Skip to main content

Application: Correlation Structure in US Industries and Regions

Application: What Correlation Structure Tells About Industries and Regions of the United States

In this section, we delve into the actual correlation structure observed in our data, focusing on industries and regions within the United States. This analysis reveals significant insights into geographical patterns and industry spaces.

Key Points:

  • Similarity Techniques: We've established that many similarity techniques are partially equivalent and can be interpreted as coexistence in continuous space. The area size influences the distance at which coexistence is detected.

  • Correlation Structure: The correlation structure is encoded in a square matrix, which can be translated into an adjacency matrix. This matrix defines a network where each industry is a node, forming an industry space. Each node has a spatial distribution across counties.

  • Community Clusters: Nodes within a community, or cluster of tightly connected nodes, share a common distribution across space. This leads to distinct geographical patterns due to the geographical units present in the cross-section.

In the following subsections, we will introduce the methods applied to develop an industry space and geographical patterns (see Section: Methods for Industry Space and Geographical Patterns). Additionally, we will present and discuss the results in Section: Network Structures and Section: Cluster Analysis.