Value-at-Risk Prediction: A Comparison of Alternative Strategies

Abstract

Given the growing need for managing financial risk, risk prediction plays an increasing role in banking and finance. In this study we compare the out-of-sample performance of existing methods and some new models for predicting value-at-risk (VaR) in a univariate context. Using more than 30 years of the daily return data on the NASDAQ Composite Index, we find that most approaches perform inadequately, although several models are acceptable under current regulatory assessment rules for model adequacy. A hybrid method, combining a heavy-tailed generalized autoregressive conditionally heteroskedastic (GARCH) filter with an extreme value theory-based approach, performs best overall, closely followed by a variant on a filtered historical simulation, and a new model based on heteroskedastic mixture distributions. Conditional autoregressive VaR (CAViaR) models perform inadequately, though an extension to a particular CAViaR model is shown to outperform the others.

Forecasting Value-at-Risk under Temporal and Portfolio Aggregation*

“…In line with our findings, Santos et al (2013) report that multivariate models perform better than univariate models, but they do not consider intermediate degrees of aggregation. The importance of the distribution choice is in line with Giot and Laurent (2004); Kuester et al (2006) ;Bao et al (2006); Clements et al (2008); McAleer and Da Veiga (2008).…”

“…5 Using an asymmetric loss function, we add a new perspective. 6 Also when tail loss is important, iterated forecasts are preferred, followed by scaled and 4 Giacomini and Komunjer (2005); Bao et al (2006); Kuester et al (2006); McAleer and Da Veiga (2008); Santos et al (2013) use daily data, whereas Giot and Laurent (2004); Brownlees and Gallo (2010); Clements et al (2008) also use intraday data.…”

Quantitative risk assessment of hypothetic investment portfolio : the case of the Baltic States

“…The main pros of Historical methods are that these methods do not assume any distribution on the asset returns and it is relatively easy to implement (e.g., Lambadiaris et al, 2003;. However, there are some cons of historical methods:…”

Forecasting value-at-risk by encompassing CAViaR models via information criteria

“…With regard to the VaR prediction, many competing methods have been proposed. For a review of those methods, we refer to Kuester et al (2006). As examples of VaR analysis for Korean time series, Shim and Hwang (2011) studied the conditional autoregressive VaR (CAViaR) models by Engle and Manganelli (2004) based on support vector quantile regression; Kim and Lee (2011) considered the VaR forecasting of portfolios using conditional copula; Lee and Noh (2010) studied the forecasting based on quantile regression for GARCH models; Choi et al (2007) applied the multivariate GARCH models.…”

“…They presented a class of CAViaR models for the parametric specification of the VaR. Many studies suggest that the CAViaR method shows satisfactory performance in various situation: see, for instance, Kuester et al (2006) and Bao et al (2006). Since the CAViaR models consist of various specifications, one may encounter a practical problem of choosing a proper specification among many of them.…”