Information As Regulation: The Effect of Community Right to Know Laws on Toxic Emissions*

Transcription

1 Ž. JOURNAL OF ENVIRONMENTAL ECONOMICS AND MANAGEMENT 32, ARTICLE NO. EE Informtion As Regultion: The Effect of Community Right to Know Lws on Toxic Emissions* SHAMEEK KONAR Deprtment of Economics, Vnderbilt Uni ersity, Nsh ille, Tennessee AND MARK A. COHEN Owen Grdute School of Mngement, Vnderbilt Uni ersity, Nsh ille, Tennessee Received September 12, 1995 There is growing cdemic nd policy-level interest in the use of informtion s qusiregultory mechnisms, such s toxic relese inventory Ž TRI. nd green lbels. Mndtory disclosure requirements hve been touted s mrket-bsed incentives tht my ffect firm behvior. We provide new evidence on the effectiveness of disclosure requirements by exmining firm behvior in response to disclosures of TRI emissions. We find tht firms with the lrgest stock price decline on the dy this informtion becme public subsequently reduced emissions more thn their industry peers. This is consistent with the view tht finncil mrkets my provide strong incentives for firms to chnge their environmentl behvior Acdemic Press I. INTRODUCTION Informtion remedies hve recently been touted s powerful supplements or lterntives to direct commnd-nd-control regultion. When President Clinton nd Vice President Gore recently issued their Reinventing Environmentl Regultion report 4, they continully stressed the power of informtion in bringing bout significnt chnge in environmentl qulity. Among their highest priority ction items were the estblishment of public ccess progrm to mke ll EPA dt nd publictions vilble through the internet, nd new center for environmentl informtion nd sttistics to ensure tht dt is vilble to the public. This recent interest in the use of informtion s qusi-regultory mechnism follows severl recent government progrms tht provide informtion to interested members of the public in n ttempt to ffect firm behvior indirectly through consumer, public, or community pressure. Exmples of recently encted informtion remedies include mndtory disclosure of toxic chemicl emissions Ž. which is the subject of the current pper, securities regultions requiring disclo- * This pper is bsed on Chp. 2 of Shmeek Konr s Ph.D. disserttion t Vnderbilt. Support from the Den s Fund for Summer Reserch, Owen Grdute School of Mngement is grtefully cknowledged. Helpful comments were received from J. S. Butler, severl nonymous referees, nd seminr prticipnts t Vnderbilt nd t University of Cliforni t Dvis. We lso thnk Scott Fenn nd Peter Chines t the Investor Responsibility Reserch Center Ž IRRC. for generously supplying us with their dt nd ssistnce. All views expressed re those of the uthors nd not necessrily those of IRRC or members of its stff $25.00 Copyright 1997 by Acdemic Press All rights of reproduction in ny form reserved.

2 110 KONAR AND COHEN sure of certin environmentl libilities, 1 nd Europen government sponsored green lbels. 2 Concurrently, severl uthors hve developed models of firm or consumer behvior when informtion is used s qusi-regultory mechnism Žsee, e.g., Aror nd Gngopdhyy 2 nd Kennedy et l Is informtion n effective regultory mechnism for controlling environmentl hzrds? If consumers, community groups, or investors cre bout firm s emissions, providing more firm-specific environmentl informtion my cuse consumers to djust their purchse decisions, community groups to pressure firms to reduce pollution beyond tht required by federl lws, or investors to chnge their portfolios. Thus, mndtory disclosure requirements might be viewed s form of mrket-bsed incentive for firms to chnge their behvior. In theory, ech firm will independently weigh the costs of public disclosure of bd informtion ginst the costs of tking ctions tht will put the firm in more fvorble light. Firms will voluntrily go beyond ny leglly mndted regultory stndrd if it is in their interest to do so. This will only work, however, if the public cres enough bout the informtion being relesed to punish firms tht re bd ctors. Before informtion remedies re used more frequently s regultory mechnisms, we need to understnd how they work nd wht effect they hve on firm behvior. This pper provides some initil nswers to these questions by exmining firm behvior in response to disclosures tht they were mong the worst polluters in the United Sttes. In 1986, Congress pssed the Emergency Plnning nd Community Right-to-Know Act, requiring mnufcturing estblishments with 10 or more employees in SIC codes 20 through 39 Žwith certin threshold sizes of chemicl emissions. to publicly disclose the quntity nd type of toxic chemicls relesed into the environment. The first reports were due to EPA no lter thn July 1, 1988 for toxic emissions in the clendr yer Dt from these reports hve been referred to s the toxic relese inventory Ž TRI.. According to recent pper by Cohen et l. 7, firms with lower TRI emissions outperformed their industry competitors in the stock mrket between 1987 nd The first public disclosure of TRI dt occurred on June 19, According to Hmilton 11, publicly trded firms whose TRI releses were first reported on tht dte experienced sttisticlly significnt negtive bnorml mrket returns. The impliction of this drop in stock price is tht investors updted their expecttion of future pollution-relted expenditures or libilities Že.g., probbility of ccidents, likelihood of exposure under other regultory progrms such s Superfund., which would reduce the future firm profitbility. We test this hypothesis by exmining how firms responded to this negtive stock price informtion. In prticulr, we sk whether or not significnt stock price reductions trnslte into significnt reductions in toxic emissions. The pper is orgnized s follows: Section II briefly reviews the theoreticl foundtions for our nlysis. In prticulr, it reviews why new informtion might ct s regultory mechnism nd how this intercts with the stock mrket. Section III describes the dt nd empiricl methods, while Section IV presents the min results. Section V presents some confirmtory evidence tht firms 1 See, for exmple, Accounting nd Disclosures Relting to Loss Contingencies, Stff Accounting Bulletin No. 92, 58 Fed. Reg. 32,843 Ž June 8, 1993., nd Wllce See, e.g., OECD See 42 U.S.C.A

