Identify trading vs banking book classification in securities holdings data

Determine, for each security in the Banca d’Italia security-by-security bank holdings dataset used in this study, whether the instrument belongs to the trading book or the banking book, given that the dataset lacks prudential portfolio classification information required to distinguish accounting portfolios.

Background

The paper analyzes interest rate risk using highly granular security-by-security holdings data collected by Banca d’Italia. These are statistical datasets where all securities are evaluated at fair value regardless of accounting treatment. However, the dataset does not include prudential portfolio identifiers, preventing the authors from distinguishing whether securities are held in the trading book or the banking book.

This classification is important for interpreting risk exposures in a prudential context and for aligning analyses with regulatory frameworks. While the paper adopts a risk management perspective that does not depend on accounting portfolios, the lack of trading/banking book information remains an explicit unknown in the data and limits certain interpretations and extensions.

References

Furthermore, for banks, there are no information on prudential portfolios and we do not know if a security belongs to the trading book or to the banking book.

Shifting the yield curve for fixed-income and derivatives portfolios  (2412.15986 - Bianchi et al., 2024) in Section 2.2 (Securities holdings)