Zugl.: Augsburg, Univ., Diss., 2007

Master-Arbeit Univ. St. Gallen, 2006.

Master-Arbeit Univ. St. Gallen, 2008.

El objetivo de este trabajo es analizar la estabilidad de las calificaciones de riesgo crediticio (ratings) de un tipo de derivados de crédito conocido como synthetic Collateralized Debt Obligations (CDO sintéticos). Durante la crisis subprime gatillada el 2007, la mayoría de los derivados de crédito tipo CDO tuvo un muy mal desempeño. Debido a que cada CDO poseía una calificación de riesgo crediticio, este mal desempeño evidenció la falta de precisión de los ratings de las agencias calificadoras. En este contexto, este trabajo se enfoca en los CDO sintéticos, por dos motivos. Primero, pues ellos tuvieron un rol protagónico en la crisis subprime al transformarse en uno de los instrumentos favoritos de los especuladores para hacer “apuestas unidireccionales”. Y segundo, dado que los CDO sintéticos se transaban en un mercado secundario no regulado, y poco transparente, esto los hace más interesantes como objetos de estudio. Tradicionalmente, los ratings de las calificadoras de riesgo se han basado en un único estimador, sin considerar el error asociado con éste. Por ello, este trabajo analiza la estabilidad de las calificaciones de riesgo, mediante la estimación de intervalos de confianza y análisis de sensibilidad en función de los distintos parámetros considerados. Este trabajo utiliza la metodología de calificación de Moody’s, una de las calificadoras de mayor participación de mercado, que emplea el concepto de pérdida esperada. En el desarrollo de los análisis, se consideró la información a la cual un inversionista habría tenido acceso previo a la crisis subprime. Los casos de estudio seleccionados corresponden a CDO sintéticos representativos del mercado global de riesgo de crédito. Este trabajo concluye que el empleo de un solo valor como medida de riesgo de crédito para los CDO sintéticos es inadecuado. Los intervalos de confianza estimados para la perdida esperada contienen consistentemente más de un rating, es decir, contienen un margen de error significativo. Además, este trabajo revela que la información disponible previa a la crisis subprime habría permitido a inversionistas sofisticados haber detectado el peligroso margen de error asociado a los ratings. Por último, este trabajo pone de manifiesto la importancia de reconsiderar la estructura de los marcos regulatorios financieros que en la mayoría de los países se basan en ratings emitidos por calificadoras de riesgo, y por lo tanto, son inherentemente inestables. Considerando la importancia de las conclusiones de este trabajo, sería interesante extender esta investigación a otras metodologías de calificación de riesgo y a otros tipos de derivados de crédito.

University, Diss.--Tübingen, 2007.

