Why can we assume that the expected value of the error term is zero?

7.5k Views Asked by At

Why can we assume that the expected value of the error term in a linear regression model is zero?

This is with regard to a simple linear regression.

1

There are 1 best solutions below

0
On BEST ANSWER

Because you estimate a constant term $\alpha$ in simple linear regression $$y_i=\alpha+\beta x_i+\varepsilon_i,$$ so that $E[\varepsilon_i]\neq 0$ just means $\alpha$ is replaced by $\hat{\alpha}=\alpha-E[\varepsilon_i]$. Hence, you might as well assume the error term has mean zero.