Factors Influencing Corporate Environmental Disclosures†Facteurs influant sur la communication d'information environnementale par les sociétés
Abstract
enWe investigate the effectiveness of the Carbon Disclosure Project (CDP), a not-for-profit organization that facilitates environmental disclosures of firms with institutional investors, thereby serving as a corporate governance mechanism for shareholders to influence the firm's environmental disclosures. We examine firm characteristics associated with firms' decisions to disclose carbon-related information via the CDP for a sample of 319 Canadian firms over a four-year period. In particular, we examine how firms' decisions to disclose via CDP are associated with shareholder activism, litigation risk, and the opportunity for low-cost positive publicity once requested by the firms' “signatory” investors. Our results also show that management's decision to release climate change data is associated with domestic, but not foreign, signatory investors. We also find that disclosing firms tend to be those from lower polluting industries with less exposure to litigation risk. This suggests that this new form of coordinated shareholder activism may not be successful at altering the behavior of firms that are heavier polluters.
Facteurs influant sur la communication d'information environnementale par les sociétés
frRésumé
Les auteurs se penchent sur l'efficacité du Carbon Disclosure Project (CDP), organisme sans but lucratif qui s'emploie à promouvoir la communication d'information environnementale par les entreprises ayant des investisseurs institutionnels, un mécanisme de gouvernance grâce auquel les actionnaires peuvent exercer une influence sur l'information environnementale publiée par les sociétés. Ils étudient les caractéristiques de l'entreprise associées à la décision de communiquer l'information relative aux émissions de carbone par l'intermédiaire du CDP, pour un échantillon de 319 sociétés canadiennes sur une période de quatre ans. Ils se demandent en particulier en quoi la décision des sociétés de communiquer des données par l'intermédiaire du CDP est liée à l'activisme des actionnaires, aux risques de litiges et à la perspective d'une publicité positive à faible coût une fois l'information demandée par les investisseurs « signataires » des sociétés. Les résultats de l'étude indiquent que la décision de la direction de publier des données sur les changements climatiques est associée aux investisseurs signataires nationaux, mais non aux investisseurs signataires étrangers. Les auteurs constatent également que les sociétés qui publient cette information sont généralement celles qui appartiennent à des secteurs d'activité moins polluants, moins exposés aux risques de litiges. Ces observations portent à croire que cette nouvelle forme d'activisme conjugué des actionnaires pourrait ne pas parvenir à modifier le comportement des sociétés qui sont d'importants pollueurs.
As focus shifts from debating the existence of global warming to formulating proposals to mitigate its damages, governments face increased pressure to regulate greenhouse gas (GHG) emissions. Investors are concerned over how businesses will respond to new emissions constraints.1 This has increased the importance of corporate environmental governance, which the Environment Agency defines as “the full range of best practice approaches to management by companies of their environmental impacts, risks, performance and opportunities” (2004: 15). To assess how stricter regulations may impact a company, investors need information about the firm's environmental practices and current GHG emissions. The Carbon Disclosure Project (CDP) is a not-for-profit organization that assists investors in gathering this information. This study examines the factors associated with management's decision to disclose environmental information through the CDP.
Previous studies have found the current rules governing financial reporting to be deficient in regulating the release of environmental information. For example, although the Canadian Securities Administrators' National Instrument 51-102 requires environmental disclosures for many firms, Li, Richardson, and Thornton (1997) report that the information released is inconsistent and not unbiased. Other studies find that mandating environmental disclosures increases the overall level of disclosures, but that firms are selective in the data provided due to differences in the interpretation of standards (Frost, 2007; Criado-Jimenez, Fernandez-Chulian, Husillos-Carques, and Larrinaga-Gonzalez, 2008). “Good” environmental performance is more likely to be reported, whereas many “bad” environmental performers tend to avoid the requirement to disclose. These shortcomings have motivated institutional investors to seek new ways to obtain the corporate environmental information they require.
The Carbon Disclosure Project (CDP), based in the United Kingdom, is one new avenue investors may pursue toward this end. Formed in 2000 as a United Nations initiative, the CDP gathers and disseminates firm level climate change information in an effort to create a unified response against global warming.2 It enlists the support of institutional investors,3 whom it refers to as “signatory investors,” to advocate for its environmental awareness agenda. In February each year, the CDP sends a questionnaire to the largest global companies requesting voluntary climate change information on behalf of the signatory investors. The CDP makes the responses accessible to the signatory investors right away and more generally to the public a bit later.
