Residential property price premia (‘premia’) have long fascinated investors, particularly in times of euphoria, but their social, climatic and urban ramifications are much wider. A proper understanding of premia is hindered by the variety of exogenous influences determining them. They occur within idiosyncratic, complex, and continuously reconfiguring metropoli, conditioned by topography, history, regime, commerce, and culture. Given imperfectly competitive housing markets, conventional explanations for premia are either restricted to their financial dissection, trawl though metrics or cast around for hedonic coefficients. However, premia illuminate affordability and other problems in the broader planning and social debate. With the general significance of premia clarified, the research question of the project becomes: ‘What drives residential property price premium evolution in a metropolis?’ A complete answer involves dissecting the nature and establishing the location of putative premia and disentangling the influence and interactions of their various price drivers. To provide it, the project conducts a property and urban literature review. Based on theory’s insight that higher order contains lower order systems, it develops and investigates a general systems model of residential premia with two modes. The system is conditioned by ideology but forced by population and capital inflows. Within it, premia mutate, influenced by a nested hierarchy of more or less contaminated information. To investigate the model and its different modes, the project employs tests across system pointers, at the macro, meso (all urban) and micro spatial resolutions. First, the turbulence and permeability of residential property markets to exogenous influences is assessed. The project then looks at the urban mosaic in the growing Sunbelt migration city of Brisbane, Australia, over the boom period from 1998-2004. Locally, it conducts a case study and survey in one micro-location, seeking clues in transaction patterns (output), property system agents (components) and the information they use (feedback mechanisms). Finally, the project draws some relevant policy implications. Its key findings are that urban housing markets are open, complex and polarised. In an exuberant economic climate, migration and debt fuel metropolitan price escalation. Public urban initiatives reinforce central incumbent affluence or spark fresh bouts of speculation. Individual premia are heterogeneous but often feed off local construction projects or iconic refurbishment. Reflecting their demographics and motives, agent risk appetites are diverse although investors are usually less averse to renewal. System feedback involves a congruence of media and local activity signals. Neither local conviviality nor Bohemian influences are, by themselves, significant. Rather, buyer rationality is validated by post-purchase infrastructure completions. The thesis of this project is, hence, that in euphoric capital markets, migration and debt accelerates the endogenous mutation of property from homes within a community towards speculative paper assets. The implication is that the excessive proliferation of premia indicates economic imbalance and urban malaise which requires recognition and treatment. While premia are paid for perceived privilege or prospects, cognitive risk representations and expectations evolve. Sometimes judgment is contaminated by media fantasy but often validated by accommodating government policy and central revitalisation projects. Yet, within a wider social and ecological remit, rampant premia suggest flaws in urban strategy, governance and planning practice. In terms of windfall events or unearned rent, the cumulative effects of ill-considered projects and price distortions can be ugly and wasteful. They alienate and accentuate spatial privilege without generating sustainable jobs. The project has procedural and substantive policy implications. The dynamics of residential premia cannot be disentangled from capital market volatility, urban fragmentation and reconstitution. Enlightened property development requires visionary urban planning beyond electoral cycles. Rather than unregulated markets or disjointed incrementalism, the project points to the advantages of cohesive projects and inclusive hubs. It impels ecological and people-focused development to nurture capable, connected and considerate edge communities. Its first steps are theoretical recognition, policy clarification, government reform, market constraints, price and tax rationalisation and spatial transparency.

