Environmental liabilities and stock price responses to FASB Interpretation No. 47.

Hunsader, Kenneth J. ; Dickens, Ross N.

INTRODUCTION

The Financial Accounting Standards Board (FASB) announced Financial Interpretation Number 47, Accounting for Conditional Asset Retirement Obligations, on March 30, 2005. The interpretation (FIN47) seeks more consistent recognition of liabilities relating to asset retirement obligations (AROs) and a standardization of information concerning carrying amounts of assets based on additional estimable retirement costs. Although FIN47 could affect all companies with future conditional obligations, it requires firms with environmental liabilities to alter the companies' reporting of such issues. We use this opportunity to judge the impact on stock prices for companies with likely environmental concerns.

Prior to 2005, companies with environmentally contaminated properties could sidestep accounting for potential liabilities through "mothballing" the property. Mothballing occurs by letting the property sit idle and not offering it for sale and/or avoiding investigating possible remediation needs. This strategy would delay any cleanup costs and also keep investors unaware of the scope of potential liabilities. FIN47 requires reporting such potential liabilities and, thus, leads companies toward greater transparency.

Managers might opt for less transparency to hide negative information and, thus, increase their ability to beat earnings benchmarks and/or deliver smoother earnings (Degeorge & Zeckhauser, 2000) while improved transparency should allow a reduction in asymmetric information, increased liquidity, and a lower cost of capital (Botosan, 1997). Thus, FIN47 could lead to negative stock price reactions if managers have been hiding negative information which now must be reported or the announcement could have a positive impact on share prices if added transparency provides the benefits Botosan (1997) notes. Given transparency failures (such as Enron and WorldCom), regulators have acted to increase corporate disclosure--with the 2002 Sarbanes-Oxley Act's being the most far-reaching example, but FIN47 is also a part.

The purpose of FIN47 is to create a fair market for all participants by providing stakeholders--such as community members, taxpayers, various government entities, and investors--with as much ARO information as possible. Community members may be most concerned about possible health implications of "toxic" sites. Taxpayers and governments worry that companies may discharge their liabilities via bankruptcy filings; leaving the public to bear the costs and risks of reclamation (Habegger, 2005). (Note that in a 2003 Accountability Office (GAO) report, the Environmental Protection Agency (EPA) is partially or wholly funding 60 of the largest 142 Superfund (toxic) sites with each site having an estimated cost of $140 million or more. A specific example of firms not bearing full cleanup costs is Asarco which filed bankruptcy in 2005 with $500 million to $1 billion in environmental liabilities for which its parent company, Grupo Mexico, set up an environmental trust fund with only $100 million to help pay cleanup costs.) Investors would want the ARO information accurately reflected in the companies' market prices.

The purpose of this study is to examine the stock market response to FASB's FIN47 announcement for companies in industries which are the most likely to have potential environmental liability possibilities: mining and manufacturing (especially the subset of chemical firms). In general, we find significant negative wealth effects for manufacturing firms. Variations in abnormal returns appear related to company-specific factors evaluated under the Environmental Protection Agency's (EPA) measures of financial distress. We also find that companies identified as top corporate polluters have significant shifts in systematic risk after the announcement.

The remainder of the paper is as follows: we provide historical background of environmental liability reporting, review related literature review, discusses our data set and methodology, and then presents the results. Finally, we provide a summary and concluding remarks.

HISTORICAL BACKGROUND

Prior to 2001, FASB Statement No. 5 "Accounting for Contingencies" (Statement 5), FASB Interpretation No. 14 (FIN14), and Statement of Position (SOP) 96-1 issued by the American Institute of Certified Public Accountants (AICPA) guided companies' handling of possible environmental liabilities. Statement 5 set forth a "two-prong" approach in which a company should recognize a liability when: 1) it is probable that the company has incurred a liability and 2) the company can reasonably estimate the amount. In response, companies addressed the first prong by following a defining approach as if the event were "more likely than not to occur". FIN14 is FASB's effort to provide an estimation approach for the second prong. Under FIN14, loss contingencies could be stated at their "most likely value". If a firm could not determine this value, the company could provide a range for a loss contingency and use the lowest amount. Finally, SOP 96-1 provided added guidance relating to environmental cleanup obligations. Rogers (2008) states that since these three items "tended to favor certainty over projections, they have been criticized for delaying recognition of contingent liabilities, understating recognized liabilities, and failing to provide users of financial statements with useful, transparent, and timely information".

As an attempt to address the above issues and provide uniformity in evaluation processes, the FASB developed FAS 143, "Accounting for Asset Retirement Obligations (AROs)" in 2001. FAS 143 requires that the liabilities for existing legal obligations be recognized when incurred-which is typically when the asset is acquired or developed through construction. This recognition assumes the company can assess the liability's fair value where the best definition for "fair value" is the "transfer" price between market participants although fair value may be set by using the best information available at the time. Soon, the FASB became concerned about accounting practice differences for recognizing conditional asset retirement obligations (CAROs)--AROs which are conditioned on a future event such as selling a currently operating production facility. In some cases, companies claimed they could not estimate the fair value given uncertainty while others claimed no legal liability since the obligation could be indefinitely deferred (mothballed).

The FASB issued FIN47 on March 30, 2005 to clarify how companies should apply FAS 143 regarding CAROs. The new interpretation states that a firm should recognize the CARO when incurred which includes acquisition, construction, or development of the asset. Also, the firm should incorporate any uncertainty regarding the timing or structure of the settlement of an obligation into the calculation of the liability's value. With regards to environmental liabilities, a company must report future environmental cleanup obligations on its balance sheet even if there are no plans to end production or sell the asset. If an ARO is not reasonably estimable through an active market for transferring the asset, applying a present value technique, or through an acquisition price to determine the value, FIN47 requires a company to disclose that the liability has not been recognized along with an explanation supporting the reasons why.

