| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 136.54 KB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
Este documento descreve o trabalho desenvolvido durante o estÔgio profissional, para obtenção do
mestrado em MatemÔtica Aplicada à Economia e Gestão, na Faculdade de Ciências da Universidade
de Lisboa. Este trabalho teve como intuito o desenvolvimento da monitorização do modelo de Credit
Conversion Factor (CCF).
De acordo com os Acordos Basileia, na abordagem Internal Ratings-Based avanƧada, o Banco pode
desenvolver modelos internos para a estimação de parâmetros de risco utilizados no cÔlculo de capital,
como é o caso do CCF. No entanto, a instituição deve validar as suas estimativas internas através de
sistemas robustos de monitorização.
Neste documento estÔ presente um resumo da definição do modelo de CCF, sendo abordada a cons trução do Reference Data Set (RDS), a definição da segmentação através de Ôrvores de decisão e o cÔlculo
das estimativas finais de CCF.
O tema principal deste trabalho é a monitorização do modelo de CCF que tem como propósito com parar as estimativas, aprovadas pelo Banco Central Europeu, do modelo em produção, com as estimativas
calculadas com o RDS atualizado, Ć data de referĆŖncia de setembro de 2022.
Desta forma, diversas anÔlises foram feitas para ajudar na validação das estimativas de CCF em
produção, sendo estas a caracterização do RDS e da carteira do Banco, a avaliação das alterações externas
e internas ao Banco que possam impactar o modelo, testes de qualidade de dados, anƔlise da qualidade
da segmentação e do desempenho das estimativas e avaliação das margens de conservadorismo.
Tendo por base as diversas anĆ”lises efetuadas no exercĆcio de monitorização, foi possĆvel concluir que
a maioria dos testes apresentou resultados positivos. No caso dos testes que não passaram os thresholds
definidos, os mesmos foram analisados, tendo sido encontrada uma justificação plausĆvel para o resultado
obtido.
Em suma, as estimativas de CCF em produção continuam a ser consideradas adequada.
This document describes the work developed during the professional internship, to obtain a masterās degree in MatemĆ”tica Aplicada Ć Economia e GestĆ£o, at the Faculdade de CiĆŖncias da Universidade de Lisboa. The purpose of this work was to develop the monitoring of the Credit Conversion Factor (CCF) model. According to the Basel Accords, in the advanced Internal Ratings-Based approach, the Bank can develop internal models for estimating risk parameters used in the capital calculation, as is the case with the CCF. However, the institution must validate its internal estimates through robust monitoring systems. This document summarizes the definition of the CCF model, covers the construction of the Reference Data Set (RDS), the definition of segmentation using decision trees and the calculation of the final CCF estimates. The main theme of this work is the monitoring of the CCF model, the purpose of which is to compare the estimates approved by the European Central Bank for the model in production with the estimates calculated with the updated RDS, on the reference date of September 2022. In this way, various analyses were carried out to help validate the CCF estimates in production, such as characterising the RDS and the Bankās portfolio, assessing the external and internal changes to the Bank that could impact the model, data quality tests, analysis of the quality of the segmentation and the performance of the estimates and margin of conservatism assessment. Based on the various analyses carried out in the monitoring exercise, it was possible to conclude that most of the tests showed positive results. In the case of the tests that did not pass the defined thresholds, they were analysed and a plausible justification was found for the result obtained. In short, the CCF estimates from the 2018 application are still considered adequate.
This document describes the work developed during the professional internship, to obtain a masterās degree in MatemĆ”tica Aplicada Ć Economia e GestĆ£o, at the Faculdade de CiĆŖncias da Universidade de Lisboa. The purpose of this work was to develop the monitoring of the Credit Conversion Factor (CCF) model. According to the Basel Accords, in the advanced Internal Ratings-Based approach, the Bank can develop internal models for estimating risk parameters used in the capital calculation, as is the case with the CCF. However, the institution must validate its internal estimates through robust monitoring systems. This document summarizes the definition of the CCF model, covers the construction of the Reference Data Set (RDS), the definition of segmentation using decision trees and the calculation of the final CCF estimates. The main theme of this work is the monitoring of the CCF model, the purpose of which is to compare the estimates approved by the European Central Bank for the model in production with the estimates calculated with the updated RDS, on the reference date of September 2022. In this way, various analyses were carried out to help validate the CCF estimates in production, such as characterising the RDS and the Bankās portfolio, assessing the external and internal changes to the Bank that could impact the model, data quality tests, analysis of the quality of the segmentation and the performance of the estimates and margin of conservatism assessment. Based on the various analyses carried out in the monitoring exercise, it was possible to conclude that most of the tests showed positive results. In the case of the tests that did not pass the defined thresholds, they were analysed and a plausible justification was found for the result obtained. In short, the CCF estimates from the 2018 application are still considered adequate.
Descrição
Trabalho de Projeto de Mestrado, MatemÔtica Aplicada à Economia e Gestão, 2024, Universidade de Lisboa, Faculdade de Ciências
Palavras-chave
CCF CLE Monitorização RDS Teses de mestrado - 2024
