.S.G. Risk Factos in a Potfolio Context Integated Modeling of nvionmental, Social and Govenance Risk Factos An Innovative Study fo Institutional Investos Novembe/Decembe, 2009 D. Steffen Höte isklab GmbH Diecto Phone +49 89 1220-7704 e-mail: steffen.hoete@isklab.com D. Wolfgang Made isklab GmbH Senio Vice Pesident Phone +49 89 1220-7759 e-mail: wolfgang.made@isklab.com Babaa Menzinge isklab GmbH Associate Phone +49 89 1220-7754 e-mail: babaa.menzinge@isklab.com
Agenda isklab Study.S.G. Risk Factos in a Potfolio Context 1. Motivation of Study 2. Modeling of.s.g. Risk Factos 3. Potfolio Optimization and.s.g. Risk Factos 4. Key Conclusions fo Investos Back-up 2
3
4
Motivation of Study 5
.S.G. Risks: the Unknown in the Investo s Potfolio Point of depatue Many institutional investos have explicitly adopted the pomotion of envionmental, social and good copoate govenance compliant investing into thei investment policy*; example: ABP views it as its obligation to achieve the highest possible etun fo clients. In doing so, it believes that companies with stategies which, in addition to financial etun, place a high value on the envionment, social factos and good copoate govenance will pefom bette in the long tem. ( ) Fo this eason, we have chosen to implement a stong.s.g. policy. Stichting Pensioenfonds ABP is the pension fund fo employes and employees in sevice of the Dutch govenment and the educational secto. Challenge Investos ae uncetain about the isk/ etun effects of.s.g. investing**. peceived as possibly beneficial in the long tem; not shot tem may delive highe etun may povide moe stable etuns in combination with a lowe isk pofile (less volatility). Conclusion While investos in theoy would suppot sustainable, esponsible investing thee is no common view to assess the impact in a potfolio and asset allocation context. * http://www.climatechangecop.com ** Souce: IP.com 18 Septembe 2009 6
Missing Link Between.S.G. Investing and Stategic Asset Allocation Focus of.s.g. Investing Reseach So fa eseach has mainly focused on.s.g. compliant equity investments fom a bottom-up investment pocess pespective. The evidence on the pefomance of SRI Funds is mixed. Usually, thee is no bottom up link of.s.g. investment eseach and potfolio level.s.g. Factos not Fully Recognized on Potfolio Level Othe top down SAA eseach has been often athe qualitative and focused on one element within the SG aconym usually the envionmental as it elates to climate change Thee exists no systematic, long-tem quantitative analysis explicitly examining.s.g. isk factos and thei potfolio impact. Impotance of Stategic Asset Allocation isklab views Stategic Asset Allocation (SAA) as the most impotant facto diving long-tem potfolio etuns. stimates conclude it accounts fo up to 90% of potfolio isks, outweighing maket timing and stock selection in impotance. 7
Modeling of.s.g. Risk Factos 8
Conestones of isklab.s.g. Study Objective Integated modeling of envionmental, social and govenance isk factos in a potfolio context Focus is the analysis of long-tem isks on a 20 yeas hoizon Key assumption:.s.g. isks do not impact expected etuns.s.g. Risk Facto Modeling Pocess.S.G. isk facto analysis and selection.s.g. isk facto modeling: definition + calibation of stochastic pocesses conomic Scenaio Geneation incl..s.g. isk facto simulation Input of.s.g. quity isk sensitivities Computation of pices fo all assets [Govies, Cash, +.S.G./Global/-.S.G.quity] Input Potfolio Analysis Futue Pojections 10,000 Paths.S.G. Risk Potfolio Analysis Robust potfolio optimization (key citeion CVaR 95%) Potfolio simulation (efficient fonties: selection of 3 altenative potfolios) Conclusions fo Investos: SAA w..t..s.g. isks 9
