ECON 104 Lecture Notes - Lecture 33: Panel Data
Document Summary
Now we understand the intuition for how the data should be used. What do we do to statistically confirm the real beer tax effect on fatality. If z were a control variable then it will be a specific regressor and it would be z it. We know that z i is an omitted variable causing trouble, but once you control it then there are no more ovb. Subtract 1988 from 1982 -> you have the difference in fatality rates which is a function of difference in beer tax (same slope) Initial population regression, have already assumed that it"s the true population regression line in slide 12 - regardless of dataset, assume that impact of beer tax will be beta 1. Assume that constant effect of beer tax, hence why beta 1 is constant throughout the years. Divide throughout 2 regressions, you have same beta 0, beta 1, beta 2. Run regression on differences, still estimate same beta 1.