Comparative Analysis of Operational Risk Approaches within Basel Regulatory Frame-Work : Case Study of Spanish Saving Bank

What the paper says

IntroductionIn June 2006, the Basel Committee on Banking Supervision, henceforth the Committee, published the last version of the New Capital Accord (Basel II)1. One of the principal novelties of Basel II is the inclusion of capital charge against operational risk, added to those traditional requirements to cover both credit and market risk. Three main methodologies are proposed by the Committee for calculating the operational risk regulatory capital:Basic Indicator Approach (BIA)Standardised Approach (SA)Advanced Measurement. Approach (AMA)More specifically, within the AMA models, the Loss Distribution Approach (LDA) is proved to be the most sensitive methodology to measure such a risk. Starting from this premise, the main hypothesis of our paper is to calibrate, by conducting an empirical application, whether the implementation of an advanced approach in financial entities provides a lower consumption of the regulatory capital, in comparison with the non-advanced ones. As previous studies on the LDA model reveal the high impact of the probability distribution parametric profile on the resulting Capital at Risk (CaR), we take this opportunity to assess this effect for each operational event type.We begin the study by establishing the theoretical framework in which those methodologies of measurement have been conceived. Having defined this framework, the next step is to test of our main hypothesis. For this purpose, we have taken the historical information on operational losses provided by a Spanish credit entity, specialised in the retail banking sector. We thus devote the third part of this paper to the detailed analysis of the inputs employed, i.e. the data of operational losses. Once the data are ready to be handled, and following the methodological sequence of the LDA approach, we have modelled separately the distributions of both frequency and severity by using different probabilistic models for fitting them. At this point, we apply an actuarial technique, such as the convolution, to mix up both frequency and severity, resulting in a third distribution of aggregated losses to which inferred a certain percentile (99.9'h). In this study, we address the operational risk within the retail banking, by computing the capital required for each event types. Lastly, for the whole entity, we conduct a comparative analysis of the capital consumption based on the methodology applied.The results of the study demonstrate a clear divergence between the capital estimated by applying the less advanced approaches and what is provided by the LDA model, giving raise to a significant capital saving for the latter.Methodology UsedThe measurement of operational risk has become the most complex and, in turn, the most important aspect when addressing such a financial risk. The Committee (2001: 3)2 defines the BIA and SA approaches as topdown methodologies. Both approaches determine the capital requirements for the global entity (BIA) or for each business line (SA). After this preliminary calculation, in a top down process, the assignment of capital is broken down by type of risk and by business unit or process, in particular. In contrast, in the AMA approaches the capital required is calculated from internal loss data split up by event type and by business line. After this specific calculation and following a bottom-up process, the global capital requirement is obtained by aggregation for the bank as a whole.According to the Committee (Basel, 2006: 148-155)', with the exception of the BIA, banks must comply with certain admission criteria for applying the SA and the AMA methodologies. In this sense, bank industry is encouraged to adopt progressively more advanced approaches, starting from the basic ones. Nevertheless, the jump evolution process through advanced techniques is conditioned to the availability of internal operational losses database (IOLD). In this respect, the LDA, based on the concept of Value at Risk (VaR) is considered the most suitable approach when calculating regulatory capital. …