.S.G. Risk Facto Sceening and Shot Listing nvionmental Risk Social Risk Govenance Risk Multiple Risk Factos Global Waming mission Waste + Pollution Labo Rights Human Rights Child Labo Bibey + Couption Unequal Conflict of shae voting Inteest Resouce Depletion Safety + Health Wong Incentives Selection Risk Dive Cabon mission Rights Spot Pice Change Relative secto cabon footpint Sick Rates Relative secto staff costs / sales Copoate Govenance Relative secto govenance atings Shot-listing.S.G. isk factos: causality, fit to modeling, data availability, SRI expet input 10
Modeling the.s.g. Influence on quity Retuns quity Retuns Q.S.G. facto modeling nvionmental: Social: Copoate Govenance: ~ ~ S ~ G stochastic pocess stochastic pocess stochastic pocess 0 Sensitivity deivation nvionmental: Social: Copoate Govenance: Secto, Secto,Global Secto,,, Secto, Secto,Global Secto,, S S S Secto, Secto,Global Secto, G, G,, G w Secto weighting S G,,, Global Global S Global G,,, S G Retun adjustment ~ Global ~ ~ ~ Global ~ ~ ~ S ~ Global ~ S G ~ ~ ~ S G Global G ~ Global ~ S ~ S ~ S ~ S G Global S G ~ ~ ~ G S G Global G ~ G Capital maket scenaios Finally, the.s.g. etun diffeence is added to the equity etun befoe inclusion of.s.g. obtained fom the conomic Scenaio Geneato. ~ ~ ~ ~ ~ ~ ~ Q Q ~ Global Global Q Q ~ Q Q 11
Modeling the.s.g. Influence on quity Retuns xample: nvionmental Risk.S.G. facto modeling The envionmental facto is modeled as a stochastic pocess. The CO2 mission ights spot pice epesents ou envionmental facto. ~ [~ ] 0 stochastic pocess Sensitivity deivation.s.g. facto modeling: Simulation Results CO-2 mission Right Spot Pice Change We model thee equity assets: quity of companies that ae in line with.s.g. citeia (+), of those that ae not (-) and of those that have an aveage exposue to.s.g. isk. A sensitivity to the envionmental facto is deived fo each secto. The secto sensitivities ae weighted accoding to the secto epesentation in the MSCI Wold. f Secto, Secto,Global Secto, whee, Secto,Global Secto Secto, w, Sectos Secto Secto, w Sectos, Global Cabon w Sectos Footpint Secto Secto, Global Sensitivity: Analysis of Cabon Footpint Data MSCI AC Aveage of Cabon Footpints MSCI AC Nomalized Cabon Footpints () +.S.G. quity () -.S.G. quity () MSCI AC (Weights) Financials 2.471 0.067 0.053 0.080 21.92% Consume Discetionay 1.826 0.108 0.087 0.130 8.82% IT 1.314 0.141 0.113 0.170 11.75% Health Cae 1.150 0.152 0.122 0.183 9.34% Telecom Sevices 0.975 0.163 0.131 0.196 5.10% Industials 0.643 0.185 0.148 0.222 9.99% Consume Staples 0.488 0.195 0.156 0.234 9.49% negy 1.048 0.294 0.235 0.353 11.19% Mateials 5.472 0.581 0.464 0.697 7.74% Utilities 11.954 1.000 0.800 1.200 4.66% Weighted Aveage 0.027 0.225 0.180 0.270 12
xpet Modeling.S.G. Risk nvionmental Risk Social Risk Govenance Risk Data Availability Compaatively good + Factal Somewhat bette o Risk Facto Risk Dive Cabon mission Rights Spot Pice Change Sick Rates Copoate Govenance Stochastic Model Regime Switching Geometical Bownian Motion Regime Switching Risk Sensitivity (quity) Relative Secto Footpint Cabon emission footpint Staff Costs / Sales Govenance Ratings Data Souce Relative Cabon Footpint in MSCI All Counties Wold based on monthly atings (2005-2009) fom Tucost Computations of staff costs / sales on the basis of Woldscope fo staff costs and Datasteam acoss GICS Relative Copoate Govenance Ratings in MSCI All Counties Wold fo diffeent sectos on monthly atings (2005-2009) fom RiskMetics. Relative Weighting qual weighting between.s.g. isk factos 13
Integation of.s.g. Geneating Futue Maket Scenaios - conomic Scenaio Geneato Cascade 1 (conomic Factos) Goss Domestic Poduct (GDP) Inflation Rate o Consume Pice Index (CPI) Cascade 2 (Yield Cuve) Teasuy Yield Cuve Cascade Model Cascade 3 (quity) Cedit Speads quity Retuns Inclusion of.s.g. isk influence on equity etuns Assets +.S.G. quity Global quity -.S.G. quity Govenment Bond Cash 14
