Cette thèse porte sur la consommation énergétique du secteur tertiaire marchand en France. Notre approche est empirique, et s’appuie sur l’Enquête sur les Consommations d’Énergie dans le Tertiaire (ECET). À partir d’une revue de littérature consacrée à l’étude économétrique des déterminants de la consommation énergétique des bâtiments (résidentiels ou tertiaires), nous mettons en évidence la coexistence de deux méthodes d’estimation, sur lesquelles nous revenons. Nous exposons également les consensus ou les débats portant sur les effets des déterminants étudiés. Dans la mesure où les données mobilisées dans cette thèse sont issues d’une enquête à plan de sondage complexe, nous présentons les outils statistiques adaptés à l’exploitation de données issues de ce type d’échantillonnage, puis nous analysons la controverse qui continue d’exister sur l’estimation de modèles économétriques sur données d’enquête. Nous mobilisons alors notre base de données pour fournir une première description statistique des consommations énergétiques du secteur tertiaire marchand en France. Cette description statistique s’appuie sur nomenclature que nous construisons afin d’obtenir des informations à l’échelle infra-sectorielle. Enfin, nous utilisons l’ensemble des méthodes et approches préalablement identifiées pour étudier les déterminants de la demande énergétique des établissements tertiaires sur la base d’une analyse économétrique des données de l’ECET. Cela nous conduit à effectuer une double lecture de nos résultats, à la fois comme éléments de réponse à la question de l’impact des variables étudiées sur la demande d’énergie des établissements, et comme éléments de comparaison des différentes méthodes. / This thesis deals with energy consumption in the French tertiary buildings. We adopt an empirical approach based on a French national survey, the Tertiary Buildings Energy Consumption Survey (l’Enquête sur les Consommations d’Énergie dans le Tertiaire (ECET)). From a literature review, we present the econometric analyses of the drivers of energy consumption in buildings (residential and tertiary). This review highlights the coexistence of two estimation methods. We discuss these methods, and then we detail consensus and debates on the effects of every driver that has already been analyzed in the literature. Because the data we use are complex sample survey data, we describe the statistical tools that must be used to analyze this type of data, and next present the still controversial issue of econometric modeling based on survey data. Then we use our database to produce a first statistical description of energy consumption in tertiary buildings. This description is based on a nomenclature that we establish to obtain information at the subsector level. Finally, we use all the methods and approaches identified previously to study the drivers of the tertiary buildings’ energy demand by implementing an econometric analysis on the ECET data. This lead to a double analysis of our results, both as elements of knowledge on the impact of each driver on energy consumption and as materials that help to compare the different methods to use.

No description available.

Submitted by Andre Maitan (andremaitan@yahoo.com.br) on 2011-07-25T18:57:36Z No. of bitstreams: 1 Tese MPFE - André Raj Maitan - 93280.pdf: 894819 bytes, checksum: cd191610788a5e242005e02d0e272ee3 (MD5) / Approved for entry into archive by Gisele Isaura Hannickel (gisele.hannickel@fgv.br) on 2011-07-25T19:18:50Z (GMT) No. of bitstreams: 1 Tese MPFE - André Raj Maitan - 93280.pdf: 894819 bytes, checksum: cd191610788a5e242005e02d0e272ee3 (MD5) / Approved for entry into archive by Gisele Isaura Hannickel (gisele.hannickel@fgv.br) on 2011-07-25T19:20:01Z (GMT) No. of bitstreams: 1 Tese MPFE - André Raj Maitan - 93280.pdf: 894819 bytes, checksum: cd191610788a5e242005e02d0e272ee3 (MD5) / Made available in DSpace on 2011-07-25T19:28:19Z (GMT). No. of bitstreams: 1 Tese MPFE - André Raj Maitan - 93280.pdf: 894819 bytes, checksum: cd191610788a5e242005e02d0e272ee3 (MD5) Previous issue date: 2011-06-27 / Our target here is to understand how non observables factors on water heating choice can affect electricity usage by individual consumers. In order to understand this, we will apply a discrete/continuous model in order to understand the impact of household water heating system discrete choice on household electricity consumption, continuous choice. First it was used a LOGIT model as a way to understand which variables influence consumer discrete choice. Hereafter with Discrete model results we verified if consumer choice is relevant on electricity consumption intensity. Hausman method results demonstrated that if we do not specify discrete choice at continuous model, it will bring biased estimators to our model, this occurs due to the fact that discrete choice variable is significant and it helps to explain continuous model. Results show that we have roughly 10% differences on estimators when we ignore discrete choice. It’s important to note that this kind of miscalculation can bring a important difference on government investment decision. This work and the results obtained reinforced Dubbin and McFadden (1984) work that it was used as principal reference to this work. / O objetivo deste trabalho passa por entender o impacto dos fatores não observáveis na escolha do sistema de aquecimento de água na intensidade do consumo de eletricidade domiciliar. Para isto será aplicado um modelo discreto-contínuo para entender o efeito da escolha discreta do tipo de aquecimento de água domiciliar sobre o consumo de energia elétrica do mesmo domicílio, escolha contínua. Primeiramente foi usado um modelo LOGIT e por meio deste foram entendidos os fatores que influenciam a escolha discreta do consumidor. Com os resultados encontrados na escolha discreta, temos em um segundo momento que constatar se esta escolha é significante para a intensidade do consumo de energia elétrica. Os resultados obtidos usando o método de Hausmann demonstraram que ignorar este fator (escolha discreta) pode levar a estimadores viesados para a parte contínua do modelo, ou seja, a escolha discreta é significante e ajuda a explicar o consumo de energia elétrica domiciliar. Os resultados mostraram que temos diferenças de até 10% no estimador obtido quando ignoramos este fator. É importante notar que este tipo de falha no cálculo, pode trazer problemas na decisão e na quantidade de investimento de um país. Além disso, os resultados reforçaram e ratificaram o trabalho desenvolvido por Dubbin e McFadden (1984) que foram usados como base para o estudo.