Are Analysts Overoptimistic about the Prospects of Sin Firms?

Transcription

1 Are Analyss Overopimisic abou he Prospecs of Sin Firms? Jin Zhang 1 & Haeyoung Shin 2 1 College of Business Adminisraion, California Sae Universiy, Sacrameno, USA 2 School of Business, Universiy of Houson Clear Lake, USA Correspondence: Jin Zhang, College of Business Adminisraion, California Sae Universiy, Sacrameno, USA. Received: Augus 30, 2017 Acceped: Sepember 14, 2017 Online Published: Sepember 19, 2017 doi: /ijfr.v8n4p99 URL: hps://doi.org/ /ijfr.v8n4p99 Absrac We invesigae he associaion beween he bias and accuracy of consensus analyss earnings forecass and wheher a firm is a sin firm or no. We measure analys forecas bias as he difference beween he consensus earnings forecas and he acual earnings, scaled by he sock price. We measure analys forecas accuracy as he negaive of he absolue value of he difference beween he firms forecased and acual earnings, scaled by he sock price. We find a posiive associaion beween he level of forecas opimism and sin firm membership. We find a negaive associaion beween he level of forecas accuracy and sin firm membership. Overall, hese resuls imply ha analyss end o issue over-opimisic and less accurae earnings forecass on sin firms. Keywords: analyss opimism, sin firms, analyss forecas accuracy 1. Inroducion Social norms are imporan in shaping economic behavior and marke oucomes. Susainable and Socially Responsible Invesmen (SRI) has coninued o grow a a faser pace han convenional invesmen asses, and is now quie popular in financial markes. Since 2005, SRI asses have increased more han 34% while he broader universe of professionally managed asses has increased only 3% (Social Invesmen Forum, 2010). A he beginning of 2010, he nearly one ou of every eigh dollars under professional managemen in he Unied Saes -approximaely 12% in oal asses-was involved in some sraegy of socially responsible invesing (Social Invesmen Forum, 2010). In general, socially responsible invesors favor corporae pracices ha promoe environmenal, consumer proecion, and human righs and avoid businesses involved in alcohol, obacco, gambling, weapons, or he miliary (Social Invesmen Forum, 2010). Meanwhile, a large number of informaion inermediaries have emerged wih he purpose of raing and ranking companies across several dimensions of environmenal, social and corporae governance performance. Sin socks (firms engaging in aciviies relaed o obacco, gambling and alcohol) are mos ofen negaively screened socks by socially responsible invesors and informaion inermediaries due o social norms. The neglec effec of sin socks has been examined by previous lieraure. For insance, Hong and Kacperczyk (2009) documen sin socks earn abnormal reurns afer accouning for marke bea, book o marke, size and momenum. They aribue he abnormal reurns o a neglec effec due o invesors social preference. They also find ha sin socks have lower level of insiuional ownership, and fewer analyss following. Kim and Venkaachalam (2011) invesigae financial reporing qualiy of sin socks. They find ha financial reporing of sin firms is superior along wo dimensions: predicabiliy of earnings for fuure cash flows and imely loss recogniion. Zhang (2012) find a negaive associaion beween sin firm membership and he magniude of absolue discreionary accruals, where discreionary accruals are adjused for firm performance, implying ha sin firms are less likely o manage heir earnings, specifically, less likely o engage in income-increasing earnings managemen. This paper examines he associaion beween he bias and accuracy of consensus analyss earnings forecass and wheher a firm is a sin firm or no. Biases in analyss earnings forecass are widely documened in he accouning and finance area. The naure and causes of analyss forecas bias has been he subjec of research for a leas hree decades. Earlier research finds ha in aggregae, analyss forecass are opimisically biased (e.g. Sickel (1991), Francis and Phibrick (1993)). Recen research indicaes ha managers do no uniformly prefer opimisic earnings forecass. Abarbanell and Lehavy (2003), hereafer referred as AL examine he properies of analyss earnings forecass, and provide insighs ino he vas seemingly mixed resuls on analyss forecas bias (see AL for an excellen summary). In paricular, AL documen wo disinc domains in he empirical disribuion of analys forecas errors ha differeniae i Published by Sciedu Press 99 ISSN E-ISSN

