Further evidence on the factors and valuation associated with the level of environmental liability disclosures.

Cox, Carol A. ; Douthett, Edward B., Jr.

INTRODUCTION

The increasing emphasis on social responsibility by various corporate stakeholders and constituents makes the reporting of environmental obligations a prominent issue for accounting regulatory and professional bodies, as well as accounting academic researchers. We aim to extend our knowledge about the determinants and market value of environmental liability disclosures by focusing on the reporting requirements outlined by the Financial Accounting Standards Board (FASB) and the U. S. Securities and Exchange Commission (SEC), which collectively form the generally accepted accounting principles for environmental liabilities (henceforth called "environmental GAAP"). Insight on the economics of environmental disclosure practices helps stakeholders to assess the impact of environmental obligations, provides regulators with input on the need for additional reporting guidance, and helps researchers develop the theoretic aspects of social responsibility disclosure.

In this study, we investigate how financial performance, presented under alternative communication strategies, is related to the firm's level of environmental GAAP disclosures, and whether these same disclosures affect investor's valuation of stock price. Previous findings on the relations between financial performance, environmental disclosures (required or voluntary), and corresponding market valuation are mixed. Studies by Neu, Warsame, and Pedwell (1998), Berthelot, Cormier and Magnan (2003a), Cowen, Ferreri, and Parker (1987), and Patten (1991) contain mixed results on financial performance and environmental disclosure. Arguments by Porter and van der Linde (1995) and Al-Tuwaijri, Christensen, and Hughes (2004) suggest that firms that make forthcoming disclosures about their environmental activities will be positively rewarded and most accounting research indicates a negative reward. We hope to provide insight on these ambiguities in the literature by studying a comprehensive set of environmental disclosures based on GAAP in the context of alternative communication strategies. Managers can use different communication strategies to frame environmental disclosures with the intent of influencing investor's perception of environment and operating performance. For example, managers could frame information in a way that educates stakeholders about environmental activities or changes the focus of stakeholder attention away from environmental activities, or managers could decide to not frame the information at all and let investors draw their own conclusions. While GAAP disclosures are presumably triggered by specified events, there is significant latitude in environmental GAAP (particularly with the qualitative disclosures) so that managers have discretion in how the environmental issues are framed and presented to stakeholders. In support of this idea, several studies provide evidence of the discretion exercised by organizations in making the required environmental disclosures (Rockness, Schlacher, and Rockness 1986, Freedman and Stagliano 1995, Berthelot, Cormier and Magnan 2003b).

Our findings indicate that the level of environmental GAAP disclosures provided in 10-K reports is significantly related to the firm's profits; however, the sign of the relation is conditional on the nature of the firm's communication strategy. We observe a positive relation between environmental GAAP and profits when the firm provides the information in the context of confirmatory environmental disclosures; negative otherwise in the context of non-confirmatory disclosures. Consistent with Neu, Warsame, and Pedwell (1998), we define a confirmatory disclosure as one intended to give "confirmation" that profitability has not been at the expense of the environment. These disclosures are intended to correct the misperceptions by stakeholders that financial success and environmental responsibility are incompatible.

Stakeholders, however, are free to interpret the implications of how the GAAP disclosures are framed, and so as a follow-up we test the market valuation of these same environmental disclosures. We find that stock returns are significantly related to environmental GAAP, but the magnitude of the impact of environmental GAAP on stock returns is also conditional on the disclosure strategy. We observe a negative relation between stock returns and environmental GAAP, but a smaller impact when environmental GAAP is provided in the context of confirmatory disclosures (although still negative overall).

In general, our results indicate that the level of environmental GAAP disclosed by managers and interpreted by investors is related to firm-specific financial success and how that financial success is framed. A confirmatory framework about environmental liabilities allows managers to address the skepticism of simultaneously achieving financial success and acting responsibly towards the environment. Thus, confirmatory disclosures are a proactive attempt to set expectations that environmental remedies will lead to operating benefits, possibly those described by Porter and van der Linde (1995), which should be credible since stakeholders can readily verify environmental information from regulatory sources. Al-Tuwaijri et al. (2004) describe the ability to verify the transparency of environmental disclosures by using data sources published by the EPA or other agencies. The negative relation between environmental GAAP and profits in the context of non-confirmatory disclosures suggests that managers cannot or will not commit to the joint pursuit of financial success and environmental responsibility, and therefore, disclose less about their environmental activities. Investors perceive as much and penalize the stock price, accordingly.

An important implication in our study is that the mixed findings on the environmental-disclosure-to-profits relation in previous research may be due to an omitted variable for communication strategies. In our empirical analysis, controlling for the use of a confirmatory disclosure framework has a significant effect on the environmental-disclosure-to-profits and the returns-to-environmental-disclosure associations. The observed results when environmental GAAP is presented in the context of confirmatory disclosures provides further support for Porter and van der Linde's (1995) proposition that good financial performance and good environmental performance can co-exist.

Finally, we provide descriptive evidence that the percentage of environmental GAAP disclosures made by Potentially Responsible Parties (PRPs) has not kept pace with the increase in the total number possible. Before 1993, firms provided about 51 percent of the total number of environmental GAAP disclosures possible. After the implementation of Staff Accounting Bulletin No. 92 in 1992, firms provided about 29 percent of the total environmental GAAP disclosures possible suggesting that there may be an issue of compliance or pertinence related to the additional environmental GAAP.

The remainder of the paper is structured as follows: the next section reviews the regulatory environment and academic literature; Section 3 develops and presents the testable hypotheses; Section 4 introduces the model and sample and Section 5 provides the empirical results; Finally, Section 6 contains some concluding remarks.

ENVIRONMENTAL REGULATIONS AND BACKGROUND

Institutional Background

Environmental regulations define environmental costs and the related disclosures. Statement of Position 96-1, Environmental Remediation Liabilities, summarizes the regulatory process with regard to the Superfund Act (AICPA, 1996). The Superfund Act adopted a "polluter pays" philosophy by establishing the right to bill firms associated with sites for their portion of the remediation costs and levying a tax on certain industries to fund orphaned sites. Several features of the Superfund Act present challenges for estimating a firm's liability under its provisions. First, the liability is comprehensive, including response and remediation costs, as well as damages, health assessments, and study costs. For estimates of the environmental costs involved, see Alciatore, Dee, and Easton (2004). Second, the Superfund Act imposes liability on a broad group of PRPs that includes the site's current owner, and anyone who: (1) owned or operated the facility when hazardous substances were disposed, (2) generated hazardous substances disposed of at the facility, or (3) transported hazardous substances to the disposal facility, and/or arranged for such transportation. Third, the Superfund Act liability is strict, retroactive, and joint and several. Thus, a study of environmental obligations under the Superfund Act is of high interest to a variety of publics because these obligations are some of the largest obligations for publicly traded corporations, can extend well into the future, and may be among the most difficult of costs to predict.

Accounting Requirements

Professional standards provide recognition and disclosure requirements with respect to environmental liabilities. The most relevant accounting guidance related to environmental liabilities is the Statement of Financial Accounting Standards Number 5 (SFAS 5), Accounting for Contingencies (FASB, 1975), which establishes both recognition rules and disclosure rules for contingent liabilities. SFAS 5 states that a loss contingency must be accrued if it is both probable and reasonably estimable.

In addition to the requirements of SFAS 5, firms must comply with guidance issued by the SEC. Most relevant to the current study is Regulation S-K (revised in 1986): Items 101, 103, and 303 (SEC, 2000), and Staff Accounting Bulletin 92 (SAB 92) (SEC, 1993). Item 101 requires a general description of the business and specific disclosure of the effects that compliance with environmental laws may have on capital expenditures, earnings, and competitive position, when material. Item 103 requires disclosure of pending or contemplated administrative or judicial proceedings, and Item 303 requires disclosure of material events and uncertainties known to management that would cause reported financial information to be unrepresentative of future operating results or financial conditions. SAB 92 was issued specifically to improve the disclosure of environmental liability information. As the sample period for the current study is 1991-1997, we do not include the requirements of FAS 143, Accounting for Asset Retirement Obligations (Issued June 2001) or FASB Interpretation No. 47 (FIN 47), Accounting for Conditional Asset Retirement Obligations (Issued March 2005).

