The Adverse Impact of Gradual Temperature Change on Capital Investment

Size: px
Start display at page:

Download "The Adverse Impact of Gradual Temperature Change on Capital Investment"

Transcription

1 The Adverse Impac of Gradual Temperaure Change on Capal Invesmen Ronald Balvers Dvson of Economcs and Fnance Wes Vrgna Unversy Morganown, WV E-mal: Phone: (304) Dng Du The W. A. Franke College of Busness, PO Box Norhern Arzona Unversy Flagsaff, AZ E-mal: Phone: (928) Xaobng Zhao The W. A. Franke College of Busness, PO Box Norhern Arzona Unversy Flagsaff, AZ E-mal: Phone: (928) Seleced Paper prepared for presenaon a he Agrculural & Appled Economcs Assocaon s 2012 AAEA Annual Meeng, Seale, Washngon, Augus 12-14, 2012 Copyrgh 2012 by Ronald Balvers, Dng Du, and Xaobng Zhao. All rghs reserved. Readers may make verbam copes of hs documen for non-commercal purposes by any means, provded ha hs copyrgh noce appears on all such copes. 0

2 Absrac Fnancal marke nformaon can provde an objecve assessmen of losses ancpaed from global warmng. In a Meron-ype asse prcng model, wh asse prces affeced by perceved changes n nvesmen opporunes due o global warmng, he rsk premum s sgnfcanly negave and growng over me, loadngs for mos asses are negave, and asse porfolos n more vulnerable ndusres have sronger negave loadngs on he global warmng facor. Average ncreases n requred reurns arbued o global warmng are 0.11 percen, mplyng a presen value loss of 4.18 percen of wealh. These coss complemen prevous esmaes of he cos of global warmng. JEL Codes: G12, Q54 Key Words: Asse Prcng, Global Warmng, Cos of Capal, Trackng Porfolos. 1

3 The Adverse Impac of Gradual Temperaure Change on Capal Invesmen 1. Inroducon Evdence n suppor of he noon of clmae change global warmng s accumulang. In a comprehensve and scenfc assessmen of clmae change, he 2007 Nobel Peace Prze recpen Inergovernmenal Panel on Clmae Change (IPCC) concludes ha Warmng of he clmae sysem s unequvocal (IPCC, 2007, Workng Group I Repor Summary for Polcymakers, p. 5). The majory of he research concernng he economc mpac of global warmng has emphaszed he fuure effec on producon or cash flows of an abrup change n clmae o occur someme afer he year Promnen examples are he work of Mendelsohn e al. (2000) and Nordhaus and Boyer (2000) examnng and aggregang he oupu losses by secor, and he work of Horowz (2001) and Nordhaus (2006) denfyng aggregae oupu losses wh a reduced-form cross-seconal approach, occurrng once emperaures ncrease by a leas 2.5 o 3.0 degrees Celsus. The hghes esmae of he adverse mpac of a 3.0 degrees Celsus ncrease n emperaure s a hree percen permanen decrease n GDP. Afer evaluang varous clmae-change mgaon sraeges, IPCC esmaes ha sablzng CO 2 concenraons and lmng he long-erm emperaure rse may resul n a 0.12 percen reducon n he annual growh rae of global GDP. Ths esmae provdes a hreshold for clmae-change mgaon polces: polces are worh consderng only f he negave economc mpac of global warmng represens a greaer cos han he The drec mpac on producon of a rse n mean emperaure sayng whn 2.5 degrees Celsus s usually small and even posve. See Tol (2008) for an excellen revew. 2

4 percen reducon n GDP growh or a presen value loss of 4.56 percen of wealh. 2 Because he GDP losses due o global warmng of up o hree percen occur so far n he fuure, he presen value of hese losses s usually oo small o jusfy aggressve mgaon polces. Tol (2008, p.5) concsely summarzes he sae of research by sang I s herefore no surprse ha cos-benef analyses of clmae change recommend only lmed greenhouse gas emsson reducon. We ake a dfferen approach n hs paper and focus on he economc and fnancal mpac of gradual emperaure change. Ths s movaed by he fndngs of IPCC. IPCC (2007) pons ou ha gradual global warmng can ncrease he nensy and frequency of exreme evens. Wh gradual clmae change, IPCC projecs ha drough, heavy precpaon, hea waves, exreme sea levels, and nense ropcal cyclone acvy wll lkely ncrease (see also NACC, 2000; Sern, 2007; IPCC, 2012). Exreme weaher evens ypcally adversely affec producon and cash flows of frms. Sern (2007) pons ou ha exreme evens (such as sorms, floods, droughs, and hea waves) could lead o sgnfcan nfrasrucure damage and faser capal deprecaon, and ha ncreases n exreme evens wll be parcularly cosly for developed economes because hey nves a consderable amoun n fxed capal each year (a good example s Hurrcane Karna). 3 Sern (2007) esmaes ha he annual losses due o exreme weaher evens have been around 0.2 percen of world GDP snce he 1990s, and ha he annual losses could reach percen of world GDP by he mddle of he cenury. 4 2 See Secon 6 below for deals. 3 The 6/30/08 (pre-hurrcane-ike) ssue of he Wall Sree Journal repors ha Bad weaher has cos U.S. propery nsurers more han $5 bllon so far n second-quarer caasrophe-relaed clams equal o abou hree-quarers of all caasrophe clams durng 2007 and could push he ndusry o an underwrng loss. 4 Asde from exreme evens, Quggn and Horowz (2003) emphasze adjusmen coss o clmae change. They argue ha coss of adjusmen wll arse f capal socks: () are dependen on clmae for her opmal locaon; and () deprecae more slowly han s requred o perm easy adjusmen o a changng 3

5 I s well known and poenally mporan for capal budgeng decsons ha mpacs of global warmng on cash flows could also be hghly unceran, whch parly sems from consderable uncerany abou he magnude of global warmng, wh he range of possble magnudes nvolvng dfferences of several hundred percen. Ths s llusraed by he IPCC s wdely-quoed esmaes for he possble ncrease n global mean emperaure. Is bes esmae for global average surface warmng a he end of he 21s cenury ranges from 1.8 C (wh 66 percen confdence nerval from 1.1 C o 2.9 C) o 4.0 C (wh 66 percen confdence nerval from 2.4 C o 6.4 C). Heal and Krsröm (2002) and Pndyck (2012) argue ha uncerany s a cenral feaure of clmae change. In bref, global warmng (negavely) affecs cash flows of frms n an unceran fashon, acng as a (negave) supply shock. Ths shock should affec frms and economy no only by s mpac on he level of cash flows, bu also by s mpac on he rsk or uncerany of cash flows and herefore he cos of capal. Global warmng (as a supply shock) nroduces a new rsk facor no he economy. As a resul, nvesors demand hgher reurns on rsky asses such as socks and corporae bonds. Hgher reurns represen a hgher cos of capal o frms, whch can have a profound and adverse mpac on capal nvesmen and economc growh. The negave lnk beween he cos of capal and economc growh s well documened n he economcs leraure (see for nsance Henry, 2003). Thus, gnorng global warmng s mpac on he cos of capal may sgnfcanly undersae s economc mpac. Ye a revew of he leraure shows ha clmae. See also Nordhaus (1991), Clne (1992), Mendelsohn, Nordhaus, and Shaw (1994), Fankhauser (1995), Tol (1995), Rosenhal, Gruenspech, and Moran (1995), Moore (1998), Mendelsohn and Neumann (1999), Schlenker, Hanemann, and Fsher (2006), Barreca (20112), and IPCC (2007) for he varous mpac mechansms of global warmng, 4

