Scribd Logo
Upload

Welcome to Scribd!

  • CSR Reporting Voluntary Nonfinancial Disclosure

    Uploaded by

    Chung Vu
    2 views42 pages
    CSR Reporting Voluntary Nonfinancial Disclosure

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Report this Document
    CSR Reporting Voluntary Nonfinancial Disclosure

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    2 views42 pages

    CSR Reporting Voluntary Nonfinancial Disclosure

    Uploaded by

    Chung Vu
    CSR Reporting Voluntary Nonfinancial Disclosure

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    of 42
    THE ACCOUNTING REVIEW American Accounting Association Vol. 86, No. 1 DOI: 10.2308/eer00000005 pe. 59-100 Voluntary Nonfinancial Disclosure and the Cost of Equity Capital: The Initiation of Corporate Social Responsibility Reporting Dan S. Dhaliwal The University of Arizona and Korea University Oliver Zhen Li The University of Arizona Albert Tsang Yong George Yang The Chinese University of Hong Kong ABSTRACT: We examine a potential benefit associated with the initiation of voluntary disclosure of comorate social responsibilty (CSA) activities: a reduction in firms’ cost of equity capital. We find that fis with e high cast of equity capital in the previous year tend to initiate disclosure of CSR activities in the current year and that initiating firms, with superior social responsiblity performance enjoy a subsequent reduction in the cost (of equity capital. Further, initiating firms with superior social responsibility performance attract dedicated institutional investors and analyst coverage. Moreover, these analysis achieve (ower absolute forecast errors and dispersion. Finally, we find that firms exploit the berieft of a lower cost of equity capital associated with the initiation of CSR discio- sure. Initiating firms are mote likely than nor-initiating firms to raise equity capital fol- lowing the initiations: among firms raising equity capital, initiating firms raise a signif. cantly laeger amount than do non-initiating fms. Keywords: comporate social responsibilty; cost of capital: voluntary disclosure. Data Availability: The data are publicly available trom the sources identied in the paper We appeciate de Vlasic commiems from Ving Cao, Zsoyang Gu, Michel Mapaan, Monte Pinces, Suresh Rahs san, 1. Wong, ant seminar garcpants at the City University of Hong Kong. Singapore Management Universi, ‘Tsinghua Univesity and 2009 AAA Annual Meeting, ad expecially the dead constetive sopgestions from Steven Kacbriaeir (editor) and two anonymous feteree. We thaak Brian Bushee for generossly sharing his casiications of insitwinal investors. We aso thank CorporteRegistercom and KLD Reseach & Analyis, Tne. for providing the ‘owporte social responsibility data, and UB/EIS for the enayst following data. We a grate 10 Xuomed HOO and ‘Gutagain Li for er excellon esearch astcence. The work decribed inthis paper wae fully supponed by 2 erat fom the Resear Grants Couneil of the Hong Kong SAR, China (reject No. CUNK@5140) ors note: Accepted by Steven Kachelineie. Submitted: April 2009 y YS9 30 Aouea) Anse A iy BUST WS J Sy CTH 2 sed 8S a.e109 so Vo FaREG (a4 tix 7 sr 68 ust a iy os oor’ o w aa wit 31 o or ose » © eae set ai if aot 186 * w $301 86 " a 901 us t st W801 as + ra 08 or " & wool 6sb 6 t oF 38 s st as UW t Re ace er it £ z sw ue ardweg jeuta 7 | hey eee iS asd ea '¥s9) ‘abi auieny “oon? @usopng ar ntg TON bushy mn Hey ‘ ones 462, q wopnquasiq = [ave January 2011 ‘American Accounting Association ‘The Accounting Review situatary Nonfinancial Disclosure and the Cust of Equity Capital 0 Empirical Models and Variable Definitions Past Cost of Equity Capital and Current-Year CSR Disclosure “To test HI, we examine whether « high cost of equity capital in the previous year gives firms un incentive for CSR disclosure in the current year. In the empirical regression model, we control tor other detertsinanis of CSR disclosure to parse out potential confounding effects. However, the ‘current literature provides limited information on what motivates a firm's CSR disclosure decision ‘As CSR disclosure is part of a firm's overall voluntary disclosure strategy, we identify potential factors from the voluntary disclasure literature that influence a firm’s decision to commit to CSR disclosure, Ovr logistic regression mode! is specified as follows: 9+ BiCOC,,.1 + ByPERFORM, 1 + ByMICONCERN,,.. + SIZE; + BLITIGATION 1 + BROA, 0 + ByCOM PETITION ya + BaF Nya + PsTOBINO, 1+ BioLEV,.1 + B\GLOBAL,.. + BrsLIQUIDITY 1+ Bi ABS_EM, so + BreCIGiy.) #SIND,, + EYEAR, + 6, o loglprob(DISC1,)(1 ~ prov(DISCI,.)] where DISCI,, is an indicator variable that equals | if firm i discloses a standalone CSR report for the first time in year ¢ (initiating firm-years or initiators), and 0 (non-initiating firm-years or hon-initiators) otherwise. Therefore, the control group (DISCI = 0), namely, non-initiators, in- cludes all years of firms that never issue CSR reports and the years before and after CSR-initiating Fars’ first-time reports ‘Our main variable of interest, the cost of equity capital in the year prior to first-time CSR disclosure, COC, is the ex ante oF implied cost of equity capital, calculated using three different models, namely, those of Gebhardt et al. (2001), Claus and Thomas (2001), and Easton (2004). ‘The mean of the thret measures (COC_AVG) serves as our proxy for the cos! of equity capital. To implement the estimation, we oblain expected future earings per share from UB/EJS and market price and dividend per share fram Compustat. ‘We include « number of controi variables in the regression. PERFORM is the total KLD score of CSR strengths, which we use 0 proxy for firms’ CSR pesformance. Firms with beter social performance have a greater incentive to disclose (Dye 1985). The KLD database is widely used in CSR research (Graves and Waddock 1994; Berman et al. 1999; Baron et al. 2009). Waddock (2003, 369) regards it as “the de facto (CSR) research standard at the moment.” "* KLD ranks firms" CSR performance in seven main categories: (1) community, (2) corporate governance, (3) diversity, (4) employee relations, (5) environment, (6) human rights, and (7) produet.'