In summary, the total number of disclosures that a PRP may have to make related to environmental liabilities is broken down as follows: 6 possible environmental disclosures required under SFAS 5, 4 possible under Regulation S-K, and 19 possible disclosures under SAB 92. Our study relies on these 29 items to measure the level of environmental disclosure in 10k reports.

RELATED LITERATURE

The focus of environmental accounting research can be either on voluntary or mandatory disclosures. The voluntary environmental disclosure stream provides important insights for required environmental disclosures, especially in light of the discretion exercised by managers with the GAAP-based reporting guidelines (Berthelot, Cormier, and Magnan 2003b, Neu, Warsame, and Pedwell 1998, and Cormier and Magnum 1999).

Early studies based on anecdotal evidence suggested firm's environmental disclosures were self-serving and inaccurate (Beams and Fertig, 1971; Estes, 1976; Churchill, 1978; Nader, 1978), while later empirical studies questioned their quality and content adequacy (Gamble, Hsu, and Radke, 1995; Freedman and Wasley, 1990; Rockness, 1985; Wiseman, 1982, Ingram and Frazier, 1980). The dependability of firms to report environmental debts is apparently subject to discretion. Rockness, Schlacter, and Rockness (1986) report that most firms in their study do not mention, let alone quantify, the possibility of an environmental debt even though the firm has been identified as responsible for at least one contaminated site. Freedman and Stagliano (1995) report that one-quarter of their sample of superfund-affected firms does not disclose any information about their superfund issues. Other descriptive research suggests that environmental disclosure quality is generally low, and that firms generally do not record environmental liabilities (Price Waterhouse, 1992, 1994; Gamble et al., 1995; Kreuze, Newell, and Newell, 1996; Walden and Schwartz, 1997).

Previous research on the market effects of mandatory environmental disclosures is mixed. Li and McConomy (1999) and Berthelot, Cormier, and Magnan (2003a) find the adoption of environmental reporting standards lowers stock price, while Blacconiere and Northcut (1997) find a positive relation between returns and environmental information in the time leading up to the adoption of the Superfund amendments. Consistent with Blacconiere and Northcut (1997), Freedman and Stagliano (1991) and Blaconniere and Patten (1994) find there is less of a stock price penalty imposed by investors on firms disclosing environmental information. However, these firms are still penalized overall, which seems to be inconsistent with arguments by Porter and van der Linde (1995) who suggest that firms that make forthcoming disclosures about their environmental activities will be positively rewarded.

Recently, Hughes and Reynolds (2001) and Bae and Sami (2005) examine earnings response coefficients (ERCs) for firms with environmental liabilities. Hughes and Reynolds (2001) find higher ERCs for high polluters than low polluters in times of higher uncertainty about environmental costs. Bae and Sami (2005) find lower ERCs for PRP firms than non-PRP firms, a finding that is seemingly at odds with Hughes and Reynolds (2001) if we assume that association with a PRP raises the uncertainty about environmental costs.

Previous research examining the association between environmental GAAP (or voluntary environmental disclosure) and profitability generally provides mixed findings. Berthelot, Cormier and Magnan (2003a) find that accounting provisions for site removal and remediation specified under Canadian Institute of Chartered Accountants' standards are positively associated with changes in earnings. For the relation between voluntary environmental disclosure and profitability, Cowen, Ferreri, and Parker (1987) and Patten (1991) document an insignificant relation, Cormier and Magnan (1999) document a positive relation, while Neu et al. (1998) document a negative relation. Our paper adds to this research by testing for differences in how managers present, and investors interpret, environmental disclosures under alternative communication strategies.

HYPOTHESES DEVELOPMENT

Hypotheses for the Level of Environmental GAAP Disclosure

We propose that corporate managers are concerned with the implications that financial performance holds for the perception of responsible social behavior. Therefore, managers will use disclosure to explain environmental actions to the firm's important stakeholders. As previous research suggests, investors could infer that profitability was at the expense of the environment (i.e., reduced environmental protection effort). Otherwise, as Porter and van der Linde (1995) imply, investors could infer that profitability was at the benefit of the environment (i.e., converting waste into saleable byproducts). Either way, a salient financial performance indicator such as profits, provides different incentives for the manager to influence perceptions of the firm's environmental behavior.

The above discussion suggests that disclosures could be used to frame information that is released to primary constituents. Similar to a dichotomy suggested by Neu et al. (1998), we propose that environmental disclosures can be confirmatory or non-confirmatory. A confirmatory communication strategy would provide information that informs constituents that financial performance and environmental responsibility are compatible, and educates constituents how that compatibility will be sustained or achieved. We also propose that in order for confirmatory disclosures to meet the needs of forward-looking investors, the disclosures would have to be credible. On the other hand, non-confirmatory disclosures could frame information in a way that redirects the focus of constituents away from the issue of financial performance and environmental compatibility, or would simply not address the compatibility issue at all. The scrutiny given to large, publicly traded corporations suggests that managers are not likely to provide disclosures that redirect attention or are intended to deceive constituents. Managers of large corporations are more likely to remain "silent" on matters they cannot explain or cannot conscientiously guarantee. Therefore, in this setting, we assume that a non-confirmatory approach is one where the manager minimizes or reduces the disclosures on the concurrent goals of corporate profitability and environmental responsibility.

Our first prediction is that profits affect the manager's decision to disclose information about corporate environmental issues. We do not predict a sign for the first prediction since our interest here is to focus on whether environmental GAAP, as opposed to voluntary environmental disclosure, is affected by profits at all. This allows us to test overall, cross-sectional variation, and tie our results to previous research. Stated in alternative form, our first hypothesis is:

H1: The level of environmental GAAP disclosed is associated with profitability.

Our second prediction is that communication strategy in terms of providing disclosures in a confirmatory or non-confirmatory framework is a conditioning factor in how profits affect the disclosure level of environmental GAAP. Under a confirmatory framework, managers will make credible disclosures to convince investors that profitability and environmental responsibility are congruent corporate objectives, and the concurrent pursuit of both will lead to improved competitive position and higher future returns. Under a non-confirmatory framework, managers will not or cannot pursue the concurrent objectives of profitability and environmental responsibility. Without a credible commitment, the manager's best option is to curtail disclosures about the relation between profits and environmental responsibility. Our second and third hypotheses are:

H2: Given a confirmatory framework, the level of environmental GAAP disclosed is positively associated with profits.

H3: Given a non-confirmatory framework, the level of environmental GAAP disclosed is negatively associated with profits.

Hypotheses for Market Effects

In capital markets, investors will determine whether the environmental disclosures made by managers are useful in valuation or not. If investors perceive the environmental disclosures as credible, they may be willing to assign positive value to those disclosures that are confirmatory about the compatibility of environmental responsibility and financial performance. Porter and van der Linde (1995) suggest that operating decisions in favor of protecting the environment could improve performance as companies become efficient in using raw materials or turn waste into saleable byproducts. If managers do not provide confirmatory disclosures (i.e., they are non-confirmatory), then investors may interpret this to mean that managers are not willing to commit to the dual objectives of environmental responsibility and higher financial performance. Our fourth hypothesis predicts a significant relation between market returns and environmental GAAP disclosures. Our test of this hypothesis will help us tie our findings to similar tests in previous research.

H4: Market returns are associated with the level of environmental GAAP disclosed.

Our fifth and sixth hypotheses suggest the market valuation of environmental GAAP will differ under confirmatory and non-confirmatory disclosure frameworks as follows:

H5: Given a confirmatory framework, market returns are positively associated with the level of environmental GAAP disclosed.

H6: Given a non-confirmatory framework, market returns are negatively associated with the level of environmental GAAP disclosed.

In sum, the predictions for investor's valuation of environmental GAAP are similarly based on the same reasoning that managers have for framing disclosures about environmental GAAP: confirmatory disclosures will be perceived favorably due to the implied future benefits, and the lack of confirmatory disclosures will not be perceived favorably.

