Abstract: This paper reports on an extensive Monte Carlo study of seven residual-based tests of the hypothesis of no cointegration. Critical values and the power of the tests under the alternative of fractional cointegration are simulated and compared. It turns out that the Phillips-Perron t-test when applied to regression residuals is more powerful than Geweke-Porter-Hudak tests and the Augmented Dickey-Fuller test. Only the Modified Rescaled Range test is more powerful than the Phillips-Perron test in a few situations. Moreover in large samples, the power of the Phillips-Perron test increases if a time trend is included in the cointegrating regression.

Abstract: This paper examines the major changes in the Canadian banking system since the Second World War, with special attention paid to the differences between Canadian and US developments over this period. An important difference between the countries is the nationwide branch banking arrangements in Canada. Two other differences are a result of regulatory dimensions of the Canadian scene: periodic reassessment and updating of banking legislation as a legislative requirement; and the absence of ceilings on interest rates on deposits or, since 1967, on loans.

Abstract: This paper details changes to the BEAM models, RRYIELD and RRECON, which are two of a series of bioeconomic models developed by the Bioeconomic Agroforestry Modelling (BEAM) project. The RRYIELD model deals with the biophysical components of a rubber-based agroforestry system. It focuses on the changes in output (latex, wood and intercropped annual and perennial crops) in response to a number of bioclimatic, topographic and silvicultural variables. Outputs are measured annually over the life of the plantation. The RRYEILD model is linked to an economic model RRECON used to determine the economic returns from a rubber plantation. The two models are used in combination as an extension and research tool to supply farmers or researchers with information on the viability of alternative rubber intercropping systems. The original model was developed for large estates and government plantations. The changes outlined in this report aimed to make the model more applicable to smallholders in Imperata infested areas of Southeast Asia, particularly Indonesia. These modifications were made progressively and were documented in a series of five papers. They are now presented here together in a single paper.

Abstract: Sind in einer statistischen Analyse mehr als eine Variationsursache zu berucksichtigen_ so zerfallt die Gesamtvariation in mehrere Komponenten. Fur diese sind in der Regel auch Beziehungen oder Ordnungen bekannt. Aber keine der bekannten Schatztheorien ist in der Lage, diese Ordnungen auch auf die Schatzfunktionen fur die theoretischen Grosen zu ubertragen. Gangigerweise kann etwa eine einzelne Komponente durchaus wesentlich groser geschatzt werden als die Summe der positiven Komponenten. Zumindest im technischen Bereich wird dieses Verhalten der bekannten Schatzfunktionen als besonders storend empfunden, da es erhebliche Anforderungen an die Interpretation stellt. Aus diesem Grunde werden hier fur die Varianz bzw. Prazision zweier unterschiedlicher Mesverfahren, welche jeweils das selbe Produkt aus einem Los genau einmal beurteilen (gedanklich: zerstorende Prufung) positive Schatzfunktionen betrachtet die eine spezizierte naturliche Ordnung einhalten. Die Bedeutung dieses Modells liegt fur die Praxis in seiner Anwendung auf typische Kontrollsituationen auch auserhalb der Technik z. B. Gewinnschatzungen seitens eines Betriebes versus Schatzungen seitens des Finanzamtes.

Abstract: We investigate and compare stable parallel algorithms for solving diagonally dominant and general narrow-banded linear systems of equations. Narrow-banded means that the bandwidth is very small compared with the matrix order and is typically between 1 and 100. The solvers compared are the banded system solvers of ScaLAPACK [12] and those investigated by Arbenz and Hegland [4, 8]. For the diagonally dominant case, the algorithms are analogs of the well-known tridiagonal cyclic reduction algorithm, while the inspiration for the general case is the lesser-known bidiagonal cyclic reduction, which allows a clean parallel implementation of partial pivoting. These divide-and-conquer type algorithms complement fine-grained algorithms which perform well only for wide-banded matrices, with each family of algorithms having a range of problem sizes for which it is superior. We present theoretical analyses as well as numerical experiments conducted on the Intel Paragon.

