The thesis contains the four chapters: Testing parameter constancy in unit root autoregressive models against continuous change; Dickey-Fuller type of tests against nonlinear dynamic models; Inference for unit roots in a panel smooth transition autoregressive model where the time dimension is fixed; Testing unit roots in nonlinear dynamic heterogeneous panels. In Chapter  1 we derive tests for parameter constancy when the data generating process is non-stationary against the hypothesis that the parameters of the model change smoothly over time. To obtain the asymptotic distributions of the tests we generalize many theoretical results, as well as new are introduced, in the area of unit roots . The results are derived under the assumption that the error term is a strong mixing. Small sample properties of the tests are investigated, and in particular, the power performances are satisfactory. In Chapter 2 we introduce several test statistics of testing the null hypotheses of a random walk (with or without drift) against models that accommodate a smooth nonlinear shift in the level, the dynamic structure, and the trend. We derive analytical limiting distributions for all tests. Finite sample properties are examined. The performance of the tests is compared to that of the classical unit root tests by Dickey-Fuller and Phillips and Perron, and is found to be superior in terms of power. In Chapter 3 we derive a unit root test against a Panel Logistic Smooth Transition Autoregressive (PLSTAR). The analysis is concentrated on the case where the time dimension is fixed and the cross section dimension tends to infinity. Under the null hypothesis of a unit root, we show that the LSDV estimator of the autoregressive parameter in the linear component of the model is inconsistent due to the inclusion of fixed effects. The test statistic, adjusted for the inconsistency, has an asymptotic normal distribution whose first two moments are calculated analytically. To complete the analysis, finite sample properties of the test are examined. We highlight scenarios under which the traditional panel unit root tests by Harris and Tzavalis have inferior or reasonable power compared to our test. In Chapter 4 we present a unit root test against a non-linear dynamic heterogeneous panel with each country modelled as an LSTAR model. All parameters are viewed as country specific. We allow for serially correlated residuals over time and heterogeneous variance among countries. The test is derived under three special cases: (i) the number of countries and observations over time are fixed, (ii) observations over time are fixed and the number of countries tend to infinity, and (iii) first letting the number of observations over time tend to infinity and thereafter the number of countries. Small sample properties of the test  show modest size distortions and satisfactory power being superior to the Im, Pesaran and Shin t-type of test. We also show clear improvements in power compared to a univariate unit root test allowing for non-linearities under the alternative hypothesis. / Diss. Stockholm : Handelshögskolan, 2004

Diss. Stockholm : Handelshögsk.

This dissertation consists of 3 essays In the first essay, A Simple Variable Selection Technique for Nonlinear Models, written in cooperation with Timo Teräsvirta and Rolf Tschernig, I propose a variable selection method based on a polynomial expansion of the unknown regression function and an appropriate model selection criterion. The hypothesis of linearity is tested by a Lagrange multiplier test based on this polynomial expansion. If rejected, a kth order general polynomial is used as a base for estimating all submodels by ordinary least squares. The combination of regressors leading to the lowest value of the model selection criterion is selected.  The second essay, Modelling and Forecasting Economic Time Series with Single Hidden-layer Feedforward Autoregressive Artificial Neural Networks, proposes an unified framework for artificial neural network modelling. Linearity is tested and the selection of regressors performed by the methodology developed in essay I. The number of hidden units is detected by a procedure based on a sequence of Lagrange multiplier (LM) tests. Serial correlation of errors and parameter constancy are checked by LM tests as well. A Monte-Carlo study, the two classical series of the lynx and the sunspots, and an application on the monthly S&amp;P 500 index return series are used to demonstrate the performance of the overall procedure. In the third essay, Forecasting with Artificial Neural Network Models (in cooperation with Marcelo Medeiros), the methodology developed in essay II, the most popular methods for artificial neural network estimation, and the linear autoregressive model are compared by forecasting performance on 30 time series from different subject areas. Early stopping, pruning, information criterion pruning, cross-validation pruning, weight decay, and Bayesian regularization are considered. The findings are that 1) the linear models very often outperform the neural network ones and 2) the modelling approach to neural networks developed in this thesis stands up well with in comparison when compared to the other neural network modelling methods considered here. / <p>Diss. Stockholm : Handelshögskolan, 2002. Spikblad saknas</p>

Diss. Stockholm : Handelshögskolan, 2002.