Sob a teoria de agência e o contexto de gerenciamento de resultados, o objetivo do trabalho foi estudar a relação entre o gerenciamento de resultados (GR) e o custo da dívida (Ki) das empresas brasileiras de capital aberto, no período de 1996 a 2007. Para isso, foram levantadas duas hipóteses de pesquisa quanto à possível relação entre GR e Ki: H1- as variáveis são inter-relacionadas; e H2- a relação é unidirecional, em que o Ki é influenciado positivamente pelo GR. Os pressupostos levantados, por meio da literatura existente, para a formulação da primeira hipótese são de que, almejando melhores condições contratuais, como o custo da dívida, o custo atual ou passado motiva a empresa a gerenciar o resultado corrente para obter melhores condições de custo futuro ou corrente. Dessa forma, as empresas, intencionalmente, procurariam gerenciar seus resultados, motivadas pelo custo da dívida. Todavia, há suporte teórico para que a relação entre essas variáveis seja unidirecional, com o gerenciamento de resultados influenciando o custo da dívida, sustentando a segunda hipótese. Um por exemplo disso é o gerenciamento de resultados, que é percebido como fator de assimetria de informação pelos credores, os quais, ao identificarem o gerenciamento, exigirão um prêmio maior pelo risco da falta de informação, elevando o custo da dívida das empresas. Assim, para testar as hipóteses do trabalho, a proxy de GR foi definida com o uso do modelo para cálculo dos accruals discricionários proposto por Kang e Sivaramakrishnan (1995) e a variável representativa do custo da dívida foi a relação entre a despesa financeira e o passivo oneroso total. A análise inicial foi feita utilizando-se o índice de correlação de Spearman e análise de média, por meio do teste não-paramétrico U de Mann-Whitney. Em seguida, utilizou-se o teste de especificação de Hausman para análise de simultaneidade, seguido da análise de regressão com dados dispostos em painel com correção de Newey-West, correção por cluster - ambas nas situações em que havia heterocedasticidade e correlação serial-, e estimador MQ2E. Os testes de correlação e de média apontaram para uma relação positiva entre GR e Ki, mas o teste de simultaneidade não indicou haver inter-relação entre as variáveis. Os resultados das regressões em painel indicaram que o Ki, atual ou passado, não influencia o GR, mas apontam para uma relação positiva, em que quanto maior o gerenciamento de resultados, maior o custo da dívida, ou seja, este é influenciado por aquele. Desse modo, os resultados permitem rejeitar a hipótese 1, mas confirmam a segunda hipótese. A princípio, tais resultados indicam que não há uma inter-relação entre GR e Ki e que as empresas não gerenciam com a intenção de impactar no custo da dívida, mas que o gerenciamento de resultados pode ter outras motivações para as empresas brasileiras. Todavia pode ser percebido pelos credores como maneira de reduzir a transparência das demonstrações contábeis, levando-os a cobrar maiores taxas de empresas com gerenciamento de resultados, gerando maior custo da dívida para as companhias. / Under the theory of agency and the context of earnings management, the objective of this research was to study the relationship between the earnings management (EM) and the cost of debt (Ki) of Brazilian open-capital companies, from 1996 to 2007. For this, two hypotheses of research have been raised as to the possible relationship between GR and Ki: H1: the variables are interrelated and H2: the relation is unidirectional where Ki is influenced positively by the GR. The hypothesis raised, by means of existing literature, to elaborate the first hypothesis are that, longing for better contractual conditions, such as the cost of the debt, the current or past costs motivate the company to manage the current result to get better conditions for future or current costs. Therefore, the companies, intentionally, would manage their results, motivated by the cost of debt. However, there is theoretical support that relations between these variables are unidirectional, with the earnings management influencing the cost of debt, supporting the second hypothesis. For example, the earnings management is perceived as a factor of asymmetric information by the creditors, who will demand a bigger prize for the risk regarding the lack of information, which raises the cost of debt of the companies. Thus, to test the hypotheses of this research, the proxy of GR was defined with the use of the model for calculation of discretionary accruals considered by Kang and Sivaramakrishnan (1995) and the representative variable of the cost of debt was the relationship between the financial expense and the total financial liability. The initial analysis was made using the index of correlation of Spearman and analysis of average, by means of non-parametric test U of Mann-Whitney. After that the test of specification of Hausman was used, for analysis of simultaneity, followed by the analysis of regression with data disposed on panel corrected by Newey-West, correction by cluster - both under the condition where there were heteroskedasticity and serial correlation, and estimator MQ2E. The average and correlation tests pointed out a positive relation between GR and Ki, but the concurrence test did not indicate to