The purpose of this thesis was to evaluate a method for reducing the bias of estimation for autocovariance estimators. Two methods are compared, one is the standard method and the other is an adjustment method. The Monte Carlo method is used within comparison. The bias and the mean squared error of the estimated autocovariance is computed for several time series models and two variations of the adjustment method of estimation. The results indicate some improvement in bias and mean squared error for the new method.

Uma correta especificação das funções densidade de probabilidade (fdp’s) de retornos de ativos é um tópico dos mais relevantes na literatura de modelagem econométrica de dados financeiros. A presente dissertação propõe-se a oferecer, neste âmbito, uma abordagem distinta, através de uma aplicação da metodologia desenvolvida em Bathia et al. (2010) a dados intradiários do índice bovespa. Esta abordagem consiste em focar a análise diretamente sobre a estrutura dinâmica das fdp’s dos retornos, enxergando-as como uma sequência de variáveis aleatórias que tomam valores em um espaço de funções. A dependência serial existente entre essas curvas permite que se obtenham estimativas filtradas das fdp’s, e mesmo que se façam previsões sobre densidades de períodos subsequentes à amostra. No artigo que integra esta dissertação, onde é feita a mencionada aplicação, encontrou-se evidência de que o comportamento dinâmico das fdp’s dos retornos do índice bovespa se reduz a um processo bidimensional, o qual é bem representado por um modelo var(1) e cuja dinâmica afeta a dispersão e a assimetria das distribuições no suceder dos dias. Ademais, utilizando-se de subamostras, construíram-se previsões um passo à frente para essas fdp’s, e avaliaram-se essas previsões de acordo com métricas apropriadas. / Adequate specification of the probability density functions (pdf’s) of asset returns is a most relevant topic in econometric modelling of financial data. This dissertation aims to provide a distinct approach on that matter, through applying the methodology developed in Bathia et al. (2010) to intraday bovespa index data. This approach consists in focusing the analysis directly on the dynamic structure of returns fdp’s, seeing them as a sequence of function-valued random variables. The serial dependence of these curves allows one to obtain filtered estimates of the pdf’s, and even to forecast upcoming densities. In the paper contained into this dissertation, evidence is found that the dynamic structure of the bovespa index returns pdf’s reduces to a R2-valued process, which is well represented by a var(1) model, and whose dynamics affect the dispersion and symmetry of the distributions at each day. Moreover, one-step-ahead forecasts of upcoming pdf’s were constructed through subsamples and evaluated according to appropriate metrics.

Uma correta especificação das funções densidade de probabilidade (fdp’s) de retornos de ativos é um tópico dos mais relevantes na literatura de modelagem econométrica de dados financeiros. A presente dissertação propõe-se a oferecer, neste âmbito, uma abordagem distinta, através de uma aplicação da metodologia desenvolvida em Bathia et al. (2010) a dados intradiários do índice bovespa. Esta abordagem consiste em focar a análise diretamente sobre a estrutura dinâmica das fdp’s dos retornos, enxergando-as como uma sequência de variáveis aleatórias que tomam valores em um espaço de funções. A dependência serial existente entre essas curvas permite que se obtenham estimativas filtradas das fdp’s, e mesmo que se façam previsões sobre densidades de períodos subsequentes à amostra. No artigo que integra esta dissertação, onde é feita a mencionada aplicação, encontrou-se evidência de que o comportamento dinâmico das fdp’s dos retornos do índice bovespa se reduz a um processo bidimensional, o qual é bem representado por um modelo var(1) e cuja dinâmica afeta a dispersão e a assimetria das distribuições no suceder dos dias. Ademais, utilizando-se de subamostras, construíram-se previsões um passo à frente para essas fdp’s, e avaliaram-se essas previsões de acordo com métricas apropriadas. / Adequate specification of the probability density functions (pdf’s) of asset returns is a most relevant topic in econometric modelling of financial data. This dissertation aims to provide a distinct approach on that matter, through applying the methodology developed in Bathia et al. (2010) to intraday bovespa index data. This approach consists in focusing the analysis directly on the dynamic structure of returns fdp’s, seeing them as a sequence of function-valued random variables. The serial dependence of these curves allows one to obtain filtered estimates of the pdf’s, and even to forecast upcoming densities. In the paper contained into this dissertation, evidence is found that the dynamic structure of the bovespa index returns pdf’s reduces to a R2-valued process, which is well represented by a var(1) model, and whose dynamics affect the dispersion and symmetry of the distributions at each day. Moreover, one-step-ahead forecasts of upcoming pdf’s were constructed through subsamples and evaluated according to appropriate metrics.

