In recent years, corporate strategy has drawn a lot of attention in the academic an practice. However, there are fewer literatures on how to put these ideas into practice, that is, how to quantify the interrelationships between the three key elements in strategic management¡Ðperformance, strategies, and environments, and how to judge and measure the opportunities and threats (O & T) when the environments change. This study is an attempt to answer these questions. The theoretical method developed incorporates a dynamic simultaneous equations model to express the interrelationship between these three elements. The method requires the identification of O & T in a three-step procedure. Step 1 relates the strategic components to the performance measures by the management¡¦s concept of business and philosophy of resource depolyment. Step 2 points out the suitable (unsuitable) environment circumstances for each of the scope and resource deployment elements. In Step 3, we link the results of Step 1 and Step 2 to identify and measure O & T. The above methodology is applied to the case of Cathy Financial Holding Company, a Taiwan largest listed financial holding company, over the period 2002Q2-2007Q3. We use the Instrument Variables Three Stage Least Square Method (IV-3SLS) to estimate them. In addition, we also use some tests to ascertain the validity of the selected instrument variables in order to obtain the more reliable results. Our empirical results indicate that both the firm strategies and the environments play significant roles in influencing the firm¡¦s performance. More specifically, whereas the diversification of products, and the debt allowance reservation rate are negatively associated with the cost/income ratio and positively associated with adjusted ROE and Tobin¡¦s Q. Additionally, the managers also can increase the investment efficiency by adjusting the content of the asset allocation, especially with regard to the holding of bonds. We also extract some major environment factors such as unemployment rate that affect the firm¡¦s performance and use the estimated results to identify and measure O & T.

Alcohol consumption may affect labor market outcomes directly through a reduction in productivity and indirectly through human capital accumulation. However, empirical results from previous studies in the economics literature are mixed and inconclusive. While some researchers found negative effects of alcohol use on labor market outcomes, quite a few studies found either positive or insignificant effects. The purpose of this dissertation is to estimate causal effects of alcohol consumption on employment status. It uses three data sets previously unexploited for this purpose and attempts to eliminate any potential estimation problems from previous studies. The results show that previous problematic heavy drinking, i.e. clinically-defined alcohol abuse and/or dependence, has no significant direct effects, but has significant indirect effects on current employment propensity for both genders through human capital components, specifically educational attainment and health status. While general alcohol consumption has only an indirect effect on employment status for females, it has both direct and indirect effects on employment status for males, though the direct effect is very small.

In this thesis, we analyse the relationship between the hedging activities and return on equity, and the relationship between profit on hedging and other factors. Fully conditional specification is used to impute the missing values. Instrumental variable estimation and finite mixture of regression models are then used to predict the return on equity and hedging gain. We find the instrumental variable estimation is better than the OLS estimation to deal with the hedging data since it eliminates the endogeneity. By finite mixture of regression models, we show that different firms have different hedging strategies, which cause different profits in hedging. We also find the companies with large total assets prefer to hedge.

This dissertation both theoretically and empirically examines the relationship between health and wealth, using proxies for health and wealth that are standard in the economics literature. We first model the endogenous interactions between life expectancy and income by modifying a standard overlapping generation model to allow individuals to directly choose their own longevity. The model displays a positive feedback between life expectancy and income that generates multiple stable equilibria. The worse equilibrium is a “poverty-trap” in which poverty and low longevity reinforce each other. The second portion of the dissertation is empirical. We first show that income has statistically significant effects on various proxies for health. The results are robust to different ways of controlling for the endogeneity of income: both instrumental variable estimation with external instruments and also generalized method of moments estimation when internal instruments are applied. We next directly test for the causal relationship between income and various proxies for health using three panel Granger causality tests. Evidence is found to support the existence of a bi-directional causal link. Sensitivity tests further suggest that middle-income countries play a more important role than low-income countries in explaining the overall wealth-health causality.