Potfolio Optimization and.s.g. Risk Factos 15
.S.G. Risk Factos: What Does it Mean fo Investos? Possible impact of.s.g. isks in the equity and potfolio context 1 Additional.S.G. quity investment isk how much? 2 Solution space altenative potfolios? - efficient fonties (+.S.G./ Global / -.S.G. quity) - example potfolios 3 Optimal stategic asset allocation? - isk eduction - etun enhancement 16
1 Risk / Retun Chaacteistics of quity Retuns Hoizon 20 yeas Retun / Risk Metic +.S.G. Global -.S.G. (aveage values p.a. ove 20 yeas) quity quity quity xpected Retun 7.6% 7.6% 7.6% CVaR 95% -26.7% -38.8% -52.3% Volatility 15.5% 19.3% 23.7% Key findings In compaison the CVaR isk of +.S.G./Global/-.S.G. quity is vey diffeent. The CVaR isk of +.S.G. quity is appox. one-thid less than Global quity*. The CVaR isk of.s.g. is appoximately double that of +.S.G. quity..s.g. isk is assumed to have no impact on expected equity etuns but is a isk dive. CVaR (95%): Conditional Value at Risk (CVaR) 95%: Aveage expected etun incued in the 5% wost case scenaios p.a. * Global quity epesents an equity allocation with an aveage.s.g. exposue 17
2 Result of Optimization: CVaR Applied as Key Citeion Hoizon 20 yeas Stating point xpected Retun + SG quity Global quity - SG quity Blue line shows efficient fontie with Govenment Bonds, Cash and Global quity. Oange line same except full allocation of equity into -.S.G. quity. Geen line same except full allocation of equity into +.S.G. quity. Optimization oppotunities nhance etun fo given level of CVaR. CVaR 95% Reduced CVaR fo given level of etun. 18
2 Fo the Analysis We Selected Thee Altenative Potfolios Hoizon 20 yeas We picked 3 potfolios xpected Retun + SG quity Global quity Potfolio Balanced : on Global quity efficient fontie (Blue) Potfolio Lowe Risk : on +.S.G. efficient fontie (Geen) Potfolio Highe Retun : on +.S.G. efficient fontie (Geen) Reasons fo selection CVaR 95% Balanced : Stating point is a compaatively consevative potfolio (equity shae 30%) Lowe Risk : equal etun expectation to Balanced but lowe isk ( Highe Retun vice vesa) 19
3 Significant Optimization Oppotunities Though +.S.G. quity Allocation Potfolio Balanced Potfolio Highe Retun Cash, 6% Cash, 6% Global quity, 30% B quities +.S.G., 40% quities +.S.G., 30% A Potfolio Lowe Risk Cash, 8% Govenment Bonds, 64% Govenment Bonds, 62% Stating Point: Potfolio Balanced Compaatively consevative potfolio with Global quity allocation of 30% Option A: Potfolio Lowe Risk Govenment Bonds, 55% Risk can be educed at same levels of etun with the same quity (+.S.G.) allocation. Option B: Potfolio Highe Retun Retun expectation can be inceased at same level of isk. 20
3 Risk / Retun Chaacteistics of Selected Potfolios Hoizon 20 yeas Retun / Risk Metic Potfolio Potfolio Potfolio "Balanced" "Lowe Risk" "Highe Retun" xpected Retun 5.5% 5.5% 5.8% CVaR 95% -7.4% -5.1% -7.4% Volatility 6.2% 5.2% 6.5% Potfolio Lowe Risk (Option A): All isks can be educed at the same level of etun compaed to potfolio Balanced. Potfolio Highe Retun (Option B): xpected etun can be inceased at simila level of isks. 21
3 +.S.G. quity ven Attactive with Lowe xpected Retun Key findings All equity asset classes (+.S.G., -.S.G., and Global quity) povide the same expected etun by assumption. xpected Retun Possible etun eduction of +.S.G. quity of 0.7% Compaed to the Balanced potfolio the Highe Retun potfolio has a highe expected etun due to the highe equity allocation (at equal CVaR 95% levels of -7.4%). CVaR 95% Theefoe, a decease in expected +.S.G. quity etun of up to 0.7% would still lead to a highe potfolio etun expectation at simila levels of isk. 22
Key Conclusions fo Investos 23
