Asymptotic Theory for Extended Asymmetric Multivariate GARCH Processes

The paper considers various extended asymmetric multivariate conditional volatility models, and derives appropriate regularity conditions and associated asymptotic theory. This enables checking of internal consistency and allows valid statistical inferences to be drawn based on empirical estimation. For this purpose, we use an underlying vector random coefficient autoregressive process, for which we show the equivalent representation for the asymmetric multivariate conditional volatility model, to derive asymptotic theory for the quasi-maximum likelihood estimator. As an extension, we develop a new multivariate asymmetric long memory volatility model, and discuss the associated asymptotic properties.

Realized Matrix-Exponential Stochastic Volatility with Asymmetry, Long Memory and Spillovers

The paper develops a novel realized matrix-exponential stochastic volatility model of multivariate returns and realized covariances that incorporates asymmetry and long memory (hereafter the RMESV-ALM model). The matrix exponential transformation guarantees the positivedefiniteness of the dynamic covariance matrix. The contribution of the paper ties in with Robert Basmann’s seminal work in terms of the estimation of highly non-linear model specifications (“Causality tests and observationally equivalent representations of econometric models”, Journal of Econometrics, 1988, 39(1-2), 69–104), especially for developing tests for leverage and spillover effects in the covariance dynamics. Efficient importance sampling is used to maximize the likelihood function of RMESV-ALM, and the finite sample properties of the quasi-maximum likelihood estimator of the parameters are analysed. Using high frequency data for three US financial assets, the new model is estimated and evaluated. The forecasting performance of the new model is compared with a novel dynamic realized matrix-exponential conditional covariance model. The volatility and co-volatility spillovers are examined via the news impact curves and the impulse response functions from returns to volatility and co-volatility.

Is alcohol consumption related to likelihood of reporting chronic widespread pain in people with stable consumption? : Results from UK Biobank

Acknowledgements This research has been conducted using the UK Biobank resource, and was funded by the University of Aberdeen. The authors have no conflicts of interest to declare.

Contribution à l'évaluation de l'équilibre quasi-statique à l'aide d'une plate-forme de force

Thèse numérisée par la Direction des bibliothèques de l'Université de Montréal.

Dynamic Programming Approaches for Estimating and Applying Large-scale Discrete Choice Models

People go through their life making all kinds of decisions, and some of these decisions affect their demand for transportation, for example, their choices of where to live and where to work, how and when to travel and which route to take. Transport related choices are typically time dependent and characterized by large number of alternatives that can be spatially correlated. This thesis deals with models that can be used to analyze and predict discrete choices in large-scale networks. The proposed models and methods are highly relevant for, but not limited to, transport applications. We model decisions as sequences of choices within the dynamic discrete choice framework, also known as parametric Markov decision processes. Such models are known to be difficult to estimate and to apply to make predictions because dynamic programming problems need to be solved in order to compute choice probabilities. In this thesis we show that it is possible to explore the network structure and the flexibility of dynamic programming so that the dynamic discrete choice modeling approach is not only useful to model time dependent choices, but also makes it easier to model large-scale static choices. The thesis consists of seven articles containing a number of models and methods for estimating, applying and testing large-scale discrete choice models. In the following we group the contributions under three themes: route choice modeling, large-scale multivariate extreme value (MEV) model estimation and nonlinear optimization algorithms. Five articles are related to route choice modeling. We propose different dynamic discrete choice models that allow paths to be correlated based on the MEV and mixed logit models. The resulting route choice models become expensive to estimate and we deal with this challenge by proposing innovative methods that allow to reduce the estimation cost. For example, we propose a decomposition method that not only opens up for possibility of mixing, but also speeds up the estimation for simple logit models, which has implications also for traffic simulation. Moreover, we compare the utility maximization and regret minimization decision rules, and we propose a misspecification test for logit-based route choice models. The second theme is related to the estimation of static discrete choice models with large choice sets. We establish that a class of MEV models can be reformulated as dynamic discrete choice models on the networks of correlation structures. These dynamic models can then be estimated quickly using dynamic programming techniques and an efficient nonlinear optimization algorithm. Finally, the third theme focuses on structured quasi-Newton techniques for estimating discrete choice models by maximum likelihood. We examine and adapt switching methods that can be easily integrated into usual optimization algorithms (line search and trust region) to accelerate the estimation process. The proposed dynamic discrete choice models and estimation methods can be used in various discrete choice applications. In the area of big data analytics, models that can deal with large choice sets and sequential choices are important. Our research can therefore be of interest in various demand analysis applications (predictive analytics) or can be integrated with optimization models (prescriptive analytics). Furthermore, our studies indicate the potential of dynamic programming techniques in this context, even for static models, which opens up a variety of future research directions.