The CDP has been reasonably successful at encouraging firms to disclose climate change information as evident by the 74 percent response rate to its questionnaires requesting environmental disclosures. This compares favorably to the 10 percent response rate for similar requests from shareholders (Rojas, M'zali, Turcotte, and Merrigan, 2009). As of January 2011, over 3000 organizations in 60 countries were disclosing through the CDP, which represented 551 signatory investors with US$71 trillion in assets under their management and included companies such as Dell, PepsiCo and Wal-Mart.4
Despite the CDP's success in eliciting voluntary corporate disclosures, little academic research has been directed to these reports. This paper examines factors associated with firms' decisions to release climate change information through this mechanism. The findings of prior studies are somewhat ambiguous. Kolk, Levy, and Pinkse (2008) attribute the CDP's success in obtaining environmental information to the strength of institutional investors. However, Stanny and Ely (2008) find no significant correlation between disclosure rates and institutional ownership.
Motivation and Expected Contribution
This paper examines whether the firm's decision to disclose information via the CDP is associated with shareholder activism, litigation risk, and the opportunity for low-cost positive publicity. We extend prior research on the CDP in four respects:
- We focus on Canadian firms, a sample not previously studied. As the CDP encompasses a global audience, it is important to observe the CDP's impact in different parts of the world. The need for independent testing in different countries is highlighted by Cormier and Magnan's (2007) finding that reaction to environmental disclosures varies across nations.
- Our study includes four years of CDP data, compared to Stanny and Ely (2008) who focus on a single year.
- We investigate the association between CDP disclosure and factors such as litigation risk that have not been examined previously.
- Unlike previous studies, we use signatory investors as a proxy for shareholder activism, after examining the proportion of each firm's ownership these signatory institutions control. Our use of CDP data to identify foreign and domestic signatory institutional investor ownership offers insight into how coordinated institutional investor activism fits into a global corporate environmental governance setting.
Focusing on signatory investors may help to explain the CDP's success in generating more voluntary environmental disclosures than shareholder activism performed by independent institutional investors.5 The contribution of this research is to assess the CDP as a new mechanism that may enhance global corporate environmental governance by encouraging firm disclosures on carbon emissions through an alternate medium to the annual report. We hope the findings will also enrich our understanding of shareholder activism.
Our study covers the period from 2006 through 2009 and is restricted to Canadian firms, because only Canadian CDP data prior to 2010 include a breakdown of each firm's percentage of signatory investor ownership. This information is not provided for the CDP's respondents from other countries. While the CDP gathers the information, it relies on domestic partners6 to create their annual reports. We leave it to future studies to construct this information for other settings to test if our findings hold in other countries.
We begin with a literature review and discuss why the CDP disclosures present a particularly interesting database for examining environmental disclosures. Next, we develop our hypotheses and outline the data collection and method of analysis. We conclude with some implications of our findings and ideas for future research.
Background
Information Disclosure as Quasi-Regulation
Konar and Cohen (1997) find that firms that suffer the largest price declines in response to disclosures labeling them as the “worst” polluters subsequently improve their environmental performance more than their industry rivals.
Two prior papers examine the environmental disclosures made in response to the CDP. Kolk et al. (2008) find that the CDP is successful in encouraging companies to divulge environmental information, but the disclosures are too complex and not comparable across firms. Stanny and Ely (2008) find the CDP disclosure rates to be positively associated with firm size, previous disclosures and foreign sales. However, unlike Kolk et al. (2008), they observe no significant relationship between disclosure and institutional ownership. This disparity may be due to differences in their samples and/or the methodology employed. Kolk et al. (2008) use the Global FT 500, while Stanny and Ely (2008) examine the U.S. S&P 500 provided by the CDP. In 2007 the Global FT 500 response rate was 77 percent, while the equivalent response rate for the U.S. S&P 500 was 56 percent. Kolk et al. (2008) base their conclusions on a comparison of regional response rates to the number of signatory investors, whereas Stanny and Ely (2008) use a logistic regression to test factors associated with the firm's decision to disclose.
We extend Stanny and Ely (2008) by introducing signatory investors as a proxy for shareholder activism. Gillan and Starks (2000) observe that large shareholders may try to participate in shaping the firm's strategic direction and monitoring its activities. Stanny and Ely (2008) include institutional investors as a proxy for increased scrutiny. However, not all institutional investors attempt to influence management's decisions or monitor the firm (Del Guercio and Hawkins, 1999). Signatory ownership is a finer proxy for shareholder activism in corporate governance, because the CDP offers institutional investors the choice to identify themselves as active investors that seek to monitor the firm's behavior. This should provide a better proxy for the institutional investor's role in corporate environmental governance.
