CAJ | 학술논문

以黄金为代表的贵金属及其金融衍生品的交易量不断增长，逐渐成为与股票和债券平行的投资和避险工具，但关于贵金属市场风险测度的研究却比较缺乏。以上海和伦敦市场的黄金和白银交易价格为样本，基于常数高阶矩模型和时变高阶矩模型建立风险测度模型，计算出不同模型的风险价值和预期损失；采用严谨的后验分析方法，在多头和空头两种头寸共10种分位数水平下对不同模型的风险测度精确性进行后验分析。研究结果表明，在测度风险价值时，时变高阶矩模型的风险测度精确性略优于常数高阶矩模型，带有杠杠效应的时变高阶矩模型优于不带杠杆效应的时变高阶矩模型；综合对比分析不同风险测度模型的后验分析结果可知，对于准确测度贵金属市场的风险，GJR-GARCHSK模型是一个相对合理的选择。
이황금위대표적귀금속급기금융연생품적교역량불단증장，축점성위여고표화채권평행적투자화피험공구，단관우귀금속시장풍험측도적연구각비교결핍。이상해화륜돈시장적황금화백은교역개격위양본，기우상수고계구모형화시변고계구모형건립풍험측도모형，계산출불동모형적풍험개치화예기손실；채용엄근적후험분석방법，재다두화공두량충두촌공10충분위수수평하대불동모형적풍험측도정학성진행후험분석。연구결과표명，재측도풍험개치시，시변고계구모형적풍험측도정학성략우우상수고계구모형，대유강강효응적시변고계구모형우우불대강간효응적시변고계구모형；종합대비분석불동풍험측도모형적후험분석결과가지，대우준학측도귀금속시장적풍험，GJR-GARCHSK모형시일개상대합리적선택。
In the previous study of VaR（ value at risk） , the risk measurements models mainly focus on the second order moments of returns distribution （variance） ,and GARCH models have been widely used in risk measurement researches. But in the frame- work of conventional GARCH family models, the time-varying three order moments （skewness） and the time-varying four order moments（kurtosis） are not included, so the GARCH model family belongs to the ＂constant higher order moments volatility mod- el＂. However, the distribution of financial asset return does not obey standard normal distribution, but is Leptokurtic and fat tailed distribution. More and more researches have begun to explore the role of the third order moments of the return distribution （skewness） and fourth order moments of the return distribution （kurtosis） in risk management, asset pricing and option pricing, where especially in the area of risk management, the influence of third order moments and fourth order moments are very intui- tive. For example, if a portfolio returns distribution has a larger kurtosis value, the probability of extreme loss occurs will be lar- So if we set kurtosis coefficient as constant in the risk measurement models, the models will underestimate the impact of ex- treme events, and the asset management based on that risk measurement models may be facing a great loss in case of an acci- dent. Therefore, the time-varying higher order moments is considered to be as an alternative for an accurate and reliable risk measurement method. The paper takes the spot price of gold and silver of Shanghai and London market as a sample. Since the trading volume of precious metals and their derivatives is continuously growing, the precious metals are gradually parallel with stock and bond as financial investment and hedging tools. However, the researches done on risk measurements of precious metals market have been relatively less. Hence, this paper uses both ＂constant higher order moments volatility model＂（ GARCH mod- els） and ＂time-varying higher order moment models＂ as risk measurement models to estimate VaR （Value at risk） values and ES （Excepted shortfall） values. Then this paper makes backtesting analysis respectively in the long and short positions. According to the empirical results, a series of important conclusions are concluded：（1）the accuracy of Time-varying higher order moment model is better than that of constant higher order moment model, （2）the time-varying higher order moment models with the lever- age effect is better than that of the time-varying higher order moment model with no leverage effect, （3）the GJR-GARCHSK is of the highest accuracy model in estimating ES.