Uma correta especificação das funções densidade de probabilidade (fdp’s) de retornos de ativos é um tópico dos mais relevantes na literatura de modelagem econométrica de dados financeiros. A presente dissertação propõe-se a oferecer, neste âmbito, uma abordagem distinta, através de uma aplicação da metodologia desenvolvida em Bathia et al. (2010) a dados intradiários do índice bovespa. Esta abordagem consiste em focar a análise diretamente sobre a estrutura dinâmica das fdp’s dos retornos, enxergando-as como uma sequência de variáveis aleatórias que tomam valores em um espaço de funções. A dependência serial existente entre essas curvas permite que se obtenham estimativas filtradas das fdp’s, e mesmo que se façam previsões sobre densidades de períodos subsequentes à amostra. No artigo que integra esta dissertação, onde é feita a mencionada aplicação, encontrou-se evidência de que o comportamento dinâmico das fdp’s dos retornos do índice bovespa se reduz a um processo bidimensional, o qual é bem representado por um modelo var(1) e cuja dinâmica afeta a dispersão e a assimetria das distribuições no suceder dos dias. Ademais, utilizando-se de subamostras, construíram-se previsões um passo à frente para essas fdp’s, e avaliaram-se essas previsões de acordo com métricas apropriadas. / Adequate specification of the probability density functions (pdf’s) of asset returns is a most relevant topic in econometric modelling of financial data. This dissertation aims to provide a distinct approach on that matter, through applying the methodology developed in Bathia et al. (2010) to intraday bovespa index data. This approach consists in focusing the analysis directly on the dynamic structure of returns fdp’s, seeing them as a sequence of function-valued random variables. The serial dependence of these curves allows one to obtain filtered estimates of the pdf’s, and even to forecast upcoming densities. In the paper contained into this dissertation, evidence is found that the dynamic structure of the bovespa index returns pdf’s reduces to a R2-valued process, which is well represented by a var(1) model, and whose dynamics affect the dispersion and symmetry of the distributions at each day. Moreover, one-step-ahead forecasts of upcoming pdf’s were constructed through subsamples and evaluated according to appropriate metrics.

Dans ce mémoire, nous avons utilisé le logiciel R pour la programmation. / Le présent mémoire porte sur les séries chronologiques qui en plus d’être observées dans le temps, présentent également une composante spatiale. Plus particulièrement, nous étudions une certaine classe de modèles, les modèles autorégressifs spatio-temporels généralisés, ou GSTAR. Dans un premier temps, des liens sont effectués avec les modèles vectoriels autorégressifs (VAR). Nous obtenons explicitement la distribution asymptotique des autocovariances résiduelles pour les modèles GSTAR en supposant que le terme d’erreur est un bruit blanc gaussien, ce qui représente une première contribution originale. De ce résultat, des tests de type portemanteau sont proposés, dont les distributions asymptotiques sont étudiées. Afin d’illustrer la performance des statistiques de test, une étude de simulations est entreprise où des modèles GSTAR sont simulés et correctement ajustés. La méthodologie est illustrée avec des données réelles. Il est question de la production mensuelle de thé en Java occidental pour 24 villes, pour la période janvier 1992 à décembre 1999. / In this master thesis, time series models are studied, which have also a spatial component, in addition to the usual time index. More particularly, we study a certain class of models, the Generalized Space-Time AutoRegressive (GSTAR) time series models. First, links are considered between Vector AutoRegressive models(VAR) and GSTAR models. We obtain explicitly the asymptotic distribution of the residual autocovariances for the GSTAR models, assuming that the error term is a Gaussian white noise, which is a first original contribution. From that result, test statistics of the portmanteau type are proposed, and their asymptotic distributions are studied. In order to illustrate the behaviour of the test statistics, a simulation study is conducted where GSTAR models are simulated and correctly fitted. The methodology is illustrated with monthly real data concerning the production of tea in west Java for 24 cities from the period January 1992 to December 1999.

