Measuring industry coagglomeration and identifying the driving forces

Agglomeration and clusters: Tools and insights from coagglomeration patterns

“…O'Sullivan and Strange () provide recent evidence using simulations of a simple coagglomeration model that—despite the large number of possible equilibria—the equilibria of the model do, on average, reflect the existence of beneficial coagglomeration of industries. Hence, measures of coagglomeration can play an important role in defining clusters, as they “embody—on average—mechanisms.” This is potentially useful in data‐poor contexts where “economic proximity” data are either not available or of low quality (e.g., developing countries; see Howard et al for Vietnam).…”

Comment to XCL co‐agglomeration index: Distance‐weighted improved DCL index

“…Following Howard et al (), let us assume six firms, p firms from industry A and q firms from industry B , located in regions X and Y . CL measure would be: where C ij is the co‐location dummy which takes value C ij = 1 when two firms from different industries ( A or B ) locate in the same region (co‐locate), and C ij = 0 in other cases, while p and q are the numbers of firms in industries A and B respectively.…”

“…Howard et al () in the Appendix in their paper propose the iterative procedure of counting the cases fulfilling the conditions. This procedure can be simplified with matrix operations.…”

“…This information on location and industry can be expressed in two matrices: ( i ) matrix of co‐localization A and ( ii ) matrix of industry B . Following Howard et al () co‐localization = 1 ( a ij = 1) when firms are located in the same region and 0 ( a ij = 0) in other cases; industry = 1 ( b ij = 1) when firms from the same industry and 0 ( b ij = 0) in other cases.…”

“…Thus, for the CL index one looks for cells with different industries ( b ij = 0) and the same region ( a ij = 1). Instead of the iterative search proposed by Howard et al () one can apply the term: where ( A⋅B ) ij is the Hadamard product (also called the Schur product) giving as a result the term a ij b ij , that requires we only multiply the corresponding cells. Noting that A is a matrix of co‐localization and B is a matrix of industry, term (2) can be read as colocalization – colocalization × industry .…”