Working Paper: CEPR ID: DP16900
Authors: Roman Inderst; Christopher Milde
Abstract: Arguably the most widely used techniques for estimating price overcharges from competition law infringements are the dummy variable and the forecasting approaches using linear regression analysis. While rarely used in practice, in this note we make use of the fully interacted dummy variable approach to review some basic properties of all three approaches. We show under which conditions and for which estimands of interest these approaches are equivalent and when they differ. We also note some interesting additional choices an interaction approach allows.
Keywords: overcharge estimation; forecasting; dummy variable; interactions; interaction model; treatment effects; damages quantification
JEL Codes: K13; C20
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| competition law infringements (K21) | price overcharges (L97) |
| fully interacted dummy variable model (C20) | average overcharge estimate (C13) |
| forecasting approach (C53) | individual overcharges (L97) |
| single dummy variable model (C29) | potential biases in estimates (C51) |
| price-determining factors (L11) | average overcharge estimate (C13) |