This dissertation both theoretically and empirically examines the relationship between health and wealth, using proxies for health and wealth that are standard in the economics literature. We first model the endogenous interactions between life expectancy and income by modifying a standard overlapping generation model to allow individuals to directly choose their own longevity. The model displays a positive feedback between life expectancy and income that generates multiple stable equilibria. The worse equilibrium is a “poverty-trap” in which poverty and low longevity reinforce each other. The second portion of the dissertation is empirical. We first show that income has statistically significant effects on various proxies for health. The results are robust to different ways of controlling for the endogeneity of income: both instrumental variable estimation with external instruments and also generalized method of moments estimation when internal instruments are applied. We next directly test for the causal relationship between income and various proxies for health using three panel Granger causality tests. Evidence is found to support the existence of a bi-directional causal link. Sensitivity tests further suggest that middle-income countries play a more important role than low-income countries in explaining the overall wealth-health causality.

In modern manufacturing, industrial robots are essential components that allow saving cost, increase quality and productivity for instance. To achieve such goals, high accuracy and speed are simultaneously required. The design of control laws compliant with such requirements demands high-fidelity mathematical models of those robots. For this purpose, dynamic models are built from experimental data. The main objective of this thesis is thus to provide robotic engineers with automatic tools for identifying dynamic models of industrial robot arms. To achieve this aim, a comparative analysis of the existing methods dealing with robot identification is made. That allows discerning the advantages and the limitations of each method. From those observations, contributions are presented on three axes. First, the study focuses on the estimation of the joint velocities and accelerations from the measured position, which is required for the model construction. The usual method is based on a home-made prefiltering process that needs a reliable knowledge of the system’s bandwidths, whereas the system is still unknown. To overcome this dilemma, we propose a method able to estimate the joint derivatives automatically, without any setting from the user. The second axis is dedicated to the identification of the controller. For the vast majority of the method its knowledge is indeed required. Unfortunately, for copyright reasons, that is not always available to the user. To deal with this issue, two methods are suggested. Their basic philosophy is to identify the control law in a first step before identifying the dynamic model of the robot in a second one. The first method consists in identifying the control law in a parametric way, whereas the second one relies on a non-parametric identification. Finally, the third axis deals with the home-made setting of the decimate filter. The identification of the noise filter is introduced similarly to methods developed in the system identification community. This allows estimating automatically the dynamic parameters with low covariance and it brings some information about the noise circulation through the closed-loop system. All the proposed methodologies are validated on an industrial robot with 6 degrees of freedom. Perspectives are outlined for future developments on robotic systems identification and other complex problems.

abstract: Total digital media advertising spending of $72.5 billion surpassed total television Ad spending of $71.3 billion for the first time ever in 2016. Approximately $39 billion, or 54% of the digital media advertising spend, involved pre-programmed software that purchased Ads on behalf of a buyer in Real-Time Bidding (RTB) settings. A major concern for Ad buyers is sub-optimal spending in RTB settings owing to biases in the attribution of customer conversions to Ad impressions. The purpose of this research is twofold. First, identify and propose a novel experimental design and analysis plan for to handling a previously unidentified and unaddressed source of endogeneity: count/quality simultaneity bias (CQB). Second, conduct a field study using data for Ad response rates, cost, and observed consumer behavior to solve for the profit maximizing daily Ad frequency per customer. One large online retailer provided data for Ad impressions, bid costs, response rates, revenue per visit, and operating costs for 153,561 unique users over 23 days. Unique visitors were randomly assigned to one of seven treatment groups with one, two, three, four, five, and six impressions per day limits as well as a final condition with no daily impression cap. Ordinary least square models (OLS) were fit to the data and a non-linear relationship between Ad impressions and site visits demonstrating declining marginal effect of Ad impression on site visits after an optimal point. The results of the field study confirmed the existence of negative CQB and demonstrated how my novel experimental design and analysis can reduce the negative bias in the estimate of impression quantity on customer response. Second, managers interested in improving the efficiency of advertising spend should restrict display advertising to only the highest quality inventory through specific site targeting and by leveraging direct buys and private marketplace deals. This strategy ensures that subsequent impressions are not of lower quality by restricting the pool of possible impressions from a homogenous set of high quality inventory. / Dissertation/Thesis / Doctoral Dissertation Business Administration 2017

