Do agglomeration forces bring productivity gains to manufacturing firms in Russian urban agglomerations?

The difficulties of agglomeration effects estimation are caused by the problem of unobserved features of heterogeneous cities, industries, enterprises and even employees, related to both the dependent variable (enterprise-level labor productivity) and the specific characteristics of agglomerations. The selection of an enterprise for the analysis of agglomeration effect may result in identification errors, given that an enterprise located in an urban community in a densely populated Western region of Russia will be observationally equivalent to an enterprise in the Siberian rarefied space in terms of its external scale economy if their sizes, specialization, political status and other urban characteristics correspond. We consider that our subject of analysis is nested in several external environments, i.e., that the enterprise is located within a city, that the city is located within a region, and that it is likely that these environments, similar by nature, would work differently, modified in turn by the nature of the enterprise. Another form of self-selection, the exit of less productive firms driven out by intense competition in urban agglomerations, can hardly be disregarded. This form of self-selection is controlled in our research with help of truncated regressions. Therefore, at the regional level of analysis, a measure of the region's involvement in international trade (exports plus imports as a percentage share of the Gross Regional Product (GRP)) is included in regression and then it is analyzing to what extent the power of urban agglomeration effects depend on the location within the region opened to trade and competition. The results suggest that plants in urban agglomerations enjoy 17-21% higher labor productivity. Productivity gained from urban agglomeration is the highest in towns with populations of 100,000 to 250,000 people. This benefit arises as a result of urbanization and external scale economy. Localization and clustering in the city is not associated with higher labor productivity. While regional own-industry clustering satisfactorily explains the productivity premium, suggesting that efficient clustering requires a scale economy larger than only a city. Another result: the urban agglomeration benefit is statistic significant in the firms with middle labor productivity and is absent in high- and low- productive firms. JEL classification: R10, R12, D24 Key words: productivity, city, urban, agglomeration