Michel G.M. Mouchart

Catholic University of Louvain | UCLouvain · School of Statistics, Biostatistics and Actuarial Science

This chapter compares minimal conditions under which a semiparametric binary response model is identified in a sampling theory and in a Bayesian framework. The impact of different forms of prior specification is explored by means of some numerical experiments.

In this chapter, we use the semi-parametric Bayesian approach in the proportional hazard model generalizing the results of Kalbfleisch (Journal of the Royal Statistical Society. Series B (Methodological), 40(2), 214–221 (1978)). Basically we consider the parameter of interest as random and assume a nonrestrictive hypothesis that the functional nuis...

Given two \(\sigma \)-algebras \(\mathcal {M}_1\) and \(\mathcal {M}_2\), necessary and sufficient conditions are given for a \(\sigma \)-algebra \(\mathcal {M}_3\) to minimally make \(\mathcal {M}_1\) and \(\mathcal {M}_2\) conditionally independent. Representations in terms of projections among \(\sigma \)-algebras are derived from those conditio...

In this chapter, the concept of invariance, standard in measure theory, is extended to the conditional case and is shown to provide a suitable framework to define invariant Bayesian experiments, even in the case of improper prior distributions. Also, the concept of conditional independence, standard in probability theory, is extended to the case of...

This paper is concerned with simultaneously regrouping regions and sectors when analyzing the relative sectorial specialization of regions and the relative regional concentration of sectors. An automatic two-mode clustering algorithm is proposed with a view toward a concept of overall localization, corresponding to a discrepancy between an actual t...

This article deals with the role of time in causal models in the social sciences. The aim is to underline the importance of time-sensitive causal models, in contrast to time-free models. The relation between time and causality is important, though a complex one, as the debates in the philosophy of science show. In particular, an outstanding issue i...

This paper presents a critical view of causal assessment as typically done in econometrics and proposes a constructive approach for improving statistical models elaborated for causal analysis. Causal inference in econometrics has been considered according to different approaches, some of which are more statistically grounded than others. Without at...

There is no unified theory of causality in the sciences and in philosophy. In this paper, we focus on a particular framework, called structural causal modelling (SCM), as one possible perspective in quantitative social science research. We explain how this methodology provides a fruitful basis for causal analysis in social research, for hypothesisi...

Background: Distinguishing between pharmacological and residual effects, this paper considers the problem of causal assessment in the case of a particular model, namely a Sure Outcome of Random Events (SORE) model developed for the analysis of data from a randomized placebo-controlled double-blind trial of a drug. Method: This model takes into a...

One method for causal analysis in the social sciences is structural modeling. Structural models, as used in this article, model the (causal) mechanism for a social phenomenon by recursively decomposing the multivariate distribution of the variables of interest. Often, however, one does not achieve a complete decomposition in terms of single variabl...

This paper proposes an integrated framework for discussing issues related to regional concentration, sectorial specialization and overall localization by considering these concepts as a row-column association– or non-independence– in a two-way contingency table ‘regions × sectors’. This is the approach of stochastic independence, in which the degre...

Determining the variables to be controlled for is usually a major problem in the social sciences when analyzing possible causal relations. A structural modelling approach, having recourse to directed acyclic graphs, is presented here as a consistent framework for determining a coherent set of guidelines when deciding what variables should be contro...

A specific concept of structural model is used as a background for discussing the structurality of its parameterization. Conditions for a structural model to be also causal are examined. Difficulties and pitfalls arising from the parameterization are analyzed. In particular, pitfalls when considering alternative parameterizations of a same model ar...

A specific concept of structural model is used as a background for discussing the structurality of its parameterization. Conditions for a structural model to be also causal are examined. Difficulties and pitfalls arising from the parameterization are analyzed. In particular, pitfalls when considering alternative parameterizations of a same model ar...

This paper develops new statistical and computational methods for the automatic detection of spatial clusters displaying an over- or under- relative specialization spatial pattern. A probability model is used to provide a basis for a space partition into clusters representing homogenous portions of space as far as the probability of locating a prim...

This paper explores the feasibility of simultaneously facing three sources of complexity in Bayesian testing, namely (i) testing a parametric against a non-parametric alternative (ii) adjusting for a partial observability situation (iii) developing a test under a Bayesian encompassing principle. Testing the normality of latent variables in the poly...

Florens, Richard and Rolin (2003) proposed a specification test of a parametric hypothesis against a nonparametric one, in the framework of a Bayesian encompassing test. Building on that work, this paper elaborates the procedure under a condition of partial observability. The general procedure is illustrated with the case where only the sign is obs...

