Two Essays about Agglomeration Dynamics and Firm Economic Performance

There is a long history of research about agglomeration economies in economic geography and regional economics. Researchers have tried and still trying to answer to questions such as does spatial clustering still matter today? how it evolves? which are their determinants? how agglomeration externalities affect the economic performance of regions and firms? The empirical literature on agglomeration economies focus on two main topics. The first is about the sources of geographic concentration of economic activities. The second is related to the effects of spatial agglomeration on firm economic performance. Research about the determinants of agglomeration economies can be date back to Marshall (1920), which identified three different sources: input sharing, labor market pooling and knowledge spillovers. Apparel manufacturer in New York is an example of input sharing, since firms can purchase a variety of relatively cheap buttons from nearby button manufacturing firms. A software company in Silicon Valley can quickly hire one skilled programmer. Meanwhile, a skilled programmer living in Silicon Valley can easily find a new job in this cluster without moving to another place. This is a good example of labor market pooling, which reduces the searching costs for both employees and employers, as well as improves the matching quality. An example of knowledge spillovers can be the random interaction between people working in similar fields who exchange tacit knowledge with each other. Research about the effects of spatial agglomeration on firm economic performance is more recent. Generally it refers to the effects of spatial agglomeration and thus of different types of local externalities on firms’ economic performance, that is whether location within an agglomerated area generates positive returns on the economic performance of firms and, consequently, of the economic dynamisms and growth of regions. This thesis intends to move along this line of research, specifically, try to contribute to this debate in two directions: [1] investigating the temporal dynamics of spatial agglomeration in the Italian manufacturing industry; [2] analysing the relationship between related variety and firm economic performance in China. In general, the thesis is a collection of two empirical studies dealing with spatial agglomeration from two different perspectives. The first chapter of this thesis, “Agglomeration over time”, which is co-authored with Giulio Cainelli (University of Padova) and Roberto Ganau (University of Padova and LSE), is aimed to investigate the space-time agglomeration dynamics that characterised the manufacturing industry during the recent period of the Great Recession. Specifically, the analysis uses a large sample of geo-referenced single-plant manufacturing firms observed over the period 2007-2012 and located in the Italian continental territory to explore the spatial and temporal dimensions of clustering processes, as well as their potential interaction. The empirical analysis is carried out by adopting three different statistical approaches. First, the index of industrial geographic concentration proposed by Ellison and Glaeser (1997). Second, the spatial K-function, originally proposed by Ripley (1976) in the context of spatial points pattern analyses. Third, the space-time K-function, that has been proposed by Diggle et al. (1995) as an extension of the univariate spatial K-function in order to analyse simultaneously the spatial and temporal dimensions of spatial points processes, as well as their potential interaction. The analysis based on EG index highlights the existence of heterogeneity in spatial agglomeration between different industries, but this region-based measure suffers from MAUP problem. To correct the MAUP, we introduce spatial point process method-K function,as well as M-function, which relying on micro-geographic data, rather than pre-defined spatial area, to test firm location patter against Completely Spatial Randomness (CSR). To address the dynamic process that evolve both over space and time, we apply space-time K-function, and some statistical diagnostics, to test the potential interaction between these two dimensions. By space-time analysis, we empirically confirm that, different space-time processes can lead to the spatial patterns which look the same. No significant interaction between spatial and temporal processes, which could be the short period we observe. The second chapter of the thesis “Related Variety and Economic Growth at Firm Level in China”, which is a single-authored paper, aims at investigating the effect of related and unrelated variety on firm level economic growth in China. As empirical results of MAR externalities and Jacobs externalities impact on economic growth are various and inconclusive. Related variety and unrelated variety, a new entropy method proposed by Frenken et.al.(2007), which focuses on the structure inside industry, was applied in this chapter. Basically, firm economic proportional growth specification-Gibrat’s Law, is extended including these two agglomeration externalities-which sectoral diversity is split into related and unrelated variety for distinguishing between sectors with cognitive or technology proximity, with a sample of 84,868 Chinese firms operating in manufactory industry observed during the period 2006-2013. Recent studies about related variety and economic growth, which indeed is the main reason for regional growth, most empirical papers are about developed countries, studies about developing countries are rare. This chapter contributes an empirical study about this debate in a typical developing country, and to our knowledge, it’s the first paper analysis Chinese firms economic growth within related variety framework; besides it’s a firm level empirical research with historical data during 2006-2013, a transformation period for China, with rapid economic development and technological innovation. The results show that, correcting only for sample-selection, unrelated variety has a negative and statistically significant impact. Accounting also for the endogeneity of the two main explanatory variables – related and unrelated variety –the negative effect of unrelated variety becomes insignificant. A positive effect for related variety and negative for unrelated variety is detected only when we consider high-developed Chinese regions. Finally, a positive effect of related variety is identified for large firms.