3 INFORMATION AS REGULATION 111 responded to this initil stock mrket rection by significntly reducing their toxic emissions. Severl concluding remrks re contined in Section VI. II. THEORETICAL FRAMEWORK: WHY SHOULD INFORMATION BE AN EFFECTIVE REGULATORY INSTRUMENT? The key issue being ddressed in this pper is the extent to which publicly provided environmentl informtion cts s n lterntive tool in the plce of direct environmentl regultion. In order to be successful policy lterntive, the provision of informtion to the public must crete n incentive structure tht pressures firms to improve their environmentl performnce. If the provision of this informtion negtively impcts the finncil performnce of the firm, it will provide strong incentive to the firm to become better environmentl ctor. We hypothesize tht the chnge Ž or the lck thereof., in the finncil performnce of the firm s result of the provision of this new informtion will provide incentives tht will ffect the ttitude of the firm towrd environmentl performnce. Finncil performnce of firms my be mesured in number of wys; the two bsic clsses of mesures of finncil performnce re ccounting-bsed mesures nd stock mrket-bsed mesures. This study uses stock mrket rther thn ccounting mesures to estimte finncil performnce. The stock mrket-bsed pproch hs number of ttrctive spects. This pproch hs strong theoreticl nd empiricl foundtions in the finncil mrkets literture. The efficient mrkets hypothesis Žsee, e.g., Fm 8, 9. predicts tht in well-functioning cpitl mrket, security prices provide the best vilble unbised estimtes of the vlue of compny s ssets. Tht is, the compny s stock price fully reflects the present discounted vlue of ll future csh flows due to ll of the tngible nd intngible ssets of the firm. 4 Providing new environmentl performnce informtion to the public my hve significnt implictions for the expected future csh flows of the firm. These implictions will be discussed in detil lter in the section. As result of this new informtion investors re-evlute the firm s expected future csh flows nd ultimtely trde the firm s securities in the stock mrket, thus ffecting both stock prices nd mrket returns. This informtion is incorported into the security prices through the cretion of new rbitrge opportunities. If the new informtion tht hs been relesed indictes n increse in the expected future csh flows, then the firm s stock is currently undervlued in the mrket. Trders nd investors wre of these rbitrge opportunities will tke dvntge nd buy this stock which in turn increses the demnd for the security nd drives its price up. Thus, this new informtion will ffect finncil performnce s reflected by security prices. Stock mrket-bsed mesures of finncil performnce re forwrd looking becuse they incorporte future performnce of the firm. Future performnce is importnt in this cse becuse most of the impcts resulting from the current environmentl performnce of firm re felt in the future. For exmple, increses in expected future libility nd future investments in pollution btement cpitl stock my be required in order to reduce pollution. On the other hnd, ccounting 4 The finnce literture provides mple support for the efficient mrkets hypothesis. A review pper by Fm 9 provides n excellent bckground to the theoreticl nd empiricl evidence vilble nd concludes tht mrkets in generl djust to publicly vilble informtion.

4 112 KONAR AND COHEN dt generlly reflect historicl performnce nd re more susceptible to mngeril mnipultion. Under the efficient mrkets hypothesis, we expect ny bnorml movement in stock prices to be the result of new informtion tht chnges the expecttions of the investing public bout the future prospects of firm. Understnding prior expecttions versus public nnouncements is importnt in the cse of TRI emissions, since the public knew tht TRI emissions were going to become public well before they were ctully relesed in Indeed, the lw requiring disclosure ws pssed in 1986, nd required tht firms report their 1987 emissions to EPA. It ws not until June 1989 tht the first disclosure of these 1987 emission levels ctully took plce. Thus, by the time the first nnouncement ws mde in 1989, the mrket my hve lredy nticipted some of these future nnouncements, nd TRI emissions my hve been incorported into existing stock prices. As result, we would expect the mrket to rect more in cses when the mrket did not nticipte firm to be high polluter nd thus did not nticipte significnt reduction in firm vlue following the TRI nnouncement. 5 Why would TRI nnouncements reduce firm vlue? One possibility is tht investors use TRI emissions s signl of the firm s productive efficiency. A firm tht hs higher emissions per dollr revenue thn its competitors my be wsting resources tht ultimtely end up in the ir or wter. Thus, disclosure of high TRI emissions my be seen by investors s n indictor of poor mngement prctices nd incresed risk of spills or ccidents. Another possibility is tht there is some form of pressure from n interested stkeholder. Community pressure to reduce emissions is likely to hve n effect on ny firm tht hs significnt stke in the community. Privte prties Ž either individully or through n environmentl group. my sue firms under vrious theories of tort ctions or government-defined right of ction. Green consumers my decide to boycott products of high polluting firms or otherwise look for lterntives. 6 Finlly, the government my step up enforcement ctions ginst high polluters. Even though TRI emissions re perfectly legl, the government might trget these firms for inspections in other res nd the result might be incresed penlties nd or the cost of new pollution btement equipment. 7 Any or ll of these externl pressures my be plced on firm forced to mke new environmentl disclosures nd my result in costly expenditures to clen-up pollution or prevent future pollution. Although it is possible tht pollution btement or prevention will ultimtely sve some firms money Že.g., through lower rw mteril purchses. to offset the originl cpitl expenditure, it is lso possible tht these ctivities will only result in net drin on firm profits. Thus, firms high on the TRI list cn be expected to spend resources to ctch up with their competi- 5 A relted question is whether or not firms nticipted the importnce of TRI disclosure s soon s Congress pssed the lw Ž or while it ws being debted., nd reduced their emissions prior to 1987, the first yer of disclosure. Unfortuntely, no dt exist to test this hypothesis. 6 Although there is n obvious problem of collective ction in sustining such boycott, severl boycotts Ž such s the tun dolphin controversy. hve ffected firm behvior. As Mkower 13, pp rgues by wy of exmples, reltively smll number of individuls cn generte significnt mount of medi ttention nd bd publicity for firm. In ddition, mrket reserch hs provided some evidence of green consumer mrket segment; see Roper Indeed, EPA recently implemented strtegy of trgeting enforcement ctivities t high risk firms nd industries bsed on TRI nd other dt 19, pp. 1 3.

