Regions between challenges and unexpected opportunities

Bernini, Cristina (Ed.); Emili, Silvia (Ed.) Book Regions between challenges and unexpected opportunities Scienze Regionali, No. 61 Provided in Cooperation with: FrancoAngeli Suggested Citation: Bernini, Cristina (Ed.); Emili, Silvia (Ed.) (2021) : Regions between challenges and unexpected opportunities, Scienze Regionali, No. 61, ISBN 978-88-351-2586-0, FrancoAngeli, Milano This Version is available at: https://hdl.handle.net/10419/251142 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by-nc-nd/4.0/ 11390.5 C. BERNINI, S. EMILI (edited by) REGIONS BETWEEN CHALLENGES AND UNEXPECTED OPPORTUNITIES Due to the Covid-19 pandemic, the XLI Annual Scientific Conference held on line in September 2-4, 2020. The Web Conference contributed to motivate the scientific debate on the regional challenges and opportunities in times of crisis. A large number of contributions have investigated the territorial impact of economic shocks and natural disaster and discussed possible trajectories for a sustainable regional development process. The book collects a selection of these contributions, covering different topics on the economic, social, and regional consequences of crises and recovery processes. The first part is dedicated specifically to the impact of the Covid-19 pandemic and to the ability of a territory to react. The challenges of the new pandemic came in addition to the economic and financial crises and the natural and environmental disasters that have occurred in recent decades. The second part then gathers contributions that discuss more broadly the resilience and regional responses to natural and economic shocks. Crises have largely affected the quality of life of citizens and may have compromised sustainable regional growth. To contribute to the discussion of these issues, the third part of the volume collects some studies that aim to analyse in depth the effects of crises in terms of individual and regional wellbeing, and their relationship with sustainability. The last part is dedicated to a discussion and empirical assessment of the role of regional and national policies in supporting recovery and resilience processes for regional development. REGIONS BETWEEN CHALLENGES AND UNEXPECTED OPPORTUNITIES edited by Cristina Bernini, Silvia Emili Cristina Bernini Full Professor of Economic Statistics and Researcher of the Center for Advanced Studies in Tourism at the University of Bologna. Silvia Emili Junior Assistant Professor of Economic Statistics and Researcher of the Center for Advanced Studies in Tourism at the University of Bologna. 61 Associazione italiana di scienze regionali Scienze Regionali FrancoAngeli La passione per le conoscenze ISBN 978-88-351-1692-9 €22,00 (edizione fuori commercio) 11390.5_1390.33 26/07/21 11:08 Pagina 1       Collana dell’Associazione Italiana di Scienze Regionali (AISRe)   L’Associazione Italiana di Scienze Regionali, con sede legale in Milano, è parte della European Regional Science Association (ERSA) e della Regional Science Association International (RSAI).  L’AISRe rappresenta un luogo di confronto tra studiosi di discipline diverse, di ambito accademico e non, uniti dal comune interesse per la conoscenza e la pianificazione dei fenomeni economici e territoriali. L’AISRe promuove la diffusione delle idee sui problemi regionali e, in generale, sui problemi sociali ed economici aventi una dimensione spaziale.  Questa collana presenta monografie e raccolte di saggi, prodotte dagli apporti multidisciplinari per i quali l’AISRe costituisce un punto di confluenza.  Comitato Scientifico della Collana di Scienze Regionali Cristina Bernini, Ron Boschma, Roberta Capello, Patrizia Lattarulo, Donato Iacobucci, Fabio Mazzola, Guido Pellegrini, Andrés Rodríguez-Pose, André Torre  Per il triennio 2019-2022 il Consiglio Direttivo è costituito da: Roberta Capello (Presidente), Cristina Bernini (Segretario), Marusca de Castris (Tesoriere). Consiglieri: Faggian Alessandra, Fregolent Laura, Lattarulo Patrizia, Mariotti Ilaria, Nisticò Rosanna, Omizzolo Andrea, Pellegrini Guido, Perucca Giovanni, Piacentino Davide, Provenzano Vincenzo, Ragazzi Elena, Rota Francesca Silvia, Scalera Domenico. Revisori dei conti: Caloffi Annalisa, Cerisola Silvia, Ciccarelli Carlo. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 Il presente volume è pubblicato in open access, ossia il file dell’intero lavoro è liberamente scaricabile dalla piattaforma FrancoAngeli Open Access (http://bit.ly/francoangeli-oa). FrancoAngeli Open Access è la piattaforma per pubblicare articoli e monografie, rispettando gli standard etici e qualitativi e la messa a disposizione dei contenuti ad accesso aperto. Oltre a garantire il deposito nei maggiori archivi e repository internazionali OA, la sua integrazione con tutto il ricco catalogo di riviste e collane FrancoAngeli massimizza la visibilità, favorisce facilità di ricerca per l’utente e possibilità di impatto per l’autore. Per saperne di più: http://www.francoangeli.it/come_pubblicare/pubblicare_19.asp I lettori che desiderano informarsi sui libri e le riviste da noi pubblicati possono consultare il nostro sito Internet: www.francoangeli.it e iscriversi nella home page al servizio “Informatemi” per ricevere via e-mail le segnalazioni delle novità. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 COPY 15,5X23 1-02-2016 8:56 Pagina 1 REGIONS BETWEEN CHALLENGES AND UNEXPECTED OPPORTUNITIES edited by Cristina Bernini, Silvia Emili Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. This work, and each part thereof, is protected by copyright law and is published in this digital version under the license Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0) By downloading this work, the User accepts all the conditions of the license agreement for the work as stated and set out on the website https://creativecommons.org/licenses/by-nc-nd/4.0 Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 3 Table of Contents Regions Between Challenges and Unexpected Opportunities 5 Cristina Bernini, Silvia Emili PART I – SOCIO-ECONOMIC IMPACT OF COVID-19 Was there a Covid-19 Harvesting Effect in Northern Italy? 15 Augusto Cerqua, Roberta Di Stefano, Marco Letta, Sara Miccoli Regional Impacts of Covid-19 in Europe: The Costs of the New Normality 35 Roberta Capello, Andrea Caragliu Lockdown and Startups Decline in the Italian Regions: the Missed New Employment 51 Marco Pini, Alessandro Rinaldi PART II – NATURAL DISASTER, ECONOMIC SHOCK AND RESILIENCE Exploring “Resiliencies” to the Great Crisis along the Peripherality Gradient in Central-southern Italy 77 Fabiano Compagnucci, Giulia Urso The High-tech Composite Indicator (HTCI). A Tool for Measuring European Regional Disparities Over Crises 97 Simona Brozzoni, Silvia Biffignandi, Matteo Mazziotta Italian NEETs: An Analysis of Determinants Based on the Territorial Districts 115 Giuseppe Cinquegrana, Giovanni De Luca, Paolo Mazzocchi, Claudio Quintano, Antonella Rocca Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 4 PART III – WELL-BEING AND SUSTAINABILITY Regional Well-being and Sustainability: Insights from Italy 133 Giovanni D’Orio, Rosetta Lombardo Analisys of Determinants of Life Satisfaction: Regional Differences 161 Barbara Baldazzi, Rita De Carli, Daniela Lo Castro, Isabella Siciliani, Alessandra Tinto Occupational Insecurity and Health Wellbeing: Does the Impact Change Across Areas? 177 Giulia Cavrini, Evan Tedeschi PART IV – THE ROLE OF POLICIES Regional Policy Out of the Trade-off: Justifications and Current Challenges 201 Ugo Fratesi Cohesion Policies, Labour Productivity, and Employment Rate. Evidence from the Italian Regions 221 Gianluigi Coppola, Sergio Destefanis Redistribution and Risk-sharing Effects of Intergovernmental Transfers: An Empirical Analysis Based on Italian Municipal Data 245 Giampaolo Arachi, Francesco Porcelli, Alberto Zanardi Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 5 Regions Between Challenges and Unexpected Opportunities Cristina Bernini*, Silvia Emili*1 In early January 2020, most European countries were called upon to actively respond to one of the most alarming and disastrous crises of the past hundred years, the Covid-19 pandemic. In the search for a new normality, the Italian Section of the Regional Science Association International – AISRe, confirmed its annual appointment with the XLI Annual Scientific Conference. The health emergency forced the event to be reorganised and switched to an online conference. The Web Conference, held on September 2-4, 2020, gathered a large number of contributions by scholars from different disciplines belonging to the Regional Sciences. These studies significantly contribute to the scientific debate on regional challenges and opportunities in times of crisis. As well as economic and natural crises, special attention is devoted to the investigation and measurement of the pandemic. Globalisation, innovation, productive transformation, economic growth, territorial transformation, disparities, well-being and sustainability are also among the major challenges for regional development in a medium and long-term perspective. Such challenges are significantly related to the relevant territorial and urban characteristics. In this contest, it is essential to identify the factors influencing local capacities to absorb and react to crises and to the socio-economic transformations that they cause. The discussion of conceptual and theoretical frameworks that enable an interpretation of urban, regional and national development can complement the understanding of regional opportunities and the proposal of policy interventions and instruments. The book collects contributions covering different topics on the economic, social, and regional consequences of crises and recovery processes. The first part is dedicated specifically to the impact of the Covid-19 pandemic and to the ability of a territory to react. The challenges of the new pandemic came in addition to the economic and financial crises and the natural and environmental disasters * University of Bologna, Department of Statistical Sciences, Bologna, Italy, e-mail: cristina. [email protected]; [email protected] (corresponding author). Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 12 significant (with and without fund allocation controls) in determining GDP per capita. National funds are basically not significant. When GDP per capita is broken down into GDP per employee and employment rate, EU funds are found to act more strongly upon the latter. The economic crisis following the Covid-19 pandemic has also hit local economies asymmetrically; thus, the increase in their needs and the drop in revenues have put subnational government budgets under strain, with differing impacts across jurisdictions. In this context, the analysis of how fiscal arrangements can absorb the idiosyncratic shocks that hit local economies and thus affect the fiscal position of subnational governments attracts further attention. Giampaolo Arachi, Francesco Porcelli and Alberto Zanardi investigate the role of intergovernmental equalization schemes in providing risk sharing and stabilization across local jurisdictions by means of local budget intervention. The empirical analysis is based on the Italian municipal equalization system that was reformed in 2015 by introducing formula grants to equalize the fiscal gap, updated yearly according to local social-economic factors. The Italian case is particularly interesting because the reform was applied only to municipalities located in the territories of standard regions; while the allocation of grants to municipalities in special regions continued using the previous system based on the equalization of historical expenditure. The results show that formula grants can produce more income redistribution across municipalities than transfers based on historical expenditures. Conversely, the new formula-based transfers continue to have very low contemporary risk-sharing effects. This result critically depends on lags in the data available for evaluating fiscal capacity and standard expenditure needs. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 PART I – SOCIO-ECONOMIC IMPACT OF COVID-19 Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 15 Was there a Covid-19 Harvesting Effect in Northern Italy? Augusto Cerqua*, Roberta Di Stefano°, Marco Letta*,1Sara Miccoli°2 Abstract We investigate the possibility of a harvesting effect, i.e. a temporary forward shift in mortality, associated with the Covid-19 pandemic by looking at the excess mortality trends of an area that registered one of the highest death tolls in the world during the first wave, Northern Italy. We do not find any evidence of a sizable Covid-19 harvesting effect, neither in the summer months after the slowdown of the first wave nor at the beginning of the second wave. According to our estimates, only a minor share of the total excess deaths detected in Northern Italian municipalities over the entire period under scrutiny (February – November 2020) can be attributed to an anticipatory role of Covid-19. A slightly higher share is detected for the most severely affected areas (the provinces of Bergamo and Brescia, in particular), but even in these territories, the harvesting effect can only account for less than 20% of excess deaths. Furthermore, the lower mortality rates observed in these areas at the beginning of the second wave may be due to several factors other than a harvesting effect, including behavioral change and some degree of temporary herd immunity. The very limited presence of short-run mortality displacement restates the case for containment policies aimed at minimizing the health impacts of the pandemic. 1. Introduction At the onset of the Covid-19 pandemic, there was much speculation in Italy as elsewhere about a potential “harvesting effect” of Covid-19, i.e. a short-term increase of mortality later followed by a corresponding decrease in deaths. The claim was that Covid-19 fatalities, whose median age was around 80 years, were, for the vast majority, very vulnerable people who, in the absence of the pandemic, would have “died anyway” shortly after they actually did, due to other causes. According to the proponents of this hypothesis, Covid-19 excess mortality would * Sapienza University of Rome, Department of Social Sciences and Economics, Rome, Italy, e-mail: [email protected]; [email protected]. ° Sapienza University of Rome, Department of Methods and Models for Economics, Territory and Finance, Rome, Italy, e-mail: [email protected]; [email protected] (corresponding author). Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 16 have been largely re-absorbed in the months after the mortality peak, as the virus would have simply anticipated a large number of occurred deaths. Stated differently, this implies that when the spread of Covid-19 progressively slows down, one should observe a significant reduction in mortality, which would counterbalance the abnormal increases experienced during the peak. Albeit this position has been quickly picked up by Covid-19 “skeptics” to protest against the social distancing policies introduced by most governments,1 research on the plausibility of this claim is still scarce as documented in Section 2, both due to lags in data availability and the short time-span observable so far. We investigate the possible harvesting effects of the Covid-19 pandemic by looking at the excess all-cause mortality trends in Northern Italy, one of the areas with the highest Covid-19 death toll in the world. Specifically, we employ the data-driven methodology introduced by Cerqua et al. (2020) to estimate excess mortality and then investigate how it has evolved during three separate periods: i) the peak of the first wave in Italy, February-May 2020; ii) the “summer break”, i.e. the tail of the first wave, going from June to September; and iii) the beginning of the second wave, i.e. October and November 2020. During the “summer break”, a “negative” excess mortality is detected. Nevertheless, this reduction in observed mortality compared to the counterfactual mortality figures predicted by our model is far too limited (corresponding to 16% of the total excess deaths observed during the first wave) to compensate for the abnormally high excess mortality of the first wave. During the onset of the second wave, new excess mortality clusters are detected, and we estimate that, as a consequence, the harvesting effect further shrinks to less than 12% when considering the entire February-November period. Still, we do observe a negative and statistically significant spatial autocorrelation between the mortality patterns of the two waves in some areas (the provinces of Bergamo and Brescia, in particular) where, remarkably, we also detect a “negative” excess mortality not only during the summer months but also in October and November. This is consistent with the well-documented lower incidence of Covid-19 during the beginning of the second wave in these Lombardy provinces, which even led some mayors to ask to be exempted from the November 3, 2020 decree, which imposed a “red zone” in the entire region. While such reversal of patterns between the two waves could be due to several factors, such as a behavioral change and some degree of herd immunity in the most affected areas, it is by no means enough to compensate for the mortality boom of the first wave. Indeed, even in these hardest-hit areas, the harvesting effect can only explain up to 17% of the Covid19 related deaths experienced during the entire period under scrutiny. 1. In Italy, early in the pandemic, there was a fierce debate, which featured heavily in media outlets, about whether Covid-19 victims had died “with coronavirus” or “from coronavirus’. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 17 On top of these period-by-period comparisons, we also compute the cumulative number of deaths over the entire period under scrutiny (February-November 2020), which sums to 49,816 deaths more than “expected” in Northern Italy. This corresponds to an increase in mortality of +20% with respect to an “ordinary” year, i.e. in a “no-Covid” counterfactual scenario. Overall, this evidence suggests that, although Covid-19 has probably anticipated the death of some of the frailest individuals of the Italian population, in the vast majority of cases, it killed relatively healthy people who did not have a short life expectancy before the pandemic’s arrival. 2. The Harvesting Effect The harvesting effect, or mortality displacement, is identified as an increase in deaths followed by fewer deaths than expected after the mortality crisis. During exogenous shocks such as heat waves or cold spells, the selective mortality among the frailest individuals increases the deaths among the total population and leaves a relevant proportion of strong survivors (Luy et al., 2020). After the shock, the number of deaths is below the expected number, and, therefore, a compensation in mortality can be observed between the crisis and the following period (Toulemon, Barbieri 2008). Several scholars studied the harvesting effect caused by particular events, such as heat waves or cold spells (e.g. Baccini et al., 2013; Cheng et al., 2018; Grize et al., 2005; Qiao et al., 2015; Stafoggia et al., 2009; Toulemon, Barbieri, 2008), seasonal influenza (e.g. Lytras et al., 2019) or air pollution (Rabl et al., 2005). Lytras et al. (2019) found out that the influenza A(H1N1)pdm09 affected the frailties individuals that would have died in the short-term because of other causes, while influenza A(H3N2) and type B caused an excess of influenza deaths among people who would not have died in the same year. Stafoggia et al. (2009), studying deaths that occurred in Rome between 1987-2005, figured out that high levels of mortality during winter periods can reduce the effect of heat waves on mortality compared with years of winters with low levels of mortality. To our knowledge, only a few papers assess the potential presence of the harvesting effect during the Covid-19 pandemic. Rivera et al. (2020) stated that in the US, the very high mortality due to Covid-19 spans over a more extended period than other influenza or pandemic, and probably no harvesting would be observed in periods following the worst waves of the Covid-19 pandemic. In fact, in a study that analyzes the 2020 life expectancy decrease in the US, Andrasfay and Goldman (2021) did not find evidence of a harvesting effect due to Covid19. Alicandro et al. (2020) indicated the possible presence of a harvesting effect at the end of the first wave of the pandemic in Italy, except for the Lombardy Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 18 region, where this effect was less pronounced at that time. Similarly, Scortichini et al. (2020), by analyzing excess mortality across Italian provinces, suggested the possible presence of harvesting effect in some areas of Central and Southern Italy at the end of the first wave of the pandemic. The Italian National Institute of Statistics (Istat) and Italian National Institute of Health (ISS) (2020b) reported some evidence of harvesting effect in some areas of Northern Italy during the summer months when the infections were minor. In contrast, the recent study by Canoui-Poitrine et al. (2021), who estimate the number of excess deaths among nursing home residents during the first wave of the pandemic in France, finds no evidence harvesting effects up to the end of August. 3. Data and Methodological Approach To determine the potential presence of a harvesting effect in Northern Italy, we first estimate the excess all-cause mortality due (directly or indirectly) to the Covid-19 pandemic at the municipality-level and then investigate its evolution over time across the three different periods described above. The first step is made necessary by the lack of reliable data on the deaths caused by Covid19, especially at a disaggregated level. Indeed, official data on the death toll of Covid-19 at the local level are scarce,2 and they are likely to suffer from substantial underreporting (Ghislandi et al., 2020). Excess mortality is defined as the difference between the observed mortality in the presence of a pandemic and the counterfactual scenario of mortality in the pandemic’s absence. It includes the number of deaths due directly to Covid-19 infections as well as the deaths due indirectly to Covid-19, i.e. the collateral effects of the lockdown. During the lockdown, the likelihood of dying for road3 and workplace accidents, pollution-related diseases, or criminal activities decreases. At the same time, the likelihood of dying for the stress on the public health system increases. The estimation of excess mortality is made possible thanks to the data released on February 3, 2021, by Istat on the number of daily certified deaths for the period January 1, 2015-November 30, 2020, for all Italian municipalities.4 2. In Italy, official data on SARS-CoV-2 reported cases are released only at the provincial level (the number of infected people) or at the regional level (the number of Covid-19 deaths). 3. The Istat, ACI (2020) report, records a decrease in victims due to road accidents in the period January-September 2020 of 1,788 (-26.3%). The percentage increases to 34% by considering the period January-June 2020. 4. Due to the creation of Mappano as a new administrative unit in 2017 and to the lack of mortality data for all years, we cannot analyze six municipalities: Borgaro Torinese, Caselle Torinese, Leini, Mappano, and Settimo Torinese. Besides, as 2020 is a leap year, we decided to ignore the deaths on February 29 for comparability with data from previous years. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 19 An accurate estimation of excess mortality requires the construction of a reliable counterfactual scenario. In the context of the pandemic mortality estimation, different approaches were used.5 The most common is what we call the “intuitive” approach. It consists of using the simple average of the numbers of deaths observed for the same unit in the past. This approach has been adopted by several national and international institutions and employed in many scientific works. It is a simple approach that does not employ any model, but it may provide excess mortality estimates which are too sensitive to outliers. Another possible approach is the use of the counterfactual approaches, such as the difference-in-differences or the synthetic control method estimators. However, these approaches are ill-suited in a setting where it is hard to find plausible control groups, i.e. municipalities potentially not affected (directly or indirectly via containment measures) by the pandemic for several months. An attractive methodological solution to such an estimation problem is the recently developed Machine Learning Control Method (MLCM) inspired by the train-test-treat-compare process proposed by Varian (2016). In the context of Covid-19, the MLCM can be applied by drawing on the predictive ability of ML algorithms to generate a no-Covid counterfactual scenario for each unit by using exclusively pre-pandemic information (Cerqua, Letta, 2020). In our setting, the use of the MLCM is made possible by constructing a comprehensive time-series cross-sectional database on Italian municipalities. The reason to prefer MLCM over the “intuitive” approach lies in its ability to estimate more accurate counterfactual scenarios. Cerqua et al. (2020) demonstrate that considering the Mean Squared Error (MSE), on average, there is a sizable gain in terms of estimation accuracy compared with the intuitive estimates, especially for small and medium-sized municipalities. For this reason, we investigate the presence of the harvesting effect on Northern Italy by applying the MLCM approach used by Cerqua et al. (2020) to retrieve excess mortality estimates at the municipality-level. The mortality scenario without the pandemic, i.e. the cumulative number of deaths per 10,000 inhabitants in an ordinary situation, is estimated using 16 selected covariates from 2015 to 2019, including the demographic, health system, economic, and contamination (air pollution) features.6 5. See Section 2 of Cerqua et al. (2020) for a review of the methodologies used to estimate excess mortality during pandemics. 