Analyst Forecast Accuracy And Nonfinancial Disclosure: Carbon Emission And Corporate Social Responsibility Disclosures In The US
Lorenzo Dal Maso
Erasmus University Rotterdam (EUR) – Erasmus School of Economics (ESE)
The University of Edinburgh Business School
June 13, 2016
We examine the association between both CO2 emission disclosure and corporate social reporting and analyst forecast accuracy for a sample of large US firms and find, following the implementation of controls for endogeneity, a significant reduction in error, bias and forecast dispersion for those firms that disclose CO2 emissions. We confirm no significant association between corporate social responsibility reporting and forecast error, which is consistent with prior evidence for the US but not with other economies (Dhaliwal, Radhakrishnan, Tsang, and Yang 2012).
Analyst Forecast Accuracy And Nonfinancial Disclosure: Carbon Emission And Corporate Social Responsibility Disclosures In The US – Introduction
We investigate the impact of disclosure of both Corporate Social Reporting (hereinafter CSR) and CO2 Emissions Disclosure (hereinafter CED) on the accuracy, bias and dispersion of analysts’ earnings forecasts. Prior research has suggested that CSR has a significant negative association with forecast error in many countries but a positive albeit insignificant association in the United States (Dhaliwal, Radhakrishnan, Tsang, and Yang 2012). We also note that sell-side analysts exhibit considerably more interest in CED than other elements of nonfinancial reporting typical of the contents of a CSR document (Eccles, Serafeim, and Krzus 2011). Our objectives are therefore to re-examine the contention that the US is a special case with regards to the importance of nonfinancial disclosures and whether CSR is a relatively blunt indicator of nonfinancial disclosures when contrasted with CED.
We contrast CO2 emissions disclosures with the general aspects of corporate social reporting as we view environmental measures as more directly related to costs than the general elements of CSR, which may include important aspects of social or ethical behaviour but may be less relevant to earnings expectations. Indeed Eccles et al. (2011) provide evidence which suggests that broker-dealers, the source of earnings forecasts, are about ten times more likely to examine CED disclosures in Bloomberg than other elements of CSR for an international sample including the US. Whilst they also show that for their full sample of Bloomberg users CED is influential outside the US whereas governance measures tend to be more influential in the US, the magnitude of the dominance of CED for broker-dealers is too large to be explained by international differences. We therefore have evidence which suggests that the analysts producing earnings forecasts are more interested in CED than CSR and hypothesize that CED may be more important than CSR in explaining forecast errors. Despite this apparent importance of CED we view such disclosure as an indicator of good environmental disclosure in general in the same way that issuing a corporate social report can be seen as an indicator of good social reporting in general.
Prior research has demonstrated a statistically significant association between CSR and reduced forecast error for a large international sample largely comprising firms in developed economies (Dhaliwal et al. 2012). It is important to note that the US was an exception and indeed, of the 31 countries in their sample, only the US had a positive coefficient on the relationship between CSR and forecast error. The authors argued that in the US social costs were less important than in most other countries in their sample and that this may explain why no significant association was found between corporate social reporting and earnings forecast accuracy for US firms. Whilst it may readily be proposed that there is greater scepticism regarding social responsibility in general in the US than in other developed economies we view the argument that social costs have a less profound impact on firms in the US than in other developed economies as more contentious. However, as acknowledged by Dhaliwal et al. (2012), attitudes to climate change, and hence the impact on the financial performance of climate change, are particularly sceptical in the US. We are therefore interested to re-examine the role of CSR in general, and CED in particular, in the US given the surprising (for us) implication that social costs are not significant in the US.
In our empirical analysis we investigate the effect of CED and CSR on mean forecast error, as in previous papers, and both forecast bias and forecast dispersion. We include dispersion, as it is possible that mean forecast error might be driven by bias, whereas low dispersion may accompany inaccurate consensus. If CED assists analysts in their forecasting activities it should both improve accuracy and reduce dispersion. We base our conclusions on a propensity score matching approach (hereinafter PSM). The association, or lack of it in the US, between CSR and forecast error is robustly demonstrated in the previous literature but it is clearly conceivable that any causal relationship may be exaggerated or reduced by endogeneity.
See full PDF below.