nÃo hà / Considerando a importÃncia dos investimentos em educaÃÃo realizados pelos 184 municÃpios cearenses, investigou-se o impacto desses gastos sobre as notas mÃdias padronizadas dos alunos das escolas pÃblicas municipais cearenses, para dois grupos de anÃlise, um para os alunos dos anos iniciais (4 sÃrie/5 ano) e outro para os alunos dos anos finais (8 sÃrie/9 ano), referentes a 2007, 2009, 2011 e 2013. A anÃlise descritiva demonstrou que as notas mÃdias padronizadas dos alunos dos anos iniciais e finais apresentam uma tendÃncia crescente nos perÃodos analisados, acompanhada por uma tendÃncia crescente nos gastos com educaÃÃo realizados pelos municÃpios cearenses. Partindo dessa evidÃncia, foram realizadas estimaÃÃes, utilizando-se variÃveis instrumentais e dados em painel para efeitos fixos, tendo como instrumento o Ãndice de EducaÃÃo para distribuiÃÃo da Cota Parte do ICMS, apresentando efeitos positivos e significantes dos investimentos em educaÃÃo sobre as notas mÃdias padronizadas da Prova Brasil de todos os grupos analisados. / Considering the importance of investments in education performed by 184 municipalities in CearÃ, this study investigated the impact of education expenditures on the standardized average scores for two groups of students enrolled in public schools, one group of students in the early years (4th and 5th grades) and another group of students in the final years (8th and 9th grades), during 2007, 2009, 2011 and 2013. Descriptive statistics for the standardized average of both early and final years students have shown an upward trend during the period analyzed, followed by a growing trend in education spending made by municipalities in CearÃ. Based on this evidence, estimations based on instrumental variable and fixed effect panel data were conducted, using an education index as instrument, which presented a positive and significant effect of investments in education on the standardized average scores of the Prova Brasil for both groups.

O total de gasto social do governo somou cerca de R$800 bilhões em 2009, quase 25% do PIB brasileiro. Dado este volume crescente e expressivo, esta dissertação apresenta uma análise do efeito do gasto social sobre a pobreza no Brasil das três instâncias de governo: federal, estadual e municipal. Para tanto, foram estimados três modelos com dados em painel de 20 anos (1987 a 2009) analisando os gastos estaduais e federais. Adicionalmente, estimou-se o efeito da despesa municipal utilizando dados de 1991, 2000 e 2010 para 5.058 municípios. Procurou-se tratar o viés de simultaneidade entre o gasto do governo e a pobreza aplicando-se o modelo de mínimos quadrados em dois estágios, utilizando variáveis de ideologia política como instrumento. Os resultados mostraram que o gasto das três instâncias de governo tem efeito sobre a queda da pobreza no Brasil. Particularmente, as funções orçamentárias de gasto que apresentaram maior efetividade sobre a queda da pobreza foram: gasto federal e municipal com saúde e saneamento e gastos agregados estaduais e municipais. Analisou-se também o efeito da interação do gasto estadual com dados das condições iniciais de cada estado em 1980 - renda familiar per capita, desigualdade, proporção de pobres, grau de educação e mortalidade infantil. Concluiu-se que tanto as condições iniciais de cada estado como suas características específicas influenciam o grau em que o seu gasto afeta a pobreza. / The total public social spending in Brazil reached about R$800 billion in 2009, almost 25% of Brazilian GDP. Giving this increasing and expressive volume, this dissertation analyzes the effect of the three levels of government social spending on poverty in Brazil - federal, state and municipal. Three models were estimated with a 20 year\'s state panel data (1987 to 2009) analyzing the federal and state spending effect. We also estimated the municipal expenditure effect on poverty using data from 1991, 2000 and 2010 in 5.058 municipalities. The simultaneity bias between government spending and poverty was treated by applying the two stages least squares method, using the political ideology as instrumental variables. The results showed that the spending of the three levels of government reduce poverty in Brazil. Particularly, the spending budget functions with higher effectiveness on reduce poverty are: federal and municipal spending on health and sanitation, state and local aggregated spending. In addition, we also analyzed the effect of the state spending data interacting with its initial conditions in 1980 - per capita income, inequality, poverty, level of education and infant mortality. We concluded that the states initial conditions as well as its specific characteristics influence the extent to which their spending affects poverty.

