doing good: can voluntary CSR reporting enhance financial performance? Journal of Indian Business Research.

Muslu, V., Mutlu, S., Radhakrishnan, S., & Tsang, A. (2019). Corporate social responsibility report narratives and analyst forecast accuracy. Journal of Business Ethics, 154(4), 1119-1142.

Nadeem, M. (2016). Intellectual capital and firm performance: evidence from developed, emerging and frontier markets of the world. Lincoln University,

https://www.iasplus.com/en/binary/resource/mckinsey.pdf

147

Nanayakkara, M., & Colombage, S. (2019). Do investors in Green Bond market pay a premium? Global evidence. Applied Economics, 51(40), 4425-4437.

Nations, U. (2015). Paris Agreement. Retrieved from https://unfccc.int/sites/default/files/english_paris_agreement.pdf

Ness, K. E., & Mirza, A. (1991). Corporate social disclosure: A note on a test of agency theory. The British Accounting Review, 23(3), 211-217.

Nguyen, Agbola, F. W., & Choi, B. (2016). Does corporate social responsibility reduce information asymmetry? Empirical evidence from Australia. Australian Journal of Management, 0312896218797163.

Nguyen, & Nguyen. (2015). The effect of corporate social responsibility on firm risk. Social Responsibility Journal, 11(2), 324-339.

Nguyen, & Nguyen, A. (2015). The effect of corporate social responsibility on firm risk. Social Responsibility Journal, 11(2), 324-339.

Nollet, J., Filis, G., & Mitrokostas, E. (2016). Corporate social responsibility and financial performance: A non-linear and disaggregated approach. Economic Modelling, 52, 400-407.

O'Bannon, D., & Preston, L. (1993). Corporate social responsibility and firm financial performance relationships: A typology and analysis of possible relationships. Paper presented at the annual meeting of the Academy of Management, Atlanta.

Ochi, N. (2018). Reporting of Real Option Value Related to ESG: Including Complementary Systems for Disclosure Incentives. International Journal of Financial Research, 9(4), 19-34.

Oikonomou, I., Brooks, C., & Pavelin, S. (2012). The impact of corporate social performance on financial risk and utility: A longitudinal analysis. Financial Management, 41(2), 483-515.

Orlitzky, M., & Benjamin, J. D. (2001). Corporate social performance and firm risk: A meta-analytic review. Business & Society, 40(4), 369-396.

Park, S., & Park, K. (2019). Can investors profit from security analyst recommendations?: New evidence on the value of consensus recommendations. Finance Research Letters, 30, 403-413.

Preston, L. E. (1982). Analysing Corporate Social Performance: Methods and Results. In Management Accountability and Corporate Governance (pp. 163-182): Springer.

Purushothaman, M., Tower, G., Hancock, P., & Taplin, R. (2000). Determinants of corporate social reporting practices of listed Singapore companies. Pacific Accounting Review, 12(2), 101.

Randolph, J. J., & Falbe, K. (2014). A step-by-step guide to propensity score matching in R. Practical Assessment, Research & Evaluation, 19.

Ratnatunga, J., & Jones, S. (2012). A methodology to rank the quality and comprehensiveness of sustainability information provided in publicly listed company reports. Contemporary Issues in Sustainability Accounting, Assurance and Reporting, 227.

Rettab, B., Brik, A. B., & Mellahi, K. (2009). A study of management perceptions of the impact of corporate social responsibility on organisational performance in emerging economies: the case of Dubai. Journal of Business Ethics, 89(3), 371-390.

Reverte, C. (2009). Determinants of corporate social responsibility disclosure ratings by Spanish listed firms. Journal of Business Ethics, 88(2), 351-366.

Reverte, C. (2012). The impact of better corporate social responsibility disclosure on the cost of equity capital. Corporate Social Responsibility and Environmental Management, 19(5), 253-272.

Reverte, C. (2016). Corporate social responsibility disclosure and market valuation: evidence from Spanish listed firms. Review of Managerial Science, 10(2), 411-435.

Rezaee, Z., Alipour, M., Faraji, O., Ghanbari, M., & Jamshidinavid, B. (2020). Environmental disclosure quality and risk: the moderating effect of corporate governance. Sustainability Accounting, Management and Policy Journal.

Roberts, M. R., & Whited, T. M. (2013). Endogeneity in empirical corporate finance1. In Handbook of the Economics of Finance (Vol. 2, pp. 493-572): Elsevier.

https://unfccc.int/sites/default/files/english_paris_agreement.pdf

148

Roodman, D. (2009). How to do xtabond2: An introduction to difference and system GMM in Stata. The stata journal, 9(1), 86-136.

Russo, M. V., & Fouts, P. A. (1997). A resource-based perspective on corporate environmental performance and profitability. Academy of management journal, 40(3), 534-559.

Sassen, R., Hinze, A.-K., & Hardeck, I. (2016). Impact of ESG factors on firm risk in Europe. Journal of Business Economics, 86(8), 867-904.

Schulz, J.-F. (2017). Does greater disclosure of information on CSR performance improve analysts’ forecast accuracy? Journal of Environmental Law and Policy, 40(2), 134-160.

Sethi, S. P. (1975). Dimensions of corporate social performance: An analytical framework. California management review, 17(3), 58-64.

Sethi, S. P., Martell, T. F., & Demir, M. (2016). Building corporate reputation through corporate social responsibility (CSR) reports: The case of extractive industries. Corporate Reputation Review, 19(3), 219-243.

