The article introduces a new method to estimate variance in economic models where the assumption of equal variance is not true. This method uses simple calculations and does not require knowing the data distribution. By testing this method on real data and simulations, researchers found that it outperforms existing tests for estimating heteroscedasticity.