This thesis investigates consumer search behavior in different contexts and its implications on certain market outcomes. It consists of three self-contained essays. Part one investigates if people search optimally and how price promotions (such as the provision of price discounts) influence search intensity and risk-taking behavior. We start with a typical sequential search task in a finite time horizon (with exogenously determined price dispersion) as the baseline treatment. In the two experimental treatments, exogenous discounts are introduced to the search process. The treatments differ in the amount of information on the discounts revealed to the subjects. Subjects’ search behavior is roughly consistent with optimality for a risk-neutral agent, but significantly influenced by the introduction of discount vouchers. We find that subjects’ search intensity is significantly reduced if they are in a shop that offers discounts, even when the monetary benefit induced by the discount has been taken into account. This suggests that people seem to gain extra non-monetary utility from buying a discounted product. Alternatively, subjects might overestimate the value of a discount. Following the findings in part one, we focus on price-framing effects of discounts on consumer search behavior in part two. In order to isolate the price-framing effect from all other possible influences, we adopt an extremely simple two-shop search model in which a consumer who sees the price for an item in a shop has to decide either to buy it or to incur a search cost to learn the ex-ante uncertain price in a second shop. The experiment is designed such that a rational buyer should make identical decisions in the base treatment (where prices are posted as net prices in both shops) and in the experimental treatments (where the price in one of the shops is framed as a gross price with a discount, holding the net-price constant). Using structural estimation of the observed risk preferences, we find that people tend to be more risk-averse and hence buy from the initial shop more often in the discount treatments, regardless of where the discount is offered. The seemingly trivial change to a discount-framing increases the complexity of the decision problem. Subjects reveal a tendency to stick with the comparatively less complex options more frequently as the complexity of the decision problem increases. However, this bias declines with experience, as subjects become more and more familiar with the framing. In part three, we study search behavior in a market experiment, where prices are determined endogenously by human players. More specifically, we examine the behavioral factors and the underlying mechanism which drive the widely observed asymmetric price adjustment to cost shocks (in a world with costly search behavior and information asymmetry). We show that price dispersion, as well as asymmetric price adjustment to cost shocks, arises in experimental markets, even though the standard theory predicts neither. We find that after controlling all the potential theoretical factors, the observed price dispersion can be explained by the presence of bounded rational play. Under price dispersion, asymmetric price adjustment arises naturally, as it is harder for buyers to learn that a negative cost shock has taken place. Learning is much quicker after a positive shock. / Thesis (Ph.D.) -- University of Adelaide, School of Economics, 2010

Thesis (Ph. D.)--University of Missouri-Columbia, 2008. / The entire dissertation/thesis text is included in the research.pdf file; the official abstract appears in the short.pdf file (which also appears in the research.pdf); a non-technical general description, or public abstract, appears in the public.pdf file. Title from title screen of research.pdf file (viewed on June 11, 2009) Vita. Includes bibliographical references.

Dans le cadre des évolutions liées à Bale II, le régulateur a rendu obligatoire la gestion du risque opérationnel en demandant aux établissements financiers d’affecter un montant en fonds propres face à ce risque.Si ce risque était géré implicitement par les banques, la nécessaire visibilité de cette gestion imposée par le régulateur va fortement influencer la gestion des risques de la banque,l’orientant dans un premier temps vers une vision quantitative et statique de cette gestion.De nombreuses recherches se sont intéressées dans un premier temps à l’aspect quantitatif du risque dans une démarche positive, à la recherche du modèle et des données pertinentes afin d’avoir l’estimation la plus vraisemblable du risque opérationnel.Or le risque opérationnel est par nature difficilement quantifiable; il est diffus, multiforme et ne repose sur aucun encours connu. De plus il est fortement influencé par la composante humaine, que le risque soit intentionnel ou non. L’individu est à la fois source du risque et acteur essentiel du processus d’identification.Il nous a semblé alors intéressant d’adopter une vision interprétative du risque, c’est-à-dire de nous intéresser non pas au résultat de l’identification, dans l’objectif de se rapprocher le plus de la réalité mais en essayant de comprendre et d’expliquer les processus à l’oeuvre lors de l’identification et plus précisément les processus d’apprentissage.Nous cherchons alors à répondre à la problématique suivante : quels peuvent être les impacts de l’identification du risque opérationnel sur le processus d’apprentissage organisationnel ?La précision du concept d’identification du risque nous amènera à comprendre les deux composantes de ce dispositif, l’une relative aux informations transmises et l’autre sur les connaissances tacites nécessaires à l’identification. C’est dans ce contexte que nous préciserons l’apport des théories relatives