June 30, 1999

Regressions of Multiples on Fundamentals: Market Wide

I have run the latest versions of these regressions without an intercept term. Intuitively, my argument would be that firms with zero values for all of the fundamentals (independent variables) should trade at zero multiples. It does mean, however, that the R-squared cannot be directly compared with standard regressions run with an intercept. All of the percentages are entered in decimal format. Thus, a firm with an expected growth rate of 30%, a payout ratio of 10% and a beta of 1.25 can be expected to have a PE of:
PE = 291.27 (.30) + 37.74 (.10) - 21.62 (1.25) = 64.13

Equity Multiples

PE = 117.94 g + 48.79 Payout + 8.40 Beta (R2 = 0.7315) [Details]

PEG = 0.04 Beta + 2.77 Payout - 0.9688 ln(g) (R2 = 0.7327) [Details]

PBV= 17.25 ROE - 1.42 Payout -0.92 Beta + 17.62 g (R2=.7338) [Details]

PS= 4.36 g - 0.18 Payout + 0.09 Beta + 18.64 Margin (R2=0.7138) [Details]

Firm Value Multiples

Value/Sales = 5.13 g + 0.04 (Net Cap Ex/Total Assets) + 14.02 (Operating Margin) - 1.15 Std dev (R2 = 0.7460) [Details]

V/EBITDA= 3.97 + 7.51 (Return on Capital) -0.01 (Tax Rate) + 0.24 g (R2=0.1686) [Details]

To see the more detailed output from the regression, click on 'Details'.

g = Expected growth in earnings over the next 5 years (enter as decimals, i.e., 15% is .15)

PE = Price/ Current EPS: Companies with negative earnings were eliminated from the sample

PBV = Price/ Book Value per share: Companies with negative BV were eliminated

PS = Price/ Sales per share

Payout = DPS/EPS: from most recent year; if negative, it is set to 100%. (enter as decimals)

Std Dev = Standard Deviation in the Stock Price

Beta = Betas based upon 5 years of monthly data

MGN = Net Income / Sales (enter as decimals)

ROE = Net Income / BV of Equity (enter as decimals)