One way social scientists explain phenomena is by building structural models. These models are explanatory insofar as they manage to perform a recursive decomposition on an initial multivariate probability distribution, which can be interpreted as a mechanism. Explanations in social sciences share important aspects that have been highlighted in the...

This paper provides an overview of structural modeling in its close relation to explanation and causation. It stems from previous works by the authors and stresses the role and importance of the notions of invariance, recursive decomposition, exogeneity and background knowledge. It closes with some considerations about the importance of the structu...

Although neural networks are commonly encountered to solve classification problems, ranking data present specificities which require adapting the model. Based on a latent utility function defined on the characteristics of the objects to be ranked, the approach suggested in this paper leads to a perceptron-based algorithm for a highly non linear mod...

L’inférence causale par contrefactuels dans les études observationnelles — Quelques épistémologiques : Cet article contribue au débat sur les vertus et les vices de contrefactuels comme base pour l’inférence causale. L’objectif est de mettre l’approche contrefactuelle dans une perspective épistémologique. Nous discutons d’un certain nombre de quest...

We study the identification and consistency of Bayesian semiparametric IRT-type models, where the uncertainty on the abilities’ distribution is modeled using a prior distribution on the space of probability measures. We show that for the semiparametric Rasch Poisson counts model, simple restrictions ensure the identification of a general distributi...

This chapter deals with causal explanation in quantitative-oriented social sciences. In the framework of statistical modelling, we first develop a formal structural modelling approach which is meant to shape causal explanation. Recursive decomposition and exogeneity are given a major role for explaining social phenomena. Then, based on the main fea...

A-t-on nécessairement besoin de données longitudinales pour inférer des relations causales ? Il est généralement admis que les causes précèdent leurs effets dans le temps. Cela justifie usuellement la préférence pour les études longitudinales par rapport aux études transversales, parce que les premières permettent la modèlisation du processus dynam...

This paper proposes a new methodology for evaluating the market imperfection and the bargaining power of each agent acting on a given market. The new methodology is an extension of DEA to the two frontiers case with respect to a privileged direction. A particular attention is paid to the treatment of bidirectional free disposability. This model is...

Whilst it might seem uncontroversial that the health sciences search for causes – that is, for causes of disease and for effective treatments – the causal perspective is less obvious in social science research, perhaps because it is apparently harder to glean general laws in the social sciences than in other sciences, due the probabilistic characte...

A statistical model is generally defined through a probability on some variables conditionally on other variables and refers to some parameters of interest. Therefore, it seems natural to ask under which conditions such a model does not lose information with respect to a model describing more variables and implying more parameters. Admissibility co...

This paper examines the potential outcome model developed by Rubin and its counterfac- tual underpinnings as developed by Lewis. Though a major contribution of Rubin's potential outcome model has been to stress the importance of the design stage, we recall the main method- ological and epistemological flaws of this approach. We argue that the study...

The heterogeneity of services in the freight transport market and the presence of imperfect information motivate the development of an empirical model for detecting market imperfection and bargaining powers of the two agents concluding a contract. The fact that agents typically bargain simultaneously on price and attributes leads to models without...

This survey is devoted to the statistical analysis of duration models and point processes. The first section introduces specific concepts and definitions for single-spell duration models. Section two is devoted to the presentation of conditional duration models which incorporate the effects of explanatory variables. Competing risks models are prese...

According to the current guidelines, it is advised not to treat patients with mild chronic hepatitis C. However, discussions as to giving immediately a treatment (direct treatment) to these patients have started and the incremental cost-effectiveness ratio (ICER) of such strategy is still unknown. The aim of this study was to estimate, in the healt...

The Valencia International Meetings on Bayesian Statistics, held every four years, provide the main forum for researchers in the area of Bayesian Statistics to come together to present and discuss frontier developments in the field. Covering a broad range of applications and models, including genetics, computer vision and computation, the resulting...

This paper proposes the construction of a Bayesian specification test based on the encompassing principle for the case of partial observability of latent variables. A structural parametric model (null model) is compared against a nonparametric alternative (alternative model) at the level of latent variables. The null extended model is obtained by i...

Cramer, Kamps and Schenk (Statist. Decisions, 2002) established conditions under which a family of joint distributions of two independent statistics is complete, and related their result with a previous one of Landers and Rogge (Scand. J. Statist., 1976). We first propose, within a sampling theory frame-work, a modification of Cramer, Kamps and Sch...

This paper contributes to the construction of a general theory for conditional models by making explicit the role of the exogenous randomness in the identification of conditional models. We start with a definition of identification in conditional models called weak identification, derived from the usual concept of identification in unconditional st...