5 INFORMATION AS REGULATION 113 tors who re not polluting s much nd to defend themselves in costly litigtion. Investors who lern tht firm is high on the TRI list my rtionlly rect to this informtion by bidding down tht firm s stock price. Following this logic, the flip side of the coin is lso true. If firm s stock price is bid down becuse it is high polluter, mngement hs strong incentive to reduce pollution nd strengthen firm vlue in subsequent yers. 8 Incresed pressure due to new informtion bout environmentl hzrds my be focused either on n entire industry or within specific firms in n industry. In some cses, consumers my look t lterntive product lines tht re less polluting or otherwise chnge their ggregte consumption hbits to reduce the demnd for n entire product ctegory. In this cse, the negtive impct following environmentl disclosures is industry-wide. Corresponding to this effect, however, might be positive impct on other industries whose products now look reltively more fvorble. The extent to which this new informtion is likely to ffect industry-wide vlution will depend on the vilbility of close substitutes, the price elsticity of demnd for tht product, nd other elsticity fctors relted to the demnd for green products. Similrly, new informtion on environmentl hzrds my hve firm-specific effects s some firms within n industry now look more or less fvorble thn others in tht sme industry. Even if ggregte demnd for product line is not ffected Ž or even positively ffected. by this new disclosure, differentil emission rtes cross firms my crete new winners nd losers in tht industry. III. EMPIRICAL METHOD Since we re interested in the effect of new informtion on subsequent firm behvior, we identified the first public nnouncement of TRI dt, June 19, On tht dte, the EPA relesed TRI dt for the clendr yer Subsequent informtion hs shown tht 1987 dt re prticulrly unrelible, nd most studies of ctul TRI emissions strt with 1988 s the bse yer. However, since we re primrily interested in the effect of stock price movements on subsequent firm behvior, we begin with the June 19, 1989 nnouncement dte even though the dt relesed on tht dte ws subsequently found to be unrelible. Wht mtters most is the fct tht on June 19, 1989, the dt ws thought to be resonbly ccurte, nd it ws relied upon by vrious stkeholders. Becuse of the time lg between ctul emissions nd public vilbility of TRI dt, we begin with 1989 emissions s the bse yer for our nlysis. Since we re interested in wht firms did following the TRI reports mde in 1989 Žfor 1987 emissions., we tke ctul 1989 emissions s the bse yer. Tht is the strting 8 Although t first glnce the stock price hit might be considered sunk cost, there re mny resons why mngement would view it otherwise. The stock price hit might be thought of s n indictor of serious problems in the firm tht will require expensive chnges in production processes or impose future libility, or tht re simply n indictor of bd mngement. Reduced firm vlue might lso increse the future cost of cpitl or cuse the firm to become potentil tkeover trget. Finlly, the fct tht the mrket rected in this mnner to TRI disclosure tells mngement tht this is now n importnt criteri to investors, nd thus mngers might tke this s new informtion for them to use in deciding the reltive trdeoff between pollution nd profits. Regrdless of the direction of custion, we expect to observe firms tht took significnt stock price hit to reduce their TRI emissions in subsequent periods.

6 114 KONAR AND COHEN point from which firms hd to begin to reduce TRI emissions subsequent to the first disclosure. We compre 1989 emission to 1992 levels three yer lg to llow for firm investment in pollution btement progrms. In ddition, in order to smooth out ny one-yer berrtions Ž nd potentil TRI reporting errors., we lso compre the verge TRI emissions for two time periods: nd Although the initil relese of dt included ll U.S. fcilities, the EPA did not provide this informtion in formt conducive to firm-level comprisons. Insted, severl environmentl groups compiled this informtion to rrive t the worst corporte polluters. In other cses, the public ws informed of prticulr compny s emissions when newspper reporter identified prticulr fcility or compny s being high polluter. Some of these nnouncements took plce subsequent to the initil TRI publiction dte of June 19. In order to determine which firms received publicity in the medi s result of the nnouncement of the TRI, we serched both the Wll Street Journl Index nd the LEXIS NEXIS dtbse for ll mentions of toxic chemicls or TRI in This serch yielded 363 fcilities nd compnies tht were mentioned in the print medi s being emitters of toxic substnces in A mjority of these medi reports were round the dtes of relese of two reports. The first report ws dted June 19, 1989 Žthe sme dy on which the EPA mde the TRI dtbse vilble to the public., nd compiled by the Nturl Resources Defense Council 14 Ž NRDC.. This publiction reported the relese of crcinogenic chemicls Ž subset of the TRI chemicls., nd ws often the impetus for medi reports implicting vrious firms s emitters of crcinogens. The second publiction ws issued by the Ntionl Wildlife Federtion 15 Ž NWF. on August 10, Unlike the NRDC report, which looked t crcinogens, the NWF report delt with ggregte releses of ll toxic substnces s listed under section 313 of the EPCRA. Thus, compnies receiving medi ttention round this time were the lrgest toxic chemicl polluters. Of the firms nd fcilities receiving medi coverge in 1989, 53% were mentioned fter the NRDC report between June 19 nd June 21, 1989; 29% followed the NWF report on August 10 nd August 11, 1989; nd the remining 18% were first mentioned in the medi on other dtes tht yer. Approximtely 59% of the fcilities or plnts referred to by the medi were operted by publicly trded firms, the rest were owned either by privte corportions or were single fcilities. After ccounting for multiple mentions to the sme prent compny, 192 publicly trded firms were found in the smple. 9 Stock price dt for 130 of these firms were vilble from the Center for Reserch in Security Prices Ž CRSP. dt tpes nd formed the core of our study Hmilton 11, Tble 3 reports on the determinnts of medi coverge. He finds higher likelihood of medi coverge mong firms with lrger emissions, more TRI reports, nd more employees. Firms with mny fcilities spred out over the country were less likely to be reported on thn those with lrger concentrted fcilities. Finlly, firms in the chemicl nd primry metls industry were less likely to be reported on even though they hd lrge emissions, presumbly becuse they were lredy expected to hve high TRI emissions, wheres firms in the pulp nd pper industry Žwhose emissions were higher thn expected. were more likely to be reported on. 10 These 130 firms represent ll NYSE nd AMEX compnies with complete stock price dt vilble on CRSP out of the 192 publicly trded firms with TRI-relted medi reports in We did not include the NASDAQ compnies in our dt set since mny of them re not s frequently trded nd re less suitble for study of dily stock price returns.