MODELS, VARIABLES, AND SAMPLE

Environmental GAAP Disclosure Model

The general form of our model is as follows:

Environmental GAAP Disclosure = [florin] (Firm-Specific Characteristics, Industry Related Characteristics, and Financial Performance).

Substituting empirical proxies, we use the following regression model to test our hypotheses related to the determination of environmental GAAP.

ENVGAAP = [alpha].sub.0] + [alpha].sub.0]ln(SIZE) + [alpha].sub.2]CHEM + [alpha].sub.3]OIL + [alpha].sub.4]PAPER + [alpha].sub.5]STEEL + [alpha].sub.6]POWER + [alpha].sub.7]ENVLIAB + [alpha].sub.8]SITES + [alpha].sub.9]POSTSAB92 + [alpha].sub.10]CAPX + [alpha].sub.11]ROA + [epsilon].

Dependent Variable for Environmental GAAP Model.

To measure the level of disclosure, we construct a comprehensive index of environmental liability disclosures based on the requirements in Regulation S-K (items 101, 103 and 303), SAB 92, and SFAS 5. Table 1 summarizes the twenty-nine environmental GAAP disclosure items that form the basis for our index. Firm 10K reports are examined for the presence or absence of specific statements as outlined in the Table 1. Two reviewers (the author and a research assistant) evaluate each 10K report independently. The reviewers met routinely to discuss independent evaluations and resolve interpretive issues. The following procedures are performed for each sample firm in each year from 1991-1997 in developing the disclosure index:

1. A score of 1 is given for each disclosure item presented in the 10K (based on the listing of disclosure items in Table 1). Thus, the environmental disclosure score ranges from 0 (for no disclosure) to a maximum of 10 for years 1991 and 1992 (prior to SAB 92), and from 0 to 29 for years 1993-1997 (including disclosures required by SAB 92).

2. The environmental disclosure score is divided by the total number of environmental disclosure index items for each year, 10 for years 1991 and 1992, and 29 for years 1993-1997. The firms' final score for each sample year represents the percentage of GAAP environmental disclosures present out of the total possible. Thus, the index variable (ENVGAAP) equally weights the disclosure items.

Experimental Variables for Environmental GAAP Model

The primary experimental variable is return on assets (ROA), a measure of the firm's profitability. The level of profitability, for a given communication strategy, can provide managers with different incentives to influence impressions through disclosure. Without knowing whether a confirmatory or non-confirmatory communication strategy is in place, we simply predict that this variable will be significantly associated with environmental GAAP (see hypothesis H1).

After controlling for communication strategy, we expect ROA to be positively associated with environmental GAAP under a confirmatory strategy (hypotheses H2), and negative under a non-confirmatory strategy (hypotheses H3). Our confirmatory variable, CONFIRM, is used as an interaction variable to test whether ROA has a different coefficient under a confirmatory versus non-confirmatory disclosure strategy. CONFIRM is constructed by examining the coding of item number 2 in our disclosure index (see Item 2 in Table 1). The coding of item 2 indicates whether a disclosure for estimated environmental capital expenditures was made under SEC Regulation S-K (Item 101). We code CONFIRM equal to one when the firm makes this disclosure and the firm's ROA is less than the mean of the sample. Constructing CONFIRM in this fashion is consistent with the proposition by Neu et al. (1998) and Herremans, Akathaporn, and McInnes (1993) who state that a confirmatory disclosure is one where "in periods of relative unprofitability these same disclosures might be directed at convincing financial stakeholders that current environmental investments will result in a future competitive advantage and future profits." Thus, we have essentially identified firms whose profits are lower than average, but are profitable none-the-less, and yet, are still willing to make environmental investments for the future. This sample cut results in partitions that are reasonably balanced between above- and below-average profitability. The resulting partitions contain a significant number of firms making the environmental capital expenditure disclosure under Regulation S-K, Item 101, as follows: 307 firms in the partition where ROA is greater than the sample mean, and 245 firms in the partition where ROA is less than the sample mean. A sensitivity analysis using these sub-samples is discussed performed later in the paper. An important assumption for the validity of our CONFIRM proxy is that disclosure about environmental capital expenditures during times of lower profitability is a leading indicator that the firm's overall disclosure framework is a confirmatory one, signifying that profitability and environmental responsibility are not a trade-off, and therefore, are compatible corporate objectives. For econometric purposes, we eliminate this item from our environmental disclosure index when using the CONFIRM as an explanatory variable, and rename the environmental disclosure index as ENVGAAPx2. Specifically, we exclude item 2 from our original disclosure index (ENVGAAP) so that it is now based on a count of 28 possible GAAP disclosures instead of the initial 29 possible. This is to avoid inducing an algebraic bias in the regression by conditioning the right-hand-side variables on the basis of the dependent variable. Excluding Item 2 from the dependent variable insures the dependent variable and the independent variables are not measuring the same construct.

To test hypotheses H2 and H3, we examine the significance and sign of the coefficients on the interaction term, ROA*CONFIRM, and the main effect, ROA. A finding in support of H2 would mean the coefficient on ROA*CONFIRM and the coefficient on ROA would sum up to a positive number, and a finding in support of H3 would mean the coefficient on ROA is negative.

Control Variables for Environmental GAAP Model

Watts and Zimmerman (1978) suggest an association between company size and social responsibility disclosure. We use the total assets to proxy for SIZE, which is logged for regression testing purposes.

We use dummy variables to identify firms in the five industries included in the Counsel on Economic Priorities (CEP) studies (CHEM, OIL, PAPER, STEEL, POWER). The CEP identified these five industries as generating significant environmental hazards. Many companies within these five industries are also voluntarily involved with the "Responsible Care" initiative started by the chemical industry. The mission of "Responsible Care" is to go above and beyond government regulations and openly communicate with the public. Part of this initiative is to track performance using standard environmental, health, safety and security measures. Therefore, for reasons identified by the CEP or the "Responsible Care" initiative, we expect that companies in these industries to disclose more than companies in other industries.

Proxies in previous research for regulatory influence or pressure include an estimated average environmental liability per PRP (Barth and McNichols, 1994) or the number of PRP sites identified per firm (Stanny, 1998). Alciatore et al. show that remediation liability, the central focus in most studies, is just one subset of the total environmental costs. Environmental exit costs are another subset of costs, and indeed can be much larger than the remediation costs. To capture the regulatory influence or pressure related to all environmental costs, we include both the estimated average environmental liability (ENVLIAB) and the number of PRP sites identified per firm (SITES). The two proxies together are likely to capture more of the components of regulatory pressure or rules pertaining to environmental disclosure than either of the proxies on an individual basis. We expect estimated liabilities (ENVLIAB) and the number of sites per PRP (SITES) to be positively associated with increased disclosure.

The environmental liability (ENVLIAB) is based on information provided in the ROD, which provides estimated costs of cleanup for Superfund sites. The ROD is obtained from the Superfund Public Information System (SPIS), which contains the full-text of the official ROD documents signed and issued by EPA from Fiscal Years 1982-1997. The following procedures are used to calculate the average liability, ENVLIAB:

3. The Superfund PRP Listing from the EPA is used to identify the number of sites to which each sample firm is named as of 12/31/1997.

4. The SPIS database is used to obtain the RODs for all sites to which sample firms are named as of 1997. The ROD is examined for each site to obtain the Present Worth Cost (PWC), which represents the present value of the estimated clean up costs for the site.

5. For each site, the total number of PRPs is determined by sorting the Superfund PRP Listing by site number.

6. The number of publicly traded PRPs for each site is obtained using the EDGAR database, and is used to compute an average liability for each site. The average liability is calculated by dividing the PWC by the number of publicly traded PRPs for each site, which indicates the potential for shared responsibility for cleanup.

7. The average liability for each site to which a sample firm is named is then added to obtain a total average liability. The current study uses total average liability as a proxy for potential environmental liability.