LITERATURE REVIEW

The passage of the Sarbanes-Oxley Act (SOX) of 2002 brought attention to the impact that environmental liabilities may have on firms given that CEO's of public companies must certify that the financial statements fairly represent the firm's financial position. Schnapf (2006) states that some firms hired environmental consultants to get actual cleanup cost data. The passage of FIN47 will likely accelerate this trend.

Regulation requiring increased disclosure and its effect on firm value has drawn attention in recent years. Recent examples in the literature concern the Sarbanes-Oxley Act (SOX) of 2002. Zhang (2007) finds that U.S. firms experience a negative wealth effect around key SOX event dates. Wintoki (2007) reports that negative wealth effects are more pervasive for firms with higher growth opportunities and greater operating environment uncertainty. Thus, the passage of increased reporting requirements seems a greater cost to growth firms (which tend to be younger, smaller in size, and have more fluid operating environments) relative to low-growth firms (which are usually older and operate in a more stable environment).

In a separate study not directly related to SOX, Cox & Douthett (2009) find that confirmatory environmental disclosures reduce negative wealth effects relative to non-confirmatory disclosures. A confirmatory disclosure is one which indicates a firm's joint strategy to act environmentally responsibly while pursuing higher financial performance. A non-confirmatory disclosure does not indicate a simultaneous pursuit of both goals. Thus, the authors conclude that investors' perception of a combined strategy affects market valuation.

Lee & Hutchison (2005) provide a survey of research regarding company characteristics related to the decision to disclose environmental information. Firm size is a significant factor. Hackston & Milne (1996), Cormier & Magnan (2003), Patten (1991), Adams, Hill, & Roberts (1998) all find a significant positive relationship between firm size and environmental reporting. However, Cowen & Parker (1987) provide conflicting evidence as they report a negative relationship in the U.S. based on size.

Industry affiliation is influential as well. Trotman & Bradley (1981), Cowen & Parker (1987), and Patten (1991) note that the more sensitive an industry is to the environment; the greater the level of disclosures. Thus, firms in an industry segment such as chemicals, would have more disclosures relative to a general manufacturing firm.

Firm systematic risk is an added explanatory variable. Trotman & Bradley (1981) find that the higher a firm's systematic risk (as measured by beta), the greater the likelihood of social transparency. Similarly, Cormier & Magnan (2003) find risk positively associated with transparency while leverage is inversely related with environmental disclosures in annual reports. The relationship between profitability and environmental disclosure is uncertain. Buhr (2002) finds that profitability is an important variable in reporting environmental issues for pulp and paper industry firms. In addition, Cox & Douthett (2009) find the level of environmental GAAP disclosure is related to firm profitability in terms of return on assets. However, Patten (1991) and Hackston & Milne (1996) find no relationship with regard to profitability.

A final potential area related to financial disclosure in regards to environmental liabilities is financial tests. Habegger (2005) reports that any firm applying for a permit for an environmentally hazardous project must demonstrate financial assurance to the issuing state's agency. This assurance that the company has the ability to fund all costs associated with environmental liabilities is to protect taxpayers as well as the environment. To obtain a state permit, firms must pass one of two alternative EPA financial tests or obtain external assurance such as insurance or letters of credit.

The EPA's first test requires a firm to meet four conditions. 1) The firm's finances must meet at least two of three ratio tests: a) total liabilities to net worth less than 2.0, b) the sum of net income plus depreciation, depletion, and amortization to total liabilities greater than 0.10, and/or c) current assets to current liabilities greater than 1.5. 2) The firm must have tangible net worth of at least S10 million. 3) Tangible net worth and net working capital must each be at least six times the current closure cost estimate for all the company's facilities. 4) Assets located in the U.S. must amount to at least 90 percent of the firm's total assets or at least six times the current closure cost estimate for the total of all facilities.

The EPA's second test requires the company to meet the second, third, and fourth requirements as the first test, but with a current bond rating of BBB (Baa) or greater by Standard and Poor's (Moody's). These firm-specific financial conditions may affect the stock market response to the passage of FIN47. (The mining industry is subjected to somewhat different financial assurance standards given alternative regulatory agencies. For example, coal mining reclamation is assured by the Department of Interior's Office of Surface Mining Reclamation. Coal mining companies must have a current ratio of greater than 1.2 times and ratio of liabilities to net worth of 2.5 times or less. Oil companies with offshore facilities guaranteeing the ability to clean spills are assured through the Department of Interior's Mineral Management Service. Corporate guarantees of financial assurance are not accepted for onshore oil and gas reclamation. We use EPA guidelines for all data screening processes.) On the one hand, firms with values that currently fail EPA standards could be considered to be in even worse positions if forced to report added CAROs. However, firms which currently pass EPA tests could arguably be in for a more negative reaction. If reporting CAROs would make a firm which currently meets requirements to no longer pass, then that newly failing firm would either lose the EPA's imprimatur or have to utilize some other assurance mechanism such as insurance. Such impacts would be, at best, costly and likely lead to negative stock price reactions.

DATA AND METHODOLOGY

The sample focuses on the mining and manufacturing industries given their likely environmental challenges and includes 1,716 firms with stock price data available from the Center of Research in Security Prices (CRSP) and accounting data for the year 2004 on Research Insight (Compustat). Of the firms included, 121 are in the mining industry (two-digit SIC codes from 10 to 14), and 1,595 firms are from the manufacturing industry (two-digit SIC codes from 20 to 39). Table 1 reports the summary statistics of the sample. (Appendix I provides a breakdown of the number of companies by industry.)

The median total debt-to-equity ratio (DE) is 0.65 while the mean is 1.37 which both meet the EPA's guideline of less than 2.0. (We emphasize median values as that informs us as to the 50th percentile company's ability to meet EPA standards.) The median value for net income plus depreciation divided by total liabilities (NIDEP/TL) is 0.16 which is above the EPA's minimum guideline of 0.10. The median current ratio (CR) is 2.57--well above the minimum EPA level of 1.5. The median market value (MV) is S345 million which is well over the S10 million floor set by the EPA. Thus, the median value comparisons show that more than 50% of all sample companies for each variable meet the EPA requirements.

