Abstract: Most of the empirical works in regional convergence are based on either crosssectional or “a-spatial” panel data models. In this paper, we propose the use of panel data econometrics models that incorporate an explicit consideration of spatial dependence effects (Anselin, 1988; Elhorst, 2001; 2003). This allows us to extend the traditional convergence models to include a rigorous treatment of regional spillovers and to obtain more reliable estimates of the parameters. We consider two models respectively based on the introduction of a spatial lag among the explanatory variables (the “spatial lag model”) and imposing a spatial autoregressive structure to the stochastic component (the “spatial error model”). We apply such a modelling framework to the long-run convergence of per-capita GDP of 125 EU-NUTS2 regions observed yearly in the period 1977-2002. A comparison of the results obtained using the two spatial panel data specifications with the main evidence available in the literature is also provided.

Abstract: The purpose of this article is to introduce a general moment-based approach to derive formal goodness of fit tests of a parametric family. We show that, in general, an approximate normal test or a chi-squared test can be derived by exploring the moment structure of a parametric family, when moments up to certain order exist. The idea is simple and the resulting tests are easy to implement. To illustrate the use of this approach, we derive moment-based goodness of fit tests for some common discrete and continuous parametric families. We also compare the proposed tests with the well known Pearson-Fisher chi-square test and some distance tests in a simulation study.

Abstract: The paper addresses the problem of robustness of regression trees with respect to outlying values in the dependent variable. New robust tree-based procedures are described, which are obtained by introducing in the tree building phase some objective functions already used in the linear robust regression approach, namely Huber’s and Tukey’s bisquare functions. The performance of the new procedures is evaluated through a Monte Carlo experiment.

Abstract: This paper considers the problem of estimating the population mean using the ratio and product methods when some observations in the sample data are missing at random and the population mean of the auxiliary characteristic is not known. Besides an unbiased estimator arising from the total discard of incomplete pairs of observations, four generally biased estimators are presented. The first two estimators arise form the partial utilization of data while the remaining two are based on full utilization. A comparative study of the efficiency properties of estimators is reported and the choice of estimators is discussed.

Abstract: We check the finite sample performance of the maximum likelihood estimators of the parameters of a mixture distribution recently introduced for modelling ranks/preference data. The estimates are derived by the E-M algorithm and the performance is evaluated both from an univariate and bivariate points of view. While the results are generally acceptable as far as it concerns the bias, the Monte Carlo experiment shows a different behaviour of the estimators efficiency for the two parameters of the mixture, mainly depending upon their location in the admissible parametric space. Some operative suggestions conclude the paer.

Abstract: The paper aims to analyse the internal effectiveness of an university educational process by means of quantile regression. In particular, the goal is to evaluate how the students features affect the outcome of the University careers taking into account that this effect can be different for students with good or bad performances.

Abstract: The basic problem considered in this paper may be stated as follows: find an N-point exact design measure which maximizes the determinant of the information matrix of a given response function. The combinatorial algorithm introduced in the paper reaches the global D-optimal design quite rapidly and a comparison against the variance exchange algorithm is indicated.

Abstract: We propose a new test for the simmetry hypothesis which is a combination of the sign statistic and the Wr Maesono statistic. The latter generalizes the Wilcoxon statistic and coincides with it for r = 2. The proposed statistic belongs to the class of non-degenerate U-statistics and hence it has asymptotically normal distribution. We calculate its Pitman efficacy and compare it with the t-test. For instance, in the normal case, for r=4, the new test, with respect t t-test, has a higher efficiency (0,9794) than the Wilcoxon test (0,9549). In the logistic case, G4 has a higher efficincy (1,0939) than the t-test.

Abstract: In this note we discuss the properties of Augmented-Dickey-Fuller [ADF] unit root tests for autoregressive processes with a unit or near-unit root in the presence of multiple level shifts of large size. Due to the presence of level shifts, the ADF tests experience severe power losses. We consider new modified ADF unit root tests which require no knowledge of either the location or the number of level shifts. The tests are based on a two-step procedure where possible level shifts are initially detected using the level shift indicator estimators suggested by Chen and Tiao (1990, Journal of business and Economics Statistics) and Chen and Liu (1993, Journal of the American Statistical Association), and later removed by a novel procedure which is denoted as “de-jumping”. Using a Monte Carlo experiment we show that the new tests, although partially oversized in samples of moderate size, have much higher power than standard ADF tests.

Abstract: In the paper we give interesting properties of Gini's mean difference. We thoroughly consider the appropriate literature taking account of book publications and articles. It constitutes an important complement to the extensive bibliography of papers based on Gini's ideas, presented by G.M. Giorgi in 1990. We show an application of the mean difference to inflated distributions which are of weight and interest in statistical problems.