Contribution à l'évaluation de l'équilibre quasi-statique à l'aide d'une plate-forme de force

Thèse numérisée par la Direction des bibliothèques de l'Université de Montréal.

Effect of Performance Feedback on Perceived Knowledge and Likelihood to Pursue Continuing Education

The purpose of this study was to assess the effect of performance feedback on Athletic Trainers’ (ATs) perceived knowledge (PK) and likelihood to pursue continuing education (CE). The investigation was grounded in the theories of “the definition of the situation” (Thomas & Thomas, 1928) and the “illusion of knowing,” (Glenberg, Wilkinson, & Epstein, 1982) suggesting that PK drives behavior. This investigation measured the degree to which knowledge gap predicted CE seeking behavior by providing performance feedback designed to change PK. A pre-test post-test control-group design was used to measure PK and likelihood to pursue CE before and after assessing actual knowledge. ATs (n=103) were randomly sampled and assigned to two groups, with and without performance feedback. Two independent samples t-tests were used to compare groups on the difference scores of the dependent variables. Likelihood to pursue CE was predicted by three variables using multiple linear regression: perceived knowledge, pre-test likelihood to pursue CE, and knowledge gap. There was a 68.4% significant difference (t101= 2.72, p=0.01, ES=0.45) between groups in the change scores for likelihood to pursue CE because of the performance feedback (Experimental group=13.7% increase; Control group= 4.3% increase). The strongest relationship among the dependent variables was between pre-test and post-test measures of likelihood to pursue CE (F2,102=56.80, p<0.01, r=0.73, R2=0.53). The pre- and post-test predictive relationship was enhanced when group was included in the model. In this model [YCEpost=0.76XCEpre-0.34 Xgroup+2.24+E], group accounted for a significant amount of unique variance in predicting CE while the pre-test likelihood to pursue CE variable was held constant (F3,102=40.28, p<0.01,: r=0.74, R2=0.55). Pre-test knowledge gap, regardless of group allocation, was a linear predictor of the likelihood to pursue CE (F1,102=10.90, p=.01, r=.31, R2=.10). In this investigation, performance feedback significantly increased participants’ likelihood to pursue CE. Pre-test knowledge gap was a significant predictor of likelihood to pursue CE, regardless if performance feedback was provided. ATs may have self-assessed and engaged in internal feedback as a result of their test-taking experience. These findings indicate that feedback, both internal and external, may be necessary to trigger CE seeking behavior.

An R-package for the Estimation and Testing of Multiple Covariates and Biomarker Interactions for Survival Data Based on Local Partial Likelihood

When we study the variables that a ffect survival time, we usually estimate their eff ects by the Cox regression model. In biomedical research, e ffects of the covariates are often modi ed by a biomarker variable. This leads to covariates-biomarker interactions. Here biomarker is an objective measurement of the patient characteristics at baseline. Liu et al. (2015) has built up a local partial likelihood bootstrap model to estimate and test this interaction e ffect of covariates and biomarker, but the R code developed by Liu et al. (2015) can only handle one variable and one interaction term and can not t the model with adjustment to nuisance variables. In this project, we expand the model to allow adjustment to nuisance variables, expand the R code to take any chosen interaction terms, and we set up many parameters for users to customize their research. We also build up an R package called "lplb" to integrate the complex computations into a simple interface. We conduct numerical simulation to show that the new method has excellent fi nite sample properties under both the null and alternative hypothesis. We also applied the method to analyze data from a prostate cancer clinical trial with acid phosphatase (AP) biomarker.