2 from a ypical bell-shaped normal disribuion: he middle and negaive ail asymmeries. The negaive ail asymmery is consisen wih opimisic forecass and he middle asymmery is consisen wih pessimisic forecass. AL inerpre heir resul by conjecuring ha such a paern is more likely o be caused by he earnings managemen raher han analyss choice, i.e., biased earnings raher han biased forecas. We invesigae he associaion beween he bias and accuracy of consensus analyss earnings forecass and wheher a firm is a sin firm or no. Sin socks are socks in he following indusries: alcoholic beverages, obacco and gaming. We follow he mehod similar wih Hong and Kacperczyk (2009) o idenify sin socks. Firs, we include all firms in SIC codes as par of obacco group and firms in he SIC codes as firms in he alcohol group. Second, we use he NAICS classificaion o idenify socks in he gaming indusry, i.e., socks in he NAICS codes of 7132, 71312, , , 72112, and We include in he sample all firm-quarer observaions spanning from 2000 o 2015 for which required daa are available on Compusa. We measure analys forecas bias as he difference beween he consensus earnings forecas and he acual earnings, scaled by he sock price. We measure analys forecas accuracy based on he negaive of absolue value of he difference beween he firms forecased and acual earnings, scaled by he sock price. Muliplying he absolue forecas error by (-1) give a measure ha increases wih greaer forecas accuracy. The ACCURACY values are negaive by consrucion (see Lang, Lins and Miller 2003)). We provide evidences on he associaion beween analys forecas and sin firm membership. We find a posiive associaion beween he level of forecas opimism and sin firm membership. We find a negaive associaion beween he level of forecas accuracy and sin firm membership. Overall, hese resuls imply ha analyss end o issue over-opimisic and less accurae earnings forecass on sin firms. The res of paper is organized as follows. Secion 2 inroduces he relaed research and develop he main hypoheses. Secion 3 describes he sample and mehodology. Secion 4 presens empirical evidence, and Secion 5 concludes he paper. 2. Hypoheses Developmen I has been well documened ha analyss earnings forecass exhibi overopimism (e.g., Abarbanell and Bernard (1992), Easerwood and Nu (1999)). Two (non-compeing) explanaions have emerged as o why analyss issue more opimisic forecass. The firs explanaion follows he sraegic bias model (Francis and Philbrick (1993); Das, Levine, and Sivaramakrishnan (1998); Lim (2001); Mes and Plummer (2003)), in which analyss have an incenive o issue favorable earnings esimaes o mainain heir relaionship wih managemen. Susaining a good relaionship wih corporae execuives is criical for analyss compeing o obain access o cerain informaion wihheld by managemen. In response o high earnings uncerainy, he sraegic bias view suggess ha analyss earnings esimaes would be upward biased. The second explanaion follows he selecion bias model (McNichols and O Brien (1997); Hayes and Levine (2000); Dieher, Malloy, and Scherbina (2002)), which argues ha because analyss are araced o make forecass when a firm s earnings reach above a cerain hreshold (self-censoring), forecas esimaes end o be opimisic. Furhermore, Dieher e al. (2002) show ha he forecas opimism resuling from self-censoring is sronger when earnings uncerainy measured by he variance of acual earnings is higher. This paper invesigaes he impac of corporae social responsibiliy on analys forecass. Sin socks (firms engaging