Maria Kubara (p) / Epainos Award Candidate

Rolf Sternberg

The location decisions of the technological startups at the urban level are still not well investigated. The gap in the literature is huge – especially compared to the amount of research regarding location decisions and their effects on more traditional businesses. Taking a sample of startups from the up-and-coming market in Central-East Europe in Warsaw, Poland their spatial organisation across the years will be tracked to investigate whether there is a defined pattern of behaviour and how it evolves in time. It will be also shown how recursive neural networks (RNN) may help in predicting the locations of technological startups’ clusters. It will be shown that the model can be successful even for short spatial panel data. It will be presented how to include the spatial dimension in the model in a computationally-effective way and how this augmentation improves the results, allowing the network to “understand” the spatial relations between neighbouring observations.

Dennis Gaus (p) / Epainos Award Candidate

Rolf Sternberg

This paper aims to answer the question whether complex accessibility indicators are considered by firms when choosing their location, or whether these decisions are made based on simpler proxies as suggested by previous literature.
In the first step, a novel accessibility indicator designed to capture the market access of firms in an intuitive way, but requiring significant data processing and computational efforts, is developed. Based on a sample of 110,083 firms in Germany, it combines the transport distance and time between firms with the relation between their respective industries and their economic size in a gravity-type model. The indicator thus describes the access to potential suppliers and customers on the company-level using the exact location of potential business partners.
In the next step, the impact of this measure on the location decision of firms is evaluated and compared with the effect of several simpler measures which are easily observable for firms (number of firms of the same industry, number of firms of relevant supplying and demanding industries, and regional economic importance). Combining these variables with the accessibility measure and company-specific characteristics, a multinomial logit model is estimated. The location choices are defined by a combination of states and the thünen-classification of German counties.
On the one hand, the easily observable variables have a strong impact on the location decision of firms. A significant positive impact is found for the number of firms from the same industry as well as from supplying and demanding industries, pointing out the existence of agglomeration effects. This is supported by a negative effect of the relative economic importance of a firm within a region, as firms tend to locate themselves in economically strong areas. On the other hand, the market access indicator does not have a significant impact on the location choice. The additional firm characteristics add to the explanatory power of the model with several variables showing significant effects. Different model specifications are estimated to check the robustness of the results.
It can be concluded that firms make their location decisions based on measures that are easy to observe instead of deriving more complex indicators. Consequently, firms locate themselves suboptimally from the perspective of accessibility, as it is found that 96% of all firms in the sample are not located in the region where they would have the highest possible market access.

Zhiwu Wei (p), Davide Luca, Cem Özüzel

Rolf Sternberg

The advent of the COVID-19 pandemic has largely shifted the work modality across European regions into telework. Utilizing a combination of data from European Union Labor Force surveys, Eurofound Living, Working, and COVID-19 surveys, and the Eurostat database, this paper finds that telework potential and actual telework uptake before and during the pandemic vary both across and within European regions. While no clear relationship was observed before the pandemic, regional telework potential can well predict actual regional telework uptake during the pandemic. Our systematic investigations into driving factors of regional telework uptake during the pandemic suggest that regions with higher shares of teleworkable workers had experienced higher levels of telework uptake during the pandemic. These workers include: those living in urban areas, those in the sectors of financial services, public administration, and education, those with a college degree, middle-aged workers, workers having children aged 0-11, and female workers. In addition, regions more densely populated and with higher shares of households with access to broadband also saw more telework uptake. (see extended abstract)