.S.G. Risk Factos: Key Conclusions fo Investos In the long-tem, ove 20 yeas,.s.g. factos ae expected to have significant isk impact on quity investments. Theefoe, investos should stive to optimize thei Global quity investments and minimize exposue to.s.g. isk. This can be achieved by choosing quity investments, whee copoate management poactively mitigates these isk factos. On the basis of a compaatively consevative potfolio with a global equity allocation of appox. one thid, optimized quity allocation offes: - ithe a potfolio isk eduction (CVaR 95%) of ca. 30% at same levels of expected etun. - O an incease of expected potfolio etun by 0.3%-pts. at simila levels of expected potfolio isk. The effects illustated amplify even moe when compaing moe isky potfolios e.g. when the equity allocation is even highe. 24
BACK UP 25
Back Up: Motivation of Study 26
Geneal Reseach Souces Used fo Scoping the Study (1/3).S.G. factos and sustainable investing (oveall) The SRI Navigato Objectively assessing nvionmental, Social, and Govenance Risks by Valeie Luclas-Leclin et al fo Societe Geneal, May 2009 the isk indicatos of this suvey seved as a good oientation fo ou study and helped to calibate the isk facto weightings. Socially Responsible Investments by Sven Hoss, Chistofe Vogt and Rudi Zagst in Wold Scientific Review, 2009 this aticle gives a geat oveview ove on SRI in geneal, maket development and the question how sustainable is SRI. A case study based on simulated etuns of an auto-egessive Makov-Switching model with undelying data fom 1992 to 2008 shows that isk-avese investos mix SRI investments in thei potfolio in ode to divesify but it also claims that the less iskavese an investo is, the moe he invests in SRI. In Pusuit of a Sustainable Wod Socially Responsible Investing and co Investments by Daius Abde-Yazani et all Bachelo Thesis by six students that summaizes vey well the ecent developments in SRI investing, intoduces a Sustainability Scoecad to help companies implement.s.g. standads, and builds the hypothesis that.s.g. can indeed lead to competitive advantage. Othe than the afoementioned Socially Responsible Investments epot by Hos, Vogt, and Zagst, it finds that isk-avese investos mix SRI/co indices to thei existing bondsstockspotfolio in ode to gain an optimal potfolio in tems of isketun measues The Mateiality of Social, nvionmental and Copoate Govenance Issues to quity Picing by UNP Finance Initiative - 11 Secto Studies Demystifying esponsible investment pefomance A Review of key academic and boke eseach on.s.g. factos by UNP Finance and Mece, Octobe 2007 Fealess Foecast by Mece 2006 - Suveys about the peceived impotance of.s.g. issues among financial pofessionals Climate Change Risk Looking ahead: Implications fo Stategic Asset Allocation lectue by Antoine de Salins fo FRR at the UN Pinciples of Responsible Investing PRI in Peson Confeence on July 3d 2009 in Sydney descibes a two level appoach to assess financial isks unde diffeent financial scenaios. Innovest Integated Oil- and Gas Secto Repot by Chistian Maede fo Innovest, 2006 - This secto epot coves a wide ange of 'non-taditional' isk factos and value dives fo the integated oil & gas secto. Aeas such as stategic govenance, envionment, stakeholdes and human capital ae coveed. A global selection of 28 companies is anked accoding to social, envionmental and combined atings, as well as on a numbe of sub-factos. The epot is notable fo its compehensive discussion of isk factos and a boad coveage of companies, including leading companies fom emeging makets. But it stops shot of assessing potential financial impacts of the descibed isk factos and poposing integated company valuation appoaches. Phamaceuticals: Integation.S.G. (Goldman Sachs Sustainability) by Saah Foest fo Goldman Sachs, 2007 - the secto-adapted.s.g. famewok is used as a poxy fo oveall management quality, and as an indicato fo cash etuns and theefoe fai value. The epot weaves the.s.g. stoy with othe, 'othodox' stategic dives, and is quite tanspaent in its.s.g. methodology. Geen Winnes The pefomance of sustainability focused companies duing the financial cisis by AT Keaney Confeence epot: New Fonties in meging Makets Investments by Who Caes Wins, 2007 an initiative to integate.s.g. issues into mainsteam investment decision-making. Povides good insight in.s.g.-awaeness in emeging makets. Clean Investo 2009 - Investing in sustainable themed funds: the new geneation of etuns? by Responsible Investo, 2009 27