6 here has been no research concernng he mpac of global warmng on he cos of capal. 5 We explore here n deal he cos-of-capal lnk. A relaed objecve s o provde a deached assessmen of possble coss of global warmng. Roll (1984) showed ha fnancal markes may be more relable ndcaors of fuure evens nfluencng cash flows han he pronouncemens of specalzed forecasers: orange juce fuures prces have forecasng power for weaher condons (nvolvng fros) beyond he forecass of he Naonal Weaher Servce. Snce he scenfc leraure s no defnve concernng he magnude of he welfare effecs of global warmng, he observable fnancal marke mpac of changes n he lkelhood of global warmng, lkewse, may provde an objecve measure of he collecve percepon of some of he economc damages from global warmng. If global warmng s economcally mporan, he rsk premum and loadngs assocaed wh he global-warmng facor should be sgnfcanly dfferen from zero. In parcular, f clmae-change mpacs are generally adverse (losses due o exreme evens and/or adjusmen coss), we expec o see a negave premum, because (1) f clmaechange mpacs are adverse, ncreases n emperaure should for ypcal frms lead o lower reurns, and so he loadngs on he global-warmng facor should be generally negave; (2) o compensae for he addonal rsk, he equlbrum reurn should be 5 Kamsra, Kramer and Lev (2003) examne how emperaure affecs human mood and nvesor behavor, bu no from he clmae change perspecve. Fankhauser and Tol (2005) sudy he mpac of global warmng on capal accumulaon, bu no from he cos of capal perspecve. Brunner (2002) analyzes he effecs of he El Nño weaher cycle on commody prces and economc acvy bu does no consder he cos of capal. Prevous fnance leraure has examned he neracon beween weaher and fnancal markes. For nsance, Roll (1984) and Flemng e al. (2006) sudy how weaher-relaed nformaon affecs, respecvely, mean reurns and reurn varances n parcular weaher-sensve markes, and Hrshlefer and Shumway (2003) consder how weaher mpnges on reurns by affecng rader audes. These sudes focus on hgher-frequency (daly) weaher evens and subsequen reurn realzaons. Our paper focuses on a very dfferen aspec of he neracon beween weaher and fnancal markes: he mpac of lowerfrequency (monhly) weaher evens, relang o changng expecaons abou clmae change, on sysemac rsk and he economy-wde cos of capal. 5

7 hgher, and wh negave loadngs hs can only be acheved by a negave rsk premum. In essence we conjecure ha global warmng s a rsk facor ha s prced because affecs nvesmen opporunes n he sense of Meron (1973). We employ wo approaches o esmae he posed rsk premum of a global warmng facor. The frs s he rackng porfolo approach of Lamon (2001). The advanage of hs approach s ha does no force us o selec an equlbrum asse prcng model and herefore may avod model specfcaon error. Bu hs approach canno provde us wh an esmae of he loadngs on a global warmng facor and he mpac on he cos of capal. We herefore also employ a second, more srucural approach he wo-pass regresson mehodology proposed by Black, Jensen, and Scholes (1972) and Fama and MacBeh (1973), as amended by Shanken (1992), Lewellen, Nagel and Shanken (2009), and Shanken and Zhou (2007). Conssen wh our conjecure, we fnd evdence suggesng ha sgnals of global warmng cause an ncrease n he cos of capal. Quanavely hs evdence suggess ha he mpac of global warmng may be larger han prevously hough. The remander of he paper s organzed as follows: Secon 2 dscusses he heorecal perspecve and saes hree esable hypoheses as well as presenng he approprae expresson for calculang he welfare loss. Secon 3 brefly dscusses he daa. Secon 4 presens a non-srucural approach he rackng porfolo approach ha does no nvolve specfcaon of a parcular asse prcng model and allows us o es wo of he hypoheses, whle Secon 5 ess a hrd hypohess and provdes quanave resuls n a srucural conex applyng he wo-pass regresson mehodology o he sandard Fama-French and CAPM asse prcng models. Secon 6 nvesgaes 6

8 specfcally he quanave mpac of clmae change on he cos of capal, and Secon 7 concludes he paper wh a bref summary. 2. Theorecal Perspecve and Hypoheses A general perspecve on he facors explanng asse reurns s provded by he Meron (1973) model. Meron s model mples ha excess reurns are deermned by he sysemac rsk of an asse, whch n urn depends on he asse s sensvy o boh marke reurns and aggregae changes n nvesmen opporunes. The applcaon of he model s no sraghforward because provdes no specfcs abou how o characerze aggregae changes n nvesmen opporunes. Of he wo mos popular asse prcng models, he Sharpe-Lnner CAPM gnores he possbly of nvesmen opporuny shfs, and he Fama-French Three-Facor model emprcally capures nvesmen opporuny shfs by means of a value and a sze facor. In a seady sae, clmae canno be consdered a deermnan of changes n nvesmen opporunes. Parcular weaher realzaons affec asse reurn realzaons hrough sandard channels bu do no affec fuure nvesmen opporunes. The naure of a ranson perod beween clmae saes arsng on accoun of global warmng, however, s ha poenal changes n clmae are boh unceran and perssen. Snce clmae srongly affecs producon and nvesmen opporunes (see for nsance Chonère and Horowz, 2006), he clmae changes assocaed wh global warmng would be key deermnans of flucuaons n nvesmen opporunes, and should be prced n he conex of he Meron model, or as a producvy shock n he conex of Balvers and Huang (2007). 7

