environmental risk management- vicram and group

8/13/2019 Environmental Risk Management- Vicram and Group

1/24

Strategic Management JournalStrat. Mgmt. J., 29: 569592 (2008)

Published online in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/smj.678

Received 26 July 2006; Final revision received 20 December 2007

ENVIRONMENTAL RISK MANAGEMENT

AND THE COST OF CAPITAL

MARK P. SHARFMAN1* and CHITRU S. FERNANDO2

1 Division of Management, Michael F. Price College of Business, The Universityof Oklahoma, Norman, Oklahoma, U.S.A.2 Division of Finance, Michael F. Price College of Business, The Universityof Oklahoma, Norman, Oklahoma, U.S.A.

Our study of 267 U.S. firms shows that improved environmental risk management is associatedwith a lower cost of capital. Our findings provide an alternative perspective on the environmental-economic performance relationship, which has been dominated by the view that improvements ineconomic performance stem from better resource utilization. Firms also benefit from improvedenvironmental risk management through a reduction in their cost of equity capital, a shiftfrom equity to debt financing, and higher tax benefits associated with the ability to add debt.These findings help build better theory regarding the outcomes of strategic improvements inenvironmental risk management.Copyright 2008 John Wiley & Sons, Ltd.

INTRODUCTION

The idea that a firms environmental (green) per-formance and overall economic performance arepositively related (Murphy, 2002) has not alwaysreceived universal acceptance within the researchcommunity. The conventional view had been that

such activities represented a cost to the firm andshould be minimized whenever possible. Specifi-cally,

(i)nvestors view pollution control expenditures as adrain on resources that could have been investedprofitably, and do not reward the companies forsocially responsible behavior (Mahapatra,1984 : 29).

Starting in the 1970s, researchers examined therelationship by looking at various approaches to

Keywords: environmental risk management; cost of cap-ital; environmental performance; environmental strategy*Correspondence to: Mark P. Sharfman, Division of Manage-ment, 307 W. Brooks, Rm. 206, Michael F. Price College ofBusiness, The University of Oklahoma, Norman, OK 73019-0450, U.S.A. E-mail: [email protected]

measuring both constructs (e.g., Christmann, 2000;Dowell, Hart, and Yeung, 2000; Hart and Ahuja,1996; Russo and Fouts, 1997; Spicer, 1978). Thetheoretical underpinning of much of this researchstems explicitly or implicitly from variants ofthe resource-based view (Bansal, 2005; Barney,1991; Darnall and Edwards, 2006; Hart, 1995;Sharma and Vredenburg, 1998). In their theoreti-cal perspectives, previous authors have argued thatif the firm makes greener (i.e., more efficient)use of its resources it will be more economi-cally effective. Such greener use can come from,for example, generating less pollution and wastefrom the resources employed, or by using fewerresources. While there have been some dissent-ing voices along the way (e.g., Chen and Met-calf, 1980; Mahapatra, 1984), when researchersfind a positive relationship between environmentaland economic performance, they generally credit

it to such improved resource utilization, which inturn leads to overall increases in organizationaleffectiveness. We propose an additional theoret-ical perspective that can enrich our understand-ing of the environmental performance-economicperformance relationship.

Copyright 2008 John Wiley & Sons, Ltd.


2/24

570 M. P. Sharfman and C. S. Fernando

Most research on the environmental-economicperformance relationship has been predicated onthe idea that (internal) strategic environmentalinvestments result in improved resource efficiency(Bansal and Roth, 2000; Branzei et al., 2004;Buysse and Verbeke, 2003). While the effects of

such strategic choices are often clear even to thefinancial markets, internal investments are not theonly phenomena that drive organizational perfor-mance. Institutional and other external factors alsohave a profound effect on the performance (sur-vival) of firms (Singh, Tucker, and House, 1986).While several researchers have examined how thestock market reacts to improved environmentalperformance through market returns (e.g., Dowellet al., 2000; Gottsman and Kessler, 1998; Maha-patra, 1984), little attention has been paid to suchexternal influences on the environmental-economic

performance relationship itself. In this article weexamine the capital markets response to improvedenvironmental risk management by a firm as mea-sured by changes in the firms cost of capital.We posit that such improved environmental riskmanagement improves the markets risk percep-tion of the firm. There is evidence in the liter-ature that investors and analysts take account ofimprovement in environmental risk factors whenmaking investment decisions and recommenda-tions (Heinkel, Kraus, and Zechner, 2001; Mackey,Mackey, and Barney, 2007). This improved per-

ception should, in turn, cause the financial marketto be willing to accept lower risk premiums onequity, or allow the firm to acquire higher levelsof leverage, all of which can result in a loweredcost of capital overall. The lowering in the firmscost of capital due to a reduction in the perceivedriskiness of its cash flows (from environmental riskmanagement) can be differentiated both conceptu-ally and empirically from an increase in its cashflows from greater revenues and/or lowered costsdue to improved resource efficiency through betterenvironmental performance. The value of the firm

can be increased due to either or both of theseeffects. In contrast to prior studies (e.g., Klassenand McLaughlin, 1996; Dowell et al., 2000) thathave looked at the broad question of whetherimproved environmental performance increases themarket value of the firm, our focus is solely onwhether or not investors risk perception of thefirm changes due to environmental performance inthe form of environmental risk management. Thisquestion is important because the cost of capital

and return on capital are fundamental variablesfrom the standpoint of the firm and capital markets,respectively. If the market perceives improvementsin resource utilization but did not perceive changesin riskiness, the cost of capital would not change.However, if the changes in perceived riskiness

lead to reduced costs of capital, the firms over-all cost base decreases while its ability to make aprofit from a given level of revenue correspond-ingly increases.

Using risk management and institutional theoriesas our conceptual underpinning with a sample fromthe Standard and Poors (S&P) 500, we examinewhether improved environmental risk managementis associated with a reduction in firms total costof capital. Further, we test several sub-hypothesesto decompose what drives any such reduction.Such findings should enrich theory by showing that

improved environmental risk management signalsthe financial markets that the firm represents a lessrisky investment that deserves less expensive debtand lower equity risk premiums. Such loweredcosts of capital should, in turn, increase the firmsoverall economic performance (Scott and Pascoe,1984) and thereby help to explain the observedpositive relationship between economic and envi-ronmental performance.

Prior research

To understand where our study fits in with priorliterature, we briefly examine earlier work on thegeneral relationship between environmental andeconomic performance (see Murphy [2002] foran extensive review of 20 well-known studiesin the area) plus what limited research has beendone on the relationship between environmentalperformance and the cost of capital.

The general research question in this literatureasks whether the investment in improved envi-ronmental performance pays off in improved eco-nomic performance. To our knowledge, the first

study on this topic was by Spicer who foundthat . . .for a sample [of firms] drawn from thepulp and paper industry, companies with bet-ter pollution-control records tend to have higherprofitability. . . (Spicer, 1978: 109). Chen andMetcalf (1980) showed that Spicers (1978) resultswere contaminated by a variety of methodologicalproblems, including inappropriate statistical tests,failure to control for size and the use of paneldata while making longitudinal/temporal causality

Copyright 2008 John Wiley & Sons, Ltd. Strat. Mgmt. J., 29: 569592 (2008)

DOI: 10.1002/smj


3/24

Environmental Risk Management and the Cost of Capital 571

arguments. Nonetheless, Spicers study focusedattention on the question that underlay the pub-lic debatewhether environmental expendituresshould be viewed as a cost or as an investment.

Several other researchers have examined thisquestion using accounting-based dependent vari-

ables. Nehrt (1996) found that firms that investedin cleaner technologies experienced higher lev-els of abnormal profit growth, while the earliestadopters had the highest levels of that growth. Hartand Ahuja (1994) found that pollution reductioninitiatives were positively related to several perfor-mance measures over a two-year horizon, and thatthe firms that made the largest improvements inenvironmental performance experienced the mostprofound economic gains. Russo and Fouts (1997)extended this body of research by examining (froma resource-based perspective) broader measures

of environmental performance and their effect onreturn on assets. They found a strong relation-ship between environmental and economic perfor-mance, particularly when including the moderatingrole of industry growth.

