Introducing cluster heatmaps to explore city/firm interactions in world cities

Abstract

This article demonstrates the added value of cluster heatmaps in urban and regional research. To this end, we apply this tool to two-mode network data on connections between world cities (mode 1) and global services firms (mode 2): we interpret structures and patterns in the location strategies of 175 globalized producer services firms across 707 world cities in 2016. To date, the analytical potential of such data has not yet been fully exploited because more standard visual-analytical tools fall short of capturing the full complexity of large datasets. Furthermore, the two-mode network data are typically projected into one-mode network data only focusing on cities. Cluster heatmaps are powerful tools because they condense large data matrices into information-thick and assumption-free visual summaries. We show how the algorithms underlying their construction are useful for exploratory data analysis and subsequent hypothesis-generation of the two-mode network data. The cluster heatmaps reveal which firms with a similar global presence co-produce similar types of world cities, and which sets of world cities mediate the location behavior of producer services firms by housing similar ecologies of firms. The heatmaps reveal subtle nuances in the position and role of cities and firms in different globalization arenas, and provide a comprehensive overview of lower-tier world cities in particular. We discuss notable nexuses, some of the technique's limitations, and avenues for further research.