® We adjust raw CSR strength scores each year by industry medians to get relative performance scores that are comparable across industries, Ofcourse, there is also no lack of eticem ofthe KD database: For example, KLD usc nist variables to describe fins’ CSR performance. This crude methodology 2nd potntally suffers from lss of informaion. Chater e 3 (2009) show that KLD environmental srenghs donot acturtely pret polation levels or eampliance violations and that KLD raings do no optimally ute publicly sealable das "The rankings we based on informaion oblaned from sirvey, nancial stslements, government docament, poe reviewed Tega jouras, sad repors from maineeam mein. KLD defioes 2 set of potential sireagihe under each ‘ategory and assigns «value of if «strength evict, and 2 value of O otherwise. See the Appondit for more deals ot KLD's rating categories, January 2011 ‘American Accounting Association The Accounting Review « Dhaliwal. Li. Teang. and Yong We control for firm size (SIZE) because size captures various factors motivating firms to issue CSR reports such as public pressure of financial resources (Lang and Lundhotm (993). We mea- sure SIZE as the natural logarithm of the market value of common equity at the beginning of each year Skinner (1997) argues that firms facing a higher level of litigation risk (LITIGATION) ase ‘more fikely to make voluntary disclosure to preempt potential lawsuits. LITIGATION is an indi- cator variable that equals 1 if a fran operates in a high-liigation industry (SIC codes of 2833~ 2836, 3570-3577, 3600-3674, 5200-596), and 7370), and 0 otherwise (Francis et al. 1994; Mat sumoto 2002). AS firms with better financial performance likely have more resources to practice CR activities and produce CSR reports, we include return on assets (ROA), computed as income before extraordinary items scaled by total assets at the beginning of each year. Dye (1985) suggests that proprietary costs arising from product market competition can reduce dixclosure incentives. Hence, we control for industry competition (COMPETITION), which is proxied by the Hesfindahi-Hirschman Index multiplied by ~1. This index is computed as the sum of the squared fractions of stles of the 50 largest firms in an industry (industries are defined based on the two-digit SIC codes). In cases where there are fewer than 50 firms in an industry, we use all firms in the industry to calculate market shares. ‘n addition, firms raising capital in the public market have a greater propensity to make voluntary disclosures (Frankel et al. 1995). We ‘contol for a firm's financing activities (FIN) by assessing the amount of debi ox equity capital ‘alsed by the firm during the year scaled by total assets atthe beginning of the year. Following Richardson et al. (2004), FIN is measured as the sale of common and preferred shares minus the purchase of common and preferred shares plus the long-term debt issuance minus the long-term debt reduction We also control for growth opportunities (TOBIN) because firms io an expansionary period are more financially constrained and have fewer resources for CSR activities and disclosure. However, growth firms also tend to have higher levels of information asymmetry, which could induce managers to make more disclosures to attract potential investors. The net effect is hence ‘unknown ex ante. TOBING is Tobin's Q, defined as the market value of common equity plus the book vaiue of preferred stock, book valve of long-term debt and current liabilities, scaled by the book value of total assets. We include the debe ratio (LEV) in the model because debt servicing plays a monitoring role and debi halders demand greater disclosure (Leftwich et ol. 1981), We dicfine LEV as the ratio of total debt divided by total assets In addition, firms with a global focus, especially those operating in emerging markets, face greater pressure to commit to social performance and are accordingly more likely to provide CSR isclosure. GLOBAL is an indicator variable that equals 1 if a firm reports foreign income, and 0 ‘otherwise, Further, managers have incentives 10 increase the liquidity f ther firms’ stock in order to issue equities or sell shares of their firm obtained from options or other incentive compensation plans. One way to increase liquidity is to improve trensparency and supply more information to investors. Our liquidity measure, LIQUIDITY, is the ratio of the number of shares traded in the ‘year to the total shares outstanding at the year-end Finally, CSR disclosure could be correlated with the general disclosure policies and financial transparency of firms. To control for this possibility, we include swo variables to proxy for firm Financia! disclosure quality and voluntary disclosure policy: earnings quality (ABS_EM and man- ‘egement eamings forecasts (C/G). We use the absolute value of abnormal accruals from the modified Jones (1991) model, based on Dechow et al. (1995), to proxy for earnings quality (Francis et al. 2008). Following prior studies that use management forecasts as a direct measure Using he orginal Jones (1991) model oc an sernatve version develope by Dechow et (2003, 359, qsation (28) eld sna esis, The Accounting Review January 2011 ‘American Aceouoting Association r oo olomtory Nonfinanciol Disclosure and the Cost of Equity Cepital of a firm's disclosure policy (Rogers and Van Buskirk 2009). we define CIG as an indicator wuriable that equals | if2 firm issues at least one earnings forecest in the yeer, and 0 otherwise. In All specifications of the model, we include industry and year indicators 10 control for potential industry and year effects Effect of CSR Disclosure on the Future Cost of Equity Capital Hypothesis 2 predicts that CSR disclosure leads to a lower cost of equity capital, We test H2 by estimating the following regression model S%COC 1 = By + ByDISCI,, + B,ASIZE,, + PABETA,, + BALEV.,+ BOMB, + BALTG,, + B,ALNDISP,,+ ZIND,, + EYEAR,, + €;, 2 where 49COC,,.) is the percentage change in the Cost of equity capital from year 1 to year /+1 ‘The control variables also adopt the change form. A negative coefficient on DISCT would support mn ‘The control variables are derived from prior research. Fama and French (1992) find that expected retums are negatively associated with firm size and positively associated with the book- tocmarket ratio, Hence, we include firm size (SIZE) and the market-to-book ratio (MB). The market model BETA, which is estimated using CRSP daily data for each year, is included to control for systematic risk. Gebhardt et al. (2001) and Gode and Mohanram (2003) find that the implied cost of equity capital is positively associated with long-term growth rate. We therefore include an empirical proxy of fong-term growth rate based on /B/E/S analyst EPS forecasts (L7G), which is measured as the difference between the (wo-year-ahead consensus EPS forecast and the one-yearahead consensus EPS forecast scaled by the one-year-ahead consensus EPS forecast. Gebhardt et al. (2001) and Dhaliwal et