have interrelation between the variables. The results of the regressions in the panel indicated that the Ki, current or past, does not influence the GR, but point a positive relationship, where the bigger the earnings management, greater is the cost of debt, that is, this is influenced by management. Then, the results allow rejecting the first hypothesis, confirming the second hypothesis. Such results indicate that there is not an interrelation between GR and Ki and that the companies do not manage with the intention of impacting the cost of the debt, but that the earnings management can have other motivations for the Brazilian companies, however it can be perceived, by the creditors, as a way to reduce the transparency of the accounting demonstrations, leading to charge greater fees from companies with earnings management, generating bigger cost of debt for the companies.

Thesis advisor: Thomas J. Chemmanur / Thesis advisor: Arthur Lewbel / The dissertation aims to investigate the role of asymmetric information in capital structure, investment, compensation of mortgage servicers, and bond and equity returns. Specifically, I evaluate the impact of credit ratings on debt issuance and investment of private and public firms, as well as the effect of asymmetric information on compensation of loan servicers in the mortgage backed securities market. Further, I study the relationship between ratings issued by investor and issuer-paid credit rating agencies and equity analyst recommendations. Finally, I evaluate the effect of the aforementioned signals on bond and equity returns as well as firm leverage and investment decisions. Chapter one in the dissertation is the first study to empirically evaluate the effect of credit ratings on capital structure and investment for private U.S. firms, relative to equivalent public firms. I find that private firms constrain debt issuance and investment by 4.5 and 6.5 percentage points more than public firms, respectively, when their credit ratings are on upgrade or downgrade thresholds. Consistent with these results, private firms that become public through an IPO constrain debt issuance by 10 percentage points before going public, if their ratings are on an upgrade or downgrade boundary. The second chapter studies the impact of asymmetric information between mortgage sellers and servicers on mortgage servicer compensation. We proxy for asymmetric information using the decision to retain mortgage servicing rights, which creates a principal-agent problem between sellers and servicers. Using loan-level data on Fannie Mae-insured, full documentation mortgages, we first find that loans in which sellers retain servicing rights default and foreclose at a significantly lower rate, and lose less in foreclosure than those in which they are not retained. Since it is more costly to service non-performing loans, these ex-post differences in default rates should be reflected in servicer compensation. However, using Fannie Mae MBS pool-level data, we find no difference in servicing fees for pools in which servicing rights are retained relative to pools in which they are not retained. In order to identify the impact of seller/servicer affiliation on servicing fees, we exploit a post-crisis regulatory change which altered the incentive to retain servicing rights for small sellers of MBS relative to large sellers. Finally, in the third chapter, we evaluate the information flows to the stock and bond markets of issuer versus investor-paid rating agencies and equity analysts. Equity analysts' forecasts and ratings assigned by issuer-paid credit rating agencies such as Standard and Poor's (S&P) and by investor-paid rating agencies such as Egan and Jones (EJR) all involve information production about the same underlying set of firms, even though equity analysts focus on cash flows to equity and bond ratings focus on cash flows to bonds. Further, the two types of credit rating agencies differ in their incentives to produce and report accurate information signals. Given this setting, we empirically analyze the timeliness and accuracy of the information signals provided by each of the above three types of financial intermediary to their investor clienteles and the information flows between these intermediaries. We find that the information signals produced by EJR are the most timely (on average), and seem to anticipate the information signals produced by equity analysts as well as by S&P. We find that changes in leverage are associated with lower EJR ratings but higher equity analyst recommendations; further, credit rating changes by EJR have the largest impact on firms' investment levels. We also document an "investor attention" effect (in the sense of Merton, 1987) among stock and bond market investors in the sense that changes in equity analyst recommendations have a higher impact than either EJR or S&P ratings changes on the excess returns on firm equity, while EJR rating changes have a higher impact on bond yield spreads than either S&P ratings changes or changes in equity analyst recommendations. Finally, we analyze differences in bond ratings assigned to a given firm by EJR and S&P, and find that these differences are positively related to the standard proxies for disagreement among stock market investors.

