Filipe Batista e Silva (p), Cristian Pigaiani, Lewis Dijkstra

Oscar Fernandez

Geospatial data about the location of population is essential for policy support and scientific assessments in many fields, including regional science and urban analysis. Since the early 2000’s, the European Commission acknowledged the usefulness of population grids, and pushed for the improvement of their coverage and quality. Specifically, the Joint Research Centre and Eurostat have made substantial contributions to advance the mapping of population distribution at fine spatial scales.
The current state-of-the-art includes the census grids for 2011 and 2021, compiled by Eurostat with input from National Statistical Institutes. These grids are based on the aggregation of address/point-based population counts to the reference European 1 km grid system. However, these so-called “bottom-up” grids are not available for earlier census years due to the lack address/point-based population counts for most EU countries. “Top-down” grids have been produced for previous or intra-census years by downscalling aggregate municipal counts of population to the 1 km grid cells in a conventional, dasymetric fashion. However, because of the inconsistency between the top-down and bottom-up methods, time-series analysis is not warranted.
Here, we present a new, chain-linked dasymetric back-casting approach to generate consistent, decennial population grids going back to the year 1961 for Europe. The approach combines known population from the 2011 census grid with historical population at municipal level and land use changes derived from Earth Observation. Independent validation corroborates the superiority of the approach vis-à-vis static dasymetric approaches. The presentation will describe the method and validation results and illustrate their application to assess long-term urbanisation and depopulation trends in the EU.
Based on preliminary results, between 1961 and 2021, the EU population increased from 359 to 456 million inhabitants. This overall demographic growth was accompanied by a steady urbanisation process, with population living in urban areas increasing from 59% to 71% at the expense of rural areas, which dropped to a share of 29% of the EU population in 2021. The high spatial detail of population grids allows to discern distinct spatial patterns. Population has increased substantially in or around the main cities. Coastal areas and coastal cities observed important population growth too, especially in the southern EU. Rural areas lost population overall. But the rural decline has been more pronounced in the southern and eastern EU, with large swaths of inner/rural parts of, for example, Portugal, Spain, Croatia, Bulgaria, Romania and the Baltic countries experiencing a strong population decline in many areas.

Fernando de la Torre Cuevas (p), Michael L. Lahr, Ana Lúcia Marto Sargento, João Pedro Ferreira

Filipe Batista e Silva

The accuracy of subnational input-output (IO) accounts tends to suffer from data scarcity. Literature on how to construct subnational IO accounts proposes several ways to estimate the magnitudes of intersectoral linkages within regions as well as for interregional trade flows. In some cases, data scarcity is so severe that information on industry productivity levels across regions of a nation is totally absent. Subnational data in certain countries/regions (e.g., the European Union, the United States, and Japan) are increasingly accessible, yet their integration into subnational IO account estimates often does not use all information available from statistical agencies. Introducing such superior data comes with trade-offs, e.g., it consumes more time and requires more funding. In this paper, we assess how much more accurate and meaningful interregional accounts might be after introducing subnational information on value-added by industry when using an interregional IO account. To do so, we propose a simple experiment that uses four sets of alternative accounts. We start by aggregating a subset of an accepted global input-output (GRIO) account to produce an EU-wide account. We try to replicate the EU’s true interregional accounts (the first of the four sets of accounts) via three (3) preferred approaches for estimating subnational accounts: (1) an integrated gravity model, (2) Flegg´s location quotient, and (3) a set of econometrically derived regional purchase coefficients. In the two latter approaches, we allocate excess supply and demand for each region using a gravity model. We ensure final coherence of the interregional accounts via biproportional balancing. For each approach, we evaluate three different scenarios to estimate regional supply and demand: (a) spatially invariant value-added/output by industry, thus using regional employment as a proxy to share out “nationwide” data (b) knowledge of only more-aggregated level regional value-added (i.e., for 11 sectors rather than for 63 industries) and (c) full knowledge of value-added including by component for each industry (in which case, we assume compensation/output only is industry-wise spatially invariant). We then assess the relative accuracy of the different approaches and scenarios.

