Spatio-Temporal Visualization of Evolving Company Networks

Switzerland has experienced significant economic changes over the past 150 years, transitioning from a primarily local economy, where businesses operated mostly in the primary sector, to one of the world’s most competitive economies, largely driven by the services sector. This evolution has attracted the attention of economists who aim to understand the factors behind this shift. Part of the data needed for economists to analyze the Swiss economy is available due to the Swiss Ordinance on the Registry of Commerce, which requires businesses listed in the Central Business Name Index to provide detailed information about their location, ownership, business purpose, and other relevant aspects. The Swiss Confederation ensures this information is made publicly accessible by publishing daily updates in the Official Gazette of Commerce. Despite the wealth of this valuable data, its potential remains largely unexploited due to its unstructured nature, compounded by the multidimensionality of the information. Therefore, to date, no large-scale analysis has ever been undertaken.

The raw data extracted from the Official Gazette of Commerce is composed by a set of text snippets, each containing information about a specific event related to a company. By combining these events together, it is possible to reconstruct the life-cycle of a company. Some of these events mention other companies, e.g., when a company acquires another company. By combining these events together, it is possible to reconstruct the networks of companies and their evolution over time. Performing this task is challenging because the data is unstructured and the relationships as well as many other attributes of the single companies change over time making traditional data analysis tools unsuitable to explore and analyse this data.

In this thesis we aim at enabling large-scale spatio-temporal analysis of the data of these evolving company networks. The approach we propose seeks to empower researchers with the means to delve into this wealth of information and gain insights beyond basic details like location, capital, and ownership. We aim to uncover how networks of companies have formed, evolved, and interacted. To address these goals, we harnessed visualization techniques capable of breaking down the complexity of the data at hand, offering researchers the means to perform exploratory and explanatory analyses. Our approach paves the way for addressing previously unanswered questions such as the impact of the series of tax reforms the canton of Luzern embarked on in 2010 with the aim of attracting businesses from other cantons—especially Zug, which at the time had a significantly lower tax rate than any other canton in Switzerland.

We implemented a tool to validate the approach we propose in this thesis and demonstrate its effectiveness. The initial phases of this thesis involved trial-and-error iterations that led us to a final design that leverages a set of numerous custom visualisation instead of a single one-size-fits-all visualisation. This is because the key to understanding the data lies in the ability to interact with the data in order to explore it from different perspectives and test different hypotheses. The tool we developed allows users to create custom populations of companies, apply filters to the dataset, and compare the characteristics of companies across different dimensions, such as canton, tax rate, or type. The tool also supports spatio-temporal analysis, enabling users to track the evolution of companies over time.