In the first essay, "Why Won't You Forgive Me? Evidence of a Financial Misreporting Stigma in Bank Loan Pricing," we examine the relation between bank loan pricing and intentional financial misreporting. Firms that misreport financial information pay greater spreads on their bank loans for five years following their restatements, whether benchmarked against their pre-restatement loans or similar loans made to matched non-misreporting firms. Misreporting firms that promptly replace certain parties who are potentially related to the misreporting see their spreads fall to benchmark levels within three years following restatement. Large fractions of firms, however, do not promptly replace the potentially related parties and continue to pay premiums over benchmark spread levels for five years following restatement. The results suggest that misreporting creates a long-lasting and costly stigma, but that certain actions can reduce the duration of the stigma. In the second essay, "Can Shareholder-Creditor Conflicts Explain Weak Governance? Evidence from the Value of Cash Holdings," we look into whether shareholder-creditor conflicts generate costs large enough to prevent improvements in governance. If firms choose to remain weakly governed, some cost must prevent improvements. We address our research question by estimating the value of cash as a function of governance, leverage, and the interaction of the two. We find that governance increases the value of cash, but that leverage reduces the gain from strong governance. However, the magnitudes are far too small to explain why weak governance firms remain weakly governed. Our estimates suggest more than 80 percent of weakly governed firms would increase the value of their cash by improving governance. In fact, half of weakly governed firms would increase the value of their cash holdings by $0.35 or more per dollar held by improving governance. Our focus on cash holdings does not seem to drive our results, nor do endogenous governance choices or nonlinearities reverse our conclusions.

<p>The costs and constraints to financing, and the factors that influence them, play critical roles in the determination of corporate capital structures.</p> <p>Chapter 1 estimates firm-specific marginal cost of debt functions for a large panel of companies between 1980 and 2007. The marginal cost curves are identified by exogenous variation in the marginal tax benefits of debt. The location of a given company's cost of debt function varies with characteristics such as asset collateral, size, book-to-market, intangible assets, cash flows, and whether the firm pays dividends. Quantifying, the total cost of debt is on average 7.9% of asset value at observed levels, reaching as high as 17.8%. Expected default costs constitute approximately half of the total ex ante cost of debt.</p> <p>Chapter 2 uses the intersection between marginal cost of debt functions and marginal benefit of debt functions to examine optimal capital structure. By integrating the area between benefit and cost functions, net benefit of debt at equilibrium levels of leverage is calculated to be 3.5% of asset value, resulting from an estimated gross benefit of debt of 10.4% of asset value and an estimated cost of debt of 6.9%. Furthermore, the cost of being overlevered is asymmetrically higher than the cost of being underlevered. Case studies of several firms reveal that, for some firms, the cost of being suboptimally levered is small while, for other firms, this cost is large, suggesting firms face differing sensitivities to the capital structure choice.</p> <p>Finally, Chapter 3 examines the role of financing constraints on intertemporal capital structure choices of the firm via a structural model of capital investment. In the model, firms maximize value by choosing the amount of capital to invest and the amount of debt to issue. Firms face a dividend non-negativity constraint that restricts them from issuing equity and a debt capacity constraint that restricts them from issuing non-secured debt. The Lagrange multipliers on the two constraints capture the shadow values of being constrained from equity and debt financing, respectively. The two financing constraint measures are parameterized using firm characteristics and are estimated using GMM. The results indicate that these measures capture observed corporate financing behaviors and describe financially constrained firms. Finally, between the two financing constraints, the limiting constraint is the debt restriction, suggesting that firms care about preserving financial slack.</p> / Dissertation

