Acessos desde 01/11/2007:
Trabalhos Aprovados
Atenção: As comunicações foram provisoriamente aceitas para apresentação no 18o SINAPE. Apenas os trabalhos em que pelo menos um dos autores tiver, até 16/06, pago o boleto de inscrição do congresso terão a aceitação definitiva, os demais não poderão ser apresentados no evento. Verifique se o seu trabalho satisfaz esse requisito.
Filtrar por categoria/formato/palavra:
| Formato | Título | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Pôster | Poder do Teste da Razão de Verossimilhança em Seqüências de DNA Inferência Estatística
Neste trabalho estudamos a distribuição da estatística do teste da razão de máxima verossimilhança que compara diferentes modelos de substituição de bases em seqüências de DNA. Nosso objetivo é a obtenção do poder do teste. Para tanto, devemos considerar a estrutura de dependência imposta pela filogenia que relaciona as seqüências. Assim, propomos um método baseado em simulações de Monte Carlo para obter o poder do teste. A metodologia é aplicada a um conjunto de seqüências de DNA de diferentes espécies de felinos. | ||||||||||||||
| Pôster | POLÍTICA DE MANUTENÇÃO PREVENTIVA ÓTIMA BASEADA NO CUSTO COMO FUNÇÃO DA CONFIABILIDADE DE SISTEMAS Estatística em Engenharia e Ciências Exatas
The main purpose of this paper is to present an alternative approach to define preventive maintenance policies which minimizes maintenance costs. In order to become the maintenance costs’ modeling more realistic, reliability engineering concepts are introduced to the problem. It provides a more realistic quantification of the uncertainty underlying the performance of the system as well as the maintenance costs (modeled as a function of reliability metrics). Block diagrams are adopted in order to model the system reliability and genetic algorithms are considering when addressing the respective mathematical programming problem. Two study cases illustrate the applicability of the proposed approach. Key-words: Reliability, Minimal cost maintenance policies, Genetic Algorithms. | ||||||||||||||
| Pôster | Political Price Cycles in Regulated Industries: Theory end Evidence Estatística em Ciências Sociais Aplicadas (Administração, Economia, Sociologia, Psicologia, etc.)
The relationship between politics and economic policy has a long tradition in economic analysis. One important approach that relates economic policy decisions to the electoral cycle focuses on industry-level issues and treats the regulatory process as the arbitration of conflicting social, economic, and political interests rather than a pure welfare-maximizing effort. This approach originated from the early work of Stigler (1971) and was further developed in Peltzman (1976) with the introduction of a formal model in which the regulated price was chosen so as to maximize political support for the incumbent government-regulator. Another important line of research links macroeconomic policy decisions to the electoral cycle and has become known in the literature as the political business cycle theory. Pioneered by Nordhaus (1975), this theory claims that macroeconomic aggregates - such as the money supply and the government budget - are likely to follow a cycle driven by electoral interests, as incumbent politicians distort economic policy to deceive myopic voters and obtain short-term political gains. Rogoff and Sibert (1988) and Rogoff (1990) refine this approach by characterizing the political business cycle as the equilibrium of a signaling game that arises due to a temporary information asymmetry between government and fully rational voters. This paper follows Paiva (1996) in combining elements of both approaches described above: the idea that policy decisions may change with the proximity of elections is borrowed from the political business cycle theory and added to enhance the traditional, static models of political regulation. More specifically, we combine the main ideas of Peltzman (1976) and Rogoff and Sibert (1988) to model the regulator's problem as a signaling game where politicians set the regulated price trying to maximize electoral support by signaling to voters a pro-consumer behavior. Political incentives and welfare constraints interact in the model yielding an equilibrium in which the real price in a regulated industry falls in periods immediately preceding an election. Besides presenting a new model of political price cycles in regulated industries, this paper also provides empirical support for this theory. Using quarterly data from 32 industrial and developing countries over the period 1978-2004, we test for the existence of political cycle in gasoline prices. The empirical results corroborate the theory. The pioneering work of Stigler (1971) claimed that industry regulation should be viewed as a way to arbitrate among competing interests rather than a way to maximize welfare. Regulators would make their decisions about price, market entry, and other relevant variables under pressure from different interest groups. The regulatory outcome is therefore determined by transactions between self-interested