Are CSR Disclosures Value Relevant? Cross-Country Evidence

Abstract

Using proprietary data that rate corporate social responsibility (CSR) disclosures of firms in 21 countries, this study examines how the strength of nation-level institutions affects the extent of CSR disclosures. We then examine the valuation implications of CSR disclosures and consider how the relation between CSR disclosures and firm value varies across countries. In contrast to prior studies, we separate CSR disclosures into an expected and unexpected portion where the unexpected portion is a proxy for the incremental information contained in CSR disclosures. We observe a positive relation between unexpected CSR disclosure and firm value measured by Tobin's Q. We also find that, while countries with strong nation-level institutions promote more CSR disclosures, the valuation of a unit increase in unexpected CSR disclosures is higher when nation-level institutions are weak.

Acknowledgements

We thank Florin Vasvari (editor) and an anonymous reviewer for their useful suggestions. We appreciate the helpful comments from Peter Clarkson, Mark Cohen, Tim Devinney, Elaine Henry, and seminar participants at Bristol University, Monash University, University of Bath, University of Bologna, University of Burgos, University of Grenoble, University of Queensland, Vanderbilt University, 2012 Annual Conference of the Accounting and Finance Association of Australia and New Zealand, and 2014 AAA International Accounting Section Midyear Meeting for their helpful comments. We thank KPMG for providing us with the CSR data used in this study.

Notes

1 We later present evidence consistent with this conjecture. For example, we show that expected levels of CSR disclosures are not significantly associated with firm value.

2 We provide a more comprehensive discussion of the nature of KPMG's disclosure measure in Section 3.1 below.

3 Although our focus is on CSR disclosure, we control for CSR performance in our additional analyses, see Section 5.4.

4 An alternative argument is that the market can be ‘fooled’ by opportunistic disclosures in which case a positive relation between CSR disclosure and firm value would not allow us to discriminate between informative and opportunistic disclosures. However, since actual CSR performance can be observed eventually, we do not expect that a firm can fool investors in the long-run. In other words, markets cannot be informationally inefficient in the long-run. If they were, all firms would engage in false disclosures on a continuous basis.

5 KPMG did not publicly state their reason(s) for reducing the number of questions in 2011 compared to 2008. However, based on private correspondence with a senior partner in sustainability at KPMG, the survey questions are reviewed each round and in 2011 the number of questions were reduced. According to the partner, KPMG considers the quality and relevance of the final report, as well as the cost of conducting the survey, when determining the questions to include.

6 Due to a lack of financial and market data on unlisted firms, it is not possible to include unlisted firms in our sample. We compared the means of CSR disclosures by listed firms in our final sample (mean = 24.217) and unlisted firms in the KPMG database (mean = 16.910) using a t-test and find that listed firms in our final sample disclose significantly (p < 0.000) higher levels of CSR information. However, care should be taken in interpreting this result as we could not control for factors that influence CSR disclosure policies, for example, size.

7 An additional advantage of estimating unexpected CSR disclosures is that it reduces the likelihood that our CSR measure is merely a proxy for the firm's overall disclosure policy since the overall disclosure policy will be reflected in the expected, rather than unexpected, part of CSR disclosures.

8 These results, discussed in Section 5.4, indicate that our conclusions are unchanged when CSR performance is included.

9 Our findings and conclusions remain unchanged when we use a one-year ahead measure of Tobin's Q.

10 Scaled accruals (Accruals) are computed using balance sheet and income statement information as (ΔDCA–ΔDCL–ΔDCash + ΔDSTD –Dep + ΔDTP)/lag(TA), where DCA is the change in total current assets; DCL is the change in total current liabilities, DCash is the change in cash, DSTD is the change in the current portion of long-term debt included in total current liabilities, Dep is depreciation and amortization expense; DTP is the change in income taxes payable, and lag(TA) is total assets at the end of the previous year.

11 Although Bloomberg, like KLD, uses corporate reports as an input to their ratings, they augment this with other formal and informal sources of information including evaluations by the United Nations and media reports.

12 In Table 8, unlike our levels tests, we find a significant positive coefficient for ΔNormCSR (p < 0.05) and a significant negative coefficient ΔNormCSR*InstStrength (p < 0.01), indicating that the changes in the normal portion of CSR disclosure also affect changes in firm value and that this relation is affected by institutional strength. Similar to the way it is more difficult to explain stock returns than stock prices, we conjecture that our models are less precise in predicting a change in CSR disclosure than in predicting the level of CSR disclosure. Thus, ΔNormCSR may be capturing some of the unexpected change in CSRDisc which would explain the similar results for ΔAbCSR and ΔNormCSR.