I examine how political geography affects firms' cost of debt. Policy risk, measured by proximity to political power reflected in firms' position in the country's political map, is negatively related to corporate bond ratings and positively related to firms' cost of debt. I find firms' policy risk can be mitigated by engaging in corporate political strategies like making campaign contributions or lobbying. Consistent with the view that such political strategies effectively protect firms against uncertainty about future policies, I find policy risk has less of an impact on the cost of debt of firms that support more powerful and well-connected politicians in the legislative co-sponsorship network or that spend more money on lobbying. Using a sample of state pension funds' equity holdings, I find that state pension funds exhibit not only local bias but also bias towards politically connected stocks. These politically connected local firms held by state pension funds do not exhibit better performance compared with their local benchmarks not held by these funds before the holding period, and the overweighting of politically connected local firms is negatively related to pension fund returns. My results do not support the information advantage hypothesis that state pension funds exhibit overweighting of local firms because they have an information advantage about home-state firms. I further examine the factors that explain local bias from political perspectives. My results show that local bias is related to public policy integrity and local politicians' congressional connections.

Cost of debt is a key cognitive anchor for managerial decisions and an important determinant of firm profitability. We extend international management research by analyzing the effects of institutional distance, institutional quality, and their dynamics on the cost of debt in the context of foreign direct investments (FDI). We test our conceptual model on a sample of companies making 3,764 greenfield foreign direct investments from developed into less developed markets. Using hierarchical linear modelling, we show that the financial consequences of internationalizing into countries with weak institutions depend on both the institutional distance between countries, as well as their institutional quality. Furthermore, we find that recent changes in institutional quality form expectations about future development and ultimately influence post investment financing costs.

O presente estudo tem como objetivo verificar se há diferença significativa entre o custo do capital de terceiros (Ki) apurado com base nas Demonstrações Contábeis (DC) e, a partir das informações contidas nas Notas Explicativas (NE) das empresas do segmento Novo Mercado de Governança Corporativa (NMGC) da Bovespa, para o período de 2001 a 2005, além de identificar os fatores que as determinam. Faz-se, assim, uma comparação entre duas formas de apuração do Ki, uma que já é utilizada e outra proposta neste estudo. O segmento NMGC foi escolhido porque as empresas que o compõem comprometem-se a melhorar a qualidade das informações prestadas aos stakeholders. No Teste de Hipóteses realizado, H0 indica que não há diferenças significativas entre as duas formas de apuração; e H1, que há diferenças significativas entre elas. O Teste de Postos com Sinais de Wilcoxon indicou que há evidência insuficiente contra a hipótese nula, ou seja, não há evidência amostral para afirmar que há diferença entre as duas formas de apuração, (&#945; = 5%). Apesar disso, a análise gráfica permite levantar algumas suposições acerca das divergências visualizadas: 1 - lançamentos herméticos das receitas e despesas financeiras; 2 - valores operacionais lançados nas despesas financeiras com a permissividade da lei; e 3 - a presença de instrumentos financeiros que interferem no total das despesas financeiras. Contudo, a apuração do Ki pelas NE apresenta-se mais satisfatória que pelas DC, pois eliminam-se as deficiências levantadas, e fornecem aos usuários informações compreensíveis, relevantes, confiáveis, comparáveis e sobretudo, úteis para a tomada de decisões. / The aim of this study is to verify whether there is any significant difference between the cost of debt (Ki), which was assessed based on Accounting Statements (AS), and that from the information contained in Explanatory Notes (EN) of the companies in the segment New Market of Corporate Governance (NMGC) of Bovespa, between 2001 and 2005. In addition, it also identifies the factors that determine this difference. Therefore, we are able to compare both methods of assessing Ki, one which is already used and another which is proposed in this study. The segment NMGC was chosen because the companies that constitute it are committed to improving the quality of information given to stakeholders. In the Hypothesis Test carried out, H0 indicates that there are no significant differences between the two assessment methods; on the other hand, H1 indicates that there are significant differences between them. The Wilcoxon Sign Rank Test showed that there is not enough evidence against the null hypothesis, that is, there is no evidence in the sample in order to state that there is any difference between the two ways of assessing (&#945; = 5%). Despite that, the graphic analysis allows us to make some assumptions about the divergences that were brought up: 1 - hermetic revenue entries; 2 - operational values released on financial expenditure with permission from the law; and 3 - the presence of financial instruments which interfere in the total financial expenditures. However, assessing Ki through NE is more satisfactory than through DC, because the deficiencies raised are eliminated, and it offers the users understandable, relevant, reliable, comparable and, above all, useful information for the decision-making process.

