Kurtosis can be measured in more than one way. A modification of Geary's measure of kurtosis is shown to be more sensitive to kurtosis in the center of the distribution while Pearson's measure of kurtosis is more sensitive to kurtosis in the tails of the distribution. The modified Geary measure and the Pearson measure are used to define a joint test of kurtosis that has high uniform power across a very wide range of symmetric nonnormal distributions.

Mémoire numérisé par la Division de la gestion de documents et des archives de l'Université de Montréal.

Mémoire numérisé par la Division de la gestion de documents et des archives de l'Université de Montréal

Submitted by Alessandra Gazzoli (alessandra.gazzoli-santos@itau-unibanco.com.br) on 2013-09-13T18:41:48Z No. of bitstreams: 1 AlessandraGazzoli_EstruturaFractal.pdf: 2516271 bytes, checksum: 6f74e2f1266b906cb221034fae87335f (MD5) / Approved for entry into archive by Suzinei Teles Garcia Garcia (suzinei.garcia@fgv.br) on 2013-09-13T18:44:12Z (GMT) No. of bitstreams: 1 AlessandraGazzoli_EstruturaFractal.pdf: 2516271 bytes, checksum: 6f74e2f1266b906cb221034fae87335f (MD5) / Made available in DSpace on 2013-09-13T18:45:22Z (GMT). No. of bitstreams: 1 AlessandraGazzoli_EstruturaFractal.pdf: 2516271 bytes, checksum: 6f74e2f1266b906cb221034fae87335f (MD5) Previous issue date: 2013-08-14 / Economic variables are often governed by dynamic and non-linear processes that can originate long-term relationship and non-periodic and non-cyclical patterns with abrupt trend changes. Commodity prices exhibit this type of behavior and the peculiarities of those markets could generate fractionally integrated time series, whose singularities could not be properly captured by the traditional analytic models based on the efficient market hypothesis and random walk processes. Therefore, this study has investigated the presence of fractal structures on some very important Brazilian commodity spot markets such as coffee, cattle, sugar, soybean and calf. Some traditional techniques were used as well as other specific for fractal time series analysis, such as rescaled range (R/S) analysis, different fractal hypothesis tests and ARFIMA and FIGARCH models. The results showed that the drift component has not shown fractal behavior, except for the calf series, however, volatility has demonstrated fractal behavior for all the commodities that were analyzed. / As variáveis econômicas são frequentemente governadas por processos dinâmicos e não-lineares que podem gerar relações de dependência de longo prazo e padrões cíclicos não-periódicos com mudanças abruptas de tendências. Para o caso dos preços agrícolas este comportamento não é diferente e as peculiaridades destes mercados podem gerar séries temporais fracionalmente integradas, cujas singularidades não seriam adequadamente capturadas pelos tradicionais modelos analíticos fundamentados na hipótese dos mercados eficientes e de passeio aleatório. Sendo assim, o presente estudo buscou investigar a presença de estruturas fractais no mercado à vista de algumas das principais commodities agrícolas brasileiras: café, boi gordo, açúcar, milho, soja e bezerro. Foram empregadas técnicas tradicionais e específicas para a análise de séries temporais fractais como a análise de R/S e a aplicação de modelos das famílias ARFIMA e FIGARCH. Os resultados indicaram que, com exceção do bezerro, o componente de drift destas séries não apresentou comportamento fractal, ao contrário do observado para o componente da volatilidade, que apresentou aspecto de estrutura fractal para todas as commodities analisadas.

風險值大多是在假設資產報酬為常態分配下計算而得的，但是這個假設與實際的資產報酬分配不一致，因為很多研究者都發現實際的資產報酬分配都有厚尾的現象，也就是極端事件的發生機率遠比常態假設要來的高，因此利用常態假設來計算風險值對於真實損失的衡量不是很恰當。 針對這個問題，本論文以歷史模擬法、變異數-共變異數法、混合常態模型來模擬報酬率的分配，並依給定的信賴水準估算出風險值，其中混合常態模型的參數是利用準貝式最大概似估計法及EM演算法來估計；然後利用三種風險值的評量方法：回溯測試、前向測試與二項檢定，來評判三種估算風險值方法的優劣。 經由實證結果發現： 1.報酬率分配在左尾臨界機率1％有較明顯厚尾的現象。 2.利用混合常態分配來模擬報酬率分配會比另外兩種方法更能準確的捕捉到左尾臨界機率1％的厚尾。 3.混合常態模型的峰態係數值接近於真實報酬率分配的峰態係數值，因此我們可以確認混合常態模型可以捕捉高峰的現象。 關鍵字：風險值、厚尾、歷史模擬法、變異數-共變異教法、混合常態模型、準貝式最大概似估計法、EM演算法、回溯測試、前向測試、高峰 / Initially, Value at Risk (VaR) is calculated by assuming that the underline asset return is normal distribution, but this assumption sometimes does not consist with the actual distribution of asset return. Many researchers have found that the actual distribution of the underline asset return have Fat-Tail, extreme value events, character. So under normal distribution assumption, the VaR value is improper compared with the actual losses. The paper discuss three methods. Historical Simulated method - Variance-Covariance method and Mixture Normal .simulating those asset, return and VaR by given proper confidence level. About the Mixture Normal Distribution, we use both EM algorithm and Quasi-Bayesian MLE calculating its parameters. Finally, we use tree VaR testing methods, Back test、Forward tes and Binomial test -----comparing its VaR loss probability We find the following results: 1.Under 1% left-tail critical probability, asset return distribution has significant Fat-tail character. 2.Using Mixture Normal distribution we can catch more Fat-tail character precisely than the other two methods. 3.The kurtosis of Mixture Normal is close to the actual kurtosis, this means that the Mixture Normal distribution can catch the Leptokurtosis phenomenon. Key words: Value at Risk、VaR、Fat tail、Historical simulation method、 Variance-Covariance method、Mixture Normal distribution、Quasi-Bayesian MLE、EM algorithm、Back test、 Forward test、 Leptokurtosis