Jean-François Ruault (p), Yves Schaeffer

The downward trend in employment in French manufacturing industry, not to mention its dramatic fall in the aftermath of the 2008 economic crisis, is reflected in significant regional disparities. These two issues – the national adverse trend and the resulting inequalities – are of great political importance and much remains to be done to better understand them. To this aim, this article proposes an update of the shift-share analysis (SSA) method and applies it to the French manufacturing industry case. The standard SSA explains regional inequalities, i.e. deviations of regional employment growth rates from the national trend, by specific regional competitive and industry-mix effects. But to implement this decomposition, all national industries are required to exist in all regions. Our update removes this unrealistic constraint and generalizes the SSA approach to the usual case where regions have missing sectors. In addition, the emergence over time of initially missing sectors and the explosive growth or degrowth of some small sectors are incorporated as specific components in the SSA decomposition, along with the competitive and industry-mix components. This generalized SSA method is then applied to better understand the decline of French manufacturing sectors over the period 1993-2015 at NUTS 2, 3 and LAU 2 geographical levels. Results show that regional inequalities over time are induced by competitive effects, i.e. inequalities in average sectoral growth rates, rather than industry-mix effects, i.e. inequalities due to differences in economic structures. In addition, the smaller the geographical scale, the greater the effects of emergence and outliers on regional inequalities.

Monica Mihaela Tudor (p), Violeta Florian

The analysis of EU citizens perceptions regarding the regional development and effectiveness of European programs and/or institutions in dealing with regional issues was performed based on quantitative methods. On the other hand, the regional experts/ practitioners point of views regarding the above-mentioned regional aspects was mostly analysed using qualitative methods. So far, the two regional actors’ visions were separately analysed, their integration being not at all an easy task having in view the methodological differences and limitations of each data collection methods.

The general objective of the paper is to propose an innovative approach in analysing the citizens’ and practitioners’ views on regional issues in order to make the two visions integrable and comparable. This methodological option belongs to the mixed methods design that makes possible to combine the qualitative and quantitative data on the same research topics into the same study. The methodological scheme proposed in our paper is based on convergence model from the mixed model triangulation design.
This methodological approach allows, in a first stage of research, to collect and analyse simultaneously and separately the quantitative and qualitative data, using a common ground regarding the research question/topic. In the second phase of research, the two results are converged (by comparing and contrasting the different findings) during the interpretation phase. Contextualization, that gives a meaning of the obtained results with reference to the specific and particular context, is used for interpreting of both qualitative and quantitative data in order to make them suitable for being comparable.

Viability of this methodological approach was validated into a case-study regarding the citizens’ and practitioners’ views on EU Cohesion Policy. The case-study analysis was conducted at the level of a representative sample of European NUTS II regions.

The comparative analysis highlighted the convergence and divergence points between citizens and practitioners regarding the public intervention needs through Cohesion Policy and evaluation of the effectiveness of these interventions, thus contributing to a better understanding of the general perception of the EU by the large public.

Acknowledgment: paper supported by European Union's Horizon2020 research and innovation programme under the project: Perception and Evaluation of Regional and Cohesion Policies by Europeans and Identification with the Values of Europe (PERCEIVE)

Alexandre Almeida (p), Carlos Azzoni

Food expenditures, mainly in low-income families, have a substantial weight in the household budget, deserving special attention. The estimation of the Engel curves, relating the share of food in total household expenditure to income levels, is relevant for public policy design, both for poverty reduction (e.g. cash-transfer programs) and to agribusiness decisions, (e.g. definition of retail prices of food). In this paper we apply parametric and non-parametric methods to pseudo-panel data from four household expenditure surveys, for six product groups (food, housing, transport, education, health and clothing). The surveys cover all 11 major metropolitan regions in Brazil. The use of panel data allows for controlling for unobserved characteristics. In addition, non-parametric methods allow verifying if the function is adequately specified according to the nature of the data and its functional form. We estimate alternative models, including fixed effects and non-parametric panel data estimators. The results for the mean values of non-parametric estimates are not different compared to fully parametric estimations. However, in different percentiles, the non-parametric estimates of household income and household components produce different impacts of these on the share spent by households on each product group. We also conclude that non-parametric methods with controls for the unobserved characteristics of household heads in panel data produce similar results to those of non-parametric methods in cross-sectional data. For the fully parametric models, similar coefficients of the independent variables were also found with cross-sectional (pooled) and panel data (fixed effects) for six consumption goods analyzed.

Ghislain Geniaux (p), Bertrand Leroux

This paper focuses on the methodological advances and scientific results of the Urbansimul project (https://urbansimul.fr). Urbansimul is a decision-making tool for urban planning and land prospecting developed as part of a long-term partnership (2009-2019) between research (INRA, Cerema), the regional public actors (PACA Region, DREAL PACA) involving local authorities in urban planning (municipalities, intermunicipalities) and public land acquisition structures (EPF).
Thanks to access to a spatial microdata on land and house property (DGFIP), land and house prices (DVF) and digitalized municipal land use plans on more than 5 million parcels monitored annually from 2007 to 2017, has enabled us:
- to built an unique learning database for the statistical analysis of land use changes and land market dynamics.
- to systematize the identification of available developable land: this type of information is essential to accurately assess the land supply, which is a key variable in urban economics (see Géniaux, Napoléone and Leroux 2015, RERU),
- to design several prospective models at the parcel scale on the probabilities of land use changes, on building capacities and land supply, and on population dynamics at the communal level.
UrbanSimul's prospective models are based on methodological advances in spatial econometrics in bigdata (Géniaux and Martinetti (2017) RSUE, Martinetti and Géniaux (2017) RSUE, see R packages mgwrsar and ProbitSpatial) and machine learning (boosting with spatial autocorrelation). These methods aim in particular to predict the constructive capacity of parcels and their probability of becoming built using a very wide range of descriptors of the land owners, the zoning and urban regulations, the physical constraints, the neighborhood contexts and past regulators behaviors in urban planning management. Without going into detail on the estimation tools, this paper will introduce a new conceptual framework for modelling the links between estimated parcels conversion probabilities and zoning change decisions. Our proposal is based on both the assessments of these probabilities of land use change and observed zoning changes over 10 years (more than 167730 parcels of 374 municipalities with zoning changes). The challenge here is to be able to identify the areas with a very high probability of zoning change in future municipal land use plans. A quantitative assessment of 10 years of evolution of local urban planning in the PACA region will also be carried out, illustrating the main mechanisms that govern zoning developments following the SRU and UH laws in France.