Other researchers have examined the questionsusing market-based measures of firm performance,also with generally consistent results. For exam-ple, Cohen, Fenn, and Naimon (1995) found thata sample of theoretical investment portfolios oflow pollution firms provided better returns than didsimilar portfolios of high pollution firms. Klassenand McLaughlin (1996) found that firms receivingstrong, positive environmental publicity increasedin market value, while negative publicity had theopposite effect. Hamilton (1995) found that firmssuffered stock market losses after particularly neg-ative Toxic Release Inventory (TRI) announce-ments. Bosch et al. (1998) found results similarto Hamilton (1995), that both firms who had beenthe target of Environmental Protection Agency(EPA) investigations and those who had appealedenforcement actions lost market value when these

outcomes were announced. Dowell et al. (2000)determined that firms with stricter environmentalmanagement systems had higher market values aswell. Essentially all of these researchers examinedthe relationship between environmental and eco-nomic performance from the perspective of variousinternal effects of improved environmental per-formance. This perspective begs the question asto whether there are external effects from suchinvestments, e.g., on costs of capitalan area that

according to the normative environmental manage-ment literature should be affected.

Nonetheless, we found considerably less workthat focused specifically on the relationship bet-ween environmental risk management and cost ofcapital. In the green literature, one sees asser-

tions that overall improvements in environmentalrisk management should lead to (among other ben-efits) lowered costs of capital. However, whilethis normative logic may be strong, the empiri-cal support for the assertion is sparse. As far aswe can determine, no prior study has examinedcomprehensively the relationship between overallenvironmental risk management and the costs ofcapital.

In the first of only three studies we found thataddressed the question in any form, Feldman,Soyka, and Ameer (1997) found positive effects on

beta and stock price from improvements in envi-ronmental risk management. Due to the proprietarynature of their model, they did not provide suffi-cient details to replicate their methodology or tocreate empirical estimates, and only provided anillustration of the inferred relationship. Addition-ally, their focus was only on beta and the cost ofequity capital, disregarding debt financing and itscost. Heinkel et al. (2001) developed a theoreticalmodel to show that exclusionary investing (basedon environmental performance) can induce somepolluting firms to undertake reforms to voluntar-

ily stop polluting when the cost of doing so isless than the cost of equity capital disadvantagethat they would otherwise have to bear. However,these authors also disregard potential effects ofenvironmental risk management on the availabil-ity and cost of debt financing, and other channelsthrough which environmental risk management canpotentially affect the cost of capital. Furthermore,they do not validate their predictions empirically.Garber and Hammitt (1998) examined the effecton costs of equity for 73 chemical firms withidentified Superfund liabilities. While they found

no relationship between the liabilities and costsof equity for small firms, they were able to finda robust positive relationship for 23 large firms.However, their focus was on balance sheet liabili-ties identified to cover Superfund charges, a single,extreme, well-publicized balance sheet item thatwas quite visible to equity analysts (e.g., Camp-bell, Sefcik, and Soderstrom, 1998) because of thenotoriety and fear that such charges evoke in thegeneral public. Superfund liabilities are far more


DOI: 10.1002/smj


4/24


visible and easy to understand than other signalssuch as TRI reports or EPA fines. Generalizedenvironmental risk management is a longer-termphenomenon and is neither as quantifiable nor astransparent. Therefore, it is important to determinewhether the capital market is able to discern dif-

ferences in environmental risk management acrossfirms. Second, by focusing only on the cost ofequity capital, Garber and Hammitt (1998) disre-gard the costs and tax effects of debt financingon a firms overall cost of capital. Therefore, thequestion of whether and how a firms environmen-tal risk management influences its cost of capi-tal remains open, and it is this question that weaddress in the present article.

THEORETICAL FRAMEWORK

In this section of the article we describe the the-oretical linkages between a firms environmentalrisk management activity and its cost of capital. Afirms cost of capital is an important determinantof its valuation for two reasons. First, the cost ofcapital is the expected rate of return demanded bya firms investors for investing in the firm. Thehigher the rate of return demanded by a firmsinvestors for the capital they provide to the firm,the more costly it is for a firm to finance itself. Ascapital is a basic input that the firm receives, the

more costly that this input is, the less chance thefirm has to make a profit regardless of its level ofrevenues.

Second, the cost of capital is the rate thatinvestors use to discount a firms future cash flows.The higher the cost of capital, the lower the presentvalue of the firms future cash flows. Therefore, allelse being equal, firms with a lower cost of cap-ital will be more highly valued than firms witha higher cost of capital and hence more attrac-tive to investors. Investors determine a firms costof capital by evaluating the riskiness of its cash

flows relative to other investment opportunitiesthat are available to them. Broadly speaking, firmsare financed through either debt or equity capital.Debt capital can come from private sources (e.g.,banks) or from public sources (the debt markets).In either case, the cost of debt is the applica-ble interest rate. (In public debt markets, investorstrade debt securities just like they trade equities ina stock market.) The cost of equity is the returninvestors in the firms shares expect as reflected

in the stock price they are willing to pay relativeto future expected cash flows. Since most publiclyheld firms typically finance themselves with bothdebt and equity, the firms overall cost of capital isgiven by the weighted average of its cost of debtand equity capital, which has come to be known

as the weighted average cost of capital (WACC).In the general case of a firm with both equityand debt financing, the firms after-tax weightedaverage cost of capital (rWACC ) (Modigliani andMiller, 1958) can be expressed as

rWACC =

E

D + E

rE

+

D

D + E

rD (1 T ), (1)

where

E = market value of the firms equity;D = market value of the firms debt;rE = the firms cost of equity capital;rD = the firms cost of debt capital; andT = the firms rate of corporate taxation.

The firms cost of equity capital (rE) is equal tothe expected investor return from holding thefirms equity, estimated with the Capital AssetPricing Model (CAPM) (Sharpe, 1964; Lintner,1965)

rE = rF+ E (rM rF), (2)

where

rF = the risk-free rate;rM= the return on the market portfolio; and

E = Cov(rE , rM)

Var(rM) measures the firms systematic

risk.

Higher levels of environmental performanceshould be viewed as environmental risk manage-

ment for a variety of reasons. At the most basiclevel, as a firm makes strategic investments thatreduce emissions and pollution, it mitigates its riskof litigation either from governmental regulatorsor from nongovernmental stakeholders (cf. Kingand Shaver, 2001). This reduces both immediaterisks from known hazards and future risks fromcurrently unknown hazards. Both current (known)and future (unknown) hazards carry an uncertainlevel of financial impact. By reducing current or


DOI: 10.1002/smj


5/24


potential hazards, the firm reduces the numberof potential claimants on its rents through fines,settlements, or other compliance/litigation costs.By reducing the number of potential claimantson its rents, more of the firms overall economicresources can be directed strategically to dividends

to stockholders, debt payments, internal invest-ments, or acquisitions. Each of these activities islikely to be rewarded by the market in terms ofimproved risk perception of the company from aninvestment standpoint.

In general, improving a firms environmentalrisk exposure (environmental risk management)can reduce its cost of capital in three ways, throughboth direct and indirect relationships:

1. By reducing the firms cost of debt capital rD;2. By increasing the firms debt capacity, and

thereby increasing the amount of income thefirm can protect from corporate taxation;3. By reducing the firms cost of equity capital.

We explore each of these possibilities using aseries of hypotheses to determine if any or all ofthe three contribute to any overall relationship thatmay exist.

One of the earliest arguments in favor of riskmanagement arises from the notion that risk man-agement reduces a firms expected costs of finan-cial distress and thereby enhances the quality of its

debt (Smith and Stulz, 1985). To understand howenvironmental risk management strategies couldaffect the cost of debt specifically, we examinethe theoretical nature of debt financing. The costof debt financing incurred by a firm is based on theassessment by the capital market (including banks,bond markets, and rating agencies) of the defaultrisk of the firm. The level of default risk that a firmpresents is a function of the uncertainty inherentin its future activities (Miller and Bromiley, 1990;Orlitzky and Benjamin, 2001). The greater theuncertainty inherent in a firms future activities, the

lower the assessed credit quality of its debt and thehigher the cost of debt financing (i.e., it will haveto pay higher interest rates). Undertaking environ-mental risk management activities by improvingenvironmental performance can reduce the likeli-hood that firms will encounter extreme environ-mental events (e.g., Union Carbides Bhopal dis-aster, the Exxon Valdez oil spill, or the hazardouswaste dumping in the Love Canal area by HookerChemical) that can require heavy cash outflows

arising from compensation and cleanup costs, andthereby make firms more vulnerable to bankruptcy.Environmental risk management can also reducethe vulnerability of firms to other adverse busi-ness developments that would reduce profitability,impair the firms reputation, or reduce the value

of its asset base. Additionally, while conventionalfinancial risk management activities (e.g., usingderivative securities) can sometimes be reversed bystockholders after new debt financing takes place(Smith and Stulz, 1985; Chidambaran, Fernando,and Spindt, 2001), environmental risk managementinvestments are usually long term in nature andcannot be easily reversed. Perhaps such stabilitymakes them more credible from the standpoint ofthe firms future debt holders. If environmental riskmanagement reduces the default risk profile that afirm presents to the debt markets, those markets

should in turn reward that improved risk profilewith lower required interest rates and subsequentlowered costs of debt capital. This leads to our firstprediction:

Hypothesis 1: The higher the level of environ-

mental risk management, the lower the firms

cost of debt capital for a given level of debt.

