Determinants and Economic Consequences of Non-financial Disclosure Quality

Abstract

This paper examines the determinants and economic consequences of non-financial disclosure quality, which is measured according to the ratings of corporate social responsibility (CSR) disclosure provided by the Ministry of Economic Affairs in the Netherlands. We find that firms with better CSR performance, greater external financing needs, and stronger corporate governance tend to provide higher quality CSR disclosures. In return, these firms gain greater analyst coverage, higher levels of institutional ownership, greater stock liquidity, higher valuations in SEOs, and lower yields to maturity in bond issuances. These benefits apply largely to firms with strong CSR performance. Collectively, our findings suggest that higher quality CSR disclosures deliver economic benefits.

Acknowledgements

We thank an anonymous referee, Laurence van Lent (editor), Oliver Zhen Li, and Arthur Reichardt for their helpful comments. All errors are the authors’ own.

Supplemental Data and Research Materials

Supplemental data for this article can be accessed on the Taylor & Francis website, 10.1080/09638180.2015.1013049.

ORCID

Fang Gao http://orcid.org/0000-0001-9880-1831

Yi Dong http://orcid.org/0000-0002-1012-7801

Chenkai Ni http://orcid.org/0000-0001-5572-8988

Renhui Fu http://orcid.org/0000-0003-3438-9483

Notes

¹An extensive stream of literature documented empirically and theoretically that investors are bounded by their limited attention which constrains their capability to process and assimilate information in public disclosures (Bloomfield 2002; Grossman & Stiglitz 1980; Hirshleifer & Teoh 2003; Indjejikian 1991; Merton 1987; Peng & Xiong, 2006). The limited attention argument generally applies to public disclosures that are less attention attracting. The variance of a firm's CSR disclosure quality is much less salient than the initiation/issuance of CSR reporting as documented in Dhaliwal et al. (2011, 2012, 2014) and hence might be subject to the limited attention constraints.

²The program has other participants, including private companies with operations in the Netherlands with an accountable turnover and/or number of employees, governmental organizations, companies that in the past voluntarily joined the program, and universities (Source: http://www.transparantiebenchmark.nl/en). We restrict our sample to public companies in the Netherlands for two reasons: first, both accounting information and market information of public companies are more available. Second, public companies are required to participate in the program, mitigating any potential self-selection bias.

³In the online supplement, OSM Appendix 1 and OSM Appendix 2 provide the full list of criteria and their descriptions for the Content-oriented Framework of Standards and the Quality-oriented Framework of Standards, respectively; OSM Appendix 3 provides two snapshots of the correspondence between rating scores and disclosure practices: for one criterion in the Content-oriented Framework of Standards, Environmental aspects of business practices, and for one criterion in the Quality-oriented Framework of Standards, Relevance. The contents originate from the website of the Transparency Benchmark program (http://www.transparantiebenchmark.nl/en).

⁴We thank the Dow Jones Sustainability Indices and RobecoSAM for providing us with data on the historical constituents of the DJSI.

⁵Our sample construction procedure can be summarized into the following steps. First, we obtain the CSR disclosure score for public firms in the Netherlands from its Ministry of Economic Affairs, Agriculture, and Innovation. The initial sample has 735 firm-year observations. Second, we match the data with Compustat Global for firms' fundamental information. The data are further supplemented with information from FactSet, leaving us 568 observations. Third, we require non-missing values for the following list of variables which will be utilized throughout different analyses in this study (SCORE, PERFORM, SIZE, MTB, LEVERAGE, ROA, and STDROA, with definitions outlined in the Appendix). The final main sample, thus, consists of 491 firm-year observations. Sample size will differ in later analyses when additional information is required, for example, accrual quality, analyst coverage, institutional ownership, liquidity, and SEO and debt issuances.

⁶Using the natural log of one plus the CSR disclosure score as the dependent variable leaves the inference unchanged.

⁷Our inference remains unchanged when we measure a firm's CSR performance using its annual healthcare expense. Such a variable directly measures a firm's protection of its employees, representing one important dimension of CSR performance. The results are omitted for the sake of brevity but are available upon request.

