54th Congress of the European Regional Science Association: "Regional development & globalisation: Best practices", 26-29 August 2014, St. Petersburg, Russia
We study the effect of international trade and freeness of trade on interregional inequality within countries. While most studies focus on single countries and only some use a small panel of countries, we use the two largest databases available to examine these issues. One database is provided by Gennaioli et al. (2013) and covers 105 countries with 1569 regions in 2005. The second data base is collected by Lessmann (2011) and covers 56 countries with 835 sub-national regions for the period 1980-2009. We estimate a model derived from a structural economic geography approach where interregional inequality depends on weighted trade shares and trade costs and where we can derive an aggregate freeness of trade measure. These measures are instrumented based on constructed trade shares and trade costs fitted from a gravity panel model of bilateral trade, that is also derived from the NEG model. We use data on 208 countries for the period 1948--2006 to estimate bilateral trade and bilateral freeness of trade. The results are used to construct proxies for trade to GDP and aggregate freeness of trade of a country. Next, we carry out cross section analyses on the huge data base of Gennailoli et al. (2013) where we use the constructed proxies as instruments. IV estimates provide ambiguous evidence that both a higher trade to GDP ratio and a higher freeness of trade raise interregional inequality. Applying the same approach to the Lessmann data confirms the findings also for his selection of countries. We take this as sign that the Lessmann data do not suffer from a selection bias in comparison to the Gennaioli et al. data. This raises our confidence that using the Lessmann data in a panel approach provides general findings. FE, Panel IV and dynamic panel regressions confirm our findings concerning the trade to GDP ratio. In contrast these within estimates do not provide significant results for the freeness of trade. Because the latter is an indicator for integration in the world markets, we conclude that more integration neutralizes the negative interregional distribution effects of the increase in trade. This implies, so our policy conclusion, that the negative impact of trade on inequality could be reduced by raising trade with countries that are highly integrated in the world market and characterized by a high freeness of trade.
Regional inequality trade gravity model panel data new economic geography