Search

We propose an Adjusted Quasi-Score (AQS) method for constructing tests for homoskedasticity in spatial econometric models. We first obtain an AQS function by adjusting the score-type function from the given model to achieve unbiasedness, and then develop an Outer-Product-of-Martingale-Difference (OPMD) estimate of its variance. In standard problems where a genuine (quasi) score vector is available, the AQS-OPMD method leads to finite sample improved tests over the usual methods. More importantly in non-standard problems where a genuine (quasi) score is not available and the usual methods fail, the proposed AQS-OPMD method provides feasible solutions. The AQS tests are formally derived and asymptotic properties examined for three representative models: spatial cross-sectional, static or dynamic panel models. Monte Carlo results show that the proposed AQS tests have good finite sample properties.

International audience; This paper analyzes the demand of French farmers for pesticides by disaggregating among the three main categories of pesticides. We estimate the demand elasticities of herbicides, insecticides, and fungicides with respect to pesticide expenditure, and also consider crop differentiation. We retain a Linear Approximated Almost Ideal Demand System specification. A Full-Information Maximum Likelihood estimation procedure is used to deal with corner solution problems and censored data. The estimation is based on two cross-sections covering pesticide use of three major crops cultivated in France in 2001 and 2006. Our results show that farmers' response to price variation is very low, especially for 2001. Furthermore, elasticities of pesticide expenditure are significantly different across categories: the highest levels are obtained for fungicides and the lowest ones for insecticides. Finally, we find higher own-price elasticities for herbicides and fungicides than for insecticides, which are less frequently used.; Cet article analyse la demande désagrégée de pesticides des agriculteurs en distinguant les trois principales catégories de traitements (herbicides, insecticides, et fongicides). Les dépenses sont modélisées sous la forme d'un système de demande AIDS linéaire (« Almost Ideal Demand System ») avec comme variables explicatives les caractéristiques des agriculteurs et leurs pratiques culturales. L'estimation par le Maximum de Vraisemblance à Information Complète permet de considérer l'hétérogénéité des niveaux de pression phytosanitaire, qui varient selon les types de traitement, les cultures et/ou les zones géographiques, ce qui se traduit par des niveaux de dépenses nulles pour certaines catégories. Cette méthode d'estimation est appliquée à des données individuelles relatives aux pratiques des agriculteurs pour trois grandes cultures (blé tendre, blé dur et orge) emblavées en France en 2001 et 2006. Les résultats obtenus révèlent que la demande est globalement peu sensible aux variations du prix des produits. En outre, les élasticités prix directes des herbicides et des fongicides sont plus élevées que celles des insecticides, qui correspondent à la catégorie de traitement la moins appliquée.

SESSION 1: Contributed lectures; National audience; This paper focuses on panel data models with spatially nested random ef- fects. This specification is useful for panel data applications which exhibit spatial dependence and a nested (hierarchical) structure. We propose to use a generalized moments estimator in the spirit of Kelejian and Prucha (1998, 1999) and Kapoor, Kelejian and Prucha (2007) for estimating the spatial autoregressive parameter and the variance components of the disturbance process. Then a spatial counterpart of the Cochrane-Orcutt transformation is defined to obtain a feasible generalized least squares procedure to estimate the regression parameters. Using Monte Carlo simulations, we show that our estimators perform well in terms of root mean square error compared to the maximum likelihood estimator. The approach is demonstrated using English house price data by district, with districts nested within counties.

This paper focuses on the estimation and predictive performance of several estimators for the dynamic and autoregressive spatial lag panel data model with spatially correlated disturbances. In the spirit of Arellano and Bond (1991) and Mutl (2006), a dynamic spatial GMM estimator is proposed based on Kapoor, Kelejian and Prucha (2007) for the Spatial AutoRegressive (SAR) error model. The main idea is to mix non-spatial and spatial instruments to obtain consistent estimates of the parameters. Then, a linear predictor of this spatial dynamic model is derived. Using Monte Carlo simulations, we compare the performance of the GMM spatial estimator to that of spatial and non-spatial estimators and illustrate our approach with an application to new economic geography.

This article focuses on the demand system of French farmers concerning pesticides uses. We estimate the demand elasticities of herbicides, insecticides and fungicides with respect to pesticide expenditure, and considering crop differentiation. Then we compare two indexes that are used in agronomic literature to measure the intensity of pesticides uses. We retain a Linear Approximated Almost Ideal Demand System (LA/AIDS) specification. A Full-Information Maximum Likelihood estimation procedure is used for dealing with the problem of censored dependent variable. We consider two cross-sections observed in 2001 and 2006 covering pesticides uses of three crops. We confirm the previous results of the literature that farmers response to price variation is very low, with higher prices response in 2006 than in 2001. Moreover, we find that conditional herbicides expenditure elasticities are often higher than insecticides expenditure elasticities, but lower than those of fungicides. We find higher own-price elasticities for herbicides and fungicides than for insecticides, which is the less used. Finally, application dose seems statistically better to explain herbicides decision, whereas treatment frequency index appears better for insecticides and fungicides. However, most of elasticities are closed for dose and treatment frequency index.

This paper compares various forecasts using panel data with spatial error correlation. The true data generating process is assumed to be a simple error component regression model with spatial remainder disturbances of the autoregressive or moving average type. The best linear unbiased predictor is compared with other forecasts ignoring spatial correlation, or ignoring heterogeneity due to the individual effects, using Monte Carlo experiments. In addition, we check the performance of these forecasts under misspecification of the spatial error process, various spatial weight matrices, and heterogeneous rather than homogeneous panel data models.

