CAJ | 학술논문

本文以成熟市场和新兴市场的六个主要的市场指数为例，将更精确反映金融资产收益率典型事实的AEPD分布和ALD分布运用于股票市场VaR的度量。并与其他常见的非参、半参和参数法VaR模型进行全面比较。实证表明，对于参数法模型，误差项服从ALD分布和正态分布的GARCH族模型分别当且仅当在度量低分位数和高分位数水平下的VaR值时表现优异；而误差项服从AEPD分布的GARCH族模型在度量各种分位数水平下的VaR值时均取得不错的效果。另外对于CAViaR模型，它们在度量VaR时与参数法中表现最好的ARrGIR-GARCH—AEPD（ALD）两个模型效果相当。
본문이성숙시장화신흥시장적륙개주요적시장지수위례，장경정학반영금융자산수익솔전형사실적AEPD분포화ALD분포운용우고표시장VaR적도량。병여기타상견적비삼、반삼화삼수법VaR모형진행전면비교。실증표명，대우삼수법모형，오차항복종ALD분포화정태분포적GARCH족모형분별당차부당재도량저분위수화고분위수수평하적VaR치시표현우이；이오차항복종AEPD분포적GARCH족모형재도량각충분위수수평하적VaR치시균취득불착적효과。령외대우CAViaR모형，타문재도량VaR시여삼수법중표현최호적ARrGIR-GARCH—AEPD（ALD）량개모형효과상당。
The paper takes the six major market indices in mature markets and emerging markets for example, applying the Asymmetric Exponential Power Distribution （AEPD） and Asymmetric Laplace Distribution （ALD） which can more accurately reflect the stylized {acts of the return on financial assets to measure Value at Risk （VaR） in the stock market. Then we compare these models with other common non-parametric, semi-parametric and parametric VaR models. The empirical results show that for the parametric models, the GARCH models which the error term obeys the ALD and normal distribution do well if and only if we measure low and high quantile VaR respectively. The GARCH models which the error term obeys the AEPD perform well both in the high and low quantile VaR measure. In addition, for the CAViaR models, they are as good as the two best parametric models that are AR-GJR-GARCH-AEPD model and AR-GJR-GARCH-ALD model respectively.