Abstract: The determination of toxicokinetic parameters is an essential component in the risk assessment of potential harmful chemicals. It’s a first step to analyse the processes which are involved in the development of DNA adducts and might therefore lead to the development of cancer. The complete research depends on investigations with animals in vivo and in vitro, so that a critical step is the extrapolation from experimental animals to the human organism. Besides the investigation of the interspecific differences, the intraspecific and the interoccasion variability have to be analysed to avoid serious errors in the determination of the human risk. The aim of extrapolation from one species to an other requires a characterisation of the interesting processes which is valid for the whole species, i.e. population mean parameters instead of sets of parameters for different individuals, occasions and concentrations of the interesting chemical. The theory of hierarchical models, basically the work of Racine-Poon et al. (1985, 1986, 1990), provides a procedure, which incorporates both, modelling of the variability structure and reduction of complexity in terms of estimates of population mean parameter vectors. This paper presents part of a strategy to determinate the processes of uptake, elimination, and metabolism of the gas ethylene, which is a natural body constituent and an important industrial chemical.

Abstract: We show that OLS and GLS are asymptotically equivalent in the linear regression model with AR (p) disturbances and a wide range of trending regressors_ and that OLS based statistical inference is still meaningful after proper adjustment of the test statistics.

Abstract: As high dimensional data occur as a rule rather than an exception in critical care today, it is of utmost importance to improve acquisition, storage, modelling, and analysis of medical data, which appears feasable only with the help of bedside computers. The use of clinical information systems offers new perspectives of data recording and also causes a new challenge for statistical methodology. A graphical approach for analysing patterns in statistical time series from online monitoring systems in intensive care is proposed here as an example of a simple univariate method, which contains the possibility of a multivariate extension and which can be combined with procedures for dimension reduction.

Abstract: In this paper the task of identifying outliers in exponential samples is treated conceptionally in the sense of Davies and Gather (1989, 1993) by means of a so called outlier region. In case of an exponential distribution, an empirical approximation of such a region also called an outlier identifier is mainly dependent on some estimator of the unknown scale parameter. The worst case behaviour of several reasonable outlier identifiers is investigated thoroughly and it is shown that only robust estimators of scale should be used to construct reliable identifiers. These findings lead to the recommendation of an outlier identifier that is based on a standardized version of the sample median.

Abstract: In the presence of variability control factors in Taguchi experiments, then the original g-method (Logothetis, 1990) is liable to lead to wrong transformations. We propose a generalization of the, β-method which should lead to correct transformations, even if there is a variability control factor which also influences the mean.

Abstract: We use the Pitman-closeness criterion to evaluate the performance of multivariate forecasting methods and we also calculate optimal matrices of weights for the linear combination of multivariate forecasts. These weights are identical with the optimal weights under the matrix-MSE criterion.

Abstract: We consider Pitman-closeness to evaluate the performance of univariate and multivariate forecasting methods. Optimal weights for the combination of forecasts are calculated with respect to this criterion. These weights depend on the assumption of the distribution of the individual forecasts errors. In the normal case they are identical with the optimal weights with respect to the MSE-criterion (univariate case) and with the optimal weights with respect to the MMSE-criterion (multivariate case). Further, we present a simple example to show how the different combination techniques perform. There we can see how much the optimal multivariate combination can outperform different other combinations. In practice, we can find multivariate forecasts e.g., in econometrics. There is often the situation that forecast institutes estimate several economic variables.

Abstract: We consider empirical autocorrelations of residuals from infinite variance autoregressive processes. Unlike the finite-variance case, it emerges that the limiting distribution, after suitable normalization, is not always more concentrated around zero when residuals rather than true innovations are employed.

Abstract: An important aim in forest-ecosystem investigation is to analyse the development of forest damages and to quantify changes in the damage-states over time of individual trees by influential factors. In addition to the ordinal measurement in such longitudinal studies one has to consider spatial correlations of the trees within an ecosystem. We present a practical method to include such dependency structures using logistic regression models. The strategy is to adopt the disposition model for correlated binary data (Bonney (1998)) and extend it to an ordinal-disposition-transitional model (ODT-model). This includes proportional-oddstransitional models (POT-model) as a special case, assuming independence over time and space given a markov model of first order. The ODT-model is used to analyse dynamic changes of damage in forestecosystems. The analysed data was sampled with infrared aerial photos by the Swiss Federal Institute for Forest, Snow and Landscape Research (Forschungsanstalt Wald, Schnee und Landschaft (WSL), Switzerland). A comparison of the independent case (POT-model) with the dependent case (ODT-model) shows that spatial correlations in forest-ecosystem should not be neglected.