à l’apprentissage organisationnel dans une approche intégrative des dimensions cognitives et comportementales. Notre méthodologie repose sur une étude menée au sein d’un établissement de crédit français qui a adopté une méthode avancée en matière de gestion du risque opérationnel, méthode la plus aboutie en termes de gestion du risque opérationnel. Au sein de cet établissement, nous avons sélectionné deux métiers du fait de leur représentativité, la banque de détail et la banque d’investissement et de financement. A partir d’une grille d’analyse des entretiens qui s’appuie sur les composantes du dispositif règlementaire d’identification du risque opérationnel, nous présentons dans la deuxième partie6nos résultats. Ces résultats sont analysés au sein de chaque métier puis dans une vision inter cas afin de faciliter une comparaison entre les cas et faire émerger des invariants dans les processus d’apprentissage identifiés.Les résultats de notre recherche mettent en avant l’intérêt du dispositif d’identification du risque sur l’apprentissage comportemental dans un processus stimulus-réponse. Cependant ils en montrent également les limites dans un environnement en constante mutation où la standardisation des processus freine l’utilisation pertinente des compétences spécifiques. En revanche, les apprentissages de type cognitif qui se dessinent génèrent de nouveaux modes de raisonnements au niveau individuel, structurés par le dispositif et développés dans le cadre d’interactions. / Our research has the ambition to study the impact of identifying operational risk within the framework of the evolution of the Basel II regulations.The absence of a theoretical reference relative to the object of our research, the identification of operational risk, the context of strong asymmetry which characterizes the banking organisation, brings us to use two theoretical routes: the contractual theory and more specifically the theory of the agency which better lets us understand the stakes in the implementation of identification. We will also base ourselves on the relative theories of organisational training in an integrative approach of cognitive and behavioural dimensions. Our methodology is based on a case study within a French finance company that has adopted the AMA approach, the approach that is the most developed in terms of operational risk management. Within this establishment, due to their representativeness, we have selected two professions, the “banquede detail” (retail bank) and investment bank. According to an analytical grid of interviews that are based on the components of the regulatory system of identifying operational risk, we present in the second part our results.These results have been analysed first for each profession and then with a vision of case comparison in order to permit a comparison between the cases and to allow the emergence of the invariables in the process of the training identified. The results of our research highlight the interest of the system of risk identification on the compartmental training in the process of response stimulus. However, it also shows the limits within an environment of constant change where the standardisation of processes stops the pertinent utilisation of specific competencies. On the other hand, the type of cognitive training that generates new methods of reasoning at an individual level, structured by the system and developed within the social interaction framework. The exchanges have come from the necessity to identify the risk and to formalize it, oblige the individual to justify him or herself and favorise the confrontation of points of view. These interactions could lead to developing new thinking tied to the needs of the need for argument and the socialisation of the individual thinking that tends to objectivise itself.

A presente pesquisa tem como objetivo identificar e analisar os principais Mecanismos de Sinalização utilizados na cadeia da carne bovina brasileira, bem como, verificar o grau de ocorrências dessas sinalizações pela Ciência, pelo Governo e pela Mídia, por meio de processos de mineração de textos e descoberta do conhecimento. Embasando-se na Economia da Informação como teoria-chave dessa análise e concentrando-se nas teorias da Assimetria de Informação, Seleção Adversa e Sinalização e, também, da revisão das interações entre Ciência, Mídia e Governo, foram definidas cinco dimensões, sob as quais os sinais sobre a carne bovina são enquadrados: Econômica, Comunicação e Marketing, Qualidade, Institucional e Sistema de Produção. Para análise, foram coletados documentos textuais em formato eletrônico ao longo de cinco anos. A busca dos documentos deu-se em bases de dados de publicações científicas, em portais do Governo e em arquivos dos jornais e magazines disponíveis na rede mundial de computadores , a partir de palavras-chave relacionadas à cadeia da carne bovina. Foram selecionados 4.281 artigos científicos para compor a base de dados da Ciência, 730 documentos para a base de dados do Governo e 5.439 documentos na base de dados da Mídia, totalizando 10.450 documentos selecionados. Esses documentos foram armazenados e classificados no software QDA Miner segundo as variáveis fonte (Ciência, Governo e Mídia) e ano (2005, 2006, 2007, 2008 e 2009). Para extrair o conhecimento das bases textuais, foi elaborada uma estrutura de análise constituída pelas dimensões, códigos (sinais) e palavras -chave. Na sequência foram analisados e codificados os documentos com os respectivos sinalizadores. Aplicando-se a estrutura de análise do software, foram utilizadas representações gráficas elaboradas a partir da frequência relativa das ocorrências de cada código nas três fontes de informação. De acordo com os resultados apresentados foi possível verificar