Hypothesis 1 assumes that the firm holds itslevel of debt fixed and benefits from increasing itslevel of environmental risk management by incur-

ring a lower cost of debt capital. Alternatively, afirm may choose to take strategic advantage ofits lower cost of debt by increasing the overalllevel of debt in its debt-equity financing mix. Thefirms ability to increase its leverage by managingrisks gives rise to another explanation for the link-age between environmental risk management andcosts of debt capital (Leland, 1998; Stulz, 1996).Risk management can be viewed as an alterna-tive strategic choice to employing equity capital.When the firm shifts its financing strategy fromequity to debt, it shifts to the debt market the bur-

den to bear the firms residual risk (Stulz, 1996).As a firm increases its level of risk management, itcan correspondingly shift its financing from equityto debt capital (i.e., increase its leverage) becausethe firm is perceived as being less risky. Leland(1998) develops a model in which a firms optimallevel of leverage rises with its level of risk man-agement. And, consistent with these arguments,Graham and Rogers (2002) provide empirical evi-dence indicating a positive relationship between


DOI: 10.1002/smj


6/24


leverage and strategic hedging. Another benefitof raising the firms level of debt is that higherleverage, in turn, increases the amount of incomethat a firm can shield from taxation. Therefore,the higher tax advantage associated with a higherlevel of debt reduces the after-tax cost of capital

of a firm, giving managers additional incentive toadd leverage by reducing the cost of borrowingeven further. This discussion leads to two addi-tional sub-hypotheses:

Hypothesis 1a: The higher the level of environ-

mental risk management, the higher the firms

leverage.

Hypothesis 1b: The higher the level of environ-

mental risk management, the higher the firms

tax advantage (shield) from debt financing.

Next, environmental risk management strategiesmay also reduce a firms overall cost of capitalby reducing its cost of equity financing. If envi-ronmental risk management results in improvedfinancial performance by way of a lower level ofsystematic risk, then the market should reward thatimproved performance with a lower cost of equitycapital. An investor makes investments by tradingoff risk and return, that is, either maximizing returnfor a given level of systematic risk or minimizing

systematic risk for a given level of return.If environmental risk management through im-

proved environmental performance augments over-all organizational effectiveness or even just increa-ses firm legitimacy (Deephouse, 1996; Russo,2003) then it would follow that the more firmsadopt such an approach, the higher their level offuture performance and the greater the willingnessof the market to invest in them. The more confi-dence the market has that a firm will provide highreturns on invested capital (assuming equal levelsof risk) or reduce the systematic risk of the invest-

ment (assuming equal levels of return), the more itwill be willing to pay for the opportunity to capturethose returns, thereby driving up the stock price.In the latter case, as firms increase their level ofenvironmental risk management and become morelegitimate in the eyes of potential stockholdersor other stakeholders (Deephouse, 1996; Hoffman,1999; Jennings and Zandbergen, 1995; Kassinisand Vafeas, 2002; Sharma and Henriques, 2005),according to institutional theorys arguments, the

market should reward these more legitimate firmswith a lower cost of equity. Thus,


mental risk management, the lower the cost of

equity capital.

A reduction in the cost of equity financing byreducing the firms systematic risk is manifestedin a lowered equity beta (Feldman, Soyka, andAmeer, 1997). The firm may be able to accomplishthis by implementing environmental risk manage-ment operating changes (i.e., better buffer sys-tems Thompson, 1967) to increase its flexibilityto manage economic downturns (Stulz, 2003). Forexample, assume the firm changes processes torequire less inputs or at least less toxic inputs.In an economic downturn, should the firms sup-

ply chains be restricted, the firm would be lesssusceptible to price increases from its reducedinput profile. By reducing the variability in perfor-mance, the firm likely will reduce its beta, whichalso should lead to lowered costs of equity cap-ital. (See Bansal and Clelland [2004] who makea comparable argument regarding environmentalperformance and unsystematic risk.) Therefore, theimpact of improved environmental risk manage-ment can be measured by a reduction in the valueof beta:

Hypothesis 2a: The higher the level of environ-mental risk management, the lower the firms

non-leveraged equity beta (systematic risk).

A second explanation for the relationship bet-ween environmental risk management and the costof equity capital stems from the types of investorsgreen firms attract. In the Heinkel et al. (2001)model with two investor classes, green and non-green, the authors showed how green investorswill only invest in firms with good environmentalrisk management (i.e., more legitimate firms) while

non-green investors are indifferent about envi-ronmental risk management and will not necessar-ily invest in green firms. Therefore, firms withpoor environmental risk management will have ahigher cost of equity capital because fewer peoplewill buy their shares (cf. Mackeyet al., 2007). Fur-ther, Merton (1987) develops a model of capitalmarket equilibrium with incomplete informationin which the firms cost of equity declines as itsinvestor base expands. In the context of Merton


DOI: 10.1002/smj


7/24


(1987), it is possible that firms engaging in higherlevels of environmental risk management will gainmore visibility and positive publicity in the media,thereby attracting more investors (Bansal, 2005).Heinkel et al. (2001) and the Social InvestmentForum (2003) estimate that approximately 10 to

11 percent of all assets under management in theUnited States are invested after applying ethicalor environmental screens, implying that a sig-nificant subset of investors intentionally includegood environmental performers in their portfolios.This increase in the number of shareholders willincrease the firms share price and decrease itscost of equity. Hence, we would once again expectthe cost of equity capital to be negatively relatedto environmental risk management; in this casedue to firms with better environmental risk man-agement having a more dispersed ownership (i.e.,

more shareholders). Therefore,

Hypothesis 2b: The higher the level of environ-

mental risk management, the more dispersed the

firms share ownership.

Finally, a third explanation for why the cost ofequity financing might be reduced can be foundin the benefits of the firm being monitored byinstitutional investors who own its shares. Underthis line of reasoning, if firms strategically man-age their environmental risks, it is possible that

this activity will gain them legitimacy even withnon-green institutional investors and increase theinvestment by institutional investors in the firmsshares. As widely documented in the literature,institutional investors actively monitor the perfor-mance of the firms they invest in, and such moni-toring can lead to a variety of benefits for the firm(Allen, Bernardo, and Welch, 2000; Carleton, Nel-son, and Weisbach, 1998; Gillan and Starks, 2000;DMello, Schlingemann and Subramaniam, 2003:Fernando, Gatchev, and Spindt, 2007; Shleifer andVishny, 1986; Smith, 1996). McConnell and Ser-

vaes (1990) show specifically that institutionalownership and firm value are positively related.A key argument in this literature is that firms witha higher concentration of institutional owners willbenefit from a lower cost of capital/higher per-formance. Additionally, in the particular case ofenvironmental risk management, it is possible thateven non-green institutional owners will attemptto steer given firms away from taking undue risks(Heinkel et al., 2001) and hence will stay away

from firms that are poor environmental perform-ers. If so, we would expect to find a positiverelationship between institutional ownership andenvironmental risk management, such that envi-ronmental risk management has an indirect effecton the cost of equity capital through institutional

share ownership. This leads to our next hypothesis:

Hypothesis 2c: The higher the level of environ-

mental risk management, the higher the percent-

age of institutional share owners.

The above hypotheses combined imply a singleomnibus prediction that improved environmentalrisk management will lower the firms weightedaverage cost of capital as in Equation 1. This leadsto our final hypothesis.


mental risk management, the lower the firms

weighted average cost of capital (WACC).

The analysis that we present in the followingsections investigates each of our predictions thatlink cost of capital components to environmen-tal risk management and our omnibus predictionin Hypothesis 3 that links the overall weightedaverage cost of capital to environmental risk man-agement. We summarize the model in Figure 1.

METHODOLOGY

Sample and datasets

To test our hypotheses, we required a dataset offirms that met two criteria: (1) Firms had to belarge enough to be traded publicly and able toaccess the capital markets regularly to enable anaccurate estimate of their costs of capital; plus(2) have a meaningful and transparent environ-

mental management function to enable measure-ment of their environmental risk management. TheS&P 500 dataset meets these criteria as it repre-sents a cross section of the largest publicly heldfirms in the United States, and has been used inother studies in the area (e.g., Dowell et al., 2000;Hart and Ahuja, 1996).