Determinants of Voluntary Environmental Disclosures
Previous researchers have examined factors associated with other voluntary environmental disclosures. Adopting a legitimacy theory approach, Patten (1992) concludes that firms increased environmental disclosures after the Alaskan oil spill in order to maintain legitimacy. Barth, McNichols, and Wilson (1997) report that regulations, litigation and negotiation concerns, and capital market requirements are associated with voluntary disclosure of environmental liabilities by firms with substantial superfund involvement. Brammer and Pavelin (2006) find larger firms with dispersed ownership and lower levels of debt more likely to make environmental disclosures. Li et al. (1997) find such disclosures positively associated with both the firm's propensity to pollute and outside knowledge of the firm's environmental liability, and negatively associated with the costs of disclosure.
Uniqueness of the Database
Two features make the CDP a particularly interesting database for studying environmental reporting. First, it presents climate change information separately from other confounding data. Second, it reflects a unique mechanism for shareholder activism.
As Stanny and Ely (2008) note, most studies of voluntary environmental disclosures examine annual reports and sustainability reports, both of which contain other confounding information that may impact investors' decisions (Ingram and Frazier, 1980; Wiseman, 1982; Hughes, Anderson, and Golden, 2001; Hedberg and von Malmborg, 2003; Al-Tuwaijri, Christensen, and Hughes, 2004; Sahay, 2004; Brammer and Pavelin, 2006; Clarkson. Li, Richardson, and Vasvari, 2008; Hossain and Reaz, 2007; Kolk, 2008). In contrast, the CDP focuses only on climate change information that is released independently from the firm's annual report or other sustainability reports.
The second and perhaps most important feature of the database is that it includes a sound proxy for shareholder activism. While institutional investors are often considered a corporate governance mechanism, Del Guercio and Hawkins (1999) find significant heterogeneity in their objectives, tactics, and success in terms of shareholder activism. The CDP disclosures are distinctive because they are made in response to investor requests. By becoming a signatory investor, an institutional investor identifies itself as an active investor. CDP data can therefore be used to distinguish active and passive institutional investors in our sample, as the CDP discloses the percentage of each Canadian firm's ownership controlled by its signatory investors. The use of signatory investors, rather than all institutional investors, provides a better proxy for studying shareholder activism. The shareholder activism displayed by the signatory investors, as well as management's response, are public in nature. Management cannot avoid the issue or conceal its decision pertaining to the investors' request, because the reluctance to disclose the climate change information is made public.
Hypotheses Development
Shareholder Activism
The use of signatory investors to pressure firms to disclose environmental information represents a relatively new stage in the evolution of shareholder activism. Gillan and Starks (2007) discuss changes in shareholder activism over time, noting various forms it can take, from selling one's shares to a hostile takeover. However, their analysis deals only with changes in the type of investors using shareholder activism and the issues targeted. They do not consider any evolution in the form of shareholder activism itself.
Prior research focuses on one form of shareholder activism, namely shareholder proposals (Ferri and Sandino, 2009; Thomas and Cotter, 2007; Gillan and Starks, 2000; Del Guercio and Hawkins, 1999). These studies usually look at institutional investors' use of Securities Exchange Act (SEC) Rule 14a-8, which allows shareholders to forward to management a proposal for inclusion in the company's proxy material, to be presented and voted upon at the next shareholder meeting.7 Success of the shareholder activism may be measured by the votes for the issue. However, Rule 14a-8 limits an investor to only one proposal per firm each proxy season, and allows management to remove a proposal if the same issue has been included in the proxy in previous years and failed to meet the required level of votes.8
While some shareholder proposals have proved an effective mechanism for corporate governance (Ferri and Sandino, 2009; Del Guercio and Hawkins, 2008; Thomas and Cotter, 2007), proposals on social reform items have been less successful. Thomas and Cotter (2007) report that environmental issues that made it to a firm's proxy for voting between 2002 and 2004 received an average 10.79 percent of the votes. Not only do these fail to reach a majority vote, but any proposal concerning environmental issues that makes it to the proxy runs the risk of receiving too few votes to be accepted in future years. Thus, Rule 14a-8 is very ineffective for soliciting climate change disclosures. When limited to only one proposal per firm, an active institutional investor will likely use Rule 14a-8 for an issue that has a higher chance of passing.
