9th Symposium on
Finance, Banking, and Insurance
|
|||
|
|||
|
|||
Ralf Flad |
|||
DG PanAgora Asset
Management GmbH |
|||
This paper
tests the forecasting ability of the residual income
valuation model for European stocks. Residual income
valuation is advocated in a series of recent articles
e.g. by Ohlson, Lee, Lee/ Myers/Swaminathan as an
alternative to traditional discounted cash flow or
dividend discount models to value firms. Based on the
reasonable "clean surplus accounting"
assumption, i.e. changes in book value are equal to
earnings minus dividends, the model concludes that the
value of a firm is equal to invested capital (book value)
plus an infinite sum of discounted "residual
earnings". In this context residual earnings are
defined as the difference between return-on-equity and
the cost-of-capital of a firm multiplied by book value.
From a practitioners point of view this model offers
several attractive features. First it links a firm's
reported book value and expected future earnings, both
measures that are commonly used in investment decisions.
The framework is very intuitive: a firm will be worth its
invested capital if the firm is not able to earn income
in the future at a rate in excess of its cost-of-capital.
The price-to-book ratio of a firm will grow with the
ability of a company to add "economic value",
i.e. to earn income with a rate in excess of
cost-of-capital and vice versa. Second, the theoretical
concept suggests cross country comparisons of firm
valuations because theoretical valuations are less
dependent on particular accounting rules as long as clean
surplus accounting holds. We test the model with a broad
sample of European stocks over the period 1990 - 1998.
The basic decisions in implementing the model are
estimating the future return-on-equity (or earnings) of a
firm and its cost-of-capital. We use I/B/E/S consensus
earnings in an 3-phase approach to forecast
return-on-equity over a 10-year horizon and develop two
different approaches to estimate a firms cost-of-capital.
The first approach estimates a firm specific
"implied" cost-of-capital by fitting the model
to past data in a nonlinear regression approach. In a
second alternative we use a country specific constant
rate which we derive from the local interest rate and a
long term European risk premium. Our major focus is on
the forecasting power of observed mispricings. We try to
explain cross sectional expected returns in various
sampling alternatives: across countries and across
sectors and mixed versions of these. We find that the
"simple" constant cost-of-capital approach
dominates the implied cost-of-capital model both in term
of forecasting power and statistical significance. The
model is best when |
|||
| Stichworte: | |||