Welcome to GEFRI


Mission and Approach DC Global Center Institute Objectives Research Focus Working Papers Executive EducationProgram Execution Global Finance Conference, April 2006Nanotech Conference,
Fall 2006

Brazilian Banks Rating and Systemic Risk

By: Theodore M. Barnhill, Marcos Rietti Souto

Due to the potentially large and widespread economic impacts associated with bank failures, the assessment and management of systemic risk is a topic of great importance. Forward-looking risk assessment methodologies can be of central importance in alleviating this problem and, more importantly, they allow for the identification and evaluation of proactive steps that may be undertaken to manage such risks. In this study we seek to develop a forward-looking methodology for assessing the systemic banking risk (the likelihood that several banks might fail, sequentially or at the same time ), that integrates market and credit risk in the same spirit of Barnhill and Maxwell (2002) – the Portfolio Simulation Approach (PSA) –, with an application to the Brazilian financial system.

For this purpose, we first simulate and rate a set of fifty Brazilian banks, via PSA, that are later aggregate into different risk categories for the systemic risk analysis. The rating methodology utilized by agencies such as Moody’s are generally designed to measure not only the likelihood of a company defaulting on its debt obligation, but also the potential monetary loss associated with this default. We will use this notion of credit rating to rate a large set of Brazilian banks and compare our results with the rating provided by Moody’s and Standard and Poor’s. It is our perception that a lack of a more conceptual framework for assessing properly the risk of default might be one determinant factor behind discrepancies found in ratings provided by different rating companies (split ratings). We thus simulate Brazilian banks through the PSA methodology and use the simulation results to rate the banks and contrast with the ratings provided by three major rating agencies: Fitch, Moody’s and Standard and Poor’s.

We also group these Brazilian banks in 4-5 different rating categories and perform a simulation exercise with the purpose of assessing the systemic risk of Brazilian banking sector. Our simulations are designed to accommodate three of the four channels, identified in the literature, through which systemic events may propagate. First, by explicitly modeling banks’ portfolios and balance sheets in details, it is possible to capture the correlation between assets returns. Second, the Monte Carlo exercise that is performed via Choleski factorization allows the possibility of a combination of draws affecting assets returns similarly to a cyclical macroeconomic downturn. As these returns are fed back to recalculate volatilities and correlations, via the EWMA process, then the cyclical downturn might deepen further. Interbank exposure is also modeled, as we assume that banks have some fraction of interbank contracts that may become delinquent if one of the counterparts fails, which might trigger a systemic failure of the other banks.