al. (2005) find that analyst forecast dispersion is negatively associated with the implied cost of equity capital. Thos, we include analyst forecast dispersion (LNDISP), which is calculated as the logarithm of the standard deviation of analyst EPS. forecasts divided by the consensus forecast. We include leverage (LEV) because Fama and French (1992) suggest that the cost of equity capital increases as the degree of leverage increases. All other variables are as defined earlier Although firms may be motivated by a possible reduction in the cost of equity capital when deciding whether to issue a CSR report, from the perspective of investors, CSR disclosure per se may not necessarily warrant a lower cost of equity capital. Corporate managers could attempt to manage public impressions through such disclosures; therefore, CSR information can be self- serving and noneredible (Cormier and Magnan 2003; Hobson and Kachelmeicr 2005). Investors are likely to have a favorable perception if a firm actually performs well in its CSR practices relative (0 its peers. To incorporate this possibility, we augment Equation (2) with a measure of a firm’s relative CSR performance from KLD (H/PERFORM): A%COC.»1 = Bo + BiDISCI,, + BHIPERFORM,, + B,DISCI,, * HIPERFORM,, + BASIZE,,+ BsABETA,,+ BpALEV,,+ BAMB,,+ PyALTC,, + ByALNDISP,, + 64, @) Where H/PERFORM is an indicator variable that equels 1 if a firm's CSR performance score, PERFORM, is higher than its industry median (in other words, if the firm is a superior CSR Performer in its industry), and 0 otherwise. All other variables are as defined earlier. We expect the effect of DISCI » HIPERFORM to be negative. In an additional test, instead of using the interac- tion term between DISCI and HIPERFORM, we estimate Equation (2) within the high and low Partitions of CSR performance scores. Using partitioned subsamples sacrifices some power due t0 January 2011 American Accounting Association ( The Accounting Review ” Dhaliwal, Li, Tzome, end Yong reduced sample size, but has the benefit of flexibility that allows the effects of other variables to also: vary based on high or low levels of CSR performance scores Endogeneity and seif-selection could potentially affect our results. In our main analysis. we use a lead-lag approach (0 tackle these issues. To further enhance inferences based on our lead-lag approach, we adapt the Heckman and Hausman (wo-stage procedures and tepeat our main analy- ses, The Heckman awo-slage procedure introduces the inverse Mills ratio into the second-stage OLS regression to control for self-selection bias that is related to CSR disclosure. We obt qualitatively similar results using the Heckman two-stage procedure.” The Hausman (est deals ‘with potential endogeneity in the-data, We conduct the Hausman test and find that endogeneity does not qualitatively affect our main results. 1V. RESULTS Descriptive Statistics Table 2, Panel A provides descriptive statistics for the variables included in Equation (1) for the full sample and separately for initiators and non-initiators. The cost of equily capital before CSR disclosure is significantly higher (p = 0.04) among CSR initiators (12.86 percent) than among non-initiators (51.98 percent). This difference is also reflected in a significantly positive correlation coefficient between DISCT and COC_AVG in Table 2, Pane! B, providing initial sup- port for H Consistent with the theory on voluntary disclosure, firms voluntarily publishing stanéalone (CSE reports tend to have superior CSR performance (PERFORM) relative (0 their industry peers. The difference in CSR performance between the «wo groups (1.613 for initiators versus ~0.166 fac non-initiators) is significant (p < 0.01). The cortelation between DISCI and PERFORM is also significantly positive though at a relatively moderate level of 0.09 based on the Spearman corre- lation and 0.13 based on the Pearson comelation (see Table 2, Panel B). This highlights the imponance of including PERFORM in our regression equations. Initiators are significantly larger (SIZE: 9.147 for initiators versus 5.783 for non-nitiaors, p < 001) and more profitable (ROA: 0.051 for initiators versus 0.015 for non-iitiators, p < 0.01) than non-intiators, lending support to the financial resources argument for CSR disclosure. Con- teary to the proprietary information argument, initiators tend to observe greater industry competi- tion than non-initiatacs (COMPETITION: ~0,060 for initiators versus —0.069 for non-initiators, p 0.02) Initiators have a significantly lower level of financing shan won- initiators (FIN: ~0.019 for initiators versus 0.043 for non-initiators, p < 001). The negative financing level fo initiators implies that these firms, in net effec, either have repurchased stock or redeemed their debts, Firms normally conduct repurchases when they believe that their stock is undervalued, indicating a high cost of equity capital, which in turn provides an incentive for managers to increase disclosure and transparency levels. Similarly, the redemption of mature debts likely implies that firms need Future financing to maintain a normal capital level. These firms would also be willing to increase their level of disclosure if doing so helped them to lower the cost of borrowing. Initiators have a higher degree of leverage than non-initiators (LEV: 0.265 for initiators versus 0.221 for non-initiators, p < 0.01). Those with a higher level of global operations are also more likely to publish CSR reports (GLOBAL: 0.460 for initiators verss 0.219 for non-intiators, p < 0.01), consistent with the notion that these firms attract more attention in the intemational com- unity. Contrary to the otion of disclosing information to improve liquidity, initiators actually "The only exception is forte test of analyst forecast eos. Among beter CSR perfrmen, she coeficint on DISCY is positive and insignificant ‘The Accounting Review January 2011 ‘American Accuuating Association vinonciat Disclosure and the Cost of Equity Capital Now hans wo o0- 800 vro~ 300 Wo0- wo~ Nd sr0- 00 wo eo0- wo 10 "NOLLLIAWOD so0- e90- ro $69, sr wo voy oro e00- ev0- 6o0~ ooo “nowvomn 6070 ce Io 100 900 92s F00~ wo see too wo wwosnae 0 woo 100 0 100 ‘av"909 100 oo 10 ero too ‘psig PNOrLLgaWOD CNonvoLn "aus = wyoriaa = av 009 ~— asia 2 = CE = 19SIC) 8211 = (0 = 19SIA) 8) aonwfasr09 uosrwagTeUIEEMS “9 (PHS c 9309) 200) 3900) Png sav st uso 9190 a0 919 ove wet ae ack ws 6z0 os zero ove zo sero ve co- L007 66"! sor ser «00 6190~ zw00 we 6300- wu 0- «o0- "woutntas 09 sort sio%9 1s00 2100 “vow wo- waz wz0 ‘NouvoULrT e305 seus 9218 s91'0~ aro- priosuad set 