Resumen: Under the Basel II Accord, banks and other Authorized Deposit-taking Institutions
(ADIs) have to communicate their daily risk estimates to the monetary authorities at the
beginning of the trading day, using a variety of Value-at-Risk (VaR) models to measure
risk. Sometimes the risk estimates communicated using these models are too high,
thereby leading to large capital requirements and high capital costs. At other
times, the
risk estimates are too low, leading to excessive violations, so that realised losses are
above the estimated risk. In this paper we propose a learning strategy that complements
existing methods for calculating VaR and lowers daily capital requirements, while
restricting the number of endogenous violations within the Basel II Accord penalty
limits. We suggest a decision rule that responds to violations in a discrete and
instantaneous manner, while adapting more slowly in periods of no violations. We
apply the proposed strategy to Standard & Poor’s 500 Index and show there can be
substantial savings in daily capital charges, while restricting the number of violations to
within the Basel II penalty limits.
Palabras clave: Daily capital charges, Endogenous violations, Frequency of
violations, Optimizing strategy, Risk forecasts, Value-at-risk.