9 We posulae ha clmae changes represen a prced Meron facor and check emprcally f he mplcaons of hs vew are confrmed by he daa. The followng hypoheses derve more or less drecly from hs posulae and wll be he arge of our emprcal analyss n followng secons. HYPOTHESIS 1. The global warmng facor has a sgnfcan and negave rsk premum. If he rsk facor s prced, should have a sascally and economcally sgnfcan rsk premum. Asses wh reurns ha ncrease wh global warmng shocks (posve loadngs on he global warmng facor) consue a hedge agans adverse shocks o nvesmen opporunes, hus requrng lower average reurns. Snce he presumpon s ha mos frms are negavely affeced by ncreased warmng (he facor loadngs on he global warmng facor are generally negave), mos asses enhance global warmng rsk raher han hedge agans. The rsk premum mus be negave o ensure ha rsker asses have hgher average reurns a decrease n fuure nvesmen opporunes s added on op of he nal negave weaher shock; and frms whose nvesmen opporunes are affeced more are accordngly rsker. 6 The awareness of he possbly of global warmng has ncreased over he las several decades and hs can be aken o mply ha changes n weaher paerns may be ncreasngly hough as arbuable o global warmng and hus more perssen, mplyng a larger rsk premum. 6 There s no real sgnfcance o he negave sgn of he rsk premum. If we defned weaher shocks wh he oppose sgn (as an unexpeced decrease n emperaure), he loadngs would be usually posve, and he rsk premum would be posve, leadng o equvalen cos of capal mplcaons. 8

10 HYPOTHESIS 2. The mpac of he global warmng facor on he average cos of capal ncreases over he global warmng perod sarng n We here ake as he sar of global warmng he year 1976 as clamed by IPCC (2007). One ssue wh global warmng, poenally conrbung o s surroundng conroversy, s ha may have an uneven mpac. In parcular, some ndusres (see IPCC, 2007 and Quggn and Horowz, 2003) are more vulnerable: agrculure and foresry, ransporaon, real, commercal servces, oursm, nsurance, hose ndusres dependen on clmae-sensve npus (such as food processng), and ndusres wh long-lved capal asses. If global warmng s ruly a rsk facor, he ndusres ha are more sensve o clmae changes should have hgher loadngs on he global warmng facor and, accordngly, hgher requred reurns and coss of capal. HYPOTHESIS 3. The ndusres ha are consdered o be more vulnerable o global warmng have hgher loadngs on he global warmng facor. To address hese hree hypoheses wh some accuracy we ask a raher narrower queson han he queson of wheher global warmng affecs he cos of capal. For reasons of mplemenaon we ask: How does news abou changes n average U.S. emperaures over he perod classfed by IPCC as he global warmng perod affec he cos of capal for U.S. eques? Our belef s ha by focusng on a case for whch we have accurae daa and a clearly defned se of quesons, we can begn o oban nsghs and deas ha are more generally applcable o he global case. 9

11 As an mplcaon, conngen on confrmaon of he above hypoheses, we can formulae a measure for he losses semmng from gradual clmae change s effec on he cos of equy capal. PROPOSITION. (A) The aggregae change n he cos of equy capal due o he global warmng facor equals he value-weghed average global warmng bea mes he global warmng rsk premum. (B) The aggregae losses n average shareholder value can be approxmaed by he average change n he cos of equy capal mes he average prce-dvdend rao. Par (B) of he proposon follows from a Gordon growh model approxmaon for equy value and by dfferenang wh respec o he cos of capal. 3. Daa IPCC (2007) pons ou ha global warmng seems o have been akng place n wo dsnc phases, he frs beween 1910 and 1945, and he second from The decrease n he global emperaure beween 1945 and 1976 may be due o subsanal sulphur emssons and changed sunspo acvy n hs perod. Therefore, he pos-wwii perod offers wo dsnc perods o examne, he pre-warmng of 1946 o 1975 and he global warmng of 1976 hrough We om he pre-1946 perod, avodng he Grea Depresson and world war perods, o oban more relable esmaes of he rsk premum of global warmng. In erms of locaon, we lm ourselves o he Uned Saes. Whereas global warmng s by defnon a world-wde phenomenon, here are complcaed dsrbuon 10

12 ssues ha we are no prepared o address. Addonally, he requred weaher and fnancal-marke daa are avalable for he U.S. and are unformly accurae n parcular, clmae measures, macro and rsk facor daa, and ndusry-specfc equy porfolo reurns. The porfolo reurns and Fama-French facor daa are from Kenneh French s webse. 7 The macro varables daa are from he Federal Reserve Bank S. Lous. The emperaure me seres s he US average emperaure seres obaned from he Naonal Clmac Daa Cener. 4. Economc Trackng Porfolos: A Non-Srucural Approach Asse reurns are drven by changng nformaon, news, abou fuure cash flows and dscoun raes. Wha maers for our purposes s he news concernng fuure global warmng conaned n he curren observaon. We could use a srucural model o esmae he news concernng fuure global warmng (and wll do so n a laer secon o check robusness). Bu dong so resuls n a jon es of he valdy of he model for wha consues news and he valdy of our hypohess ha global warmng maers for fnancal markes. To crcumven hs ssue, we use he economc rackng porfolo approach proposed by Breeden, Gbbons, and Lzenberger (1989) and Lamon (2001), and prevously appled by Vassalou (2003) and Kapada (2011). Imporanly, hs sascal approach allows us o esmae he rsk premum of global warmng whou mposng a parcular model of asse prcng A Sascal Model of Clmae News 7 hp://mba.uck.darmouh.edu/pages/faculy/ken.french/. 11

13 For measuremen purposes we equae clmae change and global warmng wh a change n average emperaures. The exan emperaure modelng leraure (see IPCC, 2007) ofen uses srucural amospherc models. As Campbell and Debold (2005) pon ou however, alhough such an approach may be bes for modelng and forecasng for a shor horzon, s no a all obvous ha s bes for longer horzons. Successful modelng and forecasng does no necessarly requre a srucural model, and n he las several decades sascans and economercans have made grea srdes n he nonsrucural modelng and forecasng of me seres rend, seasonal, cyclcal, and nose componens. We herefore follow Campbell and Debold (2005) and ake a smple me seres approach o modelng and forecasng emperaure. 8 Our me-seres model s he daly emperaure model of Campbell and Debold (2005) amended o deal wh monhly observaons. We nclude a lnear me rend and 11 monhly seasonal dummes for our monhly emperaure daa. Tha s, 11 T C A rend S seasonal error, (1a) 1 wh T represenng he average emperaure n monh, and error represenng he effecs of all oher varables. Defnng CT T 12 1 as he average emperaure over he nex year, we have CT c a rend e, (1b) See also Harvey (1989), Seaer (1993), and Vsser and Molenaar (1995). 12