83611 (@)""9a¥-909 en =) eres) ardareg wna Biase 9="0sIa 119810 uone(e:309 ue sansieig Aseumtang vaTaye. mostiedaio> way :¥ ue, ‘The Accounting Review stanuary 2013 Asntsican Accounting Association Dholinel. Li, Tsong. ond Yong 2 (280d 1x00 vo panunryo2) pss 51 vonea0> NY S31 peg i Presi 3 gue ONE iy 00 wo a0 nro tro ora soo wo sro 00 vo wo ro ero zo ao zo avo so avo wo wo- wa soo so0- coo o- w0- ico- sco 0 vo wo vo sco sro ro 00~ 00 woo 0 co O0~ wo roo 00 PoO- 800 00 “noua WOD or0- 800 S00 r¢ 90°0- sTO~ av0~ you S00 00 wo er0- aro 10 “NOLLYOLLT ate we wo 00 sro rro— “y2is woo 0 woo 100 0 0~ “Waoniss o0- wo ro o00- “ony 209 e00— 800 0 w0~ "19810 “90 "emo =| Nid soo Ir0- 0 wo- 00- “wa say woo eo 800 oO 800 “919 wo 80 100~ woe wo ALIGOOT soo ro v0 so0- soa 'nva079 stom soo io sro wo “as sro ero 00 te0~ 0 MONGOL Thouvourt = azis «= Myworaa = onv300 "sia (ete = (C= 19810) W'ZIL'TT = (0 = 19810) 8) wonEyeLs09 vosiwas\oruLNads +g jou January 2011 2 Associ sean Aceou The Accounting Review A stoner: Nonfinancial Disclosure and the Cost of Equity Capital nave ay: Way Senpsa x83 a pasapEsUeD 3 woNeU suadoud 381g pow se oy pad fa popes aygeni2e) uenooe wy sBuEGD 20) {QUE = sLLIq) wonst—ss ys pauzaid pu ypos vowed Jo esegand 2) 5 January 2011 ‘American Accounting Association The Accounting Review 4 Dhaliwal, Li, Teang, ond Yor; have higher liquidity levels than non-initiators (LIQUIDITY 1.387 For initiators versus 1.247 for ron-initiators, p = 0.04). Finally, initiators have bewer financial disclosure as manifested in thei more frequent management forecasts (CIG: 0.646 foc initiators versus 0.527 for non-initiators P< O01) and better earnings quality (ABS_EM: 0.032 for initiators versus 0.066 for non Iniiators, p< 0.01) than non-iitiatrs Cost of Equity Capital and the Likelihood of CSR Disclosure Hypothesis [ predicts that a firms likelihood of disclosing its corporate social responsibility activities is positively associated wich its cost of equity capital in the previous year. We report the regression results for Equation (1) in Table 3. fn Column I, we include alt ®ssttime ceporting Tirm-year observations. In Column Il, we exciude first-time reports that primarily discuss environ- mental issues, following Simnett et al. (2009). In Cotumn Ni, we examine the robustness of our results 69 the exclusion of the Utilities industry Across all three specifications of the dependem variable, the cast of equity capital, CO. ‘C_AVG. in year 1-1, is significantly positively associated with a firm's likelihood of voluntarily issuing a standalone CSR report in year # {coeft, = 0.089, p < 0.01; coeff. = 0.052, p < 0.01; and coeff, = 0.062, p < 0.01 in Columns 1, H and IIf, respectively), consistent with Hi, which posits that a higher past cost of equity capital is associated with a greater likelihood of voluntary CSR disclosure in the current year. In Column I, for instance, holding other factors constant, when the prior year cost of equity capita inexeases by one percentage point, the odds of initiating standalone. CSR disclosure increase by 5.02 percent ‘The coefficient estimates af the control variables are generally consistent with the univariate ‘comparisons in Table 2. One exception is LIQUIDITY, which reverses direction, The significantly negative coefficient suggrsts that firms with lower levels of liquidity are more likely to publish CSR repos, consistent with our original conjecture, The effects of financial disclosure quality, ABS_Fld, and management forecast, CIG, are no longer significant. CSR Disclosure and the Future Cost of Equity Capital Hypothesis 2 predicts that voluntary CSR disclosure leads to a lower future cost of equity capital. Table 4, Panel A compares initiators and non-initiators and Table 4, Panel B presents the regression results. In Column 1 (Equation (2)), the coefficient on DISCT is insignificant (coeff. 0.037, p > 0.50). It appears that CSR disclosure per se is not significantly associated with a ‘change in a firm's future cost of equity capital. fs Caluann IT (Equation (3)), we consider whesher firm has superior CSR performance relative to its industry peers. The interaction term between DISCI and HIPERFORM ‘s significantly vegative (Coeff. = ~4.618, p < 0.01), consistent with 1, which posits that CSR disclosure reduces the cost of equity capital."* Combining the main effect of DISCI and the effect of the interaction term between DISCI and HIPERFORM in Columa 4, we infer chat superior CSR performers enjoy a 1.833 percent reduction im the cost of equity capital when they produce standalone CSR reports for the frst time, In Columns Ill and IV, we ‘obtain similar results when we exclude environmental reports and the Uriities industry, respec: We perform a censiiviy test By restictng he analysis to fin-yearcbservatons of CSR reporters it re-post sein Specially, we foc enly on dscosing fires a ll ute the change wpeifeation of the dependent vanable ACO- CLAVG. DISC! is an indicator variable that equals | forthe fist reporting yea wil equas O before ot afer the fst ‘seponing year of aislosing fim. The purpose of this examinagn Is to show that aredyetion inthe cost of equity ‘apial ours immediatly afer tho fst repering yeas aod w lieve the concern tht the ste of into sample $l lative to he universe of fiem-yearabservations used i the man et. We blain simile ess and serene rtf hse CSR perrmanes enjoy 4 eco a he ont ey capi ey bin tne 5 eepor. ‘The Accounting Review Jenvary 2001 ‘American Aecgunting Association 18 tistuntory Nonfinanciol Disclosure ond the Cost of Equity Copital “BGR pau 20 syQeERA TY ous sont . #91 ere (1 = 780 ep) wou ov'ene ‘ney pour wea 320 at opnasa on on ssowsipy Asn sx ok pu 824, ve 190 srst- sro ue - Swasav a0 veo 910 10 + “919 wo as5500- 0 e426 0- - “auainor, wo soo na1600 wo 48100 + “vao19 ova ora 1s80- ao oreo- + agT 00 100 seeaS5E0- 000 eaB6E0~ i ao ovo wo ss60 00 v0 + Nl oro sso sort 40 a9 - “ouuiaaHod 90 “ro wort wo or + “Vou 050 «0 #10 s¥0 3020 é “NouvoUsT 00 sextl01 + “e218 sco 1s00 + “waoawad ooo 4432500 + “9Av-309 a 9D ig rae ‘pala onowny mt Taodny jruaunwaNaeg Fupouy, 0 aunsojesiq YSO Jo siuEUTUa}9g © v1avL Tosia= qn, vwapuadag January 2011 ‘American Accounting Assoviation The Accounting Review so (atad vs wo pomayoa) g 2900 100 1900 2900 al pasnipy ® PA Ph an ok soma Aan a A aK °K A ssetetou, 2% 3 0 evo oro 6010 670 L010 ore 0 “aStaNTY z ut #eSL1'0— eet eel 0 wie RELY O- Svee wnSLl 0 “OLTD & 98EI- — sael010l- eer eee lSTOl— wel SE Pl ex l€l OL ‘anv 3 tt beats59~ wz- 90d 187 au yt01'L~ 98E gus ORTL~ ‘nary 667 pg 8670 wre eee VEO Ore exe C60 ese eon te 0 ‘vEae9 wt xa OPO'S at enable