In the social sciences, most studies are concerned with the possible causes, determinants, factors, etc. of a set of observations. In particular, for planning or policy reasons, it is important to know what causes which effects. In order to attain causal knowledge, many social scientists appeal to statistical modelling to confirm or disconfirm thei...

This paper deals with the Intersection Property, or Basu's First Theorem, which is valid under a condition of no common information, also known as measurable separability. After formalizing this notion, the paper reviews gen-eral properties and give operational characterizations in two topical cases: the finite one and the multivariate normal one....

Virtually all econometric models are conditional models. Nevertheless many of such models lose information by involving non-admissible conditionings. In this article we analyse, from a Bayesian point of view, the problem of admissible conditioning. Next, we design a methodology to evaluate the loss of information when a non-admissible conditioning...

Nowcasting concerns the inference on the current realization of random variables using information available until a recent past. This paper proposes a modelling strategy aimed at the best use of data for nowcasting based on panel data with severe deficiencies, namely, short time series and many missing data. The basic idea consists of introducing...

In this paper, we propose an empirical method to measure the market imperfection and the bargaining power of the agents, by extending the methods of frontier analysis. A case study in the Þeld of freight transport illustrates the proposed method.

This article aims to determine the location and the length of road sections characterized by a concentration of accidents (black zones). Two methods are compared: one based on a local decomposition of a global autocorrelation index, the other on kernel estimation. After explanation, both methods are applied and compared in terms of operational resu...

The object of this paper is to consider specification and identification problems for the case of models involving a latent hierarchical structure. After making some characteristics of such models explicit, the paper proposes a strategy of model specification characterized by a progressive introduction of hypotheses. Such a strategy allows us a sui...

Let MΘX = (RX ,X,PΘ = {Pθ: θ ε Θ}) be a parametrized statistical model and g: Θ → G be a non-injective function characterizing a parameter of interest. The basic idea of partial sufficiency is to find a (minimal) statistic sufficient for making inference on g(θ). Following Fraser (1956), Barndorff-Nielsen (1978) has defined a concept of S-sufficien...

Guillaume Wunsch, Michel Mouchart, Josianne Duchene This book is an outcome of the activities of the Working Group on Health, Morbidity, and Mortality Differentials of the European Association for Population Studies, which was chaired for some years by one of the present editors (G. Wunsch). In collaboration with the Institute of Statistics and the...

We consider the outcomes of a clinical trial as determined by one, or several, possibly hidden causes. This paper proposes a statistical model that allows such a distinction of causes not only for the main, or therapeutic, effects but also for the side, or toxic, effects. More specifically, we focus on trials where the effects are naturally dichoto...

this paper is the analysis of identification problems arisen in the specification of mixtures models when modelling individual behaviour. In such a case, a model corresponding to a given sample size is specified, and the basic question is to know under which assumptions identification conditions for this model may be obtained from conditions define...

The object of this paper is to review the main results obtained in semi- and non-parametric Bayesian analysis of duration models. Standard nonparametric Bayesian models for independent and identically distributed observations are reviewed in line with Ferguson's pioneering papers. Recent results on the characterization of Dirichlet processes and on...

: In Belgium welfare agencies receive a subsidy to employ welfare recipients for a period sufficiently long to entitle them to benefits of the contributory social insurance program. This work experience program without any training content is called Social Employment (SE). This paper investigates the effect of SE on the exit rate from welfare. We a...

this paper we shall analyze the identification problem from the point of view of a mixture model. Indeed, a natural approach to LISREL type models is to consider a hierarchical specification of the following type (for the sake of simplicity, we take n = 1): . the structural parameter #, i.e. a parameter upon which all individuals (of sample) depend...

In this paper, minimal conditions under which a semi-parametric binary response model is identified in a Bayesian framework are presented and compared to the conditions usually required in a sampling theory framework.

This note argues that a Bayesian framework is almost inescapable when specifying statistical models of the LISREL type, i.e. models involving not only latent and manifest variables but also incidental parameters. Indeed, a careful specification, making every hypothesis explicit and interpretable both contextually and statistically, requires a fully...

In this paper, we develop a Bayesian analysis of a semi-parametric binary choice model. The prior specification of the functional parameter, namely the distribution function of a latent variable, is of the Dirichlet process type and the prior specification of the Euclidean parameter, namely the coefficients of a linear combination of exogenous vari...

We first analyse the general problem of admissible conditioning and next consider the evaluation of the loss of information when a non-admissible conditioning is used as an approximaton of the exact posterior distribution. Considering the case of Fisher test, we evaluate from a Bayesian point of view how much information is lost when the sampling p...