Michael Wögerer (p)

Fernando de la Torre Cuevas

The land-use sector plays a crucial role in achieving the EU's ambitious carbon neutrality targets. However, local land-use change can bring about many other potentially conflicting implications, including effects on, for example, food security, biodiversity, and ecosystems health. Therefore, understanding and effectively addressing local land-use changes becomes paramount for politicians and policy makers to achieve these goals. This paper presents a Bayesian downscaling approach designed to inform about future high-resolution land use patterns and help taking crucial restoration and protections decisions. The contributions to the existing literature are twofold. First, a novel Bayesian estimation approach for estimating gross land-use changes on a high-resolution grid spanning across Europe is proposed. For the estimation a multinomial logit model is formulated that relates observed land use changes to a rich set of drivers of the land allocation decision. These drivers mainly consist of economic, demographic, biophysical, and policy variables. Second, an efficient downscaling routine to relate the high resolution maps to regional targets is presented. It is based on a bias correction optimization technique to enhance the accuracy and reliability of local predictions. It is used to incorporate more detailed information from other models, which are more suitable to predict the magnitude of country-wide gross land-use changes and additionally allow to add scenario based trajectories (e.g. SSP scenarios). As a result high-resolution land use projections in line with various scenario assumptions can be generated with the proposed method helping to understand the future challenges that local authorities will face.

Marie-Estelle Binet, Jean-Sébastien Pentecôte (p)

Michael Wögerer

Empirical evidence on the Okun’s law between unemployment and economic output is still open issue. As recently surveyed by Ball et al (2017) and Porras-Arena and Martín-Román (2023), there is strong heterogeneity in the estimated Okun’s slope coefficient not only across studies, but also between countries and within their regions. Unemployment reduction from a surge in output tend to vary through time, leading to noticeable asymmetry of its strength over the business cycle.
To this regard, the aim of our empirical analysis is to go one step more in the understanding of what may explain heterogeneity in the estimated elasticities of unemployment to economic activity. We tackle this issue by having a new look at a large set of European NUTS2 regions’ record. We discuss empirical evidence from an annual panel data set of 246 regions from 2000 to 2020. Firstly, we estimate Okun’s slope coefficients by considering a new way of accounting for the “clockwise movements” in countries’ gross domestic product statistics over the business cycle from Eurostat statistics. Another original feature of our econometric approach is to find not only factors that may explain discrepancies of in unemployment gains from rising output, but also what may explain why some negative output-unemployment relationship could be at play in some instances while it might be not in others. To our knowledge, the two-step approach that we develop here is completely new in that field of research. Besides the influence of usual labor market features that were considered in previous studies at the national or regional levels (Buenda Azorin and del Mar Sanchez de la Vega, 2017, Lim et al, 2021, Mazza 2022, among others), we also consider other sources of influence that are often disregarded.
Given our preliminary results, there is supportive evidence of an asymmetric size of the effect of GDP growth acceleration on domestic unemployment: it is nearly twice stronger when economic activity slows down than when it does not. But, as it has been already questioned in the literature, the Okun’s law does not prevail everywhere at any time, it was not found statistically significant in roughly 1 in 4 cases. People’s level of education has an influence on both the existence and the strength of the Okun’s coefficient. Further work is however needed to assess the robustness of our first empirical findings.

Oscar Fernandez (p), Martin Prinz, Wolfgang Brunaer

Jean-Sébastien Pentecôte

This project aims to generate model-based regional indicators of housing prices in Austria and Germany, providing comprehensive insights into real estate market dynamics. We focus on identifying price bubbles—significant over- or under-valuation trends—and forecasting future price movements under various scenarios. Additionally, we develop indicators for interest rate elasticities of housing prices and inverse elasticities of supply. Our methodology is based on sound statistical and econometric tools (such as panel data and instrumental variable regression) to ensure reliable and precise results. These indicators are of high importance for the financial sector, guiding their financial operations and investment strategies.