Assessment of “stress tests” conducted on the French banking system

Abstract

During the first quarter of 2004, the General Secretariat of the Commission bancaire (SGCB) and the Directorate General Economics and International Relations (DGEI) of the Banque de France conducted an assessment of the stability of the French banking system and its capacity to withstand a set of macroeconomic and financial shocks, as part of a broader evaluation of the French financial system carried out under the auspices of the IMF’s Financial Sector Assessment Program (FSAP). The assessment employed a macro-prudential approach which seeks to quantify the effects of shocks to the banking system using “stress tests”. The tests measured the impact of severe shocks, deemed plausible but infrequent: e.g., a recession, a large movement in interest rates, an oil price shock, a sharp drop in stock prices. This report discusses in detail the principal characteristics of the “stress tests” and the innovations introduced during the French FSAP, including in particular the design of coherent scenarios, which were developed using the DGEI’s macroeconomic model and the SGCB’s financial models for measuring risk. The results of the assessment indicate that, given the high average solvency ratio, the French banking system is currently in a position to withstand a major macroeconomic shock, such as a prolonged recession lasting two years. This type of shock would, however, erode the quality of bank assets and reduce bank profits by 38.5% in the second year, compared with the baseline, resulting in a decline in the international solvency ratio of one percentage point (using the Basel I methodology) or two percentage points (using the new methodology proposed in the Basel II Accord). Other scenarios, such as a 32% depreciation of the dollar against the euro for two years or an increase of nearly 50% in the price of oil also for two years, would have more limited effects on net income and solvency ratios.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 27 citations exist upstream.