After 1992, the number of total, mandated disclosures possible increased from 10 to 29 with the introduction of SAB 92. We include a 1:0 indicator variable, POSTSAB92, to control for this effect.

Since our CONFIRM variable is an identifier based on disclosure about estimated environmental investment, we control for capital expenditures to separately identify the related confirmatory effects of disclosure. Therefore, we include CAPX in the regression, which is total capital expenditures scaled by total assets.

Market Model

The general form of our market model is:

Market Returns = [florin] (Firm-Specific Characteristics, Industry Related Characteristics, and Environmental Disclosure).

Substituting empirical proxies, we use the following regression model to test our hypotheses related to the valuation effects of environmental GAAP.

RETURN = [[beta].sub.0] + [[beta].sub.1]MKBK + [[beta].sub.2]CHEM + [[beta].sub.3]OIL + [[beta].sub.4]PAPER + [[beta].sub.5]STEEL + [[beta].sub.6]POWER + [[beta].sub.7]ENVLIAB + [[beta].sub.8]POSTSAB92 + [[beta].sub.9]ROS + [[beta].sub.10]UE + [[beta].sub.11]CAPX + [[beta].sub.11]ENVGAAP + [epsilon].

Dependent Variable for the Market Model

RETURN is an industry-adjusted annual return. We calculate RETURN as the 12-month change in stock price ending 3 months after the fiscal year end (adjusted for dividends), scaled by the stock price at the beginning of this 12-month window. This staggered return window is intended to capture the disclosure effects of the 10-K report, which is typically released about 3 months after the fiscal year end.

Experimental Variables for the Market Model

To assess hypothesis H4, we examine the significance of the coefficient on ENVGAAP, regardless of sign. To examine hypotheses H5 and H6, we examine the significance and sign of the coefficients on the interaction term, ENVGAAPx2*CONFIRM, and the main effect, ENVGAAPx2 (as noted before, ENVGAAPx2 is the disclosure index, ENVGAAP, excluding Item 2). A finding in support of H5 would mean the coefficient on ENVGAAPx2*CONFIRM and the coefficient on ENVGAAPx2 would sum up to a positive number, and a finding in support of H6 would mean the coefficient on ENVGAAPx2 is negative.

Control Variables for Environmental GAAP Model

MKBK is the ratio of market value of equity to the book value of equity and is a proxy for future growth opportunities. We expect a positive relation between MKBK and returns. We include control variables for the industries identified by the CEP as particularly hazardous (CHEM, OIL, PAPER, STEEL, POWER) in the event the excess industry-adjusted returns are systematically different for these industries.

POSTSAB92, an indicator variable controls for the SEC's environmental disclosures added in 1992; and ROS, return-on-sales, is included as a proxy for the impact of a firm's cost control on firm value (Al-Tuwaijri et al. 2004).

In the market model we use ENVLIAB as a control for environmental exposure. Various proxies have been used for this construct in previous research (Al-Tuwaijri et al. 2004), which controls for pollution intensity. We expect this variable to be negatively related to returns since future environmental costs should reduce the market value of the firm.

To control unexpected returns due to unexpected earnings, we include UE, a metric based on the annual change in earnings-per-share divided by the stock price at the beginning of the period. The positive association between earnings and returns is well documented in the accounting literature. We also include a capital expenditure variable, CAPX, to control for the future economic benefits from capitalized assets. CAPX is the current period reported capital expenditures scaled by total assets and should be positively associated with returns (Clarkson, Li, and Richardson 2004).

Sample Selection

Table 2 provides information about our sample selection process. Our goal is to identify a sample of big firms with a known PRP association and, therefore, a high probability of using environmental GAAP. We identified 245 Fortune 500 firms that are named as a PRP on the SPIS database. We then eliminated financial and business services firms (23), firms that had no ROD issued (27), firms that were not publicly traded during the entire sample period (11), and firms that were added as a PRP during the sample period (13). For the 171 firms in our final sample we obtained 1,187 firm-year observations with complete data for regression testing (10 firm-year observations had missing data).

RESULTS

Descriptive Statistics

The sample firms represent a broad cross-section of industries. Table 3 shows that 13 primary industries are represented at the general 2-digit SIC industry classification level. Industries that contain fewer than 10 firm-year observations are not explicitly listed. At the 4-digit SIC classification level, there are 33 industries represented. The five industries identified by the Counsel on Economic Priorities as generating significant environmental hazards, namely, oil, chemicals, power, paper, and steel, represent a combined total of 36.0 percent of the firm-year observations. The machinery industry has the largest representation with 27.6 percent of the sample.

Descriptive statistics for model variables are summarized at Table 4. The mean SIZE of sample firms, measured by total assets, is $14,208 million and the mean ROA is 4.6%. As expected, the sample firms are large with positive profits since they are drawn from the Fortune 500. However, the mean industry-adjusted market (RETURN) is negative, possibly reflecting the lower performance of larger companies during this period. Unadjusted market returns (RAWRETURN) for the sample are positive overall, with a mean and median of 19.4% and 16.6%, respectively. Consistent with the ROA statistics, the overall raw market return is positive.

The mean estimated environmental liability, ENVLIAB, is $17.2 million, which is a small percentage of the total assets of the sample firms. None of the sample firms have an estimated liability that exceeds 5% of total debt, the materiality threshold for mandatory accrual under 10k reporting requirements of the SEC. The mean number of PRP sites (SITES) per sample firm is 17.1, which is higher relative to previous research (Stanny 1998), however, our sample is taken from a later period, a time when more PRPs have been identified, and is drawn from a set of firms that are typically larger with greater output than the average firm in Compustat. The mean for POSTSAB92 is 0.73, indicating a large proportion of the sample observations occur subsequent to the implementation of SAB 92. ENVGAAP, the proportion of environmental GAAP disclosures made, has a mean of 0.35.

The mean for CONFIRM is 0.26, indicating that roughly one-fourth of the sample is making a leading, confirmatory disclosure about future environmental capital expenditures even though profits for these firms are less than the mean of the sample. The mean unexpected return (UE) is 0.04, which is consistent with previous research; however, the mean is not statistically different from zero. The median UE is 0.01 and is statistically different from zero based on a rank sum test. All other means are statistically different from zero. The means for return-on-sales (ROS), market-to-book equity (MKBK), and capital expenditures (CAPX) are 4.7%, 2.9, and $947M, respectively, and are consistent with previous research.

Tables 5 and 6 provide a graphical representation of the environmental GAAP disclosures over the sample period. In Table 5, the total number of environmental GAAP disclosures possible for the sample firms increases after 1992 from 1,710 (171 sample firms x 10 mandatory disclosures possible) to 4,959 (171 sample firms x 29 mandatory disclosures possible) while the actual number of environmental GAAP disclosures after 1992 increases from 868 to 1449. On a per firm basis, total possible disclosures increases from 10 to 29 in 1993, a 190 percent increase, while the average number of actual disclosures increases from 5.1 to 8.4, a 65 percent increase. From 1994 to 1997, the change in actual disclosures made appears to level off as the percentage increase is only 0.06 percent per year. Of course, we do not know if the firms are under-reporting with respect to the environmental GAAP disclosures since we cannot verify whether the underlying event has actually occurred. However, we note that the increase in actual mandated disclosures is not commensurate with the increase in total possible mandated disclosures, and that the gap between the two remains fairly constant after 1993.

Table 6 shows that starting in 1993 PRPs were making 3 out 4 Regulation S-K disclosures, and under SFAS 5 were making 2.7 out of the 6 mandatory disclosures possible. In terms of the average number of total environmental GAAP disclosures possible over the entire time frame, PRPs were making about 51% of the possible disclosures required under Regulation S-K and SFAS 5 (pre-1993), which then dropped off to about 29% at the time that the requirements for the additional SAB 92 disclosures were implemented. The proportionate drop is a result of an increasing denominator, however, the percentage of total disclosures made has leveled off at around 30% through 1997. Thus, the increase in actual disclosures made is apparently not commensurate with the increase in total environmental GAAP disclosures possible, which may hold implications about compliance or the pertinence (i.e., applicability) of the new disclosures required after 1992.

