Vanessa Kuentz, Yves Schaeffer (p), Tina Rambonilaza

“What we measure affects what we do. If we measure the wrong thing, we will do the wrong thing.” This introductory sentence from the recent OECD report on country-level well-being metrics (Stiglitz et al., 2018) is equally relevant to the analysis of regional inequalities and regional development policies. An overarching recommendation to monitor sustainable development policies is to rely on a set of relevant “Quality of life” and “Environmental” indicators rather than on a synthetic indicator (e.g. GDP) and to pay attention to their temporal evolution. In this respect, for the study of regional development, a multidimensional clustering approach has sometimes been used to identify “regional development profiles” and thus give meaning to this wealth of indicators. However, this clustering approach should be further improved to measure the “right (regional) thing”. The aim of this article is to make proposals in that way.
There is already a method for the clustering of a set of numerical and categorical indicators (see the R package ClustOfVar) and we believe it should be applied before the clustering of regions in order to eliminate informational redundancies in the indicator space. But a gap that still needs to be filled is the lack of treatment of spatial associations in the step of grouping indicators (i.e., the fact that, for an indicator, the values in a spatial unit and in neighboring spatial units may be related). This must be taken into account insofar regional development has a spatial dimension that should be understood to target regional policies better. Our first objective is therefore to include this spatial dimension in ClustOfVar using the Vector Quantization technique. A second gap to be filled is the way in which regional trajectories are analyzed, usually by simply comparing regional typologies established at different times. An alternative approach would be to focus on indicators’ dynamics and implement a clustering of variables followed by a clustering of regions based on temporal evolutions. This paper explores these methodological avenues to advance the analysis of regional development.

Alessandra Caputo (p), Domenico Scalera , Emanuela Sirtori

According to European Commission, Emerging Industries are industrial sectors in early stage development based on new products, services, technologies or ideas and characterised by high-growth rates and market potential. While scientific literature on the topic is still in its infancy, the European Commission is increasingly involved in launching initiatives at EU level to support clusters and individual firms in these industries. Consisting of restructured sectors that transform, evolve or merge into new industries, emerging industries usually result from inter-industry spillovers, entailing the transformation of traditional industries to respond to new market demand and exploit new key enabling technologies. The biopharmaceutical emerging industry is a momentous driver of scientific advancement. It stems from the mixing of the traditional pharmaceutical industry, dating back to the late 1800s, with the more recent biotechnology industry, developed on living cells and molecules, stemming from key innovations in the 1970s and 1980s. Whereas pharmaceutical companies are continuously developing biotech-related drugs, so that the distinction between ‘pharma’ companies and ‘biotech’ companies is progressively losing sense, biopharmaceutical companies are more and more interconnected with companies belonging to other industrial and technological areas favouring the rise of new industrial and technological trends. This paper aims at identifying geographical patterns of the biopharmaceutical industrial transformation trends at the scale of EU regions, over the time period from 2000 to 2016. Data on co-patenting, M&As, and JV&As are used as reliable proxies to capture cross-sectoral industrial transformation trends along the value chain of the industry. Combining together these three sets of data, the paper adopts a partially novel approach: a Network Analysis is performed at NUTS2 level to identify regional communities and geographical hotspots, by using data on the location of inventors and companies involved in the cross-sectoral operations. Preliminary results suggest that from 2000 to 2016, on the one hand, there has been an increase in biopharmaceutical cross-sectoral activities in Eastern Europe; on the other hand, some peripheral regions, such as those in Southern Italy, have lost their relevance. A further finding is the steady growth of cross-sectoral activities among Spanish regions. This paper contributes at the same time to both the literature on the geography of industrial activities, by employing a new method of analysis, and to the strand of literature focusing on emerging industries.

Yan Ciao Sie (p)

Understanding urban with urban morphology helps to propose a management of the urban landscape which is based on the development of history. This study explores the "plot" which is an important elements of urban morphology, and taking Tainan which is a historical city of Taiwan as a study area. It observes and analyzes the spatial distribution and morphological evolution of the 1920-2016 plots. In Taiwan, the historical cities are less observed from the plots. Additionally, a plot are also the unit of land use and control, affecting the development of the surface. It connect to the establishment of the landscape and urban functions. The larger the plot is, the greater it influence.Therefore, in the face of a city that is developing faster and faster, it is helpful to provide a reference for urban landscape management by summarizing the characteristics of urban large plots and surrounding plots.