7 INFORMATION AS REGULATION 115 Mny firm-specific events like this re best studied by evluting their effect on the firm s security prices. This technique is known s n event study nd uses mrket model prediction errors to test hypotheses Žsee Fm 10 nd Brown nd Wrner. 3. The bsic pproch is to control for the systemtic or mrket risk in the security returns by estimting the mrket model. This gives us the individul risk coefficients Ž bets. which re then used to clculte the prediction errors on the event dy. These prediction errors, lso known s bnorml returns Ž. it, re firm specific returns tht remin fter controlling for the movement of the mrket on the event dy nd hve n expected vlue of zero under the mrket model: Rit i br i mt it, it NŽ 0, s 2.. Whether n event hs significnt effect on firm or not my be nlyzed by exmining the sttisticl significnce of the bnorml returns. In this cse the provision of new informtion to the public bout the environmentl performnce of the firm my impct the future finncil performnce of the firm which would be reflected in the firm specific bnorml returns ccruing to the firm on the dy of the event. The bnorml returns were then clculted for these firms using event study methodology. The firm bets were estimted using the mrket model EŽ Rit. i br i mt, where Rit is the individul firm return for tht time period, i is the constnt term, bi is the bet for firm i nd Rmt is the mrket return t time t. Dily returns for 240 dy period, beginning with 250 trding dys before the event nd ending 10 dys prior to the event, were used to estimte the bets. These bets were then used in the mrket model to clculte the bnorml returns for the event period. We clculted the bnorml returns for 20-dy window round the event dy. The event dy hs been defined s the dy the firm ws first mentioned in the print medi. 11 As shown in Tble I, firms tht were mentioned in the medi s being toxic chemicl emitter received significnt stock price reductions on the dy of the nnouncement. The first column is tken directly from Hmilton 11, which indictes tht ll publicly trded firms with TRI emissions received significnt stock price reduction on the dy of nnouncement Ž 0.284%., lthough there ws no price effect on the dy prior to nnouncement. This suggests lck of informtion lekge prior to the initil press reports on June 19, The second column repets this nlysis for the 130 firms in our smple tht hd some firm-specific medi coverge Ž 0.299%.. Once gin, we find no significnt informtion lekge on the dy prior to nnouncement. The third column of Tble I exmines the 40 firms in our smple of 130 tht received the lrgest negtive bnorml returns, n verge of 1.324% on the event dy. 12 These firms represent bout one-third of ll firms with medi reports 11 Note tht this dte is not necessrily the sme for ech firm, s medi stories ppered t vrious times throughout the yer. Insted, these dtes represent the first time medi story ppered bout tht compny s toxic emissions. Thus, dy 0 is designted s the dy the compny ws mentioned in the medi nd ll firms re ligned using their event dy. 12 We dropped two other firms from our smple tht hd higher bnorml returns, both of which lso hd confounding events on the dy of the TRI nnouncement. Time Wrner ws involved in significnt tkeover ctivity during this time period. LTV Corp. ws involved in $2 billion dispute over unfunded pension libilities following their 1986 bnkruptcy. The Supreme Court greed to her the cse on the dy of the TRI nnouncement. Thus, these two firms were dropped from the smple to ensure we did not incorrectly ttribute bnorml returns to TRI nnouncements.

8 116 KONAR AND COHEN TABLE I Abnorml Returns Following Disclosure of TRI Emissions Window All firms in TRI Firms mentioned Top 40 firms in terms Ž dys. dtbse in the medi of event dy returns Ž Ž Ž Ž Ž Ž Ž Ž Ž Number of observtions Ž. Note. First column tken directly from Hmilton Z sttistics re reported in prenthesis. Sttisticlly significnt t p in 1989, nd their verge bnorml returns is four times the men Ž 1.3% versus bout 0.3% for the entire smple.. 13 The smllest bnorml return in the TOP 40 smple is 0.6%, twice the verge for the full smple of 130 firms. Abnorml returns re normlly reported in percentge terms to control for firm size ŽBrown nd Wrner. 3. A few million dollrs mens virtully nothing to IBM, but plenty to compny whose mrket vlue is only $50 million. Nevertheless, these percentge losses trnslte into quite significnt bsolute dollrs, with men loss of $87 million, medin loss of $29 million, nd rnge of $3.6 to $841 million for our TOP 40 smple. We use this TOP 40 smple throughout the reminder of the pper to exmine the impct of significnt negtive stock price rections on firm behvior. The smple selection process blnces our need for smple of compnies tht hd very significnt stock price reductions Žnd hence. 14 re likely to rect by reducing emissions nd for resonbly lrge smple size. Although we refer to these firms s the TOP 40 for ese of exposition, this lbel refers to the stock price hit they took, not necessrily to the level of TRI emissions. In fct, of the TOP 40 firms with negtive bnorml returns, only 11 were lso mong the top 40 lrgest TRI emitters Žhereinfter referred to s LARGEST EMITTERS. in terms of bsolute levels of pollutnts. This is consistent with our erlier conjecture tht mny of the lrgest TRI emitters would not receive significnt stock price reductions upon disclosure of their emissions, since the mrket ws lredy expecting these firms to be mong the lrgest emitters. Tht is, the negtive stock price rection hd lredy been cpitlized in the mrket vlue of the lrgest emitters by the time ctul emissions were nnounced in June The smllest bnorml returns chosen hd significnce level t bout 25%, indicting there ws only 25% chnce tht the bnorml returns were drwn from smple with zero men. The men of 1.34% hs z-sttistic of 9.23 which hs negligible probbility of hppening if the true men were This pper is primrily concerned with whether or not negtive stock price effect induces firms to reduce emissions. Note tht there re vrious pressures exerted upon firms to reduce emissions Že.g., community pressure, consumers, cost svings., nd most firms hve reduced their TRI emissions over time. Thus, in this study, we focus only on those firms tht received sttisticlly significnt stock price hit nd compre them to their industry peers. A broder study of why firms reduce TRI emissions would require different methodology nd dt. Insted, we focus on one of the resons for TRI emission reductions the finncil mrkets. 15 Additionl evidence with regrd to the vlidity of this hypothesis is discussed in Section V.

