Historical VaR, CVaR and ES to Liquidation: 'A New Tool to Measure Risk by Regulators and Fund Managers'

Abstract

Historical VaR, CVaR and ES (Expected Shortfall) to LIQUIDATION Software is a model characterized by its straightforwardness, allowing regulators measure risk using a standard database of primitive factors and portfolio positions only, leaving little error margin in comparing market risk for different financial funds. As such, it should be a tool of preference for government and supra-government regulatory institutions and is the required VaR measure of government regulators of private pension funds in Latin America. The proposed model is based in the fundamental premise that there is a limit to the amount a financial asset can be sold in a given trading day without the market detecting the fund is in unwind mode. To remain consistent with my previous models, I am not shy of dispensing with premises behind valuation models when the real world demands so, like short-term Efficient-Market-Hypothesis, EMH. In addition, the model includes a new measure of risk: a liquidity haircut for each asset and trade size. Like in my previous work, there is no Horizon H – due to different asset liquidity – in calculating the risk metrics. The algorithm developed computes simultaneously VaR and ES. (Since CVaR is defined in the specialized literature as equal to ES, I shall dispense of the former, making sole reference to ES throughout the paper.) In each simulation run Individual and Component VaR are calculated together with Historical VaR, with a good deal of information obtained in a relatively short period of time. The following version, to be release shortly, will include Individual and Component ES, as well as Marginal and Incremental VaR and ES.