In this paper a Bayesian least squares approximation is proposed for descriptive inference in a finite population when a categorical auxiliary variable is known. A hierarchical model II analysis of variance is assumed. The solution consists of a projection on the vector of group totals and on the between and within sums of squares. The approximatio...

Competing risks models are presented in the framework of single transition-multiple causes models. Particular attention is paid to general distributions of latent durations and to model identification problems, which are different from parameter identification problems. Implications for modelling and interpreting empirical findings are considered....

During the last fifteen years, statistical duration models have been increasingly used by econometricians to analyze different economic problems. The first real application of these models was probably the analysis of individual unemployment (and then employment) duration data. Papers by Lancaster [1979], Nickell [1979], Lancaster and Nickell [1980...

This paper presents recent econometric models of individual mobility on the labor market. The general framework is given by the theory of point processes which are continuous time finite state space processes. The first part of the paper is devoted to specification problems and the inference of the introduced models is not treated. The last part of...

The object of this paper is to report, for a simple testing problem of a unit root hypothesis, some experience regarding the numerical problems involved by using a Bayesian encompassing test, i.e., a Bayesian procedure that treats the null and the alternative hypotheses as different models, the null one and the alternative one, that share a same sa...

The Econometric Modelling of Individual Transitions on the Labour Market, by Jean-Pierre Florens, Denis Fougère, Thierry Kamionka and Michel Mouchart. This paper presents recent econometric models of individual mobility on the labour market. The general framework is given by the theory of point processes, which are continuous-time finite state sp...

This chapter focuses on Bayesian methods and illustrates both the intrinsic unity of Bayesian thinking, and its basic flexibility to adjust to and to cope with a wide range of circumstances. Two ideas are emphasized in the chapter. Firstly hypothesis testing and model choice have been dealt with as a single class of problems met with so strikingly...

In this paper it is shown that a subprocess of a Markov process is markovian if a suitable condition of noncausality is satisfied. Furthermore, a markovian condition is shown to be a natural condition when analyzing the role of the horizon (finite or infinite) in the property of noncausality. We also give further conditions implying that a process...

: The object of this paper is to explore how to use administrative data for econometric purposes when these data measure total annual duration spent in various states. A basic issue addressed is whether the information extracted through categorising continuous data actually defines an economically meaningful concept of an individual position on the...

The Valencia International Meetings on Bayesian Statistics, held every four years, provide the forum for researchers to come together to present and discuss frontier developments in the field. The resulting Proceedings provide a definitive, up-to-date overview encompassing a wide range of theoretical and applied research. This fourth Proceedings is...

In this paper a Bayesian least squares approximation is proposed for descriptive inference in a finite population when a categorical auxiliary variable is known. A hierarchical model II analysis of variance is assumed. The solution consists of a projection on the vector of group totals and on the between and within sums of squares. The approximatio...

This paper calculates indices of central bank autonomy (CBA) for 163 central banks as of end-2003, and comparable indices for a subgroup of 68 central banks as of the end of the 1980s. The results confirm strong improvements in both economic and political CBA over the past couple of decades, although more progress is needed to boost political auton...

Are classical test statistics of any use to Bayesian statisticians and econometricians? If so, are these statistics put to the same use by Bayesians and classical statisticians? The modest aim of this paper is to illustrate, through elementary tales, why and how Bayesians may sometimes be led to use classical test statistics — either in their usual...

Conditional completeness is shown to provide a sufficient condition for maximal ancillarity. Using properties linking ancillarity and complete sufficient statistics, the new condition is shown to be more general than another sufficient condition given by D. Basu [Sankhyā 21, 247- 256 (1959; Zbl 0091.148)]. It is also of more practical interest and,...

This paper presents preliminary results of an analysis of household load-curves, using spline functions. Bayesian methods are used to relax, in probability, smoothness restrictions underlying spline functions. The comparison of various posterior distributions, corresponding to alternative prior specifications, allows one to better appreciate how fa...

Alternative progressive strategies for specification of linear dynamic models are presented. The main theme is that specification is basically concerned with endowing the pure incidental case—i.e., the case of different moment for each observations—with progressively more structure. Linearity and exogeneity are successively introduced in that spiri...

A Bayesian experiment is defined by a unique probability on the product of the parameter space and the sample space. This joint probability determines a conditional independance relation which is used for a symmetrical analysis of sufficiency and ancillarity on the parameter and the sample. Identification is then considered as a property of minimal...

Three different price adjustment equations are compared, in the framework of a standard disequilibrium model, and are shown to provide observationally equivalent models. Implications for testing equilibrium are spelled out.