【正文】 suite of models used to assess banking sector vulnerability. Introduction Macroeconomic stress tests of the financial system have been developed in recent years (see Se (2021) for a recent survey and discussion). These tests assess the vulnerability of the banking system,or more broadly the financial system, to extreme but plausible adverse macroeconomic tests are important, from a central bank’s perspective, since they are tractable and provide a useful benchmark to assess the risks to the financial system (see Bunn et al (2021)).Recently, as part of the IMF Financial Sector Assessment Programme (FSAP), stress tests were performed on the resilience of the UK banking system. The stress test scenarios were derived from a version of the Bank of England’s structural Medium Term Macroeconometric Model (MTMM). The scenarios were then applied to UK banks’ aggregate loan book (see IMF (2021) and Hoggarth and Whitley (2021)). The main findings of this analysis were that the UK banking system was robust to a range of plausible adverse macroeconomic this paper we adopt a different approach to perform macroeconomic stress tests on the UK banking system and investigate whether the conclusions arising depend on the choice of stress test and on the fragility variable used. We attempt to account for the dynamics between banks’ writeoff to loan ratio and key macroeconomic variables using a parsimonious vector autoregression (VAR) model. Unlike most existing stress testing work on links between the business cycle and the fragility of the banking system, a direct measure of banks’ fragility – the writeoff ratio on loans – is employed. The advantage of the VAR is that it estimates how writeoffs change in the quarters following adverse business cycle shocks implying that the stress test is conditional on the historical correlation among the variables in the multivariate model. The VAR approach also allows for potential feedback effects from the fragility of banks’ balance sheets to the macroeconomy (a potentially important linkage emphasised by, inter alia, Se (2021)). No conventional theoretical macroeconomic model describes the relationship from macroeconomic variables to bank writeoffs and vice versa (for example by affecting the supply of bank credit and thus investment). We attempt to resolve this problem in two steps. First, the paper discusses a macroeconomic model to help guide the choice of a parsimonious number of macroeconomic variables to include in the specification. Second we augment this vector of macroeconomic variables with the writeoff data and let them affect one another in an autoregressive manner without restricting the dynamics. Finally the paper considers some alternative financial and economic variables that could affect the writeoff ratio and macroeconomic variables. Such robustness checks are important since the conclusions could be misleading if some key variables are left out of the paper conducts one of the first multivariate analyses of how macroeconomic developments affect UK banks’ loan writeoffs both in aggregate and at the sectoral level. The importance, from a stresstesting viewpoint, of having data on banks’ fragility covering at least one full economic cycle is emphasised. Although writeoff data are available on a quarterly basis only from 1993, annual data on aggregate writeoffs for the major UK banks are available back to the late 1980s. In this paper a number of methods for interpolating these annual data onto a quarterly basis are considered using the available information on the characteristics of the quarterly and annual data in the sample from 19932021. Using three alternative interpolation schemes, the multivariate analysis is performed on a sample of quarterly data from 1988 to 2021, therefore covering a full UK economic cycle. Since aggregate data may mask different patterns at the sectoral level, separate sectoral VARs are also estimated for corporate writeoffs and household defaults. While th