Question

You are interested in running a regression to examine the effect on stock prices after being included in the S&P500 index. Give at least two reasons why OLS might not be the best linear unbiased estimator.

Answer 1

The reasons for OLS might not be the best linear unbiased estimator in running a regression to examine the effect on stock prices after being included in the S&P500 index are:

1) The relationship between stock prices and S&P500 index may not be linear.

2) Violation of then homoscedasticity, meaning that the residuals get smaller as the prediction moves from larger to smaller (or from smaller to larger).

3. Violation of normality on residuals. In the linear regression model, we assume the errors are followed a normal distribution.