[GRAPHIC OMITTED]

[GRAPHIC OMITTED]

Regressions

Table 7 reports the regression results for our environmental GAAP model related to hypothesis H1. The overall explanatory power of model is significant. The adjusted R-square is about 54.8%, and F-statistic for the model is 132.09. The coefficients on the control variables are all significant in the expected direction. The coefficients on ENVLIAB and SITES are positive and significant, consistent with the regulatory influence hypothesis, and the coefficient on POSTSAB92 is negative and significant indicating the proportion of environmental GAAP disclosures made after the implementation of SAB 92 decreased significantly. Consistent with hypothesis H1 about profits, the coefficient on the experimental variable, ROA is significant at p < .01 suggesting it is an important determinant of the level of environmental GAAP disclosures provide by the firm. The sign of the association is negative, which is consistent with previous research by Al-Tuwaijri et al. (2004) and Neu et al. (1998), indicating that the level of environmental GAAP disclosed increases as profitability deteriorates.

Table 8 reports the regression results for an environmental GAAP model with the primary purpose of comparing the coefficients on ROA in the "confirmatory" and "non-confirmatory" partitions of the sample. The adjusted R-square is 58 percent, the F-statistic for the model is significant, and the sign and t-statistics for all controls variables are significant and in the expected direction. The coefficient on ROA is negative and significant at the 0.01 level. This is a baseline coefficient and represents the slope for ROA when CONFIRM = 0, the non-confirmatory partition. The negative sign is consistent with hypothesis H3, suggesting that managers that cannot confirm the concurrent objectives of profitability and environmental responsibility choose to disclose less about environmental activities. The coefficient on ROA*CONFIRM, as a differential coefficient, is positive and significant indicating the slope coefficient on ROA when CONFIRM = 1 is statistically larger than the slope coefficient on ROA when CONFIRM = 0. Adding the coefficient on ROA to the coefficient on ROA*CONFIRM provides an estimate of the magnitude of the slope coefficient on ROA when CONFIRM = 1. This estimate is positive overall (coefficients from Table 8: -0.003 + 0.007 = 0.004), which is consistent with hypothesis H2.

In sum, the environmental GAAP model results presented in Tables 7 and 8 suggest there is a differential effect of profits on environmental disclosure for a given disclosure framework. Under a confirmatory framework, managers are providing more environmental GAAP disclosures to persuade constituents that profits are not at the expense of the environment. Under a non-confirmatory framework, managers have not committed to future environmental investment, and therefore, do not make disclosures that profitability and environmental investment are compatible goals for the future. A related implication is that managers prefer to disclose less or remain silent about environmental issues as profitability increases (under a non-confirmatory orientation).

In Table 9 we report the regression results of a market model that is intended to test if ENVGAAP is significant in explaining market returns. The adjusted R-square is 7.6 percent, the F-statistic is 8.20 (significant at the 0.01 level), and most of the control variables are significant in the expected direction. Consistent with hypothesis H4, the coefficient on ENVGAAP is significant at the 0.10 level, suggesting that environmental GAAP disclosures are an important determinant of market returns. The negative sign on ENVGAAP is consistent with previous studies suggesting environmental GAAP disclosures are associated with lower market returns (Al-Tuwaijri et al. 2004 and Neu et al. 1998).

In Table 10, the coefficient on ENVGAAPx2 is negative and significant at the 0.05 level. In an analysis similar to that presented previously with the environmental GAAP disclosure model, the coefficient on ENVGAAPx2 is a baseline coefficient and represents the slope for ENVGAAPx2 when CONFIRM = 0, the non-confirmatory partition. The significant negative sign is consistent with hypothesis H6, suggesting investors penalize stock price when environmental disclosures are made that are not confirmatory. The coefficient on ENVGAAPx2*CONFIRM, as a differential coefficient, is positive and significant indicating the slope coefficient on ENVGAAPx2 when CONFIRM = 1 is statistically larger than the slope coefficient on ENVGAAPx2 when CONFIRM = 0. Adding the coefficient on ENVGAAPx2 to the coefficient on ENVGAAPx2*CONFIRM provides an estimate of the magnitude of the slope coefficient on ENVGAAPx2 when CONFIRM = 1. This estimate is negative overall (coefficients from Table 10: -0.108 + 0.094 = -0.014), which is not consistent with hypothesis H5.

In sum, the market model results presented in Tables 9 and 10 suggest there is a differential effect of environmental GAAP disclosure on stock returns for a given disclosure framework, but only to the extent that it mitigates negative valuation effects. Under a non-confirmatory framework, investors penalize stock price for lower levels of environmental GAAP disclosures; under a confirmatory framework, investors penalize the stock price to a lesser extent. The confirmatory disclosure framework minimizes the magnitude of the negative effect of environmental GAAP disclosures on stock price.

ROBUSTNESS CHECKS

Regression Diagnostics

We run a number of regression diagnostics to test the robustness of our results. The current estimates are reported under the assumptions of ordinary least squares and it is possible that the SIZE and ENVLIAB variables may create a heteroscedasticity problem. Therefore we run the regressions with White's (1980) robust estimator and get results that are qualitatively the same as those presented. The Durbin-Watson statistic does not indicate the presence of significant autocorrelation, however, as a sensitivity we also control for the effects of time by including a dummy variable for each year. These results are identical to the results presented, which include a dummy variable for the time period after SAB 92 implementation. Finally, multicollinearity does not appear to be an issue since all variance inflation factors are less than 2.2, and the reported findings are insensitive to influential observations as identified by Belsley, Kuh, and Welsch (1980) diagnostics.

Sensitivities for capital expenditures

Although we control for the effects of total capital expenditures, which includes environmental capital expenditures, as an additional measure of control for capital expenditures we eliminate all observations from the non-confirmatory partition that do not report an environmental GAAP disclosure under Regulation S-K, Item 101, an estimate for future environmental capital expenditures (see item # 2, Table 1). After this sample screen, the non-confirmatory and confirmatory partitions contain only those firms that disclose an estimate for current and future environmental capital expenditures. This results in 245 observations in the non-confirmatory partition and 307 observations in the confirmatory partition. Using these 552 observations, the regression results (not presented) for the environmental GAAP disclosure model are qualitatively the same as those reported in Table 8: the coefficient on ROA is negative and significant (estimate = -0.011, t-statistic = -3.52, p-value < 0.01), and the coefficient on the interaction term, ROA*CONFIRM, is positive and significant (estimate = 0.014, t-statistic = 2.75, p-value < 0.01). Adding the baseline coefficient on ROA (which represents the slope in the non-confirmatory partition) to the differential coefficient on ROA*CONFIRM provides an estimate of the total coefficient on ROA in the confirmatory partition (-0.011 + 0.014 = 0.003). Thus, the estimated coefficient on ROA in the non-confirmatory partition versus the confirmatory partition is -0.011 versus 0.003, respectively. The sign and significance of these estimates supports hypotheses H2 and H3.

We use the same 552 observations for a sensitivity analysis on capital expenditures in the market model. In terms of hypotheses H6, the regression sensitivity result (not presented) is qualitatively better than that reported in Table 9: the coefficient on ENVGAAPx2 is negative, but not significant (estimate = -0.061, t-statistic = -0.60, p-value = 0.27), however, the coefficient on the interaction term, ENVGAAPx2*CONFIRM, is positive and marginally significant (estimate = 0.130, t-statistic = 1.35, p-value = 0.08). Adding the baseline coefficient on ENVGAAPx2 (which represents the slope in non-confirmatory partition) to the differential coefficient on ENVGAAPx2*CONFIRM provides an estimate of the total coefficient on ROA in the confirmatory partition which is positive (-0.061 + 0.130 = 0.069). Thus, the estimated coefficient on ENVGAAPx2 in the non-confirmatory versus the confirmatory partition is -0.061 versus 0.069, respectively. Although there is no significant evidence in support of H5, there is significant evidence in support of H6.