We used two different datasets to create ourenvironmental risk management measure. First, weobtained access to the United States EPA TRI data


DOI: 10.1002/smj


8/24


Weighted averagecost of capital

Environmental riskmanagement

Beta

Cost of debt capital

Cost of equity capital

Tax shield from debt

Leverage

Firm size

Industry

Percentage ofinstitutionalshareholders

Dispersed shareownership

(+)(+)

(-)

(-)

(-)

(-)

(+)

(+)

Predictedrelationship

Theoretical relationshipfrom the lterature

Control variables

(-)

(-)

(-)

(+)

(+)

(+)

Figure 1. Conceptual model

through the Investor Responsibility Research Cen-ter (IRRC)1. The TRI data are reported by firms tocomply with EPA regulations about the use, emis-sion, or disposal of over 600 toxic substances. Theuse of TRI data has come under criticism (e.g.,Toffel and Marshall, 2004) because of questionsabout (among other issues) firm reporting accu-racy, the datas inability to account for relative

toxicity, and the fact that firms sometimes reportestimates as opposed to actual emissions. Regard-less of the complaints, the TRI still is the mostwidely examined set of measures of firms environ-mental activity (Toffel and Marshall, 2004). Thisextensive use occurs because it is the broadest andmost objective dataset available. As an additionalindicator of environmental risk management weused the environmental scores from the Kinder,Lydenberg, Domini & Co., Inc. (KLD) social per-formance dataset2. The use of the KLD mea-sure allowed us to triangulate between two very

different approaches to measuring environmentalrisk management. KLD is a financial advisor whoprovides social screening of firms to clients viaits reports and socially screened mutual funds.These data too have come under criticism because,for example, they were developed atheoretically(Sharfman, 1996) and are non-normal because of

1 Received directly from the IRRC in April of 2004.2 Received directly from KLD in July of 2004.

the scoring system (Mattingly and Berman, 2006).However, Graves and Waddock (1994) suggestedthat the KLD data is the best single source of socialand environmental performance data because

The group doing the rating consists of knowledge-able individuals not affiliated with any of the ratedcompanies or with researchers performing studies.Thus, this firms scaling process provides unique

access to a wide range of consistently rated firmsacross a number of important social performanceattributes(Graves and Waddock, 1994 : 1039).

Measures

The first dependent variable used in our analysiswas each firms WACC derived as in Equation 1above. The cost of equity capital is estimated usingthe CAPM (Sharpe, 1964; Lintner, 1965), whichequates the cost of equity of a firm to the risk-freeinterest rate plus the firms beta times the mar-

ket risk premium. The cost of debt is the firmsmarginal cost of borrowing and is based on esti-mates from the Bloomberg Financial dataset3. Therisk free rate (Rf) is the return an investor canearn on an essentially risk-free investment. Fol-lowing convention, we used as our risk-free ratethe 10-year U.S. Treasury bond rate of 4.79 per-cent, which was the rate at the beginning of this

3 Received directly from Bloomberg in May of 2004.


DOI: 10.1002/smj


9/24


analysis according to Bloomberg. Beta () is thecovariance of the markets return with the individ-ual companys common stock return divided by themarkets variance. We used an annual beta com-puted by Research Insight4 for each company thatwe found in the COMPUSTAT database (WRDS,

2007). The weight of equity (We) is found byusing the total market capitalization, divided bydebt plus market capitalization. Market capitaliza-tion is the product of the value of one share ofstock times the number of shares outstanding. Theweight of debt (Wd) is found by summing all debtissues of the firm, divided by the total of that debtplus market capitalization. Market capitalization,the weight of debt, and the tax rate data came fromthe Bloomberg Financial database. We set the riskpremium (Rp) at 5.6 percent based on the Famaand French (2002) estimate of average returns over

the period 18722000. Throughout the paper werefer to this as WACC-1.

To reduce the possibility of error in our WACCmeasure, we estimated it using three other methodsand then created a factor score from the alter-native approaches. First, the Bloomberg databaseprovides its own estimate of each firms WACC.While calculated using the same formula as theone we used, the key differences between ourWACC-1 measure and Bloombergs WACC arethat Bloomberg uses a beta calculated weekly asopposed to COMPUSTATs annual beta, and they

use their own, firm-specific estimate of the riskpremium rather than using the Fama and French(2002) average. As the cost of equity will beaffected heavily by the beta value used, we alsoused the Bloomberg estimate of WACC (WACC-2). Further, because the cost of equity is highlysensitive to the risk premium, we recalculatedWACC-1 based on the Bloomberg, firm-specificrisk premium (WACC-3). Finally, to reduce fur-ther any potential effect of error in the measure-ment of WACC, we conducted a factor analysisusing a varimax rotation on WACC-1, WACC-2,

and WACC-3. We extracted a single factor thataccounted for 91 percent of the variance in thethree measures suggesting a confirmatory factoranalysis was not necessary. Using the loadingsfrom the factor analysis we constructed a factorscore from the factor weights times the originalvalues. We call the factor score WACC-4 and

4 Received directly from Standard and Poors in September of2006.

estimated our equations using that measure andWACC-1.5

We estimated WACC using the multiple methodswe describe above as an attempt to limit measure-ment error. While we report the results from theequations using the WACC-1 and the aggregate

WACC-4 dependent variables, it should be notedthat the WACC-1 results are the ones we discussthroughout the paper because they are the mostconsistent with how WACC has been estimated inprior literature. For cost of equity, we rely primar-ily on our estimate (COE-1). However, in our costof equity regressions, we employ both our estimateand the Bloomberg estimate of this value (COE-2).

We also used independent theoretical and controlvariables in our analysis. The theoretical variableis the measure of environmental risk management.From the variety of ways previous researchers

(e.g., Dowell et al., 2000; Klassen and McLaugh-lin, 1996; Russo and Fouts, 1997) have measuredenvironmental risk management, and in view ofits multifaceted nature, we believed it necessary touse multiple environmental risk management indi-cators. We selected indicators that would be widelyavailable to the financial markets as well as thoserepresenting both more quantitative and more qual-itative assessments. Such an approach allows us totriangulate on the elusive notion of environmen-tal risk management and to demonstrate conver-gent validity in the measure (Campbell and Fiske,

1959). For the quantitative measures we selecteddata elements from the TRI data as compiled bythe IRRC. Once the IRRC collects the TRI datafrom the EPA, they then standardize the values bydividing them by domestic revenues for the firm.This produces scaled environmental risk manage-ment measures that are comparable across firms.The IRRC collects 25 TRI data elements; how-ever only four of them provide enough observa-tions to allow us to conduct a meaningful analy-sis. These four elements were total TRI emissions(TRI-TOTAL), total TRI emissions treated on site

to reduce their toxicity (TRI-TREAT), total TRIemissions reused or recycled on-site for energy(TRI-ERR) and total waste generation includingTRI emissions (WASTE-GEN). We standardizedeach of the first three measures by WASTE-GENto see what percentage of the firms overall wastegeneration was being closely monitored (i.e., bypublic disclosure) through the TRI (ST-TRITO),

5 The factor analysis results are available from the first author.


DOI: 10.1002/smj


10/24


what percentage of the firms discharges werebeing treated to reduce their toxicity (ST-TRITR),and what percentage was being recycled for energy(ST-TRIEN).

For a qualitative assessment of environmentalrisk management we turned to the environmental

evaluations provided by KLD. As we mentionedabove, KLD is a social investment and screen-ing firm whose evaluations are used widely withinthe financial services industry. KLD evaluates themembership of the S&P 500 along a wide rangeof social performance indicators. Specifically KLDexamines seven strength and seven concernenvironmental dimensions ranging from the extentto which the firm engages in pollution preventionprograms or uses alternative fuels to the firmsuse of ozone depleting chemicals or its genera-tion of hazardous wastes. To calculate our KLD

score we averaged the strengths and the concernsseparately. In past literature, (see Sharfman, 1996for a review) authors kept the metric consistentby subtracting the average concern score from theaverage strength score. Unfortunately there is aflaw in the assumption that each strength categoryand each concern category are equal. As such, theassumption implied in adding or subtracting them,that one is adding or subtracting equals, is notempirically verifiable. Therefore, consistent withMattingly and Berman (2006) we created a sepa-rate average strength (KLD-ENVST) and concern

(KLD-ENVCO) score for each firm.We further examined whether the various envi-

ronmental risk management indicators could becombined into a single scale. As we had no theo-retical rationale as to how to combine the measuresother than their face validity, we could not conducta confirmatory factor analysis. Instead we ran anexploratory factor analysis using z-scores of thefive measures. Using a varimax rotation we wereable to extract a factor common to the ST-TRITO,ST-TRITR, and KLD-ENVST variables. This fac-tor was the only one with an eigenvalue over 1.00

that had no cross loadings among observables, plusit accounted for 43 percent of the variance in thesedata. We used the SPSS (statistical analysis soft-ware) factor score from these three indicators asour measure of environmental risk management.