As our measure of tangible net worth (TNW), we use total assets minus total liabilities minus intangibles from Research Insight. (Research Insight's intangibles variable includes 21 items such as copyrights, goodwill, and patents.) The median value is $69.74 million. The median value of net working capital (NWC) is $65.79 million. The third EPA condition requires these two values to be six or more times the current closure cost estimate. As in Habegger (2005), we estimate total current closure cost to be 1% of the net plant, property and equipment (PPE) account. The median value of PPE is $37.47 million, thus 1% of that amount is $0.37 million. Dividing the median values for TNW or PPE by the median closure cost estimates yields values of 175 times or more (which are well over the six times requirements of the EPA).

As most firms do not have actively traded bonds, we employ Altman's bankruptcy prediction test (Z-score) as a proxy for default risk ratings. We take Z-score values from Research Insight and find a median score of 3.72. However, 328 firms fall below a score of 1.81 which indicates a high probability of bankruptcy. Overall, based on median values, the average firm in the sample would seem to have little trouble meeting the minimum guidelines of the EPA's tests for financial assurance, but some firms would not meet requirements in all areas.

Table 1 also reports the summary statistics for the manufacturing (Panel B) and mining (Panel C) industries separately. No test for statistically different mean values between the two subsamples is significant (mostly due to the relatively large values for standard deviations).

Given the common event period (the date of FIN47's enactment), we employ a multivariate regression model (MVRM) as suggested by Binder (1985a and 1985b) and Schipper & Thompson (1983) to correct for possible heteroskedasticity biases. (Under standard event study methodology, a common event period means individual asset returns will be contemporaneously correlated such that residuals across the various firm type portfolios would not be identically and independently distributed.) Following Bhargava & Fraser (1998), we employ a system of Seemingly Unrelated Regressions (SUR) and include a time lag variable to control for non-synchronous trading. The event date, [t.sub.0], is the announcement date of the passage of FIN47 (March 30, 2005), and the model specification is:

[r.sub.it] = [[alpha].sub.i] + [[alpha]'.sub.i][D.sub.t] + [[beta].sub.li][r.sub.m(t-1)] + [[gamma].sub.i][D.sub.o] + [[beta]'.sub.i][D.sub.t][r.sub.mt] + [[beta]'.sub.li][D.sub.t][r.sub.m(t-1)] + [[epsilon].sub.it]

where [r.sub.it] = the return for portfolio i on day t,

[[alpha].sub.i] = intercept coefficient for portfolio i,

[D.sub.t] = dummy which is 1.0 after the last event date; else 0.0 (= 1.0 for day +2 to day +120),

[[alpha]'.sub.i] = shift intercept coefficient for portfolio i,

[[beta].sub.i] = systematic risk coefficient on market return for portfolio i,

[r.sub.mt] = the return on the equally weighted market portfolio on day t,

[[beta].sub.li] = systematic risk coefficient on the lagged market return for portfolio i,

[[gamma].sub.i] = the wealth effect of the announcement on portfolio i for the event,

[D.sub.o] = dummy which is 1.0 in the event window; else 0.0 (= 1.0 for day -1 to day +1),

[[beta]'.sub.i] = shift in systematic risk for portfolio i,

[[beta]'.sub.li] = shift in systematic risk for portfolio i on the lagged return, and

[[epsilon].sub.it] = error term.

To compute abnormal returns, we estimate the model parameters using 120 trading days before and after the event date in the fashion of Saunders & Smirlock (1987). We calculate cumulative abnormal returns (CARs) by adding the abnormal returns for a given portfolio, across the event window (days -1 through day +1). Next, we separate the portfolios based on whether a firm meets a given EPA standard and examine CARs for firms which pass the criterion as opposed to those which fail. For each of these seven model estimation groupings, we utilize the full sample as well as subsets based on whether companies are in the mining, manufacturing (including chemical), or just chemical industries.

We predict a negative sign for debt to equity (DE) since firms with higher DE values will likely be closer to financial distress. Higher DE firms would likely be harmed more by a requirement to disclose environmental liabilities. Even firms with DE below, but near, 2.0 could have a negative reaction to FIN47 as a new disclosure could lead to failing test values. Negative reactions to FIN47 based on DE values could also come via debt covenants. Firms near debt covenant limits would have less financial flexibility and could face costly restructuring. (We would like to thank an anonymous referee for pointing out the possible affect on debt covenants.)

However, the relationship is not entirely clear. The higher DE, the more often a firm will likely need to tap the banking sector or capital market. Each round of financing brings scrutiny by analysts such that DE can be used as a proxy for firm transparency (Almazan, Suarez, & Titman, 2004). If so, higher DE firms may not see an impact from FIN47 as it is possible no new financial information will be forthcoming given prior scrutiny (Jensen, 1986). In all, a negative relationship between DE and CARs seems likely, but there may be no significant difference between firms under or over the EPA's 2.0 guideline given the transparency impact.

The EPA requires NIDEP/TL (which is, simply stated, a cash flow to liabilities measure) to be greater than 0.10. We expect NIDEP/TL to be positively related to abnormal returns since firms with less cash flow relative to their liabilities should be hurt more if the companies must report new CAROs. However, it is quite possible that firms with high NIDEP/TL may not be impacted greatly if CAROs are a relatively small amount.

The EPA test requires CR to be greater than 1.5. We expect a positive relationship between CR and FIN47's announcement based on the same idea as above that a firm in a worse financial position is likely to be harmed more by additional CARO reporting requirements as such revelations would likely be less of an impact for firms on stronger financial footing. This same general pattern should hold for TNW (tangible net worth), TNWCC (tangible net worth divided by the estimate of closing costs), and NWCCC (net working capital divided by the estimate of closing costs). For Z-scores, we expect a negative relationship to CARs given that a lower Z-score is a predictor of an earlier bankruptcy (or more current problems given higher bankruptcy risks). For each of the above variables, we divide the companies into two groups based on if the firm passes or fails the given EPA requirement.

