Anyone who has ever dealt with valuation or modern corporate
finance has probably encountered the use of beta. In modern finance beta is the
key relative risk measure used in the Capital Asset Pricing Model (CAPM) to
determine the risk premium demanded by equity investors. I’d like to take a
minute to talk about what exactly beta is measuring along with methods by which
to determine the beta used in CAPM.
Let’s begin with what exactly beta is meant to measure. Beta
is a measure of systemic risk of a security versus the market as a whole. It is
generally viewed as the covariance between the returns of the security and the
returns of the aggregate market divided by the variance of returns of the aggregate
market. Or rewritten as the correlation between the returns of the stock and
the aggregate market multiplied by the standard deviation of the stock returns
dividend by the standard deviation of aggregate market returns.
Most commonly these calculations are done
using a simple regression with the returns of the aggregate market being the independent
variable and the returns of the stock being the dependent variable. The regression
result will provide the coefficient of the x variable which will represent
beta. In essence, if we apply the simple linear equation (Y = mX + B) the “m”
is beta which show how much of a movement can be expected in the given stock
based on the movement of the aggregate market. It is common to use the returns
of the S&P 500 to represent the market and between a 3-5 year time horizons
for return data.
While much more could be said regarding
the technical side of regressing to find beta, I would like to look at one
simple fact. In any regression the coefficient of determination (R2)
indicates how much of the data can be explained by the regression model. Put simply,
it is a measure of accuracy within the regression model. In almost any regression
model to find beta, the R2 will be very low (0.30 or less), this
leads us to question the accuracy of using this method to calculate such an
important variable when calculating the cost of equity capital.
Before I get into an alternative method
for beta, I would like to shift our focus back to what beta measures for a
moment. Beta is the measure of systemic risk of a firm, and part of this risk
is both operating risk and financial risk. Operating risk can be defined as the
risk inherent in a firm that is fully financed with equity, essentially the
risk of the business if it had not required debt payments. The firm’s operating
risk is impacted by the amount of fixed versus variable costs. The other part
of this risk is financial risk, which is the risk as firm’s equity holder bear
due to the use of leverage and thus required debt payments. Now in calculating
a regression beta we will have a result of levered beta, which is the beta that
includes both operating risk and financial risk. Let’s now jump to an
alternative method for calculating beta.
An alternative that is commonly used is
to first identify the firm’s industry and collected data on a group of
comparable companies. Then take these firms and run a simple regression to
calculate beta. The betas collected, as described above, are levered betas.
Since our firm, and each firm in the group of comparable companies will most likely
have different uses of leverage, we must “unlever” these betas. The
process of unlevering beta is essentially factoring in both the amount of debt
capital used as compared to equity capital along with the benefits provided by the
tax shield offered from interest payments. Levered beta can also be calculated
from unlevered beta. The calculations for both are described below.
To continue with the alternative beta calculation, you first
must take the levered betas found through regression and unlevered each of
these based on each firms respective debt to equity capital structure and
corporate marginal tax rate. Then a mean or median of these unlevered betas can
be used to determine and estimate for the industry beta; the choice of mean of
median will be a judgment based on variance of these betas. Once the industry’s
unlevered beta is found, the beta can be relevered using our given firm’s debt
to equity structure and marginal tax rate to determine the beta used within the
CAPM.
Now you may ask “what is the benefit of completing this
process versus using a simple regression?” Well as discussed above, simple
regressions have very low predictability value and can lead to misleading
betas. By using this alternative, you are collecting date on multiple firms
within the industry which can eliminate some of the risk of having an outlier. Is
it perfect? No, but I do believe that it can provide a better estimate of the
true systemic risk of a firm.
Hope you found this interesting.