Model testing

The first tests on the beta-return relationship used two-phase linear regression (Lintner 1965). The first regression estimates the security characteristic line and beta of the individual securities as described above. In the second regression, the security's risk premium is the dependent variable, whereas beta is the explanatory variable. The null-hypothesis assumes the intercept to be zero and the slope of the curve to be the market risk premium, which is estimated as the average of the sample. The test can be extended by an additional explanatory variable: the individual variance.

Data collection

We will present the test using a sample of the US market in the pre-crisis period between 2003 and 2007. As daily data includes more short-term effects, we will apply the test on monthly returns calculated from the daily time series. So, we need the time series of the daily price of more stocks; let us download the prices of the first 100 stocks from S&P 500 in alphabetical order...