Lecture 7 summary · February 26, 2027
Weather, seasonality, and dynamic regression
Complete chapter · Laboratory ch07 · Practice · All summaries
The economic question
Economic predictors can explain demand while the residual remains dynamic. A feasible forecast must specify the future predictor path.
Central relation
Individual problem
Transform a regression with AR(1) errors. Which regressors must change? Can observed next-month weather enter the current forecast?
Worked reasoning
Transform the outcome and every design column, including the intercept. Observed future weather is unavailable now; use a forecast or scenario. Monthly-temperature transforms are exposure proxies.
A consequential failure
Only the outcome is transformed, or realized future weather is substituted for a feasible predictor.
Repair
Transform the full design consistently, declare initial-observation treatment, and distinguish predictor uncertainty from a fixed scenario.