6. The full list of covariates is the following: the share of men in the population, the share of those aged 65+ (overall as well as only men), the share of those aged 80+ (overall as well as only men), the resident population, the overall number of deaths in the previous year, the overall number of deaths in the 7 weeks before the Covid-19 outbreak in Italy, the number of employees, the share of employment in manufacturing, the PM-10 as a measure of air quality, the population density, the degree of urbanization of the municipality, the dummy of the presence of a hospital in the Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 20 For each considered period (in our case, the peak of the first wave, the “summer break”, and the beginning of the second wave), we train and test our random forest algorithm on the pooled 2015-2019 (on which, as typical in the ML literature, we apply a random split and use 80% of the full sample as the training set and the remaining 20% as the testing sample) dataset to predict, for the 2020 sample, estimates of local mortality in a counterfactual scenario without the pandemic. It is then easy to retrieve excess mortality as the difference between observed and predicted mortality. Cerqua et al. (2020) use three ML algorithms: Least Absolute Shrinkage and Selection Operator (LASSO), random forest, and stochastic gradient boosting. In this work, we apply the ML using the random forest algorithm, a fully non-linear technique based on the aggregation of many decision trees (1000, in our case), as Cerqua et al. (2020) demonstrate that it performs well for all municipality sizes. The choice to circumscribe the analysis only on Northern Italy is dictated by the fact that it was the epicenter of the pandemic in Italy during the peak of the first wave (February-May 2020) and one of the mortality hotspots in Europe. As such, we deem it a representative case study to test for a potential Covid-19 harvesting effect.7 We investigate how all-cause deaths have evolved during three separate periods: i) the peak of the first wave in Italy, February-May 2020; ii) the “summer break”, i.e. the tail of the first wave, going from June to September; iii) the beginning of the second wave, October and November 2020, according to the division made by the fourth report Istat, ISS (2020a). We do so by showing the choropleth maps of each separate period as well as by using one of the most important indexes for studying spatial relationships: the Moran’s I. Moran’s I can be of two types: the global bivariate Moran’s I and local bivariate Moran’s I (bivariate Local Indicators of Spatial Association, or more simply bivariate LISA). The former provides summary statistics for overall spatial clustering. It varies between +1 and –1: a value close to +1 indicates a strong positive spatial autocorrelation. Otherwise, a value close to –1 reveals that the spatial autocorrelation is negative, while 0 indicates a random spatial pattern. The bivariate LISA is instead applied to depict the spatiality of how the value of one variable is surrounded by values of a second variable (Anselin, 1995). Basically, the bivariate LISA measures the relationships between spatial units and their neighboring spatial units and maps statistically significant clusters of the phenomena under analysis. The neighboring structure across municipalities is measured by a spatial weights matrix based on the inverse geographical municipality, the dummy of the presence of a hospital in at least one of the neighboring municipalities, and the number of deaths due to road accidents in the previous year. For more details, see Cerqua et al. (2020). 7. On the contrary, the excess mortality observed during the peak of the first wave in Central and Southern Italy could be too mild and uneven to determine a harvesting effect. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 21 (Euclidean) distances between municipalities” centroids.8 The weight matrix is then standardized such that its rows sum to unity (in order to compute neighborhood averages) and have zeros along the leading diagonal (see Maddison, 2006). Thanks to the bivariate LISA, we will identify the following types of association: positive autocorrelation, which occurs where high values of variable 1 are surrounded by high values of variable 2 (High-High hotspots, HH) or where there is a concentration of low values (Low-Low coldspots, LL); or negative spatial autocorrelation, namely places where low values of variable 2 surround high values of variable 1 (High-Low clusters, HL), or vice versa (Low-High clusters, LH). As in Frigerio et al. (2015), we will use 999 random permutations to determine the statistical significance for each cluster. In our analysis, we will use the global bivariate Moran’s I to study the overall spatial correlation of excess mortality values of the first wave on the “summer break” (second wave) in Northern Italy and the bivariate LISA to measure the clustering patterns of excess mortality values of the first wave and the “summer break” (second wave). We will investigate whether the patterns of similarity and dissimilarity in the clustering of excess mortality values remained stable across the three time periods. 4. Results 4.1. Excess Mortality Estimates The excess mortality estimates from all-cause deaths relative to the first phase of the pandemic, the so-called peak of the first wave from February 21 to May 31, are shown in Figure 1 for Northern Italy. Compared with the counterfactual scenario, the municipalities with the highest excess mortality are located in the provinces of Bergamo, Brescia, Cremona, Lodi in the Lombardy region. Quite impressively, 40.9% of the Lombardy municipalities recorded excess mortality of over 100%. Wide clusters of municipalities with excess mortality above 100% are also present in Piacenza and Parma provinces in the Emilia Romagna region and the Lombardy region. Clusters of municipalities with an excess of deaths over 50% are located in Milan, Mantova and Pavia (Lombardy), again in Piacenza and Parma provinces (Emilia Romagna), but also in the provinces of Imperia (Liguria), Cuneo and Alessandria (Piedmont), and Trento (Trentino Alto-Adige). In many municipalities of the Liguria region and the provinces of Turin (Piedmont), Reggio Emilia, Rimini, and Forlì-Cesena (Emilia Romagna), the excess mortality is between 20% and 50%. During the first wave, 126,896 8. The distance threshold is 15.1 km, which is the minimum threshold in order to avoid neighborless municipalities. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 28 Moran’s I statistic stays negative and low (–0.091). We then use the bivariate LISA to identify clusters of the excess mortality values of the first and second waves and report them in Figure 5. Two relevant patterns emerge: i) some areas which were only moderately hit during the first wave experienced high levels of excess mortality in October and November. In particular, these areas are concentrated in the provinces of Varese, Como, and Milan in Lombardy, Belluno in Veneto, Udine in Friuli-Venezia Giulia, Cuneo and Biella in Piedmont; ii) the municipalities surrounding Bergamo and Brescia, the most harshly hit during the first wave in Italy, exhibit low levels of excess deaths at the beginning of the second wave. Overall, while our examination of overall trends in excess deaths for Northern Italy suggests very limited evidence of relevant harvesting effects, our spatial analysis gives compelling evidence that the areas of Northern Italy which were hit the hardest in the first phase of Covid-19, then experienced a decrease in the number of deaths of a larger magnitude, and over a longer time-span, with respect to the majority of the other Northern Italy municipalities. Let us, therefore, take a closer look at excess mortality dynamics in these most affected areas. By focusing on the provinces of Bergamo and Brescia, in the first wave, we observed excess mortality of +164% (16,754 individuals died in front of an “expected” number of deaths of 6,351), during the “summer break”, a drop in the number of deaths of –17.2% (6,579 individuals died in front of an “expected” number of deaths of 7,948) and at the onset of the second wave a drop in the number of deaths of –5.1% (3,936 individuals died in front of an “expected” number of deaths of 4,147). Remarkably, in these areas, the all-cause mortality balance sign is negative even at the beginning of the second wave. This evidence suggests a somewhat more pronounced harvesting effect in the most affected areas of Northern Italy during the first wave. Of the 10,403 excess deaths that occurred during the first wave in these areas, we estimate that 1,580 individuals would have died anyway by the end of November 2020. However, this still means that over 83% of the deaths due (directly or indirectly) to Covid-19 concern relatively healthy people that did not have a short life expectancy before the pandemic’s arrival. 5. Conclusions By studying mortality dynamics in the immediate aftermath of the first Covid19 wave in Northern Italy, one of the hardest-hit territories of the world, we find only limited evidence of a Covid-19 harvesting effect. The impressive Covid-19 first wave excess mortality in Northern Italian municipalities was only marginally “compensated for” by a subsequent decline in mortality. In line with Canoui-Poitrine et al. (2021) findings for nursing home residents in France, we Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 29 Figure 5 – Bivariate LISA of excess mortality values of the peak of first wave and the onset of the second wave Panel a – Cluster map Panel b – Cluster significance Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 30 do not find that Covid-19 only affected those whose health was already inevitably compromised. The vast majority of Covid-19 deaths are not “anticipated” deaths but sudden and “unexpected” ones. We document a slight reduction in total mortality during the summer months and new excess mortality clusters at the beginning of the second wave. When considering these dynamics jointly, for Northern Italy as a whole, the harvesting effect can account only for a minor share of the total excess deaths detected over the entire period. We also detect a statistically significant and negative spatial autocorrelation between the mortality trend of the first wave and that of the second, and a negative mortality balance at the beginning of the second wave, in some territories such as the provinces of Bergamo and Brescia. In these areas, the most severely affected ones during the first wave, less than 20% of the Covid-19-related deaths might have occurred anyway by the end of November 2020. However, these inverse dynamics are likely the joint outcome of a combination of causal factors, such as some degree of temporary herd immunity coupled with long-lasting behavioral consequences of the pandemic, rather than an exclusive outcome of the harvesting effect. In this respect, the recent re-explosion of cases and hospitalizations in the area of Brescia in the second half of February 2021, which led to the rapid imposition of ad hoc more severe restrictive measures, is a telltale sign that Covid-19 did not exhaust its impetus with the first wave in these territories. Finally, excess mortality estimates computed over the entire February-November 2020 period confirm that subsequent reductions did not counterbalance the initial boom in Northern Italy mortality. Indeed, total excess mortality over this time-span is still 20% above what would have happened under “ordinary” conditions. Two caveats are in order regarding the credibility of our findings. While Covid-19 incidence was extremely low throughout the summer in Italy, including Northern regions, there is a possibility that many Covid-19 survivors from the first wave may have been fatally weakened by the virus and died several months later (Canoui-Poitrine et al., 2021). We acknowledge that this mechanism may be at play, but at the same time, we do not deem it to be so substantial to significantly alter the overall mortality trend, let alone reverse the sign of the excess mortality detected. Second, we only focus on the very short-run. Even though the harvesting effect is intrinsically a short-run phenomenon, the few months for which we have data may not be sufficient for the reabsorption to arise, and mortality displacement could take place over a longer time span. Still, the very limited presence of Covid-19 induced mortality displacement in the short-run makes the health costs of the pandemic even more dramatic, suggests that Covid-19 can significantly shorten life expectancy, and restates once more the case for containment policies aimed at minimizing as much as possible the sanitary emergency and the death toll of the pandemic. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 31 Our evidence is indeed preliminary. We look at a circumscribed area, Northern Italy, and focus on the very short-run due to current data availability. Further research should extend this type of analysis to other parts of Italy, other countries, and other waves of the current pandemic. When Covid-19 is eventually brought under control, it will be possible to provide a definitive answer on whether the pandemic played a significant anticipatory role and triggered a substantial mortality displacement or not. For the moment, the answer seems to be no. References Alicandro G., Remuzzi G., La Vecchia C. (2020), Italy’s first wave of the Covid-19 pandemic has ended: no excess mortality in May, 2020. The Lancet, 396, 10253: E27-E28. Doi: 10.1016/S0140-6736(20)31865-1 . Andrasfay T., Goldman N. (2021), Reductions in 2020 US life expectancy due to COVID-19 and the disproportionate impact on the Black and Latino populations. Proceedings of the National Academy of Sciences, 118, 5: e2014746118. Doi: 10.1073/pnas.2014746118. Anselin L. (1995), Local indicators of spatial association – LISA. Geographical Analysis, 27, 2: 93-115. Doi: 10.1111/j.1538-4632.1995.tb00338.x. Baccini M., Kosatsky T., Biggeri A. (2013), Impact of summer heat on urban population mortality in Europe during the 1990s: An evaluation of years of life lost adjusted for harvesting. PLoS One, 8, 7: E69638. Doi: 10.1371/journal.pone.0069638. Canoui-Poitrine F., Rachas A., Thomas M., Carcaillon-Bentata L., Fontaine R., Gavazzi G., Laurent M., Robine J.M. (2021), Magnitude, change over time, demographic characteristics and geographic distribution of excess deaths among nursing home residents during the first wave of COVID-19 in France: a nationwide cohort study. Medrxiv, Doi: 10.1101/2021.01.09.20248472. Cerqua A., Di Stefano R., Letta M., Miccoli S. (2020), Local estimates during the Covid19 pandemic in Italy. L’Aquila: Gran Sasso Science Institute. GSSI Discussion Paper Series in Regional Science & Economic Geography n. 2020-06. Cerqua A., Letta M. (2020), Local economies amidst the Covid-19 crisis in Italy: A tale of diverging trajectories. Covid Economics, 60: 142-171. Cheng J., Xu Z., Bambrick H., Su H., Tong S., Hu W. (2018), Heatwave and elderly mortality: An evaluation of death burden and health costs considering short-term mortality displacement. Environment International, 115: 334-342. Doi: 10.1016/j.envint.2018.03.041. Frigerio I., Carnelli F., Cabinio M., De Amicis M. (2018), Spatiotemporal pattern of social vulnerability in Italy. International Journal of Disaster Risk Science, 9, 2: 249-262. Doi: 10.1007/s13753-018-0168-7. Ghislandi S., Muttarak R., Sauerberg M., Scotti B. (2020), News from the front: Estimation of excess mortality and life expectancy in the major epicenters of the Covid-19 pandemic in Italy. Medrxiv. Doi: 10.1101/2020.04.29.20084335. Grize L., Huss A., Thommen O., Schindler C., Braun-Fahrlander C. (2005), Heat wave 2003 and mortality in Switzerland. Swiss Medical Weekly, 135, 13-14: 200-205. Doi: 10.4414/smw.2005.11009. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 32 Istat, ACI (2020), Incidenti stradali. Stima preliminare. Gennaio-settembre 2020. Roma: www.istat.it. Istat, ISS (2020a), Decessi per il complesso delle cause. Periodo gennaio-novembre 2020. Roma: www.istat.it. Istat, ISS (2020b), Impatto dell’epidemia Covid-19 sulla mortalità totale della popolazione residente periodo gennaio-maggio 2020. Rome: www.istat.it. Luy M., Di Giulio P., Di Lego V., Lazarevič P., Sauerberg M. (2020), Life expectancy: frequently used, but hardly understood. Gerontology, 66, 1: 95-104. Doi: 10.1159/000500955. Lytras T., Pantavou K., Mouratidou E., Tsiodras S. (2019), Mortality attributable to seasonal influenza in Greece, 2013 to 2017: variation by type/subtype and age, and a possible harvesting effect. Eurosurveillance, 24, 14: 1800118. Doi: 10.2807/1560-7917.ES.2019.24.14.1800118. Maddison D. (2006), Environmental Kuznets curves: A spatial econometric approach. Journal of Environmental Economics and Management, 51, 2: 218-230. Doi: 10.1016/j.jeem.2005.07.002. Qiao Z., Guo Y., Yu W., Tong S. (2015), Assessment of short-and long-term mortality displacement in heat-related deaths in Brisbane, Australia, 1996-2004. Environmental Health Perspectives, 123, 8: 766-772. Doi: 10.1289/ehp.1307606. Rabl A., Spadaro J.V., van der Zwaan B. (2005), Uncertainty of Air Pollution Cost Estimates: To What Extent Does It Matter? Environmental Science & Technology, 39, 2: 399-408. Doi: 10.1021/es049189v. Rivera R., Rosenbaum J.E., Quispe W. (2020), Excess mortality in the United States during the first three months of the Covid-19 pandemic. Epidemiology & Infection, 148. Doi: 10.1017/S0950268820002617. Scortichini M., Dos Santos R.S., De’Donato F., De Sario M., Michelozzi P., Davoli M., ... Gasparrini A. (2020), Excess mortality during the Covid-19 outbreak in Italy: a two-stage interrupted time-series analysis. International Journal of Epidemiology, 49, 6: 1909-1917. Doi: 10.1093/ije/dyaa169. Stafoggia M., Forastiere F., Michelozzi P., Perucci C.A. (2009), Summer temperature-related mortality: effect modification by previous winter mortality. Epidemiology: 20, 4: 575-583. Doi: 10.1097/EDE.0b013e31819ecdf0. Toulemon L., Barbieri M. (2008), The mortality impact of the August 2003 heat wave in France: investigating the “harvesting” effect and other long-term consequences. Population Studies, 62, 1: 39-53. Doi: 10.1080/00324720701804249. Varian H.R. (2016), Causal Inference in Economics and Marketing. Proceedings of the National Academy of Sciences, 113, 27: 7310-7315. Doi: 10.1073/pnas.1510479113. Sommario Covid-19 nel Nord Italia: c’è stato un effetto harvesting? Questo lavoro testa l’ipotesi di un effetto “harvesting” (mietitura) associato alla pandemia di COVID-19, ossia un aumento temporaneo della mortalità seguito da una diminuzione della stessa, attraverso un’analisi dell’andamento degli eccessi di mortalità nel Nord Italia, un’area che ha riportato tassi di mortalità tra i più alti al mondo durante la prima ondata. Non si rileva alcuna evidenza empirica di un consistente effetto harvesting Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 33 nei comuni del Nord Italia, né durante la fase estiva di rallentamento della prima ondata, né all’inizio della seconda ondata. Le stime suggeriscono che solo una piccola percentuale della mortalità in eccesso rilevata durante il periodo in esame (Febbraio-Novembre 2020) può essere attribuita ad un’anticipazione dei decessi causata dal COVID-19. Tale quota è più alta nei territori colpiti in modo più duro (in particolare nelle province di Bergamo e Brescia), ma anche in queste aree l’effetto harvesting ammonta a meno del 20% del totale delle morti in eccesso. Inoltre, la minore mortalità registrata in queste zone all’inizio della seconda ondata potrebbe essere dovuta anche ad altri fattori causali, quali cambiamenti comportamentali o una parziale e temporanea immunità di gregge. L’assenza di supporto empirico a favore della tesi secondo cui gran parte dei decessi da COVID-19 sarebbero comunque avvenuti nel breve periodo ribadisce l’importanza di politiche di contenimento volte a minimizzare l’impatto della pandemia sulla salute della popolazione. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 35 Regional Impacts of Covid-19 in Europe: The Costs of the New Normality Roberta Capello*, Andrea Caragliu* Abstract This paper discusses the effects of the Covid-19 pandemic on growth of European regions. The impact is measured as a difference between a “New Normality” scenario, imposed by Covid, for the period 2021-2030 and a Reference scenario, whereby Covid19 did not take place. Scenarios are obtained through the MAcroeconomic, Sectoral, Social, Territorial (MASST4), built by the authors, and able to generate regional growth scenarios for regions (NUTS2) in EU member states (UK included) on the basis of the interaction bewteen macroeconomic elements and local specificities. Some EU Countries and regions will actually be capable of bouncing back and show remarkable resilience. Other regions, instead, register a high cost in terms of missed growth. 1. Introduction1 The recent and largely unexpected pandemic of Corona-19 virus has caught healthcare systems all over the world unprepared, thus exerting a dramatic toll in terms of both casualties as well as in terms of missed economic performance, mostly because of the lockdown measures enacted in many Countries to prevent the diffusion of the contagion. While countless attempts at gauging the extent of the slump caused by the pandemic have been made over the past few months, the absence of reliable real-time economic statistics and the limited availability of regional macroeconometric growth models have to date yielded scarce evidence on the regional extent of the potential economic losses engendered by the Covid-19 pandemic. Besides, insufficient information available for short-run costs makes it difficult * Politecnico di Milano, ABC Department, Milan, Italy. e-mail: [email protected]; an- [email protected] (corresponding author). 1. The Authors would like to thank Camilla Lenzi for suggestions on the presentation of results, and Chiara Del Bo for comments to an earlier version of this work. The usual disclaims matter. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 36 to foresee the likely future development paths of European regions in the aftermath of the pandemic. This paper fills this gap with the use of the fourth version of the MAcroeconomic, Sectoral, Social, Territorial (MASST4; Capello, Caragliu, 2021a) model to build scenarios for 2021-2030, since a longer simulation period would not be credible, given the substantial degree of instability of the overall situation in these difficult times. The MASST4 model merges two conceptual streams by linking regional growth determinants and macroeconomic growth elements. In order to foresee the impacts of long run regional economic development patterns for European regions, a New Normality scenario, first developed in Capello and Caragliu (2020b), is here presented. On the basis of the short-run costs of the pandemic as happening in Spring 2020, the New Normality scenario produces the regional growth rates out of the economic contraction for the period 2021-2030. The long term impact of the Covid-19 pandemic is measured as the missed growth of the New Normality scenario with respect to a Reference one, whereby Covid-19 did not take place. This offers the unique chance of highlighting the counterfactual nature of the pandemic. The achievement of this goal is not an easy task. Two long term scenarios have to be built, one of which based on short term estimates of the pandemic, which have to be estmated. In the paper, we proceed as follows. In Section 2 we present a concise description of the MASST4 model, used to derive the simulated regional economic growth rates for both scenarios. The scenario construction methodology is presented in Section 3. Section 4 illustrates national and regional results for the New Normality scenario, against the backdrop of results obtained simulating the Reference scenario. Finally, Section 5 concludes and derives a few policy implications. 2. The MASST4 Model Results presented in this paper are built through a process of simulation based on the MASST model in its fourth version. While the reader is referred to Capello and Caragliu (2021a) for a more thorough description of the latest generation of the model, it is here worth briefly recapping how the model works. In order to generate future growth rates, the MASST model first estimates structural relations among exogenous and endogenous variables; in the second stage, the equation parameters identified through econometric estimates are used to calculate predicted values for the dependent variables, with exogenous variables set to previously predetermined targets. Target values for exogenous parameters are set according to internally coherent mix of assumptions of possible future combinations of context conditions that depict specific scenarios; this approach has been termed quantitative foresight (Capello, Caragliu, 2016). Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 37 In the MASST model, a national and a regional sub-model co-exist, both contributing to the simulation of future regional differential shifts, i.e. the deviations of regional GDP growth rates from their national means (Equation 1). [1] In Equation [1], r indicates each of the 276 NUTS2 region in our sample, n represents the 27 EU Countries, while s stands for the regional differential shift. The MASST model is simultaneously generative and distributive. It is a generative growth model, in that regional growth is interpreted mainly as a competitive process (Richardson, 1973). In this class of models, regional growth is seen as a “zero-sum allocation and distribution of production” (Harris, 2011, p. 914), and a region’s growth takes place at the expense of another’s (Richardson, 1978, p. 145). In the MASST4 model, the economic performance of a region depends mainly on its institutional context, i.e. on the national performance. Institutional features, organizational quality, and competitiveness in international trade influence regional economic performance; in the MASST4 model, the global economy acts as a trigger to regional economic performance through the increase in the demand for Country’s products, within a classical Keynesian aggregate demand setting.2 The MASST model is also distributive; national growth rates are distributed to single regions depending on their factor endowments, which explain regional differential shifts (Garcilazo, Oliveira-Martins, 2015). In this sense, regional differential performance is mostly a supply-side mechanism, with both tangible (accessibility; regional policy expenditure; energy efficiency) and intangible (trust; human capital; quality of governance) assets making regions more competitive with respect to the Country mean. In the long run, exogenous variables tend instead to reach predetermined targets whose value is set depending on each scenario’s underlying assumptions. In its 4th version, the MASST model has been strengthened in many ways. The MASST4 has been reinforced in the macroeconomic part, measuring the macroeconomic changes in the period of post crisis, the regional part, inserting an endogenous productivity influenced by the 4th industrial revolution, and its urban part as well. For the last one, it now contains the role of city dynamics in stimulating national economies through their endowment of hosted functions, the quality of local governance (Peirò-Palomino et al. 2020), and the capacity to cooperate through quality long-distance scientific networks (Capello, Caragliu, 2018). A final important remark on the MASST4 model is related to the relevant effort in building a comprehensive data base covering the universe of EU NUTS2 regions. In the 2013 version, these comprise 276 administrative units, with a 2. For a historic review of the different versions of the model, see Capello and Caragliu (2020b). ,, rn Y Y sr∆=∆+ ∈  Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 44 of the Reference scenario, where the COVD-19 would have not taken place. Table 3 shows the difference in the average annual GDP growth rates between 2017 and 2030 for all EU28 Countries obtained in the New normality scenario with respect to the Reference scenario. Reconnecting to the question concluding Section 3, Table 3 shows a rather complex picture, with some of the countries hit the hardest from the immediate costs of the pandemic being actually capable of recovering faster in its aftermath. This is in particular the case of France, Italy, Belgium, and Spain. Another outcome shown in Table 3 refers to countries whose economic growth would be faster in the case of the New Normality scenario, with however a smaller difference with respect to the Reference case. This is typical of Countries that initially faced lower costs from the Spring lockdowns (e.g. Germany). A third typology of Countries shown in Table 3 encompasses those whose GDP growth substantially benefits from additional investment spurred by the EU plan devised to counterbalance the negative economic impact of Covid19, or whose initial costs incurred in Spring 2020 have been somewhat lower. These include mostly Central and Eastern European Countries, such as Romania, Estonia, Bulgaria, and Slovakia, although this does not uniformly applies. Poland, for instance, has exactly the same GDP growth rate forecasted in the two scenarios. Lastly, Table 3 suggests that some Countries will not fully counterbalance the major slump taking place in 2020, ultimately being damaged by the costs of Covid-19 more than recovery measures will be able to amend. This is the case of Austria, Croatia, and Finland. Moving to the regional set of results, Figure 1 shows the map of average annual GDP growth rates in European regions between 2020 and 2030 as a difference between the New Normality and the Reference scenarios. In Figure 1, colors are represented with darker red when the difference between the New Normality and the Reference scenarios are larger, while increasingly smaller differences are represented with increasingly more intense green shades. Not only does this map display remarkable spatial heterogeneity, as indirectly implied also by national results shown in Table 3. Also, within the same country regions present a rather substantial degree of within countries differences. For instance, this is the case of several areas (marked in dark red, i.e. regions incurring the highest long run costs due to the Covid-19 pandemic) located in peripheral regions in France, Italy, Spain, and Portugal, whose country performance will benefit from the bounce back logically following the initial slump, but whose economic growth will lack. In these Countries, other regions (e.g. Champagne-Ardenne in France, Emilia-Romagna in Italy, Galicia in Spain) will compensate for losses mostly concentrated in other peripheral and rural areas. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 45 Table 3 – Differences in Average Annual National GDP Growth Rates in the New Normality and in the Reference Scenarios, 2020-2030 Country Differential GDP growth rate (new normality vs reference) Austria -0.03 Belgium 0.12 Bulgaria 0.75 Croatia -0.39 Cyprus 0.53 Czech Republic 0.17 Denmark 0.12 Estonia 0.62 Finland -0.13 France 0.16 Germany 0.06 Greece 0.01 Hungary 0.20 Ireland 0.13 Italy 0.18 Latvia 0.57 Lithuania 0.55 Luxembourg 0.39 Malta 0.77 Netherlands 0.23 Poland 0.00 Portugal 0.04 Romania 0.42 Slovakia 0.31 Slovenia 0.10 Spain 0.12 Sweden 0.13 United Kingdom 0.00 Source: Authors’ elaboration on the basis of MASST4 simulations Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 46 The British situation shows all its drama, registering mostly all regions in the country paying a high cost due to the pandemic; especially Scotland and the rich South pay the highest cost. While in general losses do tend to be highest in rural and non-core regions, some major urban areas show significant long-run losses, despite facing initially lower health costs. This is for instance the case of the Lisbon area in Portugal, and Attiki (with the capital city Athens) in Greece. And Ile de France with the capital city Paris in France. The causes behind the positive rebound that drives regions to a higher GDP growth the respect to a reference scenario are namely: • urban areas with respect to rural ones (p-value of the t-test for mean differences equal to 0.12); this weakly suggests that urban areas basically do not lose from the New Normality scenario; Figure 1 – Differences in Average Annual Regional GDP Growth Rates in the New Normality and Reference Scenarios, 2020-2030 Source: Authors’ elaboration on the basis of MASST4 simulations Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 47 Figure 2 – Differences in Total, between and within Countries Theil Indices, in the New Normality and Reference Scenarios, 2020-2030 -0,005 0,000 0,005 0,010 0,015 0,020 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 Between countries index Total Theil index Within countries index Source: Authors’ elaboration on the basis of MASST4 simulations • quality of government (Charron et al., 2019) (p-value of the t-test for mean differences equal to 0.13), which confirms the importance of good formal and informal institutions for the efficient spending of the Recovery fund; • presence of high-tech firms and industries (Simonen et al., 2015) (p-value of the t-test for mean differences equal to 0.11), getting all advantages from the digital technologies, fundamental to do business, to entertain people and to teach during the pandemic and moving towards a 4th technological transformation of the society. A last set of analyses has been performed for verifying whether the New Normality scenario will have any effect on regional disparities. This is done by calculating a Theil index of regional inequalities, which is amenable to a useful decomposition of total disparities (green line in Figure 2) into inter-national disparities (Between Countries Index; orange line in Figure 2) and intra-national disparities (grey line in Figure 2). The Theil Index of Regional inequalities is calculated as follows (Equation 2): [2] 1 1N ii i yy Theil ln N yy =  =   ∑ Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 48 where N is the number of regions, yi is the variable of interest in the ith region (in this case, regional GDP) and is the average regional GDP calculated for all regions (OECD, 2016). Figure 2 presents the difference in the regional disparities between the two scenarios. Being the total disparity line (continuous line) always above zero (also in the last year), the first important result is that the Covid-19 has substantially generated an increase in disparities that remain over time. Moreover, between country disparities are greater in the New Normality w.r.t. the reference, in that the between country line (dashed line) is above the total disparity line, witnessing that the Covid-19 pandemic hit differently the different countries, but that the difference decreases with time. The within country disparities (dotted line) are lower in the New Normality than in the Reference, witnessing that within each country the costs of the New Normality are spatially diffused, and remain constant over time. 5. Conclusions and Policy Implications This paper presents the results of the costs of a New Normality scenario, measured as the costs of a scenario with Covid-19 and one without. Results show that, despite substantial short-run costs of the Covid-19 pandemic, in the long run European Countries and regions will not necessarily lose from the massive negative exogenous shock just happening as we write these conclusions. Some EU Countries and regions will actually be capable of bouncing back and show remarkable resilience. While further research is definitely called upon to understand the microfoundations of these effects, the two most likely causes for such resilience can be traced to the robust injection of EU money (totaling EUR 1.82 trillion for the 2020-2027 period), meant to sustain the rebound of European economies, and the reaction of European manufacturing to the further diffusion of ICT as means of long-distance communication and boosting productivity. However, our findings also hint at two sources of relevant costs. On the one hand, we do identify some net losers even after taking the two above-mentioned positive factors into account. On the other hand, spatial heterogeneity in the short-run and long-run impacts of the healthcare emergency will also cause a substantial increase in (in the short-run) international and (in the long run) intranational disparities. For both sources of costs, policymakers may want to further analyze their causes, and find suitable remedies. Policies dealing with these costs will be sorely needed not only for reasons of equity, but also to increase overall efficiency. It is in fact difficult to accept leaving countries and regions behind; the laggards are typically areas most exposed to the costs of the pandemic either because of their demographic structure, or also because of structural limitations of their healthcare systems or industrial Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 49 structure. However, it is also important to stress that by fostering a higher rebound than a GDP growth obtained in a situation without Covid-19, an important role is played by the quality of governance, which guarantees an efficient way of spending the extra budget made available by the Recovery Plan. References Asheim B. (2012), The changing role of learning regions in the globalizing knowledge economy: A theoretical re-examination. Regional Studies, 46, 8: 993-1004. Doi: 10.1080/00343404.2011.607805. Brynjolfsson E., McAfee A. (2014), The Second Machine Age: Work, Progress and Prosperity in a Time of Brilliant Technologies. London: W. W. Norton & Company. Cacciapaglia G., Cot C., Sannino F. (2020), Second wave Covid-19 pandemics in Europe: a temporal playbook. Scientific Reports, 10, 15514. Doi: 10.1038/s41598-020-72611-5. Capello R., Caragliu A. (2016), After crisis scenarios for Europe: alternative evolutions of structural adjustments. Cambridge Journal of Regions, Economy and Society, 9, 1: 81-101. Doi: 10.1093/cjres/rsv023. Capello R., Caragliu A. (2018), Proximities and the intensity of scientific relations: synergies and nonlinearities. International Regional Science Review, 41, 1: 7-44. Doi: 10.1177/0160017615626985. Capello R., Caragliu A. (2020a), Regional growth and disparities in a post-Covid Europe: A new normality scenario. Unpublished manuscript. Capello R., Caragliu A. (2020b), Modelling and Forecasting Regional Growth: The MASST Model. In: Colombo S. (ed.), Spatial Economics, Vol. 2. Cham (CH): Palgrave, McMillan. 63-89. Doi: 10.1007/978-3-030-40094-1_3. Capello R., Caragliu A. (2021a), Merging macroeconomic and territorial determinants of regional growth: the MASST4 model. The Annals of Regional Science, 66: 19-56. Doi: 10.1007/s00168-020-01007-0. Capello R., Caragliu A. (2021b), The Cost of Missed EU Integration. In: Suzuki S., Patuelli R. (eds.), A Broad View of Regional Science: Essays in Honor of Peter Nijkamp. Berlon (DE): Springer Verlag. 1-23. Doi: 10.1007/978-981-33-4098-5_1. Capello R., Lenzi C. (2018), The dynamics of regional learning paradigms and trajectories. Journal of Evolutionary Economics, 28: 727-748. Doi: 10.1007/s00191-018-0565-5. Charron N., Lapuente V., Annoni P. (2019), Measuring quality of government in EU regions across space and time. Papers in Regional Science, 98, 5: 1925-1953. Doi: 10.1111/pirs.12437. EC – European Commission (2018), Analysis of the budget implementation of the European Structural and Investment Funds in 2017 – Last access on Jan. 2021 https:// ec.europa.eu. Garcilazo E., Oliveira-Martins J. (2015), The contribution of regions to aggregate growth in the OECD. Economic Geography, 91, 2: 205-221. Doi: 10.1111/ecge.12087. Harris R. (2011), Models of regional growth: past, present and future. Journal of Economic Surveys, 25, 5: 913-951. Doi: 10.1111/j.1467-6419.2010.00630.x. Lee J., Bagheri B., Kao H.A. (2015), A cyber-physical systems architecture for industry 4.0-based manufacturing systems. Manufacturing Letters, 3: 18-23. Doi: 10.1016/j. mfglet.2014.12.001. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 50 McCann P. (2020), Perceptions of regional inequality and the geography of discontent: Insights from the UK. Regional Studies, 54, 2: 256-267. Doi: 10.1080/00343404.2019.1619928. OECD (2016), OECD Regions at a glance – Appendix C: Indexes and estimation techniques. Last access on Jan. 2021 https://www.oecd-ilibrary.org. Peiró-Palomino J., Picazo-Tadeo A.J., Rios V. (2020), Well-being in European regions: Does government quality matter? Papers in Regional Science, 99, 3: 555-582. Doi: 10.1111/pirs.12494. Richardson H.W. (1973), Regional growth theory. London: Macmillan. Doi: 10.1007/9781-349-01748-5. Richardson H.W. (1978), Regional and Urban Economics. Harmondsworth: Penguin Books. Simonen J., Svento R., Juutinen A. (2015), Specialization and diversity as drivers of economic growth: Evidence from High-Tech industries. Papers in Regional Science, 94, 2: 229-247. Doi: 10.1111/pirs.12062. Schwab K. (2017), The Fourth Industrial Revolution. New York: Crown Business. Wink R., Kirchner L., Koch F., Speda D. (2016), There are many roads to reindustrialization and resilience: Place-based approaches in three German urban regions. European Planning Studies, 24, 3: 463-488. Doi: 10.1080/09654313.2015.1046370. Sommario Impatti regionali del Covid-19 in Europa: i costi della Nuova Normalità In questo articolo viene presentato l’impatto di lungo periodo della pandemia da Covid-19 sulla crescita delle regioni Europee. L’impatto è calcolato come differenza tra uno scenario di Nuova Normalità, imposto dal Covid, per il periodo 2021-2030 rispetto a uno scenario di Reference, ottenuto nell’ipotesi che la pandemia non avvenisse. Gli scenari sono costruiti grazie al modello MAcroeconomic, Sectoral, Social, Territorial (MASST4), costruito dagli autori, e in grado di creare scenari di crescita regionale per tutte le NUTS2 dei paesi membri dell’Unione Europea (UK inclusa) sulla base di un’interazione tra elementi macroeconomici e specificità locali. I risultati mostrano come alcune aree e paesi siano in grado di riprendersi dalla crisi Covid-19 e superare in dieci anni il tasso di crescita che avrebbero avuto senza pandemia. Altre, invece, registrano alti costi dovuti a una mancata crescita. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 51 Lockdown and Startups Decline in the Italian Regions: the Missed New Employment Marco Pini*, Alessandro Rinaldi* Abstract The aim of this paper is measuring the effect of the lockdown on the startups decline and the consequences in terms of the missed new employment opportunities. We study the case of Italy through an analysis at the regional level. We found that during the two months of lockdown (March-April 2020), new business applications fell by 45.1% compared to the same period of the previous year, with greater reductions in the northern regions. In the face of this startups decline, we estimated that 30,400 people missed out on employment opportunities. Furthermore, considering all months until December 2020, we estimated 54,100 people missing out on possible employment, corresponding to 2% of total unemployed people in Italy. 1. Introduction1 The discovery of a novel coronavirus in late 2019 (Zhu et al., 2020) which led to the global pandemic of Covid-19 (WHO, 2020) in March 2020 had a massive impact on the world economies (Jorda et al., 2020; Ma et al., 2020; OECD, 2020a; Liguori, Winkler 2020), dramatically changing the political and economic environment (Winston, 2020). Since Covid-19 turned out to be a highly infectious virus that can be easily transmitted, and also involving asymptomatic or peri-symptomatic phases (Bai et al., 2020), governments had to adopt lockdown and social distancing measures (Glass et al., 2006) to combat the spread of the virus, in order to also attenuate the pressure on the healthcare system. This * Centro Studi delle Camere di Commercio “Guglielmo Tagliacarne”, Rome, Italy, e-mail: [email protected] (corresponding author); [email protected]. 1. The present paper is an in-depth analysis within the research line of the Centro Studi delle Camere di commercio “Guglielmo Tagliacarne” on the territorial impact of the Covid-19 crisis on the business demography. The views expressed in the article are those of the authors and not of the institution they are affiliated with. The Authors thank Carmine Pappalardo for valuable suggestions and the partecipants of the AISRe Conference 2020. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 52 has led to a particular shock that has affected up to one third of GDP in the major economies (OECD, 2020b). Currently, policies have been focusing on protecting the existing industries and employment, with less attention to the future of economic activities, such as startups (Kuckertz et al., 2020). The important role of startups in job creation is widely recognized in the literature (more recently, e.g., Fritsch, Wyrwich, 2017), as well as the negative economic consequences of a decline in startups during a recession (Sedláček, 2019; Ayres, Raveendranathan, 2016; Gourio et al., 2014). The combination of the Covid-19 pandemic and the lockdown measures represents an unprecedented situation that has still not been addressed in the entrepreneurship literature. Recently, Sedláček and Sterk (2020) studied the effect of the decline in startups on employment in the United States in view of the Covid-19 crisis, as well as Karimov and Konings (2020) for Belgium; and Kuckertz et al. (2020) analyzed the effect of lockdown on the survival of startups. For Italy, some scholars have studied the effects of Covid-19 by the economic geography perspective investigating the role of the geographical concentration of economic activities (Ascani et al., 2020), the socio-economic and environmental factors (Musolino, Rizzi, 2020), and the relationship with the startups activity (Pini, Rinaldi, 2021). This paper aims to enrich this new strand of literature on the connections between entrepreneurship and Covid-19 under the lenses of the economic geography by estimating, for Italy, how many employment opportunities have been missed because of the decrease in startups during the two-months of lockdown (March-April 2020). Being the first country in Europe to be hit, Italy is one of the countries most affected by Covid-19 and the consequences of the lockdown on the new entrepreneurship were very evident: in the two-months March-April 2020, new business applications fell by 45.1% compared to the same period of the previous year. The remainder of the paper is structured as follows. Section 2 reviews the literature about the role of the startups activity for the economic system. Section 3 presents the background. Section 4 illustrates the data. Section 5 describes the method. Section 6 presents the results. Section 7 concludes. 2. Literature Review The positive effect exerted by startups on employment growth is widely recognized in the literature (Fritsch, Wyrwich, 2017; Doran et al., 2016; Fritsch, Schroeter, 2011; Fritsch, Mueller, 2008). Despite new firms undergoing a high failure rate in the short-term, the surviving firms grow faster in the long-run than the average existing firms (e.g., Haltiwanger et al., 2013; Fort et al., 2013). Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 53 In the current period of economic recession, the downside is that a decline in startups may have negative effects on employment, as it may lead to a persistent void in aggregate employment (e.g., Gourio et al., 2016; Sedláček, 2019). This is because a lack of new firms today means fewer older firms in the future, which contribute the most to employment levels (Sedláček, 2019). In fact, some scholars have highlighted the relationship between the slow recovery of firm entry and the slow recovery of employment (e.g., Elsby et al., 2011; Jaimovich, Siu, 2014; Haltiwanger et al., 2013). Many studies have focused on the effects of the decline in startups on employment during the Great Recession in the United States (US). Sedláček (2019) found that if the firm entry had remained constant, the level of unemployment would have been 0.5 percentage points lower over 10 years after the crisis. Gourio et al. (2014) studied the long-run effects of a decline in startups on employment levels, finding that the reduced entry rate resulted in a loss of 1.7 million jobs between March 2006 and March 2011, compared to a loss of only 500,000 between March 2006 and March 2009. Ayres and Raveendranathan (2016) also highlighted the strong relationship between startup rate and employment, estimating that 22% of the difference in the employment levels per labor force participant between March 2012 and March 2007 (pre-recession period) was due to the lack of firm entry. With specific reference to the Covid-19 pandemic and the related lockdown, Sedláček and Sterk (2020) studied the effect of the disruption in startups activity on US employment. Focusing on three margins corresponding to the number of startups, the growth potential, and the survival rate, they estimated that a reduction in these margins for one year to their minimum levels since 1977, would lead to a 1.1% aggregate employment reduction in 2020. More specifically, they developed a calculator to compute the long-term effects on employment caused by different scenarios related to the above three margins. Lastly, several studies have studied the impact of the Covid-19 and the lockdown on unemployment (e.g., Kong, Prinz, 2020; Gregory et al., 2020), as well as on employment by combining epidemiological and macroeconomic models (e.g., Kaplan et al., 2020; for a literature review, see Bank of Italy, 2020). For Italy, there are some studies analyzing the phenomenon under the lenses of the economic geography through analyses at the provincial level: Ascani et al. (2020) investigated the relationship between the spread of Covid-19 and the geographical concentration of economic activities finding positive results; Musolino and Rizzi (2020) analyzed the influence of demographic, socio-economic and environmental factors (always on the spread of Covid-19), finding significant effects exerted by several variables such as ageing population, commuting, pollution; while Pini and Rinaldi (2021) found a high significant effect Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 60 Lombardy is the region that lost the most: both in absolute terms with 6,400 people missing out on possible employment (1st region), and in relative terms since this value represents 56.8% of the theoretical employment (1st region) (Figures 2-3). More generally, the central-northern regions showed the highest values. The top-four regions by absolute values of employed people missing out on possible employment are all central-northern, as well as in terms of the percentage on theoretical employment: Lombardy, Lazio, Emilia-Romagna and Veneto in the first case; Lombardy, Marche, Tuscany and Lazio in the second Table 1 – Employment Missed for Startups Decrease in the Lockdown Period on Italy by Region Regions Employment Missed (Thousand) % on Italy Piedmont 2.2 7.1 Aosta Valley 0.0 0.1 Lombardy 6.4 20.4 Trentino-South Tyrol 0.5 1.6 Veneto 3.2 10.2 Friuli-Venezia Giulia 0.4 1.3 Liguria 0.7 2.2 Emilia-Romagna 3.6 11.6 Tuscany 2.5 8.1 Umbria 0.3 0.9 Marche 0.9 2.8 Lazio 4.3 13.6 Abruzzo 0.5 1.5 Molise 0.1 0.2 Campania 1.8 5.6 Apulia 1.8 5.6 Basilicata 0.1 0.2 Calabria 0.6 1.9 Siciliy 0.9 3.0 Sardinia 0.6 1.9 Italy 31.4 100.0 Note: Employment missed refers to the new employment not realized because of the startups decrease in the two-months March-April 2020. The data of Aosta Valley is less than 100. Source: Authors’ estimations Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 61 case. Thus, the employment potentially missed occurred in the most developed Italian regions, considering that all the above-mentioned regions have a GDP per capita that is above the national average (source: Istat). Moreover, the regions displaying the highest values are also those most affected by Covid-19. We found a high regional bivariate correlation between the employment missed and the number of Covid-19 cases (ρ=0.8, p-value<0.01). Figures 2-3 show that most regions (14 out of 20) are situated above or below the national average in both indicators (absolute values and in percentage terms). This is the effect of the Covid-19 pandemic on the startups decline (ρ = –0.4, p-value<0.10). Analyzing this relationship using relative values, we also found a positive and significant regional bivariate correlation between the percentage of employment missed on the theoretical employment, and the number of Covid-19 cases per inhabitants (ρ = 0.5, p-value<0.05) (Figure 4). 6.2. The New Employment Missed During the Months After the Lockdown The strong startups decline registered in the two-months of lockdown (MarchApril) continued also in May with a decrease by 37.7% compared to the same month in the previous year. Then the startups decline started falling gradually in June and July (respectively –7.2% and –2.4%), arriving to register an increase in August (+1.5%) and no change in September. Nevertheless, in the last three months of 2020 the startups trend turned negative (around –8% in October and in November; –15.3% in December), also in view of the worsening of the epidemiological crisis, causing especially in November and in December new lockdown measures (e.g. restrictions for bar and restaurant activities) even if lighter than those of the March-April. This new scenario deteriorated the general climate of confidence. In this regard, we underline the existence of a strong relationship between firm births trend on the one hand, and the business and consumer confidence index on the other hand (Unioncamere, 2020b) (Figure 5) (we found the same relationship also using the Social Mood on Economy Index elaborated by Istat based on the tweet). On the basis of the startups trend in the months after the lockdown, we estimated (with the same methodology) also the employment missed because of the startups decrease for the entire period of 2020, namely from March to December. The results indicate a value of 54,100 people missing out on possible employment, of which nearly two-thirds (58%) referred to the two-months of lockdown (March-April 2020) (Figure 6). This total value corresponds to 2% of the total unemployed people in Italy. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 62 Figure 2 – Regional Ranking on the Basis of the Absolute Values of Employment Missed for Startups Decrease in the Lockdown Period in Italy 6,4 4,3 3,6 3,2 2,5 2,2 1,8 1,8 0,9 0,9 0,7 0,6 0,6 0,5 0,5 0,4 0,3 0,1 0,1 0,0 Lombardy Lazio Emilia-Romagna Veneto Tuscany Piedmont Campania Apulia Siciliy Marche Liguria Sardinia Calabria Trentino-South Tyrol Abruzzo Friuli-Venezia Giulia Umbria Basilicata Molise Aosta Valley Figure 3 – Regional Ranking on the Basis of the Percentage of Employment Missed for Startups Decrease in the Lockdown Period in Italy on the Theoretical Employment 56,8 53,2 51,6 50,9 50,6 50,6 49,4 48,5 45,9 44,1 42,8 40,1 39,0 38,8 38,7 37,7 31,8 27,6 25,0 19,8 Average Italy 46.2 Lombardy Marche Tuscany Lazio Calabria Emilia-Romagna Liguria Piedmont Friuli-Venezia Giulia Apulia Trentino-South Tyrol Veneto Sardinia Aosta Valley Abruzzo Umbria Campania Siciliy Molise Basilicata Note: Employment missed refers to the new employment not realized because of the startups decrease in the two-months March-April 2020. Theoretical employment refers to employment generated by the flow of startups in the two-months March-April 2020 in a scenario without Covid-19 pandemic. Source: Authors’ estimations Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 63 Figure 4 – Employment Missed for Startups Decrease in the Lockdown Period and Covid-19 Cases in Italy by Region ABR BAS CAL CAM ER FVG LAZ LIG LOM MAR MOL PIE APU SAR SIC TUS TST UMB AV VEN 0 3,4 6,8 10,2 0 46,2 Covid-19 cases per 1.000 inhabitants % employment missed on the theoretical employment Note: Employment missed refers to the new employment not realized because of the startups decrease in the two-months March-April 2020. Theoretical employment refers to employment generated by the flow of startup in the two-months March-April 2020 in a scenario without Covid-19 pandemic. Covid-19 cases until 30 April 2020. For each indicator, the horizontal and the vertical line is situated in correspondence to the national average. Source: Authors’ estimations Figure 5 – Firm Births Trend in Italy and Business and Consumer Confidence Index 50 60 70 80 90 100 110 120 -70 -60 -50 -40 -30 -20 -10 0 10 20 gen feb mar apr mag giu lug ago set ott nov dic gen feb mar apr mag giu lug ago set ott nov dic % change firm births (left axis) Business confidence index (2010=100) (right axis) Consumer confidence index (2020=100) (right axis) 2019 2020 Source: Authors’ elaboration on Unioncamere-Infocamere and Istat data Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 64 7. Conclusions This paper estimates the missed new employment opportunities related to the decline in startups during the months of lockdown in Italy (March-April 2020) at the regional level. To the best of our knowledge, to date in the literature there are very few studies about the Covid-19 influence on the economy from the entrepreneurship perspective. We estimated that 31,400 people missed out on employment opportunities because of the startups decline in the months of lockdown. This corresponds to 46.2% of the theoretical employment that would have been generated by a flow of startups in a scenario without the Covid-19 pandemic. Lombardy shows the highest values in both absolute and relative terms. Moreover, considering also the startups trend during the following months of the 2020 after the lockdown, we estimated 54,100 people missing out on possible employment with reference to the entire period March-December 2020, corresponding to 2% of total unemployed people in Italy. Despite the southern regions registered a smaller decrease of startups decline, however we underline that the corresponding missed new employment in this area has a stronger effect (than in the north-central regions) in view of its higher unemployment. Figure 6 – Employment Missed for Startups Decrease in the Lockdown Period and in the Following Months of 2020 Lockdown mar-apr 31,44; 58% may-dec 22,62; 42% Source: Authors’ elaboration on Unioncamere-Infocamere and Istat data Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 65 The first message that emerges from the results is the strong effect of the lockdown on the setting up of firms. This highlights the employment opportunities missed as well as the potential threat to the overall level of innovation in the economic system that new firms tend to promote. One recommendation that we propose to policy makers is the need to also sustain entrepreneurship, especially in the post-lockdown months in order to recover the decline in startups, and to prevent a lost generation of firms. The effects provided by startups to the economy are several: i) in terms of productivity growth, innovation and job creation (Nielsen et al. 2020; Liu et al. 2020); ii) in the specific case of a recession, a decline in startups may generate persistent effects at the macroeconomic level (e.g., Sedlácek, Sterk, 2017, Gourio et al., 2016) because of a “missing generation” of firms (Clementi, Palazzo, 2016; Siemer, 2016); ii) the speed of recovery post-crisis also depends on firm entry (Clementi, Palazzo, 2016). In view of these considerations, Italy cannot afford a lost generation of firms since its low level of innovation, however with wide differences across regions (Pini, Quirino, 2017), and the weak structural economic growth. In the current scenario the support of policy makers is determinant because this crisis is involving radical changes that may discourage new entrepreneurs, raising the sense of competition and creating new challenges (OECD, 2020c). Specifically, besides the various financial incentives (e.g. tax reduction, government bank guarantees, subsidies) and the reduction of the administrative burdens ‒ which would help counteract the uncertainties in times of crisis ‒, may be important: i) favoring the increased awareness concerning the new business opportunities related to the new needs that have emerged from the crisis (Fairlie, 2020); fostering both entrepreneurial training in line with the challenges of the new global scenario, and university-business collaborations to facilitate the transition from universities to entrepreneurship; iii) investing in incubators and accelerators helping new potential entrepreneurs to overcome barriers related to the lack of trust as well as of knowledge of the new market demand and global challenges. The final goal is to support a type of entrepreneurship: i) more opportunitydriven (instead of necessity-driven); ii) more equipped with the appropriate skills to address the new challenges of the competitiveness (recently, on the new challenges of the recovery, see Esposito, 2020) by focusing on innovative processes, innovative products that would really boost the economy (Padilla, Petit, 2020) in line with the new demand that has been profoundly changed by the crisis; iii) more based on a greater work and production flexibility. For such purposes, the digitalization is an essential factor, because the recovery will have to go through the digital transition as recently recognized by the EU Next Generation program Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 66 (European Commission, 2020), which could play a key role also in the entrepreneurship field. In this regard, the preliminary results of a survey carried out by Unioncamere (2020c) on Italian manufacturing firms show that the share of firms that will be to return to pre-Covid production levels by 2022 is higher for those who are increasing their level of digitalization (digital business models, digital skills, etc.) compared to the others (67% vs 55%). All these considerations should be designed under the lens of geography: if on one hand the startups decline most occurred in the north-central regions (more developed), on the other hand the policy indications above explained should primarily focus on southern regions (less developed) for several reasons. First, southern regions suffer of a gap of digitalization and innovation, also in the startup field: for instance, there is a less diffusion of innovative startups (at the end of 2020, south: 15 per 100,000 inhabitants vs 23 in the case of north-central regions: source: Infocamere for innovative startups and Istat for population). Second, in southern regions the startup activity is most driven by necessity than opportunity. Third, in southern regions the firms’ death rate is higher. Thus, supporting the born of competitive startups is determinant to favor the economic territorial convergence because, right now, there is a risk of an increase of the widening economic gap between more developed and less developed Italian regions (Meliciani, Pini, 2020): indeed, the recovery in 2021 is expected to be stronger in the north-central regions than in the southern ones (+1.2% vs +4.5%) (Svimez, 2020). Despite the difficult economic period, adversity often leads to opportunities: looking at the past, we discover that over half of the companies on the 2009 Fortune 500 list were set up during a recession or bear market (Stangler, 2009). Thus, the role of regional governments and institutions in favoring this new entrepreneurship, through for instance one-stop support shops (European Commission, 2020), such as those managed by Chambers of Commerce, is particularly important. This analysis represents a first step in a potentially fruitful line of research. The study presents several limitations. First, we assume that firms that might have been started would have had the same employment potential as existing startups. Second, we did not take into account the failure rate of the firms that might have been started, thus negatively influencing the job creation. Moreover, we should also to take into account also the future possible effects of a job loss when the layoffs block (in force since the beginning of the crisis) will finish. Future research could extend the analysis in at least two directions: studying the impact of lockdown and Covid-19 pandemic with particular regard to women entrepreneurship (for first analyses see Unioncamere, 2020d) and to youth entrepreneurship; studying the territorial differences in the recovery of startups activity together to the firms’ resilience (on the basis of the firm deaths), in the post-lockdown months. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 67 References Ascani A., Faggian A., Montresor S. (2020), The Geography of Covid-19 and the Structure of Local Economies: The Case of Italy. Journal of Regional Science: 1-35. Doi: 10.1111/jors.12510. Ayres J., Raveendranathan G. (2016), Lack of Firm Entry and the Slow Recovery of the US Economy After the Great Recession. Mimeo – www.joaoayres.com Bai Y., Yao L., Wei T., Tian F., Jin D., Chen L., Wang, M. (2020), Presumed asymptomatic carrier transmission of Covid-19. JAMA, 323, 14: 1406-1407. Doi: 10.1001/ jama.2020.2565. Bank of Italy (2020), Covid-19 and Economic Analysis: A Review of the Debate. Literature Review n. 1. Rome: Bank of Italy – www.bancaditalia.it. Box G.E.P., Jenkins G. M. (1976), Time Series Analysis: Forecasting and Control (2nd ed.). San Francisco, CA: Holden-Day. Centers for Disease Control and Prevention (2020), Clinical Management of Critically Ill Adults with Covid-19 (slides). Retrieved from – https://emergency.cdc.gov. Clementi G.L., Palazzo B. (2016), Entry, Exit, Firm Dynamics, and Aggregate Fluctuations. American Economic Journal: Macroeconomics, 8, 2: 1-41. Doi: 10.1257/ mac.20150017. Doran J., McCarthy N., O’Connor M. (2016), Entrepreneurship and employment growth across European regions. Regional Studies, Regional Science, 3, 1: 121-128. Doi: 10.1080/21681376.2015.1135406. Elsby M.W., Hobijn B., Şahin A., Valletta R.G., Stevenson B., Langan A. (2011), The Labor Market in the Great Recession – An Update to September 2011. Brookings Papers on Economic Activity, Fall: 353-384. Doi: 10.1353/eca.2011.0013. Esposito G.F. (2020), Lockdown / Upside down. Società ed economie globali in ripartenza. Roma: Fralerighe Editore. European Commission (2020), Europe’s moment: Repair and Prepare for the Next Generation, COM(2020) 456 final. Brussels: EUR-Lex – https://eur-lex.europa.eu. Fairlie R.W. (2020), The Impact of Covid-19 on Small Business Owners: Continued Losses and the Partial Rebound in May 2020. Cambridge, MA: National Bureau of Economic Research. NBER Working Paper n. 27462 – www.nber.org. Fort T.C., Haltiwanger J., Jarmin R.S., Miranda J. (2013), How Firms Respond to Business Cycles: The Role of Firm Age and Firm Size. IMF Economic Review, 61, 3: 520-559. Doi: 10.1057/imfer.2013.15. Fritsch M., Mueller P. (2008), The Effect of New Business Formation on Regional Development Over Time: The Case of Germany. Small Business Economics, 30, 1: 15-29. Doi: 10.1007/s11187-007-9067-9. Fritsch M., Schroeter A. (2011), Why Does the Effect of New Business Formation Differ Across Regions? Small Business Economics, 36, 4: 383-400. Doi: 10.1007/s11187009-9256-9. Fritsch M., Wyrwich M. (2017), The Effect of Entrepreneurship on Economic Development. An Empirical Analysis Using Regional Entrepreneurship Culture. Journal of Economic Geography, 17, 1: 157-189. Doi: 10.1093/jeg/lbv049. Glass R.J., Glass L.M., Beyeler W.E., Min H.J. (2006), Targeted Social Distancing Designs for Pandemic Influenza. Emerging Infectious Diseases, 12, 11: 1671-1681. Doi: 10.3201/eid1211.060255. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 68 Gourio F., Messer T., Siemer M. (2014), What Is the Economic Impact of the Slowdown in New Business Formation? Chicago Fed Letter n. 326, September – www.chicagofed.org. Gourio F., Messer T., Siemer M. (2016), Firm Entry and Macroeconomic Dynamics: A State-level Analysis. American Economic Review, 106, 5: 214-18. Doi: 10.1257/aer. p20161052. Gregory V., Menzio G., Wiczer D. (2020), Pandemic Recession: L-shaped or V-shaped? Cambridge, MA: National Bureau of Economic Research, NBER Working Paper n. 27105. Doi: 10.3386/w27105. Haltiwanger J., Jarmin R.S., Miranda J. (2013), Who Creates Jobs? Small Versus Large Versus Young. Review of Economics and Statistics, 95, 2: 347-361. Doi: 10.1162/ REST_a_00288. Istat (2006), La rilevazione sulle forze di lavoro: contenuti, metodologie, organizzazione. Collana Metodi e norme n. 32. Roma: Istat – www.istat.it. Jaimovich N., Siu H. (2014), The Trend is the Cycle: Job Polarization and Jobless Recoveries. Cambridge, MA: National Bureau of Economic Research, NBER Working Paper n. 18334. Doi: 10.3386/w18334. Jorda , Singh S.R., Taylor A.M. (2020), Longer-run Economic Consequences of Pandemics. London: Centre for Economic Policy Research. CEPR Covid Economics Vetted and Real-Time Papers n. 1. Doi: 10.3386/w26934. Kaplan G., Moll B., Violante G. (2020), Pandemics according to HANK. Online seminar, March 31 – https://sites.google.com. Karimov S., Konings J. (2020), How Lockdown Causes a Missing Generation of Startups and Jobs. Small Business Economics. Doi: 10.1007/s11187-020-00395-z. Kong E., Prinz D. (2020), The Impact of Shutdown Policies on Unemployment During a Pandemic. London: Centre for Economic Policy Research. CEPR Covid Economics Vetted and Real-Time Papers n. 17 – https://cepr.org. Kuckertz A., Brändle L., Gaudig A., Hinderer S., Reyes C.A.M., Prochotta A., ... Berger E.S. (2020), Startups in Times of Crisis. A Rapid Response to the Covid-19 Pandemic. Journal of Business Venturing Insights, 13. Doi: 10.1016/j.jbvi.2020.e00169. Liguori E., Winkler C. (2020), From Offline to Online: Challenges and Opportunities for Entrepreneurship Education Following the Covid-19 Pandemic. Los Angeles, CA: SAGE Publications Sage. Doi: 10.1177/2515127420916738. Liu Y., Lee J.M., Lee C. (2020), The Challenges and Opportunities of a Global Health Crisis: The Management and Business Implications of Covid-19 from an Asian Perspective. Asian Business & Management, 19: 277-297. Doi: 10.1057/s41291-020-00119-x. Ma C., Rogers J., Zhou S. (2020), Global Financial Effects. London: Centre for Economic Policy Research. London: Centre for Economic Policy Research. CEPR Covid Economics Vetted and Real-Time Papers n. 5 – https://cepr.org. Meliciani V., Pini M. (2020), La crisi pandemica in Italia e gli effetti sul divario NordSud. Luiss Open, 29 novembre – https://open.luiss.it. Musolino D., Rizzi P. (2020), Covid-19 e territorio: un’analisi a scala provinciale. EyesReg–On-line Journal of Italian Association of Regional Science, 10, 3 – www.eyesreg.it. Nielsen J.E., Stojanović-Aleksić V., Bošković A. (2020), Promoting Entrepreneurship in HEIs: Leading and Facilitating University Spin-Off Ventures. In: Babić V., Nedelko Z. (eds.), Handbook of Research on Enhancing Innovation in Higher Education Institutions. Hershey, PA: IGI Global. 216-238. Doi: 10.4018/978-1-7998-2708-5.ch010. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 69 OECD (2020a), OECD Economic Outlook, Interim Report March 2020. Coronavirus: The World Economy at Risk. Paris: OECD Publishing. Doi: 10.1787/7969896b-en. OECD (2020b), OECD Policy Responses to Coronavirus. Evaluating the Initial Impact of Covid-19 Containment Measures on Economic Activity. Paris: OECD – www. oecd.org/coronavirus. OECD (2020c). OECD Policy Responses to Coronavirus. Start-ups in the time of Covid19: Facing the challenges, seizing the opportunities. Paris: OECD – www.oecd.org/ coronavirus. Padilla J., Petit N. (2020), Competition Policy and the Covid-19 Opportunity. Concurrences Review, 2-2020, 94317: 2-6 – www.concurrences.com. Pini M., Quirino P. (2017), L’innovazione tecnologica e i divari regionali. Roma: Aracne editrice Pini M., Rinaldi A. (2020), Nuova imprenditorialità mancata e perdita di occupazione: prime valutazioni sugli effetti della pandemia sul sistema produttivo italiano. EyesReg – On-line Journal of Italian Association of Regional Science, 10, 3 – www.eyesreg.it. Pini M., Rinaldi A. (2021), Covid-19, Lockdown and decline in startups: Is there a relationship? An Empirical analysis of Italian provinces. L’industria, 1-23. Doi: 10.1430/99914. Sedláček P. (2019), Lost Generations of Firms and Aggregate Labor Market Dynamics. Journal of Monetary Economics, 111: 16-31. Doi: 10.1016/j.jmoneco.2019.01.007. Sedláček P., Sterk V. (2017), The Growth Potential of Startups Over the Business Cycle. American Economic Review, 107, 10: 3182-3210. Doi: 10.1257/aer.20141280. Sedláček P., Sterk V. (2020), Startups and Employment Following the Covid-19 Pandemic: A Calculator. London: Centre for Economic Policy Research. CEPR Covid Economics Vetted and Real-Time Papers n. 13 – https://cepr.org. Siemer M. (2016), Firm Entry and Employment Dynamics in the Great Recession. SSRN paper n. 2172594. Doi: 10.2139/ssrn.2172594. Stangler D. (2009), The Economic Future Just Happened. SSRN paper n. 1580136. Doi: 10.2139/ssrn.1580136. Svimez (2020). Rapporto Svimez. L’economia e la società del Mezzogiorno. Bologna: il Mulino. U.S. Census Bureau (2011), X-12-ARIMA Reference Manual. Washington, DC – www. census.gov. Unioncamere (2020a), Documento delle Camere di Commercio: Relazione e proposte per la ripartenza dell’economia. Documento presentato alla 10° Commissione Industria, commercio e turismo. Roma: Senato della Repubblica Italiana – www.senato.it. Unioncamere (2020b), Sul rilancio del commercio alla luce della crisi causata dall’emergenza epidemiologica. Documento presentato il 10 novembre alla 10° Commissione Industria, commercio e turismo. Roma: Senato della Repubblica Italiana – www.senato.it. Unioncamere (2020c), Third survey on Italian manufacturing firms. Unioncamere (2020d), IV Rapporto Imprenditoria Femminile 2020. Unioncamere – www.unioncamere.gov.it. WHO (2020), Coronavirus disease 2019 (Covid-2019). Situation Report 51. World Health Organization. Geneva: Switzerland – www.who.int. Winston A. (2020), Is the Covid-19 Outbreak a Black Swan or the New Normal. MIT Sloan Management Review, March – https://sloanreview.mit.edu. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 77 Exploring “Resiliencies” to the Great Crisis along the Peripherality Gradient in Central-southern Italy Fabiano Compagnucci*, Giulia Urso*1 Abstract The notion of resilience has been widely studied over the last two decades in the field of regional studies. Different dimensions have been used so far to proxy it. We suggest that the choice of these variables is not neutral in terms of evaluation of the resilience capacity, depending on the different socioeconomic structure of different territorial contexts. By using municipal data on population, employment, and personal income from 2004 to 2017 of the Abruzzo, Lazio, Marche and Umbria regions, our analysis is meant to provide empirical evidence to our assumptions by investigating the resilience of these regions in the face of the 2007-2008 Great Crisis and the subsequent recovery period. Our study intends to contribute to the production of knowledge on resilience assessment, especially with reference to peripheral areas, which are in most cases already challenged by prolonged slow-burning pressures. Results may eventually fuel both the theoretical and policy debate on the resilience of inner areas. 1. Introduction The notion of resilience has been widely studied over the last two decades in the field of regional studies, mainly because of the outbreak of the 2007-2008 crisis. Although the Great Recession has affected the entire global economy, it caused asymmetric recessionary shocks at the national, and especially, at the regional and local level (Capello et al., 2015; Groot et al., 2011). It has eventually resulted in different degrees of the magnitude of the crisis and of the extension of the recovery period depending on the different resilience capacity of places. Many scholars have corroborated these initial findings at various spatial le vels: NUTS-2 (Doran, Fingleton, 2016; Crescenzi et al., 2016), NUTS-3 (Fratesi, Perucca, 2019; Angulo et al., 2018), functional areas (Faggian et al., 2018), and municipality level (Geelhoedt et al., 2021). Along with various territorial levels, different dimensions have * GSSI – Gran Sasso Science Institute, Social Sciences, L’Aquila, Italy, e-mail: fabiano.compa- [email protected] (corresponding author); [email protected]. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 78 been used so far to proxy resilience. Indeed, despite the popularity it has gained both in the political and academic discourse, there is no unanimous consensus about what (regional) resilience precisely is (Stanickova, Melecký, 2018; Muštra et al., 2020), neither in terms of definition nor in terms of measurement (Martin, 2012). Resilience is, however, commonly assessed considering three main variables: population, employment rate and GDP (Dubé, Polèse, 2016). Together with Dubé and Polèse (2016), we suggest that the choice of these variables is not neutral in terms of evaluation of the resilience capacity. Each of them, in fact, could proxy different aspects of resilience which peculiarly react to the recessionary shock depending on the concerned territorial context. For instance, assessing resilience on the basis of employees can be more effective when carried out in urban areas than in rural/peripheral ones. Within these latter, the share of retired people is usually much larger, and with it also the share of households receiving an income independently from the crisis. It follows that, when attempting to measure resilience, we must be aware of territorial level we are analyzing, be it composed by regions, local systems, towns or villages (Compagnucci, Morettini, 2020). Against this background, the aim of this paper is twofold. First, exploiting the spatial classification provided by the National Strategy for Inner Areas (SNAI), the paper empirically investigates how the use of different variables affects the measure of resilience to the 2008 Great Crisis in the territories of the four Italian regions (Marche, Umbria, Lazio and Abruzzo) which were hit by the 2016-2017 earthquake. Second, our research aims at exploring whether using a specific variable or another can be considered more appropriate in assessing the resilience of different territorial contexts, looking at both in-between regional heterogeneity and within regional heterogeneity along the urban gradient, moving from core to more peripheral areas. The remainder of the paper is structured as follows. After having contextualized the why question of this study within the theoretical debate, we will provide empirical evidence to our assumptions through a descriptive analysis based on municipal data on population (ISTAT), employment (ISTAT) and individual income subject to taxation (Ministry of Economy and Finance) from 2004 to 2017. Relying on previous studies reflecting on the operationalization of the notion of resilience, we will finally discuss the main findings arising from the descriptive analysis about the assessment of resilience especially when investigating inner areas, which, in most cases, are territories already severely challenged by prolonged slow-burning pressures. 2. Resilience of What: The Appraisal of Context When discussing the concept of resilience several aspects need to be taken into account, hence the complexity of both its conceptualization and empirical Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 79 application (Christopherson et al., 2010; Martin, Sunley, 2015). Even when limiting the scope to Regional Science and Economic Geography, without contemplating the various interpretations within other domains, a broad agreed-upon definition of resilience of territories still remains far to be reached. The geographical scale of investigation or more appropriate boundaries to be considered – a crucial dimension, or likely the primary concern, in spatial disciplines – could be controversial and must be carefully pondered. To start with, Faggian et al. (2018), for instance, argue that answering three fundamental questions is pivotal to guide research on resilience: 1. resilience “to what?” (referring to the kind – natural disaster, economic recession, etc. – and nature of the shock – acute, one-time or chronic stress (i. e. financial crisis vs. deindustrialization, see Pendall et al., 2010); 2. resilience “of what?” (which implies the definition of what we mean by economic system or, more generally, the geographic area to be scrutinized); 3. resilience “over what period?” (in order to assess the ability of a territorial system to resist the shock, bounce back or bounce forward toward new growth paths). We add to this list a fourth question which covers another crucial, lively debated point within the knowledge produced so far on resilience – to which this study aims to contribute to – and further accounts for its complexity: 4. resilience “through what indicator?” Reviewing the huge literature on the topic (starting from the overview provided by Modica, Reggiani, 2015), the “of what” question seems to be the less investigated or, better, the one less critically scrutinized, given the (also datadriven or taken-for-granted) reliance on administrative (mostly regions) or functional areas (local labour systems). However, we deem instead essential a reflection on the geography of resilience, primarily considering the possibly different spatiality of the alternative measures to proxy it. This because, as acknowledged for instance by Ward et al. (2003) discussing more generally rural development (see also Irwin et al., 2010), theoretical and empirical tools, being biased towards urban problem definitions, are in some cases not sufficiently sensitive to account for the peculiarities and performances of non-urban areas. Also, an over-reliance on a narrow set of indicators could accentuate this issue further. When it comes to resilience this might be the case as well. In fact, as underlined by Fantechi et al. (2020), among others, referring to the academic literature on disaster resilience, the majority of studies focus on urban contexts, while research on rural areas is still a residual category that typically does not take into account the geographical characteristics of places. These latter could instead play a key role, like the degree of peripherality or accessibility moving along the urban gradient. Even analyses at the sub-regional level if, on the one side, do allow for an appraisal of the high heterogeneity of context, on the other side very rarely go beyond the Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 80 mere urban vs. rural analytical opposition, having been designed mainly on the basis of urban areas as a reference category. However, resilience may imply different dimensions relative to those which are salient in the case of urban environments and hence be more properly detected through different measures which might better capture the actual ability of non-urban or non-core places to react to a disturbance. Dubé and Polèse (2016) very well account for this issue in conceptualizing and empirically analyzing resilience. Essentially combining the two questions “over what period?” and “through what indicator?”, they assess the resilience of Canadian regions to the 2007-2009 crisis over three phases from a short to a longer-term period (1. resistance, 2. rebound and 3. recuperation) and by means of four standard metrics (1. population, 2. employment, 3. unemployment and 4. employment rate). As largely expected, they found that regional resilience varies depending on the chosen measure. More interestingly for the purposes of our research, empirical evidence also led to a further reflection: responses to a recessionary disturbance – or any kind of disturbance, we would add – cannot be unequivocally explained for all regions, because context as well matters a lot in revealing the ability of territories to react to a pressure. In the two authors’ own words, “the differing responses to shocks also invite the question whether ‘resilience’ is a concept uniformly applicable across all regions, big and small, urban and rural, industrial and resource dependant. Should the criteria be the same for a large metropolis like Toronto as for a rural region in Saskatchewan?” (Dubé, Polèse, 2016: 626). In the exploratory attempt to answer this question, we assume here that the context-specific socio-economic features of places along the urban – or more precisely peripherality – gradient might influence results in evaluating resilience. Against this backdrop, we thus add a further element of complexity which we deem as highly salient, especially from a regional science and economic geography perspective, to operationalize the notion of resilience: the spatial dimension. We aim to contribute to the advancement of knowledge in this respect by including also the “of what” question in the framework outlined by Dubé and Polèse (2016) who built on the empirical literature hitherto produced on the topic, by analysing, at a more granular level, the responses of the municipalities of 4 Italian Central regions as classified within the National Strategy for Inner Areas (henceforth, SNAI, see next section), that is according to the travel-time distance from the closest service provision centre(s). In our study, then, the spatiality issue is not tackled only as a matter of scale – i.e. sub-regional vs. regional – but also as a critical rationale in the exploitation of municipal data – i.e. overcoming the mere urban vs. rural analytical lens. The SNAI classification was utilized in other studies on resilience to the 2008 global financial crisis (Urso et al., 2019), but only for a single-metric assessment. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 81 The value of the present investigation is mainly empirical in nature in the first place, which is namely to understand what measure is more relevant to and more responsive in what territories, hence what “resilience” is more salient in what place along the peripherality gradient. Also, beyond providing insights on the operationalization of the notion from a scholarly perspective, results might help detecting the specific vulnerabilities of territories based on the dimension they are more sensitive to, and hence, policy-wise, they might input policies targeting preparedness and resistance to shocks, limiting their magnitude. 3. Data and Methodology The empirical section is based on a set of descriptive statistics providing stylised facts on resilience measured through three different metrics in four Italian central (Lazio, Marche, Umbria) and southern (Abruzzo) regions. These regions, and especially their mountain areas, form the macro-region of what is now commonly referred to as the “crater of the Central Italy 2016-17 earthquake”. Aiming in the future at investigating also the resilience of the area to this natural disaster as soon as updated data will be released, our research project seeks to realise whether some lessons can be learnt from the past, specifically from the effects caused by the outbreak of the 2007-2008 crisis. More precisely, our analysis focuses on: a) a different assessment of resilience resulting from the use of different variables for its computation; and b) the kind and level of resilience of the different areas moving along the urban gradient (from poles to ultra-peripheral areas). For the latter point (b), we adopt the classification of Italian municipalities as provided by SNAI, which is based on three breakdowns both for urban and inner areas. Urban areas are split into 3 categories: A) “poles”: single-municipality service provision centres; B) “intermunicipal poles”: multi-municipality service provision centres, the main difference with A lying in their capacity to jointly (and not individually) provide education, transportation and health services; and C) their “urban belts”: municipalities that are less than 20 minutes far from poles and intermunicipal poles. Inner areas are split into 3 classes as well: D) “intermediate”: municipalities that are between 20 and 40 minutes far from poles and intermunicipal poles; E) “peripheral”: from 40 to 75 minutes, and F) “ultraperipheral”: more than 75 minutes, areas (UVAL, 2014). To perform the descriptive analysis, building on the reflections by Dubé and Polèse (2016), we use three different variables at the municipal level to proxy resilience: 1. population (Istat, Atlante Statistico dei Comuni); 2. employment (Istat, ASIA database) – as in Dubé and Polese (2016) – and 3. total individual income subject to taxation according to the normal progressive tax rates set forth Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 82 by the financial administration1. We were not able to also use GDP, since in the Italian context its estimation is not available at the municipal level. The eighteen-year period has been further subdivided into three periods with respect to the outbreak of the Great Crisis: the pre-crisis period, between 2004 and 2007, the crisis period, between 2007 and 2009, and the post-crisis period, between 2009 and 2017. Regarding this last period a comment should be made. Even though it can be considered a quite long period after the Great Crisis to assess resilience, it is worth noting that Italy also suffered from the sovereign debt crisis in 2010-2011. The sovereign debt crisis, whose effects lasted until 2014, slowed down a merely embryonic and very fragile recovery process which eventually started only in 2015. Although the three variables can be alternatively used to describe the resilience capacity of places, as is commonly found in the empirical literature on the topic, each of them can capture different dimensions of resilience, stressing different functions taking place at the local level. More in depth, population trends can be used to describe the capacity of a place to keep its inhabitants, thus pointing to the local/residential function; employment trends, since employees are recorded on the basis of the municipality where they do work, describe the job attractiveness of a place; and finally, total individuals’ income denotes the trends in the purchasing power of a given territory. These different aspects must be taken into account when performing a territorial analysis on resilience. For instance, it is important to consider that for some municipalities the residential function might be more important than attractiveness. In a functional perspective, in fact, considering the metric of local systems or cluster of municipalities, some places can be primarily specialised in hosting households whereas some other can play as local economic engines providing job opportunities. This means that in the first case resident population might be the most affected variable in the aftermath of the shock, whereas in the latter this might be the case for employment. These metrics will be analysed and compared on the basis of sequential variations (∆) of annual mean growth, calculated as the geometric mean of annual variations through the following equations: [1] [2] 1. Because of data availability constraints related to “employment” we had to limit our analysis to the period 2004-2017. 1 1 k tk t pop pop pop +  ∆= −   1 1 k tk t emp emp emp +  ∆= −   Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 83 [3] In equation [1] we consider the ratio between population at the end of the period (t+k) on the value of the population at the beginning of the period (t), we raise the result to the power of one divided by the period length (k) and we subtract one from the subsequent result. In equations [2], [3] we perform the same calculation using the variables employment (emp), and the sum of individual income at municipal level (Σinc)2. Building on and partially rearranging Dubé and Polèse (2016), the different trends (under the three scrutinized variables) followed by the selected municipalities can be classified into 8 categories. More specifically, we can consider two blocks of trends: a first block (1-4, see Table 1) includes different crisis and post-crisis trends following a pre-crisis negative variation, whereas in the second block (5-8, Table 1) the pre-crisis variation is positive. 4. Discussion of Results In 2007, in Abruzzo, Lazio, Marche and Umbria there are 1003 municipalities, most of which are located in inner areas (67,5% against 32,5% belonging to core areas). In terms of population the situation is completely reversed: only 26,2% of the total population live in peripheral areas, while 73,8% in urban ones. Looking at the different typologies identified by SNAI, data show that the most common category is that of intermediate inner areas (43,4%), followed by urban belts (27,7%), peripheral inner areas (20,8%), ultra-peripheral (3,3%), poles (3,1%) and intermunicipal poles (1,7%) (See Statistical Appendix, Table A1 and A2). A first stylised fact arising from the descriptive analysis revolves around one of the main why question inspiring this contribution. Does considering different variables affect the measure of resilience? Results show that, at the regional level, using population or employees or income leads to different results in terms of degrees of resilience (as identified in Table 1): a resistance trend when using population, or a severely hit trend, when employees or income are concerned. This is true for all the regions considered but Lazio, where employees and population followed a resistance path while total income proved to be severely hit by the crisis (Table 2). The differences in terms of resilience arising from the choice of one of the three variables, however, become striking when considering the local level. Table 3, which reports the number of municipalities following the same trend independently from the variables used in assessing resilience, suggests that 2. Regional values of population, employment and individual income have been calculated by summing their respective municipal values. In the case of individual income, for the calculation of the regional value, we obviously considered the total amount of individual income at the municipal level and not the mean of the municipal individual income. 1 1 k tk t inc inc inc +  ∑ ∆∑ = −  ∑  Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 84 selecting different variables affects the measure of resilience. When considering population, employment and total income, in fact, only 48 municipalities (4,8% out of total) follow the same sequential variations, independently from the concerned variable. Moreover, although the existence of some common macro-pattern related to pre-crisis, crisis and post-crisis periods, the choice of a given variable results in different territorial outcomes, both in the number and the typology of the concerned municipalities (Figures 1 and 2). As for population trends, among the municipalities following a negative pre-crisis performance, which were 42% out of total, about half of them (208 units) were affected by a systemic declining (– – –) (Tables B, C and D, Statistical Appendix). It is worth noting that 90,1% of these municipalities belong to inner areas (Figure 1), a share which is quite higher than their relative weight on the total number of municipalities. The second most common trend was the one labelled as counter cyclical (– + –), which characterized 170 municipalities. Here again, the phenomenon has affected inner areas more than proportionally. Concerning the municipalities that were following a positive growth path before the crisis (58% out of total), we find that the most common trends are resistance (+ + +, 302 units) and lagged shock (+ + –, 211 units). Unlike the first block, and particularly regarding resistance (+ + +), these trends are more common in all the typologies of urban areas (poles, intermunicipal poles, and belt areas). From a regional perspective, each of the four regions behaved accordingly with the average outlined above described. The only difference concerns Umbria region, where systematically declining municipalities were substantially fewer than counter-cyclical ones (Figures 1 and 2). When considering employment, the picture changes considerably. First of all, the difference between municipalities following a growth path and those following a declining path before the crisis is larger. Municipalities with a negative performance between 2004 and 2007, in fact, amount to 24,7%, which is about half of those showing a declining demographic trend (Tables B, C and D, Statistical Appendix). Here again the trends with the highest frequencies are those characterised by a systemic declining path (– – –, 96 municipalities) and a counter cyclical one (– + –, 80 municipalities). As for the former, it affected more than proportionally the belt areas, whereas the counter cyclical trend is a feature of mostly all the three classes of inner areas. Among the municipalities that experienced an employment growth before the crisis, the highest number of them followed the severely hit trend (+ – –), which affected more than proportionally all the typologies of urban areas. On the contrary, the lagged shock trend (+ + –), with 214 municipalities, is not linked with any specific typology of municipalities (Tables B, C and D, Statistical Appendix). Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 85 Table 1 – Pre-crisis, Crisis and Post-crisis Resilience Trends Trend Periods (2004-2007) (2007-2009) (2009-2017) 1 Systemic declining – – – 2Turnaround – – + 3Counter cyclical –+– 4Positive jolt –+ + 5 Resistance + + + 6Severely hit + – – 7Standard resilience + –+ 8 Lagged shock + + – Note: + indicates positive sequential variations of annual mean growth; – the opposite Source: Authors’ elaboration building on Dubé and Polèse (2016) Table 2 – Resilience Trends per Region and Variable Region Population Employees Income Umbria + + + + – – + – – Marche + + + + – – + – – Lazio + + + + + + + – – Abruzzo + + + + – – + – – Source: Authors’ elaboration on Istat and MEF data Table 3 – Resilience Trends along the Peripherality Gradient Urban areas Inner areas Total Poles Belt areas Intermediate areas Peripheral areas + – – 0 13 2 6 + – + 0 0 2 0 2 + + – 0 10 15 4 29 + + + 1 4 4 2 11 Total 1 15 24 848 % on total municipalities 4,8 Note: Table 3 reports only the typologies of municipalities within which at least one municipality followed the same trend independently from the concerned variable. Source: Authors’ elaboration on Istat and MEF data Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 92 Sommario “Resilienze” dell’Italia centro-meridionale alla Grande Crisi lungo il gradiente di perifericità La nozione di resilienza è stata ampiamente studiata negli ultimi due decenni nel campo delle scienze regionali. In letteratura sono state individuate diverse dimensioni per indagare la resilienza dei sistemi territoriali. L’assunto da cui muove questa riflessione è che la scelta delle variabili utilizzate per descriverla non sia neutra in termini di valutazione della capacità di resilienza, poiché questa può dipendere anche dalla diversa struttura socio-economica dei diversi contesti territoriali. Utilizzando dati a livello comunale sulla popolazione, sull’occupazione e sul reddito delle persone fisiche dal 2004 al 2017 delle regioni Abruzzo, Lazio, Marche ed Umbria, la nostra analisi intende fornire evidenza empirica alle ipotesi formulate rispetto alla Grande Crisi del 2007-2008, contribuendo così a produrre conoscenza sulla valutazione della resilienza. Questo obiettivo assume una rilevanza particolare rispetto alle aree periferiche, che, nella maggior parte dei casi, hanno subito fenomeni di declino socio-economico prolungato. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 93 Statistical Appendix Table A1 – Number of Municipalities per Region and SNAI Classification Poles Intermunicipal poles Belt areas Intermediate areas Peripheral areas Ultraperipheral areas Total Abruzzo N. 6 4 65 115 84 31 305 % 2,0 1,3 21,3 37,7 27,5 10,2 100 Lazio N. 10 0 78 205 83 2 378 % 2,6 0,0 20,6 54,2 22,0 0,5 100 Marche N. 11 8109 75 25 0 228 % 4,8 3,5 47,8 32,9 11,0 0,0 100 Umbria N. 4 526 40 17 0 92 % 4,3 5,4 28,3 43,5 18,5 0,0 100 Table A2 – Population per Region and SNAI Classification Poles Intermunicipal poles Belt areas Intermediate areas Peripheral areas Ultraperipheral areas Total Abruzzo N. 362619 67005 415824 330447 125240 21112 1322247 % 27,4 5,1 31,4 25,0 9,5 1,6 100 Lazio N. 3399140 0 881741 1374679 238089 4475 5898124 % 57,6 0,0 14,9 23,3 4,0 0,1 100 Marche N. 568524 134053 626010 182087 27381 0 1538055 % 37,0 8,7 40,7 11,8 1,8 0,0 100 Umbria N. 373330 72166 221701 190436 31275 0 888908 % 42,0 8,1 24,9 21,4 3,5 0,0 100 Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 94 Table B – Crosstab between Typology of Municipalities and Total Income Trends Poles Intermunicipal poles Belt areas Intermediate areas Peripheral areas Ultraperipheral areas Total - - - N. 0 0 2 6 8 1 17 % inc. 0,0% 0,0% 11,8% 35,3% 47,1% 5,9% 100% % mun. 0,0% 0,0% 0,7% 1,4% 3,8% 3,0% 1,7% - - + N. 0 0 2 2 509 % inc. 0,0% 0,0% 22,2% 22,2% 55,6% 0,0% 100% % mun. 0,0% 0,0% 0,7% 0,5% 2,4% 0,0% 0,9% - + - N. 0 0 2 85015 % inc. 0,0% 0,0% 13,3% 53,3% 33,3% 0,0% 100% % mun. 0,0% 0,0% 0,7% 1,8% 2,4% 0,0% 1,5% - + + N. 0 0 1010 2 % inc. 0,0% 0,0% 50,0% 0,0% 50,0% 0,0% 100% % mun. 0,0% 0,0% 0,4% 0,0% 0,5% 0,0% 0,2% + - - N. 20 10 92 159 84 23 388 % inc. 5,2% 2,6% 23,7% 41,0% 21,6% 5,9% 100% % mun. 64,5% 58,8% 33,1% 36,6% 40,2% 69,7% 38,7% + - + N. 2 2 57 65 23 5154 % inc. 1,3% 1,3% 37,0% 42,2% 14,9% 3,2% 100% % mun. 6,5% 11,8% 20,5% 14,9% 11,0% 15,2% 15,4% + + - N. 84 77 162 69 4 324 % inc. 2,5% 1,2% 23,8% 50,0% 21,3% 1,2% 100% % mun. 25,8% 23,5% 27,7% 37,2% 33,0% 12,1% 32,3% + + + N. 1 1 45 33 14 0 94 % inc. 1,1% 1,1% 47,9% 35,1% 14,9% 0,0% 100% % mun. 3,2% 5,9% 16,2% 7,6% 6,7% 0,0% 9,4% Total N. 31 17 278 435 209 33 1003 % inc. 3,1% 1,7% 27,7% 43,4% 20,8% 3,3% 100% % mun. 100% 100% 100% 100% 100% 100% 100% Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 95 Table C – Crosstab between Typology of Municipalities and Employment Trends Poles Intermunicipal poles Belt areas Intermediate areas Peripheral areas Ultraperipheral areas Total - - - N. 1 1 35 43 13 3 96 % emp. 1,0% 1,0% 36,5% 44,8% 13,5% 3,1% 100% % mun. 3,2% 5,9% 12,6% 9,9% 6,2% 9,1% 9,6% - - + N. 0 0 11 20 52 38 % emp. 0,0% 0,0% 28,9% 52,6% 13,2% 5,3% 100% % mun. 0,0% 0,0% 4,0% 4,6% 2,4% 6,1% 3,8% - + - N. 0113 38 23 580 % emp. 0,0% 1,3% 16,3% 47,5% 28,8% 6,3% 100% % mun. 0,0% 5,9% 4,7% 8,7% 11,0% 15,2% 8,0% - + + N. 0 0 4 18 10 2 34 % emp. 0,0% 0,0% 11,8% 52,9% 29,4% 5,9% 100% % mun. 0,0% 0,0% 1,4% 4,1% 4,8% 6,1% 3,4% + - - N. 13 10 98 104 55 10 290 % emp. 4,5% 3,4% 33,8% 35,9% 19,0% 3,4% 100% % mun. 41,9% 58,8% 35,3% 23,9% 26,3% 30,3% 28,9% + - + N. 53 46 62 37 1154 % emp. 3,2% 1,9% 29,9% 40,3% 24,0% 0,6% 100% % mun. 16,1% 17,6% 16,5% 14,3% 17,7% 3,0% 15,4% + + - N. 92 53 98 45 7 214 % emp. 4,2% 0,9% 24,8% 45,8% 21,0% 3,3% 100% % mun. 29,0% 11,8% 19,1% 22,5% 21,5% 21,2% 21,3% + + + N. 3 0 18 52 21 3 97 % emp. 3,1% 0,0% 18,6% 53,6% 21,6% 3,1% 100% % mun. 9,7% 0,0% 6,5% 12,0% 10,0% 9,1% 9,7% Total N. 31 17 278 435 209 33 1003 % emp. 3,1% 1,7% 27,7% 43,4% 20,8% 3,3% 100% % mun. 100% 100% 100% 100% 100% 100% 100% Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 96 Table D – Crosstab between Typology of Municipalities and Population Trends Poles Intermunicipal poles Belt areas Intermediate areas Peripheral areas Ultraperipheral areas Total - - - N. 3 0 16 100 70 19 208 % pop. 1,4% 0,0% 7,7% 48,1% 33,7% 9,1% 100% % mun. 9,7% 0,0% 5,8% 23,0% 33,5% 57,6% 20,7% - - + N. 0 0 131 1 6 % pop. 0,0% 0,0% 16,7% 50,0% 16,7% 16,7% 100% % mun. 0,0% 0,0% 0,4% 0,7% 0,5% 3,0% 0,6% - + - N. 12 36 77 49 5170 % pop. 0,6% 1,2% 21,2% 45,3% 28,8% 2,9% 100% % mun. 3,2% 11,8% 12,9% 17,7% 23,4% 15,2% 16,9% - + + N. 3 1 8 20 8 1 41 % pop. 7,3% 2,4% 19,5% 48,8% 19,5% 2,4% 100% % mun. 9,7% 5,9% 2,9% 4,6% 3,8% 3,0% 4,1% + - - N. 0 0 11 23 17 556 % pop. 0,0% 0,0% 19,6% 41,1% 30,4% 8,9% 100% % mun. 0,0% 0,0% 4,0% 5,3% 8,1% 15,2% 5,6% + - + N. 0 0 2 3 4 0 9 % pop. 0,0% 0,0% 22,2% 33,3% 44,4% 0,0% 100% % mun. 0,0% 0,0% 0,7% 0,7% 1,9% 0,0% 0,9% + + - N. 2 567 102 34 1211 % pop. 0,9% 2,4% 31,8% 48,3% 16,1% 0,5% 100% % mun. 6,5% 29,4% 24,1% 23,4% 16,3% 3,0% 21,0% + + + N. 22 9137 107 26 1302 % pop. 7,3% 3,0% 45,4% 35,4% 8,6% 0,3% 100% % mun. 71,0% 52,9% 49,3% 24,6% 12,4% 3,0% 30,1% Total N. 31 17 278 435 209 33 1003 % pop. 3,1% 1,7% 27,7% 43,4% 20,8% 3,3% 100% % mun. 100% 100% 100% 100% 100% 100% 100% Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 97 The High-tech Composite Indicator (HTCI). A Tool for Measuring European Regional Disparities Over Crises Simona Brozzoni*,1Silvia Biffignandi*, Matteo Mazziotta° 2 Abstract Composite indicators are a tool for territorial economic policy strategies as they allow the dimensional reduction of complex socio-economic phenomena not directly measurable with single elementary indicators. This paper focuses on the measurement and study of high technology in European regions: a new indicator of high-tech is constructed and used for the spatial and temporal analysis. In particular, the proposed indicators consider the period 2006-2016 for regions of Europe. Through the statistical analysis of this indicator have been verified some hypothesis of territorial disparities of high-tech, of their trend and development factors. 1. Introduction This research provides a new definition and measurement of high technology in European Regions, classified according to NUTS 2 Regulation. A new and innovative composite indicator has been constructed to underline European Regional disparities (in a spatial and temporal comparison) and what are the determinants of the development of high technology. The period covered in the analysis is from 2006 to 2016. The data source is Eurostat, the Statistical Office of the European Union. The analysis on two levels (spatial and temporal) allows an adequate understanding of the European regional breakdown according to the high technology content, as well as highlighting possible changes over time (particularly in relation to economic and financial crises). What is obtained is a synthetic and robust measurement of this phenomena, expressed as a combination of elementary indicators which, independently, represent specific dimensions of the concept to be measured. * University of Bergamo, Department of Economcs, Bergamo, Italy, e-mail: simona.brozzoni@ outlook.it (corresponding author); silvia.bif[email protected]. ° Istat – Italian National Institute of Statistics, Rome, Italy, e-mail: [email protected]. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 98 To structure composite indicators in a proper and transparent way, every step of the construction methodology is deeply studied, different techniques are applied and results compared. It came out that the best method in this case is the Adjusted Mazziotta-Pareto Index (AMPI) because it allows a spatial and temporal comparison, as well as ensuring robust results. The scores of the new computed high-tech indicator (based on AMPI method) have been compared through the construction of maps of European regional geography in order to obtain a discrimination of the different territories in terms of high technology. The results have highlighted a constant disparity between the regions of Eastern and Northern Europe, where the latter are the most performing. Moreover, the crisis of 2007-2008 has negatively affected the whole European regional scenario, slowing its high technology content. However, regions have experienced an increase over time in their level of high-tech (in 2016, have been registered higher index scores). 2. Backgrounds 2.1. Composite Indicators According to Saisana and Tarantola (2002), a composite indicator is a combination of elementary indicators representing different dimensions of a phenomenon and it is usually applied when a multidimensional concept cannot be measured by a single one. Composite indicators are widely used by various national and international organizations to analyse economic, environmental and social scenarios (i.e., industrial competitiveness, sustainable development, quality of life assessment, globalisation, etc.) (OECD, 2008). Maximum benefits can be obtained if and only if the composite indicator is structured correctly and transparently: just think that each choice made in the construction phase will have a direct impact on both the quality and reliability of the results. For this reason, ten steps have been defined for the construction of a composite indicator (OECD, 2008; Mazziotta, Pareto, 2017). 2.2. High Technology The phenomenon of high technology diversifies according to specific factors such as research and development spending, intellectual property rights, specific capabilities but also external and territorial influences. For the construction of composite indicators, this work will focus on various aspects that influence the growth of high technology in European Regions. The definition of “high technology” is quite complex and several problems have been addressed by the literature to define it, also considering the context Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 99 in which it is applied. Although recognised in almost all countries, there is little harmonisation between the various definitions (Joseph, 1988). However, by reviewing the researchers’ definitions, it is possible to give a new characterization of high technology. The concept could uniquely be expressed as the ability of a firm to remain competitive, to renew itself, to be innovative, as well as to steer investment in science and technology and in R&D. On the other hand, from a general point of view, high technology could be interpreted as the tight network of social, political and economic forces that interact each other, leading to economic growth. This new interpretation can be traced back to the common features of the actual literature overview (Porter et al., 1996; Johnson et al., 2010; Steenhuis, De Bruijn, 2006; Erlhoff, Marshall, 2007; Eurostat, 2018; 2020). In addition, thanks to its enormous growth, high-tech arouses considerable economic and social interest. The existing literature has brought to light several definitions, triggering problems related to the understanding of the phenomenon. In addition to this ambiguity, it is essential to determine whether it exists and which are the differences with “digitalization”, a concept often associated or used as synonymous. According to Salento (2018), there is a close relationship between high technology and digitalization: the former is seen as a key element for economic development and value creation; while the latter, on the basis of technological progress, should create a society compatible with this progress and avoid the increase of inequalities that lead to significant negative consequences. 3. High-tech Research: Objectives and Data Description Existing literature discusses various definitions of high-tech. This paper doesn’t go through them, which would need extended comments. On the contrary, considering some common characteristics emerging from the literature, some hypotheses are fixed that should hold across different definitions. The first hypothesis is about the fact that the territory plays a prominent role in the development of high technology. Thus, considering Europe, it is expected that regional disparities in high-tech intensities exist, as well as disparities inside the territories of a single country. The second hypothesis is about the localization factors of high-tech. It is expected that high-tech is especially affected from externalities. In particular, high-tech is relevant when some factors are present on the territory like existence of universities, laboratories, large enterprises, services and demographic intensity. Summing up, a complex interaction between social, political and economic forces is the critical positive factor of high-tech. The third hypothesis is that high-tech is a phenomenon leading to economic growth, in particular it implies the ability of a firm to remain competitive, to Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 100 renew itself, to be innovative, thus the hypothesis is that a high level of the indicator is a signal of the ability to recover of a territory as well of better resilience in crisis periods. The aim of the paper is to construct in a rigorous statistical way a high-tech indicator by verifying the above-mentioned hypotheses. The analysis has been carried out using data on socio-economic characteristics at European regional level provided by Eurostat. To support its primary objectives (growth and employment, promotion of territorial cooperation and reduction of the disparity between the European Regions) Eurostat has organised the European Union into territorial units for statistics, establishing the so-called NUTS (Nomenclature of Territorial Units for Statistics) classification1. As previously mentioned, among the most important statistics developed by Eurostat are regional statistics. As they are better able to highlight the disparities and similarities between EU Member States than in a comparison between nations, where there is often a risk of comparing small states with large ones, a strong focus has been placed on them in this work. Eurostat provide a wide range of socio-economic data, covering different area: the “science and technology” one is the most interesting to build a composite indicator. In detail, the indicators and filters considered are shown in Table 1. In the analysis carried out, starting from the NUTS 2 classification, 238 regions belonging to 25 of the 28 Member States of the European Union were taken into account. Due to a lack of data, such countries have been excluded: Greece, Lithuania and Slovenia. Furthermore, candidate countries and potential candidates for accession to the European Union were not taken into account. The analysis period starts in 2006 and ends in 2016. This time period makes it possible to investigate the possible impact of crisis phases on the development of high technology. Moreover, the choice of this period of time allows to have available all the elementary indicators set out in Table 1 and, consequently, to build a more complete picture of the phenomenon under study. 