Damodaran dismisses the dividend yield model arguing it is too simplistic. Moreover Goyal and Welch (2007) find that the predictive power is low and future equity risk premia (ERP’s) are often negative.
Of course the ERP we calculate with any method is per definition wrong. By now everyone knows the ERP and its underlying model (CAPM) are very limited, however as an investor I’d prefer some quantitative (and qualitative) guidance on the opportunity cost of capital.
Fortunately Damodaran helps me here. Again. He backtests which approach had the most predictive power. So instead of triangulating we can pick a winner: current implied ERP’s, with a close second for 5yr historical implied ERP’s.
The test clearly dismisses historical ERP’s (e.g. 75 year equity returns vs. T-bonds) showing there is a negative correlation between ERP and the next year and next 10 yrs realized premium. McKinsey’s Koller et al seem to come to the same conclusion. Both Koller and Damodaran propose the calculation of an implied equity risk premium.
Damodaran’s paper also gives guidance on what implied ERP to use. If I believe that the direction of the market cannot be predicted (I have to believe markets are efficient) I should use a current implied ERP. But if I ‘believe’ that…markets can be significantly over and undervalued you should use the average historical implied premium.
As I find it hard to accept that irrational short term market movement influence the value of a company (perhaps the stock price, not the value) I would use historical implied ERP’s to determine intrinsic value and current implied ERP to determine price. Price vs.Value.
So how to calculate the ERP? It depends on what you need it for.
Damodaran’s paper on ERP:http://ssrn.com/abstract=1769064