- Hypothesis 1. The higher the percentage of signatory investor ownership, the more likely a firm will respond to the CDP and disclose environmental information.
- Hypothesis 2. The higher the percentage of domestic signatory investor ownership, the more likely a firm will respond to the CDP and disclose environmental information.
We use two logistic models to test the relationship between domestic signatory investors and management's decision to disclose climate change information. The first model includes both TOTSIG and a variable that defines the relative domestic signatory firm ownership by dividing the firm's domestic signatory ownership by its total signatory investor ownership (RelativeDOMSIG). A second model uses the actual percentage of domestic signatory investor ownership (DOMSIG), as well as the foreign signatory investor ownership (FORSIG = TOTSIG − DOMSIG). We predict that RelativeDOMSIG, DOMSIG and FORSIG will all be positively associated with the decision to disclose.
Litigation Risk
For public companies, executives and directors (among others) need to consider the potential for lawsuits in Canada whereby they can be at risk for providing misleading public disclosures or failing to make timely disclosures. (CICA, 2008: 10)
- Hypothesis 3. The firm's exposure to litigation risk will influence the firm's decision to disclose environmental information through the CDP.
We use an indicator variable (LIT) to proxy for litigation risk, which is not directly measurable. We assign a code of 1 to firms listed in environmental litigation in the five-year period prior to the CDP request and 0 to other firms.
Low-Cost Publicity
Clarkson, Li, and Richardson (2004) identify several benefits associated with being considered a “green” firm. Firms in industries such as retail or information technology create more emissions through processing, packaging, and transportation than by their own business practices. Because a firm's carbon footprint goes beyond its own direct emissions, some companies, such as Wal-Mart, PepsiCo, and IBM, use supply-chain management to choose suppliers with lower emissions levels.13 Walton, Handfield, and Melnyk (1998) report that pulp and paper companies with lower polluting levels fare better than industry rivals in securing long-term contracts with “green” customers. Kristrom and Lundgren (2003) discuss the concept of “green goodwill” which implies that consumers who prefer to buy a “greener” product are willing to pay a premium for it. Blend and Ravenswaay (1999) find a significant demand for “greener” eco-labeled apples.
There is a wide disparity in pollution levels among firms responding to the CDP. For example, in 2008 Merrill Lynch & Co. had a carbon intensity14 of 6, while the utility firm Ameren Corporation had a carbon intensity of 9,036.15 Obviously, the low carbon intensive firm will incur less cost to collect the emissions data. According to the Greenhouse Gas Protocol, the leading body in setting greenhouse gas reporting standards, the only sources of direct emissions for office-based organizations come from owning vehicles, stationary combustion devices, refrigerators, and air conditioning equipment.16 It is much more expensive for high carbon intensive firms to calculate direct emissions from their physical or chemical processes and identify emissions caused by the wearing down of equipment. Furthermore, a company such as Merrill Lynch & Co. need not fear that the CDP disclosures will lead the market to reevaluate its susceptibility to the risks associated with global warming. Firms in low carbon intensive industries (such as financial services; hospitality, leisure, and business services; retail and consumer; and technology, media, and telecommunications) are unlikely to be held responsible for global warming and will have little difficulty transitioning to a carbon constrained economy. For these firms, the CDP may represent an inexpensive means to get positive publicity by enhancing its “green” reputation.
- Hypothesis 4. Firms that are not expected to produce scope 1 direct GHG emissions will be more likely to respond to the CDP by disclosing environmental information.
We use an indicator variable (LOWGHG) to proxy for low emission firms, coding firms identified as high GHG producers as a 0 and all others as 1. To construct the variable, we use the U.S. Environmental Protection Agency's 2009 Mandatory Reporting of Greenhouse Gases: Proposed Rule to identify firms expected to be affected by future GHG regulations. We also code as 0 firms the CDP lists as being in carbon intensive industries, such as manufacturing, oil and gas, mining and transportation. Where available, we check information about a company's direct emissions to ensure we have coded the firm properly.
Control Variables
In testing our hypotheses, we control for other firm characteristics and time effects that may also influence the firm's response to the CDP. It has been well documented that larger firms tend to incur higher political costs and receive more public scrutiny and media attention than smaller firms. Prior studies find environmental disclosures positively related to firm size (Clarkson et al., 2008; Stanny and Ely, 2008). We expect larger firms more likely to respond to the CDP, and predict a positive sign on our proxy for firm size (SIZE), the natural log of a firm's total assets averaged over the prior 2 years.17
As environmental regulations tighten, firms that have not adequately addressed the issue of climate change will face additional costs. Less profitable firms may be more reluctant to release climate change data that may suggest a decline in future earnings. More profitable firms may be inclined to respond to the CDP to assure investors that their high quality earnings can withstand a more constrained regulatory environment (Stanny and Ely, 2008). We predict a positive sign on a profitability variable (ROA), computed as net income averaged over the prior two years divided by total assets averaged over the prior two years.