9 INFORMATION AS REGULATION 117 Since we wnt to exmine the effect of significnt stock price reductions on subsequent firm behvior, we compre TRI emissions mong the TOP 40 to tht of their industry peers. Given the discussion bove concerning inter- versus intrindustry effects, nd the fct tht the TOP 40 compnies re primrily from trditionlly polluting industries, comprison smple of firms from clener industries would not llow us to drw legitimte inferences on the behvior of firms in the TOP 40. Thus, insted of compring the TOP 40 to ll publicly trded compnies, we dopt n industry-weighted pproch. To construct this comprison smple, we first determined the Stndrd & Poors industry clssifiction for ech of the TOP 40 compnies bove. We obtined list of 1500 publicly trded firms from IRRC Žwhich lso contined TRI nd relted environmentl informtion., 16 of which 942 were in industries where t lest one firm reported TRI emissions. Of these 942 compnies, 512 were in the sme industry ctegory s one of our TOP 40. After excluding few compnies with missing dt nd the TOP 40, totl of 455 compnies remined in our comprison group. Finlly, ech firm in the smple of 455 compnies ws given weight so tht we could replicte the industry clssifictions of the TOP 40. For exmple, since the TOP 40 bnorml returns smple included two steel compnies, firms in the steel industry were given weight of 1 20 for purposes of constructing the industry-weighted smple. 17 Since firms vry drmticlly by size nd opportunities to pollute, ny mesurement of environmentl performnce must somehow djust emission levels so tht firms cn be compred on n equl footing. Unless we explicitly control for firm size, ny results would be extremely misleding. For exmple, multi-billion dollr firm with mssive emissions could reduce emissions by smll percentge nd become the firm with the lrgest bsolute emission reductions while still being one of the most inefficient polluters. Thus, we dopt two mesures of performnce: Ž. 1 the bsolute level of emissions per thousnd dollrs revenue, nd Ž. 2 firm rnk within its industry ctegory Ž normlized by the number of firms in the industry., where rnks re determined by the level of emissions per dollr revenue. 18 In ddition to TRI dt, we lso obtined two other mesures of environmentl performnce from the IRRC dtset: oil nd chemicl spills nd governmentimposed fines for environmentl violtions. We collected this informtion to 16 IRRC selects this list of 1500 publicly trded firms to include both the Stndrd nd Poors 500 nd the Fortune 500 compnies, s well s the lrgest cpitliztion stocks. Thus, the firms in our popultion re mong the lrgest nd most visible publicly trded compnies in the United Sttes. Since we re interested in the effect of stock price declines on subsequent firm behvior, this is n pproprite popultion for our study. Moreover, we re interested in compring similrly situted compnies. Thus, it is importnt to compre firm tht took significnt stock price hit with one tht could potentilly hve received the sme hit in the mrket. It lso llows us to ignore different underlying incentives nd governnce structures of privtely held orgniztions. 17 Note tht lthough we report the industry-weighted smple of 455 firms throughout the pper, we lso compred the TOP 40 to the unweighted full smple of 873 firms with TRI dt, s well s to smller mtched smple of 80 firms. The results reported in Tbles III nd IV remin essentilly unchnged regrdless of the comprison group. 18 An lterntive pproch is to exmine the percentge reduction in emissions. The problem with this mesure of performnce is tht, empiriclly, percentge chnges of TRI vry drmticlly, nd often for resons tht hve little to do with firm rection to the mrket. For exmple, firm might expnd by cquiring or building new plnt or merging with n existing firm. In such cse, TRI emissions might double or triple. Alterntively, firm tht closes plnt might cut its emissions by hlf. But tht firm is lso producing less output. Given the huge vrition in percentges, we opted for our mesure of emissions per dollr revenue.

10 118 KONAR AND COHEN exmine the extent to which there is ny positive spillover between TRI emission reductions nd other res of environmentl performnce. If firms reduce their TRI emissions s result of receiving negtive bnorml returns, do they lso tke ctions to reduce the probbility of other bd environmentl news? Even though spills re primrily ccidents, firms cn tke ctions ex nte to reduce their likelihood nd severity Žsee, e.g., Cohen. 6. Thus, reduction in the number nd the mount of spills my indicte better environmentl mngement techniques nd tht the firm is ttempting to improve its environmentl performnce. The other trend vrible is the dollr vlue of fines per 1000 dollrs revenue. Although government imposed penlties re certinly n indiction of environmentl performnce, they re lso prtly bsed on discretionry nd or rndom enforcement. Both vribles re mesured s the verge of two different time periods, nd The next section exmines TRI emissions nd the rnking of firms within their industry ctegory, both before nd fter public nnouncement of TRI dt. IV. RESULTS Tble II lists the TOP 40 firms tht received the lrgest negtive bnorml returns in 1989 s result of TRI nnouncements. Among these TOP 40 firms, Owens Corning Fiberglss Corp. received the lrgest negtive bnorml returns Ž 2.7%., while McDonnell Dougls Corp. received the smllest stock price hit Ž 0.6%.. Next, firms re rnked within their industry ctegory on the bsis of pounds of TRI emissions per thousnd dollrs revenue. For exmple, in 1989 Owens Corning Fiberglss Corp. ws the 7th lrgest TRI emitter in the Building Mterils industry on per dollr bsis out of the 19 firms in the Building Mterils industry, while in 1992 it ws the 2nd lrgest emitter per dollr revenue. These rnkings re not bsed on bsolute TRI emissions becuse lrger firms would tend to hve higher emissions nd hence bis the rnkings. The number of firms in ech industry is tken from the list of 1500 publicly trded firms in the IRRC dtset. As shown in Tble II, the men rnk for TOP 40 compnies ws 7.20 in The men normlized rnk in 1989 Žcompny rnk divided by industry size. ws 0.339, indicting tht firms in the TOP 40 were in the top 1 3 of their industry in terms of TRI emissions per dollr revenue. 19 The verge rnk in 1992 ws 7.85, while the normlized rnk ws 0.396, indicting tht these firms becme reltively lower TRI emitters following the public nnouncement of TRI dt, both in bsolute nd reltive terms within their industries Although these firms were mong the lrgest TRI $ emitters in their industry ctegories, they were not necessrily the lrgest emitters. Only 11 of the TOP 40 firms tht received significnt negtive bnorml returns were lso mong the 40 LARGEST EMITTERS in terms of bsolute pollution volume. 20 The chnge in bsolute rnking is not significnt Ž p This is not relly n pproprite mesure of comprison, however, since the size of industries rnge from four to 53 firms. Thus, two unit chnge in rnking, for exmple, might indicte n importnt shift in reltive performnce in n industry with four firms, but only minor chnge in n industry with 53 firms. Thus, more pproprite mesure Ž nd the one used in subsequent empiricl nlyses. is the normlized rnking, rnk divided by number of firms in the industry. As shown lter in Tble V, the improvement in normlized firm rnking is not sttisticlly significnt t norml levels of significnce Ž p

