This paper investigates the power properties of the Sargan test in the presence of measurement errors in dynamic panel data models. The conclusion from Monte Carlo simulations, and an application on the data used by Arellano and Bond (1991), is that in the very likely case of measurement errors in either the dependent or any of the independent variables, we will, if we rely on the Sargan test, quite likely accept a mis-specified model and end up with biased results.

CoordenaÃÃo de AperfeiÃoamento de Pessoal de NÃvel Superior / A tese à composta de trÃs artigos e em todos eles utilizam-se modelos de estimaÃÃo para dados em painel dinÃmico. Os mÃtodos de estimaÃÃo empregados sÃo os Momentos Generalizado-sistema (MMG-sistema) desenvolvido por Arellano-Bond (1991), Arellano-Bover (1995) e Blundell e Bond (1998). O primeiro artigo intitulado âO Sistema de Seguridade Social e a Pobreza Rural no Brasilâ analisa o impacto da aposentadoria da seguridade social na pobreza. Os resultados indicaram que a aposentadoria rural nÃo tem impacto significativo na reduÃÃo da pobreza, nÃo corroborando assim a hipÃtese daqueles que afirmam que a seguridade social rural resolve significativamente a reduÃÃo da pobreza. O segundo artigo intitulado âDeterminantes da Desigualdade de Renda no Brasilâ propÃe-se analisar as contribuiÃÃes de diferentes determinantes para a reduÃÃo da desigualdade de renda no paÃs. Os resultados apontaram que as transferÃncias de renda do governo federal nÃo afetaram a dinÃmica da desigualdade de renda no perÃodo analisado. Em relaÃÃo aos outros determinantes, a educaÃÃo foi o fator preponderante na reduÃÃo da desigualdade. A segunda contribuiÃÃo mais importante foi a renda de todos os trabalhos. A carga tributÃria do governo contribuiu para aumentar a desigualdade de renda no Brasil. O terceiro artigo intitulado âCrescimento EconÃmico e ConcentraÃÃo de Renda: Seus Efeitos na Pobreza no Brasilâ verifica o impacto de variaÃÃes no crescimento econÃmico e da desigualdade de renda sobre as alteraÃÃes da pobreza no Brasil. Uma vez que somente o crescimento nÃo à capaz de explicar alteraÃÃes da pobreza, considera-se tambÃm a desigualdade de renda como fator complementar no estudo sobre esta Ãltima, buscando avaliar a hipÃtese de Bourguignon (2002) de que quanto mais desigual for o paÃs menor seria a efetividade do crescimento econÃmico em reduzir a pobreza. Os resultados mostram que a elasticidade desigualdade-pobreza à maior do que a elasticidade renda-pobreza e a elevada desigualdade e o baixo nÃvel de desenvolvimento inicial da maioria dos estados brasileiros sÃo empecilhos para a reversÃo do quadro de pobreza, via crescimento da renda. / This thesis is composed of three works, and they all use dynamic panel data estimation models. The estimation models used are the Generalized Moments system, developed by Arellano-Bond (1991), Arellano-Bover (1995) and Blundell and Bond (1998). The first work is entitled âThe Social Security system and Rural Poverty in Brazilâ, and it analyzes the impact of retirement through social security on poverty. The results indicate that rural retirement has no significant impact on poverty reduction, thus not corroborating the hypothesis which states that rural social security significantly decreases poverty. The second work is entitled âInequality Determinants in Brazilâ, and it aims to analyze the contribution of different determinants of income inequality reduction in the country. The results show that income transfers from the federal government do not affect the dynamics of income inequality in the period studied. When considering the other determinants, education was the main factor in reducing inequality. The second most important contribution was from income of all kinds of labor. The governmentâs tax burden contributes to increase income inequality in Brazil. The third work is entitled âEconomic Growth and Income Concentration: Its effects on Poverty in Brazilâ, and it considers the impact of variations in economic growth and income inequality on poverty alterations in Brazil. Since economic growth in itself is not capable of explaining alterations in poverty, the work takes income inequality into account as a complementary factor in analyzing poverty, aiming to evaluate Bourguignonâs (2002) hypothesis, which states that the higher the inequality in a country, the lesser the effectiveness of economic growth in reducing poverty will be. The results show that the inequality-poverty elasticity is greater than the income-poverty elasticity and the high inequality and low initial development levels of the majority of states are obstacles for reversing the poverty situation via income growth.

We estimate the speed of income convergence for a sample of 196 European NUTS 2 regions over the period 1985-1999. So far there is no direct estimator available for dynamic panels with strong spatial dependencies. We propose a two-step procedure, which involves first spatial filtering of the variables to remove the spatial correlation, and application of standard GMM estimators for dynamic panels in a second step. Our results show that ignorance of the spatial correlation leads to potentially misleading results. Applying a system GMM estimator on the filtered variables, we obtain a speed of convergence of 6.9 per cent and a capital elasticity of 0.43. / Series: EI Working Papers / Europainstitut

In this paper, we argue that shadow activities and different levels of marketization of household production systematically distort international comparisons of aggregate gross household saving rates (HSRs): Higher shares of hidden income increase observed HSRs. Panel data for 18 (24) OECD-countries covering a period of a decade show that gross HSRs are positively related to the degree of corruption(used as a proxy for the propensity to shift economic activities into the shadow) and to the share of income from property and self employment. At the same time, gross HSRs are negatively related to the female employment rate, the ratio of indirect taxes to direct taxes, and to the tax wedge. One plausible story behind these phenomena might be that unobserved consumption and wages in the shadow labor market induce an upward bias in observed HSRs and profit shares, while the price level effects of a higher share of indirect taxes and a 'welfare state' effect lower observed HSRs. (authors' abstract) / Series: Department of Economics Working Paper Series