To take advantage of the SUR specification, we estimate Equation (1) several times based on changing characteristics. First, we estimate the model using the whole sample by differentiating between mining and manufacturing firms. Then, we estimate the model for the different industries by separating the sample into whether each firm passes or fails the individual EPA financial assurance standards. Finally, we combine the six separate EPA-inspired measures into an all-in variable which separates companies which pass or fail the EPA standards. (We calculate the overall EPA pass/fail variable using the EPA's first test with conditions 1, 2, and 3 as stated above. We do not use condition 4 as it requires the percentage of each firm's assets located in the U.S. which is beyond our data source.) Equation (1) enables us to test if the independent variables are significantly different than zero as well as to test across portfolios for differences in the variables such as the wealth effect (Ho: [[gamma].sub.1i] = [[gamma].sub.2i]), and differences in systematic risk (Ho: [[beta]'.sub.1i] = [[beta]'.sub.2i]).

RESULTS

Table 2 reports the results from estimating Equation (1). Both regressions are highly significant with the systematic risk coefficient on market return, [[beta].sub.i], being the main contributing explanatory variable. The estimated equation explains 37.7% (85.4%) of the variation in stock returns for the mining (manufacturing) firms based on adjusted [R.sup.2] values. The market beta for the mining (manufacturing) firms is significantly positive, as expected, with an estimated coefficient of 1.48 (1.18). Thus, mining firms have nominally higher systematic risk. The results also show a significant shift in the systematic risk for the mining subset.

Of most interest, however are the tests for wealth impacts. Tests based on [[gamma].sub.i]s show no significant change in wealth based on the FIN47 announcement. There also is no difference in the [[gamma].sub.i] values for mining versus manufacturing firms. Given the discussion above, we believe it is possible that separating firms based on financial characteristics may reveal underlying issues.

Table 3 compares the CARs based on the pass/fail values for various EPA test variables. Our general expectation is that firms with variable values showing the firm less able to pass will have lower CARs. Using the whole sample, we do find firms with DE > 2.0, NIDEP/TL < 0.10, and Z < 1.81--which means the firms fail those tests--have CARs significantly less than 0.0. However, we also find firms with TNW > $10 million, TNWCC > 6.0, and NWCCC > 6.0--which means the firms passed those tests--have CARs significantly less than 0.0. However, only in the case of comparing the CARs for TNW < $10 million as compared to firms in the TNW > $10 million categories do we find a significant difference between the two groups. Still, the test-passing group has the lower average CARs.

We believe a plausible explanation is that the test-failing firms' known financial bad news is such that investors do not believe the possibility of reporting CAROs will harm the firm significantly more. However, having to report CAROs might harm the test-passing firms. This condition would explain why better firms react with lower CARs.

To examine the impact from industry type, we repeat the above tests from the whole sample, but divide the companies into mining and manufacturing subsets. We, then, segment chemical firms from the manufacturing group given chemical firms greater potential environmental issues. In general, the most striking result for the mining and manufacturing subsets is that there are few statistically significant test results. However, the three significant test results are all in keeping with firms with the test-passing results having negative CARs (for TNW > $10 million for manufacturing) or the test-passing group having significantly lower CARs than the test-failing group (for NWCC for mining firms and TNW for manufacturing firms).

The desirability of separating the chemical industry shows from the eleven significant test results for that group. The DE and Z-score values again have the expected negative relationship to CARs, although there is no significant difference between the CARs for test-passing and test-failing firms for either variable. In general, we also continue to find results for NIDEP/TL, CR, TNW, TNWCC, and NWCCC that would support the idea that passing firms in danger of becoming failing firms if they must report CAROs, face likely higher costs. However, only for NIDEP/TL are the test-failing firms' CARs significantly different from the test-passing firms'. The possible explanation that the better firms have more to lose, thus, gets further support.

Table 4 reports the results from estimating Equation (1) when we sort the sample into firms which pass EPA standards as compared to those which do not pass. The results for systematic risk ([[beta].sub.i]) and the shift in the systematic risk for lagged returns for the mining subset are the same as those reported in Table 2. Of greater interest are the results for the tests on wealth impact ([[gamma].sub.i]) from the FIN47 announcement. The [[gamma].sub.i] coefficient estimates for firms passing EPA requirements are significantly negative both for the "all firms" set and the manufacturing firms subset. Also, F-test results show that the wealth effects are significantly worse for firms which pass EPA requirements than for firms which do not. These results are consistent with the explanation that passing firms are hurt more by the possibility of having to report CAROs than non-passing firms.

As a robustness check, we examine manufacturing firms known to have environmental problems and included on the "Toxic 100" list compiled by the University of Massachusetts Political Economy Research Institute (PERI) for which we have the needed stock and financial statement data. (The website for the PERI Toxic 100 list is: http://www.peri.umass.edu/ToxicIndex.430.0.html.) We believe it is quite likely that firms with known environmental issues will not suffer (or, at least, not suffer as much) from FIN47's passage. Table 5 reports the results from the 46 firms (21 passing and 25 not) in this data set. We again estimate Equation (1) based on whether the firms pass the EPA's test or not. We find no difference in the wealth effects ([[gamma].sub.i]) between those that pass or fail the EPA's test. Thus, for those firms which already seem to have well-publicized environmental problems, the FIN47 announcement does not have significant wealth impacts. However, there is a shift in systematic risk for EPA-pass firms and the increase in systematic risk is significantly different than for non-EPA-pass firms. Taking this result along with those in the earlier tables, we conclude that FIN47 is most likely to impact firms which financial tests support as being in sound shape, but for which investors fear unknown environmental issues as investors appear to already have discounted the value of firms with known environmental issues.