Because any cost of capital effects would be theresult of prior environmental risk management, itwas also necessary to determine whether more thanone year of lagged environmental risk managementdata was necessary for our analysis. Using 2002

as our reference year (t), we estimated our modelsusing data from 1999(t3)to 2001 (t1)as well asthe average values from 20002002 to see if anyother lag in this data was appropriate. We obtainedmeaningful results only with the data from 2001(a one-year lag) so we used those values in our

analysis. We should note that list-wise deletion dueto missing data resulted in a final sample of 267.We also required measures for shareholder con-

centration and the percentage of institutional own-ers. For our indicator of shareholder concentrationwe used the total number of shareholders at theclose of 2002. The logic here is that the higherthe number of shareholders, the less concentratedthe shares. We used the published percentage ofinstitutional holders at the same point as a mea-sure as well. Both the number of shareholders andthe percentage of institutional holders came from

COMPUSTATs Research Insight database.In terms of control variables, our review of thecost of capital literature suggests that the three con-structs most likely to affect any sample of firmscosts of capital are firm financial leverage, indus-try membership, and firm size (Gebhardt, Lee, andSwaminathan, 2001). Additionally, controlling forindustry and size differences is also important torule out a spurious correlation between environ-mental risk management and cost of capital thatwould stem from these differences. For example,it is possible that different industries may have

systematically different costs of capital and lev-els of environmental exposure/environmental riskmanagement, causing a spurious correlation. Withregard to size, we know from Bansal (2005) thatlarger firms are more likely to engage in envi-ronmental management because they attract morestakeholder attention and have more resources gen-erally. As such, we believed it particularly impor-tant to control for both size and industry.

Our first control variable is the firms levelof financial leverage, since leverage affects costof capital in multiple ways. Consistent with our

market-based perspective, we used the firms levelof long-term debt as reported in COMPUSTAT,standardized by market capitalization. Financialleverage is also a theoretical variable in the testof Hypothesis 1a.

The question of whether industry membershipis a valid control variable is an empirical one.Gebhardt et al. (2001) found industry effects oncost of capital in their study looking at all pub-licly held firms. With a sample less than three


DOI: 10.1002/smj


11/24


percent of the size of Gebhardt et al. (2001), itwas ill-advised to assume that there would be anindustry effect and categorize industries in someway to test for such an assumed effect. How-ever, it is important that we assess industry effectsbecause different industries might have system-

atically different costs of capital and levels ofenvironmental exposure/environmental risk man-agement, leading to a spurious correlation betweenenvironmental risk management and WACC. Thereare a variety of ways to assess industry effects.Often authors use either the Standard IndustrialClassification (SIC) or the North American Indus-try Classification System (NAICS) code that iden-tifies each firms primary industry as a categor-ical variable. Unfortunately, because of the sizeof our dataset and the number of two-digit SICcodes (gathered from COMPUSTAT) represented

in the sample (39), adding the appropriate num-ber of dummy variables was not practical. Variousmethods exist for coding such a large number ofindustry codes into a manageable set (Amburgeyand Miner 1992). However, we worried that if wedid not use the classes as originally distributed, wewould not be able to rule out the explanation thatthe coding method itself led to any detected resultrather than it being a bona fide industry member-ship effect. As such, we treated the question ofan industry effect as an empirical one so that wecould determine if indeed there was an effect inthis sample. So using the two-digit codes we ranan analysis of variance (ANOVA) with WACC-1 asthe dependent variable and the two-digit SIC codesas the independent variable. Missing data reducedthis sample to 331. There was a significant effect,so we ran a post hoc analysis using DunnettsT3 test (Toothaker, 1992) to determine which SICcode groups were different. This analysis is morelikely to find differences among groups (be lessconservative) as it does not assume equal variancesacross cells. In this analysis, six two-digit groups

were not homogeneous with the other groups so wecreated a dummy variable with the homogeneousgroups in one category and the nonhomogeneousgroups in the other. This resulted in 129 firms inthe nonhomogeneous group and 202 in the homo-geneous group and we used this dummy variablein the analysis we describe below. Our approachallows us to identify in advance where differenceslie and to include such differences explicitly inour analysis. By doing so, we have parsed out any

industry effect on cost of capital in a rigorous andconservative way.

Finally, given that larger firms are more likely toengage in environmental risk management (Bansal,2005) because they attract more stakeholder atten-tion, have more resources generally, and have

lower costs of capital (Gebhardt et al., 2001); weincluded a measure of size in our analysis. Becausewe are working with financial market measuresin the cost of capital variables, we selected theBloomberg Financial databases measure of marketcapitalization as our size measure. While revenue,employee, and asset-based measures for size arequite common, market capitalization is the mea-sure with which financial market analysts likelywould work. Given that the measure is highlyskewed, we conducted a natural logarithmic trans-formation on it before using it in our analysis. This

reduced the skewness to conventional levels (cf.Muthen and Kaplan, 1985).

RESULTS

Tables 1 and 2 present the correlations and descrip-tive statistics, respectively, for the key variablesin our study. Even though there was no obvi-ous reason to suspect that our results might beaffected by linear dependencies across the inde-

pendent variables, we ran a colinearity diagnosticfor each equation. The VIF statistics for each inde-pendent variable were at or only slightly above 1.0(Neter, Wasserman, and Kutner, 1985), providingevidence that no variable caused undue influenceon the results because of multicolinearity.

We analyzed the data using hierarchical regres-sion because of the need to assess the marginalpredictive contribution of our theoretical vari-ables over and above that of the control vari-ables. Table 3 presents the results of the test ofHypothesis 1, which predicts a negative relation-

ship between the cost of debt capital and envi-ronmental risk management. In the analysis, thishypothesis is not supported. Our results suggesta significant positive relation between the cost ofdebt and environmental risk management, evenafter controlling for size, leverage, and industryeffects. As we predict, the cost of debt decreasessignificantly with size and increases significantlywith leverage. As one can see, while the controlvariables account for 4.2 percent of the variance


DOI: 10.1002/smj


12/24


Table 1. Correlations

WACC-1

Cost ofequity-1

Cost ofdebt

Beta Financialleverage

Log-marketcapitali-zation

Number ofshareholders

Institutionalpercentage

Cost of equity-1 0.857

N 330Cost of debt 0.290 0.168

N 330 330Beta 0.857 0.959 0.168

N 330 335 330Financial leverage 0.307 0.034 0.170 0.034N 330 335 330 335Log-market

capitalization0.193 0.281 0.114 0.281 0.056

N 330 335 330 335 335Number of

shareholders0.074 0.104(+)0.128 0.104 0.002 0.888

N 269 274 269 274 274 274Institutional

percentage0.002 0.021 0.038 0.021 0.034 0.119 0.084

N 269 274 269 274 274 274 274Environmental risk

management0.200 0.142 0.220 0.142 0.140 0.150 0.192 0.188

N 268 271 268 271 271 271 270 270

Correlation is significant at the 0.01 level (two-tailed). Correlation is significant at the 0.05 level (two-tailed).+Correlation is significant at the 0.10 level (two-tailed).

Table 2. Descriptive statistics

N Mean Std. dev.

WACC-1 330 0.09 0.04Cost of equity-1 335 0.10 0.05Cost of debt 330 0.04 0.01Beta 335 0.96 0.81Financial leverage 335 0.61 1.25Log-market cap 336 8.88 1.15Number of shareholders

(in thousands)274 426.47 227.78

Institutional holderspercentage

274 65.58 16.61

Environmental riskmanagement

546 0.00 1.00

in the cost of debt, the environmental risk man-agement indicator adds an additional five percent.Because leverage also increases with environmen-tal risk management (see below), it is possible thatwe are unable to control properly for the effect ofleverage on the cost of debt. We examined possiblenonlinearities in the cost of debt-leverage relation,but the significant positive relationship betweencost of debt and environmental risk managementpersists, so the result appears robust. Note that we

conducted this analysis with a before tax estimateof the cost of debt. As a firms effective cost ofdebt is affected by its marginal tax rate, we reesti-mated this equation with the cost of debt adjusted

for the firms marginal tax rate. The coefficientfor environmental risk management was signifi-cant at a similar level and in the same directionbut accounted for a bit less variance.