Multivariate Appoximations to Portfolio Return Distribution

This article proposes a three-step procedure to estimate portfolio return distributions under the multivariate Gram-Charlier (MGC) distribution. The method combines quasi maximum likelihood (QML) estimation for conditional means and variances and the method of moments (MM) estimation for the rest of the density parameters, including the correlation coefficients. The procedure involves consistent estimates even under density misspecification and solves the so-called ‘curse of dimensionality’ of multivariate modelling. Furthermore, the use of a MGC distribution represents a flexible and general approximation to the true distribution of portfolio returns and accounts for all its empirical regularities. An application of such procedure is performed for a portfolio composed of three European indices as an illustration. The MM estimation of the MGC (MGC-MM) is compared with the traditional maximum likelihood of both the MGC and multivariate Student’s t (benchmark) densities. A simulation on Value-at-Risk (VaR) performance for an equally weighted portfolio at 1% and 5% confidence indicates that the MGC-MM method provides reasonable approximations to the true empirical VaR. Therefore, the procedure seems to be a useful tool for risk managers and practitioners.

Locally quasi-nilpotent elementary operators

Let A be a unital dense algebra of linear mappings on a complex vector space X. Let φ = Σⁿ _i=1 M_ai,_bi be a locally quasi-nilpotent elementary operator of length n on A. We show that, if {a1, . . . , an} is locally linearly independent, then the local dimension of V (φ) = span{b_ia_j : 1 ≤ i, j ≤ n} is at most n(n−1) 2 . If ldim V (φ) = n(n−1) 2 , then there exists a representation of φ as φ = Σⁿ _i=1 M_ui,_vi with v_iu_j = 0 for i ≥ j. Moreover, we give a complete characterization of locally quasinilpotent elementary operators of length 3.

Modelos autorregressivos com valores inteiros não negativos : aplicação em séries económicas de Cabo Verde

No estudo de séries temporais, os processos estocásticos usuais assumem que as distribuições marginais são contínuas e, em geral, não são adequados para modelar séries de contagem, pois as suas características não lineares colocam alguns problemas estatísticos, principalmente na estimação dos parâmetros. Assim, investigou-se metodologias apropriadas de análise e modelação de séries com distribuições marginais discretas. Neste contexto, Al-Osh and Alzaid (1987) e McKenzie (1988) introduziram na literatura a classe dos modelos autorregressivos com valores inteiros não negativos, os processos INAR. Estes modelos têm sido frequentemente tratados em artigos científicos ao longo das últimas décadas, pois a sua importância nas aplicações em diversas áreas do conhecimento tem despertado um grande interesse no seu estudo. Neste trabalho, após uma breve revisão sobre séries temporais e os métodos clássicos para a sua análise, apresentamos os modelos autorregressivos de valores inteiros não negativos de primeira ordem INAR (1) e a sua extensão para uma ordem p, as suas propriedades e alguns métodos de estimação dos parâmetros nomeadamente, o método de Yule-Walker, o método de Mínimos Quadrados Condicionais (MQC), o método de Máxima Verosimilhança Condicional (MVC) e o método de Quase Máxima Verosimilhança (QMV). Apresentamos também um critério automático de seleção de ordem para modelos INAR, baseado no Critério de Informação de Akaike Corrigido, AICC, um dos critérios usados para determinar a ordem em modelos autorregressivos, AR. Finalmente, apresenta-se uma aplicação da metodologia dos modelos INAR em dados reais de contagem relativos aos setores dos transportes marítimos e atividades de seguros de Cabo Verde.

Modélisation en bandes C et X de la rétrodiffusion de couverts de neige sèche : évaluation de l’apport de l’approximation quasi-cristalline pour les milieux denses.