Sob a teoria de agência e o contexto de gerenciamento de resultados, o objetivo do trabalho foi estudar a relação entre o gerenciamento de resultados (GR) e o custo da dívida (Ki) das empresas brasileiras de capital aberto, no período de 1996 a 2007. Para isso, foram levantadas duas hipóteses de pesquisa quanto à possível relação entre GR e Ki: H1- as variáveis são inter-relacionadas; e H2- a relação é unidirecional, em que o Ki é influenciado positivamente pelo GR. Os pressupostos levantados, por meio da literatura existente, para a formulação da primeira hipótese são de que, almejando melhores condições contratuais, como o custo da dívida, o custo atual ou passado motiva a empresa a gerenciar o resultado corrente para obter melhores condições de custo futuro ou corrente. Dessa forma, as empresas, intencionalmente, procurariam gerenciar seus resultados, motivadas pelo custo da dívida. Todavia, há suporte teórico para que a relação entre essas variáveis seja unidirecional, com o gerenciamento de resultados influenciando o custo da dívida, sustentando a segunda hipótese. Um por exemplo disso é o gerenciamento de resultados, que é percebido como fator de assimetria de informação pelos credores, os quais, ao identificarem o gerenciamento, exigirão um prêmio maior pelo risco da falta de informação, elevando o custo da dívida das empresas. Assim, para testar as hipóteses do trabalho, a proxy de GR foi definida com o uso do modelo para cálculo dos accruals discricionários proposto por Kang e Sivaramakrishnan (1995) e a variável representativa do custo da dívida foi a relação entre a despesa financeira e o passivo oneroso total. A análise inicial foi feita utilizando-se o índice de correlação de Spearman e análise de média, por meio do teste não-paramétrico U de Mann-Whitney. Em seguida, utilizou-se o teste de especificação de Hausman para análise de simultaneidade, seguido da análise de regressão com dados dispostos em painel com correção de Newey-West, correção por cluster - ambas nas situações em que havia heterocedasticidade e correlação serial-, e estimador MQ2E. Os testes de correlação e de média apontaram para uma relação positiva entre GR e Ki, mas o teste de simultaneidade não indicou haver inter-relação entre as variáveis. Os resultados das regressões em painel indicaram que o Ki, atual ou passado, não influencia o GR, mas apontam para uma relação positiva, em que quanto maior o gerenciamento de resultados, maior o custo da dívida, ou seja, este é influenciado por aquele. Desse modo, os resultados permitem rejeitar a hipótese 1, mas confirmam a segunda hipótese. A princípio, tais resultados indicam que não há uma inter-relação entre GR e Ki e que as empresas não gerenciam com a intenção de impactar no custo da dívida, mas que o gerenciamento de resultados pode ter outras motivações para as empresas brasileiras. Todavia pode ser percebido pelos credores como maneira de reduzir a transparência das demonstrações contábeis, levando-os a cobrar maiores taxas de empresas com gerenciamento de resultados, gerando maior custo da dívida para as companhias. / Under the theory of agency and the context of earnings management, the objective of this research was to study the relationship between the earnings management (EM) and the cost of debt (Ki) of Brazilian open-capital companies, from 1996 to 2007. For this, two hypotheses of research have been raised as to the possible relationship between GR and Ki: H1: the variables are interrelated and H2: the relation is unidirectional where Ki is influenced positively by the GR. The hypothesis raised, by means of existing literature, to elaborate the first hypothesis are that, longing for better contractual conditions, such as the cost of the debt, the current or past costs motivate the company to manage the current result to get better conditions for future or current costs. Therefore, the companies, intentionally, would manage their results, motivated by the cost of debt. However, there is theoretical support that relations between these variables are unidirectional, with the earnings management influencing the cost of debt, supporting the second hypothesis. For example, the earnings management is perceived as a factor of asymmetric information by the creditors, who will demand a bigger prize for the risk regarding