in aciviies relaed o obacco, gambling and alcohol) are mos ofen negaively screened socks by socially responsible invesors and informaion inermediaries due o social norms. We expec ha forecas biases are differen for sin firms and non-sin firms, as sin firms are less socially responsible firms, and analyss may have social preferences when hey evaluae firms On one hand, Hong and Kacperczyk (2009) provide evidence on he poenial downside of having an invesmen philosophy ha adheres o social norms by focusing heir aenion on he invesmen environmen of sin socks. They documen ha sin firms have lower insiuional ownership and analys coverage, consisen wih he noion ha such socks are eschewed by an imporan group of capial marke paricipans. Hong and Kacperczyk (2009) and Kim and Venkaachalam (2011) documen ha hese sin firms behave like value socks and ouperform he marke afer conrolling for facors ha deermine expeced reurns. Collecively, heir evidence suggess ha sin socks are boh negleced and undervalued. I is plausible ha analyss issue more opimisic forecass on undervalued sin socks. Moreover, previous lieraures (Francis and Philbrick (1993); Das, Levine, and Sivaramakrishnan (1998); Lim (2001); Mes and Plummer (2003)) documen ha analyss have an incenive o issue favorable earnings esimaes o mainain heir relaionship wih managemen. Sin firms, which have lower insiuional ownership and analys Published by Sciedu Press 100 ISSN E-ISSN

3 coverage, have an incenive o provide more privae informaion o analyss. The analyss would be relaively opimisic abou he fuure profiabiliy of sin firms. On he oher hand, Lim s (2001) model of raional bias posis ha analyss issue more favorable forecass for firms wih less predicable earnings. Consisen wih Lim s (2001) predicion, Das e al. (1998) provide empirical evidence ha forecass are more opimisically biased when earnings are less predicable. Sin firms repor higher qualiy accouning informaion o arac sophisicaed marke paricipans, i.e., insiuional invesors and analyss (Kim and Venkaachalam, 2011). I is of a less concern o sin firms ha managing earnings o mee or bea analys forecas due o he neglec effec of sin firms (Zhang 2012). Since analyss issue opimisic earnings forecas when earnings are uncerain, sin firms, wih higher qualiy accouning informaion, may have less opimisic earnings forecass, compared wih non-sin firms. Moreover, corporae social responsibiliy (CSR) becomes favorably evaluaed by sell-side analyss in recen years (Ioannou and Serafeim (2015)). I is plausible ha analyss issue less opimisic forecass on sin firms, which are highly social irresponsible firms. Thus, i is an empirical quesion on wheher analyss are relaively more opimisic or less opimisic on he fuure of sin firms. The following hypoheses summarize my expecaions: H1a. Analyss are relaively more opimisic abou he fuure of sin firms. H1b. Analyss are relaively less opimisic abou he fuure of sin firms. 2.1 Conribuion This paper conribues he lieraure by exploring he impac of social norms on he capial marke and more specifically, by undersanding he underlying sociological processes ha deermine how hese analyss perceive and evaluae he firms wih poor corporae social responsibiliy, i.e., sin firms. Moreover, his paper conribues analyss bias lieraure by showing he significan difference of forecas bias beween sin firms and non-sin firms. 3. Daa and Research Design 3.1 Sample Selecion The firs objecive in he sample selecion process is o idenify a se of sin socks. Sin socks are socks in he following indusries: alcoholic beverages, obacco and gaming. We follow he mehod similar wih Hong and Kacperczyk (2009) o idenify sin socks. Firs, we include all firms in SIC codes as par of obacco group and firms in he SIC codes as firms in he alcohol group. Second, we use he NAICS classificaion o idenify socks in he gaming indusry, i.e., socks in he NAICS codes of 7132, 71312, , , 72112, and We obain all financial saemens informaion from he Compusa daabase. We