CEPII, Recherche et Expertise sur l'economie mondiale

The new RCA database contains revealed comparative advantage measures (RCA) obtained by using the methodology suggested by Costinot et al. (2012) - adjusted to some extend for a proper worldwide comparison of indexes.

Several indexes have been proposed as proxies for country-sector specific comparative advantage; among them Balassa index (Balassa 1965) has been widely used in the literature. However Balassa Index has not a proper theoretical micro foundation and some problems related with its empirical distribution.

The new index proposed by Costinot et el (2012) – and here systematically computed for several countries and products – solves both theoretical and empirical distribution problems with former measures of comparative advantages.   

The index is then provided for 20 countries over the period 1995-2010 at two different product disaggregation levels (HS-2 and HS-4). Each database contains 5 variables:
  • country is the country of interest (i.e. the exporter country);
  • isocode is the ISO 3166-1 numeric code of the country;
  • year is the year dimension;
  • hs2 or hs4 indicates the sector of interest (respectively HS-2 and HS-4);
  • RCA is the index of revealed comparative advantage.

 Download RCA index for several industries and countries

 Methodology: a Short Description
The RCA index is based on econometric technique (detailed description on the related CEPII working paper). All in all, it is a three steps procedure:
  1. First, we estimate country pair-product specific export flows using exporter-product and importer-product fixed effects.
  2. Second, we use the coefficients associated with the exporter-product fixed effects to recover a productivity parameter (which is then country-product specific)
  3. Third, we normalize the previous productivity parameter to the sake of easy interpretation.    


Reference document to mention
Leromain, E. and Orefice.G (2013), "New Revealed Comparative Advantage Index: dataset and empirical distribution", CEPII Working Paper 2013-20

Download Page


Gianluca Orefice