The factor content of regional bilateral trade: The role of technology and demand

The Heckscher-Ohlin-Vanek (HOV) model in its strict form has been strongly rejected by the data. Relaxing some assumptions of the standard HOV model is key to find improvements in its performance. We apply the Davis and Weinstein (2001) methodology to analyse the validity of the HOV model using regions rather than countries. Surprisingly, our results using data for 17 Spanish regions are similar to theirs with international data for OECD countries. Accounting for technological differences improves the predictive capacity of the factor proportions model and including trade costs and geography reduces significantly the missing trade problem. However, relaxing the assumption of factor price equalisation does not improve the performance of the HOV model in a regional setting.