Made available in DSpace on 2016-03-15T19:31:20Z (GMT). No. of bitstreams: 1 Patricia Marilia Ricomini e Almeida.pdf: 431610 bytes, checksum: 0efa171a28be7631027cccbcc8a1bc23 (MD5) Previous issue date: 2010-03-08 / Fundo Mackenzie de Pesquisa / The concept of efficient market hypothesis has been the focus of finances for a long time, with the development of powerful theoretical reasons to explain why the hypothesis should remain. Since the beginning of 1930, the first papers about the analysis of securities were in evidence, it s been emerging the idea that the fundamental value of any security should be equal to the discounted cash flow from it and prices would vary around their fundamental values. Despite of security s present value beeing the best indicator to reflect their true value, this model covers expectations about future income, discount rate and people s racionality, becoming dificult the aplication of this model uses. As consequence, Campbell e Shiller (1987) developed the cointegration model, a powerful framewoork for testing expectations and racionality in financial markets. In this context, the literature about panel with unit roots and cointegration have been extended in a fast way. In part, this is happening due to the complex nature of interactions and dependences that, generally, ocurrs during the time and between individual units in the panel. The major recent concern of econometric literature, related to the cointegration tests and the unit roots of the dynamic panels, has been the development of tests that control the cross sectional dependence. In such case, an econometric model was adopted based on the application of the unit roots and the cointegration tests in panel, with the firm beeing the unit of analysis. To deal with the serial correlation, problems of nonstationary series as well as problems of small sample, recent techniques were applied in this study: panel dynamic OLS (DOLS) and fully modified OLS (FMOLS). Nine stocks, that compose the São Paulo Stock Exchange Index, have been analyzed throughout the period between 1994 and 2008. Summarizing, in spite of some conflicting results, it s possible to prove that there is a cointegration process between the prices of equities traded at BM&FBOVESPA and the dividends. The results obtained in this study allow the partial validation of the present value model at the firm level. However, the prices seemed not to reflect the expectation of dividends for the brazillian market. Therein, the prices of equities are over valued in relation to the payment of the dividends. Future researchs about the present value model for the brazillian market should be done. / A Hipótese de Eficiência de Mercado tem sido a proposição central das finanças durante muitos anos, com o desenvolvimento de razões teóricas poderosas explicando porque a hipótese deveria permanecer. A partir da década de 30, destacam-se os primeiros trabalhos relacionados à análise de títulos, surgindo a idéia de que o valor fundamental de qualquer título seria igual ao fluxo de caixa descontado deste, e que os preços atuais iriam variar em torno dos valores fundamentais. Apesar do valor presente de um ativo ser o melhor indicador para refletir seu verdadeiro valor, ele envolve expectativas sobre a renda futura, a taxa de desconto e a racionalidade das pessoas, tornando difícil a aplicação na prática desse modelo. Surge, então, o modelo de cointegração de Campbell e Shiller (1987), uma ferramenta útil para o teste de expectativas e racionalidade nos mercados financeiros. Nesse contexto, a literatura sobre painel com raízes unitárias e cointegração começa a crescer rapidamente. Em parte isso ocorre em função da natureza complexa de interações e dependências que geralmente existem ao longo do tempo e entre as unidades individuais no painel. A maior preocupação da literatura, no que diz respeito à análise de dados em painéis dinâmicos, tem sido o desenvolvimento de testes que controlem a dependência cross sectional. Nesse estudo,adotou-se um modelo econométrico baseado na aplicação de raízes unitárias e testes de cointegração em painel, tendo a empresa como unidade de análise. Para tratar a correlação serial, problemas de não estacionariedade das séries bem como problemas de pequena amostra, foram adotadas as técnicas de Panel Dynamic OLS (DOLS) e Fully Modified OLS (FMOLS). Analisaram-se nove ações (Klabin, Petrobrás, Bradesco, Itáu, Cemig, Ambev, Souza Cruz, Aracruz e Vale) para o período de 1994 a 2008. Apesar de alguns resultados conflitantes, é possível afirmar que existe um processo de cointegração entre os preços das ações negociadas e os dividendos. Os resultados permitem a validação parcial do modelo de valor presente a nível individual, já que os preços das ações parecem não refletir integralmente a expectativa dos dividendos para o mercado brasileiro. Os preços das ações estão superavaliados com relação aos dividendos. Recomendam-se pesquisas futuras sobre o modelo de valor presente para o mercado financeiro brasileiro.

Cette thèse se compose de trois articles empiriques appliquant à divers sujets des techniques d'économétrie sur données de panel. L'article principal étudie l'évolution de la transmission des prix des trois principales céréales (blé, maïs, riz) des marchés internationaux vers les producteurs domestiques pour 52 pays sur la période 1970-2013, et cherche à identifier les principaux facteurs expliquant l'hétérogénéité des pass-through. Le second article mesure l'élasticité du trafic aérien au produit intérieur brut dans le monde et met en évidence la grande stabilité de la relation tant dans le temps que d'une région à l'autre. Enfin, le troisième article modélise le mécanisme de fixation du taux de rendement servi par les assureurs français à leurs clients sur les produits d'assurance-vie. / This dissertation is composed of three empirical articles resorting to econometric methods in panel data analysis to address various research questions. The main article investigates the evolution of the level of price transmission for the three major cereals (wheat, maize and rice) from the international commodity markets down to the local producers for 52 countries between 1970 and 2013 while attempting to identify the main drivers of the heterogeneity in pass-through. The second article measures the elasticity of air-traffic to GDP around the world and demonstrates that the relationship is very stable across régions and through time. Eventually, the third article models the mechanisms through which French life-insurers set the rate of return they pay annually to their policyholders.