Geneal Reseach Souces Used fo Scoping the Study (2/3) Climate Change and mitigation costs with a view on global economy The Sten eview on the conomics of Climate Change by Nicholas Sten (Baon Sten of Bentfod) (and updates): A 700-page epot fo the Bitish govenment, which discusses the effect of climate change and global waming on the wold economy. Its main conclusions ae that one pecent of GDP pe annum is equied to be invested in ode to avoid the wost effects of climate change, and that failue to do so could isk global GDP being up to twenty pecent lowe than it othewise might be. It povides pesciptions including envionmental taxes to minimize the economic and social disuptions. In June 2008 Sten inceased the estimate to 2% of GDP to account fo faste than expected climate change. The Global Deal by Nicholas Sten (Baon Sten of Bentfod), 2009 newest update on political and economic plans to mitigate climate change and fight global waming and povety. Pathways to a low-cabon economy V2 Global Geenhouse Gas Abatement Cost Cuve, McKinsey & Company, 2009 The conomics of Climate Change by the Select Committee on conomics of the UK House of Lods, 2006 A question of Balance by W. Nodhaus about the mitigation costs of global waming Climate Change: The costs of inaction and the cots of adaption by the uopean nvionment Agency, 2007 A Climate fo Recovey by HSBC, Febuay 2009 - eviews 20 economic ecovey plans published by then to combat the cedit cisis: 15% of the assets (o $432bn)of a total $2.8tn in fiscal measues could be associated with investments consistent with stabilizing and subsequently cutting global emissions of geenhouse gases. Povides insight how the cisis effects the combat against climate change. 28
Geneal Reseach Souces Used fo Scoping the Study (3/3) ffects of Climate Change on diffeent sectos / egions Utilities 2020 Vision: favo low cabon geneatos, cautions on high cabon intensity by Gaeme Moyse, 2008 fo Goldman Sachs - a longtem (2020), geneally quantitative analysis that tests vaious scenaios. The epot takes a wide-anging look at enegy povision and its implications in uope, including the ole of clean tech, cabon captue & stoage and nuclea enegy. The authos ae tanspaent egading thei assumptions and how they aive at thei conclusions. Adaption and Vulneability to Climate Change: Role of the Finance Industy by UNP Finance Initiative Climate Change Woking Goup, Novembe 2006 - co-authoed by Amin Sandhövel of Allianz Climate Solutions, this is a good summay of potential theats and challenges of climate change to the financial secto. Cabon Cunch: Meeting the Cost by UNP Finance Initiative Climate Change Woking Goup, Decembe 2007 it continues the wok fom the pevious aticle now with Amin Sandhövel as chai of the woking goup, now with moe details and figues about the finance secto. Climate Change and the ASX100: An Assessment of Risks and Oppotunities by Buce Rolph fo Citigoup, 2006 - A compehensive climate impact study, which coves not only the impact of ising cabon pices on ASX100 companies, but also the effects of potential physical impacts. The analysis distinguishes between two scenaios fo cabon pices and two scenaios fo physical impacts. A Climate fo Change by Mece a bief discussion on climate