4. Methodology This section deals with the construction of high-tech composite indicators, followed by the analysis of the results obtained. First, an overview of the theoretical framework has been provided, other than different techniques to select elementary indicators, i.e., those that are sufficient and suitable to describe the phenomenon (Mazziotta, Pareto, 2019a). In fact, by means of multivariate 1. The classification assigns a specific code and name to each territorial unit and subdivides the EU Member States into NUTS level 1 territorial units, each of which is subdivided into NUTS level 2 territorial units, which in turn are subdivided into NUTS level 3 territorial units. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 101 analysis, the variables needed to describe different dimensions of the phenomenon are appropriately selected and inserted in the composite indicator model. The construction process occurs using different methods that are then compared in order to identify the one that meets the requirements of temporal and spatial comparability as well as robust results. 4.1. Composite Indicator Construction The first step for the construction of the composite indicator is the development of a theoretical framework. A formative model of measurement has been developed in which elementary indicators are the cause of the phenomenon (Mazziotta, Pareto, 2019b). The elementary indicators of regional science and technology selected have been explained in Table 1 and all of them have positive polarity with the phenomenon (i.e., there is a positive relationship between indicators and “hightech”) and, therefore, no mathematical transformation is required. As previously stated, several studies (in particular Marullo, Perugi, 2011) define high technology through the level of R&D spending, the high technology employment, the specialized human resources and the ability to exploit the results of innovation (i.e., the patents intensity). Table 1 – Regional Science and Technology Statistics Indicator Filters Unit of Measure Intramural R&D expenditure (GERD) • NUTS 2 regions • Sector of performance (all sectors) % GDP Employment in technology and knowledge-intensive sectors • NUTS 2 regions • Employed people between 15 to 74 years • High technology sectors (high technology manufacturing and knowledge-intensive high technology services) (NACE classification) % total employment Human resources in science and technology • NUTS 2 regions • People between 15 to 74 years • Tertiary level education (ISCED classification) and/or employed in science and technology • Managers excluded (ISCO classification) % active population (in 15-74 age group) EU trademark applications • NUTS 2 regions % total population Community designs • NUTS 2 regions % total population Source: Author’s elaborations from Eurostat Regional Statistics Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 108 Figure 3 – High-tech Composite Indicator (AMPI method) in 2016 Notes: The classes and the correspondent range of high-tech indicator values (AMPI Method) are: first class from 87.395 to 90.874, second from 90.875 to 94.353, third from 94.354 to 97.832, fourth from 97.833 to 101.311, fifth from 101.312 to 104.790, sixth from 104.791 to 108.269, seventh from 108.270 to 111.748, eight from 111.749 to 115.227, ninth from 115.228 to 118.706 and tenth from 118.707 to 122.185. Source: Author’s elaborations Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 109 Figure 4 – Ten Worst and Best Regions in 2006/2016 and Their Trends Worst Best 2006 80 82 84 86 88 90 92 94 96 98 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 RO21 (Nord-Est) RO41 (Sud-Vest Olt enia) RO31 (Sud - Munte nia) RO22 (Sud-Est) PT16 (Centro - PT) RO11 (Nord-Vest) PT18 (Alentejo) PT11 (Norte) RO12 (Centru) PL33 (Swietokrzysk ie) 106 108 110 112 114 116 118 120 122 124 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 FI1B (Helsinki-Uusimaa) DE21 (Oberbayern) DK01 (Hovedstaden) SE11 (Stockholm) BE31 (Prov. Brabant wallon) FR1 (Île de France) DE11 (Stuttgart) UKJ1 (Berkshire, Buckinghamshire and Oxfordshire) DE12 (Karlsruhe) SE22 (Sydsverige) 2016 80 82 84 86 88 90 92 94 96 98 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 RO21 (Nord-Est) RO22 (Sud-Est) RO31 (Sud - Muntenia) RO41 (Sud-Vest Oltenia) BG34 (Yugoiztochen) ITF6 (Calabria) ITF4 (Puglia) ITG1 (Sicilia) RO11 (Nord-Vest) CZ04 (Severozápad) 106 108 110 112 114 116 118 120 122 124 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 DK01 (Hovedstaden) SE11 (Stockholm) FI1B (Helsinki-Uusimaa ) LU00 (Luxembourg) UKJ1 (Berkshire , Buckinghamshir e and Oxfordshire) DE21 (Oberbayern) BE31 (Prov. Brabant wallon) CZ01 (Praha) DE3 (Berlin) DE11 (Stuttgart) Source: Author’s elaborations Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 110 The ten best regions in 2006, in descending order of technological content, belong to Northern European countries and are: FI1B (Helsinki-Uusimaa), DE21 (Oberbayern), DK01 (Hovedstaden), SE11 (Stockholm), BE31 (Prov. Brabant wallon), FR10 (le de France), DE11 (Stuttgart), UKJ1 (Berkshire, Buckinghamshire and Oxfordshire), DE12 (Karlsruhe) and SE22 (Sydsverige). The interval of values is between 108.22 and 115.624. All regions show a fluctuating trend over the entire period and their positions in the ranking are almost unchanged. As for the worst regions, there is a decrease in scores in 2008, coinciding with the recessionary phase that affected the entire European economy. Two peculiarities that immediately appear from the trend graph are the behaviours assumed by UKJ1 (Berkshire, Buckinghamshire and Oxfordshire) and BE31 (Prov. Brabant Wallon). The first shows significant growth from 2008 onwards and then decreases again in 2015, while the second has a positive peak of about 6 points in 2011 compared to 2010. Moreover, the Province Brabant Wallon (BE31) exhibits a significant growth and decrease over time. As regards the variation they have undergone over time, both positive and negative differences can be noted (Figure 4). Regions in the top two positions of the ranking have experienced a slight positive increase: Helsinki-Uusimaa (FI1B) and Oberbayern (DE21) reach respectively 0.207 and 0.449. Different is the positive change of Hovedstaden (DK01), the greatest among the ten regions (equal to 3.865). Finally, the decreases in technological content occurring between 2006 and 2016 are slight and in a range between -0.606 and -1.402. Analysing 2016, among the ten worst regions there are still those of Romania that had already been in this position in 2006. However, the positions covered by the regions of Portugal and Poland have been replaced by BG34 (Yugoiztochen), ITF6 (Calabria), ITF4 (Puglia), ITG1 (Sicily) and CZ04 (Severozápad). The range of values is comprised between 87.97 and 92.537. From the lowest to the highest rating, there are: RO21 (North-East), RO22 (South-East), RO31 (South-Muntenia), RO41 (South-West Oltenia), BG34 (Yugoiztochen), ITF6 (Calabria), ITF4 (Puglia), ITG1 (Sicily), RO11 (NorthWest) and CZ04 (Severozápad). These regions show a fluctuating trend over time. For what concern the best regions, there are: DK01 (Hovedstaden), SE11 (Stockholm), FI1B (Helsinki-Uusimaa), LU00 (Luxembourg), UKJ1 (Berkshire, Buckinghamshire and Oxfordshire), DE21 (Oberbayern), BE31 (Prov. Brabant wallon), CZ01 (Praha), DE30 (Berlin) and DE11 (Stuttgart). Part of them have already been the best in 2006 and, as underlined in Figure 4, a growing trend can be seen in most cases since 2008. Looking at the best and the worst, both show positive and negative fluctuations compared to 2006. In particular, the Italian regions have worsened over time, as for Severozápad (CZ04), while the others Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 111 report a positive change. Among the top ten, only Stuttgart (DE11) records a negative change of -0.606, while the others show favourable variations. The shift observed in Luxembourg (LU00) is interesting and equivalent to +7.273. Summing up, as regards regional disparities, the territorial distribution of the values of the high-tech indicator confirms that more intense presence is observed where there is a interaction of context factors and structural characteristics. In fact, it is possible to notice that high technology content is recorded in European capitals or big cities where statistical data show high population density, elevated household disposable income, low unemployment rate, presence of universities and research centers. For example, Northern capitals show these characteristics: Berlin, defined as ‘Digital Friendly capital’ and Silicon Valley of Europe; Munich is another city constantly evolving in terms of innovation and high technology. Focusing on the trend, a downturn or stability during the crisis is observed in several regions; an upward is observed soon after the crisis. Thus, the tendency supports the hypotheses of the important and critical role of high-tech in the recovering phase. Indeed, most relevant upward is registered where high-tech indicator shows a considerable importance of this phenomenon. Focusing on Italy, in 2006 and 2016 the worst region is Calabria (ITF6), positioned in the European ranking at 214th and 233rd respectively with a composite indicator value of 93,024 and 92,064. In 2006, the best is Friuli Venezia Giulia (ITH4) in 33rd position with a high-tech score of 104,578, while in 2016 Lombardia (ITC4) is the one to gain the supremacy with its 66th position and a value equals to 102,544. These results underline that the presence on Italian soil of a gap between North and South in terms of technology is in line with the hypotheses considered in this paper about regional factors disparities in high-tech. 5. Conclusions This research has constructed, in a rigorous way, a high-tech composite indicator, devoting particular interest to different methods that can be used, as well as the steps needed to compute an indicator correctly. The literature argues that there are ten steps to construct a composite indicator, starting from the theoretical definition to the various techniques of representation and dissemination of results. There are many advantages to using these methods of dimensional reduction, but it is equally true that there can be many problems if the choices made during the construction process are not dictated by solid and well understood study bases. In particular, accuracy and adequacy in the selection of elementary indicators assumes a relevant role: data has to be of good quality and in relevant quantities, so sufficient and appropriate to describe the phenomenon. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 112 The composite indicator approach has been applied to the high-tech phenomenon in European Regions, using Eurostat data from 2006 to 2016. Following a careful methodological analysis, composite indicators have been constructed with different methods and subsequently compared: the study has demonstrated that, in this case, the best one is the Adjusted Mazziotta-Pareto Index (AMPI) and therefore, has been used for the analysis of the results. This aggregation technique satisfies the requirements of spatial and temporal comparability, as well as robustness of the results (an analysis of the variation coefficients has identified that the AMPI shows lower values than the others and this indicates that the estimates are the most precise). High technology is a multidimensional phenomenon that has no single definition: literature states that several meanings can be attributed to this concept, depending on the various approaches used. However, a number of authors stress the importance of defining high technology unambiguously, given its considerable relevance as an indicator of economic development. Indeed, researchers claim that it can be conceived as the result of an interaction of social, political and economic forces, as well as the ability of companies to remain competitive in the market. The indicators values confirm the hypothesis of the existence of regional disparities in Europe, the localization of high-tech seems to be related to the characteristics of the external context. Moreover, high-tech is suffering crisis effects in a smoothed way and is fast recovering. In this regard, the results show that the “high-tech” regions of Europe are those of the North, while those of the East have the lowest scores of the composite indicator (i.e., low high-tech). Trend in the considered period is different. The detailed analysis of the results (paragraph 4.2) is useful to get specific insights for general conclusions. This research has reached its objectives through a theoretical study of the composite indicators’ construction process and then the implementation of what has been learned. The analysis provided a synthetic measure of the high technology phenomenon by building a new composite indicator: this innovative approach is an informative contribution at regional policy level. This makes it possible to identify the disparities between European Regions in technological terms and to understand their development and competitive capacities. This new composite indicator has been constructed by including in the model factors that have been chosen on the basis of a detailed analysis of the literature: the variables used gauge different dimensions of the concept and so, the synthetic measure is well representative. Furthermore, it is important to point out again that, although the geographical coverage of this research is wide, it is limited by the lack of some European Regions and countries (Greece, Lithuania and Slovenia) in the data matrix. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 113 References Assolombarda (2018), La Lombardia nel confronto europeo. Milano: Assolombarda, Booklet Ricerca e Innovazione n. 03/2018 – www.assolombarda.it. Assolombarda (2019), La Lombardia nel confronto europeo. Milano: Assolombarda, Booklet Ricerca e Innovazione n. 4/2019 – www.assolombarda.it. Erlhoff M., Marshall T. (eds.) (2007), Design dictionary. Berlin, Basel: Birkhäuser. https://doi.org/10.1007/978-3-7643-8140-0. Eurostat (2018), Statistics explained: high-tech statistics – economic data – https://ec.europa.eu/eurostat. Eurostat (2020), Glossary: high tech – https://ec.europa.eu/eurostat. Johnson D.M., Porter A.L., Roessner D., Newman N.C., Jin X.Y. (2010), Hightech indicators: Assessing the competitiveness of selected european countries. Technology Analysis and Strategic Management, 22, 3: 277-296. https://doi. org/10.1080/09537321003647313. Joseph R.A. (1988), The politics of defining high technology in Australia. Science and Public Policy, 15, 5: 343-353. https://doi.org/10.1093/spp/15.5.343. Kotro T., Pantzar M. (2002), Product development and changing cultural landscapes-is our future in “snowboarding”? Design Issues, 18, 2: 30-45. https://doi. org/10.1162/074793602317355765 Marullo C., Perugi R. (2011), High tech e territorio. Il ruolo delle città nelle dinamiche di localizzazione delle imprese ad alta tecnologia – www.academia.edu. Mazziotta M., Pareto A. (2017), Synthesis of indicators: The composite indicators approach. In: Maggino F. (ed.), Complexity in society: From indicators construction to their synthesis. Cham: Springer – Social Indicators Series Research. 159-191 – https://doi.org/10.1007/978-3-319-60595-1_7. Mazziotta M., Pareto A. (2019a), Metodi per la costruzione di indici sintetici: teoria e pratica (ISTAT). Slides presented at Corso SAES held in Palermo. Italy: June. Mazziotta M., Pareto A. (2019b), Use and misuse of PCA for measuring well-being. Social Indicators Research, 142: 451-476. https://doi.org/10.1007/s11205-018-1933-0 Mendona S., Pereira T.S., Godinho M.M. (2004), Trademarks as an indicator of innovation and industrial change. Research Policy, 33, 9: 1385-1404. https://doi. org/10.1016/j.respol.2004.09.005 Millot V. (2009), Trademarks as an indicator of product and marketing innovations. Paris: Oecd, Statistical Analysis of Science, Technology and Industry, STI Working Paper n.2009/6. OECD (2008), Handbook on constructing composite indicators. Methodology and user guide. Paris: Oecd Publications. Porter A.L., Roessner J.D., Newman N., Cauffiel D. (1996), Indicators of high technology competitiveness of 28 countries. International Journal of Technology Management, 12, 1: 1-32. https://doi.org/10.1504/IJTM.1996.025477. Saisana M., Tarantola S. (2002), State-of-the-art report on current methodologies and practices for composite indicator development, EUR 20408 EN – Italy. Brussels: European Commission-JRC. Salento A. (ed.) (2018), Industria 4.0: Oltre il determinismo tecnologico. Bologna: TAO Digital Library. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 114 Steenhuis H.J., De Bruijn E.J. (2006), High technology revisited: definition and position. Paper presented at IEEE International Conference on Management of Innovation and Technology, Singapore. 1080-1084. https://doi.org/10.1109/ICMIT.2006.262389. Sommario L’indicatore composito dell’alta tecnologia (HTCI). Uno strumento per misurare le disparità regionali europee Gli indicatori compositi sono uno strumento per le strategie politico economiche territoriali in quanto consentono la riduzione dimensionale di fenomeni socio-economici complessi non direttamente misurabili con singoli indicatori elementari. Questo paper si focalizza sulla misurazione e studio dell’alta tecnologia nelle regioni Europee: un nuovo indicatore dell’alta tecnologia è costruito e utilizzato per un’analisi spaziale e temporale. In particolare, gli indicatori proposti considerano il periodo 2006-2016 per le regioni d’Europa. Attraverso l’analisi statistica di questo indicatore sono verificate alcune ipotesi relative alle disparità territoriale dell’high-tech, al loro trend e fattori di sviluppo. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 115 Italian NEETs: An Analysis of Determinants Based on the Territorial Districts Giuseppe Cinquegrana*, Giovanni De Luca°, Paolo Mazzocchi°, Claudio Quintano§, Antonella Rocca° Sommario This paper aims at analyse the NEET phenomenon (young people not in employment, education or training) in the post-2007 financial crisis in Italy, using municipality as unit of analysis. Through new databases with high territorial detail made available by the Italian National Institute of Statistics (ISTAT) and Ministry of Education, we regress the municipal NEET rate on a selection of municipal and provincial social, economic and education indicators. We used a multi-level model to account for the nested structure of data (municipalities nested into the provinces). Results highlight the importance on NEETs, in particular, of indicators measuring the effectiveness of the education system. 1. Introduction1 NEETs, that is young people not in employment, education and training, are a significant share of the total youth population, especially in such Southern European countries – and in particular in Italy – which are among the countries more hit by the 2007 financial crisis. NEETs represent a relevant economic loss for each country because this condition may affect also their future career prospects. First pioneering studies on the current socio-economic crisis due the Covid-19 pandemic – which occurred when the recovery from the previous crisis * Istat – Italian National Institute of Statistics, Rome, Italy, e-mail:[email protected]. § University of Naples Suor Orsola Benincasa, Department of Legal Sciences, Naples, Italy, email: [email protected]. ° University of Naples Parthenope, Department of Management and Quantitative Studies, Naples, Italy, e-mail: [email protected]; [email protected]; roc- [email protected] (corresponding author). 1. A preliminary version of this paper was presented at the XLI Scientific Web Conference of the Italian Association of Regional Sciences, held on the 2nd-4th of September 2020. The authors wish to thank all seminar participants and, above all, Antonella Bianchino, for some valuable suggestions. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 116 was still not completed – have demonstrated that the crisis is producing, besides strong economic effects on firms, relevant increases in socio-economic inequalities, because the most vulnerable segments of population, such as young people, migrants and women, result more invested (Tamesberger, Bacher, 2020; Eurofound, 2020; Shanahan et al., 2020). Young people are very disadvantaged in comparison to their adult peers because in entering the labour market they lack of job experience and in experience in job search. Even young workers are in a disadvantaged condition because they are usually more widespread among precarious jobs, more easily fired in time of crisis (Quintini et al. 2007; Scarpetta et al., 2010). However, young generations are the key drivers for future economic growth and development (Boulianne, Theocharis, 2018; ILO, 2020) and their contribution to each country economic growth and development is nowadays still more urgent, considering the digital revolution in progress, still more stimulated by the Covid-19 pandemic (Iivari et al., 2020). The aim of this paper consists in identifying the factors which mainly affect the distribution of young NEETs in Italy. In the last years, a wide stream of literature has studied young people condition in the labour market. Economists, sociologists and psychologists have till now mainly investigated the causes leading to the NEET status acting at individual level and due to the socio-economic context, that is the labour market conditions, education and the institutions regulating the school to work transition. However, these studies were unable to explain why young people living within the same country, sharing the same institutions and having identical personal characteristics manifest so many different propensities to the NEET condition. In this paper, we propose a new approach, focused on a spatial perspective, in order to account for the influence in the NEET propensity exerted by the place where the individuals live. Recently, the Italian National Institute of Statistics (ISTAT) has made available data at provincial and municipal level. Referring in particular to “A misura di comune”, that is a multi-source experimental statistical information system, we look at the share of NEETs in each municipality and verify the relationship with many other socio-economic indicators observed with a municipal or a provincial detail. The empirical evidence shows a strong concentration of NEETs in the South of Italy, but we demonstrate that the North-South divide is not the only key-lecture in explaining the stronger variability in the NEET phenomenon. Focusing the analysis on a single national domain, with identical policies and laws, but also strictly homogeneous in the cultural and social aspects, we can better identify the role of the place of residence in terms of degree of urbanisation and socioeconomic aspects directly ascribable to it. The outline of the paper is as follows. Section 2 clarifies the concept of NEETs while Section 3 focuses on the Italian regional disparities. Section 4 Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 117 shows methodology and data. Finally, Section 5 presents the results and Section 6 concludes. 2. Inactivity and Unemployment The NEET indicator refers to the condition of unemployment or inactivity, out of education. Even if unemployment and inactivity are very different conditions, their effect is the same and consists in total disengagement from the labour market. The age class involved in the NEET identification, initially limited to 16-18 years, has been extended to 15-24 or even 15-29 years (see Yearly Report 2019, Istat, and for a detailed collection of NEET data by age-class, see on http://dati-congiuntura. istat.it) in reason of the recent more prolonged stay of young people in education and of the increase of the mean duration of the school-to-work (STW) transition. STW represents the period from the end of studies to the attainment of a stable job and, in this paper, we refer to 15-29 age class. When individuals leave school, they may decide to enter the labour market and starting the job search or may decide to remain inactive. Therefore, after completing the studies and until the achievement of a stable job, young people are in the NEET status if they are not involved in occasional jobs or in brief experiences of training and apprenticeship. More prolonged is the period of STW transition, higher is the share of NEETs. This explains why high shares of NEETs are usually linked to the long-standing structural problems of the youth labour market, which holds to high levels the youth unemployment rates and makes the transition from school to work slow and problematic (Bratti et al., 2008; Caroleo, Pastore, 2012; Hadjivassiliou et al., 2018; Piopiunik, Ryan, 2012; Choudhry et al., 2012). Many economists have studied the NEET issue referring almost exclusively to unemployment. However, the expansion of the focus from unemployment to the broader concept of NEET responds to the need to involve in the analysis also youth who have given up looking for a job or who are unwilling to join the labour market (UCW, 2013). According to the consequences of the NEET status, at individual level, it drives towards marginalization and exclusion from the labour market (Eurofound, 2012; Thompson, 2011), impoverishing human capital and reducing the probabilities of future engagement at work, with potential scarring effects on successive generations and concomitant economic and social impacts (Ryan, 2001; Manfredi et al., 2010; Gregg, Tominey, 2004). At macro-level, it induces to a loss of economic productivity and growth (Eurfound, 2012). However, the distinction between unemployment and inactivity assumes relevance because when the status of NEET derives from unemployment, it depends only by the incapacity of the labour market to satisfy the labour offer or to stimulate the match between the demand and labour offer. Conversely, inactivity is a Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 124 useless, and simple regression is sufficient. Otherwise, when the ICC approaches 1, no variance exists to explain the share of NEETs at the municipal level. 5. Results In the first step, we limit to analyse separately the effects on the NEET rates of the economic, educational and social dimensions. These relationships should be interpreted only in terms of correlation because we cannot exclude some endogeneity issues. The economic dimension allows better than the other dimensions to catch the differences among the groups. The ICC indicated that the 42% of the total variability in the share of NEETs is captured by the groups identified. Municipalities with the higher entrepreneurial rates, where the share of individuals employed in the high-technology sector is higher, with a high patent production and with a high degree of attractiveness show the lower NEET rates. The direct relationship between the NEET rate and the endowment of capital resources suggests instead that the only economic assets are not able to create the conditions in reducing the share of NEETs. Model 2 analyses the connection between NEET rates and the characteristics of the education system. It presents the highest Wald statistic, demonstrating the very strong connection between, on the one side, a high educational attainment and low NEET rates and, on the other side, lower NEET rates where the education system shows a major capacity to transfer the educational competencies measured by tests. The very high correlation between the scores for numerical and literal competences (0.985) suggested to introduce into the model only one of them. Finally, with reference to the social dimension, Model 3 shows the best fit according to the AIC and BIC indicators. The regressors included in the model concern the percentages of electoral participation, of separate waste collection and no-profit organisations. They all show an inverse and significant relationship highlighting lower resilient attitude towards the environment and the society in places with higher NEET rates. The last model includes a selection of all the previous indicators chosen according to their statistical significance. In a first step, we inserted all the variables included in models 1, 2 and 3 and added also the indicators of the degree of urbanisation. Subsequently, we proceeded removing those variables which appeared no more significant. Surprisingly, living in a densely populated area (city) increases the probability of being NEET. In other words, Italian cities, instead of representing job opportunities catalysts for young people, are the places where the lack of job opportunities and social exclusion are maximum. