Essays in taxation and savings

• Ano de defesa: 2013
• Autor(a) principal: Monteiro, Marcos José Pérez
• Orientador(a): Ferreira, Pedro Cavalcanti Gomes
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Palavras-chave em Português: Tributação; Poupança; Econometria de dados em painel
• Palavras-chave em Inglês: Panel-data econometrics; Taxation; Savings
• Link de acesso: https://hdl.handle.net/10438/11356

This dissertation is composed of three articles. Two of them investigate taxation issues and the third is a paper on savings. Although the subject of analysis is different among them, they all share the com- mon feature of applying panel-data econometric techniques on newly assembled datasets. Two of the essays employ GMM estimation on dynamic panel frameworks and the remaining one is an application on panel limited-dependent variable models. A short summary of each paper is provided below. It starts with the two essays on taxation, which share a common contextualizing section on the Brazilian state level value-added tax (ICMS), and finishes with the one on savings. Essay 1 makes an assessment of the importance of enforcement as an instrument to deter tax evasion and increase tax revenue, in the case of a developing country Value-Added Tax (VAT). It uses data from the Brazilian State of São Paulo. To cope with inertia in the revenue series and potential endogeneity, dy- namic panel-data techniques are employed. The level of regional GDP and two proxies for enforcement, namely the quantity and the value of penalties inflicted, were used as covariates. The results indicate a significant impact of the enforcement on tax revenues. The paper indirectly provides evidence on how non-compliance is affected by the penalty for detected evasion. Its conclusions are also relevant for the discussions on Brazilian fiscal federalism, specially in the case of a potential tax reform. Essay 2 examines one of the main tasks of a tax revenue service, which is to periodically choose which taxpayers will be subject to auditing. Improved efficiency in the audit selection mechanism is likely to impact positively the probability of fraud detection, resulting in better allocation of scarce fiscal resources. This paper attempts to design such a mechanism by computing the taxpayers probabilities of non-compliance. This is done by optimally combining on latent-dependent variable formulations several already existing indicators with information of ex-post audit results on a restricted sample of audited companies. With the estimated coefficients, tax non-compliance probabilities are calculated to the entire universe of taxpayers. The method was employed on a panel of firm-level micro-data from the state of São Paulo VAT tax (ICMS), corresponding to the fiscal region of Guarulhos. Essay 3 revisits the stylized fact of low saving rates in Latin American countries over the last decades. To investigate this situation, it employs panel data techniques to identify savings determinants and per- form counterfactual analysis using China, whose savings rate have been booming in the same period. Special attention is given to Brazil, which has fallen far behind its BRIC peers on this matter. The pa- per contributes to the existing literature in several ways. It combines two different and comprehensive datasets to encompass a vast array of savings determinants, including social security and demographic factors. It restates previous findings in the literature, albeit benefiting from the robustness conferred by richer datasets. For some Latin American countries, it reveals that their saving rates would increase if they perform more like China in other areas, but the increment would not be so dramatic.