Le présent mémoire porte sur les séries chronologiques qui en plus d’être observées dans le temps, présentent également une composante spatiale. Plus particulièrement, nous étudions une certaine classe de modèles, les modèles autorégressifs spatio-temporels généralisés, ou GSTAR. Dans un premier temps, des liens sont effectués avec les modèles vectoriels autorégressifs (VAR). Nous obtenons explicitement la distribution asymptotique des autocovariances résiduelles pour les modèles GSTAR en supposant que le terme d’erreur est un bruit blanc gaussien, ce qui représente une première contribution originale. De ce résultat, des tests de type portemanteau sont proposés, dont les distributions asymptotiques sont étudiées. Afin d’illustrer la performance des statistiques de test, une étude de simulations est entreprise où des modèles GSTAR sont simulés et correctement ajustés. La méthodologie est illustrée avec des données réelles. Il est question de la production mensuelle de thé en Java occidental pour 24 villes, pour la période janvier 1992 à décembre 1999. / In this master thesis, time series models are studied, which have also a spatial component, in addition to the usual time index. More particularly, we study a certain class of models, the Generalized Space-Time AutoRegressive (GSTAR) time series models. First, links are considered between Vector AutoRegressive models(VAR) and GSTAR models. We obtain explicitly the asymptotic distribution of the residual autocovariances for the GSTAR models, assuming that the error term is a Gaussian white noise, which is a first original contribution. From that result, test statistics of the portmanteau type are proposed, and their asymptotic distributions are studied. In order to illustrate the behaviour of the test statistics, a simulation study is conducted where GSTAR models are simulated and correctly fitted. The methodology is illustrated with monthly real data concerning the production of tea in west Java for 24 cities from the period January 1992 to December 1999. / Dans ce mémoire, nous avons utilisé le logiciel R pour la programmation.

Ce travail porte sur les séries spatiales. On étudie les phénomènes dont l’observation est un processus aléatoire indexé par un ensemble spatial. Dans cette thèse on s’intéresse aux données bidimensionnelles régulièrement dispersées dans l’espace, on travaille alors dans un rectangle régulier (sur Z2) . Cette modélisation vise donc à construire des représentations des systèmes suivant leurs dimensions spatiales et à ses applications dans de nombreux domaines tels que la météorologie, l’océanographie, l’agronomie, la géologie, l’épidémiologie, ou encore l’économétrie etc. La modélisation spatiale permet d’aborder la question importante de la prédiction de la valeur d’un champ aléatoire en un endroit donné d’une région. On suppose que la valeur à prédire dépend des observations dans les régions voisines. Ceci montre la nécessité de tenir compte, en plus de leurs caractéristiques statistiques, des relations de dépendance spatiale entre localisations voisines, pour rendre compte de l’ensemble des structures inhérentes aux données. Dans la plupart des champs d’applications, on est souvent confronté du fait que l’une des sources majeures de fluctuations est la saisonnalité. Dans nos travaux on s’intéresse particulièrement à ce phénomène de saisonnalité dans les données spatiales. Faire une modélisation mathématique en tenant en compte l’interaction spatiale des différents points ou localités d’une zone entière serait un apport considérable. En effet un traitement statistique qui prendrait en compte cet aspect et l’intègre de façon adéquat peut corriger une perte d’information, des erreurs de prédictions, des estimations non convergentes et non efficaces. / This thesis focuses on the time series in addition to being observed over time, also have a spatial component. By definition, a spatiotemporal phenomenon is a phenomenon which involves a change in space and time. The spatiotemporal model-ling therefore aims to construct representations of systems taking into account their spatial and temporal dimensions. It has applications in many fields such as meteorology, oceanography, agronomy, geology, epidemiology, image processing or econometrics etc. It allows them to address the important issue of predicting the value of a random field at a given location in a region. Assume that the value depends predict observations in neighbouring regions. This shows the need to consider, in addition to their statistical characteristics, relations of spatial dependence between neighbouring locations, to account for all the inherent data structures. In the exploration of spatiotemporal data, refinement of time series models is to explicitly incorporate the systematic dependencies between observations for a given region, as well as dependencies of a region with neighboring regions. In this context, the class of spatial models called spatiotemporal auto-regressive models (Space-Time Autoregressive models) or STAR was introduced in the early 1970s. It will then be generalized as GSTAR model (Generalized Space-Time Autoregressive models). In most fields of applications, one is often confronted by the fact that one of the major sources of fluctuations is seasonality. In our work we are particularly interested in the phenomenon of seasonality in spatiotemporal data. We develop a new class of models and investigates the properties and estimation methods. Make a mathematical model taking into account the spatial inter-action of different points or locations of an entire area would be a significant contribution. Indeed, a statistical treatment that takes into account this aspect and integrates appropriate way can correct a loss of information, errors in predictions, non-convergent and inefficient estimates.

Thèse numérisée par la Division de la gestion de documents et des archives de l'Université de Montréal.

Thèse numérisée par la Division de la gestion de documents et des archives de l'Université de Montréal