Zsolt Csáfordi (p), Frank van Oort

Does an embedding in a skilled region, city or sector indeed lead to competitive performance advantages, in what time frame do advantages occur (immediately or on the long run), and which types of firms, cities, regions and sectors select themselves into this firm-context agglomeration process? To what extent are the effects of agglomeration economies and regional endowments maintained when also controlling for the structure of actual labor flows, including work experience acquired in a more productive company and in a technologically related industry?
Present research aims to disentangle the effects of regional endowments and agglomeration economies from the firm-level effects of new hires with an experience of a more efficient production and experience in a technologically related industry. These effects will be analyzed together in a multilevel framework. We expect that the cross-level effects between firm size and agglomeration economies found in Van Oort et al. (2012) may significantly be lowered due to the inclusion of more firm-level variables, especially the characteristics of actual labor inflows. We cannot exclude however, that some of the newly included firm-level variables might show cross-level effects with urbanization and localization economies, and that other regional characteristics might also interact with labor flows in facilitating productivity spillovers between firms.

Yusuke Teraji (p)

Through the investment in railroads, cities in Hokkaido, the northern island of Japan, have experienced the significant decrease in the access time to Sapporo, the capital of Hokkaido. Although the most of cities have experienced the reduction in the access time, the growth in the population is observed in some cities while the other face the significant decline. This indicates that the benefit of the investment in the transportation network shared unequally among the cities. Furthermore, through the observation of data in 50 years, the increase in the access time to Sapporo is observed after the decline in the population.
Based on the data observation, we construct a three-city model. In the model, firms and the provider of the transportation service exist. Firms choose their locations of office: namely, a single office at one of the two cities, two offices at two of the three, or offices at each city. Firms’ choice depends on the access cost between the two cities, and the access cost is determined according to the investment by the transportation service provider. Based on this model, we address the following questions: i) whether the benefits of the investment in the transportation is unequally shared among cities; and ii) whether the investment in the transportation generates the population growth.
Our model has shown that, as in the observation in Hokkaido, the benefits of the investment is unequally shared. Furthermore, whether the population of the city grows depends on the access condition to the other cities as well as the center. At a city, the population grows once the access condition to the center is improved while the condition to the rest remains. In contrast, if the access condition in the entire network is improved, cities other than the center will decline.

Emmanouil Tranos (p), Andre Carrascal Incera , George Willis

This paper brings new data and methods in order to expose the structure and the evolution of regional interdependencies. Although our focus is on the UK regions, the proposed research framework lends itself to applications to other European countries. Inter-regional trade relationships have traditionally been very difficult to capture, as national statistics do not monitor intra-country trade links at the firm level. Recently, the EUREGIO project (Thissen et al., 2018) has published spatially disaggregated Input-Output (I-O) information for 37 NUTS2 UK regions at a 14 industries level, which includes imports and exports by region of origin and destination. This database was developed using the interregional trade data from the PBL Netherlands Environmental Assessment Agency, freight transport data from Eurostat (for goods), and business flight ticket information (for services). Nevertheless, building such a regional I-O dataset has been a costly and non-trivial process. In this paper, we are proposing a research framework to predict such flows by utilising freely available web data. Specifically, we are employing the JISC UK Web Domain Dataset in order to extract hyperlinks between geolocated commercial websites in the UK. This dataset is a subset of the Internet Archive - the most complete archive of online web content, which includes all the archived webpages under the .uk country code Top Level Domain (ccTLD). We are able to geolocate these webpages by searching the archived web text for the inclusion of a UK postcode. In addition, this dataset also contains the hyperlinks included in the archived webpages. So, we are able to aggregate these data and create an inter-regional network based on the hyperlinks between geolocated commercial webpages. Formally, we approach the prediction of the I/O flows as a link prediction problem. Hence, we employ some well-established machine learning models, such as Random Forests, to predict the I-O flows using, among other features, the network of digital interdependencies between the UK regions.