at ee 6b8P ot re 8697 “a2is0 i9t ene SOO a eeS109~ BET~ eee ITP ‘WaOLMIdIH #19810 eo 500 120 109 oo oso ‘wwosuadii srl 9s8't se 96 69F vst ae eS8LT o- ‘Le00- ‘rosia 7a TEx} as a 7 a wey BD Seo ‘Sn Bunou, auedoy —awes ang revuauuaainy SoyouDy a it © 1 (aa¥ 2000 =agrHeA, wopEadad) adeD Aambg Jo 9D aunsopsig ASO-I80a +H HRA 00 6000 00 "SSIQNTY 00 1so0- x00 ‘on1v sto0- zt 0- SiWO- ‘any «z00- wz00- ez00- as79 oz D- 1909 6100 ‘waay €000- 2000- 000- ‘aziSV aot e6ET 169 (%)"Dav 2099 = o= 181d T= "9810 aidureg (104 uosundaiog ura :¥ aR de Aynbg jo 1909 amsopesia US0-I804 PatavL %6 January 2011, ‘American Acepunting Aseociation The Accounting Review it Disclosure and the Cost of Equity Capital tuna Nonfinan (eed nav wo panywo2) 2m 380K =!"'9qV-D0IT A O12 OSI soRRyed agEUeA 3h, ssoveaypuy Aasnpuy eK ssoveorput 2, aco 096'0- 860 wa “gSIONTV wt x8L 0 we saaTEEO- ‘ony ert- xB svpt- 44850 71~ “aN one 8g1~ alS89~ ag79 evo ae 440900 "vagay 668 wa0fl9S ws wx e002E ‘z2sv ital ocre foz- allt t~ ‘sia + 1D Tes) B09) SRK HYOT HIE OS wORER WHORIaE %05 FL, ssuwaoyng ¥S2 sway Bo foaoRYpLOD pauonRAE, TeIdeD Amba Jo wo AMSOPSIE YS)-1904 {9 [eNO ee S06 wre 1506 u (= 'nwosuaane ost wet sor vat vot ae siz oo) ey “uR0 aes “90 Wes "BD SgeDEA anoway adres Gm at 1 oAV-D09¥=9198"4 1wapuadad) [ede synbg J0 2809 aUNSORESIG YSD-W80d “A UE The Accounting Review January 2011 ‘American Accounting Association Dholivwol, Li, Tsong, and Yong. | ® January 2015, ‘The Accounting Review ‘American Accounting rr Voluntary Nonfinancial Disclosine und the Cost of Equity Capital 1” tively. Overall, the evidence is consistent with our H2 that CSR-disclosing firms with superior CSR performance achieve 2 redvetion ia the cost of equity capital. ‘Table 4, Panel C presents the results from estimating Equation (2) within the 1wo subsamples panttioned based on annual industry medians of CSR performance (PERFORM). Consistent with the results in Panel B, we find a significantly negative coefficient on DISCT (coeff. = 1.777. p (0.04) ia the high CSR performance subsample. This coefficient indicates that voluntary CSR disclosure yields 2 1.77 percent reduction in the cost of equity capital. In the low CSR perfor ‘mance subsample, there is no significant association between CSR disclosure and the change in the-cost of equity capital.” Potential Mechanisms Linking CSR Disclosure and the Cost of Equity Capital ‘The above results suggest that CSR disclosure combined with superior CSR performance is associated with a reduction in the cost of equity capital. Below, we provide evidence on the potential underlying mechanisms througk which voluntary CSR disclosure lowers the cost of ‘equity capital. We focus on two types of financial intermediaries: institutional investors and finan- cial analysts. CSR Disclosure and Institutional Investors Shleifer and Vishny (1986) suggest that the large equity stakes in the invested firms and the high levels of sophistication of these investors enable them to reduce agency cost problems and the extent of information asymmetry between managers and shareholders, an effect that leads (0 a reduction in the cost of equity capital. We consider three different types of institutional investors dedicated (DED), transient (7RA), and quasi-indexer (QIX) institutional investors. Dedicated in- stitutional investors are more likely to play monitoring and governance roles than the other (wo types (Bushee 1998), To determine whether CSR disclosure attracts institutional investors, we follow Bushee and Noe (2000) and estimate the following model AINST,, = B+ ByDISC, + ByHIPERFORM,, + ,DISCI,, * HIPERFORM,,+ ByINST 4+ BsMMRET,,+ BgTVOL,»-1 + B)AMVis + PyBETA,. 1 + RIRISK;,. + BigALEV,,+ By ADP, + Bi2QEP,, + BiyAMB,,+ BuASGR,,+ BysARATE,, + BygASHRS, + SIND, + EYEAR«,+ 6. @) "tn aneraivespeifietione of the model, we examine the effects of eo other vasables proxying for firme effort and ‘omuniment to beter CSR iisclosure (1) We kdeny firms Ia provide asuranee (ASSURANCE) of ther ceports ‘vough odependeot tid partes, most often Big ¢ accounting fms nnd intematona conalng conpanis. Stet et a1. 2008) provide evicence ta firms secking 1 enhance the ereditability of ther report ad their corporate reputasion ‘cc more tly to have thie sustainability eepods assured. (2) We also aes the effet ofthe length ofeach CSR report (LENGTH) rele tothe average rept lengh ofthe ascosing fem’ indusey (Leuz and Sehrand 2008). OF course, ASSURANCE and LENGTH a7e #0t independant of DISCI. We Gud tat, condioad on fratcime CSR dislorure (DICH, external acura and long report Teng further reduce the cost of equity capital. Specially, when we wie {he ASSURANCE indieator (equals | with an assurance, and O oterwise) the coefficient on DISC! » HIPERFORM i rogative and signiiant (ect: ~ ~2523, p ~ ODI) andthe coetScient on DISCI + HIPERFORH » ASSURANCE i egaste and signifcam (cool = ~3.540, p = O08). Therefore, usurance doubles the eflect of CSR disloeue, Whea Wwe use the LENGTH incestr (equals | if looger than the indsuy-year median, and O cierwis), te coetcient on DISCI + HIPERFORM is negative aod signifeant (coef. = ~2574, p = 002) and the coefficiens on DISCI » HIPERFORM » LENGTH is negative and sigifican (Coe. = —3,920, p= 006) Therefore, along repet more. than Aovbles the eect of CSR dielonire The ooeticien on DISCY is not significant but positive, if anything, for poor CSR performers It is possible that disclosing pooe CSR performance could actually be a sina of high sk or fm weakness and, terefore. the cost of guy capltal could scully go up. Tas does explain why the diet effect of DISCT is insignificant Io the pooled regression Jenuary 2011 American Accounting Association La The Accounting Review 0 Dhofiw. Li. Teng, and Yang ‘where & denotes change from year 1 to year (+1. INST represents stock ownership by dedicated (DED), transient (TRA), oF quasiindexer (QIX) institutional investors. MRET is the market- adjusted buy-and-hold stock return measured over the year. TVOL, «liquidity proxy. is the average monthly tcading volume relative to cotal shares outstanding, JRISK is the logarithmic transforma tion of the standard deviation of markct-mode! residuals calculated using daily stock returns. Beta (BETA), debt ratio (LEV), and IRISK capture firm risk along different dimensions. DP is the ratio of dividends to the market value of equity. EP is the ratio of income before extraordinary items to the market value of equity. SGR is the percentage change in annual sales. We include DP, EP, MB, and SGR to control for changes in firms" fundamentals that can affect the investment decisions of institutional investors (Bushee 2001). RATE is Ure S&P stock rating (9 = A+,8 = A,7 = A 6 B+, 5 = B,4 = B~,3 = C,2 =D, | = not rated), which captures the preference of institutional investors for well-reputed firms (Del Guercio 1996). SHRS is the logarithmic trans- formation of shares outstanding, and is change Form proxies for equity isswance or repurchases that affect both institutional investor following and firms” disclosure policies. All other variables are as defined earlier “Table 5, Panel A presents comparisons of one-year-ahead holdings and changes in holdings by the three types of institutional investors between initiators and non-initiators. Overall, the univari- ale comparisons do rot reveal significant differences between initiators and non-initators. 