suppliers and demanders. The suppliers are the regulators (with no distinction from the government) that sell wealth transfers in the form of regulated prices and industry entry. The demanders are the interest groups, namely consumers and firms, who bid to be favored in the wealth transfer process. The equilibrium of Stigler's model is one in which, in the words of Peltzman (1993), "cohesive minorities tax diffuse majorities", and the process ends with the capture of the regulatory agency by the firms. This conclusion refers to the fact that consumers have weak incentives to acquire information and actively defend their interests in a particular industry since each individual regulatory decision has only a limited impact on consumers' utility. In contrast, producers have a strong enough interest on the regulatory outcome in their industry so as to form pressure groups to influence the political process. Building on Stigler's work, Peltzman (1976) and Becker (1983) developed models that introduce a broader range of political influence of interest groups. In particular, Peltzman's model formalizes the idea that a regulatory agency chooses a price that maximizes political support for the incumbent government-regulator. Political support depends on the profits of the regulated firm(s) and on consumers' surpluses. Given that profits (consumers' surpluses) increase (decrease) at a decreasing (increasing) rate, the regulated price that maximizes political support will lie between the monopoly and the competitive price. This conclusion differs from the previous work of Stigler in which the regulated price should be close to the monopoly price, reflecting pure industry protection. Paiva (1996) modifies Peltzman model by adding a time dimension to the regulator's maximization problem. The main idea is that regulator's behavior changes with the proximity of elections, securing higher profits and campaign donations from firms in non-election periods and becoming more concerned with consumers' interests in periods immediately preceding elections. In line with Stigler's view of diffused consumer interests, Paiva's model assumes that consumers do not keep track of past prices and that their voting decisions are more affected by current prices and utility. In contrast, firms have a large stake in the regulatory process and have (or acquire) perfect information about past prices. The equilibrium of the model is one in which the solution to the intertemporal maximization problem of the regulator yields lower prices in periods preceding the election, characterizing an electoral cycle in regulated prices. One important weakness of this model is the reliance on myopic consumer-voters or their heavy discounting of past regulatory decisions. In his ground-breaking work on political business cycle, Nordhaus (1975) predicts that governments will inflate the economy in election years in order to exploit a Philips curve trade off that is beneficial in the short but not in the long run. His model assumes that consumers have adaptive expectations about inflation and are myopic in the sense that they partially forget, or heavily discount, the past. Regardless of their party affiliation, politicians turned policy makers are assumed to be opportunistic and only care about being re-elected. The model implies that (1) in an election year the government uses expansionary policies to decrease unemployment; (2) concerns about inflation only arise after the election; and (3) every government follows the same policy. A common criticism of this type of model is that rational voters would understand the motivation for and the consequences of distorting the optimal policy in election years, therefore making politicians less likely to adopt such strategy. However, Tufte (1978), among others, continued to provide empirical evidence of the occurrence of political business cycles and new models were developed to explain the phenomenon. Abandoning the old assumptions of adaptive expectations and myopic voters, Rogoff and Sibert (1988), Rogoff (1990) and Persson and Tabelini (1990) rationalize the political business cycle as the equilibrium of a signaling game originated from a temporary information asymmetry between government and voters. Governments can be differentiated by their level of competency, with more competent governments providing the same amount of services using less resources. Government competency is modeled as a serially correlated stochastic variable that receives a new shock every period. Information asymmetry arises from the fact that the government learns its competency shock before voters do. In election years, the incumbent party has an incentive to act as if it had received a high-competency shock, otherwise voters may prefer an opposition party that can do better in the next period. In equilibrium, most governments will lower non-distortionary taxes (collected before the elections) and increase distortionary taxes (collected after the elections) in election years in order to signal their types. The least efficient type of government is the only