que os termos e dimensões utilizados pela Ciência, Governo e Mídia diferem nas frequências em praticamente todos os códigos , demonstrando pouca sinergia entre as fontes de informação . No decorrer do período analisado, percebeuse maior oscilação na frequência das publicações que incluíam os sinais Sanidade e Sustentabilidade, os quais foram mais enfatizados nos últimos anos da presente análise. Os códigos Certificação e Rastreabilidade que pareciam ter maior destaque para fornecer informações sobre aspectos ligados à carne bovina brasileira, de acordo com estudos existentes, não são os mais emitidos, de acordo com esta pesquisa. A maior identidade da Ciência ao longo do período de anál ise foi para a Dimensão Sistema de Produção, demonstrando que as publicações científicas possuem um caráter mais voltado à melhoria da produção em si, como alimentação, genética, reprodução, idade de abate, dentre outros. Em relação ao Governo, observou-se maior identidade dessa fonte de informação para a Dimensão Institucional, com exceção do ano de 2007, onde se percebe maior es ênfases das publicações do Governo nas Dimensões Qualidade e Sistema de Produção . Já em relação ao que se observa na Mídia, há maior interesse em publicações nas Dimensões Qualidade e Sistema de Produção. Ao analisar os resultados entre as três fontes de informação, verifica-se maior proximidade entre o que se observa na Mídia e no Governo, visto a predominância da Dimensão Sistema de Produção na Ciência. Analisando comparativamente as fontes de informação foi possível observar que o maior grau de similaridade entre os códigos foi observad o nos documentos publicados pelo Governo. Embora os códigos adotados para sinalizar informações sobre a cadeia da carne bovina sejam os mesmos, enquadrados igualmente em cinco Dimensões, determinados sinalizadores são utilizados com maior destaque, similaridades e agrupamentos de códigos e, de acordo com as peculiaridades de cada fonte de informação investigada (Ciência, Governo e Mídia). / The goal of this research is to identify and analyze the main signaling mechanisms utilized in the Brazilian beef chain, as well as to investigate the degree of occurrence of these signalings by science, government, and media, through text mining and knowledge discover processes. Using Information Economy as key theory of this analysis and concentrating on the theories of Information Asymmetry, Adverse Selection and Signaling and, also, through revision of the interactions between science, media and government , five dimensions have been defined which encompass the signals about beef: economic, communication and marketing, quality, institutional and production system. To perform the analysis, electronic text documents over a five-year period have been collected. The documents have been searched in scientific journal databases, government portals, and news paper and magazine files available on the Internet, based on keywords connected to beef chain. The selection included 4.281 scientific articles to compose the science database, 730 documents to the government database, and 5.439 documents to the media database, in a total of 10.450 documents selected. These documents have been stored and classified in the QDA Miner software through the source variables (science, government and media ) and year (2005, 2006, 2007, 2008 and 2009). To extract the knowledge from the text databases, it has been elaborated an analysis framework composed by dimensions, codes (signals) and keywords. The next step was to analyze and codi fy the documents with the respective signallers. With application of the software analysis framework , it has been used graph representations elaborated from the relative frequency of each code occurrences in the three sources of information. According to the results found, it h as been evidenced that the terms and dimensions used by science, government an d media differ in frequency in practically all codes, demonstrating little synergy between the information sources. Over the period analyzed, a greater oscillation has been perceived in publications that included the sani ty and sustainability signals, which have been more emphasized in certain years. According to this research, the certification and traceability signals which, judging from existing studies, seemed to have the greatest emphasis to supply information on aspects related with Brazilian beef, are not the most issued. Science, over the analysis period, has been more identified with the production system dimension, demonstrating that scientific journals are more focused on the enhancement of production itself, such as feeding, genetics, breeding, slaughter age, among others. With regard to the government, it has been observed that this source of information is more identified with the institutional dimension, with exception of the year 2007, where the greatest emphasis of government publications is perceived in the quality and production syste m dimensions. In the media, quality and production system dimensions are the most emphasized in publications. Analyzing the results between the three sources of information, a closer connection between media and government is perceived, given the predominance of production system dimension in science . Comparative analysis between the sources of information revealed that the greatest degree of similarity between the codes has been found in the documents published by the government. Although the codes adopted to signaling information about beef chain are the same, equally framed in five dimensions, certain signallers are utilized with a greater emphasis, similarities and code groupings, and in accordance with the peculiarities of each source of information investigated (science, government and media).

