Construction of a financial risk engineering model for banking supervision in the face of systemic crises

Abstract

The paper develops a model that improves the measurement of correlations present in stress scenarios through the use of copulas, reorders the propagation of shocks and involves expert judgments to improve predictions through a Bayesian VAR, it is shown that, under scenarios of a systemic crisis, losses can reach high percentages. Considering a loss rate associated with counterparty default of 45% and a default threshold between 4% and 8% suggested by the Basel Committee on Banking Supervision, it is estimated that an external shock can generate falls of more than 10% in the different financial variables: the savings rate, stock market indexes, the exchange rate, among others. If a counterparty defaults, it generates 45% of the associated losses (exposure), each institution absorbs 45% of its exposures. The model constructed is applicable to regulatory agencies because it exposes a propagation mechanism through a financial contagion resulting from an external shock and subjected to stress tests. Taking into account the above assumptions, it is found that real variables can be affected by more than 15%. Although these rates are extreme and the stress scenario unlikely, it is necessary to consider these effects for the prevention of systemic crises, so it is advisable for regulatory authorities to emphasize the regulatory capital required from financial institutions.