ABSTRACT
In this article, we empirically investigate regional unemployment in Russia. We first detect the existence of two unemployment clubs, that is, regions with high (low) unemployment surrounded by regions with high (low) unemployment, and a group that comprises the remaining regions. We then apply a specially designed class of spatial-econometric models to regional data 2005–2012, using difference GMM, and we obtain partial confirmation of our two main hypotheses: (i) spatial effects for the high-high and low-low clubs regions differ significantly; and (ii) the determinants of unemployment of the two clubs significantly differ with respect to those of the remaining regions. Our results have key implications for the national- and regional-level policies.
Notes
1. In a pioneering research study, Overman and Puga (Citation2002) clustered 150 NUTS-2 European regions according to several characteristics.
2. On the measurement of polarization, see Esteban and Ray (Citation1994).
3. Few attempts have been made to combine the regional dimension and features with the national-level institutions (e.g., Perugini and Signorelli Citation2007).
4. For the Sakhalin region, these boundaries are measured by sea.
5. As suggested by one of the anonymous referees, this insignificant result could be partly explained by the well-known high percentage of people with higher education in Russia, jointly with a possible lower quality of tertiary education and, especially, a significant mismatch between the supply of graduates and the needs of companies.