include in he sample all firm-quarer observaions spanning from 2000 o 2015 for which required daa are available on Compusa. Table 1 provides a year by year look a my daa se of sin socks beginning in 2000 and ending a There is a oal of 408 sin companies, comprised of 50 disinc obacco companies, 108 disinc alcohol companies and 250 disinc gaming companies. Our sample is less han Hong and Kacperczyk (2009) s sample since hey use Compusa segmen daa o include firms ha have segmens operaing in any hese SIC or NAICS groups. We only include he firms whose main businesses are engaging in alcoholic beverages, obacco and gaming. There are no a lo of sin firms relaed o all oher firms. The number of sin socks doesn no change a lo in 2010 s. Table 1. Number of sin firms by year Year All Tobacco Alcohol Gaming Published by Sciedu Press 101 ISSN E-ISSN

4 Toal Noe: This able repors year by year he number of sin socks ha fall ino hree sub-groups of obacco, alcohol and gaming. The daa span he period of Main Variables We measure bias (BIAS) consisen wih prior research (Duru and Reeb (2002); Lang, Lins and Miller (2003); Herrmann, Hope and Thomas (2008)). The variable BIAS is he difference beween he laes consensus earnings forecas and he acual earnings, scaled by he sock price. BIAS FORECAST ACTUAL (OPTIMISM) PRICE where FORECAST is he laes consensus earnings forecas of he period. ACTUAL is he acual earnings per share before exraordinary iems of he period. Price -1 is he sock price a he ime of -1. A posiive (negaive) BIAS indicaes analys opimism (pessimism). Forecas opimism increases as BIAS becomes larger, while opimism decreases as BIAS becomes smaller. We measure forecas Accuracy based on absolue value of he difference beween he firms forecased and acual earnings, divided by sock price a ime. ACCURACY FORECAST PRICE 1 1 ACTUAL where ACCURACY is he negaive of he absolue forecas error a ime. Muliplying he absolue forecas error by (-1) give a measure ha increases wih greaer forecas accuracy. The ACCURACY values are negaive by consrucion (see Lang, Lins and Miller 2003)). The closer he accuracy value is o zero, he more accurae he forecas. 3.3 Empirical Models Consisen wih Zhang, Lin and Shin (2000), we esimae he following regression o es H1(a) and H1(b): BAIS j,q = α 0 + β 0 SINFIRM j,q + β 1 ANALYSTS j,q-1 + β 2 DISPERSION j,q-1 +β 3 LNSIZE j,q-1 + β 4 RETQ1 j,q-1 + β 5 UE j,q-1 + β 6 LOSS j,q + ε j,q. (3) BAIS is he difference beween he consensus earnings forecas and he acual earnings, scaled by he sock price. SINFIRM is our main variable of ineres. I equals one if he firm engages in aciviies relaed o obacco, gambling and alcohol. A significanly posiive coefficien on SINFIRM would be consisen wih our H1a. A significanly negaive coefficien on SINFIRM would be consisen wih our H1b. We include a number of addiional variables in he model o conrol for various facors ha may confound he relaionship beween forecas bias and sin firm membership. We conrol for he number of analyss (ANALYST) since Lys and Soo (1995) sugges ha he number of analyss proxies for he inensiy of compeiion in he marke. Greaer analys following indicaes greaer ex ane incenives for analyss o make accurae forecass and a beer ex pos informaion environmen of he followed company. ANALYST is he number of analyss following. We also conrol for analys disagreemen (DISPERSION). DISPERSION is measured as sandard deviaion earnings forecas scaled by he sock. (1) (2) Published by Sciedu Press 102 ISSN E-ISSN