Sensitivity for Industry Effects

Including the dummy variables for the five industries identified by the CEP increases the adjusted R-square by approximately 16 percentage points for the environmental GAAP model, but only 1 percentage point for the market model. Outside of the five CEP industries, the industry distribution statistics show that the Machinery industry is the largest industry group, followed by Food and Transportation. Sensitivities that include dummies for these industries do not improve or change the explanatory power of the regressions, and so it appears that the industries identified by the CEP are the ones making the additional environmental GAAP disclosures.

Sensitivity for Self-Insurance Disclosure

In constructing ENVGAAP, our implicit assumption is that a higher score represents a higher level of disclosure. However, the disclosure items related to insurance may not represent a meaningful disclosure. Many large firms self-insure, and in some cases a firm's insurance may not cover contingent liabilities. Thus, it is not clear that a lack of disclosure relative to the insurance items is necessarily a deficiency in disclosure. To examine if these insurance items have an impact on our results, we construct the variable ENVGAAPx21_24 by eliminating disclosure items 21 through 24 from the index and rerun the empirical tests. The results are not qualitatively different.

Sensitivity for Estimated Environmental Liability

A potential model specification issue is related to the fact that our liability estimate is an average and is not weighted toward those PRPs that are likely to have a larger portion of the liability on each site. Thus, the negative sign on our profitability proxy (ROA) could be a consequence of bigger firms having bigger liabilities that tend to disclose more, and therefore, lower profitability because they have to accrue more. Prior research has shown that the chemical industry has a relatively large number of PRPs identified that are associated with a large number of sites suggesting PRPs in the chemical industry are more likely to pay (Campbell, Sefcik, and Soderstrom 2003). As a sensitivity we interact ROA with the chemical industry dummy to determine if a differential coefficient on ROA exists for the chemical industry. The estimated coefficient on the interaction term is insignificant suggesting the coefficient on ROA in the chemical industry is no different than the rest of the cross-section.

CONCLUSION

Our study contributes to the understanding of environmental disclosure practices of publicly traded U.S. firms by examining the factors associated with the extent of environmental GAAP disclosures, as well as the valuation effects of the environmental GAAP disclosures in 10K filings. Identifying factors that are systematically related to environmental disclosure helps regulators, investors, and other users of 10k information to be on guard in reading the financial information where the model predicts a potential lack of disclosure. Consistent with previous research, our findings indicate that profitability is significantly associated with environmental GAAP disclosures, and these same disclosures are value-relevant to investors. We also find that the effect of profitability on the level of environmental GAAP disclosed is contingent on how the disclosures are framed. A confirmatory disclosure framework indicates the manager's credible intent to simultaneously achieve financial success and environmental responsibility, as suggested by Porter and van der Linde (1995). In this scenario, the manager discloses more information about the added benefits of pursuing a joint strategy of higher financial performance and increasing environmental responsibility in the future. Under a non-confirmatory disclosure framework, the manager cannot or will not credibly indicate that simultaneous pursuit of profits and environmental responsibility is possible, and so he reduces his disclosures about environmental activities. Investors perceive differences in confirmatory versus non-confirmatory disclosures, mitigating the negative valuation impact of environmental GAAP disclosures under the confirmatory framework.

We also provide some striking descriptive evidence about a sample of firms that have been specifically identified as potentially responsible parties by the EPA. We find that the percentage of environmental GAAP disclosures actually made by these firms decreases from 51 percent to 29 percent at a time when the required disclosures for environmental liabilities virtually triples. Although the level of environmental disclosures may be appropriate, the drop in the rate of reporting environmental GAAP disclosures raises questions about the usefulness of the additional GAAP or the compliance rate by PRPs, both of which question the completeness of environmental disclosures.

REFERENCES

Alciatore, M., C. C. Dee, and P. Easton (2004). 'Changes in Environmental Regulation and Reporting: The Case of the Petroleum Industry from 1989 to 1998,' Journal of Accounting and Public Policy, Vol. 23, pp. 295-304.

Al-Tuwaijri, S.A., T. E. Christensen, and K. E. Hughes, II (2004). 'The Relations Among Environmental Disclosure, Environmental Performance, and Economic Performance: A Simultaneous Equations Approach,' Accounting, Organizations, and Society Vol. 29, pp. 447-471.

American Institute of Certified Public Accountants (AICPA) (1996). Statement of Position 96-1: Environmental Remediation Liabilities (Including Audit Guidance). New York, NY.

Bae, B. and H. Sami (2005). 'The Effect of Potential Environmental Liabilities on Earnings Response Coefficients,' Journal of Accounting, Auditing & Finance Vol. 20, No. 1, pp. 43-70.

Barth, M. E., and M. F. McNichols (1994). 'Estimation and Market Valuation of Environmental Liabilities Relating to Superfund Sites,' Journal of Accounting Research Vol. 32, (Supplement), pp. 177-209.

Beams, F. A., and E. Fertig (1971). 'Pollution Control Through Social Cost Conversion,' Journal of Accounting Vol. 132, No 5, pp. 37-42.

Belsley, D.A., E. Kuh, and R. E. Welsch (1980). Regression Diagnostics. New York: John Wiley and Sons.

Berthelot, S., D. Cormier and M. Magnan (2003a). 'Les Provisions Environnementales et la Gestion Strategique des Resultats: Une Etude Canadienne,' Comptabilite-controle-audit Vol. 9, No. 2, pp. 109-136.

Berthelot, S., D. Cormier and M. Magnan (2003b). 'Environmental Disclosure Research: Review and Synthesis,' Journal of Accounting Literature Vol. 22, pp. 1-44.

Blacconiere, W., and W. Northcut (1997). 'Environmental Information and Market Reactions to Environmental Legislation,' Journal of Accounting, Auditing, and Finance Vol. 12, (Spring), pp. 149-178.

Blacconiere, W. and D. Patten (1994). 'Environmental Disclosures, Regulatory Costs, and Changes in Firm Value,' Journal of Accounting and Economics Vol. 18, pp. 357-377.

Campbell, K., S. Sefcik and N. S. Soderstrom (2003). 'Disclosure of Private Information and Reduction of Uncertainty: Environmental Liabilities in the Chemical Industry,' Review of Quantitative Finance and Accounting, Vol. 21, pp. 349-378.

Churchill, T. A. (1978). 'Reporting on Corporate Social Responsibility: A Progress Report,' Journal of Contemporary Business, Vol. 7, No. 1, pp. 5-18.

Clarkson, P. M., Y. Li and G.D. Richardson (2004). 'The Market Valuation of Environmental Capital Expenditures by Pulp and Paper Companies,' The Accounting Review, Vol. 79, No. 2, pp. 329-353.

Committee on Commerce. (1995). Reform of Superfund Act of 1995: Hearings Before the Subcommittee on Commerce, Trade and Hazardous Materials of the Committee on Energy and Commerce. Washington DC, U.S. Government Printing Office.

Cormier, D., and M. Magnan (1999). 'Corporate Environmental Disclosure Strategies: Determinants, Costs and Benefits,' Journal of Accounting, Auditing, and Finance Vol. 14, No. 4, pp. 429-451.

Cowen, S., L. Ferreri and L. Parker (1987). 'The Impact of Corporate Characteristics on Social Responsibility Disclosure: A Typology and Frequency-Based Analysis,' Accounting, Organizations, and Society Vol. 12, No. 2, pp. 111-122.

Estes, P. (1976). 'Standards for Corporate Social Reporting,' Management Accounting, Vol. 58, No. 5, pp. 19-22. Financial Accounting Standards Board. (1975). Accounting for Contingencies, FASB Statement No. 5, Stamford, CT: FASB.

Freedman, M., and A. Stagliano (1995). 'Disclosure of Environmental Cleanup Costs: The Impact of the Superfund Act,' Advances in Public Interest Accounting, Vol. 6, pp. 163-176.

Freedman, M., and C. Wasley (1990). 'The Association Between Environmental Performance and Environmental Disclosure in Annual Reports and 10Ks,' Advances in Public Interest Accounting, pp. 183-193.

