Heteroscedasticity: as a symptom of omitted variable bias - part 1

[@SpartacanUsuals]

Hi, thanks for your comment. I am (slowly) working my way through the graduate econometrics material, and hope to cover GMM and Bayesian over the next few months. Thanks, Ben

[@saqibkhan-nw5yy]

this is very nice effort that you have mad and I m so pleased to see all the materials are given in a very appropriate way

[@sudiptabanerjee6262]

Hi Ben, I was unable to understand the concept of constant variance in ols and varying variance can you please give a simpler explaination, How does it look like in a regression graph? For me the variance is different for all data points on the regression line.

[@appliedeconometricsqa8495]

which software did you use to make video ?

[@troysullivan9404]

Hi Ben, I hear what you are saying, that you are simplifying a concept without necessarily being strictly correct, but several times you say that the population heteroscedastic term is not correlated with the explanatory variable. I think this is not correct and important to draw the distinction. The heteroscedasticity is showing that the omitted variable is in fact correlated to Xi, but as you say the omitted variable may not be important for our regression, as in the example of an individual's taste for food. So in the end we would just want to adjust for the omitted population variable and be done with it. In a 3D graph the two independent variables would form a plane, which has slopes in two dimensions. If we graphed the points on the plane we would find that all of the points lie closer to the plane than they did to the 2D regression line, and there is constant variance with respect to the plane. Is this a valid way of thinking about it? Thank you again,