To test Hypothesis 1a, which predicts that im-proved environmental risk management wouldallow the firm to increase its leverage therebyincreasing the potential tax benefit of debt financ-ing, we use the financial leverage measure asthe dependent variable and regress the remain-ing control variables plus the environmental riskmanagement variables on it. As we see in Table 4,

the control variables account for one percent of thevariance in financial leverage, and the environmen-tal risk management scale accounts for 1.6 percentof additional explained variance. The coefficientfor the environmental risk management indica-tor is in the correct direction and significant (atp


13/24


Table 3. Hypothesis 1: Cost of debt regression results

Stage Variable Cost of debt Standardized coefficients

Beta tvalue and significance level1 (Constant) 7.629()

Financial leverage 0.162 2.671()Log-market capitalization 0.116 1.914Industry 0.017 0.286

Initial R-squared R-squared F value and significance level0.042 3.89()

2 (Constant) 8.179Financial leverage 0.132 2.225()Log-market capitalization 0.162 2.679()Industry 0.045 0.736Environmental risk management 0.238 3.813()

Test of the change R-squared change F value and significance levelin R-squared of the R-squared change

0.05 14.54()

Total R-squared 0.093

Coefficient is significant at the 0.001 level (two-tailed). Coefficient is significant at the 0.01 level (two-tailed). Coefficient is significant at the 0.05 level (two-tailed).

Table 4. Hypothesis 1A: Leverage regression results


Beta tvalue and significance level1 (Constant) 1.806

Log-market capitalization 0.022 0.363Industry 0.094 1.536

Initial R-squared R-squared F value and significance level

0.010 1.337

2 (Constant) 1.989()Log-market capitalization 0.047 0.751Industry 0.058 0.917Environmental risk management 0.132 2.079()


0.016 4.321()6



To test for a positive relation between thelevel of environmental risk management and thefirms tax advantage from debt financing as pre-dicted by Hypothesis 1b, we first compute thetax reduction (TR) associated with debt financ-ing as TR = tax ratecost of debtleverage. We

6The overall F statistic for the final equation is 2.343 which issignificant at p = 0.073

then regress TR on environmental risk manage-ment, using market capitalization and industryas our control variables (see Table 5). We findsupport for Hypothesis 1b, with the environ-mental risk management coefficient being posi-tive and significant. Thus, firms that undertakea higher level of environmental risk manage-ment reap higher tax benefits arising from debtfinancing.


DOI: 10.1002/smj


14/24


Table 5. Hypothesis 1B: Tax advantage from debt financing regression results

Stage Variable Tax advantage Standardized coefficients


Log-market capitalization 0.083 1.353Industry 0.128 2.091()

Initial R-squared R-squared F value and significance level0.026 3.53()

2 (Constant) 2.982()Log-market capitalization 0.106 1.713Industry 0.094 1.479Environmental risk management 0.124 1.955()7


0.014 3.823()8


Coefficient is significant at the 0.001 level (two-tailed).

Coefficient is significant at the 0.01 level (two-tailed). Coefficient is significant at the 0.05 level (two-tailed).

While our tests of Hypothesis 1 indicate thatthe cost of debt increases with environmental riskmanagement, the results for Hypothesis 1b indicatethat this cost increase may be offset by the higherlevel of tax reduction associated with environmen-tal risk management, driven by the combinationof higher leverage and higher cost of debt. There-fore, it is important to check how the after-tax costof debt, that is, the cost of debt net of associated

tax offsets, relates to environmental risk manage-ment. We address this question by substituting theafter-tax cost of debt for the dependent variablein Table 3 and then repeating the regression (seeTable 6). Notwithstanding the higher debt-relatedtax advantage associated with environmental riskmanagement, the net cost of debt continues to bepositively and significantly related to environmen-tal risk management and Hypothesis 1 continuesto be unsupported.

In our examination of Hypothesis 2, whichpredicts that improved environmental risk man-

agement will lead to reduced costs of equitycapital, we estimate the equations for both ver-sions of our cost of equity measure. Regard-less of the approach, (see Table 7) the controlvariables account for a significant amount ofvariance (14.7% for COE-1 and 10.3% for

7P = 0.0528P = 0.052

COE-2). The environmental risk management mea-sure also adds a significant amount of variance

(2.6% in each case). In both equations the coef-

ficients for the indicator are in the correct direc-

tion and significant. Hence, Hypothesis 2 is sup-

ported.To examine further the cost of equity results

through the three sub-hypotheses, we first test

whether beta is affected by environmental risk

management (see Hypothesis 2a). In Table 8, wesee that the control variables account for 10.3 per-cent of explained variance, while the environmen-

tal risk management measure adds 2.6 percent. The

coefficient for the environmental risk management

indicator is in the correct direction and is signif-

icant, so Hypothesis 2a is supported. As we cansee from Equation 2, beta and the cost of equity

measures are, by definition, linearly related. There-

fore, if environmental risk management leads to a

lower beta it also leads to a lower cost of equity

(Feldman, et al., 1997).To test Hypothesis 2bs prediction concerning

improved environmental risk management leadingto lower share ownership concentration (more dis-

persion), which in turn should lead to a reduced

cost of equity capital, we use the number of share-

holders at the end of the target year as our measure

of share dispersion. In this analysis (see Table 9)we see the vast majority of variance in the num-

ber of shareholders accounted for by the size


DOI: 10.1002/smj


15/24


Table 6. Hypothesis 1B: After tax cost of debt regression results



Financial leverage 0.124 2.030 ()Log-market capitalization 0.066 1.084Industry 0.061 0.984

Initial R-squared R-squared F value and significance level0.026 2.39(+)

2 (Constant) 5.924 ()Financial leverage 0.098 1.620Log-market capitalization 0.107 1.747 (+)Industry 0.004 0.071Environmental risk management 0.212 3.351 ()


0.04 11.23()


Coefficient is significant at the 0.001 level (two-tailed). Coefficient is significant at the 0.01 level (two-tailed). Coefficient is significant at the 0.05 level (two-tailed).+Coefficient is significant at the 0.10 level (two-tailed).

Table 7. Hypothesis 2: Cost of equity-1 and cost of equity-2 regression results

Stage Variable COE-1Standardized coefficients

COE-2Standardized coefficients

Beta tvalue and Beta t value andsignificance level significance level

1 (Constant) 9.651() 9.913()

Financial leverage 0.023 0.401 0.003 0.050Log-market 0.262 4.575() 0.242 4.142()capitalization

Industry 0.244 4.233() 0.182 3.101()

Initial R-squared F value and R-squared F value andR-squared significance level significance level

0.147 15.132 0.103 10.225()

2 (Constant) 9.428() 9.710()Financial leverage 0.044 0.775 0.023 0.404Log-market 0.229 3.975() 0.210 3.576()

capitalizationIndustry 0.289 4.907() 0.226 3.770()Environmental risk

management0.172 2.887() 0.170 2.805()

Test of the R-squared change F value and R-squared change F value andchange in significance level significance levelR-squared of the R-squared of the R-squared

change change0.026 8.332() 0.026 7.87()

TotalR-squared

0.173 0.129



DOI: 10.1002/smj


16/24


Table 8. Hypothesis 2A: Equity beta regression results

Stage Variable Equity beta Standardized coefficients


Financial leverage 0.003 0.050Log-market capitalization 0.242 4.142()Industry 0.182 3.101()

Initial R-squared R-squared F value and significance level0.103 10.225

2 (Constant) 7.379()Financial leverage 0.023 0.404Log-market capitalization 0.210 3.576()Industry 0.226 3.770()Environmental risk management 0.170 2.805()


0.026 7.87()



Table 9. Hypothesis 2B: Shareholder concentration regression results

Stage Variable Shareholder concentration (number of shareholders)Standardized coefficients

Beta tValue and Significance Level1 (Constant) 22.516()

Financial leverage 0.028 1.001Log-market capitalization 0.893 31.719 ()

Industry 0.021 0.739Initial R-squared R-squared F value and significance level

0.793 338.82()

2 (Constant) 22.323()Financial leverage 0.021 0.760Log-market capitalization 0.883 30.909()Industry 0.007 0.223Environmental risk management 0.055 1.873(+)


0.003 3.508 (+)Total R-squared 0.795

Coefficient is significant at the 0.001 level (two-tailed). Coefficient is significant at the 0.01 level (two-tailed). Coefficient is significant at the 0.05 level (two-tailed).+Coefficient is significant at the 0.10 level (two-tailed).

control variable (79.3%). Because of the massiveamount of variance explained by the market capi-talization variable, the environmental risk manage-ment indicator only added 0.3 percent additionalexplained variance. Nonetheless, in the equationthe coefficient for environmental risk management

is in the correct direction but only significantat p


17/24


4.9 percent additional variance. As size is so pow-erfully related to share ownership and so mod-estly related to environmental risk management(r = 0.15), it appears to have a suppressor effecton the relationship between the two. This set ofanalyses then suggests that as firms improve envi-

ronmental risk management, their share ownershipbecomes more dispersed, thus supporting Hypoth-esis 2b and the conjecture in Heinkel et al. (2001)and Mackey et al. (2007) of exclusionary environ-mental investing.