14 where c C A S, a A, and e 12 error , capures he effecs of all oher varables beween +1 and +12. Therefore, E CT ) c a rend E ( e ). (1c) 1( Now defne E CT ) E ( CT ) E ( CT ) ( as he news componen n he emperaure observaon n monh. Then we have he auology CT, (2) 12 E 1( CT 12) E ( CT 12) 12 whch decomposes he annual emperaure a he end of he followng year n a prevously expeced componen, a news componen, and nose, wh nose e E e ) ( An Economc Trackng Porfolo for Clmae Change If global warmng maers for asse prcng, nnovaons n excess reurns of base asses reflec nnovaons n expecaons abou fuure global warmng. Tha s, ~ E ( CT 12 ) br, (3) where R ~ ~ represens a column vecor of unexpeced reurns R R E ( R ), wh R a 1 column vecor of excess reurns of base asses n monh, and he componen of news ha s orhogonal o he unexpeced reurns of he base asses. Assume ha he base asse reurn n monh s a lnear funcon of Z 1, a vecor of condonng economc varables known a perod -1, and ha e 12 s a lnear funcon Z, and Z ' 1 a vecor of condonng clmae varables known a perod -1. Tha s, of 1 E 1 ( R ) d Z 1 and E 1 ( e 12) f Z 1 g Z' 1. Then we have from equaons (1b), (2), and (3) ha 13

15 CT 12 E 1( CT 12) E ( CT 12) 12 c a rend f Z 1 gz' 1 b ( R dz 1). 12 Ths provdes he Prmary Specfcaon: CT c a rend gz br ez, (4) 12 ' wh e = - bd + f, and Here CT +12 represens he average annual emperaure from +1 o +12; he vecor R = (SL, SM, SH, BL, BM, BH ) represens excess reurns: he Fama-French sze-bm porfolos ne of he one-monh T-Bll rae; Vecor Z -1 = (RF -1, DEF -1-,TERM -1 ), represens respecvely, he lagged one-monh T-Bll rae, he lagged defaul rsk premum, and he lagged erm premum; Z -1 here s he lagged average emperaure from -12 o -1. The global warmng facor s he rackng porfolo reurn GWF br. excess reurns Trackng porfolo reurns are defned here as he facor mmckng porfolos of br. The OLS regresson gven by equaon (4) can be used o esmae he porfolo weghs b so as o oban br. The rackng porfolo reurns of changes n expecaons abou global warmng, he global warmng facor, s hus GWF br. The uncondonal mean of he rackng porfolo reurns be R ) represens he facor rsk premum (see Lamon, 2001, Vassalou, 2003, and Kapada, 2011); n our applcaon he facor rsk premum of he global warmng facor. The nuon s ha he esmaed coeffcens b represen he base asse loadngs on he emperaure news. The porfolo wh weghs b on he base asses has a mean excess reurn be R ) ha reflecs he rsk due o emperaure news and can be nerpreed as he rsk premum on he global warmng facor. If global warmng s economcally mporan, he rsk premum ( ( 14

16 assocaed wh global warmng should be sgnfcanly dfferen from zero. Furhermore, f clmae-change mpacs are generally adverse (losses due o exreme evens and/or hgher adjusmen coss), we expec o see a negave premum: because he loadngs on he global-warmng facor should be normally negave and o compensae for he addonal rsk he equlbrum reurn should be hgher, whch, wh negave loadngs, can only be acheved by a negave rsk premum. We focus on news concernng nex year s cumulave average emperaures ( CT 12). The mplc assumpon s ha asse reurns are affeced by news regardng global warmng over he upcomng one-year perod. Ths s a reasonable smplfyng assumpon because, even f asse reurns are affeced by global warmng news over a longer horzon, we would expec ha much of hs news perans also o nex year s global warmng. In our prmary specfcaon we largely follow Vassalou (2003) and use he sx Fama-French sze/book-o-marke porfolos as he base asses. 9 To oban he condonng varables he Z 1 n equaon (4) we agan follow Vassalou (2003) and use macro varables whch are known o predc equy reurns. They are he rsk-free rae (RF), he erm premum (TERM), and he defaul premum (DEF). We use he lagged average emperaure over he pas one year as he sngle clmae conrol varable n Z ' 1. 9 Alhough Vassalou (2003) also ncludes he erm premum and he defaul premum as her base asses n R, we do no do so for wo reasons. Frs, wh our monhly daa we fnd srong frs-order auocorrelaon n hese wo me seres. As a resul, ncludng hem as base asses nroduces srong mulcollneary no Eq. (4). Noe ha he one-monh lagged erm and defaul premum are used as he condonng varables n Z. The resulng rackng porfolo s domnaed by me-seres varaon n he defaul premum, and has lle value for rackng he varaon n emperaure. The resuls are avalable upon reques. Second, n heory, we wan o exrac he news abou fuure global warmng ha he sock marke perceves and hen esmae he rsk premum on socks. Therefore, he bond porfolo reurns are no necessary as base asses. They are sll useful as condonng varables because he leraure shows ha hey can predc equy reurns (see Fama and French, 1988). 15

17 4.3. Trackng Porfolo Properes The prmary specfcaon n Table 1 presens he consrucon and dagnosc ess of he global warmng facor mmckng rackng porfolos based on equaons (4) wh he sx Fama-French sze-bm porfolos as he base asses and he macro varables used also by Vassalou (2003) as condonng varables. The -raos are based on Newey-Wes HAC sandard errors wh he lag parameer se equal o 12. The resuls are for he IPCC global warmng perod 1976:1-2010:12 (noe ha we lose he 2011 perod as we need for he fnal sample pon of CT 12). The coeffcen esmaes n Panel A ndcae ha SL and SH have sgnfcan rackng ably for fuure global warmng. Ths seems o be reasonable because small companes are generally more sensve o global warmng or any oher rsk. Growh frms may poenally benef from global warmng as hey do no have her capal n place ye (SL coeffcen > 0). Value frms face adjusmen coss, as argued by Quggn and Horowz (2003), snce her capal s already n place (SH coeffcen < 0). In he nex panel, he ch-square es rejecs he hypohess ha he coeffcens on he base asses are jonly zero a he 5% level, ndcang ha he base asses have sgnfcan rackng ably. Panel B provdes he mean of he rackng porfolo reurn, whch s percen wh -sasc of for he 1976: :12 perod. Ths provdes drec suppor for Hypohess 1, ha he rsk premum on global warmng s sgnfcanly negave. Noe ha he resul s no dependen on he choce of a parcular asse prcng model. Fgure 1 16

