Article Detail

How to Back-test Operational Risk: An Empirical Basic Analysis

Journal 35: Zicklin-Capco Institute Paper Series in Applied Finance

José Manuel Feria-Dominguez, Enrique J. Jimenez-Rodriguez, Paz Rivera-Pérez

Measuring, controlling and managing operational risk have played an important role for bank industry endeavors since the publication of Basel II (2006) as the regulatory framework for the effective management and supervision of financial risks. More recently, due to the current financial crisis, the Basel Committee on Banking Supervision (henceforth, the Committee) has revised such international standards giving rise to Basel III (2010) in order to strengthen the regulation, supervision and risk management of the banking sector.

Inherited from market risk, the application of the Value at Risk (VaR) concept to the Loss Distribution Approach (LDA) has been encouraged by the Committee for measuring operational risk. Moreover, complementary analysis has also been recommended to calibrate the soundness of such estimates by assessing the exceedances beyond Operational Value at Risk (OpVaR) forecasts. In this paper, we conduct an empirical back-testing on the LDA model by using an Internal Operational Losses Database (IOLD) provided by a medium-sized Spanish savings bank. For this purpose, we calculate daily OpVaRs to which we applied the basic analysis, based on the binary indicator and extremal index. Our empirical results reveal that the implementation of the LDA model has proven to be unreliable for the savings banks analyzed under the Basel regulatory framework.

For ages financial institutions have been exposed to financial risks such as credit and market risk, but also to operational risk whose measurement and control have recently gained significant attention for regulators, supervisors, managers and investors.
Traditionally there has been a lack of consensus on a standard definition for operational risk (henceforth, OR). Moreover, it has been considered the firm’s residual risk after other sources of risk, such as market risk and credit risk, have been taken into account [Allen and Bali (2007b)].

With the publication of Basel II [BCBS (2006)] the Committee introduced an explicit definition of such financial risk as: ‘‘the risk of loss resulting from inadequate or failed internal processes, people and systems or from external events’’. In addition, the main novelty of this international regulatory framework was the introduction of capital charges for covering operational risk, modifying, in consequence, the traditional solvency coefficient.

The current financial crisis has forced the Committee to revise those international standards for banking regulation giving rise to Basel III [Basel Committee (2010)]. More specifically, it implies a set of reform measures in order to strengthen the regulation, supervision and risk management of the banking sector with the aim of: improving the banking sector’s ability to absorb shocks arising from financial and economic stress, whatever the source; improving risk management and governance; and strengthening banks’ transparency and disclosures.

In practice, the heterogeneity of factors surrounding operational risk increases the complexity when measuring, controlling and managing such a risk within a financial institution. Being aware of that, the Committee proposes three main methodologies (Basic Indicator, Standardized Approach and Advanced Measurement Approaches) to calculate capital requirements for covering operational risk [Medova and Berg-Yuen (2009)]. In this paper, we will focus on the most sophisticated one, that is, the Loss Distribution Approach (LDA), an actuarial model to which the Value at Risk (VaR) concept is applied. Inherited from market risk, this is a statistical estimate that indicates the maximum operational loss, expressed in economic terms, which a bank can incur within a certain period of time (one year) for a given confidence level (99.9%). Moreover, apart from estimating the corresponding Operational Value at Risk (OpVaR), “banks should also regularly review actual performance after the fact relative to risk estimates (i.e. back-testing) to assist in gauging the accuracy and effectiveness of the risk management process and making necessary adjustments” [Basel Committee (2010)]. In other words, it is essential to carry out complementary analysis on the internal risk model to demonstrate its soundness.