We expect that more highly leveraged firms will have stricter debt covenants to restrain shareholders' actions, and creditors will likely demand more information to monitor management behavior (Leftwich, Watts, and Zimmerman, 1981). Thus, highly leveraged firms may be motivated to make voluntary disclosures in order to reduce contracting costs. Clarkson et al. (2008) find a significant positive correlation between leverage and voluntary environmental disclosures, but Stanny and Ely (2008) observe a negative, but insignificant relationship in regards to the CDP disclosures. We compute the firm's leverage (LEV) by dividing the firm's total debt averaged over the prior two years by the total assets averaged over the prior two years.
Healy and Palepu (2001) suggest managers may make voluntary disclosures to reduce information asymmetry in an attempt to decrease the firm's cost of capital. One source of information asymmetry is the presence of unrecorded intangible assets that may be associated with good future prospects, but whose value is difficult for investors to gauge. Tobin's Q measures the ratio of a firm's market value of assets to its book value of assets. A Tobin Q higher than 1 implies the existence of valuable unrecorded intangible assets associated with growth opportunities (Barth and Kasznik, 1999). Following Clarkson et al (2008) and Stanny and Ely (2008), we include this as a control for information asymmetry. We expect firms with higher Tobin Q scores will more likely respond to the CDP. We compute Tobin Q (TOBINQ) as the total market value of assets averaged over the prior two years divided by total book value of assets averaged over the prior two years.
Finally, because our study covers a four-year period, we include indicator variables to control for time effects in order to control for variations in the data (such as differences in response rates) that are related to the year of the survey. The variables CDP5, CDP6, and CDP7 represent data from 2007, 2008, and 2009 respectively, with 2006 being used as the base year.18
Data Collection and Method of Analysis
Data Collection
The sample used in this study is taken from the CDP Canada 200 data sets for the period 2006 to 2009. This database provides information about firms' responses to its questionnaires, firm characteristics such as market capitalization and industry, and the signatory investors' ownership stake in each firm. Additional information was retrieved from LawSource,19 the U.S. Environmental Protection Agency's 2009 Mandatory Reporting of Greenhouse Gases: Proposed Rule, and DataStream.
Despite being called Canada 200, this data set includes more than 200 companies per year. The firms included in this data set change annually. The CDP simply sends surveys to the largest firms, regardless of whether or not it sent them a survey in the past. This causes a substantial amount of change within the companies approached from one year to the next. The original sample consists of 858 firm-year observations. However, as shown in Table 1, we remove 40 observations due to incomplete financial information, leaving a sample consisting of 818 firm year observations for the logistic regression. The final sample represents 319 unique firms.
| Disclosing firms | Nondisclosing firms | Total | |
|---|---|---|---|
| Initial sample | 366 | 492 | 858 |
| Observations removed due to lack of financial information | 15 | 25 | 40 |
| Final sample for logistic regression | 351 | 467 | 818 |
Note:
- The table displays the original sample size, the reasons for the removal of observations, the number of observations removed, and the final sample sizes for the logistic regression analysis.
Logistic Regression
()
()where:
TOTSIG = the percentage of signatory firm ownership.
RelativeDOMSIG = the domestic signatory firm ownership divided by the total signatory firm ownership.
DOMSIG = the percentage of domestic signatory firm ownership.
FORSIG = the percentage of foreign signatory firm ownership.
LIT = 1 if a firm has been listed in an environmental lawsuit within the previous five years.
LOWGHG = 1 if a firm's business does not produce a significant level of GHG emissions.
SIZE = the natural log of total shareholder's equity averaged over the prior two years.
ROA = the net income averaged over the prior two years divided by the total assets averaged over the prior two years.
LEV = the total debt averaged over the prior two years divided by the total assets averaged over the prior two years.
TOBINQ = the total market value of assets averaged over the prior two years divided by the book value of assets averaged over the prior two years.
CDP5 = 1 for observations from the 2007 sample.
CDP6 = 1 for observations from the 2008 sample.
CDP7 = 1 for observations from the 2009 sample.