change effect on vaious asset classes Climate Change and quity valuations a biefing fo quity analysts by PicewatehouseCoopes fo the Cabon Tust and the Institutional Investos Goup on Climate Change, 2007 good summay with focus on uope and the US, stesses out that egulation and maket esponse ae still vey uncetain and impacts vay widely between sectos. Up in Smoke Theats fom, and esponses to, the impact of global waming on human development by Andew Simms et al fo The Woking Goup on Climate Change and Development, 2004 vey good epot with inteesting case studies, stong bias on developing counties. Afica up in Smoke by Andew Simms et al fo The Woking Goup on Climate Change and Development, 2005 follow up on the pevious epot, good souce fo climate change-elated issues in Afica 29
Back Up: Modeling of.s.g. Risk Factos 30
xpet Modeling.S.G. Risk (1/2) nvionmental isks Fo envionmental facto data availability and quality is compaatively good isklab expet modeling of missions Rights Spot Pice Change (egime switching popety) On the basis of U missions Right Spot Pice data quity isk sensitivity deived on the basis of Relative Cabon Footpint in MSCI All Counties Wold fo diffeent sectos based on monthly atings (2005-2009) fom Tucost Social isks The challenge is that thee is only factal data available to model social isk factos Divese intepetation of social isk (fatality ates, sick ates, staff tunove ates, ) No time seies available to deive a stochastic pocess fo the etuns/pice changes (like the CO 2 emission ights spot pices) Assumptions have to be made egading the type of the stochastic pocess isklab expet modeling of the social isk facto epesents the geneal expected etun impact of company standads and policy w..t. social aspects on quity (positive o negative) It is modeled with a Geometical Bownian Motion, i.e. a nomally distibuted pocess chaacteized by mean and volatility quity isk sensitivity is deived though computations of staff costs / sales on the basis of Woldscope fo staff costs and Datasteam acoss sectos 31
xpet Modeling.S.G. Risk (2/2) Govenance Risks Thee is somewhat bette data available to model govenance isk factos compaed to social isks Divese intepetation of govenance isk (bibey, insufficient copoate govenance boads, ) No time seies available to deive a stochastic pocess fo the etuns/pice changes (like the CO 2 emission ights spot pices) Assumptions have to be made egading the type of the stochastic pocess isklab expet modeling of the govenance isk facto epesents the geneal etun impact of company policy w..t. govenance aspects on quity (positive o negative) Like envionmental isk it is modeled with Regime Switching popety quity isk sensitivity is deived though Relative Copoate Govenance Ratings in MSCI All Counties Wold acoss diffeent sectos on the basis of monthly govenance atings (2005-2009) fom RiskMetics. SRI xpet Coss Checks The modeling and calibating of.s.g. isk by isklab has been challenged and as a esult patly adapted upon expet input and eview of AllianzGI uope, in paticula the AllianzGI Fench quity team. 32
Modeling nvionmental Risk: Key Data Souces.S.G. facto modeling isklab expet modeling on the basis of U missions Rights Spot Pice Change. The initial idea was to stat with egional CO2 pices fo uope, US and China and to mege them in a single common CO2-pice in the yea 2020. This idea was disegaded, as only few sectos ae pat of an established emissions tading scheme and the pice can be passed on to the end-consume in diffeent ways - so we used the U missions Rights Spot Pice Change as the most impotant input vaiable as sudden pice changes pose bigge isks to the companies than long-tem pice-hikes. Also, companies that invest in sustainable techniques ealy on should be less susceptible to CO2 pice changes. uopean emission data