CONCLUSION

We examine the impact on mining and manufacturing firms' stock returns from the announcement of the FASB's FIN47 March 30, 2005. In general, we find marginally negative CARs, but with many returns insignificantly different from 0.0%. When examining CARs relative to financial variables utilized by the EPA, in general, we find firms with better financial variables have lower stock returns. Separating firms into mining and manufacturing firms shows little differences in wealth impacts. It is possible that our generally insignificant findings for mining firms could be related to the fact that their assurance process differs from the average manufacturing firm. That issue is an avenue for future research.

Comparing firms which passed EPA tests to firms which did not, we find passing firms generally had lower stock returns than the non-passing group. We interpret this result in regards to the financial issue of transparency. Our results seem to indicate that investors expect firms with known problems will not worsen in any significant way while seemingly stronger firms may now have to report environmental problems that had previously been undisclosed. Thus, the market's reaction to FIN47 supports the idea that investors consider some companies had not been fully disclosing potential environmental issues.

Comparing firms on a "Toxic 100" list provides added support to the above argument. We find no difference between stock returns for firms which do or do not pass the EPA's test requirements. Thus, the specter of having to improve financial transparency by reporting environment-related CAROs bring a wealth impact to relatively stronger firms, but more so for those firms with fewer existing environmental disclosures.

Appendix A: Industries

The following table provides the two-digit SIC code and the number
of firms from each industry that were included in the study.

SIC # Firms Industry Name

Mining

10 6 Metal Mining
12 6 Coal Mining
13 102 Oil and Gas Extraction
14 7 Mining and Quarrying of Nonmetallic
 Minerals, Except Fuels

Manufacturing

20 74 Food and Kindred Products
21 4 Tobacco Products
22 9 Textile Mill Products
23 34 Apparel and Other Finished Products
 Made from Fabrics, etc.
24 12 Lumber and Wood Products, Except
 Furniture
25 21 Furniture and Fixtures
26 30 Paper and Allied Products
27 47 Printing, Publishing, and Allied
 Industries
28 326 Chemicals and Allied Products
29 14 Petroleum Refining and Related
 Industries
30 35 Rubber and Miscellaneous Plastic
 Products
31 17 Leather and Leather Products
32 17 Stone, Clay, Glass, and Concrete
 Products
33 39 Primary Metal Industries
34 48 Fabricated Metal Products, Except
 Machinery and Transport Equipment
35 212 Industrial and Commercial Machinery
 and Computer Equipment
36 312 Electronic and Other Electrical
 Equipment and Components Except
 Computers
37 66 Transportation Equipment
38 248 Measuring, Analyzing and Controlling
 Instruments
39 30 Miscellaneous Manufacturing Industries

REFERENCES

Adams, D., W. Hill, & C.B. Roberts (1998). Corporate social reporting practices in western Europe: Legitimating corporate behaviour?, The British Accounting Review 30(1), 1-21.

Almazan, A., J. Suarez, & S. Titman (2004). Stakeholder, transparency and capital structure. CEPR Discussion Paper No. 4181. Available at SSRN: http://ssrn.com/abstract=491143

Altman, E. (1968). Financial ratios, discriminant analysis and the prediction of corporate bankruptcy, The Journal of Finance, 23, 589-609.

Altman, E, R. Haldeman, & P. Narayanan (1977). Zeta analysis: A new model to identify bankruptcy risk of corporations, Journal of Banking and Finance, 1, 29-54.

Altman, E. (1993). Corporate financial distress and bankruptcy: A complete guide to predicting and avoiding distress and profiting from bankruptcy (Second edition) New York: John Wiley & Sons, Inc.

Bhargava, R. & D.R. Fraser (1998). On the Wealth and Risk Effects of Commercial Bank Expansion into Securities Underwriting: An Analysis of Section 20 Subsidiaries, Journal of Banking and Finance, 22(4), 447-465.

Binder, J.J. (1985a). Measuring the Effects of Regulation with Stock Price Data, Rand Journal of Economics, 16(1), 167-183.

Binder, J.J. (1985b). On the Use of the Multivariate Regression Model in Event Studies, Journal of Accounting Research, 23(2), 370-383.

Botosan, C. (1997). Disclosure level and the cost of equity capital, The Accounting Review, 72, 323-349.

Buhr, N. (2002). A structuration view on the initiation of environmental reports, Critical Perspectives on Accounting, 13(1), 17-38.

Cormier, D. & M. Magnan (1999). Corporate environmental disclosure strategies: Determinants, costs and benefits, Journal of Accounting, Auditing & Finance, 14(4), 429-451.

Cowen, S., L. Ferreri, & L. Parker (1987). The impact of corporate characteristics on social responsibility disclosure: A typology and frequency based analysis, Accounting, Organizations and Society, 12(2), 111-122.

Cox, C.A. & E.B. Douthett, Jr. (2009). Further Evidence on the Factors and Valuation Associated with the Level of Environmental Liability Disclosures, Academy of Accounting and Financial Studies Journal, 13(3), 1-26.

Degeorge, F., J. Patel, & R. Zeckhauser (2000). Earnings management to exceed thresholds. Journal of Business, 72, 1-33.

Financial Accounting Standards Board (FASB) (2005). Accounting for conditional asset retirement obligations. Interpretation no. 47. Financial Accounting Standards Board, Stanford, CT.

Habegger, W. (2005). An investigation offinancial assurance mechanisms for environmental liabilities. Dissertation, Florida State University.

Hackston, D. & M. Milne (1996). Some determinants of social and environmental disclosures in New Zealand companies, Accounting, Auditing & Accountability Journal, 9(1), 77-108.

Jensen, M. (1986). Agency costs of free cash flow, corporate finance and takeovers, American Economic Review, May, 323-329.