1f anything, we observe a greater decrease in ansient institutional holding. among initiators com- pared t0 non-initiators (P = 0.04), even though the level of this type of holding is stil slightly higher among initiators than among non-inititors (Pp = 0.07) ‘Table 5, Panel displays the regression cesults. There is weak evidence that initiating firms with superior CSR performance attract more dedicated institutional investors. The coefficient on DISC! HIPERFORM is marginally significantly positive (coeff. = 0.414, p = 9.16). To further examine this issue, we run regressions without the interaction term in the two subsamples parti- tioned based on annual industry medians of CSR performance for dedicated institutional investors. We report the results in Table 5, Panel C. We observe a significantly positive coefficient on DISC! (coelf. = 0.438, p = 0.01) for the superior-performance group, whereas the coefficient on DISCT for the low-performance group is insignificane. In uotabulated tests, we do not find a significant association between transient or quasi-indexer institutional ivestor holdings and the initiation of CSR disclosure for the full sample or the partitioned subsamples. In sum, the evidence in this subsection suggests that voluntary CSR disclosure atracts ded cated institutional investors, who have long investment horizons and play monitoring and gover- nance roles. Consistent with our previous evidence that superior CSR performers enjoy a reduction in the cost of equity capital through CSR disclosure, the effect of CSR disclosure on dedicated instirutional ownership is stronger if disclosing firms have CSR performance superior to their industry peers. CSR Disclosure and Analyst Forecasts We also examine three questions related to financial analysts and CSR disclosure. Firs, we explore whether financial analysts are more willing to cover firms after they initiate CSR disclo sure. Second, we investigate whether the level of forecast accuracy increases and finally we determine whether forecast dispersion decreases when CSR reports are available. Increased levels of analyst coverage and forecast accuracy and a reduction in the level of forecast dispersion have Ue potenti! to lower the cost of equity capital. To determine the impact of CSR disclosure on the behavior of financial analysts, we run the following three regressions following Lang and Lund- hholm (1996) and Ali et al. (2007) January 2011 —___| The Accounting Review ‘American Accounting Association (and sau vo pamnuoa) s80- sseo- 0 zo wo ero ei 4199 ez sexSE6 11 wi zoe £90 60 ori w260— 0 s000- wo 1800- wo 5900 «90 9500 s#0- 2400- s00- o190- aL 0- *500- 01> srt o- +0 09 0 we rel 9900 oro 00 so 6100 rr0- B100- wi wvo- 20 S100 wst- 46500- sus- sea 200- 3 vs 448100- & zs00- wee e600- ort v1r0 ‘waosradiH »'19810 a sa00 wie exro~ svo- Leg0- ‘waoasaaiH = 500 £90 w7a oso oto & 9 er 8 a a0 Sarg 3 vay RoW waaay g SSuypjop sosaau] peuoAMASE] NSO ASTI s10 YSD-A80A sangvi. — ‘The Accounting Review January 2011 ‘Arierican Accossting Association Dholiwal, Li. Tsang, and Yong (280d 20 wo panunuo>) 370 00 10- ‘wy wo" wvo w0- ‘av owe a ‘aa0 661'0- 360 ‘any sero- ist “ysrar 600 wie yaaa 9800 ot ‘aw 99000 wo "OAL ‘o0- 090 “aw ene l200- (6e- aaa £900- we ‘nosta “wep “Te sigen%, waaay WUOTUSA %05 woRoE Saaoy WuOTIAd %05 AOL dounuuo}ng ¥SD Sugg Wo revoRTPUED SBEIPIOR ZODACT UNAMINSHT SIMEEIHG ASO-HeA =D UR re wee we6 u oz set out oui out re 2600 wo A, 5 3 suoyeotpuT Aas, 5K SK 8K ssoveopoy 83x, wato- wo- eor0- 48670 ‘saHsy £200 1 +1600 anet800 ‘auve9 1000~ wo zoo too0- ‘yosy 961 +5000 990 Hoo o1w0- ‘any ia ‘we, ae a) HD Sane, Py PIV . Waaae sBujpjoy{ 2015940] feoopRIASH AIOSO}IsCE YSI-WOd #4 FUE January 2011 ‘Aroerican Accounting Association The Accounting Review tnt Nonfinancial Disclosure and the Cost of Equity Capital (8nd wesw wo panwivo>) 40 aes fo tp OF (9A) SPUR yo OFM! a Ht ga aU ga —'4d = "4D sve [mot @ popu A772 4 mak og 1-1 EAC ay BION BMD = ATR ‘nat tt ac so] gud jo sBewaarad ur HD = “yaze BUR MO az 141 ak G Tapani wojsen pe soxopuy-sonb ponepop 6a Rupoy dlyss2uMO Jo aBeUEDId = yay FUE ID EI gee wepeadan 2 ‘aimpasosd any soqny Sui p09 a8 ‘eanaodsat sana] wont 1 pue waned ¢ 420d O| a sg sig 0 te sx (i="9sia) 8 w0s0 iso a pasnlpy 3% in ssottompu, aenpst seh oh ‘HOW sak eo oro sz wxss¥0 ‘saNsY evo 8000 ore 920100 “uae oi oo wee 00000~ ‘wos we wD 787 a) wiaeneA 7 GaaN 7qgqy HONEA “05 Wood AUOTUTE %05 AOL 2puewa0}eg YSD ULE uo jeuoNpUED SBUNPION s0ySIAUT reUORRURSUT SURSO|SIC YSIS “3 tee Jonuary 2011 ‘American Accounting Assoviation The Accounting Review Dhalivol. Ls, Tsong, and Yang 1HS~'SUHS = 'SaHIST oe pai We P= 29 See PaaS a "y9$~ "40s = "vos *@aQ) Amba Jo ‘an = "ana 49~'dg ='430 1 0909 20 Ag pop 84 January 2011 ‘American Accounting Association The Accounting Review olimary Nowfinancial Disclosure and the Cost of Equity Capital 8s ACOVERAGE,,,, = B+ ByDISCI,,+ BxHIPERFORM, + ByDISCI,,* HIPERFORM + BAMSIZE,, + BsMSTDROF, «+ BySINVPRICE, + ByARETVAR,, + BARD), + ByMROA,,+ BighCORR,, + Ein, (s) AIFEl:sc1 = Bo + B)DISCI,,+ 2>HIPERFORM, + B,DISCI,,* HIPERFORM,,+ BeASIZE,, + BsASTOROE,, + BACHEPS,, + BARD), + BySROA,, + BACORR,, + €1,. (6) ADISP, 41 = By+ B,DISCI,, + SzHIPERFORM,,+ PyDISCI,, * HIPERFORM,. + PyASTZE;, + BASTDROE,, + BACHEPS,, + ByARD;,+ BDROA,,+ BpACORR,, + €;,, a where COVERAGE is the 12-month average of the number of analysts who issue annual earings forecasts captured in the /B/E/S database for a specific firm: IFA) is the absolute value of the 12-month average of analyst forecast errors, which is defined as actual earnings minus the mean forecast, deflated by the stock price atthe beginning of the fiscal year, and DISP is the 12-month average of the standard deviations of analyst forecasts, deflated by the sock price at the beginning ofthe fiscal year We include number of control variables derived from