Suppose that the true population regression (data generating process) is given by Y_i = B₀ + B₁X_i+u_i. Further suppose that the population covariance between X_i and u_i is equal to some negative value A, rather than zero:

COV(X_i ,u_i)=A<0.

Derive the expression showing the bias of the OLS estimator for B₁.

Discuss why this covariance might be non-zero (give an example).

Include a sketch that shows this result (negative bias).