18 shows raw average emperaures and he derved global warmng facor seres (he rackng porfolo reurns) from 1976 hrough If global warmng does no occur n he perod before 1976, emperaures for hs perod ough o be rrelevan o asse prcng mean asse reurns are no drven by emperaure news and he rackng porfolo wll have lle power o predc fuure emperaure. To es hs conjecure, we repea he above exercse for he IPCC prewarmng perod of 1946 o The resuls are repored n Table 1 n he columns labeled 1946:1-1975:12. The rackng porfolo for hs perod has some ably o rack fuure emperaure (bu sgnfcanly less han for he global warmng perod). However, he rsk premum for hs perod s an nsgnfcan percen as ancpaed. 4.4 Alernave Specfcaons We consder hree alernave approaches o consrucng he global warmng facor n order o evaluae he robusness of our resuls. Frs, we apply he more recen approach of Kapeda (2011) nsead of Vassalou (2003) for choosng he bass asses and condonng varables used n he rackng porfolo approach. Alernave Specfcaon 1 CT c a rend gz br ez 12 ' (4a) CT +12 represens he average annual emperaure from +1 o +12; R = (MKT, SMB, HHL, LCMLG, LGMRF) represens he hree reurns of he Fama-French facors and he excess reurn of long erm corporae bonds over long-erm governmen bonds as well 10 The resuls are robus o he somewha arbrary choce of condonng varables. The resuls presened are for he condonng varables used n Vassalou (2003). However, he global warmng rsk premum esmae s nearly dencal f we use lagged values of he es asses as condonng varables, or even f we drop he condonng varables alogeher. These resuls are avalable from he auhors. 17

19 as he excess reurn of long erm governmen bonds over he rsk free rae; Z -1 = (RF -1, DEF -1,TERM -1 ), represens respecvely, he lagged one-monh T-Bll rae, he lagged defaul rsk premum, and he lagged erm premum; Z -1 s he lagged average emperaure from -12 o -1. The global warmng facor s he rackng porfolo reurn GWF br. Second, we avod he specfcs of he clmae rend model of Campbell and Debold (2005) by assumng nsead a sochasc rend model of emperaure wh unknown drf erm o denfy he rackng porfolo. Alernave Specfcaon 2 Esmae he followng model over he enre sample: CTGRTH c gz br ez 12 ' , (4b) where CTGRTH +12 = 100 log(ct +1, +12 /CT -11, ), he log dfference n he average annual emperaure. As n he benchmark specfcaon R = (SL, SM, SH, BL, BM, BH ) represens excess reurns: he Fama-French sze-bm porfolos ne of he one-monh T- Bll rae; and Z -1 = (RF -1, DEF -1-,TERM -1 ), represens respecvely, he lagged onemonh T-Bll rae, he lagged defaul rsk premum, and he lagged erm premum; Z -1 s he lagged growh rae n he average emperaure. The global warmng facor s he rackng porfolo reurn GWF br. Thrd, we avod he rackng porfolo approach alogeher n favor of he wopass approach wh specfc rsk correcons based on he CAPM and he Fama-French hree-facor model, usng raw emperaure changes as he global warmng facor. Alernave Specfcaon 3 18

20 Esmae he followng model wh he 10 years of daa precedng a parcular monh, say January 1960: 11 T C A rend S seasonal L T e (4c) 1 1 n where T represens he monhly emperaure, and he lag lengh s seleced by he Schwarz-Bayes Creron. Then, calculae he expeced emperaure n January 1960 based on he esmaed parameers. The global warmng facor s he unexpeced change n he monhly emperaure, GWF e. The specfcs of he rackng porfolos generaed by he frs and second robusness approaches (ALT1 and ALT2) are n Table 1 (he hrd approach does no use a rackng porfolo). Panel B n hs able llusraes ha he mean rackng porfolo reurn connues o be sgnfcanly negave even for he alernave se of bass asses and he dfferen rend specfcaon for emperaures. 5. Sandard Asse Prcng Models: Srucural Approaches The rackng porfolo approach of Lamon (2001) n Secon 4 esmaes he rsk premum of he global warmng facor whou specfyng an equlbrum asse prcng model. In hs secon, we supplemen hese resuls by esmang he rsk premum of global warmng whn a mul-facor model. Ths approach enables us o denfy he aggregae quanave mpac on he cos of capal and how he mpac changes over me, o address Hypohess 2, and o oban he sensves o he global warmng facor of parcular ndusres, o address Hypohess The Emprcal Model 19

21 For our asse prcng specfcaons we ake he Fama-French (1996) hree-facor model and he Sharpe-Lnner CAPM. These are he mos common sysemac rsk models. Boh can be vewed as specal cases of he Meron (1973) model. The global warmng facor s added o hese specfcaons as an addonal facor (he only such facor n he CAPM case) affecng nvesmen opporunes over me. If clmae-change mpacs are adverse we expec o fnd negave loadngs for ypcal frms or porfolos and a negave rsk premum (o compensae for he addonal rsk he equlbrum reurn should be hgher; wh negave loadngs hs can only be acheved wh a negave rsk premum). The mpac of global warmng on he average cos of capal can naurally be measured by he produc of he premum and he average loadng of asses on he global-warmng facor. We use he Black, Jensen, and Scholes (1972) and Fama and MacBeh (1973) wo-pass mehodology esmang facor sensves n he frs pass, and usng hese o oban rsk prema n he second pass wh sandard refnemens: he Shanken (1992) correcon o oban errors-n-varables-robus sandard errors, accounng for he fac ha facor sensves are esmaed, and he Shanken and Zhou (2007) correcon o generae msspecfcaon-robus sandard errors. 11 Lewellen, Nagel and Shanken (2009) argue ha usng as es asses only sze BM porfolos, as s common n he leraure, can be hghly msleadng due o he srong facor srucure of hese porfolos. They propose o expand he se of es asses o nclude oher porfolos, such as ndusry porfolos. Includng ndusry porfolos also s parcularly aracve for our research snce he global warmng leraure has predcons for how dfferen ndusres should be affeced. Lewellen, Nagel and Shanken (2009) 11 See also Km (1995) and Jagannahan and Wang (1998). 20

22 furher argue, snce he problems are exacerbaed by he fac ha emprcal ess ofen gnore resrcons on he cross-seconal slopes, ha researchers ake he predced magnudes of he slopes serously. For example, he rsk premum on a facor porfolo should be close o s average excess reurn. We ake her advce and use 55 sze-bm and ndusry porfolos (25 sze-bm porfolos and 30 ndusry porfolos) nsead of jus 25 sze-bm porfolos Furher Emprcal Resuls Snce global warmng represens a new sysemac rsk, adversely affecng nvesmen opporunes n he economy, we expec o see ha he global-warmng facor s prced and ha s premum s negave n he 1976 o 2010 perod. The resuls are repored n Table 2. The frs-pass resuls for each of he four models: CAPM, Fama- French model, he CAPM plus he global warmng facor, and he Fama-French model plus he global warmng facor are smlar. In each case he GRS es rejecs he model, alhough leas clearly for he CAPM plus global warmng. The second-pass resuls reveal ha he rsk premum for he global warmng facor s sgnfcanly negave as predced. The model has a hgher average cross-seconal R-square han he Fama-French model n he Fama-MacBeh regressons. The premum esmae of s n he ballpark of s mean value repored n Table 1 whch s We confrm Hypohess 1. Table 3 shows ha he alernave specfcaons perform smlarly as he prmary specfcaon: he global warmng rsk premum s sgnfcanly negave, alhough only 21