This thesis consists of three essays on banking in the U.S. The first two chapters study how supervisors and regulators influence bank behavior. The third chapter explores how bank CEOs allocate credit. The first chapter uses a quasi-natural experiment, the closure of regulatory offices, to identify the effects of supervision on bank behavior. Under the decentralized structure of U.S. bank supervision, banks in the same geographic area may be supervised by different regulatory offices. The chapter shows that, following the closure of a regulatory office, banks previously supervised by that office increase their solvency risk and lending compared with banks in the same counties that are supervised by a different regulatory office. Further, these banks exhibit lower risk-adjusted returns, lower asset quality, and opportunistic provisioning behavior for loan losses. Information asymmetry between banks and supervisors partly explains the results. The second chapter documents that nearly 30% of U.S. banks employ at least one board member who currently or previously served on a regulator’s advisory council or on the board of a regulator as a form of public service. The chapter shows that connections to regulators undermine regulatory discipline by decreasing the sensitivity of bank risk to capital. Connected banks are able to extract larger public subsidies than non-connected banks by shifting risk to the financial safety-net, resulting in wealth transfers from taxpayers to shareholders of risk-shifting connected banks. One potential reason for these effects is that connected banks receive preferential treatment in supervision from regulators. The third chapter uses the birthplace of U.S. bank CEOs to investigate the effect of hometown favoritism on bank business policies. Exploiting within-bank variation in distances to a CEO’s hometown, the chapter shows that banks make more mortgage and small business lending as well as branch expansions in counties that are proximate to the hometown of the CEO. This is due to the CEO’s altruistic attachment to her hometown; the effects are stronger during economic downturns, among altruistic CEOs, in poorer counties and marginal mortgage applicants. Further, hometown favoritism does not lead to worst bank performance. However, it is associated with positive economic outcomes in counties exposed to greater favoritism.

A presente pesquisa tem como objetivo identificar e analisar os principais Mecanismos de Sinalização utilizados na cadeia da carne bovina brasileira, bem como, verificar o grau de ocorrências dessas sinalizações pela Ciência, pelo Governo e pela Mídia, por meio de processos de mineração de textos e descoberta do conhecimento. Embasando-se na Economia da Informação como teoria-chave dessa análise e concentrando-se nas teorias da Assimetria de Informação, Seleção Adversa e Sinalização e, também, da revisão das interações entre Ciência, Mídia e Governo, foram definidas cinco dimensões, sob as quais os sinais sobre a carne bovina são enquadrados: Econômica, Comunicação e Marketing, Qualidade, Institucional e Sistema de Produção. Para análise, foram coletados documentos textuais em formato eletrônico ao longo de cinco anos. A busca dos documentos deu-se em bases de dados de publicações científicas, em portais do Governo e em arquivos dos jornais e magazines disponíveis na rede mundial de computadores , a partir de palavras-chave relacionadas à cadeia da carne bovina. Foram selecionados 4.281 artigos científicos para compor a base de dados da Ciência, 730 documentos para a base de dados do Governo e 5.439 documentos na base de dados da Mídia, totalizando 10.450 documentos selecionados. Esses documentos foram armazenados e classificados no software QDA Miner segundo as variáveis fonte (Ciência, Governo e Mídia) e ano (2005, 2006, 2007, 2008 e 2009). Para extrair o conhecimento das bases textuais, foi elaborada uma estrutura de análise constituída pelas dimensões, códigos (sinais) e palavras -chave. Na sequência foram analisados e codificados os documentos com os respectivos sinalizadores. Aplicando-se a estrutura de análise do software, foram utilizadas representações gráficas elaboradas a partir da frequência relativa das ocorrências de cada código nas três fontes de informação. De acordo com os resultados apresentados foi possível verificar que os termos e dimensões utilizados pela Ciência, Governo e Mídia diferem nas frequências em praticamente todos os códigos , demonstrando pouca sinergia entre as fontes de informação . No decorrer do período analisado, percebeuse maior oscilação na frequência das publicações que incluíam os sinais Sanidade e Sustentabilidade, os quais foram mais enfatizados nos últimos anos da presente análise. Os códigos Certificação e Rastreabilidade que pareciam ter maior destaque para fornecer informações sobre aspectos ligados à carne bovina brasileira, de acordo com estudos existentes, não são os mais emitidos, de acordo com esta pesquisa. A maior identidade da Ciência ao longo do período de anál ise foi para a Dimensão