Lee, T., & P. Hutchison (2005). The decision to disclose environmental information: a research review and agenda, Advances in Accounting, 21, 83-111.

Patten D. (1991). Exposure, legitimacy, and social disclosure, Journal of Accounting and Public Policy, 10(4), 297-308.

Rogers, C. (2008). Environmentally insolvent: fair value measurement of environmental liabilities poses solvency risk, Business Law Today, 17(6) 41-46.

Saunders, A. & M. Smirlock (1987). Intra- and interindustry effects of bank securities market activities: the case of discount brokerage, Journal of Financial and Quantitative Analysis, 22, 467-482.

Schipper, K. & R. Thompson (1983). The Impact of Merger-related Regulations on the Shareholders of Acquiring Firms, Journal of Accounting Research, 21(1), 184-221.

Schnapf, L. (2006). New accounting standard will have far-reaching consequences for environmental disclosure, ABA Environmental and Energy Business Law e-newsletter.

Trotman, K. & G. Bradley (1981). Associations between social responsibility disclosure and characteristics of companies, Accounting, Organizations and Society, 6(4), 355-362.

United States Government Accountability Office (2005). Environmental liabilities: EPA should do more to ensure that liable parties meet their cleanup obligations, Report to congressional requesters, August.

Wintoki, M. (2007). Corporate boards and regulation: the effect of the Sarbanes-Oxley Act and the exchange listing requirements on firm value, Journal of Corporate Finance, 13, 229-250.

Zhang, I. (2007). Economic consequences of the Sarbanes-Oxley Act of 2002, Journal of Accounting and Economics, 44, 74-115.

Kenneth J. Hunsader, University of South Alabama

Ross N. Dickens, University of South Alabama

Table 1: Summary Statistics

The following table provides the summary statistics for the sample.
DE is the total debt to total equity ratio. NIDEP/TL is the (net
income + depreciation) divided by total liabilities. CR is the current
ratio (defined as total current assets divided by total current
liabilities) of the company. MV is the market value of the company in
millions of dollars. TNW is the tangible net worth (defined as total
assets minus total liabilities minus intangibles) in millions of
dollars. NWC is the net working capital (defined as total current
assets minus total current liabilities) in millions of dollars. PPE is
the net plant, property and equipment in millions of dollars. Finally,
Z-score is Altman's measure of bankruptcy prediction as computed by
Research Insight using Compustat data.

Panel A: All Firms

Variable Number Mean Median Standard
 of firms Deviation

DE 1,716 1.37 0.65 9.21
NIDEP/TL 1,716 -0.07 0.16 1.46
CR 1,716 3.65 2.57 3.77
MV 1,716 2,993.54 344.97 12,256.18
TNW 1,716 440.09 69.74 2,137.27
NWC 1,716 301.33 65.79 1,039.28
PPE 1,716 561.62 37.47 2,303.98
Z-score 1,716 5.54 3.72 10.59

Panel B: Mining

Variable Number Mean Median Standard
 of firms Deviation

DE 121 1.29 0.97 3.24
NIDEP/TL 121 0.32 0.24 0.81
CR 121 2.58 1.50 5.53
MV 121 2,117.95 749.57 3,728.41
TNW 121 771.47 247.09 1,420.31
NWC 121 120.12 8.96 381.09
PPE 121 1,357.87 401.01 2,671.13
Z-score 121 4.15 2.45 8.39

Panel C: Manufacturing

Variable Number Mean Median Standard
 of firms Deviation

DE 1,595 1.37 0.61 9.51
NIDEP/TL 1,595 -0.10 0.15 1.49
CR 1,595 3.73 2.67 3.59
MV 1,595 3,059.96 330.82 12,669.15
TNW 1,595 414.95 64.75 2,180.33
NWC 1,595 315.07 65.59 1,071.67
PPE 1,595 499.84 31.05 2,262.74
Z-score 1,595 5.64 3.89 10.73

Table 2: Multivariate Regression Model Results

We estimate the following model: [r.sub.it] = [[alpha].sub.i] +
[[alpha]'.sub.i][D.sub.t] + [[beta].sub.1i][r.sub.mt] +
[[beta].sub.1i][r.sub.m(t-1)] + [[gamma].sub.i][D.sub.o] +
[[beta]'.sub.i][D.sub.t][r.sub.mt] + [[beta]'.sub.1i][D.sub.t]
[r.sub.m(t-1)] + [[epsilon].sub.it] return for portfolio i on day t,
[[alpha]'.sub.i] = intercept coefficient for portfolio i, [D.sub.t]
= dummy which is 1.0 after the last event date; else 0.0 (= 1.0 for
day +2 to day +120), [[alpha]'.sub.i] = shift intercept coefficient
for portfolio i, [[beta].sub.i] = systematic risk coefficient on
market return for portfolio i, [r.sub.mt] = the return on the equally
weighted market portfolio on day t, [[beta].sub.1i] = systematic risk
coefficient on the lagged market return for portfolio i,
[[gamma].sub.i] = the wealth effect of the announcement on portfolio i
for the event, [D.sub.o] = dummy which is 1.0 in the event window;
else 0.0 (= 1.0 for day -1 to day +1), [[beta]'.sub.i] = shift in
systematic risk for portfolio i, [[beta]'.sub.1i] = shift in
systematic risk for portfolio i on the lagged return, and
[[epsilon].sub.it] = error term. We estimate the model utilizing the
full 1,716 firms (separated into 121 mining firms and 1,595
manufacturing firms).