one who will not distort the taxation. This paper applies the ideas of Peltzman and Rogof and Sibert to rationalize a possible political cycle in the regulatory process. It re-defines the regulator problem as a preference-signaling game. We combine the main ideas of Rogoff and Sibert (1988) and Peltzman (1976) and model the government-regulator's problem as a signaling game where the cycle is generated by politicians trying to signal pro-consumer behavior to voters when setting the regulated price. We deal with the discrete version of the problem where there are only two possible types of government-regulator: pro-industry and pro-consumer. Contrary to the continuous case developed in Rogoff and Sibert, here we are able to derive a closed-form solution to the model in which regulators' strategies may lead to a political cycle in the regulated price. We model the problem of a government-regulator who sets prices in a regulated industry seeking to maximize an intertemporal objective function which includes social welfare but also the government's chance of being reelected. The social welfare function in the regulated market is a weighted average of consumers' utility and industry profits in which the weights follow a stochastic process. Elections take place every other period. In a non-electoral year the incumbent government-regulator chooses the price that maximizes welfare in the regulated market. In an election year, however, the incumbent may lower the regulated price to increase its chance of being reelected. It is assumed that firms in the regulated industry cannot affect the result of the election. Consumers decide their votes influenced in part by the incumbent's pricing decision and in part by an exogenous shock to their utility function. This shock to consumers' utility could be related to the performance of the government in other areas, for instance. We model the government-regulator's social welfare function as a stochastic process in which the relative weight of consumers' utility and industry profits is determined by a shock occurring at the beginning of every election period. In principle, a shock that determines a higher (lower) weight on firms' profits characterizes a pro-industry (pro-consumer) type of regulator and results in the choice of a higher (lower) regulated price. For instance, a shock that makes the incumbent more pro-industry could be thought of as resulting from (1) a recently formed political coalition that attributes higher importance to securing future investment in the sector; (2) external constraints that limit the country's capacity to import the regulated product, thus requiring an increase in domestic production and the profits to sustain it; or (3) a negative fiscal shock elsewhere that would increase the need for extracting additional income taxes from firms in the regulated industry. The shock that determines whether the incumbent government-regulator is pro-consumer or pro-industry is not observed by the voters until after the election. This temporary information asymmetry makes it possible for a pro-industry regulator to try and mimic the behavior of a pro-consumer type by setting a lower price and increasing its chances of winning the election. However, since deviations from the welfare-maximizing price generate a decrease in welfare at increasing rates, pro-consumer regulators can achieve a lower price than pro-industry regulators for any given level of welfare. Therefore, in equilibrium, the pro-consumer regulator will set a price unachievable by the pro-industry type in order to unequivocally signal its type to consumer-voters. Unable to match this lower price, pro-industry regulators have no reason to distort the regulated price in equilibrium. The order of events is what drives the cycle. At the beginning of the period $t$ the incumbent receives its preference shock and sets the regulated price. Afterward, the election happens, the winner is announced, and the period ends. At the beginning of the following period $t+1$, the winner of the election takes office until the end of the next period $t+2$. Note that a new preference shock will define a new orientation for the incumbent at the beginning of period t+2, and another election will then take place. The incumbent type is revealed to the public at period t+1, since the regulator will set the post-election price that maximize its social welfare function, with no electoral concerns. | ||||||||||||||
| Oral | Power Series Generalized Linear Models Modelos de Regressão
We introduce in this paper a new class of discrete generalized linear models to extend the binomial, Poisson and negative binomial models to cope with count data. This class of models includes some important models such as log-linear models, logit, probit and negative binomial models, generalized Poisson and generalized binomial regression models, among other models, which enables the fitting of a wide range of models to count data. We derive an iterative process for fitting these models by maximum likelihood and discuss inference on the parameters. The usefulness of the new class of models is illustrated with an application to a real data set. | ||||||||||||||