Sistema de Produção, demonstrando que as publicações científicas possuem um caráter mais voltado à melhoria da produção em si, como alimentação, genética, reprodução, idade de abate, dentre outros. Em relação ao Governo, observou-se maior identidade dessa fonte de informação para a Dimensão Institucional, com exceção do ano de 2007, onde se percebe maior es ênfases das publicações do Governo nas Dimensões Qualidade e Sistema de Produção . Já em relação ao que se observa na Mídia, há maior interesse em publicações nas Dimensões Qualidade e Sistema de Produção. Ao analisar os resultados entre as três fontes de informação, verifica-se maior proximidade entre o que se observa na Mídia e no Governo, visto a predominância da Dimensão Sistema de Produção na Ciência. Analisando comparativamente as fontes de informação foi possível observar que o maior grau de similaridade entre os códigos foi observad o nos documentos publicados pelo Governo. Embora os códigos adotados para sinalizar informações sobre a cadeia da carne bovina sejam os mesmos, enquadrados igualmente em cinco Dimensões, determinados sinalizadores são utilizados com maior destaque, similaridades e agrupamentos de códigos e, de acordo com as peculiaridades de cada fonte de informação investigada (Ciência, Governo e Mídia). / The goal of this research is to identify and analyze the main signaling mechanisms utilized in the Brazilian beef chain, as well as to investigate the degree of occurrence of these signalings by science, government, and media, through text mining and knowledge discover processes. Using Information Economy as key theory of this analysis and concentrating on the theories of Information Asymmetry, Adverse Selection and Signaling and, also, through revision of the interactions between science, media and government , five dimensions have been defined which encompass the signals about beef: economic, communication and marketing, quality, institutional and production system. To perform the analysis, electronic text documents over a five-year period have been collected. The documents have been searched in scientific journal databases, government portals, and news paper and magazine files available on the Internet, based on keywords connected to beef chain. The selection included 4.281 scientific articles to compose the science database, 730 documents to the government database, and 5.439 documents to the media database, in a total of 10.450 documents selected. These documents have been stored and classified in the QDA Miner software through the source variables (science, government and media ) and year (2005, 2006, 2007, 2008 and 2009). To extract the knowledge from the text databases, it has been elaborated an analysis framework composed by dimensions, codes (signals) and keywords. The next step was to analyze and codi fy the documents with the respective signallers. With application of the software analysis framework , it has been used graph representations elaborated from the relative frequency of each code occurrences in the three sources of information. According to the results found, it h as been evidenced that the terms and dimensions used by science, government an d media differ in frequency in practically all codes, demonstrating little synergy between the information sources. Over the period analyzed, a greater oscillation has been perceived in publications that included the sani ty and sustainability signals, which have been more emphasized in certain years. According to this research, the certification and traceability signals which, judging from existing studies, seemed to have the greatest emphasis to supply information on aspects related with Brazilian beef, are not the most issued. Science, over the analysis period, has been more identified with the production system dimension, demonstrating that scientific journals are more focused on the enhancement of production itself, such as feeding, genetics, breeding, slaughter age, among others. With regard to the government, it has been observed that this source of information is more identified with the institutional dimension, with exception of the year 2007, where the greatest emphasis of government publications is perceived in the quality and production syste m dimensions. In the media, quality and production system dimensions are the most emphasized in publications. Analyzing the results between the three sources of information, a closer connection between media and government is perceived, given the predominance of production system dimension in science . Comparative analysis between the sources of information revealed that the greatest degree of similarity between the codes has been found in the documents published by the government. Although the codes adopted to signaling information about beef chain are the same, equally framed in five dimensions, certain signallers are utilized with a greater emphasis, similarities and code groupings, and in accordance with the peculiarities of each source of information investigated (science, government and media).