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 125 Table 1 – Multilevel Models on NEET Determinants Observed on 7,842 Italian Municipalities NEET Model 1 Model 2 Model 3 Model 4 Economic dimension Entrepreneurial rate -0.007** Attractiveness index -0.12* Patent intensity -0.035*** -0.004 High tech employees -0.072*** -0.042*** Endowment capital resources 0.001* Education dimension Numerical competences -0.317*** -0.099** High secondary school graduated 2564 (%) -0.151*** -0.154*** Tertiary educated 30-34 % -0.037*** -0.041*** Social dimension Separate waste collection -0.094*** -0.081*** Electoral participation -0.316*** -0.215*** No-profit organisations -0.044*** Geographical indicators (ref. rural area) City 2.562*** Town 0.417** Constant 34.803*** 104.195*** 57.812*** 77.557*** Var(_cons) 24.09 12.81 9.674 8.875 Var(res) 33.43 31.62 33.558 31.523 Wald chi2 62.49*** 568.09*** 181.12*** 704.58*** AIC 49509 49661 49441.93 48936.30 BIC 49565 49703 49483.65 49019.73 ICC 0.419 0.288 0.224 0.220 Source: Authors’ ad hoc elaborations on ISTAT and MIUR data Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 126 6. Conclusions Analysing the NEET phenomenon is very challenging because of the complexity of the causes originating it. In this paper, we have proposed a new key lecture, analysing the share of NEETs using as unit of analysis the municipalities and referring to municipal and provincial indicators connected to the labour market, education and social dimensions. Further, the municipal detail also allowed us to control for the effect on the NEET rates of the degree of urbanisation. Results highlight the strong effect on the NEET rates of factors linked to the economic vitality of territory (entrepreneurial rate, attractiveness, share of employees in the high-tech sector and patent intensity), to the outcomes connected with the education systems (in terms of share of high educated and of mean scores got for numerical competences) and to the social participation (electoral participation, separate waste collection and no-profit organisation). Finally, as the NEET rates appear to be higher in densely populated areas than in the rural ones, this result suggests that Italian cities are a catalyst for social exclusion, rather than hubs for innovation and job opportunities. However, all these results need to be elaborated on. In reason, above all, of the significant gap between the North and the South of Italy, it should be interesting to analyse separately the Italian macro-regions (North and South) or to adopt a different hierarchical structure in the multi-level models, accounting also for the regional dimension. References Aiello F., Bonanno G. (2017), Multilevel Empirics for Small Banks in Local Markets. Papers in Regional Sciences, 97, 4: 1017-1037. Doi: 10.1111/pirs.12285. Bacher J., Koblbauer C., Leitgöb H., Tamesberger D. (2017), Small differences matter: how regional distinctions in educational and labour market policy account for heterogeneity in NEET rates. Journal for Labour Market Research, 51, 4: 1-20. Doi: 10.1186/s12651-017-0232-6. Balan M. (2015), Methods to Estimate the Structure and Size of the “NEET” Youth. Procedia Economics and Finance, 32: 119-124. Doi: 10.1016/S2212-5671(15)01372-6. Bertola G., Blau F.D., Kahn L.M. (2007), Labor Market Institutions and Demographic Employment Patterns. Journal of Population Economics, 20: 833-867. Doi: 10.1007/s00148-007-0137-8. Bickel R. (2007), Multilevel Analysis for Applied Research, It’s Just Regression!. Guilford Press: New York. Billari F. (2004), Becoming an Adult in Europe: A Macro/micro-demographic perspective. Demographic Research, 3, 2: 15-44. Doi: 10.4054/DemRes.2004.S3.2. Boesler M. (2014), What the Unemployment and Labor Force Participation Rates Would Be If All of the Discouraged Workers Came Back. Business Insider, accessed on January 2019, www.businessinsider.com/unemployment-rate-i. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 127 Bonanomi A., Luppi F. (2020), A European Mixed Methods Comparative Study on NEETs and Their Perceived Environmental Responsibility. Sustainability, 12, 2: 515. Doi: 10.3390/su12020515. Boulianne S., Theocharis Y. (2018), Young People, Digital Media, and Engagement: A Meta-Analysis of Research. Social Science Computer Review, 38, 2: 111-127. Doi: 10.1177/0894439318814190. Bratti M., Checchi D., de Blasio G. (2008), Does the Expansion of Higher Education Increase the Equality of Educational Opportunities? Evidence from Italy. Labour, 22 (special issue): 53-88. Doi: 10.1111/j.1467-9914.2008.00411.x. Brofenbrenner U. (1979), The Ecology of Human Development, Experiments by Nature and Design. Harvard University Press. Brunello G., De Paola M. (2014), The Costs of Early School Leaving in Europe. IZA Journal of Labor Policy, 3, 22: 1-31. Doi: 10.1186/2193-9004-3-22. Caroleo F.E., Pastore F. (2012), Talking about the Pigou Paradox. Socio-Educational Background and Educational Outcomes of AlmaLaurea. International Journal of Manpower, 33, 1: 27-50. Doi: 10.1108/01437721211212510. Caroleo F.E., Quintano C., Mazzocchi P., Rocca A. (2020), Being NEET in Europe before and after the Economic Crisis: An Analysis of the Micro and Macro Determinants. Social Indicator Research, 3. Doi: 10.1007/s11205-020-02270-6. Cefalo R., Scandurra R., Kazepov Y. (2020), Youth Labour Market Integration in European Regions. Sustainability, 12, 9, 3813. Doi: 10.3390/su12093813. Choudhry M., Marelli E., Signorelli M. (2012), Key Determinants of Youth Unemployment in OECD Countries – www.aiel.it Davig T., Mustre-del-Río J. (2013), The Shadow Labor Supply and Its Implications for the Unemployment Rate. Federal Reserve Bank of Kansas City Economic Review, Third quarter – www.KansasCityFed.org. De Luca G., Mazzocchi P., Quintano C., Rocca A. (2019), Italian NEETs in 2005-16: has the Youth Guarantee Fund Recent Labour Market Reforms Produced any Effect?. CESIFO Economic Studies, 65, 2: 154-176 . Doi: 10.1093/cesifo/ifz004. Elder S. (2015), What Does NEETs Mean and Why Is the Concept So Easily Misinterpreted? Work 4 Youth Technical Brief. Geneva: ILO. Eurofound (2012), NEETs: Young People Not in Employment, Education and Training: Characteristics, Costs and Policy Responses in Europe. Luxembourg: Publications Office of the European Union. Eurofound (2020), Living, Working and Covid-19. Covid-19 series. Luxembourg: Publications Office of the European Union. Finegan T.A. (1978), Should Discouraged Workers Be Counted as Unemployed? Challenge, 21, 5: 20-25. Doi: 10.1080/05775132.1978.11470464. Gregg P., Tominey E. (2004), The Wage Scar from Youth Unemployment. Labour Economics, 12, 4: 487-509. Doi: 10.1016/j.labeco.2005.05.004. Hadjivassiliou K.P., Tassinari A., Eichhorst W., Wozny F. (2018), How Does the Performance of School-to-Work Transition Regimes in the European Union Vary? In: O’Reilly J., Leschke J., Ortlieb R., Seeleib-Kaiser M., Villa P. (eds.), Youth Labor in Transition. New York: Oxford University Press. Doi: 10.1093/oso/9780190864798.003.0003. Heck R.H., Thomas S.L. (2000), Quantitative Methodology Series: An Introduction to Multilevel Modeling Techniques. Mahwah: Lawrence Erlbaum Associates Publishers. Doi: 10.4324/9781410604767. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 128 Iivari N., Sharma S., Ventä-Olkkonen L. (2020), Digital Transformation of Everyday Life – How Covid-19 Pandemic Transformed the Basic Education of the Young Generation and Why Information Management Research Should Care? International Journal of Information Management, 55: 102183. Doi: 10.1016/j.ijinfomgt.2020.102183. ILO (2020), Global Employment Trends for Youth 2020, Technology and future of jobs. Geneva: ILO. Istat (2019), Le differenze territoriali di benessere – Una lettura a livello provinciale. Roma: Istat – www.istat.it. Manfredi T., Scarpetta, S., Sonnet A. (2010), Rising Youth Unemployment During the Crisis: How to Prevent Negative Long-Term Consequences on a Generation? Paris: OECD Social, Employment and Migration Working Papers, n.106. Pastore F. (2019), Why So Slow? The School-to-work Transition in Italy. Studies in Higher Education, 44, 8: 1-14. Doi: 10.1080/03075079.2018.1437722. Pastore F., Quintano C., Rocca A. (2020), Stuck at a Crossroads? The Duration of the Italian School-to-work Transition. International Journal of Manpower. Doi: 10.1108/IJM-05-2020-0199. Piopiunik M., Ryan P. (2012), Improving the Transition between Education/Training and the Labour Market: What Can We Learn from Various National Approaches? Report for the European Commission. EENEE Analytical Report n. 13 – www.eenee.de. Quintano C., Mazzocchi P., Rocca A. (2018), The Determinants of Italian NEETs and the Effects of the Economic Crisis. Genus, 74, 1:1-24. Doi: 10.1186/s41118-018-0031-0. Quintini G., Martin J.P., Martin S. (2007), The Changing Nature of the School-to-work Transition Process in OECD Countries. Bonn: Institute for the Study of Labor, IZA Discussion Paper, 2582. Doi: 10.2139/ssrn.1884070. Rabe-Hesketh S., Skrondal A. (2008), Multilevel and Longitudinal Modeling Using Stata. College Station (TX): Stata Press. Ryan P. (2001), The School-to-Work Transition: A Cross-National Perspective. Journal of Economic Literature, 39, 1: 34-92. Doi: 10.1257/jel.39.1.34. Scarpetta S., Sonnet A., Manfredi T. (2010), Rising Youth Unemployment During The Crisis: How to Prevent Negative Long-term Consequences on a Generation? Paris: OECD. Social, Employment and Migration Working Papers, 106. Shanahan L., Steinhoff A., Bechtiger L., Murray A., Nivette A., Hepp U., Esner M. (2020), Emotional distress in young adults during the Covid-19 pandemic: Evidence of risk and resilience from a longitudinal cohort study. Psychological Medicine, 1-10. Doi: 10.1017/S003329172000241X. Simoes F., Brito do Rio N. (2020), How to Increase Rural NEETs Professional Involvement in Agriculture? The Roles of Youth Representations and Vocational Training Packages Improvement. Journal of Rural Studies, 75, April: 9-19. Doi: 10.1016/j.jrurstud.2020.02.007. Spielhofer T., Benton T., Evans K., Featherstone G., Golden S., Nelson J., Smith P. (2009), Increasing Participation: Understanding Young People Who do Not Participate in Education or Training at 16 or 17. London: National Foundation for Educational Research. Tamesberger T., Bacher J. (2020), Covid-19 Crisis: How to Avoid a “Lost Generation”. Intereconomics, 55: 232-238. Doi: 10.1007/s10272-020-0908-y. Thompson T. (2011), Individualisation and Social Exclusion: The Case of Young People Not in Education, Employment or Training. Oxford Review of Education, 37, 6: 785-802. Doi: 10.1080/03054985.2011.636507. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 129 UCW – Understanding Childrens’ Work (2013), NEET Youth Dynamics in Indonesia and Brazil A Cohort Analysis. Rome: UNICEF. Walsh K. (2010), Youth Measures. United Kingdom. EEO Review: European Employment Observatory. European Parliament Observatory. Sommario I giovani NEET in Italia: un’analisi delle determinanti a livello comunale Il presente lavoro propone una nuova chiave di lettura per l’identificazione delle determinanti del fenomeno dei NEET (giovani che non studiano e non lavorano) nel periodo post-crisi finanziaria del 2007 in Italia, il paese che, a livello europeo, presenta i tassi più elevati. Sulla base, infatti, di alcune banche dati dell’ISTAT e del Ministero dell’Istruzione che presentano un elevato dettaglio territoriale, si è proceduto ad identificare le determinanti della quota di giovani NEET osservata a livello comunale. A tal fine, si è adoperato un modello multilevel, con indicatori aventi dettaglio municipale e provinciale. I risultati evidenziano l’importanza dei fattori legati, oltre che al tessuto economico-produttivo del territorio, al funzionamento del sistema educativo. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 PART III – WELL-BEING AND SUSTAINABILITY Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 133 Regional Well-being and Sustainability: Insights from Italy Giovanni D’Orio*, Rosetta Lombardo*1 Abstract It is widely recognized that, to go beyond the usual income-related aspect of wellbeing, it is fundamental to consider well-being as a multidimensional phenomenon concerning several dimensions of people’s lives. Until recently, a number of countries and organizations proposed their own well-being measures and multidimensional well-being has been mainly studied at country level. However, the well-being of individuals living in the same country might differ from one region to another. The focus of the chapther is centered on the possibility of connecting the Well Being generated in all the Italian Regions in the period 2010-2015, estimated through a factor analysis, to some aspects of economic, social and environmental sustainability. 1. Introduction The Gross Domestic Product per capita has been considered, for a long time, the main instrument to measure a country’s economy. The awareness of the limitations of economic measures for assessing a country’s living conditions and overall well-being has spread in recent years. It is also emerged that an exclusive focus on the economic dimension of well-being gives no relevance to social, environmental and economic sustainability (Altken, 2019; Heys, 2019). The literature on well-being revolves around physical limitations which might inhibit the achievement of the desired level of well-being, while one of the fundamental aims of sustainability studies, for example, is to highlight ways to increase or maintain intergenerational well-being (Quasim, 2017). Sustainability is a relatively new concept emerging in the late 1980s around the time of the report of the UN World Commission on Environment and Development, * University of Calabria – Department of Economics, Statistics and Finance “Giovanni Anania” Arcavacata di Rende (CS), Italy, e-mail: [email protected]; [email protected] (corresponding author). Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 140 The construction of composite indicators is complex because of two principal criticalities. Firstly, the selection of important domains of well-being and the weights given to each domain in the aggregation procedure. Secondly, the choice of an adequate method of the aggregation. In order to try to limit arbitrariness in choosing the well-being dimensions, we consider the insights that emerge from the Equitable and Sustainable Wellbeing (BES) project, resulting from the collaboration between the Italian National Institute of Statistics (ISTAT) and the National Council for Economics and Labor (CNEL). In order to do not incur in the criticism of having chosen in an arbitrary manner the weights of the relevant well-being domains, and to limit the subjectivity in attribution of weights to each domain, we opt for equal weighting. Decancq and Lugo (2013) identify equal weighting as the preferred procedure when the theoretical scheme assigns to each indicator the same adequacy in defining the variable to measure and it does not allow hypotheses consistently derived on differential weightings and when the empirical knowledge is not sufficient for defining specific weights. We compute, in fact, a composite well-being index, for Italian regions, by using the factorial analysis (FA, hereafter). The FA is a statistical technique that aims at simplifying a complex data set by representing it in terms of a smaller number of underlying variables. It allows the study of correlations between large numbers of variables, grouping them around factors, so that they are arranged on factors highly correlated with each other (Dillon, Goldstein, 1984). This methodology permits to explain the variance of the phenomenon under analysis and it can summarize a set of sub-indicators while preserving the maximum possible proportion of the total variation in the original set. If we have p variables X1, …, Xp measured on a sample of n subjects, then variable xs can be written as a linear combination of m factors F1, …, Fm where m < p: x s = ks1F1 + … + ksmFm + w [1] where ks are the factor loadings for variable xs; w is the variability of x_s not explained by the factors. FA condenses the information contained in a matrix of correlation or variance/ covariance; it aims to identify statistically the latent, not directly observable dimensions of the observed phenomenon (Ivaldi et al., 2016). We compute our composite well-being index starting from a panel data of 49 variables1 that synthesize – through the AMPI method adopted by ISTAT – the original variables grouped into the twelve domains of the BES project. 1. The variables that synthetize the 129 variables of the 12 BES dimensions are 63. We do not consider some of those variables because data do not cover the needed time interval. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 141 3.2. Data Table 1 describes the variables2, presents the sector of pertinence in well-being (Economics, Eco; Social, Soc; Environment, Env), the average value at national level in 2010, the average value at national level in 2015 and the expected sign of effect of each variable on well-being. It is noteworthy to highlight some potential effects that these variables can have on well-being and to consider possible effects of variables that in a “well-being” approach are very important, but have been neglected in previous works. We will not discuss in details the role of some variables commonly used in studies on well-being, such as “income”, “employment” or “innovation”, since it is widely recognized that, for instance, “income” matters for well-being. Economic Well-being The set of indicators on economic well-being contains information on Average disposable income (per capita) of consumer households and on an Index of inequality of disposable income (Table 1). While the impacts of income inequality differ across various dimensions of well-being, reducing economic inequality will generally help to improve the well-being of a society. For us inequality is not just “economic”. Other indicators on “social” inequality are contained in the set of indicators labelled Well-being and minimum conditions that will be discussed in next paragraph. Well-being and minimum conditions, education, health and participation in the labour market The variables for this topic, illustrated in Table 1, are Well-being & minimum conditions from 1 to 4, Education from 1 to 5, Health from 1 to 5, Labour occupation 1 and Labour quality from 2 to 6. Here we want to focus on some effects on the need to associate the concept of social exclusion with a specific set of indicators. This will be useful to assess and monitor the problem of exclusion as a proxy of “negative” economic and social sustainability, as stated in the definition of the definition of the so-called “Laeken indicators”, established by the European Council in December 2001. These indicators are helpful to measure the progress made by European Regions on some agreed objectives in areas deemed crucial, such as the fight against poverty and social exclusion, health, education and participation in the labour market. In one word, “inequality”. 2. The full explanation of each of them is provided by ISTAT (https://www.istat.it/it/ benessere-e-sostenibilità/misure-del-benessere). Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 142 Table 1 – Variables, Meaning and Expected Signs Indicator Meaning BES Type Ita 2010 Ita 2015 Exp Sign Economic Well-being 1 Average disposable income (per capita) of consumer households eco 17,78 17,88 + Economic Well-being 2 Index of inequality of disposable income soc 5,7 6,3 - Education 1 Children of 4-5 years attending kindergarten soc 94,7 92,1 + Education 2 People aged 25-64 who completed at least second grade secondary school soc 55,1 59,9 + Education 3 People aged 30-34 who have obtained a university degree soc 19,9 25,3 + Education 4 People aged 18-24 who have only completed middle school and are not included in a training program soc 22,0 25,7 - Education 5 People aged 25-64 who participated in education and training activities in the 4 weeks prior to the interview soc 6,2 7,3 + Environment 2 Urban waste sent to waste disposal site env 46,3 26,5 - Environment 4 Availability of urban green env 31,1 30,9 + Environment 5 People ≥ 14 very or fairly satisfied of environmental situation where they live env 69,0 69,8 + Environment 7 Electricity consumptions generated by renewable sources env 22,2 33,1 + Environment 8 Urban waste subject to recycling env 35,3 47,5 + Health 1 Life expectancy at birth Soc 81,7 82,3 + Health 2 Life expectancy in good health at birth Soc 57,7 58,3 + Health 5 Life expectancy without limitations in activities at the age of 65 years Soc 9,0 9,7 + Innovation 1 Research intensity Eco 1,2 1,4 + Innovation 2 Employees with scientific-technological university degree Eco 13,4 15,9 +/- Innovation 3 Employees in creative businesses Eco 2,8 2,8 + Labour quality 2 Employees on temporary contracts and employees who started their current job at least five years before Eco 19,7 19,5 - Labour quality 3 Rate of incidence of employees with low pay Eco 11,2 10,5 - Labour quality 4 Rate of incidence of non-regular employees Eco 12,3 13,5 - Labour quality 6 Share of involuntary part-time employees on total employees Eco 7,3 11,8 - Labouroccupation 1 Employment rate of the population aged 20-64 years Eco 61,0 60,5 + Landscape 2 Index of illegal buildings Env 12,2 19,9 - Landscape 3 Share of agri-touristic farms Env 6,6 7,4 +/- Politics 1 People ≥14 that express confidence in the Italian Parliament Soc 3,4 3,4 + Politics 2 People ≥14 that express confidence in the judicial system Soc 4,6 4,0 + (Continues...) Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 143 Indicator Meaning BES Type Ita 2010 Ita 2015 Exp Sign Politics 3 People ≥14 that express trust in the political parties Soc 2,6 2,3 + Politics 7 Index of overcrowding of prisons Soc 151,0 105,0 - Security 1 Burglaries rate Soc 12,0 16,5 - Security 2 Pickpocketing rate Soc 5,1 7,7 - Security 3 Robbery rate Soc 1,4 1,4 - Securitymurders Murders rate Soc 0,9 0,8 - Services quality 2 Children 0-2 years old who have used the services for children Soc 13,6 12,6 + Services quality 3 Families that signalled difficulties to access at least 3 essential services Soc 7,0 7,4 - Services quality 4 Families that reported shortages in water supply Env 11,4 9,3 - Services quality 5 Seats-km available in all types of transportation Env 4983,7 4502,7 + Services quality 6 Percentage of users who expressed a satisfaction grade ≥8 for the public transportation they use Soc 16,0 14,2 + Social relationship 1 People ≥14 that express satisfaction with family relationships Soc 35,7 34,6 + Social relationship 2 People ≥14 that express satisfaction with their friendships Soc 25,4 24,8 + Social relationship 4 People ≥14 that participated at least one social activity in the last 12 months+ Soc 26,9 24,1 + Social relationship 5 People ≥14 that are very or fairly satisfied with the environmental situation of the area in which they live Soc 26,9 24,1 + Social relationship 6 People ≥14 that talk about politics or who are informed about politics at least once a week, who have participated online in consultations or votes on social or political problems or have read and posted opinions on social or political problems on the web in the last 3 months Soc 67,4 66,4 + Social relationship 8 People ≥14 who have financed associations in the last 12 months Soc 17,6 14,9 + Subjective Well-being 1 People ≥ 14 with a satisfaction score for life between 8 and 10 Soc 43,4 35,1 + Well-being & minimum conditions 1 People living in families with severe material deprivation Soc 7,4 11,5 - Well-being & minimum conditions 2 People living in overcrowded housing without services and with structural problems Eco 7,0 7,6 - Well-being & minimum conditions 3 Subjective evaluation index of economic difficulty Eco 17,4 15,4 - Well-being & minimum conditions 4 People < 60 living in very low labour intensive families Eco 10,6 11,7 - Source: ISTAT (...follows) Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 144 There are forms of non-economic inequality that may have significant implications in terms of well-being. In order to assess issues related to economic and social sustainability, it is therefore essential to consider, in a well-being indicator, information concerning equal opportunities as well as policies for the family and child poverty, unemployment benefits and poverty among mature workers, old-age pensions and poverty among the elderly. Full understanding of the real level of economic and social sustainability of well-being requires a proper investigation of the institutional system in the level and distribution of social rights. Social relationship and subjective well-being The variables for this topic, illustrated in Table 1, are Social Relations from 1 to 8 and Subjective well-being 1. Here we want to focus on some potential effects on social sustainability of “relational goods”. The theory of modern relational goods raises questions that are simple but of fundamental importance for the definition of specific targets in the realization of a well-being indicator. The production of relational goods, the multiplication of socialization and support opportunities that may reduce the discomfort of minors, young people, the elderly and families are, in all respects, essential areas of well-being. For these reasons, social relationships, as relational goods, are very important since, in this perspective, they may assume “materiality” when they are perceived as a “well-neing good”. Subjective well-being and social non-instrumental relations could signal, for example, that the time spent in personal relationships (affective, family, social), regardless of intrinsic motivations strongly influences our happiness. Therefore, using relational goods and subjective well-being into economic analyses produces important effects on the overall well-being. Environment, landscape and crime The variables for this topic, illustrated in Table 1, are Environment from 1 to 8, Landscape from 1 to 4, Security from 1 to 3 and Security-murders. Concepts of well-being and its connection with landscape and environmental features provide a wealth of information for popular phrases including “exercising outside is better than gym,” “a nice view from your hospital bed will aid recovery” and “living in a greener environment affects happiness.” These variables are related to environmental sustainability of well-being. Providing precise evidence for these statements and analysing what the real relationships are, is an ongoing challenge but it is quite evident that environmental and landscape factors may influence people’s quality of life. Landscape, natural beauty & scenery are connected to psychological well-being. A bulk of literature exists about people’s mental health and state of relaxation when looking at natural Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 145 landscape images or when being outside in areas of parkland, gardens or the ‘wilderness.’ At the same time, a person or group of people could feel a deep loss and grief when the environment in their community has dramatically changed. The physical health of a person increases with greater contact with nature. Ecosystems play a critical role in the recycling and redistribution of nutrients. Disruption of nutrient cycling can impair soil fertility, resulting in reduced crop yields. This impairs the nutritional status of households (medium certainty) and diet deficiencies (both macro and micro-nutrients) harm children’s physical and mental development. In turn, this can impair the livelihoods of farmers and limit the options open to their children. Toxic chemicals in water (especially the one derived from percolation of rubbish dump sites) and food can have adverse effects on various organ systems. Exposure to low concentrations of some chemicals (such as PCBs, dioxins and DDT) may cause endocrine disruption, interfering with normal human hormone mediated physiology and impairing reproduction. People are expected to be more satisfied with their life and happier if they feel safe and secure in well-kept, tidy and pleasant business or residential area. Understanding if crime is associated with well-being is also important. Criminal victimization and well-being may be linked to health outcomes. Experiencing violence or theft victimization is normally associated with significantly lower happiness and life satisfaction. This may influence negatively a victim’s overall quality of life and results in diminished well-being. Politics This set of variables, presented in Table 1 – Politics 1-7, includes people attitude towards Parliament, Judicial System and Political Parties. Furthermore, a variable (Share of women elected to Regional Councils) can be seen as a proxy of equal opportunities in politics, and two extra variables are relative to the effectiveness of judicial system (average duration of trials defined in ordinary courts) and to the rate of prison overcrowding. Innovation This set of variables, illustrated in Table 1 – Innovation 1-3, illustrates the Research intensity of each Region, the share of Employees with scientific-technological university degree and the share of Employees in creative businesses. Services quality This set of variables, illustrated in Table 1 – Service quality 1-6, is related to the use and perception of quality of some public services like Transportation, Childcare, Residential services for elderly people, and Water. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 146 4. Results and Discussion We utilize the factor analysis in order to obtain an overall well-being indicator (WB) for each Italian region from 2010 to 2015. 4.1. Application of Factor Analysis Table 2 reports the results of the factorial analysis employed to obtain the indicator. For an easier reading, we decided to present the first 9 of the 49 factors since they explain a high level of variability and they are the ones with an eigenvalue greater than 1. Just to be sure that “unclear” results like the previous one could not affect the stability and reliability of the Factor Analysis performed, we conducted a test of “rotation” of the factors. The so-called rotation (Ivaldi et al., 2016) is an important issue in the factorial analysis stability since it causes the reduction of factor loadings that already, in the first phase, were relatively small, and the increase of the absolute values of factor loadings that predominated in the first phase. In order to avoid some mathematical problems, a process of rotation of the axes can transform the factors. In fact, in an un-rotated solution every variable is explained by two or more common factors, while in a rotated solution each variable is summarized by a single common factor (Ivaldi et al., 2016). In Table 3, we employ the Varimax rotation matrix as robustness check and we find no significant differences with un-rotated results (Abdy, 2003). Note Table 2 – Factor Analysis to Calculate Our WB Indicator Factor Eigenvalue Difference Proportion Cumulative Factor 1 21.815 16.037 0.484 0.484 Factor 2 5.778 2.291 0.128 0.612 Factor 3 3.487 1.005 0.077 0.690 Factor 4 2.482 0.479 0.055 0.745 Factor 5 2.003 0.582 0.045 0.789 Factor 6 1.421 0.190 0.032 0.821 Factor 7 1.231 0.096 0.027 0.848 Factor 8 1.135 0.084 0.025 0.873 Factor 9 1.051 0.214 0.023 0.897 Source: Authors elaborations on ISTAT data Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 147 Table 3 – Robustness Check on WB Measurement: Rotation Matrix Factor Variance Difference Proportion Cumulative Factor 1 19.445 14.597 0.432 0.432 Factor 2 4.848 0.245 0.108 0.539 Factor 3 4.603 1.599 0.102 0.641 Factor 4 3.004 0.919 0.067 0.708 Factor 5 2.086 0.117 0.046 0.754 Factor 6 1.969 0.366 0.044 0.798 Factor 7 1.603 0.159 0.036 0.834 Factor 8 1.443 0.041 0.032 0.866 Factor 9 1.402 0.031 0.897 Notes: orthogonal varimax; LR test: independent vs. saturated: chi2 (1176) = 8963.33 Prob>chi2 = 0.000. Source: Authors elaborations on ISTAT data. that un-rotated factor analysis for results of Table 2 are on 120 observations; 9 retained factors and 405 parameters. The eigenvalue shows the variance of the factor. In the initial factor solution, the first factor will account for the most variance, the second will account for the next highest amount of variance, and so on. Some of the eigenvalues are negative because the matrix is not of full rank, that is, although there are 49 variables the dimensionality of the factor space is much less. To choose which are the important factors to be considered we used the Kaiser method, retaining factors with eigenvalue greater than 1. In our case, this leads to consider the first nine factors with a cumulative variance explanation of well-being of almost 90% (0,897). The column Difference gives the differences between the current and following eigenvalue, the column Proportion gives the proportion of variance accounted for by the factor and finally the column Cumulative gives the cumulative proportion of variance accounted for by this factor plus all of the previous ones. Given these results, Table 4 reports the pattern matrix of the nine retrieved factors. The factor loadings for this orthogonal solution represent both how the variables are weighted for each factor but also the correlation between the variables and the factor. The higher the load the more relevant in defining the factor’s dimensionality. A negative value indicates an inverse impact on the factor. Most importantly, in the column ‘Uniqueness’ of Table 3 we report the proportion of the common variance of the variable not associated with the generated factors. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 148 Table 4 – Factor Loadings (Pattern Matrix) and Unique Variances Variable Factor Uniqueness 123456789 economic well-being 1 0.934 0.124 0.208 0.006 0.067 0.102 -0.121 0.012 0.011 0.040 economic well-being 2 -0.764 0.240 0.103 0.023 0.268 0.083 -0.120 -0.059 0.017 0.251 education 1 0.219 -0.672 0.243 -0.137 0.204 0.128 0.207 0.210 -0.198 0.239 education 2 0.689 0.439 -0.176 0.398 -0.059 0.046 0.083 -0.064 -0.170 0.098 education 3 0.549 0.517 -0.278 0.421 -0.137 0.157 0.060 0.011 -0.103 0.119 education 4 -0.520 -0.308 0.343 -0.408 0.276 0.191 -0.251 -0.009 0.261 0.107 education 5 0.593 0.211 -0.425 0.232 0.316 0.072 0.067 -0.029 -0.020 0.258 environment 2 -0.504 -0.247 -0.023 0.272 -0.392 0.405 -0.062 -0.003 0.203 0.249 environment 4 -0.030 -0.283 -0.288 0.414 -0.021 -0.559 0.017 -0.273 0.275 0.201 environment 5 0.726 -0.440 -0.313 0.157 -0.201 0.131 0.028 -0.039 0.027 0.096 environment 7 0.211 -0.598 -0.257 0.147 0.133 0.161 -0.431 0.139 -0.052 0.258 environment 8 0.765 0.067 -0.190 -0.349 0.292 -0.222 0.013 0.156 -0.140 0.074 health 1 0.671 0.112 -0.418 0.092 -0.050 -0.020 0.320 0.113 0.266 0.165 health 2 0.795 -0.018 0.099 0.184 0.178 0.034 -0.148 0.294 0.124 0.167 health 5 0.687 0.090 -0.124 0.037 -0.084 -0.006 -0.036 0.070 0.255 0.425 innovation 1 0.441 0.596 0.338 -0.067 0.245 -0.094 -0.062 -0.040 0.033 0.256 innovation 2 -0.252 0.778 -0.090 0.291 0.193 0.099 -0.172 -0.106 -0.115 0.138 innovation 3 0.585 0.452 0.293 0.135 -0.005 0.228 -0.093 -0.083 0.066 0.277 labour quality 2 -0.742 -0.158 -0.048 0.289 0.236 0.120 0.036 -0.163 0.136 0.222 labour quality 3 -0.932 -0.063 -0.001 0.089 0.086 -0.059 0.188 0.050 0.006 0.070 labour quality 4 -0.920 0.102 -0.058 0.107 0.226 0.095 0.081 -0.007 -0.160 0.036 labour quality 6 -0.589 0.466 -0.441 -0.076 0.180 0.207 -0.063 -0.133 0.181 0.107 labouroccupation 0.979 0.024 0.069 0.003 -0.081 0.080 -0.040 0.025 -0.001 0.022 landscape 2 -0.873 0.020 -0.209 0.191 0.032 0.028 0.023 0.002 -0.158 0.130 landscape 3 0.587 -0.025 -0.047 0.330 0.416 0.111 0.270 0.191 0.083 0.242 politics 1 -0.318 0.194 0.425 0.589 -0.176 -0.071 -0.025 0.113 0.086 0.277 politics 2 -0.457 -0.297 0.604 0.225 0.083 0.140 -0.041 -0.092 0.148 0.229 politics 3 0.031 -0.262 0.588 0.632 0.019 -0.042 -0.007 0.096 -0.171 0.145 politics 7 0.035 -0.043 0.707 -0.120 -0.347 -0.096 0.255 0.102 -0.028 0.278 security 1 0.446 0.573 -0.020 -0.223 -0.144 0.173 0.262 0.315 0.197 0.166 security 2 0.429 0.749 0.307 -0.055 0.025 -0.027 -0.124 -0.059 0.063 0.133 (Continues...) Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 149 Variable Factor Uniqueness 123456789 security 3 -0.348 0.308 0.365 0.210 0.396 -0.228 0.044 0.322 0.177 0.261 securitymurders -0.624 -0.080 0.175 -0.006 0.215 0.181 0.356 -0.154 -0.106 0.333 services quality 2 0.828 0.014 0.113 0.033 -0.148 0.270 -0.147 -0.003 0.144 0.165 services quality 3 -0.878 -0.022 0.007 0.128 0.073 -0.180 0.133 -0.021 0.125 0.141 services quality 4 -0.798 -0.063 -0.066 0.007 0.085 0.371 0.247 -0.230 -0.093 0.087 services quality 5 0.400 0.478 0.358 -0.177 0.152 -0.193 0.018 -0.381 -0.101 0.236 services quality 6 0.522 -0.664 -0.210 0.081 0.084 -0.050 -0.086 0.038 -0.027 0.217 social relationship 1 0.838 -0.115 0.102 -0.081 0.185 0.159 0.267 -0.138 0.184 0.084 social relationship 2 0.835 -0.170 0.053 -0.006 0.188 0.143 0.237 -0.163 0.164 0.107 social relationship 3 0.856 -0.322 0.032 0.051 0.187 -0.098 0.004 -0.140 0.006 0.097 social relationship 4 0.876 0.166 0.057 -0.223 0.009 0.164 0.006 -0.112 -0.021 0.112 social relationship 5 0.806 -0.270 -0.098 -0.011 0.393 -0.159 0.032 -0.059 0.063 0.079 social relationship 6 0.734 -0.192 0.113 0.172 0.329 0.104 -0.248 -0.083 -0.064 0.192 subjective well-being 0.621 -0.570 0.244 0.070 0.081 0.086 0.122 -0.125 -0.057 0.177 well-being & minimum conditions 1 -0.834 0.011 -0.082 -0.039 0.078 -0.035 -0.015 0.155 0.349 0.142 well-being & minimum conditions 2 -0.490 0.194 -0.232 -0.011 0.236 0.101 0.015 0.382 -0.250 0.393 well-being & minimum conditions 3 -0.826 0.043 0.075 -0.168 0.162 0.138 -0.045 0.077 0.106 0.217 well-being & minimum conditions 4 -0.920 -0.068 -0.152 -0.046 0.124 -0.009 -0.076 -0.071 0.126 0.083 Source: Authors elaborations on ISTAT data (...follows) Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 252 municipalities located in the centre-north are above the national average, and most of the municipalities located in the south (including the main islands) are below the average. As shown in Figure 3 (panel a), SEN are not correlated with municipal income because expenditure determinants associated with local income constitute marginal components. In particular, on average, input prices explain 5.2% of standard expenditure, variables that capture the structure of the local economy 4.6%, and finally, variables related to deprivation only 1% (the source of the impact of determinants of the standard expenditure is www.opencivitas.it, a governmental web repository of all data used for SEN evaluation). Instead, in panel b of Figure 3, we observe a strong positive correlation between FC and income, because declared income is the tax base of the local income tax and significantly correlates with the cadastral values representing the tax base of the property tax. Figure 2 – 2018 Reported per Capita Income, Municipal Average Source: Italian Ministry of Economy and Finance Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 253 Figure 3 – Correlation between Municipal Declared Income, Standard Expenditure Needs, and Fiscal Capacity a) Income Vs Standard Expenditure Needs -31 235 67 100 Std. expenditue needs 2020 % dev. from nat. mean -50 050 Income per capita % dev. from nat. mean b) Income Vs Fiscal Capacity -77 -33 12 56 100 Fiscal capacity 2020 % dev from nat. mean -50 050 Income per capita % dev. from nat. mean Source: our elaboration on data of the Italian Ministry of Economy and Finance Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 254 Figure 4 – Municipal Solidarity Funds Evolution of Grants Redistributions (Gross Endowment), per capita values a) 2020 MSF Gross Endowment b) 2030 MSF Gross Endowment Simulation Source: our elaboration on data of the Italian Ministry of Interior Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 255 As a result, the fiscal gap will be more pronounced in municipalities located in the southern regions. Therefore, as Figure 4 shows, the flow of equalization grants will distribute in favour of municipalities located in the southern regions, especially at the end of the transition period. In particular, in panel a) of Figure 4, we report the distribution of per capita MFS equalization grants in deviation from the national mean as they appear in 2020 at the 27,5% of the transition, in panel b) we show how the distribution should change in 2030 according to 2020 regulations. 3. Empirical Strategy Our study is based on the collection of financial and socio-economic data of the municipalities located in ordinary regions (OR municipalities) and the special statute regions of Sicily and Sardinia (SR municipalities) over nine years, from 2012, which marked the MSF, up to 2020. Therefore, the complete sample will be a balanced panel that includes 7,240 municipalities for nine years (we exclude from the dataset municipalities that underwent an amalgamation process between 2010 and 2020). Our empirical strategy aims at identifying the impact of dynamic fiscal gap equalization on the redistributive and risk-sharing effect of intergovernmental grants. Italian data allows us to use a difference-in-difference technique, where OR municipalities will constitute the treated group and the SR municipalities the control group. Instead, the introduction of dynamic fiscal gap equalization in 2015 will represent our treatment effect. Finally, to make the municipalities in the treated and the control groups more comparable and satisfy the pre-treatment common trend conditions, our final regression sample will include only OR municipalities located in the southern regions. Table 3 reports the descriptive statistics of the variables included in the dataset. General statistics are presented for four distinct groups: all municipalities, only municipalities in ordinary regions, municipalities in special statute regions (Sicily and Sardinia), municipalities in ordinary southern regions. Data sources are from the Ministry of Interior, the Ministry of Economy and Finance and ISTAT (the Italian Institute of National Statistics). Table 3 shows three group of variables: MFS grants, considering 2020 values and their projection at the end of the transition period; declared income (tax base of the personal income tax) that we use as a proxy of GDP at municipal level; control variables related to the structure of the resident population. Variables means are comparable between municipalities in special statute regions (Sicily and Sardinia) and municipalities in ordinary southern regions, respectively our control and treated groups. Figure 5 reports the time series of MSF grants and municipal declared income, expressed in real per capita terms, to support the difference-in-difference Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 256 Table 3 – Descriptive Statistics for the 2012-2020 Period Ordinary Regions Municipalities Total Municipalities Obs. Mean Std. Dev. Min Max Obs. Mean Std. Dev. Min Max MSF (euro per capita) 58,266 77.47 209.66 -4,439.00 2670.71 65,160 80.80 204.94 -4,439.00 2,670.71 MSF simulation 100% (euro per capita) 58,266 53.91 240.88 -5,756.76 2670.71 65,160 59.73 234.04 -5,756.76 2,670.71 Declared PIT income (euro per capita) 58,266 12,216 3,019 1,988 46,272 65,160 11,834 3,105 605 46,272 Population 0-2 (% of total pop.) 58,266 2.27 0.70 0.00 6.98 65,160 2.26 0.69 0.00 6.98 Population 3-14 (% of total pop.) 58,266 10.34 2.22 0.00 18.69 65,160 10.30 2.21 0.00 18.69 Population over 75 (% of total pop.) 58,266 12.72 4.19 2.38 46.75 65,160 12.72 4.15 2.38 46.75 Net population variation 58,266 -5.02 7.15 -80.51 37.04 65,160 -5.04 7.02 -80.51 37.04 Net migration 58,266 1.99 14.07 -202.02 243.90 65,160 1.74 13.80 -202.02 243.90 Resident population / 1000 58,266 7.823 45.509 0.029 2,873.494 65,160 7.920 44.175 0.029 2,873.494 Special Regions Municipalities Southern Odinary Regions Municipalities Obs. Mean Std. Dev. Min Max Obs. Mean Std. Dev. Min Max MSF (euro per capita) 6,894 108.97 156.78 -1001.47 1031.99 15,984 169.59 179.42 -1743.44 2670.71 MSF simulation 100% (euro per capita) 6,894 108.97 156.78 -1001.47 1031.99 15,984 153.92 169.23 -2390.27 2670.71 Declared PIT income (euro per capita) 6,894 8,533 1,444 605 17,002 15,984 8,735 1,572 4,446 17,409 Population 0-2 (% of total pop.) 6,894 2.13 0.62 0.00 5.22 15,984 2.20 0.67 0.00 5.87 Population 3-14 (% of total pop.) 6,894 9.93 2.03 0.62 16.84 15,984 10.10 2.34 0.79 18.69 Population over 75 (% of total pop.) 6,894 12.68 3.81 3.48 35.59 15,984 12.98 4.77 3.03 45.63 Net population variation 6,894 -5.21 5.77 -44.87 16.95 15,984 -5.29 6.84 -69.31 15.75 Net migration 6,894 -0.41 11.05 -127.60 110.22 15,984 -0.24 12.69 -120.67 133.81 Resident population / 1000 6,894 8.737 30.660 0.083 678.492 15,984 7.837 28.061 0.079 989.111 Source: Ministry of Economy and Finance, Ministry of Interior and ISTAT Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 257 Figure 5 – Time series of Intergovernmental Grants of Municipal Solidarity Fund and Average Declared Income, Comparison between the Control Group and the Treated Group. Only Municipalities Located in Southern Regions a) Municipal Solidarity Fund (MSF) b) Average Declared Income Source: Ministry of Economy and Finance and Ministry of Interior Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 258 empirical strategy. We compare the average values recorded in the treated and control groups to verify the common trend assumption in the pre-treated period. The treated group is restricted only to municipalities located in southern regions. The presence of a pre-treatment common trend is particularly evident in MSF grants. After the 2015 reform, in the treated group we observe a substantial increase in grants compared to the average amount allocated to SR municipalities. Moreover, both groups show a drop in 2014 and 2015 caused by the fiscal consolidation process. Instead, in 2016, we observe an increase due to the transformation of the property tax’s revenue on the owner-occupied main residence in grants from the central government. However, the fiscal consolidation process and the 2016 property tax reform do not operate as confounding factors since their effects are commonly spread in municipalities belonging to both groups. Therefore, our results did not change if we depurate MFS grants from these components and, for the sake of simplicity, we decided to consider only the gross flow of MFS grants. A more formal analysis of the pre-treatment common trend assumption is reported in the Figure A1 of the Appendix. 4. The Estimation of the Redistributive Effect of Formula Grants The redistributive effect is estimated through OLS applied to the following two-periods linear model specified in equation (3). ' 01 2 3 4 it it it it it it i it Y X XD D Z R t = γ +γ +γ +γ +γ + + + ε [3] where: •t = zero before 2015 and one from 2015; •Yit = income and equalization transfers (euro per capita), average before and after 2015; •Xit = income (euro per capita), average before and after 2015; •Dit = treatment dummy (one after 2015 for municipalities located in southern ordinary regions municipalities); •Zit = control variables, average before and after 2015; •Rit = ordinary region dummy; ••εit = idiosyncratic error component. In particular, the redistributive effect of historical transfers will correspond to  1 1 −γ and the redistributive effect of formula transfers will correspond to   12 1 −γ −γ . Table 4 reports the point estimates of the relationship between income and intergovernmental grants and its interaction with the treatment dummy. In Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 259 column (1) we report the point estimates related to the 2020 structure of grants, whereas in column (2) we simulate the level of grants that will be distributed at the end of the transitional period. We simulate the full implementation of the new equalization system computing the distribution of grants setting the parameter α of equation (2) equals one (in 2020, instead, α = 0.275). Subsequently, we use the point estimates reported in Table 4 to evaluate historical and formula grants’ redistributive effect decomposing this effect between the contribution of standard expenditure needs and fiscal capacity contribution. These final computations are reported in Table 5. Formula grants that dynamically equalize the fiscal gap generate a stronger redistributive effect than static equalization grants based on historical expenditure. However, this divergence is visible only when we simulate formula grants at the end of the transition period. In this case, we register an increase of the redistributive effect from 4.5% to 5.7% moving from historical expenditure equalization to fiscal gap equalization. Moreover, it is interesting to notice that fiscal capacity shows a positive redistributive Table 4 – Point Estimates of the Relationship between Municipal Declared Income and Intergovernmental Grants (Only Southern Regions) (1) (2) MFS MFS 100% simulation Income 0.95 0.96 [0.000]*** [0.000]*** Income X Treatment -0.00 -0.01 [0.722] [0.018]** Observations 5,084 5,084 Controls yes yes Estimator OLS OLS Note: OLS estimates with robust std. error p-value in brackets *=p<0.10; **=p<0.05; ***=p<0.01 Table 5 – Computation of the Redistributive Effect of Intergovernmental Grants MSF MSF 100% simulation Redistributive effect historical grants 4.6% 4.5% Redistributive effect formula grants 4.6% 5.7% of which standard expenditure needs -1.4% of which fiscal capacity 7.1% Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 260 effect because of the correlation with local income. Instead, standard expenditure needs show a negative redistributive effect since they aim to equalize provision costs. For the decomposition of the redistributive effect, we estimate the grants’ distribution considering SEN uniform in per-capita terms to estimate the impact produced exclusively by FC. Then we obtain the impact of SEN by difference. 5. The Estimation of the Risk-sharing Effect of Formula Grants As a second step, we estimate the income elasticity of formula grants that can then be used to evaluate the risk-sharing effect. In this case, we specify a linear panel data model, and the point estimates of the income elasticity are obtained using the Within-the-Group estimators. The model is reported in equation (4). ' 01 2 3 4 it i i it it it i t it Y X XD D Z κκ κκ ∆ = β +β ∆ +β ∆ +β +β +α +τ + ε [4] where: •k is replaced by: t, t–1, t–2, t–3, t–4, t–5; ••∆Yit = % deviation of equalization transfers (euro per capita) from the national mean; ••∆Xit = % deviation of income (euro per capita) from the national mean; •Dit = treatment dummy (one after 2015 for municipalities located in southern ordinary regions municipalities); •Zit = control variables lagged by one period; ••α i = municipal fixed effect; ••τ i = year fixed effect; ••ε i = idiosyncratic error component. In particular  1κ β corresponds to the estimated average income elasticity of equalization grants based on historical expenditure at different lags that, in turns, approximates the risk-sharing effect of historical grants. Instead,   12κκ β +β corresponds to the estimated average income elasticity of equalization grants based on the dynamic fiscal gap at different lags that, in turns, approximate the risk-sharing effect of formula grants. Table 6 reports the point estimates of the average income elasticity of grants considering different income lags. In column (1) we report the point estimates related to the 2020 structure of grants. In column (2) we simulate the level of grants at the end of the transitional period. Figure 6 summarises the final estimates of the income elasticity at different intertemporal lags and its decomposition between the two components: standard expenditure and fiscal capacity. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 261 Table 6 – With-in-the Group Point Estimates of the Income Elasticity of Intergovernmental Grants (1) (2) MSF MSF 100% simulation Income 0.53 0.65 [0.000]*** [0.002]*** Income X Treatment -0.08 -0.82 [0.048]** [0.000]*** Income lag 1 0.29 0.13 [0.001]*** [0.481] Income lag 1 X Treatment -0.06 -0.78 [0.096]*[0.000]*** Income lag 2 0.00 -0.49 [0.142] [0.001]*** Income lag 2 X Treatment 0.00 -0.74 [0.143] [0.000]*** Income lag 3 -0.28 -0.64 [0.001]*** [0.000]*** Income lag 3 X Treatment -0.06 -0.72 [0.087]*[0.000]*** Income lag 4 -0.20 -0.28 [0.020]** [0.098]* Income lag 4 X Treatment -0.07 -0.72 [0.067]*[0.000]*** Income lag 5 0.00 0.00 [0.160] [0.328] Income lag 5 X Treatment -0.06 -0.70 [0.080]*[0.000]*** Observations 22,878 22,878 Municipal fixed effect yes yes Time fixed effect yes yes Controls yes yes Estimator Within-the-Group Within-the-Group Note: Within-the-Group estimator with std. error clustered at municipal level, p-value in brackets *=p<0.10; **=p<0.05; ***=p<0.01. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 268 parallel trend assumption requires that we do not reject the null hypothesis that the treatment effect is equal to zero in all periods between 2012 and 2014. In other words, the distance of the outcome variables between the treatment and control group should remain constant in the pre-treatment periods. This evidence is verified for MFS grants and income. Copyright © 2021 by FrancoAngeli s.r.l., Milano, Italy. ISBN 9788835125860 11390.5 C. BERNINI, S. EMILI (edited by) REGIONS BETWEEN CHALLENGES AND UNEXPECTED OPPORTUNITIES Due to the Covid-19 pandemic, the XLI Annual Scientific Conference held on line in September 2-4, 2020. The Web Conference contributed to motivate the scientific debate on the regional challenges and opportunities in times of crisis. A large number of contributions have investigated the territorial impact of economic shocks and natural disaster and discussed possible trajectories for a sustainable regional development process. The book collects a selection of these contributions, covering different topics on the economic, social, and regional consequences of crises and recovery processes. The first part is dedicated specifically to the impact of the Covid-19 pandemic and to the ability of a territory to react. The challenges of the new pandemic came in addition to the economic and financial crises and the natural and environmental disasters that have occurred in recent decades. The second part then gathers contributions that discuss more broadly the resilience and regional responses to natural and economic shocks. Crises have largely affected the quality of life of citizens and may have compromised sustainable regional growth. To contribute to the discussion of these issues, the third part of the volume collects some studies that aim to analyse in depth the effects of crises in terms of individual and regional wellbeing, and their relationship with sustainability. The last part is dedicated to a discussion and empirical assessment of the role of regional and national policies in supporting recovery and resilience processes for regional development. REGIONS BETWEEN CHALLENGES AND UNEXPECTED OPPORTUNITIES edited by Cristina Bernini, Silvia Emili Cristina Bernini Full Professor of Economic Statistics and Researcher of the Center for Advanced Studies in Tourism at the University of Bologna. Silvia Emili Junior Assistant Professor of Economic Statistics and Researcher of the Center for Advanced Studies in Tourism at the University of Bologna. 61 Associazione italiana di scienze regionali Scienze Regionali FrancoAngeli La passione per le conoscenze 11390.5_1390.33 26/07/21 11:08 Pagina 1