Two features in a dataset are perfectly correlated: x₂ = 3x₁. What happens when both are included in an OLS regression?