since 2005 fom http://datasevice.eea.euopa.eu/datasevice/metadetails.asp?id=1078 (fo an oveview see the pivotal application "uopean Union missions Tading Scheme Data Viewe ). As the Cabon Footpint data doesn t show the actual amount of tons of CO2 but just the tilts between diffeent sectos and egions, we wee looking fo absolute data. In this pecise fom, they exist only in uope and cove only the sectos with a tading scheme. Sensitivity deivation Relative Cabon Footpint in MSCI All Counties Wold: tilts of the Cabon Footpint fo diffeent sectos against MSCI AC Wold povided by IDS GmbH based on monthly atings (2005-2009) fom Tucost. Accoding to thei own epots, Tucost has geneated envionmental impact pofiles fo ove 464 diffeent business activities. Tucost uses these pofiles, along with financial and segmental analysis, to poduce an estimate of a company's diect impacts. An input-output model is used to quantify the indiect impacts that a company has. Tucost then seaches fo any public disclosues that have been made by the company and incopoates them. Once the quantity pofile has been calculated, an extenal cost is applied to each esouce and emission to geneate the extenal cost pofile. Once the analysis has been completed, a veification sheet is sent to the company fo feedback. Feedback is analysed and elevant additional data is incopoated, with Tucost monitoing any new envionmental disclosues fom the company. All in all, thei database contains envionmental data fo 4,500 companies globally coveing all the majo investable indices including the MSCI AWD. In the MSCI AWD 22% of companies povide full disclosue and those that povide patial disclosue take the total to 48%. Fo data on the emaining 52% of the companies, they elay on thei own model that calculates the likely emissions fo each company in the index. 33
Modeling nvionmental Risk: Additional Souces Analyzed Selection Modeling the pice dynamics of CO2 emission allowances by Benz,. and Tück, S., in negy conomics 31, 4 15, 2009. 34
Modeling Social Risk: Data Souces.S.G. facto modeling isklab expet modeling. Sensitivity deivation AllianzGI uope SRI quity Reseach, Fance: computations of staff costs / sales on the basis of Woldscope fo staff costs and Datasteam: - Woldscope collected the public data fo aound 800 stocks evey yea, mainly uopean (2/3). - Using GICS, the evolution of the atio ove the last ten yeas fo evey stock was computed and then the data was aggegated (equally-weighted) pe secto. The peiod coves 1999 to 2008. - Then a coss-peiod aveage and standad deviation was computed. - To be sue that the data is not too eatic e.g. if some stocks wee "out of contol", only those statistics fo each data type and yea wee consideed with data anging fom -2 to +2 standad deviations (keeping 95% of the data, oughly). 35
Modeling Social Risk: Additional Souces Analyzed Selection Costs of Sick Days to UK Business by Bupa Foundation 2006 (http://www.bupa.co.uk/about/html/p/110806_sickdays.html) and conomic Advises Unit fo UK Teasuy 2004 (http://www.hm-teasuy.gov.uk/d/5(1).pdf) Sick days is anothe potential social isk facto as it is possible to asset the economic costs fa bette than fo othe factos, but sick ates ae consideed to be significantly influenced by the cuent unemployment ate and a county s social secuity policy than by a company s individual employee standads. uopean Social Statistics: Accidents at wok and wok elated health poblems Data 1994-2000 by the uopean Communities 2002. Repoting on Human Rights by the Global Repoting Initiative and the Robets nvionmental Cente (Claemont McKenna College), 2008 this suvey coves many diffeent aea such as "investment in human ights", "child labo" and "non-discimination and secuity pactices", but the dataset coves only 100 companies with a stong bias on uope. It is