Submitted by Felipe Castor (felipecastor15@gmail.com) on 2014-08-29T00:17:59Z No. of bitstreams: 1 dissertaçãonovafinal.pdf: 887468 bytes, checksum: 73ab0f9aa41de8d5561dc1eacdc1ad2e (MD5) / Approved for entry into archive by BRUNA BARROS (bruna.barros@fgv.br) on 2014-09-04T13:53:50Z (GMT) No. of bitstreams: 1 dissertaçãonovafinal.pdf: 887468 bytes, checksum: 73ab0f9aa41de8d5561dc1eacdc1ad2e (MD5) / Approved for entry into archive by Marcia Bacha (marcia.bacha@fgv.br) on 2014-09-22T19:02:19Z (GMT) No. of bitstreams: 1 dissertaçãonovafinal.pdf: 887468 bytes, checksum: 73ab0f9aa41de8d5561dc1eacdc1ad2e (MD5) / Made available in DSpace on 2014-09-22T19:10:00Z (GMT). No. of bitstreams: 1 dissertaçãonovafinal.pdf: 887468 bytes, checksum: 73ab0f9aa41de8d5561dc1eacdc1ad2e (MD5) Previous issue date: 2014-05-30 / Some studies in economics and political science argues that preferences for redistribution are able to explain different redistributive efforts around the world. It is natural to think that the voting decision is the most direct channel to validate this argument. This work provided empirical evidence that there is a correlation between preferences for redistribution and individual voting decision in the United States. First, I present a theoretical model that links party identification with individual self-interest. Based in this model, I use General Social Survey data to assess such relation. The results show that preferences for redistribution are related to party identification and, consequently, related to individual voting decision for redistributive policies. There is also some evidences that this influence is growing in the last decades. / Parte dos estudos em economia e ciência política argumenta que preferências por redistribuição são capazes de explicar os diferentes esforços redistributivos ao redor do mundo. É natural pensar que a decisão de voto é o único canal relevante para que esta argumentação seja válida. Este trabalho buscará evidências empíricas de que haja correlação entre preferências por redistribuição e a decisão individual de voto nos Estados Unidos. Primeiro, apresento um modelo teórico que faz ligação entre a identificação partidária do indivíduo com seus interesses próprios e coletivos. Com base neste modelo, serão usados dados do General Social Survey para encontrar tais relações. Nos resultados encontra-se que as preferências por redistribuição estão relacionadas à identificação partidária e, consequentemente, à decisão de voto por redistribuição. Há também alguma evidência de que esta influência esteja crescendo nas últimas décadas.

Cette thèse s'intéresse à l'analyse des variables d'opinion. Les opinions couvertes concernent spécifiquement des questions économiques comme le niveau de bien-être, la situation financière, le niveau minimum de revenu nécessaire pour vivre dignement, la préférence pour la redistribution. Le traitement de ces variables d'opinion et leur mise en relation avec les grandeurs économiques traditionnelles comme le niveau de revenu ou sa dynamique nécessitent des techniques micro-économétriques spécifiques. Dans cette thèse, les modèles dynamiques de panel sont utilisés pour étudier la mobilité des revenus et la mobilité des statuts d'emploi. Dans le dernier chapitre, nous discutons également le modèle Tobit dynamique et l'importance des conditions initiales. Dans les modèles de panel, se pose la question du choix entre effet fixe et effet aléatoire. Parce que les variables subjectives sont ordinales et discrètes, les effets fixes sont difficilement identifiables. L'identification des effets aléatoires est moins problématique, mais l'estimation devient pourtant difficile quand la dimension d'intégration augmente. Pour résoudre cela, je l'utilise intensivement les techniques de simulation, soit pour le modèle dynamique multinomial logit, soit pour le modèle dynamique Tobit. La technique de simulation est également appliqué au modèle trivarié probit pour mesurer les corrélations conditionnelles entre trois (ou même plus) variables ordinales. / This dissertation studies option variables. These opinions cover specially economic questions such as the level of wellbeing, financial situations, the minimum income question and the preference for redistribution. The treatment of these opinion variables and their relation to conventional economic questions such as income level or dynamics requires the use of special micro-econometric models. In this dissertation, the dynamics panel models are used to study the job status and income mobility. In the 6 chapter, we discuss the dynamic Tobit model with an emphasis on initial conditions. The choice between fixed or random effect is another question. Because subjective variables are ordinal and discrete, the identification of fixed effects is problematic. Random effects are better identified while the estimation difficulty increases with the integration dimension. To solve that, I use intensively simulation method in the study of dynamic multinomial logit model or dynamic tobit model. It is also been applied in the trivariate probit model to measure the conditional correlations among more than 2 ordinal variables.