Empirical Results
Table 2 displays the correlations among the variables within the logistic regression models. As expected, TOTSIG is highly correlated with DOMSIG and FORSIG, because TOTSIG is the sum of DOMSIG and FORSIG. Because we do not use TOTSIG in the same model as DOMSIG and FORSIG, this is not a concern. RelativeDOMSIG is also highly correlated with both DOMSIG and FORSIG, but not a concern, because they are not used in the same regression models. As there is a significant correlation (0.635) between SIZE and LEV, we run a diagnostic to ensure no multicollinearity exists. All variables have a tolerance greater than 0.5 and the highest variance inflation factor is 2.05. Because all tolerance values are greater than 0.2 and all VIF values are less than 5, multicollinearity should not be an issue. There are also significant correlations between AQ and CDP6 as well as CDP7, providing support for including control variables for the time effects. LIT is significantly negatively correlated to LOWGHG. This is expected because lower emission firms are less likely to be involved in environmental litigation. LIT has a significant positive correlation with SIZE.
| AQ | TOTSIG | RelativeDOMSIG | DOMSIG | FORSIG | LIT | LOWGHG | SIZE | ROA | LEV | TOBINQ | CDP5 | CDP6 | CDP7 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| AQ | 1 | |||||||||||||
| TOTSIG | 0.076* | 1 | ||||||||||||
| RelativeDOMSIG | 0.089* | −0.212** | 1 | |||||||||||
| DOMSIG | 0.143** | 0.693** | 0.432** | 1 | ||||||||||
| FORSIG | −0.042 | 0.673** | −0.727** | −0.068 | 1 | |||||||||
| LIT | 0.215** | 0.032 | 0.061 | 0.065 | −0.022 | 1 | ||||||||
| LOWGHG | −0.073* | 0.063 | 0.114** | 0.135** | −0.051 | −0.236** | 1 | |||||||
| SIZE | 0.374** | 0.166** | 0.212** | 0.268** | −0.044 | 0.165** | 0.183** | 1 | ||||||
| ROA | 0 | −0.042 | 0.137** | 0.011 | −0.069* | 0.038 | −0.079* | −0.075* | 1 | |||||
| LEV | 0.144** | 0.084* | 0.220** | 0.211** | −0.100** | 0 | 0.333** | 0.635** | −0.127** | 1 | ||||
| TOBINQ | −0.033 | −0.04 | −0.034 | −0.053 | −0.001 | −0.05 | −0.035 | −0.128** | 0.107** | 0.044 | 1 | |||
| CDP5 | 0.027 | −0.106** | 0.461** | 0.192** | −0.344** | 0.004 | −0.036 | 0.042 | 0.06 | 0.036 | 0.007 | 1 | ||
| CDP6 | 0.132** | −0.312** | 0.072* | −0.204** | −0.223** | 0.038 | 0.002 | 0.126** | 0.026 | 0.031 | 0.061 | −0.282** | 1 | |
| CDP7 | 0.083* | 0.340** | −0.212** | 0.149** | 0.317** | 0.003 | 0.002 | 0.115** | −0.090** | 0.026 | −0.051 | −0.298** | −0.292** 1 | 1 |
Notes:
- This table lists the correlations between all variables used in both of the logistic regression models. AQ is the dependent variable, it is a dummy variable coded as a 1 for firms that disclosed through the CDP and 0 otherwise. TOTSIG is the percentage of signatory firm ownership. RelativeDOMSIG is the firms' domestic signatory ownership divided by the firms' total signatory ownership. DOMSIG is the percentage of domestic signatory ownerhip. FORSIG is the percentage of foreign signatory firm ownership. LIT is a dummy variable representing litigation risk, it is coded 1 for all firms that were listed as defendants in environmental litigation within five years prior to being approached by the CDP. LOWGHG is a dummy variable coded 1 for firms that produce low levels of greenhouse gases and 0 otherwise. SIZE is the natural log of the total assets averaged over the prior two years. ROA is the firms' net income averaged over the prior two years divided by total assets averaged over the prior two years. LEV is the firms' total debt averaged over the prior two years divided by the total assets averaged over the prior two years. TOBINQ is the firms' market value of equity averaged over the prior two years divided by the firms book value of equity averaged over the prior two years. CDP5, CDP6, and CDP 7 are dummy variables included to capture time effects, they represent 2007, 2008, and 2009 respectively.
- *Correlation is significant at the 0.05 level (two-tailed).
- **Correlation is significant at the 0.01 level (two-tailed).