also vey difficult to asses the costs of ignoing this facto. We disegaded the idea to use the facto "Child Labo": child labo typically occus only in developing counties, Westen companies typically ae only connected to this poblem via sub-contactos. Thee is little data on these sub-contacto elationships and it is difficult to estimate the costs of child labo as they mostly consist of "eputation damage". Safety Spotlight: ASX 100 companies and moe Injuy and Fatalities Data Pesented and Intepeted by laine Pio fo Citigoup, June 2009 this thoough epot on accident epots in Asian companies (2005-2009) led to the idea to use fatalities ates as a social facto descibing labo conditions. The wold-wide database http://labosta.ilo.og/ lists fatal occupational injuies by county and by yea - and also by diffeent sectos. We voted against this isk facto as the data is not complete and it poved too difficult to assess the costs fo evey secto and county and its impact on investment pefomance. 36
Modeling Govenance Risk: Data Souces.S.G. facto modeling isklab expet modeling. Sensitivity deivation Relative Copoate Govenance Ratings in MSCI All Counties Wold: tilts of the Copoate Govenance Quotient (CGQ ) fo diffeent sectos against MSCI AC Wold povided by IDS based on monthly atings (2005-2009) fom RiskMetics. This data poves to be vey eliable and is available in a simila matix as the Cabon Footpint data: elative monthly atings since Januay 2005 fo diffeent sectos (Consume Disc, Consume Staples, negy, Financials, IT, Industials, Mateials, Health Cae, Telecom, and Utilities). RiskMetics employ a bottom up appoach to collect and analyze data fom public disclosue documents, pess eleases and copoate websites and veify it with thei in-house expets. The CGQ coves 7.400 companies woldwide, with undelying data points fo up to 65 individual copoate govenance vaiables in eight aeas of focus: Boad of Diectos; Audit pactices; Chate and bylaw povisions; Anti-takeove povisions; xecutive and diecto compensation; Pogessive pactices; Owneship stuctue; Diecto education. These vaiables ae weighted in the scoing methodology based on thei statistical coelation to a ange of isk and pefomance metics. In some cases, vaiables ae eviewed togethe based on the pemise that copoate govenance is enhanced when specific combinations of these factos ae adopted. The exact weighting method was not evealed to us, but the esulting atings poved to be simila to the othe govenance atings. 37
Modeling Govenance Risk: Additional Souces Analyzed Selection The conomic Costs of Couption: A Suvey and new vidence by Axel Dehe and Thomas Hezfeld, June 2005 links couption level with GDP gowth ate. Aggegate Govenance Indicatos 1996-2008 fom by the Wodbank (www.govindicatos.og) anks 212 counties by voice & accountability, political stability & no violence, govenment effectiveness, egulatoy quality, ule of law, and contol of couption. We liked the quality of those ating but decided against using those data as thee was no secto beakdown available. The KPMG Suvey on intenational copoate esponsibility epoting by KPMG available fo the yeas 2002, 2005 and 2005 - it summaizes how many companies submit epots on copoate govenance, diffeentiated by sectos and counties. Fo a shot oveview, look at table 3.1 and 3.3. as well as 4.3 and 4.4. We decided not to use this as a souce as it coves only companies that adhee to ethical standads and do egulaly publish thei effots. The Bibe Payes Index by Tanspaency Intenational fo 1999, 2002, 2006, and 2008 - unfotunately, the sample and the method of calculation have changed ove time, so it is difficult to compae the 2008 BPI diectly with ealie editions of the index. Same is tue fo the Couption Peceptions Index which goes back until 1995 we used both to veify the othe available atings. 38