5 We furher conrol for firm size (LNSIZE), as a proxy for a firm s general informaion environmen (Aiase, 1985). In addiion, various oher facors poenially relaed o firm size (Hope, 2003). Firm size (LNSIZE) is measured as he naural log of book value of oal asses a he beginning of he fiscal quarer. We also conrol for sock reurn over he previous quarer (RETQ1). RETQ1 is measured as he buy-and-hold raw reurn cumulaed over he previous quarer. Hayn (1995) argues ha he valuaion of loss firms may be differen from ha of profi ones because invesors are more likely o exercise heir liquidaion opion. Because he informaiveness of earnings for fuure cash flows is weaker among loss han among profi firms analyss will have greaer difficuly in forecasing he earnings of hose firms. We include an indicaor variable, LOSS, o conrol for his effec. LOSS equals one if he firm repors negaive earnings in he quarer and zero oherwise. We also conrol for price deflaed earning change (UE). UE is measured as he difference in earnings per share before exraordinary iems beween fiscal quarer o quarer -1, scaled by he sock price a he beginning of quarer. We also esimae he following regression o es he associaion beween analys forecas accuracy and sin firm membership: ACCURACY j,q = α 0 + β 0 SINFIRM j,q + β 1 ANALYSTS j,q-1 + β 2 DISPERSION j,q-1 +β 3 LNSIZE j,q-1 + β 4 RETQ1 j,q-1 + β 5 UE j,q-1 + β 6 LOSS j,q + ε j,q. (4) ACCURACY is he negaive of he absolue forecas error a ime. Muliplying he absolue forecas error by (-1) give a measure ha increases wih greaer forecas accuracy. SINFIRM is our main variable of ineres. I equals one if he firm engages in aciviies relaed o obacco, gambling and alcohol. A significanly posiive coefficien on SINFIRM would be consisen wih predicion ha analyss issue more accurae earnings forecass on sin firms. A significanly negaive coefficien on SINFIRM would be consisen wih predicion ha analyss issue less accurae earnings forecass on sin firms. We include he same conrol variables in he equaion (3) o conrol for various facors ha may confound he relaionship beween forecas accuracy and sin firm membership. 4. Resuls Table 2 provides descripive saisics of he variables used in he empirical analysis for boh sin firms and all oher firms. Sin socks are larger and are characerized by higher reurn (mean of for sin firms vs for all oher firms, median of for sin firms vs for all oher firms), consisen wih he conjecure ha sin socks are perhaps value socks. The number of analys following is significanly differen beween sin firms and all oher firms (mean of for sin firms vs for all oher firms, median of for sin firms vs for all oher firms). Bias is larger for sin firms (mean of for sin firms vs for all oher firms), alhough he difference is no saisically significan (=-1.05). Many of he conrol variables are significanly differen beween sin and non-sin firms. Thus i is imporan o conrol hese variables o draw proper conclusions abou relaion beween analys opimism and wheher a firm is a sin firm or no. Table 2. Descripive saisics Panel A: Mean Sin firms (1) All oher firms (2) Difference -sa (2-1) BIAS (%) ACCURACY (%) *** ANALYSTS *** DISPERSION *** LNSIZE *** RETQ UE LOSS *** # of OBS 1, ,063 Published by Sciedu Press 103 ISSN E-ISSN

6 Panel B: Median Sin firms (1) All oher firms (2) Difference Z-sa (2-1) BIAS (%) *** ACCURACY (%) *** ANALYSTS *** DISPERSION *** LNSIZE *** RETQ ** UE * LOSS *** # of OBS 1, ,063 Noes: ***, **, and * indicae significance a he 1%, 5%, and 10% level, respecively. SINFIRM equals one if he firm engages in aciviies relaed o obacco, gambling and alcohol. BIAS is he difference beween he consensus earnings forecas and he acual earnings, scaled by he sock price. ACCURACY is he negaive of he absolue forecas error, scaled by he sock price. ANALYSTS is he number of analyss following. DISPERSION is sandard deviaion earnings forecas scaled by he sock. LNSIZE is measured as he naural log of marke value of he firm a he beginning of he fiscal quarer. UE is he difference in earnings per share before exraordinary iems beween fiscal quarer and quarer 1, scaled by he sock price a he ending of quarer 1. RETQ1 is he buy-and-hold raw reurn cumulaed over he previous quarer. LOSS equal o 1 for observaions wih negaive earning, zero oherwise. Table 3 presens he regression esimaes of Equaion (3). The coefficien on SINFIRM is posiive and significan in he BIAS model (0.157, =3.96). The coefficiens on he conrol variables, whenever