Abstract: Center sampling is useful in finite population surveys when exhaustive lists of all units are not available and the target population is naturally clustered into a number of overlapping sites spread over an area of interest such as, for instance, the immigrant population illegally resident in a country. Center sampling has been successfully employed in official European surveys; nevertheless few systematic theoretical results have been given yet to support empirical findings. In this paper a general theory for Center sampling is formalized and an unbiased estimator for the mean of a quantitative or dichotomous characteristic is proposed together with its exact variance. A suitable estimator for the variance, unbiased under simple random sampling, is also derived and the optimum allocation of the sample size among centers subject to linear cost constraints is discussed. Other sampling designs, useful under operational aspects, are also considered.

Abstract: The estimation of the size of an elusive population is a problem frequently addressed in many fiels of applications In the paper a sampling strategy for the estimation of the size is proposed, the properties of the estimator are evaluated in a design-based approach The estimator is unbiased, admissible and consistent and the design is measurable The expressions of the variance of the population size estimator and of its unbiased sample estimator are also proposed The strategy is applied to a simulated population

Abstract: In this paper we extend the concept of Value-at-risk (VaR) to bivariate return distributions in order to obtain measures of the market risk of an asset taking into account additional features linked to downside risk exposure. We first present a general definition of risk as the probability of an adverse event over a random distribution and we then introduce a measure of market risk (b-VaR) that admits the traditional b of an asset in portfolio management as a special case when asset returns are normally distributed. Empirical evidences are provided by using Italian stock market data.

Abstract: The paper presents a non-parametric test of significance regarding the equality of the distribution functions of two independent n-dimensional random variables, on the basis of samples, one of which is one-element. Kernel estimators and order statistics have been used to solve this problem. A fully elaborated procedure is provided for numerical computations.

Abstract: The Authors consider the general problem of similarity and dissimilarity measures in Symbolic Data Analysis. First of all they examine the classical definitions of elementary event, assertion object, hierarchical dependences and logical dependences. Then they consider some well-known measures of similarity and dissimilarity between two objects (Sokal-Michener, Roger-Tanimoto, Sokal-Sneath, Dice-Czekanowski-Sorenson, Russel-Rao). For resemblance measures based on aggregation functions, the authors consider the proposal of Gowda-Diday, De Baets et al., Malerba et al., Vladutu et al., and Ichino-Iyaghuchi. A paragraph is dedicated to the general algebraic structure; particularly to intervals and vector lattices in Banach space.

Abstract: Motivated by a Bayesian inference problem, a modification of the Bessel function distribution is introduced. Various particular cases, expressions for its moments and estimation procedures are derived. An application is illustrated to failure data.

Abstract: The aim of the paper is to give an estimate of the population size when neither a complete list nor some incomplete lists are available and the populution is partially undetectable. In such a framework the center sampling technique proves an essential tool. A first estimate relies on the likelihood approach assuming the complete detectability of the units. Then, this unrealistic hypothesis is removed introducing a prior distribution on detectability, so that a second estimate is given resorting to integrate likelihood. Finally, the previous estimates are compared through a simulation study.

Abstract: The aim of this paper is to propose a general nonparametric model to estimate latent variables with scores non indeterminate; in this paper, following other approaches (PLS, RCD, RCDR), a latent variable (LV) is conceived as a linear combination of predictors (causes) which best predicts a set of dependent variables (indicators), maximising, in this manner, all available information about a LV in the soecified model. The model is also extented to categorical variables (nominal, ordinal) by means of optimal scaling methodology and applied to the estimate of a bidimensional LV as a proxy of human capital for US families in 1983.

Abstract: We study Italian households’ portfolio choices, with a special focus on equity investments, by analysing jointly time series and cross-sectional portfolio data. We investigate the temporal evolution of the actual composition of Italian households’ investments in order to explain their portfolio choices and to detect possible determinants of the observed disequilibria phenomena. Moreover, we model the stock market participation choice by using probit regression techniques and we test for parameter stability over time. Instability of participation parameters and a peculiar evolution of Italian households’ portfolios pointed out by our concurrent analysis of cross-sectional and time series data seem to confirm the distance of Italian households’ financial decisions from the rational choice predicted by the Markowitz model. In particular, we find that the housing market bubbles interact strongly with the stock market and financial institutions seem to be unable to advise investors suggesting optimal portfolio choices. The deep reason behind these facts may be the bounded education of investors, in particular the low financial literacy of Italian households.

Abstract: In this paper we study the rate of convergence to the normal approximation of the least squares estimators in a regression model with long range dependent errors. The method of investigation used is based on the asymptotic analysis of orthogonal expansions of non linear functionals of stationary Gaussian processes and on Kolmogorov's distance.