11 INFORMATION AS REGULATION 119 TABLE II Chnge in TRI Industry Rnk for Top 40 Compnies with TRI Relted Abnorml Returns Industry Industry Chnge Chnge Rnk by rnk rnk in rnk in TRI $ Compny bn. ret. Industry b Owens-Corning Fiberglss Corp. 1 Building Mteril Qunex Corp. 2 Steel Cooper Industries Inc. 3 Mchinery Crown Cork & Sel Co. Inc. 4 Continers Ž Metl Armco Inc. 5 Steel Estmn Kodk Co. 6 Photogrphic Thoms Industries Inc. 7 Electricl Equip Hewlett-Pckrd Co. 8 Electronics Instr Rhone-Poulenc Rorer 9 Drugs Costl Corp. 10 Nturl Gs Generl Motors Corp. 11 Auto Phelps Dodge Corp. 12 Metls Springs Industries Inc. 13 Textiles Scott Pper Co. 14 Household Prod Jmes River Corp. of Virgini 15 Pper Illinois Tool Works Inc. 16 Mnufcturing Inlnd Steel Industries Inc. 17 Steel Eton Corp. 18 Hevy Duty Truck Interntionl Business Mchines 19 Computers Bethlehem Steel Corp. 20 Steel Mytg Corp. 21 Household Furn Whirlpool Corp. 22 Household Furn Unocl Corp. 23 Oil Prker Hnnifin Corp. 24 Mnufcturing Olin Corp. 25 Chemicls Chrysler Corp. 26 Auto Generl Electric Co. 27 Electricl Equip Allied-Signl Inc. 28 Mnufcturing Rytheon Co. 29 Aerospce Lockheed Corp. 30 Aerospce Boeing Co. 31 Aerospce Upjohn Co. 32 Drugs Ferro Corp. 33 Specilty Chem Atlntic Richfield Co. 34 Oil Westvco Corp. 35 Pper Mobil Corp. 36 Oil Armstrong World Industries Inc. 37 Household Furn Union Cmp Corp. 38 Pper Federl Pper Bord Co. Inc. 39 Pper McDonnell Dougls Corp. 40 Aerospce Men vlues Normlized industry rnk Ž compny rnk industry size Positive number indictes improvement reltive to industry; rnk 1 for worst polluter. b Negtive number indictes improvement through reduced emissions.

12 120 KONAR AND COHEN TABLE III Comprison of TRI $ nd Industry Rnk for Top 40 Firms nd Industry Peers Top 40 firms t test vlues rnked by Industry compring the event dy weighted top 40 firm bnorml smple of mens with returns firms industry peers Smple size TRI $ revenue in TRI $ revenue in Chnge in TRI $ revenue t test vlues compring 1989 to 1992 b within ech ctegory Note. t sttistics re reported bsed on one-tiled tests. Significnt t p b Significnt t p The sixth column of Tble II reports the chnge in industry rnk between 1989 nd A positive number indictes tht the firm lowered its rnk in terms of TRI $, nd is now reltively clener firm thn before. Of the 40 firms shown, 21 lowered their rnkings, 7 firms remined unchnged, nd 12 incresed their reltive TRI emissions. The lst column shows the bsolute chnge in TRI emissions per dollr of revenue, with negtive number indicting reduction. Of the 40 firms, 32 reduced their TRI $ revenue, while only 8 firms incresed emissions. The verge firm reduced emissions by 1.84 pounds per thousnd dollrs revenue. It is cler from Tble II tht the 40 firms who received the most significnt negtive stock price rections following nnouncement of TRI emissions significntly lowered both their TRI $ emissions nd their TRI $ rnking within their industry. As further evidence, nd to help rule out the possibility of other unknown fctors cusing the TRI reductions, Tble III compres industry rnkings nd TRI $ reductions of the TOP 40 firms to our industry-weighted smple of 455 compnies. TRI $ revenue for the TOP 40 firms ws nerly twice tht of the industry-weighted smple in 1989 Ž p However, by 1992, the TOP 40 firms hd reduced their TRI $ revenue nerly in hlf, so tht the difference between those firms nd the industry-weighted smple ws no longer sttisticlly significnt. While the TOP 40 firms reduced TRI $ revenue by 1.84 pounds per thousnd dollrs revenue, 21 the industry-weighted smple hd only reduced their TRI $ revenue by 0.17 pounds per thousnd. Since the nlysis bove only compred two yers of dt Ž 1989 nd 1992., we lso compred verge TRI $ revenue over two different time periods in order to smooth out ny errors in reporting or nnul fluctutions. Tble IV reports on the verge TRI $ revenue during the time periods nd The results re eqully striking, with verge TRI $ revenue decresing from 4.48 to 21 As noted in Tble III, this reduction Ž from 4.38 to 2.54 pounds per thousnd dollr revenue. is sttisticlly significnt t p Although the results shown in Tble IV re bsed in current dollrs, there is virtully no difference once they re converted to constnt dollrs.

