Agglomeration Effects in a Developing Economy: Evidence from Turkey

Productivity differences across Turkish provinces is one of the highest among the OECD countries. In this paper, I estimate agglomeration effects for Turkish provinces to shed light on the causes of productivity differences and provide evidence on the importance of such effects in a developing country context which literature needs. I use a novel administrative dataset recently made available at NUTS-3 level, for 81 provinces of Turkey for the period 2008-2013 and carry out a two-step estimation. Using a variety of panel data techniques and historical instruments to deal with estimation concerns, I estimate an elasticity of labor productivity with respect to the density of 0.057-0.06, which is higher than in developed countries and around the levels observed in developing countries. I find that domestic market potential matters even more than density and is the most significant determinant of the productivity differences across Turkish provinces. Finally, in stark contrast with the evidence coming from developed countries, I do not find any effects for positive sorting of workers across provinces. This finding suggests that urbanisation patterns may be operating differently in developing countries, indicating the need for further evidence from such countries.