Using corporate social responsibility performance to evaluate financial disclosure credibility

Abstract

Due to the paucity of immediate and direct information about financial disclosure credibility, it is often difficult for investors to assess the credibility of financial disclosures (e.g. whether reported earnings are biased). Given this situation, the present study proposes and finds that investors use additional cues, such as information about corporate social responsibility (CSR) performance, to form overall impressions about management's honesty, credibility, and trustworthiness. Similar to other findings in the halo effect literature, we find that these overall impressions subsequently influence both investors' assessments of financial disclosure credibility and the prices they are willing to pay for a company's stock. The findings support the theoretical framework on financial disclosure credibility by (1) showing that management credibility is an important tool that investors use to assess disclosure credibility and (2) suggesting that management credibility is a multidimensional latent construct for which CSR performance can be one of several relevant indicators.

Keywords:

• corporate social responsibility
• financial disclosure credibility
• halo effects
• management credibility

Acknowledgements

We thank James Hunton, Lisa Koonce, Shana Proell, Andrew Spicer, Jennifer Winchel, Valentina Zamora, two anonymous reviewers, and workshop participants at the University of South Carolina for their valuable comments. We also thank Victoria Glackin, Scott Jackson, Linda Quick, and Tammie Rech for their valuable assistance. Data availability: Please contact the first author.

Notes

1. Though investors do not detect earnings management in many situations, they appear to see through earnings management in the banking and insurance industries (Healy and Wahlen 1999).

2. According to Mercer's (2004) framework, management credibility is an enduring trait that refers to investors' perceptions of management's competence and trustworthiness. Essentially, this construct refers to information source credibility.

3. Mercer (2004) defines financial disclosure credibility as investors' perceptions of the believability of a particular financial disclosure, such as an earnings report. Therefore, this construct captures investors' subjective assessments of the decision-usefulness of various earnings reports. In this study, we operationalise the construct of financial disclosure credibility with the measurement of PEC.

4. In defining earnings quality, Dechow et al. (2010) argue that reported accrual-based earnings are a function of the ‘fundamental’ earnings process as well as the accounting system that imperfectly measures it. Consistent with their view and that of Hodge (2003), we refer to higher-quality earnings as those that more faithfully represent the features of the firm's fundamental earnings process.

5. In the economic literature, such activities are called screening.

6. An alternative perspective is that managers undertake CSR activities opportunistically to advance their careers or other personal agendas (Petrovits 2006).

7. We acknowledge that perceptions regarding management credibility derived from other sources, such as investor briefings, may cause investors to ignore information from CSR performance. Similarly, earnings news may be inherently credible, such as the case with negative news, thus causing investors to ignore CSR information.

8. Four concepts expressed in this study need to be differentiated: (1) the actual CSR performance achieved by the company, (2) the presence/absence of CSR disclosures, (3) the information content of CSR disclosure, and (4) investor perceptions of CSR performance based on #3. This study controls for #1 by having an NGO review the CSR report, as is the practice in the oil and gas industry, thus providing an independent view of the firm's CSR performance. We also control for #2 in that CSR disclosures are present for all participants and this concept of CSR disclosures is not varied. We do vary the information content of the CSR disclosure (#3), which is referred as the CSR performance proxy in the early part of the paper and we measure perceptions (#4).

9. 1.0 metric tonne = 1000 kilograms = 2205 US pounds; 1.0 Gigajoule = 278 kilowatt hour.

10. Industry guidelines recommend nine social responsibility performance measures as key report indicators: human right, bribery and corruption, political contributions, non-discrimination and equal opportunity, training and development, non-retaliation and grievance system, labour practices, community relationships, and security.

11. Participants were selected in four groups following four manipulation conditions: the combination of high CSR performance and frequent turnover, the combination of high CSR performance and infrequent turnover, the combination of low CSR performance and frequent turnover, and the combination of low CSR performance and infrequent turnover. Each group experienced only one of four manipulation combinations.

12. To evaluate whether participants are successfully randomised across experimental conditions, we examine whether the source of the data, i.e. in-class or online, and the demographic variables (i.e. age, years of experience, ownership of investment, and gender) differ across experimental conditions. All the differences are insignificant.

13. The results of the confirmatory factor analysis are all satisfactory in terms of construct dimensionality, convergence validity, discriminant validity, and construct reliability. Therefore, we conclude that the measured constructs as shown in Table 1 have sufficient reliability and validity to proceed with hypothesis testing.

14. Inferences from SEM will be the same as those from SEM using Item 1–12 to represent overall PCSRP.

15. In Table 6, a null model is included in which both PCSRP and REP are indistinguishably parts of a larger construct. Both the orthogonal model and the oblique model reject the null model.

16. The mediation procedures in Baron and Kenny (1986) are for a two-path mediated effect. Regressions 1–3 collectively predict a two-path mediation effect on perceptions of earnings credibility, while Regressions 4 and 5 predict a two-path mediation effect on price. We also test Hypotheses 1–4 with a three-path mediated effect using procedures recommended in Taylor et al. (2008). Our findings are consistent in the three-path mediation model as well.

17. Inferences are identical when non-standardised data are used.

18. When we remove management turnover from Equation (3), we find that after controlling the effects of REP, the path coefficient between PEC and PCSRP remains insignificant.

19. The path coefficient between price and PCSRP is positive and significant (path coefficient = 0.21, p-value = .03) after controlling for the effects of both perceptions of earnings credibility (path coefficient = 3.69, p-value <.001) and management turnover (path coefficient = −0.22, p-value = .02).