A spatial regression methodology for exploring the role of regional connectivity in knowledge production: Evidence from Chinese regions

We explore the role of alternative types of connectivity between regions in knowledge production. Past literature has criticized exclusive focus on the role played by spatial proximity in knowledge production. We introduce a methodology that allows for simultaneous consideration of multiple dependence sets, based on a convex combination of multiple n × n weight matrices. Each matrix represents a different set of connections between regions on which knowledge production exhibits dependence. We show how to estimate spatial panel regression models based on the convex combination of alternative dependence sets, and to draw inferences regarding the relative importance of each type of dependence. We illustrate our method using a model of knowledge production for 263 Chinese regions.     

Natural amenities and regional tourism employment: A spatial analysis

This research examines the spatial linkages between natural amenities and tourism employment spillovers under alternative neighbourhood structures. A main contribution is to contrast models where economic distance defines neighbouring regions with a model where the spatial structure is defined by geographical proximity. Significant spillovers are found not only between geographical neighbours, but also between economic neighbours with important implications for policy design. Internet, population, regional parks and state forests and World Heritage Areas are found to be important drivers of regional spillovers. This study uses data for Queensland, Australia, however the approach and conclusions can be extrapolated to similar economies.     

Territorial determinants and NUTS 3 regional performance: A spatial analysis for Italy across the crisis

This paper analyses the differential impact of several territorial determinants of the economic performance of Italian provinces (NUTS 3 level). as measured by per capita GDP, export and employment growth from 1999 to 2014. It covers both the pre-crisis and the crisis period and stresses the role of geographical proximity in shaping local performance over a wide set of explanatory variables. In order to do so, we employ, firstly, a spatial Durbin model which enables us to discriminate between direct and indirect effects and to highlight the possible contagion or crowding-out spatial effects for each territorial dimension affecting growth. Then, we extend the analysis by allowing for the possibility of two regimes (pre-crisis and post-crisis). The performance of the provinces before and during the crisis relates to specific territorial components and geographic proximity appears to influence differently the results and their interpretation.     

EU regional development policy and territorial capital: A systemic approach

The territorial capital of regions is an important determinant of growth, but also of the impact of EU Cohesion Policy. First, because it can act as a filter, enhancing the impact of regional policies. Second, the latter can help building territorial capital which, in turn, will foster regional development. This work analyses the medium and long-run relationship between the territorial capital of EU NUTS 3 regions and Cohesion Policy using data from the programming period 2000–2006. Results point out complementarities between different territorial assets of regions and the impact of EU regional policy: Cohesion Policy effectiveness is higher when investments are focused on the assets complementary to those already abundant in the region.     

R&D networks and regional knowledge production in Europe: Evidence from a space‐time model

In this study we estimate space‐time impacts of the embeddedness in R&D networks on regional knowledge production using a dynamic spatial panel data model with non‐linear effects for 229 European NUTS 2 regions in the period 1998–2010. Embeddedness refers to the positioning in networks where nodes represent regions that are linked by joint R&D projects funded by EU Framework Programmes. We find evidence that increasing embeddedness in EU funded R&D networks leads to positive immediate impacts on regional knowledge production, and that regions with lower levels of own knowledge endowments more likely exploit the positive effects. However, the long‐term impacts of a region's embeddedness in these R&D networks are comparatively small.     

Regional characteristics and the decision to innovate in a developing country: A multilevel analysis of Ecuadorian firms

This is the first study that uses multilevel modelling to analyse regional influence on the decision to invest in innovation activities of firms in a single developing country, Ecuador. Our results indicate that the decision to invest in R&D and in other innovation activities are conditioned by the region in which the firm is located. Regional loan volume, orientation towards knowledge exploitation and intra-regional-sectorial R&D spillovers are positively associated with both types of innovation activities, while regional levels of co-operation and inter-regional R&D spillovers are only positively associated with the probability of investing in other innovation activities.     

Employment,income, migration and public services: A simultaneous spatial panel data model of regional growth*

We estimate a regional spatial panel simultaneous‐equations growth model, using a five‐step new estimation strategy that generalizes an approach outlined in Kelejian and Prucha. The study region consists of the 418 Appalachian counties 1980?2000. Estimates show feedback simultaneities among the endogenous variables, conditional convergence with respect to the respective endogenous variables, and spatial autoregressive lag and spatial cross‐regressive lag effects with respect to the endogenous variables. A key policy conclusion is that sector‐specific programs should be integrated and harmonized and that regionally differentiated development policies may yield greater returns than treating all locations the same. Estimamos un modelo de crecimiento de panel espacial regional por ecuaciones simultaneas, utilizando una nueva estrategia de estimación de cinco pasos que generaliza una enfoque descrito en Kelejian y Prucha. La región de estudio comprende los 418 condados Apalaches 1980‐2000. Las estimaciones muestran simultaneidades de retroalimentación entre las variables endógenas, convergencia condicional con respecto a las variables endógenas respectivas, y un retardo espacial autorregresivo y efectos de retardo regresivos cruzados espaciales con respecto a las variables endógenas. Una conclusión clave sobre políticas es que los programas sectoriales específicos deberían estar integrados y armonizados, y que las políticas de desarrollo diferenciadas regionalmente podrían producir mejores retornos que si se tratasen todas las localizaciones del mismo modo.