Submitted by isabela.moljf@hotmail.com (isabela.moljf@hotmail.com) on 2017-06-29T14:40:18Z No. of bitstreams: 1 filiperodriguesvianna.pdf: 2477695 bytes, checksum: 7e55b9eb9b2d53b58522606533f2a699 (MD5) / Approved for entry into archive by Adriana Oliveira (adriana.oliveira@ufjf.edu.br) on 2017-08-08T12:02:06Z (GMT) No. of bitstreams: 1 filiperodriguesvianna.pdf: 2477695 bytes, checksum: 7e55b9eb9b2d53b58522606533f2a699 (MD5) / Made available in DSpace on 2017-08-08T12:02:06Z (GMT). No. of bitstreams: 1 filiperodriguesvianna.pdf: 2477695 bytes, checksum: 7e55b9eb9b2d53b58522606533f2a699 (MD5) Previous issue date: 2017-04-18 / Essa dissertação tem o objetivo de identificar e mensurar o efeito dos pares sobre o desempenho escolar, medido pela proficiência em testes de Português e Matemática, de alunos de 1ª à 4ª série do Ensino Fundamental, de escolas públicas e privadas participantes do Estudo Longitudinal Geração Escolar (GERES, 2005-2008). A base de dados do GERES acompanha alunos ao longo da primeira etapa do ensino fundamental e fornece, além das informações dos alunos, informações sobre escolas, professores e diretores. O estudo parte de um modelo linear onde o desempenho escolar dos alunos é explicado pelo desempenho médio de seu grupo de pares, além de características próprias e pela média das características de seu grupo. O método aplicado, proposto por Lee (2007) utiliza variações no tamanho dos grupos e desvios das características em relação à média para a identificação dos efeitos. Os resultados são obtidos pelo estimador de Mínimos Quadrados em Dois Estágios (2SLS), utilizando os instrumentos propostos por Bramoullé, Djebbari e Fortin (2009). As regressões foram feitas para cada etapa do GERES, com o intuito de verificar a evolução dos coeficientes ao longo do tempo. Além disso, comparam-se os resultados obtidos com o uso de uma amostra de dados em painel balanceada e uma base de dados não balanceada, em que se aproveita informações de alunos com informações faltantes através de uma correção proposta por Davezies et al. (2009). Seja usando a amostra balanceada, seja usando a amostra não balanceada, os resultados indicam que os alunos são negativamente impactados pelos pares que possuem desempenho escolar superior. As características contextuais do grupo de pares também exercem influência sobre o desempenho escolar, sendo o nível socioeconômico e o gênero dos alunos as características mais importantes. / This dissertation aims to identify and measure the peer effects on scholar performance of students, measured by proficiency in Portuguese and Mathematics tests, from 1st to 4th grades of public and private schools who attended the Estudo Longitudinal Geração Escolar (GERES, 2005-2008). The GERES database follows students throughout the first stage of elementary school and provides, apart from the students' information, data about schools, teachers and principals. The study starts from a linear model where the students' performance is explained by the average performance of his/her peer group, as well as his/her own characteristics and average characteristics of his/her group. The applied method - proposed by Lee (2007) - uses variations in the size of the groups and deviations of the characteristics from the mean as well, in order to identify the peer effects. The coefficients are estimated through the Two-Stage Least Squares (2SLS) estimator, using the instruments proposed by Bramoullé, Djebbari and Fortin (2009). The regressions consider each wave of the GERES to verify the change of the coefficients over time. Further, we compare the results between a balanced database with an unbalanced database, where the correction method follows Davezies et al. (2009). In both samples, the results show that the average performance of the peers negatively impacts student’s performances. The background characteristics of the peer group also influence school performance, specifically the socioeconomic level and the gender of the students.