Argentine regions based on dynamic criteria

In this paper, we use the Stock and Watson methodology to estimate economic coincident indexes for each of the twenty-four Argentine provinces. We extract business cycle components from these indexes using the Christiano-Fitzgerald filter and then we group Argentine provinces into regions according to the dynamic behavior of their economies, applying a Ward-like hierarchical clustering algorithm under different scenarios. We found very varied results, but with a certain regularity that can be highlighted, since there are some groups of provinces that are clustered together in every scenario. However, neither scenario produce any regionalization similar to the statistical regions determined by the Argentine Institute of Statistics and Censuses (INDEC). When we assign equal weights to the contiguity and business cycle dimensions in the clustering process, the resulting clusters are similar to what we can expect from a economic regionalization, that is, complete contiguity, business cycle similarities and a relatively balanced size. Another particularly interesting result is that the provinces that concentrate the country's agro-industrial production and exports (Córdoba and Santa Fe) appear together in almost all the clustering procedures As a whole, the results show that regionalization based on static criteria may not be the most appropriate approach when dynamics matter.