European Central Bank publishes research proposing a new method to identify relationship-level credit shocks and finding bias in standard fixed-effects estimates

The European Central Bank has published a working paper by Olivier De Jonghe and Daniel Lewis, produced within the ChaMP research network, that presents a new econometric framework for identifying relationship-specific effects in bipartite networks, such as firm-bank credit demand and supply shocks, using covariance restrictions from jointly observed price and quantity data. In author research that does not represent ECB views, the paper argues that standard two-way fixed-effects approaches can break down when shocks vary across individual relationships rather than being homogeneous at firm or bank level. Applying the method to AnaCredit data covering thousands of firms and banks across nine euro area countries from 2019 to 2023, the authors reject the standard fixed-effects assumptions in 25 of 27 country-period combinations and find that within-firm and within-bank shock variation is economically large, with within-bank supply variation exceeding between-bank variation. The paper also reports substantial bias in conventional estimates of credit demand and supply and finds that the post-2022 ECB monetary tightening had materially negative effects on firms exposed through floating-rate or short-term borrowing, reducing credit supply and subsequently total assets, turnover and credit-to-assets ratios by more than standard fixed-effects methods would indicate, while also revealing heterogeneous transmission across firm-bank relationships.