We are also able to provide modest support forthe prediction that higher levels of share owner-ship dispersion lead to lower costs of equity. Asthe reader can see from Table 1, the correlationbetween the number of shareholders and COE-1 is in the correct direction, although it is sig-nificant only at p


18/24


Table 11. Hypothesis 3: WACC regression results

Stage Variable WACC-1standardized coefficients

WACC-4Standardized coefficients

Beta tvalue and Beta tvalue andsignificance level significance level

1 (Constant) 8.961() 5.102()Financial leverage 0.332 5.892() 0.377 6.864()Log-market 0.177 3.129() 0.206 3.743()

capitalizationIndustry 0.188 3.312() 0.187 3.378()

Initial R-squared F value and R-squared F value andR-squared significance level significance level

0.169 17.95 () 0.211 23.50()

2 (Constant) 8.731() 4.844()Financial leverage 0.308 5.525() 0.357 6.532()Log-market 0.139 2.463() 0.175 3.157()

capitalizationIndustry 0.240 4.147() 0.230 4.054()Environmental risk

management

0.196 3.346() 0.162 2.827()

Test of the R-squared change F value and R-squared change F value andchange in significance level significance levelR-squared of the R-squared of the R-squared

change change0.034 16.79() 0.023 20.09()

TotalR-squared

0.203 0.234


management is associated with decreased WACC,after controlling for size, leverage, and industryeffects.

DISCUSSION

The purpose of this article is to examine an addi-tional theoretical perspective to explain the grow-ing consensus in the literature concerning the pos-itive relationship between firm environmental andeconomic performance. Using arguments devel-

oped out of risk management and institutional the-ories, we propose an external explanation for theimproved economic performance previous authorshave found. As we argue, improved environmentalrisk management is theoretically synonymous withstrategic risk management. The strategic choicesthat lead to improvements in environmental riskmanagement make the firm less risky, that is, thereare fewer potential claimants on its assets/rentsthrough potential litigation or compliance costs. In

addition, we predict that the marketplace wouldreward the firms improved risk posture throughlowered costs of debt and equity capital. Finally,we examine complementary theoretical argumentsas to why the firm would experience improvementsin the cost of debt and equity capital when it under-takes environmental risk management.

Our key contribution is that we are able to findstrong support for the basic prediction of this arti-cle that firms that develop a strategy to improvetheir risk management through improved environ-mental performance reduce their weighted average

cost of capital regardless of the way in whichwe estimate WACC. Our second set of contri-butions comes from the examination of severalarguments to see if we could determine why wefind support for our omnibus prediction. We firsthypothesize a beneficial impact of environmentalrisk management on the firms cost of debt, butobtain results contrary to prediction. Higher levelsof environmental risk management, although per-mitting firms to carry more debt, increase the cost


DOI: 10.1002/smj


19/24


of debt capital. While this cost increase is par-tially offset by higher tax benefits, the net resultis still the same. There are three alternative expla-nations for this contrarian result. First, it may bethe case that the debt markets continue to see evenstrategic investments in environmental risk man-

agement beyond that necessary for compliance asinefficient and punish firms who engage in suchbehavior. Another perspective suggested by ourdata arises from our observation that high levelsof environmental risk management are positivelyrelated to leverage. As the firm reduces its envi-ronmental risk profile through improved environ-mental risk management, debt markets appear tobe more willing to provide debt financing, whichcan improve overall organizational performance.Such increases in leverage also increase the costof debt capital, and it is possible that we are

unable to separate the effect on the cost of debt ofhigher leverage from that of higher environmen-tal risk management. Also, the increased leverageand decreased cost of capital may be key drivers inthe overall improvement in economic performancethat firms with strong environmental risk manage-ment experience. Finally, of the over $2 trillionin socially invested assets, less than one percentis in corporate debt instruments and very littleof that is screened on environmental risks. Assuch, given that institutional debt holders evalu-ate things strictly on current risk, firms making

long-term investments in environmental risk man-agement may be seen as more risky in the shortterm. Also, given that there is very little demandfor debt from socially screening investors, therewould be little or no resulting upward price pres-sure (cf. Mackey et al., 2007).

We also carry out a comprehensive examinationof the relation between cost of equity and envi-ronmental risk management. The results for ourregressions on the costs of equity measures showthat the markets amply reward increased environ-mental risk management by reducing the firms

cost of equity capital. As expected from CAPMtheory, the negative relationship between environ-mental risk management and costs of equity isconfirmed by the observed decrease in beta. Asthe firm lowers its systematic risk profile throughimproved environmental risk management, it expe-riences less volatility in its performance, and themarket appears to reward such behavior with lowercosts of equity capital and, by implication, a lowerWACC.

We decompose the cost of equity relationshipfurther through the application of risk managementtheories about the effects of improved environ-mental risk management on both share ownershipconcentration and institutional ownership. We wereable to confirm that firms with more dispersion

in the number of shareholders experienced lowercosts of equity capital, and that improved envi-ronmental risk management increases the disper-sion of shares as more individual investors wishto acquire the stock of less environmentally riskyfirms (Mackey et al., 2007). We also predict thatgood environmental performers should becomemore attractive to institutional holders resulting inmore concentration. However, as we show in ourresults, improved environmental risk managementleads to lower levels of institutional holders, butthere is no relationship between the level of insti-

tutional holdings and the cost of equity capital.The key implication of these results is that ifimproved environmental risk management leads todecreases in WACC in one period, then this low-ered cost of capital will lead to improved overalleconomic performance in later periods (e.g., Scottand Pascoe, 1984). Like any input, if firms candecrease the costs of the capital they use, theyhave the opportunity to gain higher margins oruse margins/pricing as a more effective strategicweapon. It is important to remember that the effecton economic performance is lagged. In this study

we find effects from 2001 environmental risk man-agement on 2002 WACC. As we discuss below,we examine the effect of 2002 WACC on 2002economic performance and find a nontrivial nega-tive relationship as one would predict. However wefind no direct effect from 2001 environmental riskmanagement on 2002 economic performance. Assuch, in this sample, WACC mediates the effectof environmental risk management on economicperformance.

When one examines our results in the contextof the earlier literature, several key advancements

are evident. First, we argue that in addition to theinternal efficiency effects of investments in envi-ronmental performance that have been the focusof most prior work in the area, there are exter-nal reactions to such investments by both retailand institutional investors. Our results support thisposition and show that the effects of such invest-ments are much wider than previously theorized.Second, our results help clarify why environmen-tal performance and economic performance have


DOI: 10.1002/smj


20/24


been linked in previous work. Because investmentsin environmental risk management lead to WACCreductions, the firms cost basis is reduced, andhence profits will be larger for any given incomelevel. Because of these results, we can add an insti-tutional explanation to the resource-based view

type arguments that have been proposed as theoret-ical justifications of the linkage. Finally, our resultsalso support the position taken by those normativearticles that argue that managers should improveenvironmental risk management because by doingso they will, among other things, lower the firmscost of capital.

Possible alternative explanations for the results

The development of strong theory from empir-ical results rests in part on the elimination of

alternative explanations for ones findings (Stinch-combe, 1968). We address the potential for alter-native explanations of our results in two ways.First, the use of hierarchical regression allows usto determine the marginal effect of our theoreti-cal variables after any effects of the control vari-ables that prior literature indicated should have aneffect on cost of capital. Two of the control vari-ables, size and industry, also have been shownin prior literature to predict environmental per-formance, so their inclusion is doubly importantas we try to rule out any spurious relationships

through the use of these controls. Our use of alagged design is a strength because it helps ruleout the alternative explanation that any relation-ship between environmental risk management andWACC is coincidental based on single-year datacollection.