Cette thèse comprend trois essais dont l’objectif est d’étudier les déterminants et les implications de la qualité de l’information narrative des entreprises cotées. L’originalité de ce travail par rapport aux études antérieures réside dans le fait que nous nous intéressons à la dimension qualitative de l’information divulguée. Afin d’appréhender la qualité de cette information nous recourons à des techniques de traitement automatique de langage naturel qui permettent de construire des indices de lisibilité des rapports annuels.Le premier essai étudie l'effet de la complexité textuelle des rapports annuels sur la liquidité des titres. L’utilisation d’un échantillon d’entreprises françaises cotées en bourse sur la période 2002-2013 montre l’existence d’une relation positive entre le degré de lisibilité des rapports annuels et la liquidité des titres. Ces résultats suggèrent que la complexité textuelle de l’information narrative affecte les investisseurs sur le marché des actions.Le deuxième essai étudie l’effet de la lisibilité des rapports annuels sur le coût des fonds propres des entreprises. Nous menons notre étude empirique sur un échantillon d’entreprises américaines cotées en bourse sur la période 1995-2012. Les résultats montrent que les entreprises font face à un coût de financement plus élevé lorsque leurs rapports annuels sont moins lisibles, ce qui indique qu’un degré faible de lisibilité réduit la capacité des investisseurs à prévoir les performances futures de l’entreprise et leur amène par conséquence à demander un rendement de fonds propres plus élevé.Le troisième essai examine l’effet des pratiques de réduction d’impôt des entreprises sur la lisibilité de leurs divulgations financières. La littérature mobilisant la théorie d’agence montre que ces pratiques de réduction d’impôt créent un cadre permettant aux dirigeants d’extraire des bénéfices privés aux dépens des autres parties prenantes. Afin de s’assurer que leurs actions opportunistes ne soient détectées, les dirigeants réduisent la qualité de l’information divulguée, ce qui détériore l’environnent informationnel de l’entreprise en question. En utilisant un échantillon d’entreprises américaines cotées en bourse pour la période 1995-2012, nous constatons que les entreprises qui s’engagent dans des politiques de diminution d’impôt publient des rapports annuels moins lisibles et plus ambigus.Mot Clés: Information narrative; lisibilité des rapports annuels; Risque d’information ; Liquidité; Coût des fonds propres; Optimisation fiscale / This thesis comprises three separate but interconnected essays that focus on the determinants and economic implications of corporate narrative disclosure. The first essay examines the effect of annual report textual complexity on firms’ stock liquidity. Using techniques from computational linguistics, we predict and find that less readable filings are associated with lower stock liquidity. Our study provides evidence that difficult-to-read annual reports can act as a non-trivial impediment to investors’ ability to process information into useful trading signals. The findings are robust to a battery of sensitivity tests, including endogeneity, use of alternative regression techniques, and use of alternative liquidity and readability proxies.Using a large panel of U.S. public firms, the second essay presents the first evidence highlighting the relation between annual report readability and cost of equity capital. We hypothesize that complex textual reporting deters investors’ ability to process and interpret annual reports, leading to higher information risk, and thus higher cost of equity financing. Consistent with our prediction, we find that greater textual complexity is associated with higher cost of equity capital. Our results are statistically significant and economically important. We also show that disclosure tone exerts a non-trivial bearing on the cost of equity. Our findings are robust to a battery of sensitivity checks, including use of multiple estimation methods, alternative proxies of annual report readability and cost of equity capital measures, and potential endogeneity concerns. Overall, our study contributes to the research examining the relation between disclosure quality and cost of capital.The third essay investigates the effect of firms’ tax avoidance practices on the textual properties of their annual filings. Using a large sample of U.S.-listed firms, we document a positive and statistically significant relation between corporate tax avoidance and annual report textual complexity. In addition, we show that managers of tax-avoiding firms tend to hide their avoidance behavior in more ambiguous language. Our results prove to be robust to the use of numerous alternative proxies of corporate tax avoidance and annual report readability. The findings are also robust to a number of checks, including, using additional control variables, employing alternative regression methodologies, and addressing endogeneity concerns.Keywords: Narrative disclosure; Annual report readability; Disclosure tone; Information risk Stock liquidity; Cost of equity capital; Corporate tax avoidance