Gamble, G. O., K. Hsu, and R. Radke (1995). 'Environmental Disclosures in Annual Reports and 10Ks: An Examination,' Accounting Horizons Vol. 9, No. 3, pp. 34-54.

Herremans, I., P. Akathaporn, and M. McInnes (1993). 'An Investigation of Corporated Social Responsibility Reputation and Economic Performance,' Accounting, Organizations, and Society, Vol. 18, pp. 587-604.

Hughes, II, K. E., and J. K. Reynolds (2001). 'Uncertainty Associated with Future Environmental Costs and the Market's Differential Response to Earnings Information,' Journal of Business Finance & Accounting, Vol. 28, No. 9, pp. 1351-1386.

Ingram, R., K. Frazier (1980). 'Environmental Performance and Corporate Disclosures,' Journal of Accounting Research, Vol. 18, No. 2, pp. 614-622.

Kreuze, J. G., G. E. Newell, and S. J. Newell (1996). 'What Companies are Reporting,' Management Accounting, Vol. 78, No. 1, pp. 37-43.

Li, Y. and B. McConomy (1999). 'An Empirical Examination of Factors Affecting the Timing of Environmental Accounting Standard Adoption and the Impact on Corporate Valuation,' Journal of Accounting, Auditing, and Finance, Vol. 14, No 3, pp. 279-313.

Nader, R. (1978). 'Corporate disclosure: The public right to know,' Journal of Contemporary Business, Vol. 7, No. 1, pp. 25-32.

Neu, D., H. Warsame, and K. Pedwell (1998). 'Managing Public Impressions: Environmental Disclosures in Annual Reports,' Accounting, Organizations, and Society, Vol. 23, No. 3, pp. 265-282.

Patten, D. M. (1991). 'Exposure, Legitimacy and Social Disclosure,' Journal of Accounting and Public Policy, Vol. 10, pp. 297-308.

Porter, M. and C. van der Linde (1995). 'Toward a New Conception of the Environment-Competitiveness Relationship,' Journal of Economic Perspectives, Vol. 9, No. 4, pp. 97-118.

Price Waterhouse. (1994). Progress on the environmental challenge: A survey of corporate America's environmental accounting and management. New York, Price Waterhouse.

Price Waterhouse. (1992). Accounting for environmental compliance: Crossroad of GAAP, Engineering, and Government. New York, Price Waterhouse.

Rockness, J. W. (1985). 'An Assessment of the Relationship Between US Corporate Environmental Performance and Disclosure,' Journal of Business, Finance & Accounting, Vol. 12, No. 3, pp. 339-354.

Rockness, J., P. Schlachter, and H. Rockness (1986). 'Hazardous Waste Disposal, Corporate Disclosure, and Financial Performance in the Chemical Industry,' Advances in Public Interest Accounting, Vol. 1, pp. 167-191.

Securities and Exchange Commission. (1993). Staff Accounting Bulletin No. 92. Federal Register (June 14, 1993).

Securities and Exchange Commission. (2000). Regulation S-K. Federal Register (April 11, 2000).

Stanny, E. (1998). 'Effect of Regulation on Changes in Disclosure of Reserved Amounts for Environmental Liability,' Journal of Financial Statement Analysis, Vol. 3, No. 4, pp. 34-49.

Walden, W. D., and B. N. Schwartz (1997). 'Environmental Disclosures and Public Policy Pressure,' Journal of Accounting and Public Policy, Vol. 16, No. 2, pp. 125-154.

Watts, R. L., and J. L. Zimmerman (1978). 'Towards a Positive Theory of the Determination of Accounting Standards,' The Accounting Review, Vol. 58, No. 1, pp. 112-134.

White, H. (1980). 'A Heteroskedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroskedasticity,' Econometrica, Vol. 48, No. 4, pp. 817-838.

Wiseman, J. (1982). 'An Evaluation of Environmental Disclosures Made in Corporate Annual Reports,' Accounting, Organizations and Society, Vol. 7, No. 1, pp. 53-63.

Carol A. Cox, Middle Tennessee State University

Edward B. Douthett, Jr., George Mason University

Table 1: Components of the Environmental Disclosure Index (ENVGAAP

Item Source GAAP Disclosures
#

1 Reg. S-K A general description of the business and specific
 Item 101 disclosure of the effects that compliance
 with environmental laws, when material.

2 Reg. S-K Estimated amount disclosed for capital expenditures
 Item 101 representing current and succeeding fiscal years
 in which those expenditures may be material.

3 Reg. S-K Disclosure of pending or contemplated
 Item 103 administrative or judicial proceedings.

4 Reg. S-K Disclosure of material events and uncertainties
 Item 103 known to management that would cause reported
 financial information to be unrepresentative of
 future operating results.

5 SFAS5 Nature of accrual.

6 SFAS5 Accrued amount

7 SFAS5 Nature of loss contingency

8 SFAS5 Estimate of additional possible loss or range.

9 SAFS5 Statement that estimate cannot be made.

10 SAFS5 Nature of probable unasserted claims that are
 possibly unfavorable.

11 SAB92 Whether an asset is recorded for probable recovery.

12 SAB92 Whether the accrual is undiscounted.

13 SAB92 The discount rate used.

14 SAB92 Expected payments for each of 5 succeeding years

15 SAB91 Reconciliation of the undiscounted to recognized
 amounts

16 ASB92 Material changes in expectation explained

17 SAB92 precision of loss estimates

18 SAB92 Extent to which unasserted claims are reflected in
 any accrual or may affect th magnitude of the
 contingency

19 SAB92 Uncertainties with respect to joint and several
 liability

20 SAB92 Nature and terms of cost sharing arrangement with
 other PRPs

21 SAB92 Uncertainties with respect to insurance claims

22 SAB92 The extent to which disclosed but unrecognized
 contingent losses are expected to be recoverable
 through insurance, etc.

23 SAB92 Uncertainties about the legal sufficiency of
 insurance claims or solvency of insurance carriers.

24 SAB92 The time frame over which accrued or unrecognized
 amounts may be paid out

25 SAB92 Material components of accruals and significant
 assumptions.

26 SAB92 Recurring costs associated with managing hazardous
 substances and pollution in ongoing operations

27 SAB92 Mandated expenditures to remediate previously
 contaminated sites.

28 SAB92 Other infrequent or nonrecurring cleanup
 expenditures, anticipated but not required in the
 present circumstances.

29 SAB92 Loss disclosure with respect to particular
 environmental sites that are individually material.

Table 2: Sample Selection Procedures

Selection criteria No. of
 Firms

1997 Fortune 500 firms named as PRP on SPIS database 245
Firms eliminated:
 Financial and Business Services (23)
 Firms named to a site with no ROD issued (27)
 Firms not publicly traded during entire sample period (11)
 Firms added as PRP during the sample period (13)
Final sample of firms for analyses 171
No. of Firm-Year Observations
Total firm-year observations possible (171 firms x 7 years) 1197
 Observations with missing data (10)
Final sample of firm-year observations 1187

Table 3: Industry Distribution of Sample Observations

Industry No. Of % of Sample
 Observations

Oil 79 6.7
Chemicals 139 11.7
Power 76 6.4
Paper 63 5.3
Steel 70 5.9
Food 80 6.7
Wood 28 2.4
Printing 28 2.4

Plastic, Glass & Cementt 42 3.5
Machinery 328 27.6
Transportation 70 5.9
Wholesale 21 1.8
Retail 69 5.8
All Other (observations [less 94 7.9
 than or equal to] 10 per industry
Total Observations 1187 100

Table 4
Descriptive Statistics for 1,187 observations from 1991 to 1997

Variable Mean (1) Std Dev Median

ENVGAAP (proportion) 0.35 0.23 0.34
ENVGAAPx2 (proportion) 0.38 0.32 0.33
RETURN (percentage) -10.3 66.7 -3.5
RAWRETURN (percentage) 19.4 30.6 16.6
SIZE ($ millions) 14,208 30,069 6100
ROA (percentage) 4.6 5.6 4.5
ROS (percentage) 4.7 7.6 4.3
UE (proportion) 0.04 3.57 0.01
MKBK (ratio) 2.9 6.2 2.4
CAPX ($ millions) 947 1,966 398
ENVLIAB ($ millions) 17.1 28.6 6.0
SITES (number of PRP 17.2 19.3 10.0
 sites per firm)
POSTSAB92 (1:0 indicator 0.73 0.44 1.0
 for SAB 92 rules or not)
CONFIRM (1:0 indicator for 0.26 0.44 0.0
 confirmatory disclosure
 or not)

[1.sub.t]-statistics for all means are significant at p [less than
or equal to] 0.01, except for UE (p = 0.70).

