6 Whilerevenuescanneverbenegative,theycanbezeroand
thereareabout 100 firmsinthesamplewithnorevenuesbut
withsomemarketvalueforequity.Inaddition,thedefinition
of revenues is hazy for financial services firms.
7 Thereisalsoasecondaryeffect.Theretentionratioinstable
growthalsochangestoallowthefirmtocontinuegrowingat
4 percentforever.Asthereturnonequitydrops,theterminal
value of equity will also decrease as a consequence.
8 Usingthenaturallogoftheexpectedgrowthratenarrows
thedifferences across companies on thegrowth dimension
and makestherelationshipbetweenPEGand growthmore
linear.
9 Thiscreatesa phenomenonknownasmulticollinearityin
the regression. To illustrate the problems this will create,
assume (asisreasonable) thathigh growth companiesalso
havehighbetasandlowpayoutratios.Thebetathenbecomes
a proxy not only for risk but also for growth, and the
coefficientintheregressionwillreflectthedominantfactor.
In 2003 and2004,betaswerebetterproxiesforgrowththan
risk,whichexplainsthepositivecoefficientonbetasinthe
regressions from those years.