Machine learning and neural networks have recently become very popular in a large category of domains, partly thanks to their ability to solve complex problems by finding patterns in data, but also due to an increase in computing power and data availability. Successful applications of machine learning include for example image classification, natural language processing, and product recommendation. Despite the potential upside of machine learning applied to financial data there exists relatively few articles published while the ones that do exist exhibit that there exists a potential for the tools that it provides. This thesis utilizes neural networks to price collateralized loan obligations which is a type of bond that is backed by a large pool of corporate loans, rather than being issued by a single company or government like a regular bond. The large pool of corporate loans and structure of a collateralized loan obligation makes it a good candidate for this type of research as it involves regressing a large number of variables into a final single real-valued price of the bond where the relations are not necessarily linear. The thesis establishes a relatively simple model and builds upon this using a state-of-the-art ensemble method while also exploring a volatility scaled loss function. The findings of this thesis are that artificial neural networks can price collateralized loan obligations using only their structural and loan pool data with an accuracy close to that of a human. Ensemble methods outperform non-ensemble methods and boost performance by up to 28% when only considering mean squared error while scaling the loss function with the inverse of market volatility does not boost performance. The best performing model can price a collateralized loan obligation tranche rated AAA with an average absolute error of 0.88 and an equity tranche with an average mean absolute error of 4.67. / Under de senaste åren har maskininlärning samt artificiella neurala nätverk blivit väldigt populära i många olika domäner. Detta är delvis tack vare deras förmåga att lösa komplexa problem genom att hitta mönster i data, men även tack vare en ökning i beräkningskraft samt att tillgängligheten av data har blivit bättre. Några exempel på områden där maskininlärning har applicerats med framgång är klassificering av bilder, språkteknologi samt produktrekommendationer. Trots att maskininlärning skulle kunna erbjuda en stor potentiell uppsida vid lyckad tillämpning på finansiella data finns relativt lite studier publicerade kring ämnet. De studier som däremot är publicerade visar på stora möjligheter inom området. Den här studien använder artificiella neurala nätverk för att prissätta ”collateralized loan obligations” (CLOs), som tyvärr inte har någon bra svensk översättning. En CLO utfärdar obligationer vars underliggande värde härstammar från en portfölj av företagslån, och är därmed ett finansiellt instrument. Strukturen av en CLO och dess underliggande lånportfölj ger upphov till en stor mängd data, vilket gör instrumentet till en bra kandidat för maskininlärning. Studien etablerar ett relativt enkelt neuralt nätverk som sedan används för ett jämföra med en ensemblemetod samt en modifierad loss funktion som tar höjd för volatilitet. Slutsatserna av den här studien är att neurala nätverk lyckas prissätta instrumenten näst intill lika bra som vad en människa skulle kunna göra med befintliga metoder som bygger på Monte Carlo simulering. Däremot är studiens metod inte lika beroende av antaganden som gör den befintliga metoden väldigt känslig. Vidare så bidrar ensemblemetoden som används till att minska det genomsnittliga felet i kvadrat med upp till 28%. Att ta höjd för volatilitet vid inlärning bidar inte till att minska felet.

Master of Science / Department of Industrial & Manufacturing Systems Engineering / Chih-Hang Wu / The recent economic crisis has been partially blamed on the decline in the housing market. This decline in the housing market resulted in an estimated 87% decline in value of collateralized debt obligations (CDOs) between 2007 and 2008. This drastic decline in home values was sudden and unanticipated, thus it was incomprehensible for many investors how this would affect CDOs. This shows that while analytical techniques can be used to price CDOs, these techniques cannot be used to demonstrate the behavior of CDOs under radically different economic circumstances. To better understand the behavior of CDOs under different economic circumstances, numerical techniques such as Monte Carlo simulation can be used instead of analytical techniques to price CDOs. Andersen et al (2005) proposed a method for calculating the probability of defaults that could then be used in the Monte Carlo simulation to price the collateralized debt obligation. The research proposed by Andersen et al (2005) demonstrates the process of calculating correlated probability of defaults for a group of obligors. This calculation is based on the correlations between the obligors using copulas. Using this probability of default, the price of a collateralized debt obligation can be evaluated using Monte Carlo simulation. Monte Carlo simulation provides a more simple yet effective approach compared to analytical pricing techniques. Simulation also allows investors to have a better understanding of the behaviors of CDOs compared to analytical pricing techniques. By analyzing the various behaviors under uncertainty, it can be observed how a downturn in the economy could affect CDOs. This thesis extends on the use of copulas to simulate the correlation between obligors. Copulas allow for the creation of one joint distribution using a set of independent distributions thus allowing for an efficient way of modeling the correlation between obligors. The research contained within this thesis demonstrates how Monte Carlo simulation can be used to effectively price collateralized debt obligations. It also shows how the use of copulas can be used to accurately characterize the correlation between obligor defaults for pricing collateralized debt obligations. Numerical examples for both the obligor defaults and the price of collateralized debt obligations are presented to demonstrate the results using Monte Carlo simulation.

I re-examine “rating shopping” and “rating catering” in the market for AAA rated collateralized debt obligations (CDOs) by replicating the study of Griffin and Tang (2013) using post-crisis data. I find a sharp increase in the amount of CDOs that received a single rating, suggesting that CDO underwriters were more cautious about formally soliciting multiple ratings. However, I also find a decrease in AAA rating disagreements between S&P and Moody’s, implying that issuers shopped their CDOs through informal conversations with agencies. Finally, I find investors correctly accepted tighter credit spreads for dual-rated CDOs because dual-rated CDOs experienced fewer rating downgrades than single rated deals. These results differ from the pre-crisis findings of Griffin and Tang (2013) and are consistent with the existence of rating shopping and disappearance of rating catering during the post-crisis period.