Variable definitions:

ENVGAAP = a count of environmental GAAP disclosures actually made
as a proportion of total possible (source: 10k).

ENVGAAPx2 = same as ENVGAAP except it excludes disclosure item
101 under regulation S-K (which disclosure #2 listed in Table 1).

RETURN = an industry-adjusted annual return calculated as the 12-month
change in stock price ending 3 months after the fiscal year end
(adjusted for dividends), scaled by the stock price at the beginning
of this 12-month window (source: Compustat).

RAWRETURN = same as RETURN above except not adjusted for industry mean
return.

SIZE = total assets (source: Compustat).

ROA = income before extraordinary items divided by total assets
(source: Compustat).

ROS = net income divided by total sales revenue. (source: Compustat).

UE = annual change earnings-per-share divided by beginning stock
price (source: Compustat).

MKBK = market value equity divided by book value equity (source:
Compustat).

CAPX = total capital expenditures for the period (source: Compustat).

ENVLIAB = estimated average environmental liability per PRP. (Source:
Environmental Protection Agency's Records of Decision).

SITES = number of superfund sites associated with a PRP. (Source:
provided by a representative at the Environmental Protection Agency.

POSTSAB92 = 1 for SAB environmental disclosures made after 1992,
and 0 otherwise.

CONFIRM = 1 CONFIRM is constructed by examining the coding of item
number 2 in our disclosure index (see Item 2 in Table 1). The coding
of item 2 indicates whether a disclosure for estimated environmental
capital expenditures was made under SEC Regulation S-K (Item 101).
We code CONFIRM equal to one when the firm makes a disclosure about
future environmental capital expenditures and the firm's ROA is less
than the mean of the sample, 0 otherwise.

Table 7
Environmental GAAP Disclosure Model Regression Results.
The Dependent Variable is a disclosure index (ENVGAAP).

(1) Variables (2) Predicted Sign (3) Coefficients

 N = 1187
 0.247 (c)
INTERCEPT
 (5.82)
 0.034 (c)
ln(SIZE) -2
 (6.83)
 0.159 (c)
CHEM 0
 (10.40)
 0.197 (c)
OIL 0
 (10.45)
 0.130 (c)
PAPER 0
 (6.21)
 0.204 (c)
STEEL 0
 (10.34)
 0.260 (c)
POWER 0
 (13.66)
 0.021 (c)
ENVLIAB 0
 (8.47)
 0.003 (c)
SITES 0
 (8.83)
 -0.204 (c)
POSTSAB92 -
 (-19.69)
 0.053 (c)
CAPX 0
 (4.82)
 -0.004 (c)
ROA -/+
 (-4.79)
Adjusted [R.sup.2] 0.55
F-Value 132.09

The total number of observations used is 1,187 and the sample period
is 1991 through 1997. Variables defined in Table 4. Superscripts a, b,
and c, indicate one-tail statistical significance at
0.10, 0.05, and 0.01 levels, respectively.

Table 8
Environmental GAAP Disclosure Model Regression Results.
The Dependent Variable is a disclosure index (ENVGAAPx2).

(1) Variables (2) Predicted Sign (3) Coefficients

 n = 1187
 0.352 (c)
INTERCEPT
 (6.30)
 0.033 (c)
ln(SIZE) -2
 (5.08)
 0.148 (c)
CHEM 0
 (7.08)
 0.217 (c)
OIL 0
 (8.54)
 0.091 (c)
PAPER 0
 (3.24)
 0.211 (c)
STEEL 0
 (7.48)
 0.255 (c)
POWER 0
 (9.60)
 0.018 (c)
ENVLIAB 0
 (5.56)
 0.003 (c)
SITES 0
 (8.03)
 -0.366 (c)
POSTSAB92 -
 (-26.80)
 0.047 (c)
CAPX 0
 (3.18)
 -0.003 (c)
ROA -/+
 (-2.52)
 0.103 (c)
CONFIRM 0
 (5.38)
 0.007 (a)
ROA*CONFIRM 0
 (1.49)
Adjusted [R.sup.2] 0.58
F-Value 125.93

The total number of observations used is 1,187 and the sample period
is 1991 through 1997. Variables defined in Table 4. Superscripts (a),
(b),
and (c), indicate one-tail statistical significance at 0.10, 0.05, and
0.01 levels, respectively. (1) CONFIRM is interacted with ROA to
estimate the differential coefficient on ROA when the disclosures are
of a confirmatory nature. When CONFIRM=1, the observation is
considered to be of a confirmatory nature, non-confirmatory otherwise.
CONFIRM is assigned a value of 1 when a firm makes a disclosure about
future environmental expenditures under Regulation S-K, Item 101, and
the firm has an ROA less than the mean of the sample.

Table 9
Market Model Regression Results.
The Dependent Variable is Industry-Adjusted Market Returns (RETURN).

(1) Variables (2) Predicted Sign (3) Coefficients

 N = 1187
 -0.154 (c)
INTERCEPT
 (-3.00)
MKBK 0.005 (c)
 -2
 -2.56
 -0.001
CHEM -/+
 (-0.03)
 -0.112 (c)
OIL -/+
 (-2.80)
 0.041
PAPER -/+
 -1.06
 0.019
STEEL -/+
 -0.5
 0.080 (b)
POWER -/+
 -2.11
 -0.006 (a)
ENVLIAB -
 (-1.38)
 0.056 (c)
POSTSAB92 -/+
 -2.47
ROS 0.497 (c)
 0
 -3.07
UE 0.168 (c)
 0
 -2.61
 0.046b
CAPX 0
 -2.1
 -0.074a
ENVGAAP -/+
 (-1.43)
Adjusted [R.sub.2] 0.076
F-Value 8.2

The total number of observations used is 1,187 and the sample period
is 1991 through 1997. Variables defined in Table 4. Superscripts (a),
(b), and (c), indicate one-tail statistical significance at 0.10,
0.05, and 0.01 levels, respectively.

Table 10
Market Model Regression Results.
The Dependent Variable is Industry-Adjusted Market Returns (RETURN).

(1) Variables (2) Predicted Sign (3) Coefficients

 N = 1187
 -0.147 (c)
INTERCEPT
 (-2.91)
MKBK 0.005 (c)
 -2
 -2.55
 0.001
CHEM -/+
 -0.03
 -0.111 (c)
OIL -/+
 (-2.79)
 0.053a
PAPER -/+
 -1.37
 0.007
STEEL -/+
 -0.19
 0.077b
POWER -/+
 -1.98
 -0.007 (a)
ENVLIAB -
 (-1.63)
 0.047 (b)
POSTSAB92 -/+
 -1.88
ROS 0.530 (c)
 0
 -3.17
UE 0.161 (c)
 0
 -2.49
 0.045 (b)
CAPX 0
 -2.03
ENVGAAPx2 -0.108 (b)
 -/+
 (-2.25)
 -0.032
CONFIRM 0
 (-0.81)
 0.094 (a)
ENVGAAPx2*CONFIRM 0
 -1.44
Adjusted R2 0.077
F-Value 7.25

The total number of observations used is 1,187 and the sample period
is 1991 through 1997. Variables defined in Table 4. Superscripts (a),
(b), and (c), indicate one-tail statistical significance at 0.10,
0.05, and 0.01 levels, respectively. 1CONFIRM is interacted with
ENVGAAPx2 to estimate the differential coefficient on ENVGAAPx2 when
the disclosures are of a confirmatory nature.