Identification of Relationship Lending in Bank–Borrower Networks

ABSTRACT Relationship lending refers to lending based on close bank–borrower relationships, expected to increase credit availability and support real economic activity. This paper proposes the ST‐B filter, a novel framework identifying significant ties in temporal bipartite networks. Applying our method to Japanese financial data (1977–2021), relationship lending comprises about 10%–35% of total lending and is associated with 33% larger loan volume than transaction lending. Our method uniquely enables the analysis of relationship lending using only transaction data, without requiring detailed information on bank–borrower relationships.