La compréhension et la modélisation de l’interaction de l’onde électromagnétique avec la neige sont très importantes pour l’application des technologies radars à des domaines tels que l’hydrologie et la climatologie. En plus de dépendre des propriétés de la neige, le signal radar mesuré dépendra aussi des caractéristiques du capteur et du sol. La compréhension et la quantification des différents processus de diffusion du signal dans un couvert nival s’effectuent à travers les théories de diffusions de l’onde électromagnétique. La neige, dans certaines conditions, peut être considérée comme un milieu dense lorsque les particules de glace qui la composent y occupent une fraction volumique considérable. Dans un tel milieu, les processus de diffusion par les particules ne se font plus de façon indépendante, mais de façon cohérente. L’approximation quasi-cristalline pour les milieux denses est une des théories élaborées afin de prendre en compte ces processus de diffusions cohérents. Son apport a été démontré dans de nombreuses études pour des fréquences > 10 GHz où l’épaisseur optique de la neige est importante et où la diffusion de volume est prédominante. Par contre, les capteurs satellitaires radar présentement disponibles utilisent les bandes L (1-2GHz), C (4-8GHz) et X (8-12GHz), à des fréquences principalement en deçà des 10 GHz. L’objectif de la présente étude est d’évaluer l’apport du modèle de diffusion issu de l’approximation quasi-cristalline pour les milieux denses (QCA/DMRT) dans la modélisation de couverts de neige sèches en bandes C et X. L’approche utilisée consiste à comparer la modélisation de couverts de neige sèches sous QCA/DMRT à la modélisation indépendante sous l’approximation de Rayleigh. La zone d’étude consiste en deux sites localisés sur des milieux agricoles, près de Lévis au Québec. Au total 9 champs sont échantillonnés sur les deux sites afin d’effectuer la modélisation. Dans un premier temps, une analyse comparative des paramètres du transfert radiatif entre les deux modèles de diffusion a été effectuée. Pour des paramètres de cohésion inférieurs à 0,15 à des fractions volumiques entre 0,1 et 0,3, le modèle QCA/DMRT présentait des différences par rapport à Rayleigh. Un coefficient de cohésion optimal a ensuite été déterminé pour la modélisation d’un couvert nival en bandes C et X. L’optimisation de ce paramètre a permis de conclure qu’un paramètre de cohésion de 0,1 était optimal pour notre jeu de données. Cette très faible valeur de paramètre de cohésion entraîne une augmentation des coefficients de diffusion et d’extinction pour QCA/DMRT ainsi que des différences avec les paramètres de Rayleigh. Puis, une analyse de l’influence des caractéristiques du couvert nival sur les différentes contributions du signal est réalisée pour les 2 bandes C et X. En bande C, le modèle de Rayleigh permettait de considérer la neige comme étant transparente au signal à des angles d’incidence inférieurs à 35°. Vu l’augmentation de l’extinction du signal sous QCA/DMRT, le signal en provenance du sol est atténué d’au moins 5% sur l’ensemble des angles d’incidence, à de faibles fractions volumiques et fortes tailles de grains de neige, nous empêchant ainsi de considérer la transparence de la neige au signal micro-onde sous QCA/DMRT en bande C. En bande X, l’augmentation significative des coefficients de diffusion par rapport à la bande C, ne nous permet plus d’ignorer l’extinction du signal. La part occupée par la rétrodiffusion de volume peut dans certaines conditions, devenir la part prépondérante dans la rétrodiffusion totale. Pour terminer, les résultats de la modélisation de couverts de neige sous QCA/DMRT sont validés à l’aide de données RADARSAT-2 et TerraSAR-X. Les deux modèles présentaient des rétrodiffusions totales semblables qui concordaient bien avec les données RADARSAT-2 et TerraSAR-X. Pour RADARSAT-2, le RMSE du modèle QCA/DMRT est de 2,52 dB en HH et 2,92 dB en VV et pour Rayleigh il est de 2,64 dB en HH et 3,01 dB en VV. Pour ce qui est de TerraSAR-X, le RMSE du modèle QCA/DMRT allait de 1,88 dB en HH à 2,32 dB en VV et de 2,20 dB en HH à 2,71 dB en VV pour Rayleigh. Les valeurs de rétrodiffusion totales des deux modèles sont assez similaires. Par contre, les principales différences entre les deux modèles sont bien évidentes dans la répartition des différentes contributions de cette rétrodiffusion totale.

Likelihood-Based Inference for Partially Observed Multi-Type Markov Branching Processes

Thesis (Ph.D.)--University of Washington, 2016-08