the lack of information, which raises the cost of debt of the companies. Thus, to test the hypotheses of this research, the proxy of GR was defined with the use of the model for calculation of discretionary accruals considered by Kang and Sivaramakrishnan (1995) and the representative variable of the cost of debt was the relationship between the financial expense and the total financial liability. The initial analysis was made using the index of correlation of Spearman and analysis of average, by means of non-parametric test U of Mann-Whitney. After that the test of specification of Hausman was used, for analysis of simultaneity, followed by the analysis of regression with data disposed on panel corrected by Newey-West, correction by cluster - both under the condition where there were heteroskedasticity and serial correlation, and estimator MQ2E. The average and correlation tests pointed out a positive relation between GR and Ki, but the concurrence test did not indicate to have interrelation between the variables. The results of the regressions in the panel indicated that the Ki, current or past, does not influence the GR, but point a positive relationship, where the bigger the earnings management, greater is the cost of debt, that is, this is influenced by management. Then, the results allow rejecting the first hypothesis, confirming the second hypothesis. Such results indicate that there is not an interrelation between GR and Ki and that the companies do not manage with the intention of impacting the cost of the debt, but that the earnings management can have other motivations for the Brazilian companies, however it can be perceived, by the creditors, as a way to reduce the transparency of the accounting demonstrations, leading to charge greater fees from companies with earnings management, generating bigger cost of debt for the companies.

The work deals with the role of cost of debt capital in the process of market valuation. Analyses used as a basis for determining the cost of debt, simple and advanced methods of cost of debt capital estimation, as well as the issue of determining the value of debt itself are explained. Further, the work briefly examines the impact of aspects that are not directly related to the credit quality of the firm on its cost of debt; by these aspects, legal conditions for interest payments tax deductibility and third-party guarantees are meant. The pivotal part of the work designs its own model for debt rating and a premium over the risk-free rate of return estimation, with the premium consisting of a premium for expected and unexpected loss of a potential creditor.

This dissertation thesis deals with estimation of the cost of debt that is applied for DCF valuation of businesses in the specific environment of the Czech Republic. Neither theory nor practice has focused on this issue in detail. There is no systematic and comprehensive list of the cost of debt estimation methods usable in practice. Valuers and analysts disregard estimation of the cost of debt arguing that it can be easily estimated from the current bank loan interest rates or from the traded bonds. Neither method is perfect and fully applicable in the Czech Republic. This dissertation thesis shows how the DCF value of an enterprise is shaped by the cost of debt. Almost all formulas analysed caused the value of an enterprise to change with the cost of debt, which makes the estimation of the cost of debt very important. Further, based on an analysis of 183 valuation reports, the thesis documents the practice of the cost of capital estimation in the Czech Republic. It shows that while the cost of equity is mainly estimated in compliance with theory, estimation of the cost of debt is done by simple and sometimes inappropriate methods, because there is no guidance in the literature. Finally, this thesis suggests various methods of the cost of debt estimation that can be used in practice in the Czech Republic. It provides full application guidance to these methods, data that was calculated by the author or taken from other sources and evaluates advantages and disadvantages of these methods. This thesis concludes by recommending methods that should be used for the cost of debt estimation in the Czech Republic as opposed to the simple methods currently used.