Bootstrap historical simulation

Abstract

In this paper the authors present a new VaR model for the estimation of market risk in banks and other financial institutions. The model is labeled BootstrapHS500, since it is theoretically based on historical simulation and implementation of the bootstrap method. The aim of the paper is to provide answers to the question whether incorporating the bootstrap method in the standard model of historical simulation contributes to the improvement of the historical simulation's applicability in terms of meeting the backtesting rules of the Basel II standards. In order to obtain the answer to this question, in this paper we conducted and dealt with the testing of applicability and comparison of performances of the HS500 and the BootstrapHS500 models at the capital markets of Serbia, Croatia, Greece, Spain, Germany, Slovakia, the Czech Republic, Romania and Hungary. The research methodology involves the use of the appropriate quantitative analysis and tests of unconditional and conditional coverage. The results of the research show that the BootstrapHS500 model achieved a better performance than the standard model of historical simulation, from the perspective of both backtesting tests. It can be concluded that incorporating the bootstrap method in the standard model of historical simulation contributes to the improvement of the historical simulation' applicability.

U radu autori predstavljaju novi model VaR za procenu tržišnog rizika u bankama i drugim finansijskim institucijama. Model je označen kao BootstrapHS500, budući da se zasniva na teorijskim osnovama istorijske simulacije VaR i primeni bootstrap metode. Glavni cilj rada jeste da se dobije odgovor na pitanje da li inkorporiranje bootstrap metoda u standardni model istorijske simulacije vrednosti pri riziku dovodi do unapređenja aplikativnosti istorijske simulacije u kontekstu zadovoljenja pravila validnosti internih modela uz procenu tržišnog rizika Bazel II standarda. Kako bi se odgovorilo na ovo pitanje u radu je izvršeno testiranje aplikativnosti i komparacija performansi HS500 i BootstrapHS500 modela na tržištu kapitala Srbije, Hrvatske, Grčke, Španije, Nemačke, Slovačke, Češke, Rumunije i Mađarske. Metodologija istraživanja podrazumeva primenu odgovarajuće kvantitativne analize i testa bezuslovnog i uslovnog pokrića. Rezultati istraživanja pokazuju da BootstrapHS500 model, na testiranim tržištima, postiže bolje performanse iz vizure oba testa, te da se može zaključiti da inkorporiranje bootstrap metode u standardni model istorijske simulacije doprinosi unapređenju njegove aplikativnosti.