README This README file documents all steps and files necessary to replicate the results of the paper: “The Benefits of Remoteness - Digital Mobility Data, Regional Road Infrastructure, and COVID-19 Infections” Previous discussion paper version: “The Benefits of Remoteness - Digital Mobility Data, Regional Road Infrastructure, and COVID-19 Infections”, CEGE Discussion Paper No. By: Krenz, Astrid; Strulik, Holger Abstract: We investigate the regional distribution of the COVID-19 outbreak in Germany. We use a novel digital mobility dataset, that traces the undertaken trips on Easter Sunday 2020 and instrument them with regional accessibility as measured by the regional road infrastructure of Germany’s 401 NUTS III regions. We identify a robust negative association between the number of infected cases per capita and average travel time on roads to the next major urban center. What has been a hinderance for economic performance in good economic times, appears to be a benevolent factor in the COVID-19 pandemic: bad road infrastructure. Using road infrastructure as an instrument for mobility reductions we assess the causal effect of mobility reductions on infections. The study shows that keeping mobility of people low is a main factor to reduce infections. Aggregating over all regions, our results suggest that there would have been about 55,600 infections less on May 5th, 2020, if mobility at the onset of the disease were 10 percent lower. In case of questions, please contact: Astrid Krenz, a.m.krenz@sussex.ac.uk. This documentation contains information about: 1. Data sources and data access 2. Programme codes 3. Simulation exercise (Table 4) 1. Data sources and data access All the data that we used for the analyses are publicly available. They can be directly downloaded from the following sources. - INKAR Data, from the Bundesinstitut fuer Bau-, Stadt- und Raumforschung im Bundesamt fuer Bauwesen und Raumordnung, Link to the Data: https://www.inkar.de/ - Mobility Data, from the Mobility Monitor, Robert Koch Institute and Humboldt University Berlin, Link to the Data: https://www.covid-19-mobility.org/mobility-monitor The data, at the level of NUTS III regions (Kreise and kreisfreie Staedte), has been directly extracted from the Monitor. For a handful of cases a deeper disaggregation of regions than the NUTS III level was shown by the Mobility Monitor. In that case we took the regions that constitute an ordinary NUTS III region and formed the average of the mobility number. - Infection and death cases of COVID-19, from the Robert Koch Institute, COVID-19-Dashboard, Corona Landkreise, Link to the Data: https://npgeo-corona-npgeo-de.hub.arcgis.com/datasets/917fc37a709542548cc3be077a786c17_0/data?page=1 - Laboratory tests, from the Robert Koch Institute, Wochenbericht vom 6.5.2020, Laborbasierte Surveillance von SARS-CoV-2, Link to the report and data: https://ars.rki.de/Docs/SARS_CoV2/Wochenberichte/20200506_Wochenbericht.pdf - Regional data from GENESIS, from the German Statistical Office, Link to the Data: https://www.regionalstatistik.de/genesis/online For all these data sources, investigators should follow the rules of citing the Institutions and the websites from which they obtained the data, respectively. 2. Programme Codes The programme codes (Stata do-files) are provided for replication purposes. 3. Simulation Exercise To get results for the effects of a 1 Percent Pre-Disease Mobility Reduction on Infections, the following simulation exercise was executed. For selected regions (NUTS III), we 1. take the number of infected cases 2. take the number of population (RKI numbers) 3. compute the cases per 1000 population, that is dividing the number of cases by the number of population and by 1000 4. compute a 3.39 percent decrease in the cases per 1000 population, that is multiplying the number from 3. by 0.039 5. compute the number of new cases per 1000 population, that is substracting the number in 4. from the number in 3. 6. compute the Counterfactual number of total cases, that is multiplying the number in 5. with the number of population in 2. 7. compute the Counterfactual reducation of cases, that is substracting the number in 6. from the number in 1.