Quantifying the Effect of Transaction Costs on Automobile Replacement

Abstract

This study extends the structural dynamic demand model for new and used durable goods proposed by Schiraldi (2011) and applies the random coefficients discrete choice approach to model consumer heterogeneity in demand and consumer forward-looking behavior. Specifically, I estimate the time- and product-specific transaction costs incurred in the purchase of automobiles with vehicle registration data from South Korea. A combination of the contraction mapping technique (Berry et al., 1995) and the nested fixed point algorithm (Rust, 1987) enables separate identification of the unobserved transaction costs and the unobserved product characteristics. Counterfactual experiments with arbitrary transaction costs as an input imply the significant impact of transaction costs on demand, which offers rich marketing implications.