| Pôster | Predição de Valores Latentes com Erros de Medida Endógeno e Exógeno Modelos de Regressão
Obtemos um preditor do valor latente das unidades amostrais selecionadas de uma população finita na presença de erros de medida: endógenos (atribuíveis à variabilidade natural da resposta) e exógenos (atribuíveis às condições de medição). | ||||||||||||||
| Pôster | Prediction Intervals for Artificial Neural Networks Modelos de Regressão
This article describes the application of a nonlinear regression technique (least squares) to create prediction intervals on artificial neural networks (ANN´s). Through Monte Carlo’s simulations it is shown a way of choosing the set of parameters (weights) to a Neural Network, according to a selection criteria based on the magnitude of the prediction intervals provided by the net. With this technique it is possible to obtain the prediction intervals with the desired amplitude and with known coverage probability (test points that fell within the interval), according to the chosen confidence level. The associated results and discussions indicate to be possible and feasible to obtain these intervals, thus making the network response more informative and consequently increasing its applicability. | ||||||||||||||
| Pôster | Presença da Estatística no Ensino Fundamental e Médio Educação Estatística
O desenvolvimento de tópicos relacionados a Estatística estão previstos nos Parâmetros Curriculares Nacionais em várias séries da Educação Básica. Em geral, os professores de Matemática seriam os responsáveis pela introdução desses tópicos em suas aulas. Como a formação dos professores é bastante diferenciada, a presença da Estatística não tem um local consolidado na grade curricular e, não raras vezes, é simplesmente ignorada. O objetivo deste artigo é relatar os resultados parciais de um projeto de avaliação sobre a presença de Estatística no ambiente do Ensino Fundamental (5a a 8a) e Médio. Em particular, nosso interesse se concentrou em escolas estaduais da capital de São Paulo. | ||||||||||||||
| Pôster | Previsão da Margem de Contribuição Financeira do Cliente na Indústria de Cartões de Crédito do Brasil por meio de Modelos Lineares Hierárquicos Estatística em Ciências Sociais Aplicadas (Administração, Economia, Sociologia, Psicologia, etc.)
O modelo do Valor do Cliente e o cálculo do Valor Vitalício do Cliente vêm ganhando destaque como referência entre os modelos para a tomada de decisão dos investimentos de marketing. Apesar de na literatura acadêmica existirem diversos artigos descrevendo o cálculo do Valor do Cliente, poucos sugerem técnicas estatísticas que poderiam ser úteis para este fim. O modelo de cálculo usado como referência neste artigo combina as técnicas de análise de sobrevivência com modelos lineares hierárquicos. Neste trabalho será demonstrado como os Modelos Lineares Hierárquicos podem ser utilizados como uma importante ferramenta para predizer a margem de contribuição gerada por um cliente. A técnica permite levar em consideração características individuais dos clientes e a tendência da margem de cada cliente ao longo do tempo. Como ilustração, a técnica será aplicada à Unicard, uma das cinco principais administradoras do Brasil. Vale observar que o mesmo modelo poderia ser aplicado em outras indústrias com características semelhantes, como telefonia, provedores de internet, bancos e financeiras. | ||||||||||||||
| Pôster | Previsão da Produção Industrial do Refino de Petróleo e Álcool no Brasil: Uma abordagem comparativa entre Holt-Winters e Box & Jenkins Estatística em Meio Ambiente
Este estudo tem por objetivo traçar um paralelo entre o Desenvolvimento Sustentável e o refinamento de álcool e petróleo através da série temporal que o Instituto Brasileiro de Geografia e Estatística (IBGE) divulga mensalmente. Dada a importância energética que estes dois produtos proporcionam ao Brasil, bem como outros países, faz-se necessário um estudo que tente prever a realidade acerca de seu refinamento para o final ano de 2008 e início de 2009 à luz da conceitualização de desenvolvimento sustentável. Neste trabalho, abordaremos uma comparação entre dois métodos para as estimativas, a clássica metodologia de Box & Jenkins e o método de Holt Winters ambos com ajustes feitos pelo software R versão 2.6.2. | ||||||||||||||
| Pôster | Previsão dos preços das Culturas de Aveia e Azevém praticados no Rio Grande do Sul. Iniciação Científica e Concurso de Iniciação Científica
The main purpose of this research is to determine the level of price of the azevem and oats that are used in the field of Rio Grande do Sul, by means of ARIMA models in the way to supply the farmers with an estimation of costs to plant this kind of pasture. The model SARIMA (0,1,1)(0,1,1 represents the azevem pasture and the SARIMA (2,1,0)(0,1,1)12 represents the oats pasture. The price levels are similar with that they are practiced in the region. | ||||||||||||||
| 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 | |||||||||||||||