Table 3 displays the results from the binary logistic regression. Nagelkerke's pseudo R square and the model likelihood ratio test the models' goodness of fit. The first model's likelihood ratio is significant with p < 0.000. The Nagelkerke R square of 0.259 represents an adequate fit. The second model slows a slightly better fit with a likelihood ratio that is significant at p < 0.000 and a pseudo R2 of 0.262.
| Variable | Predicted sign | Model 1 | Model 2 | ||||
|---|---|---|---|---|---|---|---|
| Estimate | Standard error | Wald statistic | Estimate | Standard error | Wald statistic | ||
| Intercept | −2.556*** | 0.839 | 9.287 | −2.764*** | 0.811 | 11.625 | |
| TOTSIG | + | 0.012 | 0.01 | 1.428 | |||
| RelativeDOMSIG | + | −0.085 | 0.286 | 0.089 | |||
| DOMSIG | + | 0.026** | 0.013 | 3.919 | |||
| FORSIG | + | −0.005 | 0.015 | 0.11 | |||
| LIT | +/− | −0.916*** | 0.272 | 11.335 | −0.88*** | 0.273 | 10.414 |
| LOWGHG | + | 0.409** | 0.174 | 5.527 | 0.45** | 0.175 | 6.618 |
| SIZE | + | 0.597*** | 0.077 | 60.382 | 0.594*** | 0.077 | 59.747 |
| ROA | + | 0.208 | 0.903 | 0.053 | 0.074 | 0.894 | 0.007 |
| LEV | − | −0.959* | 0.525 | 3.329 | −1.054** | 0.527 | 4.002 |
| TOBINQ | + | 0.01 | 0.019 | 0.27 | 0.011 | 0.019 | 0.36 |
| CDP5 | +/− | −0.571** | 0.248 | 5.301 | −0.402* | 0.24 | 2.806 |
| CDP6 | +/− | −0.892*** | 0.239 | 13.953 | −0.836*** | 0.238 | 12.347 |
| CDP7 | +/− | −0.573** | 0.224 | 6.525 | −0.574** | 0.225 | 6.525 |
| N | 819 | 819 | |||||
| Pseudo R2 | 0.259 | 0.262 | |||||
| Model L.R. (df, p-value) | 175.170 (11, 0.000) | 177.804 (11, 0.000) | |||||
Notes:
- This table lists the results from two logistic regressions where the dependent variable for both is a dummy variable coded as a 1 for firms that disclosed through the CDP and 0 otherwise. Predicted sign, the coefficient estimate, standard error and Wald statistic are given for the intercept as well as all independent variables. The first model includes TOTSIG which is the percentage of signatory firm ownership. RelativeDOMSIG is the firms' domestic signatory ownership divided by the firms' total signatory ownership. LIT is a dummy variable representing litigation risk, it is coded 1 for all firms that were listed as defendants in environmental litigation within five years prior to being approached by the CDP. LOWGHG is a dummy variable coded 1 for firms that produce low levels of greenhouse gases and 0 otherwise. SIZE is the natural log of the total assets averaged over the prior two years. ROA is the firms' net income averaged over the prior two years divided by total assets averaged over the prior two years. LEV is the firms' total debt averaged over the prior two years divided by the total assets averaged over the prior two years. TOBINQ is the firms' market value of equity averaged over the prior two years divided by the firms' book value of equity averaged over the prior two years. CDP5, CDP6, and CDP 7 are dummy variable included to capture time effects; they represent 2007, 2008, and 2009 respectively. The second model replaces TOTSIG and RelativeDOMSIG with DOMSIG and FORSIG. DOMSIG is the percentage of domestic signatory ownership. FORSIG is the percentage of foreign signatory firm ownership. 819 firm year observations were used for both models. Model 1 had a likelihood ratio of 168.641 with 11 degrees of freedom and was significant at all levels and a pseudo R2 of 0.258. Model 2 had a likelihood ratio of 177.804 with 11 degrees of freedom and was significant at all levels and a pseudo R2 of 0.262.
- ***, **, and * indicate significance at the 1, 5, and 10 percent levels, respectively.
The coefficient for the percentage of total signatory investors has the expected positive sign, but is not significant. This fails to support Hypothesis 1. While the coefficient on the relative level of domestic signatory investors is insignificant, the domestic signatory firm ownership in the second model is positive and significant at p < 0.05. This suggests that a firm responds to shareholder activism by domestic institutional investors more than to foreign institutional investors.