prior research. We include firm size (126) because larger firms have more potential brokerage or investment banking businesses for analysts’ brokerage houses (Bhushan 1989), which affects analyst forecasting behavior. We in- clude the inverse of stock prices (INVPRICE) because Brennan and Hughes (1991) sugecst that it proxies for the brokerage commission rate. Analysts are more likely to follow firms with higher levels of return variability because the anticipated trading benefits based on private information on these stocks are greater (Bhushan 1989). We therefore include STDROE, which is measured as the standard deviation of ROE in the preceding four quarters, and RETVAR, computed as the daily stock retum variance over the 200 days prior to the year-end. We include research and develop rent expense (RD) as a proxy for the level of information asymmetry (Aboody and Lev 2000) because analysts have relatively stronger incentives to follow firms with higher levels of informs: tion asymmetry (Barth et al, 2001), The earnings-retur (Pearson) correlation (CORR) between ROE and annual stock returns in the preceding four quarters captures the difficulty in predicting a firm's earings. In addition, ROA controls for firm profitability. Finally, annual changs in EPS (ACHEPS) controls for the magaitude of the forthcoming earnings information (Ali etal. 2007) All other variables are a8 defined earlier Table 6, Panel A presents a comparison of the levels of and changes in the three main analyst variables in the year following first-time CSR disclosures. Initiators are covered by more analysts than non-initiators (COVERAGE: 26.08 for initiators versus 15.72 for non-inihators, p < 0.01), and achieve greater improvernent in forecast accuracy than non-initiators {AIFEY: ~0.137 for initiators versus 0.120 for non-initators, though at a more marginal statistical significance level with p = 0.07) We present the multivariate regression results for Equations (5), Equation (6), and Equation (7) in Panels B, C, and D of Table 6, respectively. Column I of Pasel B shows that there is a significantly positive coefficient on DISC/ * HIPERFORM (coeff. = 1.052, p = 0.05), which suggests chat analyst following increases for initiators with supericr CSR performance. When we "un the regression separately in the two subsamples portioned based on industry medians of CSR The Accounting Review January 2011 ‘American Accounting Astoviation (00d rau wo panwuuo>) : uo pion soo vapstne fom 0 1000 ovo 0 fennel 3 ag. ead oe ene 6UT, re aeelSLT "vONR 2 98 44950 oro 10 oe ent ‘aay sve ven l200 #9 0- 100- 20 2000 ‘vain gw eae6it 9 woraTS56 8 santo? ‘godant9 wv0- 9100- Bo 1020 £6 0- uro- "owas we onl? ss 498880 8 aestsP0 ‘wis so 97501 “yaioawaattis'Osia oso- ozo0- ‘wuoawaatn 00 sooo zz axt60 to1- esr o~ ‘rosta See 144 9080~ 168 wag ]0¥O— SRI sgt" O- doosary as aD as HOD, eT “B05 - AOI 2505 WORD HuOT IIa is HH, dian wn 7 r (go vaan09¥= a9eHe, wapuadag) ses00D whyeRY aOR YSO-8OA Ha PUN siseos0g 1sAjBuy UF s9BueyD sunsoPsIG YSD-IS0d 2 9aT8va. 6000 Ti00- oro aa oe rr) jal eset wee "govea009 soco soz0 azo sizo eo 61097 89L'St “79¥vaIAOD iosia corsa wospiedwoy avayy -y fated January 2011 ‘American Acceunting Association The Accounting Review a7 ohumery Nowfinaneial Disclosure and the Cost of Equity Capital (2800 rau wo ponnive2) 6559 289 a a (v=1osia) 8 3500 09 al eaisnlpy. 9e0- too0- veo 1000 er0- ‘3¥099 ws 045580 a Bro ore "vouy 6x0 0s eee sestlS TH wie ‘aw eor~ ex 00T 0 se0- s900- oz ‘saaHove ori 46800 zo 600 wt 'sowaisy 9s5— ane8bZ0- ors- sea8810- weoi- "aay ro “nuoautalH «19810 Ts- ‘waosTaiH ai- sevo- wi- sIst0- wie 649600 ‘sia 8999 weet@'l L0%s sea €260 ore evel Moose 78s a) ais a) a4 “uD, Sane, WUOTITA 05 WORT Hewosdad 05 Moy, sais Tm m 7 I Igcrly =a1aeHe4 wapurdag) Sia14g iseaz0y ysATETY auNsopSIG YD" 19 [UE vee 3951 Borst 7 Fa so at {v="498I0) Tea U9D) Tee "1909 cn "5 SIaRA WHOTYTI %05 WHR wioTiad %0s Foy ‘ardureg wo a 7 ~ i ('zovuaAIV=aIg1TeA wWapusdag) aBe2409 IsKjouy aunso}>sIC YSO-ISOd “A ER January 2011, ‘American Accounting Association The Accounting Review (Sed mou uo pamquno) 3 sae. Suv 80 320 « OSD) Ab as 10 WADA IRD Fz a 4 For sramie# weswoy Sunune owe peas ays sce Jo sgune 29 jo eae uous at zovAAAO sR “29¥YEKGD -"ZQVEENOD = 3 (ep 1 uo So We pay 2 vu a OH = UOLABAIH sb tea sane ose! = OST some 81%, sino! aan 9 so fom ae putt gg pert peony Sg SRO IY sjosann sol SBE (por UD § OL ae t coer a rs so a ig 00 #200 we to00- wt 1000 ov0- ‘ooo- z- see65010- e- ee800- Orr pea 00 set- 4!L00- out “soa sot zoo t eat G00- te 5s001'0~ re9- #447900- 'sagHOWY wt 100 wo- sooo eco 00 ‘gouaiso 9. eee 1200- ore +8000 9- e510 0— ‘3zis9 sor yats00- ‘weoseaate +9810 0 1000 ‘WeOLaA TH sor vz00- wee vne8H00- eo s000 so #6000 wr #489000 oss L000 “Te 29 78s “IRD 1859 a =a, WIOTVd HOS WOHOR WYOINA %WOS O01 adues 1 in 7 T S (asta = age, ww2puedag) vojsiadaig sse0g is4euY NSO ASD-804 } suogqugec AREA ‘paso! anya sognH mp Bsn pstaason axe Suse seWaDIad 6S 2 pmuosulm am s2}gees snonuBHO> ITY ‘tonnsadss va] Worsd | pur sie s wand OW 943 WED Aiea 29200 PIER 3 EI PT a ae us 88 e108! u S38 wy eal wo ost sue ac195 ueo 200 oto ee sot senxl00 wo vero- wo 3109 soz ez aac £00 aaf8P2- 00 wg l68 I~ al we eestO1'0 00 weatito 000x080 v0 6 412070 aro vse ovo 191 oro 9s 6500 000 ees0IET 9 wewaStt-Z or a2 oro s2850- 000 +a 8960- so sse 00 wader 00 e652 we et s40lS70 070 a dH0Z- 00 enwSU¥T~ ‘vou oi (91 eaatt0 0 eax8EO- 00 eeaPtvo- sams oF i Bee #448000 sco £100 10 a4?200 ‘aw wet “989.0 set weet 0 850 wo $3080 ‘asia S99 pn 08E0- HLOI- ——44907O— 000 wund082- 00 wxs000'T= esi, 7a 0 Tes 2) cor aD aad a SIgeHA PARIS PiWVanSsr Foas Woas TL van} rewnjeD, Tr HEINE Teamiey sas 19 mg Pung sos J0 299p)01 *Y joaeg sBuysey Kuinba pauosees aunsojssid YSI-80d LaTavL, January 2011 American Accounting Association The Accounting Review 3 Voluntary Nenfaoncial Disclosure und the Cost of Equity Capital (pm (436 Bunenn vedas yg) svodsr > sons oy ay & 0 al Buns ware ap 51 (panupuos) 1 ITaVE [ senbe vey ate srrgpal = ‘psig January 2011 ‘American Accounting Association ‘The Aecounting Review 9 Dhaliwal, Li Teong. and Yong Dhaliwal et al. 2006). To provide assurance that our results are not Sensitive fo the choice of these measures, we repeat our analyses using the three measures separately and obtain similar results. Correlation between KLD Performance and CSR Disclosure KLD performance scores (PERFORM) are correlated with the decision of firms to issue CSR reports for the fist time (DISCI), but their correlation coetficients (Table 2, Panel B) aze moderate at about 10 percent.” As discussed earlier. we contol for CSR performance