© BARRA Predicted Beta DCL Exhibit 1148Cpr © 2000 BARRA, Ine No part of hi pubiaéen maybe reproduced or ranean ery erm by any mea (electro phtecoping or eberie) wehoat te por wlan conto PARRA. BARRA ina regard tear and other BARRA prod mes see mes, gu or ges frend nt fubleadon ae vaderaria or registred aderavs of BARRA, he, BARRA tears clue but are nc Ire te BARRA Aas Systen, Ags eg Rik ogee Glob Rk Manager Asp Opis, Aes Peforance Aya As PAN Aes Devers Tele Aes Autcmagon Asia BARRA [county reame] Equity Heda BARRA European Equiy Moca BARRA Srp Equity Mode BARRA Gal Euty Mode GEM, BARRA Equity ating Model Mart Ipset Mot Apher BARRA Macrosconomc Model BARRA Custom Performance, BARFA Cosmo Sata, Comms, Cosvor Global Ris Manage Gel Se ‘Arar BARRA Wer erkets Hod BARRA A Sten, BARRARboute Exo: Deradves BARRA To Pan Ri Sst Toe, BARRA WoriVew Inver Wera The Extmate Decry TED TEDae, BARRA Geb Exrates, Dre, BARRA RogeCareyEQuert and QSUM.Al afer company prod or seven names ference’ ubleton are ued for dean parpse na may be deere of hr repecve ower ‘AGRA AND SUPPERS OSCLAM ALL VARRANTES EXPRESS OX PALED REGARDING THE PROGUCTS AND TH SERVCES REFERENCE INTIS PURLCATION AND ANT ESULSTO EE OBTANAD FROM TH USE TEED NCLUONG BUT NOT LISTED TO ALLVAREANTES OF MERCHANTABLTY ATRSS FORA PETICLAR URECEEOR {SLANG OHGNAITAND ALL WARRANTIES ALEING ROM COURS OF PERFORMANCE COURSE OF DEALNG AND USAGE OF TRADE ORTH EQUALINT UNO THE LAWS OF ANY RRSCICTION THE PROCLICTS AND SERVICE ARE PROVED "ASI" Aca conlngy and complete of ta ae nee pared, Nether BARRA or af supper warrant tht he products or serves refereed In is pubhaten wl be uniaapad or reo ever er fom unaeraed aden programe rede nts nich produc wou thr now. BARRA prodacts contin a meer of sje tos ut nul be urd onl by soghat= cated irvine proton: Thre sno mac ht he ran nares Hrd by he prot wit perorm is marmar at coninant with ir moral craracorisics or aire the pfeil or wl of erect cr expect wales. Except a expres aged by BARRA, BARRA al be dened to be orig inesentrarageene supervon or avr services. “othe extent ts pabiaton dicts yor 2000 redness be cones Year 2000 Rennes Dire andl notes 3 combat Warren \y oF the bs of my ssbapon eens o her tag oF WarectonBARRA Predicted Beta Historical Beta Beta is a gauge of the expected response of a stock, bond, or portfolio to the ‘overall market. For example, a stock with a beta of 1.5 has an ex; ‘excess retum of 1.5 times the market excess retum. if the market is up 10% ‘over the risk-free rate, then—other things held equal—the portfolio is ‘expected to be up 15%. Beta is one of the most significant means of measuring portfolio risk and shows a strong relationship to expected return. vs. Predicted Beta Historical betais calculated after the fact by running a regression (often over {60 months) on a stock's excess retums against the market's excess retums. There are two important problems with this simple historical approach: * ltdoes not recognize fundamental changes in the company’ operations. For example, when RJR Nabisco spun off its tobacco holdings in 1999, the company's risk characterises changed sgniicanly. Historical bela ‘would recognize this change only slowly, over time. * Itis influenced by events specific to the company that are unlikely to be repeated. For example, the December 1984 Union Carbide accident in Bhopal, India, took place in a bull market, causing the company’s historical beta to be artificially low. Predicted beta, the beta BARRA derives from its risk model, is a forecast of a stock's sensitivity to the market. Itis also known as fundamental beta, because it is derived from fundamental risk factors. In the BARRA model these risk factors include 13 attributes—such as size, yield, and price/eamings ratio—plus industry exposure allocated across a maximum of 6 of 55 industry groups. Because we reestimate these risk factors monthly, the predicted beta reflects changes in the company’s underlying risk structure ina timely manner. BARRA programs use predicted beta rather than historical beta because it is a better forecast of market sensitivity. BARRA Predicted Beta 1BARRA, Computing Predicted Beta Below we show how the predicted beta of a portfolio is computed. The beta of a portfolio p with respect to the market mis defined as the covariance of the portfolio return with the market return divided by the variance of the market: COV(t5 sm 0 eee The covariance between two portfolios is decomposed into two parts: 2), the pat explained by factors, caled common factor covariance, and b) the part unexplained by factors, called specific covariance. The factor covariance between portfolio pand the return on the market m the product of the transposed vector of the factor exposures for the portfo the factor covariance matrix, and the vector ofthe factor exposures forthe mar (2) CF COV(tyslm)=Xp F Xm The specific covariance is: (3) SPCOV(Fyslm) =>. Am? a Now, combining equations (1) and (4) COV(r) = VAR(1) we have the formula for the BARRA predicted beta of a portfolio: COV(ty fm) 6) By oF cov(r,, CF COV(fns )+#SP COV(fystm) )+SP COV(tinsfm) EAC NAC rn SF SE FIX + Mla? fm fo Se xk Xu tS MBso? fa et ms BARRA Predicted Beta -2‘Technical Foundations is the number of factors (88 in U.S. E2) is the number of assets in the market portfolio is the portfolio's exposure to factor j is the covariance between factors kand j Is the market's exposure to factor j is the holding of the portfolio in asset / is the holding of the market in asset / is the specific variance of asset i is the variance of the market BARRA Predicted Beta -3