Estimation of Cost of Capital 89Holding period return for the stock =
1–( – 1– ) Div
ttt
PPP +
where Pt and Pt–1 are current and prior period prices, and Div is the dividend received.
Similarly,
Index return =
Beginning value Dividend yield indexonValue of index at the end – Value at the beginning
+The estimate of beta depends on four factors:
1.The period over which the beta is estimated. The model does not specify the time period over which
betas have to be estimated. An analyst has the liberty to choose an appropriate time period. Typically,
analysts use 2-year and 5-year data. The latter is more popular. The longer the period, the more the
number of observations and more meaningful the estimate. But macroeconomic changes may be better
reflected in recent data. Further, the company itself might have changed in risk complexion. So there is a
trade off between number of data points and accuracy.
2.The starting and ending points of the estimation interval. One could randomly choose a starting and
ending point or choose some deliberate interval. Some start with January, some with July. There are no
rules except that an analyst should avoid data that are abnormal. For instance, for some reason if the
returns in January are consistently more than returns in other months, one could avoid January data.
3.The market proxy used. A stock market index is generally used as a proxy for the market portfolio that
is supposed to cover all risky assets in the economy. The beta estimate is sensitive to the market proxy
used. We could use the Bombay Stock Exchange sensitive index (Sensex), National Stock Exchange
50 (NIFTY), CRISIL 200 or BSE 200 or any other market proxy. As long as these indices are highly
correlated, it doesn’t really matter which index is chosen. But the problem is that none of these indices
might be good proxies for the market portfolio. This issue will probably never be resolved. The beta
estimates for some well-known companies are shown in Exhibit 4.1.
Exhibit 4.1 Beta estimates with different market proxies
Company BSE 200 NIFTY Sensex
Ashok Leyland 0.78 0.89 0.77
Bajaj Auto 0.90 0.92 0.85
TELCO 0.97 1.04 1.00
SBI 1.59 1.20 1.30
ACC 1.28 1.23 1.23
Colgate Palmolive 0.76 0.68 0.68
Reliance 1.42 1.24 1.38
The differences in beta estimates are not dramatic although different proxies do give different results.
4.The choice of return interval. The analyst can choose weekly, monthly, quarterly, or annual data to
estimate beta. The normal practice is to use weekly returns. The choice is made depending on whether
weekly or monthly returns have significant correlation with beta. CAPM can be applied only if there is
correlation between holding period returns and beta. A recent study suggests that there is significant
correlation between quarterly returns and beta. As usual, the beta estimate is sensitive to the return interval
chosen. Exhibit 4.2 presents the beta estimates of the same companies for various return intervals with
Sensex as proxy.