significan, are consisen wih he signs prediced by previous sudies. Overall, he resuls sugges ha here is a posiive associaion beween analys opimism and wheher his firm is a sin firm. Table 3. Sin firms and analys forecas bias, Dependen Var. =BIAS Coeff. -sa. Inercep *** SINFIRM *** ANALYSTS *** DISPERSION *** LNSIZE *** RETQ *** UE *** LOSS *** Adjused R 2 (%) No. of Observaions 156,557 Noes: 1. Model: BAIS j,q = α 0 + β 0 SINFIRM j,q + β 1 ANALYSTS j,q-1 + β 2 DISPERSION j,q-1 +β 3 LNSIZE j,q-1 + β 4 RETQ1 j,q-1 + β 5 UE j,q-1 + β 6 LOSS j,q +ε j,q. 2. ***, **, and * indicae ha he esimaed coefficiens are saisically significan a he 1%, 5%, and 10% level, respecively. Degree of Freedom is 156,549. Published by Sciedu Press 104 ISSN E-ISSN

7 3. SINFIRM equals one if he firm engages in aciviies relaed o obacco, gambling and alcohol. BIAS is he difference beween he consensus earnings forecas and he acual earnings, scaled by he sock price. ANALYSTS is he number of analyss following. DISPERSION is sandard deviaion earnings forecas scaled by he sock. LNSIZE is measured as he naural log of marke value of he firm a he beginning of he fiscal quarer. UE is he difference in earnings per share before exraordinary iems beween fiscal quarer and quarer 1, scaled by he sock price a he ending of quarer 1. RETQ1 is he buy-and-hold raw reurn cumulaed over he previous quarer. LOSS equal o 1 for observaions wih negaive earning, zero oherwise. Table 4 presens he regression esimaes of Equaion (4). The coefficien on SINFIRM is negaive and significan in he ACCURACY model (-0.095, =-2.29). The coefficiens on he conrol variables, whenever significan, are consisen wih he signs prediced by previous sudies. Overall, he resuls sugges ha here is a negaive associaion beween analys forecas accuracy and wheher his firm is a sin firm. Table 4. Sin firms and analys forecas accuracy, Coeff. Dependen Var. =ACCURACY -sa. Inercep *** SINFIRM ** ANALYSTS *** DISPERSION *** LNSIZE *** RETQ *** UE *** LOSS *** Adjused R 2 (%) No. of Observaions 156,557 Noes: 1. Model: ACCURACY j,q = α 0 + β 0 SINFIRM j,q + β 1 ANALYSTS j,q-1 + β 2 DISPERSION j,q-1 +β 3 LNSIZE j,q-1 + β 4 RETQ1 j,q-1 + β 5 UE j,q-1 + β 6 LOSS j,q + ε j,q. 2. ***, **, and * indicae ha he esimaed coefficiens are saisically significan a he 1%, 5%, and 10% level, respecively. Degree of Freedom is 156, SINFIRM equals one if he firm engages in aciviies relaed o obacco, gambling and alcohol. ACCURACY is he negaive of he absolue forecas error, scaled by he sock price. ANALYSTS is he number of analyss following. DISPERSION is sandard deviaion earnings forecas scaled by he sock. LNSIZE is measured as he naural log of marke value of he firm a he beginning of he fiscal quarer. UE is he difference in earnings per share before exraordinary iems beween fiscal quarer and quarer 1, scaled by he sock price a he ending of quarer 1. RETQ1 is he buy-and-hold raw reurn cumulaed over he previous quarer. LOSS equal o 1 for observaions wih negaive earning, zero oherwise. 5. Conclusions This paper provides evidences on he associaion beween he bias and accuracy of consensus analyss earnings forecass and wheher a firm is a sin firm or no. We measure analys forecas bias as he difference beween he consensus earnings forecas and he acual earnings, scaled by he sock price. we measure analys forecas accuracy as he negaive of absolue value of he difference beween he firms forecased and acual earnings, scaled by he sock price. We find a posiive associaion beween he level of forecas opimism and sin firm membership. We find a negaive associaion beween he level of forecas accuracy and sin firm membership. Overall, hese resuls imply ha analyss end o issue over-opimisic and less accurae earnings forecass on sin firms. This paper conribues he lieraure by exploring he impac of social norms on he capial marke and more specifically, by undersanding he underlying sociological processes ha deermine how hese analyss perceive and evaluae he firms wih poor corporae social responsibiliy, i.e., sin firms. Moreover, his paper conribues analyss bias lieraure by showing he significan difference of forecas bias beween sin firms and non-sin firms. Published by Sciedu Press 105 ISSN E-ISSN