23 margnally so for ALT3. 12 The able also shows he oal value-weghed mpac for each of he CAPM plus global warmng specfcaons (he rsk premum mes he average global warmng loadng) s economcally sgnfcan and wh he same order of magnudes. They are a weghed average annual 0.11 percen ncrease n he cos of capal from global warmng n he Prmary specfcaon, and 0.14 percen for ALT1, 0.22 percen for ALT2, and 0.20 percen for ALT Tme Pah of he Impac To oban more nformaon abou he pah of he global warmng rsk premum over me, we repea he above exercse wh a rollng sample. The rsk premum a each me s esmaed wh 30 years of daa o oban meanngful esmaes. Our earles observaon for he erm premum s Consequenly he es perod sars n We updae esmaes monhly by droppng he earles observaon and addng he laes observaon. We wan o see f Hypohess 2 can be confrmed by examnng wheher he mpac of global warmng has ncreased from an early par of he es perod o a laer par. Table 4 shows he average mpac over wo evenly dvded sub-perods: and To properly evaluae he sgnfcance of he dfference n mpac n he wo perods we use he approach of Cooper, Guerrez, and Hameed (2004) whch provdes approprae sandard errors n hs case where we use he rollng 30-year sample. In he prmary specfcaon he average global warmng mpac n perod s a sgnfcanly posve 0.05 percen. The mpac n he perod s also 12 The second-pass resuls n Tables 2 and 3 are based on OLS. The GLS resuls are avalable from he auhors. They are dfferen only n ha (1) he R-squares are lower; (2) he sgnfcance of he rsk prema, for global warmng n parcular, s hgher. 22

24 sgnfcanly posve, bu much larger a 0.24 percen. The dfference s clearly sgnfcanly posve. The resuls for he hree alernave specfcaons are qualavely dencal, also showng sgnfcan mpac n boh perods ha s subsanally larger n he second perod. The sgnfcan dfference for he 2 nervals nomnally confrms Hypohess 2. However, a closer look a he me pah of he mpacs over me llusraes ha he ssues are no as smple as hypoheszed. The specfc me pahs of he mpac are dsplayed n Fgure 2 and are que conssen for he varous specfcaons. The evdence furher confrms ha global warmng maers for asse prcng and ha s mpac s sgnfcanly posve and, unl recenly, growng over me as s conssen wh Hypoheses 1 and 2. However, he mpac esmae appears o decrease afer 2004, whch s no conssen wh Hypohess 2. The possble reasons, oher han a rue abaemen n he rsk of global warmng, are as follows: (1) Samplng error. Whle emperaure shocks ndeed have been mosly negave snce 2004, he perod from 2004 s smply oo shor o reveal relable rends. (2) The uncerany abou he exen and he mporance of global warmng has been decreasng, and/or he ably of frms o handle global warmng hreas has mproved as radng n weaher dervaves has aken off dramacally n he early 21 s cenury. 6. Global Warmng and he Cos of Capal 6.1. Esmaon To esmae he mpac of clmae change on he cos of capal, we repor he facor loadngs n Table 5 for he 1976 o 2010 sample perod. The sgnfcan facor 23

25 loadngs (a he 10% level for wo-sded ess) are n bold. We dscuss he ndusry porfolo facor loadngs o gauge he suppor for Hypohess 3. The leraure (see IPCC, 2007, Workng Group II Repor Impacs, Adapaon and Vulnerably, Chaper 7, and Quggn and Horowz, 2003) suggess ha he mos vulnerable ndusres are Agrculure and Foresry, Transporaon, Real, Commercal Servces, Toursm, Insurance, hose ndusres dependen on clmae-sensve npus (such as Food Processng), and ndusres wh long-lved capal asses. The resuls n Table 5 seem generally conssen wh hs classfcaon, confrmng Hypohess For he sze- and value-sored porfolos s clear ha (1) value frms, whch end o have long-lved capal asses, are far more sensve o he global warmng facor han growh frms, as follows by examnng he dfference beween SB5-SB1 for all sze classes, and (2) small frms are more sensve o he global warmng facor han larger frms, as follows by he dfference S1Bj-S5Bj for all value classes j. In all, 20 ou of 30 ndusry porfolos have sascally sgnfcan negave loadngs on he global-warmng facor. They nclude Meals, Fnance, Transporaon, Auos, Real, and Consrucon. The loadngs on he global-warmng facor are posve for fve of he ndusres only wh hree of hose sgnfcan: he Busness Equpmen, Servces and Healh ndusres. Ol and Telecommuncaons are posve also. The uly ndusry has a small demand elascy, and herefore may provde a good hedge agans negave supply shocks. The elecommuncaons and healh ndusres appear also o have small demand 13 We only presen he loadngs on he global warmng facor for he prmary specfcaon. The loadngs for alernave specfcaons 1 and 2 are very smlar, havng correlaon of 94 percen and 96 percen, respecvely, wh he loadngs for he prmary specfcaon. For alernave specfcaon 3 he loadngs have lower, bu sll posve, correlaon wh hose of he prmary specfcaon of 29 percen. 24

26 elasces. Servces have few long-lved capal asses. Busness Equpmen benefs from adjusmen due o global warmng as hs ncreases demand for her produc. The economy-wde average mpac on he cos of capal s obaned by mulplyng he global warmng rsk premum by he weghed average loadng on hs facor. As was shown n Table 3 hs produces an average mpac of 0.11 percen per year. A convenen means of capalzng he annual mpac o accoun for he presen value coss of global warmng s o use he Gordon growh model approxmaon arsng when we equae asse prces wh he expeced ne presen value of fuure dvdends, and proxy he laer by seng he dynamc pahs of dvdends, dvdend growh, and coss of capal equal o her averages. Aggregae sock marke values can hen be expressed as P D /( R G), n whch P s he curren sock marke prce ndex, D he end-of-perod expeced dvdends (consdered a consan proporon of GDP), G he average ancpaed growh rae of dvdends, and R s he average cos of capal Implcaons We hen draw he followng conclusons based on he exsng leraure and our own resuls: (1) The esmaed cos of global warmng n erms of permanen fuure decreases n GDP, as emphaszed by prevous leraure, vary from around 0.2% o 1.0% (Sern, 2007) and below 1% (IPCC, 2007) o 2.0% o 3.5% per 1 degree Celsus ncrease n emperaure (Chonère and Horowz, 2006) and a maxmum of 3.75% (Heal and Krsröm, 2002). Even hough hese esmaes deal wh permanen changes n GDP he presen value of he cos, assumng ha dvdends reman a consan proporon of GDP 25

