Estimating expectations-based reference-price effects in the used-car retail market

This paper estimates the reference-price effects in consumers’ car purchase decisions exploiting a unique panel data-set on the non-negotiable daily list prices for used cars advertised by a large national dealership. Guided by the theory of Köszegi and Rabin (The Quarterly Journal of Economics, 1133–1165, 2006), the paper identifies the reference price for a car as the car’s list price on the previous day, a point with a large probability mass in car buyers’ price expectations. The paper finds that — controlling for a car’s actual price, its detailed attributes, the competitive environment, and a rich set of fixed effects — a positive (negative) deviation of the car’s actual price from the previous day’s list price lowers (increases) its daily sale probability significantly more than what can be accounted for by the standard effect of the car’s resulting actual price. The magnitudes of the estimated reference-price effects are comparable to that of the standard price effect and are significantly stronger for cars with more exposures to prior consumer search. Implications for the dynamic pricing strategy for car dealers (and retailers in general) and inflation dynamics are discussed.