The article investigates how multicollinearity affects linear regression estimates. The researchers used Monte-Carlo simulation to create highly collinear variables and compared different regression methods. They found that ordinary least squares had consistent bias across sample sizes, while lasso estimators varied. Ridge regression had the lowest mean square error at small sample sizes, while OLS performed best with larger sample sizes. The study suggests that OLS is reliable at specific sample sizes, and ridge regression is effective with small to moderate sample sizes.