13 INFORMATION AS REGULATION 121 TABLE IV Comprison of Environmentl Performnce Trends Pired t test Averge Averge for vlues for period period compring Ž 1992 for TRI. two time periods TRI $ revenue TOP 40 Industry weighted smple Ž n Spills $ revenue TOP 40 Industry weighted smple Ž n Fines $ revenue TOP 40 Industry weighted smple Ž n Note. t sttistics re reported bsed on two-tiled tests. Significnt t p in the TOP 40 group Ž p 0.01., compred to reduction from 2.43 to 2.14 Ž p for the industry-weighted smple. Tble IV lso reports on two other mesures of environmentl performnce: oil nd chemicl spills, nd government-imposed fines. 23 While the verge volume of spills per dollr sles decresed in the TOP 40 group Ž from 4.00 to 3.14, p 0.10., it ctully incresed slightly in the industry-weighted smple Žfrom 2.04 to 2.15, p Trends in government-imposed fines re little more complicted to ssess, since they not only serve s mesure of firm environmentl performnce, but lso reflect chnges in government enforcement progrms. Both the TOP 40 firms nd the industry smple increse their fines $ revenue over time, with verge fines being higher for the TOP 40 firms. However, only the increse in industry-weighted fines is sttisticlly significnt Ž p The fct tht fine levels increse for both smples my be n indiction of chnges in EPA penlty nd or enforcement trends. 25 V. CONFIRMATORY EVIDENCE ON THE ROLE OF INFORMATION We hve rgued tht the reson for significnt reduction in stock mrket vlution following relese of TRI dt is the fct tht this is new informtion nd 23 Unlike the TRI trend dt which compre nd , these dt compre nd Since there is longer lg in reporting TRI dt, it is not s current s spill dt nd enforcement dt. 24 Note tht there re more thn twice s mny firms with incresed fines $ thn decresed fines $ mong the industry-mtched smple. This rtio is bout 1.5 for the TOP 40. Using Wilcoxon sign test, we find tht the industry-mtched smple hs significnt number of increses t p 0.01, wheres the TOP 40 smple does not hve sttisticlly significnt number of increses Ž p Thus, even if enforcement trends nd fine levels re incresing for the TOP 40 group, they re incresing t much lower rte thn their industry peers. 25 Cohen 5 nd U.S. Environmentl Protection Agency 19 provide evidence of incresing monetry penlties nd enforcement trends during this time period.

14 122 KONAR AND COHEN ws unnticipted. Thus, if firm ws expected to be the highest polluter nd ws subsequently found to be so, we re likely to see little or no chnge in stock mrket vlution on the ctul dte of TRI. Tble V compres the TOP 40 group to the 40 LARGEST EMITTERS. As noted erlier, we found tht only 11 of our TOP 40 were lso mong the 40 LARGEST EMITTERS. The 40 LARGEST EMITTERS were lso lrger emitters in terms of TRI $. Wheres the TOP 40 smple hd verge emissions of 4.38 pounds per thousnd dollrs revenue in 1989, the 40 LARGEST EMITTERS verged pounds per thousnd dollrs revenue. Moreover, while the TOP 40 significntly reduced their TRI $ Ž from 4.38 to 2.54, p 0.10., the LARGEST EMITTERS did not reduce TRI $ t sttisticlly significnt level Žfrom to 10.05, p This sme result crries over to the industry TRI $ rnkings, where the normlized rnk Ž in terms of TRI $. of the 40 LARGEST EMITTERS remined virtully unchnged Ž to , while the rnking for the TOP 40 smple improved from 0.34 to 0.40 Ž p These findings re consistent with the view tht medi ttention nd resultnt stock price effects hve more of n effect on subsequent firm behvior thn simply being known s the lrgest emitters in the United Sttes. These results re consistent with our hypothesis tht the mrket rected more to unexpected TRI disclosures thn to those tht were lredy expected to be very lrge. Further evidence is presented by Hmilton 11 in n nlysis of the dollr vlue of bnorml returns on the dy of TRI disclosure. After controlling for size of firm, he found positive correltion between the number of existing Superfund sites Ž lredy known to the public. nd bnorml returns, which is consistent with the mrket lredy nticipting tht these firms would be mjor polluters. He lso found little correltion between industry dummies for the five most polluting TABLE V Comprison of TRI $ nd Industry Rnk for Top 40 Abnorml Returns nd Top 40 Emitters Top 40 firms rnked by event dy bnorml returns Top 40 firms rnked by bsolute TRI levels TRI $ revenue in TRI $ revenue in Chnge in TRI $ revenue Ž t sttistic in prenthesis. Ž Ž Normlized industry rnk in Ž compny rnk industry size. Normlized industry rnk in Ž compny rnk industry size. Decline in normlized industry rnk Ž Ž Ž t sttistic in prenthesis. Note. The top 40 firms rnked by bsolute TRI emissions hve higher TRI $ nd higher TRI $ normlized industry rnkings thn the top 40 rnked firms rnked by bnorml returns, in both 1989 nd 1992 Ž p t sttistics re reported bsed on one-tiled tests. Significnt t p 0.10.