Sample Sort [[alpha].sub.i] [[alpha]'.sub.i]
 Variable

All Mining 0.0009 0.0004
Companies -0.9 -0.25

 [F-test = 25.28 ***, Adjusted [R.sup.2] = 0.377]

 Manuf. -0.0003 0.0003
 (-0.96) -0.8

Sample [[beta].sub.i] [[beta].sub.1i] [[gamma].sub.i]

All 1.4816 -0.2808 -0.0017
Companies (8.22) *** (-1.53) (-0.26)

 1.1802 0.039 -0.0026
 (26.34) *** -0.85 (-1.58)

Sample [[beta]'.sub.i] [[beta]'.sub.1i]

All 0.0571 0.6589
Companies -0.23 (2.60) ***

 -0.0156 -0.027
 (-0.78) (-0.43)

 [F-test = 234.78 ***, Adjusted [R.sup.2] = 0.854]

***, **, and * indicate significance at the 0.01, 0.05 and 0.10 level,
respectively.

Table 3: Analysis of Cumulative Abnormal Returns (CARs)

We compare CARs utilizing the full 1,716 firms (separated into
121 mining firms, 1,595 manufacturing firms, and 326 chemical
firms). DE is the debt to equity value, NIDEP/TL is the net
income plus depreciation divided by total liabilities, CR is
the current ratio defined as total current assets divided by
total current liabilities, TNW is the tangible net worth (defined
as total assets minus total liabilities minus intangibles), NWC
is the net working capital (defined as total current assets minus
total current liabilities), TNWCC is tangible net worth divided by
1% of net plant, property and equipment, NWCCC is net working
capital divided by 1% of net plant, property and equipment, and
Z-score is Altman's measure of bankruptcy prediction from Research
Insight. We report t-statistics in parentheses () and F-statistics
in brackets [].

Sort Variable Whole Sample Mining

 CAR t- or F- CAR t- or F-
 statistic statistic

DE>2 -0.350 (-1.80) * -0.805 (-1.17)
DE<2 -0.238 (-1.59) -0.073 (-0.11)
Difference -0.112 [0.41] -0.732 [1.93]
NIDEP/TL<0.10 -0.370 (-1.69) * -0.689 (-1.01)
NIDEP/TL>0.10 -0.178 (-1.40) 0.000 (-0.01)
Difference -0.192 [1.16] -0.689 [1.72]
CR<1.5 -0.177 (-0.98) 0.033 (0.04)
CR>1.5 -0.269 (-1.64) -0.358 (-0.58)
Difference 0.092 [0.18] 0.391 [1.28]
TNW<10 mil. 0.005 (0.02) -0.541 (-0.48)
TNW>10 mil. -0.320 (-2.21) *** -0.117 (-0.18)
Difference 0.325 [3.27] -0.424 [0.21]
TNWCC<6 -0.250 (-1.33) -0.567 (-0.65)
TNWCC>66 -0.250 (-1.71) * -0.145 (-0.22)
Difference 0.000 [0.00] -0.422 [0.32]
NWCCC<6 -0.072 (-0.22) 0.224 (0.30)
NWCCC>6 -0.269 (-1.71) * -0.502 (-0.81)
Difference 0.197 [0.27] 0.726 [4.23] **
Z<1.81 -0.462 (-1.73) * -0.379 (-0.52)
Z>1.81 -0.207 (-1.55) -0.085 (-0.13)
Difference -0.255 [1.46] -0.294 [0.48]

Sort Variable Manufacturing Chemical

 CAR t- or F- CAR t- or F-
 statistic statistic

DE>2 -0.316 (-1.52) -0.810 (-2.25) **
DE<2 -0.251 (-1.47) -0.552 (-1.97) **
Difference -0.065 [0.12] -0.258 [0.64]
NIDEP/TL<0.10 -0.356 (-1.53) -0.926 (-2.67) ***
NIDEP/TL>0.10 -0.194 (-1.33) -0.013 (-0.06)
Difference -0.162 [0.85] -0.913 [10.11] ***
CR<1.5 -0.225 (-1.42) -0.724 (-1.85) *
CR>1.5 -0.265 (-1.48) -0.572 (-1.99) **
Difference 0.040 [0.06] 0.152 [0.13]
TNW<10 mil. 0.028 (0.13) -0.522 (-1.42)
TNW>10 mil. -0.338 (-2.03) ** -0.615 (-2.27 **)
Difference 0.310 [3.94] ** -0.093 [0.10]
TNWCC<6 -0.240) (-1.21) -0.707 (-1.57)
TNWCC>66 -0.262 (-1.56) -0.573 (-2.17) **
Difference 0.022 [0.02] -0.134 [0.14]
NWCCC<6 -0.262 (-0.95) -0.787 (-1.08)
NWCCC>6 -0.259 (-1.52) -0.581 (-2.11) **
Difference -0.003 [0.00] -0.206 [0.08]
Z<1.81 -0.473 (-1.54) -0.893 (-2.16) **
Z>1.81 -0.214 (-1.43) -0.473 (-1.87) *
Difference -0.259 [1.21] -0.420 [1.65]

***, **, and * indicate significance at the 0.01, 0.05, and
0.10 level, respectively

Table 4: Multivariate Regression Model Results Utilizing EPA Test
Groups--Pass versus Fail

We estimate the following model: [r.sub.it] = [[alpha].sub.i] +
[[alpha]'.sub.i][D.sub.t] + [[beta].sub.i][r.sub.mt] + [[beta].sub.1i]
[r.sub.m(t-1)] + [[gamma].sub.i][D.sub.o] + [[beta]'.sub.i][D.sub.t]
[r.sub.mt] + [[beta]'.sub.1i][D.sub.t][r.sub.m(t-1)] +
[[epsilon].sub.it] where [r.sub.it] = the return for portfolio i on
day t, [[alpha].sub.i] = intercept coefficient for portfolio i,
[D.sub.t] = dummy which is 1.0 after the last event date; else 0.0
(= 1.0 for day +2 to day +120), [[alpha]'.sub.i] = shift intercept
coefficient for portfolio i, [[beta].sub.i] = systematic risk
coefficient on market return for portfolio i, [r.sub.mt] = the return
on the equally weighted market portfolio on day t, [[beta].sub.1i] =
systematic risk coefficient on the lagged market return for portfolio
i, [[gamma].sub.i] = the wealth effect of the announcement on
portfolio i for the event, [D.sub.o] = dummy which is 1.0 in the event
window; else 0.0 (= 1.0 for day -1 to day +1), [[beta]'.sub.i] = shift
in systematic risk for portfolio i, [[beta]'.sub.1i] = shift in
systematic risk for portfolio i on the lagged return, and
[[epsilon].sub.it] = error term. We utilize the full 1,716 firms, the
121 mining firms, and the 1,595 manufacturing firms (separating each
set into firms passing or failing the EPA tests). The values in {}
report the F-test comparing the wealth effects ([[gamma].sub.i]) for
passing and failing groups.