Diversas teorias tentam explicar o que determina a política de financiamento adotada pelas empresas. Uma das abordagens existentes, a Teoria de Pecking Order, foca a assimetria de informação como um importante determinante da estrutura de capital. No presente trabalho, busca-se analisar o pressuposto central da referida teoria e verificar se a assimetria de informação influencia nas decisões de financiamento de empresas brasileiras de capital aberto no período 1997-2007. Para tanto, são utilizadas variáveis proxies para assimetria de informação, além de variáveis de controle representativas de características das empresas. Utiliza-se a metodologia de análise de dados em painel. Como resultado principal, encontra-se que empresas consideradas com menor grau de assimetria informacional são mais endividadas que as demais na análise do nível de endividamento total, resultado contrário a teoria de pecking order, onde estas empresas teriam a oportunidade de captar recursos por meio da emissão de ações devido à baixa probabilidade de ocorrência dos problemas derivados da assimetria de informação. No entanto, este resultado apóia a relação esperada alternativa, onde empresas com menor assimetria informacional propiciam maior facilidade de avaliação do seu risco por parte dos credores, o que poderia aumentar a oferta de crédito, favorecendo a utilização de dívidas por meio da redução dos custos de seleção adversa e pelo aumento da capacidade de financiamento destas empresas. Nesse sentido, torna-se importante analisar também o papel do risco ao estabelecer uma hierarquia de preferências por fontes alternativas de financiamento utilizadas por empresas brasileiras. / Several theories try to explain what determines the financing policy adopted by the firms. One of the existent approaches, Pecking Order Theory, focuses on the information asymmetry as an important determinant of the capital structure. In the present work, we try to analyze the central presumption of the referred theory and to verify if the information asymmetry influences the financing decisions of Brazilian open capital firms in the period 1997-2007. In order to do so, proxies variables are used for information asymmetry, besides control variables which represent of firms characteristics. Panel data analysis is the methodology used. As main result, it was found that companies considered to have the lowest degree of information asymmetry are higher leveraged than the others analyzing the level of total debt, the result is contrary to the pecking order theory, where these companies would have the opportunity to raise resources by emitting shares due to the low probability of occurring problems derived from information asymmetry. However, this result supports the expected alternative relation, where companies with lower asymmetric information make it easier to their debt holders to evaluate their risk, what could increase the credit offer, favoring the use of debts by reducing the adverse selection costs and increasing the debt capacity of these companies. In doing so, it becomes important to also analyze the role of the risk when establishing a hierarchy of preferences adopted by Brazilian firms regarding alternative financing sources.

O propósito do presente estudo foi investigar a relação entre qualidade das informações contábeis, restrição financeira e decisões de investimentos na América Latina. Para isso analisou-se uma amostra com 10.318 observações de 958 empresas não financeiras de capital aberto, oriundas de sete diferentes países, entre os períodos de 1992 a 2009. Com o uso do modelo probabilístico Logit estimou-se o impacto da qualidade dos lucros na probabilidade das firmas apresentarem superinvestimentos, subinvestimentos ou restrição financeira. Em seguida, utilizando o método dos momentos generalizados (GMM) e uma adaptação do modelo acelerador de investimentos, estimaram-se os determinantes dos investimentos. Os resultados indicam que uma pior qualidade nos lucros possa reduzir a probabilidade de decisões eficientes de investimentos e aumentar a possibilidade de subinvestimentos. Verificou-se haver indícios de que a pior qualidade nos lucros possa exacerbar os problemas causados pelo conflito de interesses entre os tomadores de decisões e os investidores das firmas latino-americanas, pois a baixa qualidade nos lucros mostrou-se positivamente associada às taxas de investimentos das firmas com superinvestimentos e negativamente relacionada às taxas de investimentos das firmas com subinvestimentos. Por fim, os resultados indicam que a qualidade dos lucros afeta a sensibilidade dos investimentos ao caixa das firmas que estejam em situação de restrição financeira. Assim, conclui-se que a qualidade das informações financeiras, como ferramenta para redução de assimetria de informações, pode impactar as decisões de investimentos das empresas da América Latina. / The purpose of this study was to investigate the relationship among accruals quality, financial constraint and investment decisions in Latin America. For this, it was analyzed a sample with 10,318 observations of 958 non-financial companies from seven different countries, between 1992 to 2009. Using the probabilistic model Logit it was estimated the impact of earnings quality on the probability of overinvestment, underinvestment and financial constraints. Then, using the generalized method of moments (GMM) and an adaptation of the accelerator model of investment, it was estimated the investment determinants. The results indicate that a poor accruals quality can reduce the probability of efficient investment decisions and increase the possibility of underinvestment. The results show that poor accruals quality could exacerbate the problems caused by the conflict of interest between decision-makers and investors from Latin American firms, because poor accruals quality was positively associated with investment rates for firms with overinvestment and negatively related to firms investment for firms with underinvestment. Finally, the results indicate that the accruals quality affects the investment-cash flow sensitivities for firms in financial constraint situation. Thus, we concluded that quality of financial information, as a tool to reduce information asymmetries, might affect the investment decisions in Latin America.