This thesis examines some statistical procedures in the frequency domain to analyze long-memory series. We define a long-memory series and review part of the literature. Then we proceed by analyzing different estimation procedures for H, the parameter that characterizes the existence of long-memory. Parametric estimates have as a main drawback that they can lead to inconsistent estimates of H if the parametric model is misspecified. Therefore we focus on semiparametric estimates in the frequency domain. In our case, semiparametric means that we only need to assume a parametric model for the spectral density in a neighbourhood of zero frequency. We focus mainly on a multivariate framework. First we analyze estimates based on the average periodogram. We prove the consistency of the average cross-periodogram for the cumulative cross-spectrum. We also establish the asymptotic distribution in the scalar case. Then we focus on an implicit estimate based on a discrete approximation of the Gaussian likelihood in a neighbourhood of zero frequency. We prove the consistency and asymptotic normality of this estimate. Based on this estimate we establish a Lagrange multiplier test for weak dependence. We finish with an application of these methods to financial data.

Strategic interaction has traditionally been modelled in economics with game theoretic equilibrium models. In these models, strategies constitute best responses to beliefs that are consistent with other players' strategies. While this consistency is realistic in settings familiar to the players, it is less appropriate in situations that are encountered for the first time. This shortcoming has led to the conception of models of bounded rationality, in particular the level-k model of levels of reasoning. While experimental studies usually employ only action data to test the level-k model, in this thesis, a team setup with electronic communication between participants allows for a qualitatively richer insight in actual reasoning processes. Two different games are played to investigate different notions of strategic thinking. The first study uses a dominance-solvable 'beauty contest' game in which 6-8 teams compete for a prize. This game lends itself naturally to the observation of levels of reasoning. In addition, the communication allows to analyse the anchoring level-0 belief and the population belief of individual players. The second study uses a zero-sum 'hide and seek' game that two teams play against each other. Both the influence of non-neutral framing on the level-0 belief and the task-dependence of the level of reasoning can be brought to light in this study. The third and final chapter considers an application of the equilibrium concept in the theory of implicit incentives, a situation of complex strategic interaction. The method and results of the study are viewed against the background of the limitations of equilibrium models to reflect a situation of inherent one-shot nature.

This thesis presents extensions to formal theories of rationality in order to analyse intertemporal decisions. It offers multiple-self models of the decision-maker's personal identity over time. These models complement decision and game theory and are used to develop the new accounts of time discounting, backward induction, and preference change that are presented in this thesis. The first part of the thesis develops multiple-self models of personal identity over time. These models depict a rational decision-maker as a series of different but interconnected temporal selves. The models allow one to relax the assumption that a rational decision-maker is a diachronically stable entity. Moreover, they structurally cohere with key problems and distinctions in theories of personal identity over time. In the second part of the thesis, three problems of time in decision and game theory are analysed. Firstly, the problem of time discounting is considered. General foundations of time discounting are given in a measurement-theoretic framework. In the multiple-self interpretation of a decision-maker, the discounting factor represents the degree of connectedness between temporal selves in a person. Secondly, the reasoning method of backward induction in interactions over time is considered. Sufficient conditions for backward induction are given by formulating a belief revision policy on the basis of intrapersonal connectedness of players. Thirdly, preference change is considered. A new characterisation of diachronic inconsistency in terms of conflicts in intrapersonal connectedness is given. The multiple-self models presented here allow one to represent the internal temporal structure of decision-makers. They capture problems of the interplay between rationality, identity, and time, thereby elucidating new accounts of time discounting, backward induction, and preference change. More generally, this thesis offers a new approach to modelling the intertemporal aggregation of value, which possesses broader relevance for decision theory, the foundations of economics, social epistemology as well as environmental ethics.

This thesis proposes and justifies parameter estimates in two semiparametric models for economic time series. In both models the parametric component consists of a linear regression model. The nonparametric aspect consists of relevant features of the distribution function of the disturbances. In the first model the disturbances follow a possibly non-linear autoregressive model, with autoregression function of unknown form. In the second model the disturbances are both linearly serially correlated and heteroscedastic, the serial correlation and heteroscedasticity being of unknown form. For both models estimates of the regression coefficients of generalized least squares type are proposed, and shown to have the same limiting distribution as estimates based on correct parameterization of the relevant features of the disturbances. Monte-Carlo simulation evidence of the finite sample performance of both estimates is reported.

This thesis deals with certain problems in nonparametric estimation and testing. In the first part of the thesis, we propose a method to improve nonparametric regression estimates of regression functions with a similar shape. This is achieved by first estimating the unknown parameters in the parametric relationship between the regression functions, and subsequently using the estimated transformation to pool the two data sets. The second part is concerned with nonparametric tests for serial independence. We extend an idea by Robinson (1991a) to use the Kullback-Leibler information criterion to measure the distance between the joint and marginal densities of consecutive observations in a stationary time series, and we also propose an entirely new test in which the joint and marginal characteristic functions of afore-mentioned observations are used.

No description available.

In recent years, general state space models have been proven to be extremely useful in modelling wide range of economic and financial time series. Subsequently, particle filters, a computational simulation based method along with its related techniques had burst into our spectrum and fill our expectation of estimating general state space models. However, particle methods can be computationally intensive, as well as possibly requiring stringent restrictions on the parameters space to achieve timely convergence. In this thesis, I propose several improvements to particle methods on different aspects. A list of the improvements are: general computational time reduction in particle filters, modified particle smoothing algorithm, more accurate parameter and state variable estimation through the utilizations of Modified Entropy particle filter, and apply novel general state space model estimation method to real economic and financial time series.

No description available.

No description available.

No description available.