This work presents a new perspective on characterizing the similarity between elements of a database or, more generally, nodes of a weighted and undirected graph. It is based on a Markov-chain model of random walk through the database. More precisely, we compute quantities (the average commute time, the pseudoinverse of the Laplacian matrix of the graph, etc) that provide similarities between any pair of nodes, having the nice property of increasing when the number of paths connect- ing those elements increases and when the “length” of paths decreases. It turns out that the square root of the average commute time is a Eu- clidean distance and that the pseudoinverse of the Laplacian matrix is a kernel (its elements are inner products closely related to commute times). A procedure for computing the subspace projection of the node vectors of the graph that preserves as much variance as possible in terms of the commute-time distance – a principal component analysis (PCA) of the graph – is also introduced. This graph PCA provides a nice interpreta- tion to the “Fiedler vector”, widely used for graph partitioning. The model is evaluated on a collaborative-recommendation task where sug- gestions are made about which movies people should watch based upon what they watched in the past. Experimental results on the MovieLens database show that the Laplacian-based similarities (the pseudoinverse of the Laplacian matrix and the “random-forest matrix”) perform well in comparison with other methods. The model, which nicely fits into the so-called “statistical relational learning” framework, could also be usedto compute document or word similarities, and, more generally, it could be applied to machine-learning and pattern-recognition tasks involving a database.

This work presents some general procedures for computing dissimilarities between nodes of a weighted, undirected, graph. It is based on a Markov-chain model of random walk through the graph. This method is applied on the architecture of a Multi Agent System (MAS), in which each agent can be considered as a node and each interaction between two agents as a link. The model assigns transition probabilities to the links between agents, so that a random walker can jump from agent to agent. A quantity, called the average first-passage time, computes the average number of steps needed by a random walker for reaching agent k for the first time, when starting from agent i. A closely related quantity, called the average commute time, provides a distance measure between any pair of agents. Yet another quantity of interest, closely related to the average commute time, is the pseudoinverse of the Laplacian matrix of the graph, which represents a similarity measure between the nodes of the graph. These quantities, representing dissimilarities (similarities) between any two agents, have the nice property of decreasing (increasing) when the number of paths connecting two agents increases and when the "length" of any path decreases. The model is applied on a collaborative filtering task where suggestions are made about which movies people should watch based upon what they watched in the past. For the experiments, we build a MAS architecture and we instantiated the agents belief-set from a real movie database. Experimental results show that the Laplacian-pseudoinverse based similarity outperforms all the other methods.

Multi-Agent Systems (MAS) architectures now appear to be more appropriate than traditional ones for building latest generation software that is typically concurrent, distributed, and dynamic. Since the fundamental concepts of multi-agent systems are social and intentional, rather than object, functional or implementation-oriented, the design of MAS architectures should be eased by using what we call social patterns rather than object-oriented design patterns. Social patterns are idioms inspired by social and intentional characteristics used to design the details of a system architecture. The paper presents a framework called SKWYRL used to gain insight into social patterns and help to conduct the design of a MAS architecture in terms of these newly proposed idioms. We define the Booking social pattern, to illustrate the modeling dimensions of SKWYRL. A framework for code generation is also presented.

This paper is the second part of the object-oriented study of the Carsid cokerie. The aim of this project was to redesign the data models from the Marchienne plant of CARSID and to integrate all of them in one single data model, while improving its design and showing the possibilities offered by OLAP technologies in the bettering of the data exploitation carried out in this plant.

Multi-Agent Systems (MAS) architectures now appear to be more appropriate than traditional ones for building latest generation software that is typically concurrent, distributed, and dynamic. Since the fundamental concepts of multi-agent systems are social and intentional, rather than object, functional or implementation-oriented, the design of MAS architectures should be eased by using what we call social patterns rather than object-oriented design patterns. Social patterns are idioms inspired by social and intentional characteristics used to design the details of a system architecture. The paper presents a framework called SKWYRL used to gain insight into social patterns and help to conduct the design of a MAS architecture in terms of these newly proposed idioms. We define the Booking social pattern, to illustrate the modeling dimensions of SKWYRL. A framework for code generation is also presented

Multi-Agent Systems (MAS) architectures are gaining popularity over traditional ones for building open, distributed, and evolving software. Since the fundamental concepts of multi-agent systems are social and intentional rather than object, functional, or implementation-oriented, the design of MAS architectures should be eased by using what we call social patterns rather than object-oriented design patterns. Social patterns are idioms inspired by social and intentional characteristics used to design the details of a system architecture. The paper presents a framework called SKwyRL used to gain insight into social patterns and help design a MAS architecture in terms of these new idioms. The framework is integrated in the Tropos agent methodology. It is developed according to the five modeling dimensions provided by Tropos: social, intentional, structural, communicational, and dynamic. We consider the Broker social pattern as a combination of patterns and use it to illustrate the modeling dimensions of SKwyRL. A framework for code generation is also presented as well as an e-business broker module.

A design pattern describes a structure of communicating components that solves a commonly occurring design problem. Designing with patterns offers the possibility of raising the abstraction level at which design is performed, with improvements in clarity, understanding, and facility of maintenance of applications. However, in their most common presentation, design patterns are informal pieces of design process, which application is not reflected in the operational system, and the potential advantages of a more principled design are not realized. This work proposes to organize design in such a way that pattern applications remain explicit in the operational systems. A reflective architecture is proposed, where patterns are reified as metalevel constructs., Sociedad Argentina de Informática e Investigación Operativa