8 References Abarbanell, J., & R. Lehavy. (2003). Biased Forecass or Biased earnings? The role of Repored Earnings in explaining Apparen Bias and Over/Underreacion on Analyss Earning Forecass. Journal of Accouning & Economics, 36, hps://doi.org/ /j.jacceco Abarbanell, J.S., & V.L. Bernard. (1992). Tess of Analyss Overreacion/Underreacion o Earnings Informaion as an Explanaion for Anomalous Sock Price Behavior. Journal of Finance, 47, Aiase, R.K. (1985). Predisclosure Informaion, Firm Capializaion, and Securiy Price Behavior around Earnings Announcemens. Journal of Accouning Research, 23(1), hps://doi.org/ / Das, S., C.B. Levine, & K. Sivaramakrishnan. (1998). Earnings Predicabiliy and Bias in Analyss Earnings Forecass. Accouning Review, 73, Dieher, K.B., C.J. Malloy, & A. Scherbina. (2002). Differences of Opinion and he Cross-Secion of Sock Reurns. Journal of Finance, 57, hps://doi.org/ / Duru, A., & D.M. Reeb. (2002). Inernaional Diversificaion and Analyss Forecas Accuracy and Bias. Accouning Review, 77, hps://doi.org/ /accr Easerwood, J.C., & S.R. Nu. (1999). Inefficiency in Analyss Earnings Forecass: Sysemaic Misreacion or Sysemaic Opimism?. Journal of Finance, 54, hps://doi.org/ / Francis, J., & D. Philbrick. (1993). Analyss Decisions as Producs of a Muli-Task Environmen. Journal of Accouning Research, 31, hps://doi.org/ / Hayes, R.M., & C.B. Levine. (2000). An Approach o Adjusing Analyss Consensus Forecass for Selecion Bias. Conemporary Accouning Research, 17, hps://doi.org/ /j b00911.x Hayn, C. (1995). The Informaion Conen of Losses. Journal of Accouning and Economics, 20, Herrmann, D., O. Hope, & W.B. Thomas. (2008). Inernaional Diversificaion and Forecas Opimism: The Effecs of Reg FD. Accouning Horizons, 22(2), hps://doi.org/ /acch Hong, H., & M. Kacperczyk. (2009). The Price of Sin: The Effecs of Social Norms on Markes. Journal of Financial Economics, 93, hps://doi.org/ /j.jfineco Hope, O. (2003, May). Disclosure Pracices, Enforcemen of Accouning Sandards, and Analyss Forecas Accuracy: An Inernaional Sudy. Journal of Accouning Research, 41, hps://doi.org/ / x Ioannou, L., & G. Serafeim. (2015). The Impac of Corporae Social Responsibiliy on Invesmen Recommendaions: Analyss' Percepions and Shifing Insiuional Logics. Sraegic Managemen Journal, 36(7), Kim, I., & M.Venkaachalam. (2011). Are Sin Socks Paying he Price for Accouning Sins. Journal of Accouning, Audiing & Finance, 26(2), hps://doi.org/ / x Lang, M., K. Lins, & D. Miller. (2003). ADRs, Analyss, and Accuracy: Does Cross Lising in he U.S. Improve a Firm s Informaion Environmen and Increase Marke Value?. Journal of Accouning Research, 41, Lim, T. (2001). Raionaliy and Analyss Forecas Bias. Journal of Finance, 56, hps://doi.org/ / Lys, T., & L. Soo. (1995). Analyss Forecas Precision as A Response o Compeiion. Journal of Accouning, Audiing and Finance, 10, hps://doi.org/ / x McNichols, M., & P. O Brien. (1997). Self-Selecion and Analys Coverage. Journal of Accouning Research, 35, hps://doi.org/ / Mes, D.P., & E. Plummer. (2003). Analyss Raionaliy and Forecas Bias: Evidence from Sales Forecass. Review of Quaniaive Finance and Accouning, 21, hps://doi.org/ /a: Social Invesmen Forum. (2011) Repor on Socially Responsible Invesing Trends in he Unied Saes. Rerieved from hp://ussif.org/resources/research/documens/2010trendses.pdf Sickel, S. E. (1991). Common Sock Reums Surrounding Earnings Forecas Revisions: More Puzzling Evidence. Accouning Review, 66(2), Zhang, J, E. Lin, & H. Shin. (2010). S&P 500 Index Inclusions and Analyss Forecas Opimism. The Journal of Invesing, Winer, hps://doi.org/ /joi Zhang, J. (2012). Are Sin Firms Less Likely To Manage Their Earnings?. Theorical Accouning Research, Fall, Published by Sciedu Press 106 ISSN E-ISSN