Tests of serial independence based on Kendall's process

The authors propose new rank statistics for testing the white noise hypothesis in a time series. These statistics are Cramér‐von Mises and Kolmogorov‐Smirnov functionals of an empirical distribution function whose mean is related to a serial version of Kendall's tau through a linear transform. The authors determine the asymptotic behaviour of the underlying serial process and the large‐sample distribution of the proposed statistics under the null hypothesis of white noise. They also present simulation results showing the power of their tests.     

Inference in model-based cluster analysis

A new approach to cluster analysis has been introduced based on parsimonious geometric modelling of the within-group covariance matrices in a mixture of multivariate normal distributions, using hierarchical agglomeration and iterative relocation. It works well and is widely used via the MCLUST software available in S-PLUS and StatLib. However, it has several limitations: there is no assessment of the uncertainty about the classification, the partition can be suboptimal, parameter estimates are biased, the shape matrix has to be specified by the user, prior group probabilities are assumed to be equal, the method for choosing the number of groups is based on a crude approximation, and no formal way of choosing between the various possible models is included. Here, we propose a new approach which overcomes all these difficulties. It consists of exact Bayesian inference via Gibbs sampling, and the calculation of Bayes factors (for choosing the model and the number of groups) from the output using the Laplace–Metropolis estimator. It works well in several real and simulated examples.     

On robust forecasting in dynamic vector time series models

In this article, robust estimation and prediction in multivariate autoregressive models with exogenous variables (VARX) are considered. The conditional least squares (CLS) estimators are known to be non-robust when outliers occur. To obtain robust estimators, the method introduced in Duchesne [2005. Robust and powerful serial correlation tests with new robust estimates in ARX models. J. Time Ser. Anal. 26, 49–81] and Bou Hamad and Duchesne [2005. On robust diagnostics at individual lags using RA-ARX estimators. In: Duchesne, P., Rémillard, B. (Eds.), Statistical Modeling and Analysis for Complex Data Problems. Springer, New York] is generalized for VARX models. The asymptotic distribution of the new estimators is studied and from this is obtained in particular the asymptotic covariance matrix of the robust estimators. Classical conditional prediction intervals normally rely on estimators such as the usual non-robust CLS estimators. In the presence of outliers, such as additive outliers, these classical predictions can be severely biased. More generally, the occurrence of outliers may invalidate the usual conditional prediction intervals. Consequently, the new robust methodology is used to develop robust conditional prediction intervals which take into account parameter estimation uncertainty. In a simulation study, we investigate the finite sample properties of the robust prediction intervals under several scenarios for the occurrence of the outliers, and the new intervals are compared to non-robust intervals based on classical CLS estimators.     

Statistics on statistics: Measuring research productivity by journal publications between 1985 and 1995

Sixteen international journals publishing statistical theory were surveyed over the 11-year period beginning in 1985. Paper, author and adjusted page counts yield cursory measures of productivity for institutions and countries that contributed to fundamental statistical research during that period. These data clearly identify Canada as one of the main contributors to the development of the discipline in the past decade. They also provide valuable information on the evolution of publication habits, in terms of the volume of research, the length of papers, coauthorship practices, etc.     

Firm size and board diversity

This study seeks to understand the relation between firm size and supervisory board composition. Specifically, we ask if and how firm size influences occupational and international background diversity in supervisory boards. Relying on resource dependence theory and theories of organizational behavior, we hypothesize that board diversity with respect to directors’ occupational background will increase with firm size, while the relation between firm size and board diversity with respect to directors’ international background will be concave. Using archival data for supervisory board members of 151 German firms listed in the German stock exchange indices DAX, MDAX, SDAX and TecDAX for the business year 2005, we find empirical support for our hypotheses: Both, occupational and international background diversity increase with increasing firm size, but international background diversity does so at decreasing rates.     

A structured variational learning approach for switching latent factor models

A data-driven approach for modeling volatility dynamics and co-movements in financial markets is introduced. Special emphasis is given to multivariate conditionally heteroscedastic factor models in which the volatilities of the latent factors depend on their past values, and the parameters are driven by regime switching in a latent state variable. We propose an innovative indirect estimation method based on the generalized EM algorithm principle combined with a structured variational approach that can handle models with large cross-sectional dimensions. Extensive Monte Carlo simulations and preliminary experiments with financial data show promising results.     

Historical regional patterns of infant mortality in Austria

First an overview is given of the secular decline of infant mortality in Austria between 1820 and 1950 is given. Second the study analyses the historical regional differences of infant mortality in Austria comparing data for all 99 political districts (Politische Bezirke) for the period 1900 to 1950. The most important results are: At the turn of the 19th century infant mortality rates were generally lower in Alpine than in non-alpine regions and lower in cities than in their surrounding areas. This geographical pattern which remained constant until the 1950s was solely determined by differences in post-neonatal mortality rates. Plausible explanations for these regional differences are discussed.En premier lieu cette étude fournit une vue d'ensemble du déclin séculaire de la mortalité infantile en Autriche en 1820 et 1950. Elle analyse ensuite les différences régionales de mortalité infantile en comparant les données des 99 districts politiques (Politische Bezirke) autrichiens, de 1900 à 1950. Les principaux résultats sont les suivants: à la fin du XIX^e siècle les taux de mortalité infantile étaient en général plus faibles dans les régions alpines que dans les autres et plus faibles dans les villes que dans les aires adjacentes. Ce modèle géographique, qui est resté le même jusqu'en 1950, était uniquement déterminé par les différences entre les taux de mortalité post-néonatale. Des explications plausibles de ces différences régionales sont discutées ici.