This study analyzes the overall plots in different periods by spatial auto-correlation analysis and percentage rating analysis, to find out the spatial aggregated and characteristics of the large plots. Calculating the openness and compactness of the large plots to analyze the cutting situation in different periods. Then using the town-plan analysis of Conzenian to observe how the large plots are affected by complexes of plan elements (plots, streets and town plan), building types and land use. Finally, it examines the morphological evolution and classification principle of urban large plots and surrounding plots based on land ownership. As a result, we can know which types of large plots are most possibly developed or need to be protected. So that they can be used as the basic information for urban design guidelines, land use control and development proposals, to manage urban landscape.

Stevi Vafeiadou (p), Marie Noelle Duquenne, Dimitris Kallioras, Maria Tsiapa

Fueling the relative academic debate and providing insight to the evaluation of the relative policies, the evolution of regional inequalities is an issue of utmost importance. Hence, the study of regional inequalities - in particular, the study of regional convergence / divergence - is at the heart of regional science. From the policy viewpoint, the study of regional convergence / divergence may interpret as a sign with respect to the evaluation of the effectiveness and the efficiency of the implemented regional policy mix. From the theory viewpoint, the study of regional convergence / divergence may serve as an empirical exercise with respect to the affirmation of regional development theories.
Referring to the diachronic decrease of the overall dispersion of a (regional) dataset, σ-convergence is a dominant concept in the empirical regional convergence / divergence literature. Σ-convergence may, usually, apprehend through the coefficient of variation (CV) and the weighted coefficient of variation (wCV) measures. CV is a standardized (relative) measure of dispersion and may express as the ratio of the standard deviation of a (regional) dataset to the corresponding mean, at a given date. Including a weighting factor in the CV formula, so as to account for the corresponding relative (regional) size in the treatment of the (regional) dataset, allows for the compilation of the wCV formula, the weighted CV counterpart.
The paper revisits the σ-convergence concept, on the, purely, statistical rationale that the mean is a central tendency measure highly sensitive to the presence of outliers. To this end, the paper specifies the conventional CV and wCV measures against the backdrop of the median, proposing the corresponding CVmd and wCVmd measures. Supported from an illustrative empirical analysis of regional inequalities in the EU countries (NUTS III spatial level), the paper indicates that different expressions of the σ-convergence concept may lead to different inferences with respect to regional inequalities. Such an indication, besides its scientific importance per se, provides important policy implications given that different expressions of the σ-convergence concept may mask the magnitude of the actual regional problem.

Bart Los (p), Sarina van Doorn

How have economic growth rates in the regions of the UK (defined at NUTS1 and NUTS2 levels) been affected by the outcome of the Brexit referendum?
Proponents of Brexit have repeatedly argued that the negative consequences of the referendum outcome as predicted by most economists have been overstated. They refer to the fact that UK growth remained at levels comparable to its pre-referendum pace. Many economists, however, mentioned that post-referendum economic growth in other G7 countries has been considerably more rapid than in the UK. To resolve this disagreement, one should come up with a useful and plausible "reference scenario".
Born et al. (2017) and Springford (2019) have adapted the Synthetic Control Method (SCM. introduced by Abadie and Gardeazabal, 2003) to this end. They compare the post-referendum growth performance of the UK economy to the performance of a "lookalike". This lookalike consists of a linear combination of other countries (that were not hit by the referendum shock). A regression-based algorithm ensures that the lookalike had a growth performance very similar to the UK before the referendum, and/or had a very similar performance in terms of determinants of growth. According to Springford (2019), the UK economy was about 2.3% smaller in September 2018 than its lookalike.
The economic performance of UK regions is very heterogeneous, as stressed by McCann (2016), and several researchers are afraid that the consequences of Brexit will mainly hurt already lagging regions (Chen et al., 2018). To see whether economic uncertainty and lowered business and consumer confidence have affected UK regions differently, we construct SCM-lookalikes for these regions and study the differences between the regional GDP growth rates of the regions and these lookalikes for the one year and a half after the referendum. The data are from Eurostat.


Abadie, A. and J. Gardeazabal (2003), "The Economic Costs of Conflict: A Case Study of the Basque Country", AER 93(1), 113–132.
Born, B., G. Muller, M. Schularick and P. Sedlacek (2017), "The Economic Consequences of the Brexit Vote", CESifo Working Paper 6780, Munich.
Chen, W., B. Los, P. McCann, R, Ortega-Argiles, M. Thissen and F. van Oort (2018), "The Continental Divide? Economic Exposure to Brexit in Regions and Countries on Both Sides of The Channel", PiRS 97(1), 25-54.
McCann, P. (2016), The UK Regional–National Economic Problem: Geography, Globalisation and Governance. Routledge.
Springford, J. (2019), "The Cost of Brexit to September 2018", CER Insight, London.