The robustness of VaR models during the Ukrainian crisis and the impact of ESG scores on the results of US ETF backtests.

Abstract

Under the supervision of the Commission de Surveillance du Secteur financier (CSSF), the financial regulator in Luxembourg, the management of tail risk for UCITS funds relies on the daily computation of Value at Risk (VaR), which is then backtested monthly to ensure its reliability. Consistency in VaR models is determined by the level of exceptions aligning with the confidence interval and their independence from each other. The robustness of VaR models has faced challenges during the Ukrainian crisis and the subsequent announcements by central banks regarding interest rate hikes. During extreme market events, VaR may not fully capture the risk, leading to the rejection of VaR models in the backtesting process. This paper argues that parametric VaR methodologies applied to US ETFs exhibit greater robustness compared to historical and Monte Carlo methods during the shocks caused by the Ukrainian war. By comparing backtesting results, it has been demonstrated that the exceptions were independent from each other according to the Haas test, and the proportion of failure aligned more coherently with the estimated confidence level. This is attributed to the ability of parametric models to select the skewed student-t distribution, which effectively captures extreme downward movements in asset returns. Previous research has examined the relationship between ESG scores and VaR levels, particularly during financial crises, and indicated a downward effect. This study aimed to assess whether there is a downward relationship between ESG scores and backtesting results, as a reduction in VaR level and lower sensitivity to extreme market shifts should correspond to a lower number of exceptions and an accepted VaR model. Surprisingly, the variables related to backtesting results, such as the number of exceptions and the results of the Kupiec proportion of failure test, were positively correlated with ESG scores and their sub-pillars (Economic, Social, and Governance). Furthermore, the fund with the lowest ESG score exhibited the fewest exceptions, and its VaR models were generally accepted by the different backtesting procedures, unlike other ETFs with higher scores. Therefore, for US ETFs during the Ukrainian crisis, no downward relationship between ESG scores and backtesting results was found.