27 and even f he changes occur whou delay, s smply he equvalen percen change n sock marke value: he Gordon growh equaon mples ha % P % D. (2) Henry (2003) fnds an approxmae one-o-one relaonshp beween he cos of capal and GDP growh. Ths s conssen wh he Gordon growh perspecve n whch he mpac of a change n he cos of capal s equvalen o he mpac of an oppose change n he growh rae of dvdends or oupu: % P / R % P / G. Accordngly, he sock marke wealh mpac of our 0.11 percen ncrease n he cos of capal s equvalen o ha of a 0.11 percen reducon n he growh rae of dvdends or oupu. Snce he IPCC cos of lmng he long-erm emperaure rse s a 0.12 percen reducon n annual GDP growh, he mpac of global warmng on he cos of capal alone s of smlar magnude as he oal polcy cos, whch allows room o conemplae more aggressve clmae-change mgaon polces. (3) The Gordon growh approxmaon also mples ha % P ( P / D) R. If we se P/D (he prce-dvdend rao) equal o 38, whch s s average value over he global warmng perod, snce 1976 (see Rober Shller s webse) hen he presen value of he cos of global warmng s 38 x 0.11% = 4.18%. Ths negave mpac n solaon s of smlar magnude as he 4.56 percen cos of reducng global warmng based on he IPCC esmae Perspecves To pu our esmae of a 4.18 percen loss due o global warmng n he perspecve of prevous esmaes, consder he followng: 26

28 Frs, he leraure has consdered manly he specfc presen value coss of permanen fuure losses n GDP due o abrup clmae change, and, more recenly, he coss of losses n GDP relaed o exreme weaher due o gradual clmae changes. However, he uncerany abou emperaures, exreme weaher, and assocaed regulaon adds subsanally o he overall coss and shows up n he hgher cos of capal whch prevous sudes have gnored n hs conex. Thus our resul amouns o addng an addonal presen value cos of global warmng of larger magnude o prevously denfed coss. Second, our esmae of a 4.18% loss presumes ha he change n he cos of capal s permanen. Our rollng regresson resuls for he rsk premum show how dangerous hs assumpon s clearly he rsk premum changes subsanally over me and has decreased n recen years. However, we have aken conservavely he average rsk premum snce he sar of he global warmng perod. Whle s ceranly possble ha he cos of capal arbuable o global warmng may connue o decrease over me, may also ncrease over me, and here s no reason o assume ha uncerany abou clmae change s dmnshng n he near fuure. Thrd, an approxmae confdence nerval of wo sandard devaons around our annual mpac esmae would be an annualzed 1.27%, a he upper bound an annualzed 6.93%. Thus, esmaon uncerany s subsanal. Fourh, s no clear how represenave he Uned Saes sock marke values are of he world. Developed economes experence hgher losses due o he adjusmen coss of capal and so he coss for he Uned Saes may overesmae hose for he world. A furher mplc assumpon s ha he sock marke losses are represenave for losses n 27

29 he overall economy. I s possble for nsance ha ceran secors of he economy are over or under represened n he sock marke or ha here are dsrbuon effecs whn a secor. For nsance, farmers may suffer from global warmng bu hs need no lower prof margns n he food ndusry. Ffh, our esmaes do no ake no consderaon me varaon n he sensves o he global warmng facor. Indusres wh larger ncreases n he cos of capal may shrnk over me as nvesmen n hese ndusry falls, causng he valueweghed loadngs o move closer o zero. On he oher hand, such reducon n nvesmen lowers growh, an ndrec effec ha we have no ye ncorporaed n he cos calculaon. Sxh, he oal presen value loss of 4.18% s based on he sandard dscounng paradgm. Oher socal dscounng perspecves, such as zero dscounng or Dez and Ashem s (2012) susanable dscounng, would ypcally generae a larger loss and movae sronger abaemen effors Uncerany abou regulaon One poenal explanaon for he sensvy of parcular ndusres o our global warmng measure s ha he global warmng ndcaor has lle nrnsc economc mporance (eher because s a sascal lluson, has only mnor economc mpac, or s oo remoe o affec curren presen values) bu ha markes fear he polcal pressures arsng from common percepons of a global warmng hrea ha may lead o unoward regulaon hurng busness profs. Ths unoward regulaon lne of reasonng mples ha hose ndusres whch are mos vulnerable o regulaon desgned o reduce global 28

30 warmng would be he mos sensve o he global warmng facor. A prospecve regulaory mpac generally enals raonng or axaon of carbon-doxde emssons. Hence he predcon s ha ndusres ha are he mos sensve o he global warmng facor are hose ndusres ha have he hghes share of carbon doxde emssons. Schpper (2006, pp ) provdes daa on carbon-doxde emssons n U.S. manufacurng. Manufacurng accouns for around 84 percen of energy-relaed carbon doxde emssons. The Peroleum (Ol), Chemcals, and Prmary Meals (mosly Seel) ndusres have he hghes carbon doxde emssons, ogeher generang more han half of hese emssons. However, as s apparen from Table 5 hese ndusres have les han average negave sensves o he global warmng facor. Furhermore, many of he ndusres n Table 5 wh srong negave exposures o he global warmng facor are non-manufacurng ndusres such as Fnance, Games, Real, and Meals, wh obvously low carbon-doxde emssons. These observaons do no suppor he unoward regulaon explanaon. 7. Concluson The severy of mpac and even he exsence of global warmng are heavly debaed. On one end of he debae are he envronmenalss who care deeply abou he negave long-run mpac of clmae change and may be emped o under apprecae he economc sacrfces requred o comba global warmng; on he oher end are he ndusralss who may be owners of capal and overly occuped wh he coss necessary o mplemen polces of fghng global warmng. The conroversy makes dffcul o oban an unbased measure of even he perceved coss of global warmng. 29

31 Fnancal markes may be helpful n uncoverng rue percepons. When nvesng n fnancal asses, ndvduals rrespecve of her background or polcal convcons have an ncenve no o pu her money where her mouh s bu o pu her money where her mnd s. Asse prces provde objecve measures of perceved value. We aemp o explo he nformaon embedded n asse prce reacons o news abou global warmng o nfer an objecve measure of perceved coss of global warmng. We focus on gradual emperaure change, whch s movaed by he recen emphass of Sern (2007) and IPCC (2007) on he ongong coss of exreme weaher evens ha sem from gradual warmng. 14 Our approach es no he gradual warmng perspecve and sresses he hgher cos of capal ha arses from uncerany abou he exen of gradual warmng and he ncreasng prevalence of exreme weaher evens. The coss of global warmng ha we derve, herefore, are a complemen o he coss found n he earler work and should be added o prevous cos esmaes o oban a more comprehensve cos oal. Whle we may draw nferences abou he broad queson of wha fnancal markes n general can ell us abou global warmng, we oban a more relable answer o a narrower queson: how do US equy markes reac o news abou average US emperaures? If we work n he conex of () he Meron (1973) asse prcng model and () presume ha he IPCC (2007) s jusfed n markng he pos-1976 perod as one of global warmng, hen we hypohesze ha a sgnfcan rsk premum exss on a global warmng facor, whch s rsng n he pos-1976 perod, and ha loadngs a he ndusry 14 In analogy o he ermnology used n evoluonary bology dsngushng beween he sandard gradual Darwnan perspecve and Gould s puncuaed equlbrum heory of abrup evoluonary changes, we may refer o he dsncon here as global warmng by creeps and global warmng by jerks. 30

