Models compare actual prices to a but-for benchmark using control variables for cost and demand factors. Benchmark periods may precede or follow the conspiracy, and yardstick approaches compare unaffected markets.
Specification choices materially affect results.
Alternative Names:
Antitrust Regression|Overcharge Regression
Why it Matters?
Benchmark period selection is the most consequential specification choice, since a period contaminated by the conspiracy or by different market conditions produces a biased estimate, and demonstrating that the chosen period was atypical undermines the entire model. Control variable omission and functional form assumptions are separately attackable, and requiring the expert to run alternative specifications exposes sensitivity.
Frequently Confused with
Related terms
Frequently asked questions
What is the most consequential choice?
What exposes model sensitivity?