Sethi, S. P., Martell, T. F., & Demir, M. (2017). Enhancing the role and effectiveness of corporate social responsibility (CSR) reports: The missing element of content verification and integrity assurance. Journal of Business Ethics, 144(1), 59-82.

Shamsuddin, A., & Kim, J. H. (2010). Short‐horizon return predictability in international equity markets. Financial Review, 45(2), 469-484.

Shane, P. B., & Spicer, B. H. (1983). Market response to environmental information produced outside the firm. Accounting Review, 521-538.

Shapiro, S. P. (2005). Agency theory. Annual review of sociology, 31. Siew, R. Y., Balatbat, M. C., & Carmichael, D. G. (2016). The impact of ESG disclosures and

institutional ownership on market information asymmetry. Asia-Pacific Journal of Accounting & Economics, 23(4), 432-448.

Stanny, E., & Ely, K. (2008). Corporate environmental disclosures about the effects of climate change. Corporate Social Responsibility and Environmental Management, 15(6), 338-348.

Tagesson, T., Blank, V., Broberg, P., & Collin, S. O. (2009). What explains the extent and content of social and environmental disclosures on corporate websites: a study of social and environmental reporting in Swedish listed corporations. Corporate Social Responsibility and Environmental Management, 16(6), 352-364.

Tamimi, N., & Sebastianelli, R. (2017). Transparency among S&P 500 companies: An analysis of ESG disclosure scores. Management Decision, 55(8), 1660-1680.

Testarmata, S., Ciaburri, M., Fortuna, F., & Sergiacomi, S. (2020). Harmonization of Non-financial Reporting Regulation in Europe: A Study of the Transposition of the Directive 2014/95/EU. In Accountability, Ethics and Sustainability of Organizations (pp. 67-88): Springer.

Thomas, A. (2001). Corporate environmental policy and abnormal stock price returns: an empirical investigation. Business Strategy and the Environment, 10(3), 125-134.

Ullmann, A. A. (1985). Data in search of a theory: A critical examination of the relationships among social performance, social disclosure, and economic performance of US firms. Academy of management review, 10(3), 540-557.

Varadarajan, P. R., & Menon, A. (1988). Cause-related marketing: A coalignment of marketing strategy and corporate philanthropy. The Journal of Marketing, 58-74.

Velte, P. (2020). Institutional ownership, environmental, social, and governance performance and disclosure–a review on empirical quantitative research. Problems and Perspectives in Management, 18(3), 282.

Venturelli, A., Caputo, F., Cosma, S., Leopizzi, R., & Pizzi, S. (2017). Directive 2014/95/EU: Are Italian companies already compliant? Sustainability, 9(8), 1385.

Vollero, A., Conte, F., Siano, A., & Covucci, C. (2019). Corporate social responsibility information and involvement strategies in controversial industries. Corporate Social Responsibility and Environmental Management, 26(1), 141-151.

149

Waddock, S. A., & Graves, S. B. (1997). The corporate social performance–financial performance link. Strategic management journal, 18(4), 303-319.

Wagner, T., Lutz, R. J., & Weitz, B. A. (2009). Corporate hypocrisy: Overcoming the threat of inconsistent corporate social responsibility perceptions. Journal of marketing, 73(6), 77-91.

Wang, Z., & Sarkis, J. (2017). Corporate social responsibility governance, outcomes, and financial performance. Journal of cleaner production, 162, 1607-1616.

Wei, Z., & Xue, J. (2015). Fair value accounting of financial assets and analyst forecasts. China Journal of Accounting Studies, 3(4), 294-319.

Wintoki, M. B., Linck, J. S., & Netter, J. M. (2012). Endogeneity and the dynamics of internal corporate governance. Journal of financial economics, 105(3), 581-606.

Wokutch, R. E., & McKinney, E. W. (1991). Behavioral and perceptual measures of corporate social performance. Research in corporate social performance and policy, 12, 309-330.

Xu, S., & Liu, D. (2018). Do Financial Markets Care about Corporate Social Responsibility Disclosure? Further Evidence from China. Australian Accounting Review, 28(1), 79-103.

Yang, M., Cheng, X., Sun, Q., & Lu, C. (2019). How does analyst forecast dispersion affect SEO discounts in uniform-price auction system? Evidence from investor bids in China. International Review of Economics & Finance, 63, 198-208.

Yoon, B., & Chung, Y. (2018). The effects of corporate social responsibility on firm performance: A stakeholder approach. Journal of Hospitality and Tourism Management, 37, 89-96.

Yu, E. P. y., Guo, C. Q., & Luu, B. V. (2018). Environmental, social and governance transparency and firm value. Business Strategy and the Environment, 27(7), 987-1004.

Yu, M. (2010). Analyst forecast properties, analyst following and governance disclosures: A global perspective. Journal of International Accounting, Auditing and Taxation, 19(1), 1-15.

Yusoff, R., Yusoff, H., Abd Rahman, S. A., & Darus, F. (2019). Investigating Sustainability Reporting from the Lens of Stakeholder Pressures and Isomorphism. Journal of Asia-Pacific Business, 20(4), 302-321.

Zeng, H., Zhang, T., Zhou, Z., Zhao, Y., & Chen, X. (2020). Water disclosure and firm risk: Empirical evidence from highly water‐sensitive industries in China. Business Strategy and the Environment, 29(1), 17-38.

Non-Financial Information Disclosures and Firm Risk

Documents