in all regression equations. To further alleviate the conccm regarding the correlation between the KLD perfor- mance scores and CSR disclosure of firms and the potential impact of this corelation on our results, we conduct the following robustness analyses. First, we cemove the transparency-related category, that is, “Comporate Governance,” from the KLD performance ranking scores, as this category contains a subcategory, “Transparency,” which is a dimension that is likely to reflect the CSR disclosure policy of firms. Our main inferences are unchanged, Second, we use the perfor: mance score of each KLD CSR category (o measure firms’ social performance. Our main infer ences are unchanged. Finally, we match each DISCI observation with a non-disclosing firm that has the closest industry-adjusted KILD CSR performance score in the same year and industry and run regression Equation (2). The coefficient on DISCT is significantly negative atthe conventional level, suggesting that CSR initiators enjoy a subsequent reduction in the Cost of equity capital Alternative Measures of CSR Performance To determine if our results are sensitive to alternative measures of CSR performance, we use ‘ovo additional measures. One measure is an indicator DJS/ that equals {if a firm appeared in the Dow Jones Sustainability Index in any year during the period 2002-2007, and 0 otherwise. The other is an indicator CRO that equals | ifa firm was on the “100 Best Corporate Citizens” list for 2007 from the Corporate Responsibility Officer. and 0 otherwise.” We do not consider year-to- year variation because of data constraint. I turns out DJS! (CRO) is correlated with our KLD. performance scores with a Pearson coefficient of 30 percent (23 percent). Using these two me sures in place of the KLD scores produces qualitatively similar results \V. SUMMARY AND CONCLUSIONS ‘We examine a potential benefit associated with the initiation of voluntary disclosure of CSR activities: a reduction in the cost of equity capital. We find that the likelihood of a firm initiating standalone disclosure of CSR activities is associated with a higher prior yeas cost of equity capital Firms with CSR performance superior to that oftheir industry peers enjoy a reduction in the cost of equity capital after they initiate CSR reports, Further, firms initiating CSR disclosure with superior CSR performance attract dedicated institutional investors and analyst coverage, and these analysts achieve lower absolute forecast errors and dispersion following such disclosure. Finally, CSR disclosure initiators appear to exploit this potential benefit ofa reduction in the cost of equity ® Consistent with tis low conelation, KLD provides he following information reacting how it tes the CSR perfor. Tanee of Firms (se hip.wwwSld.comvrescarchmethodsiogy nal): “KLD reaches the #9, environmental. and ovemance performance of eoporaions, KUD research relies oh Bve distinct datasources inform our ratings and nasi. Data are colete ina disciplined process from a wide variety of company, goverment and Hon-government tnganization and media sources. KLD tracks each company through more than 14,000 global media soures daily “These five distinct daa sources include (I) dee commanicaion with company offices: (2) nctwork of global ESG ‘esearch firms that cover non-US. markets: (3) review of rote then 14.000 globel news sources: (3) public Socuments of companis. including annual fepors and groxy sttemens: and (5) information ebsined from government and on-government organizations ielading the US, Depanaent of Labor, EPA. Human Rights Watch, OSHA, CANICOR, Ceres. ICCR, and Dab. Hence, st spears it dhe CSR reports oF frms conte only ane ofthe merous information a foures employed by KLD. See ip. wsvrthero ear The Accounting Review January 2011 ‘American Accounting Association siotamars Nonfinancial Disclosure and the Cost of Equity Capital 95 capitol. They are more likely than non-disclosing fiems to conduct SEOs to raise capital in the «wo years following the disclosure, In addition, among firms conducting SEOs. CSR disclosure initia- tors raise & significantly larger amount of equity capital than non-initators. This stady adds (o the voluntary disclosure literature by extending the traditional research on voluntary diselosure beyond the narrow focus of financial disclosure, Our analyses enhance our understanding of the rationales behind and the consequences of the recent trend in voluntary CSR disclosure. These results have important implications for companies, regulators, and investors 'A few caveats are worth noting. Most of the control variables that we use in the CSR determination mode! are obtained from the standard voluntary disclosure literature. To the extent that CSR disclosuce is distinct from other forms of voluntary disclosure examined in the literature, we may have missed important determinants of CSR disclosure. In addition, itis postible that we missed some reports on stale websites because of thcir lack of maintenance, which would add noise (© our results. Also, we do not examine the content of the CSR reports. To the extent that the detailed information of these reports is not fully captured by the KLD scores, we fail to capture some important characteristics of CSR reports. Further, itis important to control for the other isclosure policies of firms when examining the impact of CSR disclosure. Our empirical proxies using management guidance and earings quality may not be sufficient to capture these potentia! confounding effects. Finally, although the KLD rating is widely used in the management Jerature, 4 significant amount of future research is warranted to further establish its validity in measuring the social performance of firms. ‘These caveats notwithstanding, we believe that our study opens various venues for future research. For example, CSR disclosure and performance could have a different impact on the cost of debt as debtholders have a payoff function different from that of equityholders. Further, the cffect of CSR disclosure could be a function of differences in legal envionment and institutional setting. Therefore, an international study could help us better understand CSR disclosure. Last, 8 mentioned previously, it would be worthwhile to investigate the effect af the information content of CSR reports on the valuation decisions of investors January 2013, American Accounting Ass (rr The Accounting Review | Dhaliwal, Li Tsong. and Yong 96 5000 oyoGns som pur souoFeta wie Uanoe ou sop 2 sinsaMo 9 ue ‘ssa yfuans 42409 ep a Sy JO 2BUEULOPN 4S a ae 1 92 (S10

    You might also like