From Classroom Algebra to Modern Econometrics: How Equation Subtraction Powers Regression
Consider the classic econometric modeling challenge of assessing a minimum wage hike across state borders. Suppose New Jersey raises its hourly floor, while neighboring Pennsylvania leaves its rate unchanged. Comparing post-policy employment between the two states provides misleading results because New Jersey and Pennsylvania have structurally different economies, living costs, and labor densities.
Economists solve this with difference-in-differences estimation, a method built on two distinct rounds of equation subtraction:
First, analysts take the post-treatment employment equation for New Jersey and subtract its pre-treatment baseline. This removes all permanent, baseline characteristics unique to New Jersey that existed before the law changed. Second, analysts perform the identical subtraction on Pennsylvania's timeline, netting out regional trends that occurred over the same calendar window. Finally, subtracting the second equation from the first leaves behind only the treatment effect. Unobserved state characteristics vanish from the ledger because they entered both time periods with identical values.