⁸One caveat worth mentioning is that we assume managers believe that high-quality disclosures will reduce financing costs. We find support for this assumption in later sections when we examine the economic outcome of CSR disclosure quality. Although the sequence is reversed, we choose to tabulate the results here because our analyses of firms' capital needs belong in the section dedicated to determinants of CSR disclosure quality.

⁹We thank Prof. Isil Erel for making the corporate governance data available on her website (http://faculty.msb.edu/aggarwal/Gov.xls). The 41 attributes from RiskMetrics used in the construction of governance measure cover 4 broad subcategories: (1) Board (24 attributes), (2) Audit (3 attributes), (3) Anti-takeover provisions (6 attributes), and (4) Compensation and ownership (8 attributes). Board attributes capture the aspects of the board of directors such as board independence, composition of committees, size, transparency, and how the board conducts its work. Audit includes questions on the independence of the audit committee and the role of auditors. Anti-takeover provisions are drawn from the firm's charter and by-laws and refer to dual-class structure, role of shareholders, poison pills, and blank cheque preferred. Compensation and ownership deals with executive and director compensation on issues related to options, stock ownership and loans, and how compensation is set and monitored.

¹⁰In untabulated results, we find that effect of governance on CSR disclosure quality is weakened for firms with low CSR performance. The evidence is similar when we use the lagged institutional ownership as the proxy of corporate governance quality.

¹¹Note that our sample's standard deviation is a constant and that this step has no effect on the statistical significance of coefficient estimates.

¹²Alternatively, we use the absolute value of the discretionary accrual, multiplied by minus one, as our accrual quality measure. Results are qualitatively similar.

¹³One concern related to causality is that CSR disclosure quality and the variables capturing economic consequence including analyst coverage, institutional ownership, and stock liquidity are measured contemporaneously, rendering it possible that any association found between the two can be due to reverse causality. The following two points address this concern: (1) A firm's CSR disclosure quality in the current year is based on past CSR reports. For example, the transparency benchmark requires firms to assess their CSR disclosure, starting on 1 June in each year, at which point, the firm's existing CSR reporting is that of last year. The lagged nature of CSR disclosure quality captured in the current year's rating justifies our selection of contemporaneous variables; and (2) we conduct change analyses in Section 4.4, finding a positive association between the change in CSR disclosure quality and the change in a firm's economic consequence including analyst coverage, institutional ownership, and stock liquidity.

¹⁴Our results are insensitive to excluding PERFORM and its interaction with SCORE.

¹⁵In unreported results, we find that there are 12 main types of institutional owners. Among them, investment advisors account for the largest average ownership stake, with a mean value of 21.03% and a median value of 20.2%. Ranked after the investment advisors are mutual funds, which hold, on average, 5.1% of the sample firms' shares. Pension funds have an average ownership of 1.67%, while insurance companies have an average ownership of 1.16%. Of the remaining institutional owners, none hold larger than 1% of the sample firms' shares on average.

¹⁶The standard error becomes 0.000 due to rounding at the third digit after the decimal. It is 0.00043 if we require rounding at the fifth digit after the decimal.

¹⁷We check whether our results are robust towards the consideration that both P/E ratio and CSR disclosure quality are correlated with growth. We include the growth dummy variable, coded one for observations with higher-than-median MTB and zero otherwise, and then interact the growth dummy variable with SCORE. Untabulated results show that the coefficient of CSR disclosure quality is significantly positive for firms with high and low growth, suggesting that our results are not affected by the inclusion of growth in the model.

¹⁸Our investigation here hinges on how changes in CSR disclosure quality are associated with changes in institutional ownership, analyst coverage, and stock liquidity. As equity and debt issuances occur no more than once for most of our sample firms in the period, these two issues are not covered in the change analyses.

Additional information

Funding

We acknowledge financial support from the China National Natural Science Funds [Project No. 71402026, 71172050, and 71372072], the China National Social Science Fund [Project No. 12AZD048], and the Project sponsored by SRF for ROCS, CEM.