Abstract: In this paper we introduce shrinkage estimators for the estimation of the Unemployment rate in small domains of the Italian Labour Force Survey. The proposed estimators are based on Hierarchical Linear Mixed Models and on the borrowing strength on both time series and cross section. Auxiliary information from source external to the Labour Force Survey is not considered. A Hierarchical Bayesian approach is adopted, in which models are solved by means of MCMC sampling algorithms. This allows to measure variability associated to estimators accounting, in a simple way, for all e uncertainty sources. Results highlight how, simple hierarchical models allows for remarkable gain in efficiency with respect to published estimates, and that models with a time series component perform better than those based exclusively on data from the same repetition of the survey.

Abstract: We consider the problem of comparing, by simulations, the robustness as regards heteroschedasticity GARCH (1,1) of some of the most important procedures proposed in literature for the cointegration analysis. In particular, we consider the Johansen test and some "two steps" "procedures", i.e. Dickey-Fuller test, Sargan-Bhargava test and an External Bootstrap test. The Bootstrap test performs very well, particularly for the lowest sample size.

Abstract: The Bernstein-type approximation using the beta-binomial distribution is proposed and studied. Both univariate and multivariate Bernstein-type approximations are studied. The uniform convergence and the degree of approximation are studied. The Bernsteintype estimators of smooth functions of population means are also proposed and studied.

Abstract: Extensive and persistent geographic variability of the unemployment rate within the same region has been attributed to various causes. Some theories identify the “thickness” of markets as the source of positive externalities affecting labour market by improving the ability to match the skills requested by firms with those offered by workers. A recent paper by Gan and Zhang (2006) empirically confirms this hypothesis for the US labour markets. Agglomeration can be defined as aggregation of people, basically measured by city size, or as aggregation of firms, measured by cluster size (employment or number of plants). However, the population location and the industrial location are by far more similar in United States than in Europe and in Italy. Our paper aims to evaluate the effects of agglomeration on the local unemployment rate. The new methodological contribution of the study is the identification of both urban and industrial cluster agglomeration effects, using a wide set of control variables. Adjusting the system for the effects of sectorial and size shocks, as well as those relating to geographic structure and policy interventions, the results of our analysis differ from that for the United States. The study stresses the presence of negative and significant urbanisation externalities. We obtain, instead, positive effects concerning the geographic agglomeration of firms, and their thickness, in a specific area. Furthermore, positive and significant effects can be found in local systems with features of a district. Finally, the model distinguishes the negative effects of urban agglomerations (in terms of population density) from positive firm’s agglomerations (in terms of density of local units).

Abstract: A method is presented for the estimation of the power of permutation tests when F is unknown It is based on the natural plug-in of the empirical distribution in the structure of the statistical test, giving the bootstrap power The consistency of the permutational bootstrap test is shown Moreover, to determine the sample size m of permutation tests starting from a pilot sample of size n, the "Mapped Bootstrap" is introduced The Mapped Bootstrap works for a fixed m and is consistent as the pilot sample size n tends to infinity

Abstract: We propose a Poisson-based model that uses both infallible data and fallible data subject to misclassification in the form of false negatives that yield visibility bias. We than derive maximum likelihood estimators for the Poisson rate parameter of interest and the misclassification parameter under two different sampling scenarios. We also derive expressions for the information matrices and the asymptotic variances of the maximum likelihood estimators for the rate parameter and the maximum likelihood estimators for the false-negative parameter. Finally, we also study our new models via a simulation experiment and then apply our new estimation procedures to a real data set.

Abstract: In the frame of a statistical survey, the identification of non respondent units thatshould be object with priority of a reminder action (Intensive Follow Up - IFU), with the aimto produce enough good estimates, represents a relevant, but quite not deeply analysed methodological aspect. In this context, we propose and compare some score functions -that can be all reconnected to a generalised function – evaluating how much is dangerousthe exclusion from calculations of each unit. Moreover, we evaluate and compare somecriteria aimed at identifying IFU units by means of suitable statistical tests or thresholdsderived by parametric or non parametric methods. A comparative empirical applicationon a panel of Italian retail trade businesses has been carried out and commented.

Abstract: A two unit parallel system where in the failure rate of a unit is a constant and the repair time distribution is a two stage erlangian distribution is considered. Measures of system performance such as reliability, MTBF, system availability and steady state availability are derived. Also, a consistent asymptotically normal (CAN) estimator, a 100(1-α) % asymptotic confidence interval for the steady state availability of the system and the maximum likelihood estimator (MLE) of the system reliability are obtained.

Abstract: The "aggregation points" sampling design applies, for instance, in survey of irregular immigrants ie of populations composed by a finite but unknown number of units which do not consent labelling and that can be reached only through a set of known but overlapping frames called "aggregation points" Dealing with the "aggregation points" sampling design, the problem of estimating the mean of a quantitative character is concerned; an estimate of the estimator's variance is also proposed Some results from a simulation study are presented Simulations indicate that estimators proposed perform better in case of not too large number of aggregation points but extensively overlapping