The coefficient for the litigation risk variable is negative and significant at p < 0.01 level in both models. This suggests firms do not believe early disclosures will be beneficial in the defense of environmental litigation. Instead, firms concerned with potential lawsuits appear less likely to disclose environmental information.
The results are consistent with Hypothesis 4 that predicts low emission firms will disclose to the CDP for low-cost good publicity. The estimated coefficient for the variable representing low GHG emitting firms is positive and significant at p < 0.05 in both models. This finding is consistent with Stanny and Ely (2008), and indicates that firms with low compliance costs are more likely to disclose than firms that need to be more concerned with future regulations.
The results from both models show a positive coefficient on SIZE that is significant at p < 0.01. As predicted, larger firms appear more likely to disclose information to the CDP. The coefficient on LEV is negative and significant at p < 0.10 and p < 0.5. This finding is inconsistent with our prediction that highly leveraged firms are more likely to provide voluntary disclosures. For this sample of Canadian companies, it appears firms with higher leverage are more reluctant to make disclosures to the CDP. All of the time effects variables have negative coefficients that are significant under both models, at levels ranging from p < 0.10 to p < 0.00. While the coefficients on the variables ROA and TOBINQ are in the expected direction, they are not significant. This finding is consistent with that of Stanny and Ely (2008).
Discussion and Conclusions
We test four hypotheses to study management's decision to disclose climate change information through the CDP. The first two hypotheses consider the CDP and its signatory investors as a new mechanism for disclosures related to corporate environmental governance. The results suggest that aggregate efforts of domestic institutional investors are positively associated with management's decision to disclose (Hypothesis 2). However, the failure to find a significant correlation between disclosure and participation by all (both foreign and domestic) signatory investors (Hypothesis 1) highlights a limitation of the CDP as a new source of corporate environmental governance information. While the evidence suggests that the success of shareholder activism for social issues can be improved, a globally unified response will likely require support from investors in each domestic market.
Results from the other hypotheses suggest a further qualification regarding the CDP's success. This new form of shareholder activism appears most effective at altering the behavior of firms that are least likely to endure a major negative impact from the tightening of GHG regulations. While we find evidence that environmental litigation risk impacts the decision to release climate change data (Hypothesis 3), the negative coefficient on this variable suggests that firms exposed to lower levels of environmental litigation risk are more likely to disclose this information voluntarily. Similarly, the support for Hypothesis 4 indicates that low polluting firms are more likely to respond to the CDP, perhaps hoping to use this as an outlet to gain low-cost publicity. While these findings do not fare well for the CDP's informativeness, they are based on results employing dichotomous variables. Future studies should examine the impact of litigation risk and the potential for low-cost publicity using finer variables.
When considering all of these results collectively, we start to form a picture of the CDP, its signatory investors and the effectiveness of the corporate environmental governance device they represent. Even though the CDP is an international corporate environmental governance mechanism, the evidence suggests that it may require the participation of local institutional investors in order to be successful. In addition, this mechanism appears less influential in obtaining voluntary disclosures from high polluting firms that may be most adversely affected by stricter environmental regulations. We believe these results enhance our understanding of how institutional investors fit into corporate governance. They may be of interest to parties looking to duplicate the CDP's efforts, to active investors who wish to promote change within an organization, and to future researchers.
In order to examine the influence of institutional shareholders, we restricted our sample to Canadian firms, because only the Canadian reports include information about the percentage of each firm's signatory investors. This limits the generalizability of our findings, as we do not know if these results would hold true for other countries. Thus, further investigation of the CDP's impact on participants from other jurisdictions is needed before one can assess its success as a viable substitute for mandatory reporting.
Another avenue for future research is to study the market response to CDP disclosures. Unfortunately, the CDP data currently available does not specify the dates on which questionnaire answers are received or the dates on which the information is disclosed to signatory investors or made publicly available. Also, it is not clear whether signatory investors get access to the information before it becomes publicly available, or whether it becomes available to signatory investors and the public at large at the same time. At present, these limitations restrict our ability to estimate the value relevance of the CDP disclosures from a market perspective. However, since its inception, the CDP has shown significant improvement in terms of managing the information provided by its respondents. This is evident in a recent statement made by the CDP addressing the demand of its stakeholders for data verification and the setting of a framework for respondents to encourage data verification.20 As more information becomes accessible, future researchers may be able to examine the equity market response to CDP disclosures of environmental information and the determinants of such responses, both for companies who decide to participate and those who refuse to answer the questionnaire.