32 level on hs facor are generally negave and more so for ndusres ha are consdered o be more sensve o global warmng. On he whole we are able o confrm he hypoheses and nfer, akng he average rsk premum for he pos-1976 perod, ha he average cos of capal s 0.11 percenage pons hgher on an annual bass due o global warmng. Thus, markes expec ha due o he uncerany, he coss of adjusmen, and he ncreased ncdence of exreme-weaher evens, each caused by global warmng, poenal projecs wll on average have a 0.11 percenage pon lower reurn. The mpled coss of global warmng amoun o a pon esmae of a 4.18 percen loss n value larger han, and addng o, mos prevous esmaes. Acknowledgemen The auhors wsh o hank semnar parcpans a he Unversy of Canerbury, Monash Unversy, he Fnancal Managemen Assocaon meengs, and Glenn Boyle, Erc Crampon, Mke Dempsey, Rober Faff, John Founan, Phlp Gharghor, Seamus Hogan, Peko Kalev, and Ed Maberly for helpful commens. 31

33 References Balvers, R. J., and D. Huang, 2007, Producvy-Based Asse Prcng: Theory and Evdence, Journal of Fnancal Economcs 86, 2007, Barreca, A. I., 2012, Clmae Change, Humdy, and Moraly n he Uned Saes, Journal of Envronmenal Economcs and Managemen 63, Black, F., M. C. Jensen, and M. Scholes, 1972, The Capal Asse Prcng Model: Some Emprcal Fndngs; n Mchael C. Jensen, ed.: (Praeger, New York). Sudes n he Theory of Capal Markes Breeden, D. T., M. R. Gbbons, and R. H. Lzenberger, 1989, Emprcal Tes of he Consumpon-Orened CAPM, Journal of Fnance 44, Brunner, A. D., 2002, El Nño and World Prmary Commody Prces: Warm Waer or Ho Ar? Revew of Economcs and Sascs 84, Campbell, Sean D; Debold, Francs X, 2005, Weaher Forecasng for Weaher Dervaves, Journal of he Amercan Sascal Assocaon 100, 6-16 Chonère, C. J. and J. K. Horowz, 2006, A Producon Funcon Approach o he GDP- Temperaure Relaonshp, workng paper, Deparmen of Agrculural and Resource Economcs, Unversy of Maryland, Maryland. Clne, W.R., The Economcs of Global Warmng. Insue for Inernaonal Economcs, Washngon, DC. Cooper, M. J., R. C. Guerrez, and A. Hameed, 2004, Marke Saes and Momenum, Journal of Fnance, 59, Dez, S. and G. B. Ashem, 2012, Clmae Polcy under Susanable Dscouned Ularansm, Journal of Envronmenal Economcs and Managemen 63, Fama, E. F., and J. D. MacBeh, 1973, Rsk, Reurn and Equlbrum: Emprcal Tess, Journal of Polcal Economy 81, Fama, E. F., and K. French, 1988, Dvdend Yelds and Expeced Sock Reurns, Journal of Fnancal Economcs, 22, Fama, E. F., and K. French, 1996, Mulfacor Explanaons of Asse Prcng Anomales, Journal of Fnance 51, Fankhauser, S., Valung Clmae Change. The Economcs of he Greenhouse. Earhscan, London. Fankhauser, S. and R. S. J. Tol, 2005, On Clmae Change and Economc Growh, Resource and Energy Economcs, 27, Flemng, J., C. Krby, and B. Osdek, 2006, Informaon, Tradng, and Volaly: Evdence from Weaher-Sensve Markes, Journal of Fnance, 61,

34 Harvey, A.C., 1989, Forecasng, Srucural Tme Seres Models and he Kalman Fler. Cambrdge: Cambrdge Unversy Press. Heal, G., and B. Krsröm, 2002, Uncerany and Clmae Change, Envronmenal and Resource Economcs 22, Henry, P. B., 2003, Capal-Accoun Lberalzaon, he Cos of Capal, and Economc Growh, Amercan Economc Revew 93, Hrshlefer, D., and T. Shumway, 2003, Good Day Sunshne: Sock Reurns and he Weaher, Journal of Fnance 58, Horowz, J. 2001, The Income Temperaure Relaonshp n a Cross-Secon of Counres and Is Implcaons for Global Warmng, workng paper, Unversy of Maryland, Maryland. Inergovernmenal Panel on Clmae Change, 2007, The IPCC Fourh Assessmen Repor, hp:// Inergovernmenal Panel on Clmae Change, 2012, Specal Repor on Managng he Rsks of Exreme Evens and Dsasers o Advance Clmae Change Adapaon, hp:// Jagannahan, R., and Z. Wang, 1998, An Asympoc Theory for Esmang Bea-Prcng Models Usng Cross-Seconal Regresson, Journal of Fnance 53, Kamsra, M. J., L. A. Kramer, and M. D. Lev, 2003, Wner Blues: A SAD Sock Marke Cycle, Amercan Economc Revew 93, Kapada, N., 2011, Trackng Down Dsress Rsk, Journal of Fnancal Economcs 102, Km, D., 1995, The Errors n he Varables Problem n he Cross-Secon of Expeced Sock Reurns, Journal of Fnance 50, Lamon, O. A., 2001, Economc Trackng Porfolos, Journal of Economercs 105, Leau, M., and S. Ludvgson, 2001, Consumpon, Aggregae Wealh, and Expeced Sock Reurns, Journal of Fnance 56, Lewellen, J. W., S. Nagel, and J. A. Shanken, 2009, A Skepcal Apprasal of Asse Prcng Tess, Journal of Fnancal Economcs, forhcomng. Mendelsohn, R.O., W.N. Morrson, M.E. Schlesnger, and N.G. Andronova, 2000, Counry- Specfc Marke Impacs of Clmae Change, Clmac Change 45, Mendelsohn, R.O., Neumann, J.E. (Eds.), The Impac of Clmae Change on he Uned Saes Economy. Cambrdge Unversy Press, Cambrdge. Mendelsohn, R. O., W. D Nordhaus, and D. Shaw, 1994, The Impac of Global Warmng on Agrculure: A Rcardan Analyss, Amercan Economc Revew 84,