The home-market effect and bilateral trade patterns: a reexamination of the evidence

In this paper, we reexamine closely the empirical evidence for the home-market effect (HME) found by Hanson and Xiang (American Economic Review, 2004). We first show that evidence for the HME from their difference-in-difference gravity equation is sensitive to the way the independent variable of interest (i.e. the log of the ratio of GDPs of exporter pairs) is created. Moreover, regardless of how the sample is configured, the HME is found only in particular sub-samples of country pairs. Second, we find no evidence of the HME when we estimate the difference-in-difference gravity model on a truncated sample of positive trade flows. We also find that the magnitude of the estimated HME is sensitive to the value imputed to zero trade flows. Monte Carlo simulations show that the truncated OLS, the Eaton-Tamura Tobit and the Heckman sample-selection estimators outperform (in terms of both the bias and variation of the gravity estimates) Hanson and Xiang?s difference-in-difference estimator. Truncated OLS, ET-Tobit and Heckman estimation of the gravity equation using Hanson and Xiang?s data yield no evidence of the HME. Finally, evidence from data on exports from non-OECD countries and from Canadian provinces to the US states provides no support for the home-market effect.