15 INFORMATION AS REGULATION 123 industries Ž in terms of ggregte toxic emissions. nd bnorml returns, suggesting tht investors lso expected these industries to be highly polluting. 26 Finlly, lthough most of the 130 publicly trded compnies with medi nnouncements of TRI emissions in 1989 hd negtive Ž or essentilly zero. bnorml returns on their respective event dys, some ctully hd positive bnorml returns. Twenty-two firms hd positive bnorml returns with t sttistic t the 25% level or below Žthe sme criteri used in generting the TOP 40 smple of compnies with negtive bnorml returns.. Of these 22, we hd to eliminte three tht did not hve subsequent environmentl dt to compre. Thus, we constructed smple of 19 firms with the lrgest positive bnorml returns Žhereinfter lbeled POSITIVE 19. comprble to our TOP 40 negtive returns. Although the smple size is smll nd we hve not controlled for industry chrcteristics, we cn mke few tenttive comprisons. Wheres the TOP 40 firms hd n verge normlized industry rnk of in 1989, the POSITIVE 19 firms were rnked Thus, the POSITIVE 19 were reltively low emitters in their industry groups compred to the TOP 40 in 1989, lthough the difference is only significnt t p By 1992, the TOP 40 group cught up with the POSITIVE 19, with the normlized rnk being for the TOP 40 nd for the POSITIVE 19. This is consistent with our hypothesis tht the mrket rected to new informtion some of which positively ffected firms tht looked more fvorble thn their industry peers nd more fvorble thn investor s expecttions. VI. CONCLUDING REMARKS This pper hs exmined firm behvior in response to significnt stock mrket rection to new informtion on toxic chemicl emissions. We identified ll firms with significnt negtive bnorml returns upon the public nnouncement of their TRI emissions in Firms with the lrgest negtive stock price effects following nnouncement of their TRI emissions were found: Ž. 1 to be mong the top 1 3 of polluting firms Ž per dollr revenue. in their industries; Ž 2. not to be the lrgest bsolute TRI emitters; Ž. 3 to subsequently reduce their TRI emissions more thn other firms in their industry Žincluding those firms with the lrgest TRI $ revenue prior to the disclosure of TRI levels.; Ž 4. to lso mke other significnt ttempts t improving their environmentl performnce by reducing the number nd severity of oil nd chemicl spills; nd Ž. 5 to hve lower chnce of receiving higher fines from the government in subsequent yers. We hve provided prtil nswer to the fundmentl question concerning the role of informtion in reducing the emission of environmentl contminnts. New informtion concerning firm s toxic emissions tht hs significnt effect on mrket vlution is likely to induce tht firm to significntly reduce subsequent emissions nd to otherwise improve its environmentl performnce. Thus, in some instnces, providing informtion to the public my be n effective remedy to reduce environmentl externlities beyond regultory stndrd. Nevertheless, severl importnt questions re left unnswered. Are these reduced emissions the worst pollutnts for these firms? Are these emissions the most cost-effective 26 The only significnt industry dummy ws primry metls, which ctully hd positive coefficient. Thus, the mrket my hve been surprised tht this industry ws not s lrge polluter s expected.

16 124 KONAR AND COHEN ones to reduce or lterntively, re they simply those tht re most subject to public scrutiny? 27 Do the stock price nd subsequent emissions reductions represent responses to public pressure environmentl concerns or to previously inefficient production processes? Wht other fctors cuse firms to reduce emissions beyond the regultory stndrd? Are mngers expecttions tht decresing TRI emissions following significnt stock price hit relized; in other words, does it py these firms to reduce TRI? Finlly, do the socil benefits of this form of regultion exceed its costs, nd does it dominte direct regultion or other forms of mrket-bsed incentives? Answers to these questions must wit further reserch. REFERENCES 1. S. Aror nd T. N. Cson, An experiment in voluntry environmentl regultion: prticiption in EPA s progrm, J. En iron. Econom. Mngement, Ž S. Aror nd S. Gngopdhyy, Towrd theoreticl model of voluntry overcomplince, J. Econom. Beh. Orgniz Ž S. J. Brown nd J. B. Wrner, Using dily stock returns: The cse of event studies, J. Finncil Econom Ž B. Clinton nd A. Gore, Reinventing Environmentl Regultion Ž Mrch 16, M. A. Cohen, Environmentl crime nd punishment: Legl economic theory nd empiricl evidence on enforcement of federl environmentl sttutes, J. Criminl Lw Criminology, Ž M. A. Cohen, Optiml Enforcement Strtegy to Prevent Oil Spills: An Appliction of Principl- Agent Model with Morl Hzrd, J. Lw Econom Ž M. A. Cohen, S. A. Fenn, nd J. S. Nimon, Environmentl nd Finncil Performnce: Are They Relted? Investor Responsibility Reserch Center, April E. F. Fm, Efficient Cpitl Mrkets II, J. Finnce Ž E. F. Fm, Efficient Cpitl Mrkets: Review of Theory nd Empiricl Work, J. Finnce Ž E. F. Fm, L. Fisher, M. Jensen, nd R. Roll, The Adjustment of Stock Prices to New Informtion, Internt. Econom. Re., 21 Ž J. T. Hmilton, Pollution s News: Medi nd Stock Mrket Rections to the Toxics Relese Inventory Dt, J. En iron. Econom. Mngement Ž P. W. Kennedy, B. Lplnte, nd J. Mxwell, Pollution Policy: The Role for Publicly Provided Informtion, J. En iron. Econom. Mngement Ž J. Mkower, The E-Fctor: The Bottom-Line Approch to Environmentlly Responsible Business, Penguin Books, New York, N.Y. Ž Nturl Resources Defense Council Ž NRDC.. The Who s Who of Toxic Polluters Ž Ntionl Wildlife Federtion, The Toxic 500 Ž OECD, Environmentl lbeling in OECD Countries, Orgniztion for Economic Coopertion nd Development, Pris Ž Roper Orgniztion, The Environment: Public Attitudes nd Individul Behvior, Commissioned by S. C. Johnson & Son, Rcine, Wisconsin Ž July P. E. Wllce, Disclosure of environmentl libilities under the securities lws: the potentil of securities-mrket-bsed incentives for pollution control, Wshington & Lee Lw Re iew 50, 1093 Ž Summer U.S. Environmentl Protection Agency, Enforcement Accomplishments Report, FY 1993, Office of Enforcement, U.S. Environmentl Protection Agency, EPA 300-R Ž April One of the risks ssocited with using informtion s regultory tool is tht providing one piece of informtion might stifle the flow of nother type. For exmple, by focusing on the bsolute level of TRI emissions insted of the hrmfulness of emissions, some hve rgued tht firms who pper to be the worst polluters re ctully not.