Roberto Patuelli (p), Giacomo Galiazzo , Aura Reggiani

Spatial interaction models (SIMs) are extensively used, in many socio-economic contexts (such as commuting, migration, tourism, international trade, communication networks, FDI, citation networks, etc.) to model all kinds of bilateral interactions between spatial units. Typically, municipalities, regions or nations are the reference level of spatial aggregation. However, in this wide literature making use of SIM, very few studies make use of a complete, more proper estimation based on origin and destination constraints (ODC). In particular, ODCs allow to carry out comparative statics (e.g. simulating shocks) while maintaining consistent totals of the origin-destination matrix containing all flows. While ODCs can be numerically approximated for a specific cross-section by sets of fixed or random effects, or by what in international trade are known as multilateral resistance terms, they depend on the explanatory variables in the model, and therefore are dynamic. Their typical nonlinear estimation represents an obstacle, for most researchers, to their utilization. The analysis of tourism flows by means of SIMs has been limited by such difficulty as well. In this paper, we propose an easy-to-use algorithm and software for the estimation and calibration of doubly constrained SIMs. We employ a 12-year panel of domestic tourism flows for the 20 Italian regions, as well as a related database of regional characteristics and tourism push/pull factors for empirical verification and comparative statics. We stress that the proposed calibration method/software may be applied in SIMs for any other type of origin-destination flow data. In addition, based on the cost-sensitivity parameters emerging from the calibration of the doubly constrained SIMs, the accessibility of the most touristic regions will be computed, in order to test whether the most attractive regions in Italy are also the most accessible.

Carlos Espinel (p), Eduardo Castro , Angelo Gurgel

The Brazilian grain production is exported mostly through the Southeast and South ports, which represents a large distance to transport the grains produced in the Center West region. The objective of this research is to analyze the economic impacts of increasing the grain exports from the Center West (CW) region through the Brazilian northern ports using the PAEG computable general equilibrium (CGE) model, to identify the direct and indirect effects in the Brazilian regions. PAEG is a static, multi-regional and multi-sectorial model, representing the production and distribution of goods and services in the world economy. The database comprises bilateral flows of trade between countries and regions, as well as transport costs, import tariffs and export taxes or subsidies. The first step was to calculate the reduction in transportation costs due to the smaller freight prices and distances to the Northern ports considering the potential capacity of these ports. Increasing the grain exports through the North region would generate efficiency gains about 8% in the transportation costs, which supported the shocks applied to the input “transport” in the agricultural sector of the CW region. The main results indicated that increasing the CW agricultural exports through the Northern ports increases Brazilian GDP and increases bilateral exports from Brazil to China, Europe and USA by up to 2%. The results also indicate an income concentration: Center West GDP and welfare grow in all the analyzed scenarios and the gains come from the other regions of the country. The economic impacts are not larger because "transport" is only an input of the agricultural sector in the PAEG model and the investments to construct or repair the roads from CW region to the northern ports were not considered in this analysis. In spite of the positive impacts to the agricultural sector a negative indirect effect (not measured) might happen: increasing the transportation through the North region tends to attract more business, which might contribute to increase the Amazon deforestation. The alternative of increasing the transport flow through the North has been shown effective, but it does not solve the logistic deficit that challenges the Brazilian agribusiness performance nor does it reduce the regional differences in Brazil.

Rémy Le Boennec (p), Julie Bulteau , Thierry Feuillet

In this empirical article, we seek to reveal the effect of accessibility to passenger railway stations in land capitalization. The stakes are crucial, as urban development in many OECD countries has been built around the individual car, favoring sprawled cities. The proximity to a railway station may thus be sought by households engaged in a land acquisition for residential purposes.

The method we used in this article consists of estimating the effects of accessibility to passenger train stations on land prices in the periurban and rural areas of the Loire-Atlantique Département, France. Data on land prices come from the Perval database for transactions recorded in 2012 in both the suburban ring of Nantes (431 observations selected) and in isolated municipalities of the Loire-Atlantique Département (395 observations selected). The database integrates intrinsic attributes: surface area, status concerning the viabilisation, presence of constructions to demolish ...To constitute our database, contextual variables likely to be capitalized in land prices were quantified as well.

Three regression models were estimated afterwards, all adjusted to intrinsic and contextual variables. At first, the average effect of the distance to the nearest station was revealed through a global log-linear model. In a second step, we were willing to test the assumption according to which the function between this distance and the prices would be decreasing. Thus, a generalized additive model (GAM) has been produced. This type of nonparametric model makes it possible to estimate nonlinear functions in a multiple regression. Finally, we explored the possible spatial heterogeneity of the relationship between the distance to the nearest station and the price of land using a mixed geographically weighted regression (MGWR). The underlying assumption is that this relationship is likely to vary depending on local contexts, including the attractiveness of the municipality of the train station, as well as human flows in this station.

Our recommendations support decisions made in the development of local mobility. The results concerning the significant radius of influence of a train station should help decision-makers to make these latter evolve into multimodal exchange poles. In the absence of significant influence of the station considered, a broad incentive to carpooling practices should be driven by local authorities. The range of our results regarding the types of territory considered also prepares for tomorrow's autonomous mobility for “last-mile trips” towards train stations, a solution envisaged by many French local authorities in various types of areas.