Sample Sort [[alpha]. [[alpha]'. [[beta].
 Variable sub.i] sub.i] sub.i]

All Firms EPA -0.0003 0.0004 1.2364
 Pass (-1.34) (1.04) (28.51) ***

 EPA 0.0002 0.0002 1.1243
 Fail (0.73) (0.040) (23.65) ***

Mining Firms EPA 0.0004 0.0008 1.6271
 Pass (0.44) (0.56) (9 45) ***

 EPA 0.0013 0.0001 1.3829
 Fail (1.13) (0.06) (7.13) ***

Manufacturing EPA -0.0004 0.0003 1.2198
Firms Pass (-1.35) (0.91) (25.91) ***

 EPA 0.0000 0.0002 1.0848
 Fail (0.12) (0.38) (20.51) ***

Sample [[beta]. [[gamma]. [[beta]'. [[beta]'.
 sub.1i] sub.i] sub.i] sub.1i]

All Firms -0.0284 -0.0034 -0.0378 -0.0204
 (0.64) (-2.13) ** (-0.63) (-0.33)

 -0.0086 -0.0006 -0.0482 0.1102
 (-0.18) (-0.37) (-0.73) (1.65) *
 {2.71} *
Mining Firms -0.3372 -0.0019 0.0155 0.6341
 (-1.92) * (-0.31) (0.06) (2.62) ***

 -0.2427 -0.0015 0.0851 0.6754
 (-1.23) (-0.21) (0.32) (2.48) **
 {0.02}
Manufacturing 0.0439 -0.0035 -0.0401 -0.0482
Firms (0.92) (-2.00) ** (-0.61) (-0.73)

 0.0271 -0.005 -0.0685 0.0240
 (0.50) (-0.26) (-093) (0.32)
 {3.53}*

Sample Adj F-test
 [R.sup.2]

All Firms 0.874 277.87 ***

 0.824 188.36 ***

Mining Firms 0.437 32.04 ***

 0.319 19.71 ***

Manufacturing 0.851 229.48 ***
Firms

 0.781 139.03 ***

***, **, and * indicate significance at the 0.01, 0.05 and 0.10 level,
respectively.

Table 5: Multivariate Regression Model Results Utilizing Firms on
Toxic 100 List

We estimate the following model: [r.sub.it] = [[alpha].sub.i] +
[[alpha]'.sub.i][D.sub.t] + [[beta].sub.i][r.sub.mt] + [[beta].sub.li]
[r.sub.m(t-1)] + [[gamma].sub.i][D.sub.o] + [[beta]'.sub.i][D.sub.t]
[r.sub.mt] + [[beta]'.sub.li][D.sub.t][r.sub.m(t-1)] +
[[epsilon].sub.it] where [r.sub.it] = the return for portfolio i on
day t, [[alpha].sub.i] = intercept coefficient for portfolio i,
[D.sub.t] = dummy which is 1.0 after the last event date; else 0.0 (=
1.0 for day +2 to day +120), [[alpha]'.sub.i] = shift intercept
coefficient for portfolio i, [[beta].sub.i] = systematic risk
coefficient on market return for portfolio i, [r.sub.mt] = the return
on the equally weighted market portfolio on day t, [[beta].sub.1i] =
systematic risk coefficient on the lagged market return for portfolio
i, [[gamma].sub.i] = the wealth effect of the announcement on
portfolio i for the event, [D.sub.o] = dummy which is 1.0 in the event
window; else 0.0 (= 1.0 for day -1 to day +1), [[beta]'.sub.i] = shift
in systematic risk for portfolio i, [[beta]'.sub.1i] = shift in
systematic risk for portfolio i on the lagged return, and
[[epsilon].sub.it] = error term. We utilize the 46 firms for which we
have data that appear on the University of Massachusetts Political
Economy Research Institute's Toxic 100 list (separating the firms as
passing (21 firms) or failing (25 firms) the EPA tests). The value in
{} reports the F-test comparing the wealth effects ([[gamma].sub.i])
for passing and failing groups.

Sample Sort [[alpha]. [[alpha]'. [[beta].
 Variable sub.i] sub.i] sub.i]

Manufacturing EPA 0.0002 -0.0013 1.3809
Firms on the Pass (0.36) (-2.16) ** (19.16) ***

 EPA 0.0000 -0.0002 1.4719
Toxic 100 List Fail (0.00) (-0.36) (18.60) ***

Sample [[beta]. [[gamma]. [[beta]'. [[beta]'.
 sub.1i] sub.i] sub.i] sub.1i]

Manufacturing -0.1911 -0.0032 0.2029 0.1640
Firms on the (-2.60) *** (-1.21) (2.03) ** (1.62)

 -0.2012 -0.0034 0.0365a 0.1703
Toxic 100 List (-2.50) ** (-1.16) (0.33) (1.53)
 {0.01}

Sample Adj F-test
 [R.sup.2]

Manufacturing 0.787 148.68 *
Firms on the **

 0.753 123.22 *
Toxic 100 List **

***, **, and * indicate significance at the 0.01, 0.05, and 0.10 level,
respectively.

(a) There is a marginal statistical difference between the
[[beta]'.sub.i] values for the passing and failing subgroups (F-test =
3.52 which is significant at the 0.10 level).