In addition to design strategies, we examinedempirically some additional alternative explana-tions proposed by our thoughtful reviewers. Thefirst argument was that the markets are not reward-ing improved environmental risk management, butrather simply recognizing better economic per-

formance that has often been associated withimproved environmental performance. If the mar-ket was rewarding improved resource utilizationrather than lower risk with lower WACC, then wewould expect that higher levels of environmen-tal risk management in year t would be associ-ated with both lower WACC and higher economicperformance in t+1. To examine this question welooked at six 2002 firm-specific economic per-formance indicators for our dataset. We did a

lagged analysis for the economic-environmentalperformance link to be consistent with our envi-ronmental risk management-WACC analysis. Wecorrelated our 2001 environmental risk manage-ment measure and the 2002 performance measures.In none of these cases did we detect a signif-

icant correlation. There is a strong (r = 0.42,p


21/24


of its coefficient were identical. While unmea-sured causes are critical issues in any analysis, thedesign choices and additional analyses we con-ducted give us confidence that our results arerobust.9

Future research

As with most research, the results from this studyopen as many new areas of inquiry as they answer,with two questions being the most intriguing. Themost obvious future research implication is thefurther examination of the cost of debt findings.Given how strong the cost of equity and WACCrelationships were, the fact that the cost of debtresult is opposite is intriguing. While clearly debtmarkets look at risk differently than do equitymarkets, we would like to examine further why the

relationships we found are opposite of predictionand contrary to the cost of equity results.A second interesting question stems from the

fact that these results were from U.S. firms. Inmarkets where the pressure for firms to improvetheir environmental risk management is potentiallystronger (e.g., Europe and Australia) both fromregulation and from societal pressure, do theseresults hold? Specifically, we would predict thatthe equity results would be similar, but would thecost of debt results change in more environmen-tally sensitive locations?

Implications for managers

These results have implications for managers mak-ing strategic financial choices. Like all strategicinvestments, those that firms make to improvetheir environmental risk management have a costthat must be offset by commensurate benefits. Incalculating the benefits of such investments, ourresults suggest that managers can also includethe potential for reductions in costs of capi-talparticularly those who finance primarily with

equity. Given that we know from previous lit-erature that improved environmental performanceleads to increases in market value, the cost ofequity changes we identified should have far-reaching financial strategy effects.

For those firms whose strategy is to financewith debt, our results are informative as well.

9 The specifics of these additional analyses are available fromthe first author.

While investing in environmental risk manage-ment leads to higher costs of debt, the firmsthat do so are able to carry higher levels ofleverage and are able to reap more tax benefitsfrom debt financing. Depending on the specificsof the financing strategy a manager chooses, the

positive effects of investments in environmen-tal risk management may outweigh the increasesin the cost of debt. Particularly for managersunder strong pressure from competitive or insti-tutional sources to make environmental risk man-agement changes, the increases in the cost ofdebt may be simply short-term costs that mustbe absorbed for the longer-term gains that envi-ronmental improvements may generate (e.g., withstakeholders or regulators). In any case, our resultsprovide managers with more complete informa-tion with which to make their strategic financial

choices.

Potential limitations

There are two key elements to these analyses thatmay be seen as limiting the robustness or gen-eralizability of the results. First, concerns mightbe raised about our measure of environmental riskmanagement. The TRI and KLD data are limited intheir ability to reflect environmental risk manage-ment accurately. While the measures are subjectto industry differences (among other problems) we

limit the potential problems from either data sourcefirst by using them in concert (i.e., triangulating onenvironmental risk management with two types ofindicators) and by including industry differencesexplicitly.

Second, it could be argued that structural equa-tions modeling (SEM) might have provided a morerigorous test of the model. However, using SEMwas not appropriate for two reasons. First, the rela-tionships between firm size and number of share-holders, plus those among beta, cost of equity andWACC are so large that when using them all in a

structural equation, massive colinearity is unavoid-able. Such colinearity causes uncorrectable insta-bility in SEM equations. By estimating hierarchicalregressions on the different paths in the model, thecolinearity problems can be avoided or addresseddirectly. Secondly, because in this analysis wewere interested in the marginal contribution of thetheoretical variables over any variance explainedby the control variables, hierarchical regressionwas again more appropriate.


DOI: 10.1002/smj


22/24


CONCLUSION

This articles main contribution is that overall,firms that develop a strategy that improves theirtotal risk management through better environmen-tal risk management are rewarded by the financial

markets for their efforts. The main driver for thisrelationship comes from the resultant reduction inthe cost of equity, particularly through the lower-ing of the volatility of the firms stock as measuredby beta. There is also some limited effect on thecost of equity capital due to the fact that moreindividuals purchase the high environmental per-formers stock, further driving down the cost ofequity capital. While increased environmental riskmanagement seems to increase the cost of debtcapital, it does so by allowing the firm to takeon increased leverage, thereby increasing the tax

subsidies resulting from debt financing and poten-tially improving the firms overall economic per-formance. Our results suggest that in addition tothe improved resource utilization that comes withimproved environmental risk management, suchactions are legitimated (rewarded) by the equitymarkets and, in some ways, by the debt marketsas well. These findings help us build a better theo-retical understanding of the outcomes of strategicchoices to improve environmental risk manage-ment. Not only does an improved environmentalrisk management strategy result in resource effi-

ciencies, but it also has a payoff in terms of themarkets perception of the risk profile of the firmand helps explain why better environmental per-formers are also better financial performers.

ACKNOWLEDGEMENTS

The authors wish to thank Editor Richard Bet-tis, Parthiban David, Rick Johnson, Jesus Salas,Teresa Shaft, Steve Wartick, and the reviewers fortheir insightful comments on this and earlier drafts

of this article. The data collection and the firstauthor were supported in part by National ScienceFoundation grant IOC 423352. We also wish tothank the Investor Responsibility Research Centerplus Kinder, Lydenberg, and Domini for access tothe respective datasets. Finally we wish to thankDiane Keefe of the Pax World High Yield Fund forher ideas regarding costs of debt capital. Portionsof this paper were presented at the 2005 AnnualMeeting of the Academy of Management.

REFERENCES

Allen F, Bernardo AE, Welch I. 2000. A theory ofdividends based on tax clienteles. Journal of Finance55: 24992536.

Amburgey TL, AS Miner. 1992. Strategic momentum:the effects of repetitive, positional, and contextual

momentum on merger activity.Strategic ManagementJournal 13(5): 335348.Bansal P. 2005. Evolving sustainably: a longitudinal

study of corporate sustainable development. StrategicManagement Journal 26(3): 197218.

Bansal P, Clelland I. 2004. Talking trash: legitimacy,impression management and unsystematic risk inthe context of the natural environment. Academy ofManagement Journal 47(1): 93103.

Bansal P, Roth K. 2000. Why companies go green:a model of ecological responsiveness. Academy ofManagement Journal 43(4): 717736.

Barney J. 1991. Firm resources and sustained competitiveadvantage.Journal of Management 17: 771792.

Bosch JC, Eckland E, Woodrow J, Lee I. 1998. Environ-mental regulation and stockholder wealth. Managerialand Decision Economics 19: 167177.

Branzei O, Ursacki-Bryant TJ, Vertinsky I, Zhang W.2004. The formation of green strategies in Chinesefirms: matching corporate environmental responsesand individual principles. Strategic ManagementJournal 25(11): 10751095.

Buysse K, Verbeke A. 2003. Proactive environmentalstrategies: a stakeholder management perspective.Strategic Management Journal 24(5): 453470.

Campbell DT, Fiske DW. 1959. Convergent and discrim-inant validation by the multitrait-multimethod matrix.Psychological Bulletin 56: 81105.

Campbell K, Sefcik SE, Soderstrom NS. 1998. Site

uncertainty, allocation uncertainty, and superfundliability valuation. Journal of Accounting and PublicPolicy 17(45): 331367.

Carleton W, Nelson J, Weisbach M. 1998. The influenceof institutions on corporate governance through privatenegotiations: evidence from TIAA-CREF. Journal ofFinance 53: 13351362.

Chen KH, Metcalf RW. 1980. The relationship betweenpollution control record and financial indicatorsrevisited.Accounting Review 55(1): 168177.

Chidambaran NK, Fernando C, Spindt P. 2001. Creditenhancement through financial engineering: Freeport-McMoRans gold-denominated depositary shares.Journal of Financial Economics 60: 487528.

Christmann P. 2000. Effects of best practices ofenvironmental management on cost competitiveness:the role of complementary assets. Academy ofManagement Journal 43(4): 663880.

Cohen MA, Fenn SA, Naimon J. 1995. Environmentaland financial performance: are they related? Workingpaper. Investor Responsibility Research Center:Washington, D.C.

Darnall N, Edwards D Jr. 2006. Predicting the cost ofenvironmental management system adoption: the roleof capabilities, resources and ownership structure.Strategic Management Journal 27(4): 301320.


DOI: 10.1002/smj


23/24


Deephouse DL. 1996. Does isomorphism legitimate?Academy of Management Journal 39: 10241039.

DMello R, Schlinge

environmental risk management- vicram and group

Documents