Abstract: In the present study we investigate the data provided by the karstwater level monitoring system set up in the Transdanubian Mountains, more precisely in the Bakony, the Keszthelyi Mountains and the Balaton-Highland (Here, like in the sequel, the term karstwater is used for groundwater in karstic areas) The detailed description of the monitoring system itself and the geological and hydrogeological situation in which the system was planned to function and collect data about the water level can be found in Markus et al (1997) as well as the results of our previous study in determining the underlying (called also latent or background) effects driving the karstwater fluctuations

Abstract: In this paper we derive asymptotic x2-tests for general linear hypotheses on variance components using repeated variance components models. In two examples, the two-way nested classification model and the two-way crossed classification model with interaction, we explicitly investigate the properties of the asymptotic tests in small sample sizes.

Abstract: In their paper, Davies and Gather (1993) formalized the task of outlier identification, considering also certain performance criteria for outlier identifiers. One of those Criteria, the maximum asymptotic bias, is carried over here to multivariate outlier identifiers. We show how this term depends on the respective biases of estimators which are used to construct the identifier. It turns out that the use of high breakdown robust estimators is not sufficient to achieve outlier identifiers with bounded maximum asymptotic bias.

Abstract: This paper surveys the empirical literature on the benefits of low inflation emphasizing contributions since 1990. It follows the framework of a section in the Bank's 1990 Annual Report, "The Benefits of Price Stability." This framework looks at the costs of inflation, or the benefits of price stability, in the context of four themes: inflation creates uncertainty about the future; there are costs of having to cope with inflation; inflation affects equity and fairness; and 'living with inflation' is no answer.

Abstract: The financial service industry has been undergoing significant change in recent years. This paper analyzes some key developments affecting the industry and examines some important issues facing the industry and its regulators. Changes discussed include the way services are provided, the instruments used to provide servces, and the nature of the financial service providers. Factors driving these changes include technological developments, the changing role of competition, and demographically led changes in household portfolios.

Abstract: This work presents the results from a case-control study about occupational exposures as risk factors for bladder cancer. Odds ratio a nalysis and logistic regression give results, which show the influence of dif ferent exposures to polycyclic aromatic hydrocarbons and an exposure to paint s o bladder cancer etiology. Further an outlook on the upcoming studies about gen etic predispositions as additional risk factors is given.

Abstract: The behaviour of group sequential tests in the two-sample problem is investigated if one replaces the classical non-robust estimators in the t-test statistic by modern robust estimators of location and scale. Hampel's 3-part redescending M-estimator 25A used in the Princeton study and the robust scale estimators length of the shortest half proposed by Rousseeuw & Leroy and Q proposed by Rousseeuw & Croux are considered. Of special interest are level, power and the average sample size number of the tests. It is investigated, whether commerical software can be used to apply these tests.

Abstract: Noisy observations form the basis for almost every scientific research and especially in environmental monitoring. The Noise is often an effect of imprecise instruments which cause measurement errors. If the noise variance is known it is possible to filter out the contaminating noise from the observations and then to predict the latent signal process. Solutions for this problem exist for time series application and will be briefly reviewed. In the geostatistical literature, i.e. for the analysis of spatial data, similar methods have been foreshadowed in the literature and will be outlined in this work.

Abstract: We describe a computer intensive method for linear dimension reduction which minimizes the classification error directly. Simulated annealing Bohachevsky et al (1986) is used to solve this problem. The classification error is determined by an exact integration. We avoid distance or scatter measures which are only surrogates to circumvent the classification error. Simulations in two dimensions and analytical approximations demonstrate the superiority of optimal classification opposite to the classical procedures. We compare our procedure to the well-known canonical discriminant analysis (homoscedastic case) as described in Mc Lachlan (1992) and to a method by Young et al (1986) for the heteroscedastic case. Special emphasis is put on the case when the distance based methods collapse. The computer intensive algorithm always achieves minimal classification error.

Abstract: Sliced Inverse Regression (SIR) is a promising technique for the purpose of dimension reduction. Several properties of this relatively new method have been examined already, but little attention has been paid to robustness aspects. We show that SIR is very sensitive towards outliers in the data. Therefore a generalized estimation procedure which allows for robustness properties, especially for a high breakdown point, is proposed.

Abstract: We simulate forecast errors with different variance-covariance structures based on macroeconomic data. The simulations are used to compare the performance of different forecast combining techniques.

Abstract: If there are various forecasts for the same random variable, it is common practice to combine these forecasts in order to obtain a better forecast. But an important question is how to perform the combination, especially if the system under investigation is subject to structural changes and consequently the best combination method is not the same all of the time. This paper presents a data driven approach, which for each point of time selects a combination technique from a given set of combination techniques. Properties and limitations of this selection procedure are investigated using simulated data from normal distributions.