Analyzing price efficiency using machine learning generated price indices: The case of the Chilean used car market

Abstract

This paper examines how the prices of newly imported cars affect the valuation of used vehicles in Chile’s secondary car market, offering a novel perspective on price efficiency within durable goods markets. Previous studies analyze substitution effects between new and used vehicles in the context of equilibrium models with demand-side heterogeneity. Leveraging a dataset of 2.7 million used car advertisements, we employ Machine Learning techniques to construct synthetic price indices, which serve as the foundation for an event study. Our findings reveal a prompt and statistically significant adjustment in used car prices, particularly among newer and higher-end models, even before the public release of import price data. These results suggest a high degree of informational efficiency in Chile’s used car market and are consistent with demand substitution effects between new and used cars and the incorporation of supply-side shocks by market participants into price valuations.