我國財務會計準則委員會於2004年7月1日發佈財務會計準則第35號公報《資產減損之會計處理準則》，此號公報允許公司可選擇於2004年提前認列資產減損，且公司擁有認列資產減損金額之裁量權。當公司擁有認列資產減損時點與金額的裁量權時，資產減損之認列可能成為一項新的盈餘管理工具。由於我國公司的代理問題是起因於控制股東剝削外部股東之財富，故本研究擬探討控制股東之股權結構與公司提前認列資產減損之關聯性。 / 本研究以2004年年報與2005年第一季季報宣告資產減損的上市櫃公司為研究對象。研究目的有三，首先探討控制股東之控制權偏離現金流量權引發的代理問題，是否會影響公司提前認列資產減損的可能性與金額。其次探討公司的控制股東持股質押的現象，是否會加深公司的代理問題，因而影響公司提前認列資產減損的可能性與認列減損的金額。最後本研究將探討市場對於宣告資產減損資訊的反應，是否會受到公司治理結構不同所影響。 / 研究結果發現，控制股東之控制權偏離現金流量權幅度愈大的公司，提前認列資產減損的可能性愈大，但是與提前認列資產減損的金額無顯著的關聯性。其次，控制股東有持股質押的公司，提前認列資產減損的可能性與金額皆愈大。最後，本研究並未發現股票市場對於不同公司治理結構的公司宣告資產減損的訊息有不同的反應。 / Statement of Financial Accounting Standards (SFAS) No.35 “Accounting for Assets Impairments” was released on July first of 2004. The provision of this Statement should be effective for quarterly financial statements ending on or after first quarter of 2005, but earlier application on the forth quarter of 2004 is encouraged. In Taiwan, the agency problem arises from the divergence between controlling owner’ voting right and cash flow right. Thus the regulation on the early adoption provides controlling owner with stronger incentive to manage reported earnings by manipulating the recognition timing and amount of write-off of impaired assets. / The primary objective of this paper is examine the association ownership structures, measured as the control divergence and the percentage of collateralized stock by controlling owners, and the likelihood of the early adoption of SFAS No. 35 and amount of assets impairments. Second, we further examine whether market valuation on assets impairment systematically varies depending on the agency problem level. / The results show that, as predicted, firms with greater control divergence and greater percentage of collateralized stock by controlling owners are more likely to early recognize impartment of assts. In addition, we find that the amount of asset impairment is associated with the percentage of collateralized stock by controlling owners, as predicted, but is not related to the control divergence. Finally, contrary to our prediction, we find no evidence that the market reaction to the announcement of asset impairments is significantly associated with the corporate governance structures.

Background: The financial crisis of 2007-2010 had a massive impact on the financial markets worldwide. The crisis was partly blamed on the credit derivatives collateralized debt obligations and credit default swaps. These instruments were used to create leverage and speculation, which led to uncertainty in the financial system worldwide. There has been no recent documentation of how credit derivatives are used in Swedish banks, and what risks and opportunities they bring along. Purpose: The purpose of this thesis is to describe the use of credit derivatives in Swedish banks, what benefits and risks they may generate and how the recent financial crisis has affected their use. Research Method: This is a qualitative multiple case study which uses an inductive approach. The study covers four cases, three of the largest Swedish commercial banks, and a bank that specializes on international financing. Seven people working in different fields in these banks have been interviewed. Conclusions: Credit derivatives are mostly used for hedging in Swedish banks, which mainly involves the use of credit default swaps, and sometimes iTraxx. Purely speculative trades are rare. The risks that arise are mainly due to lack of transparency in OTC trading, and abusive use of these instruments. Credit derivatives greatly facilitate risk management in banks. Regulations have increased since the financial crisis and the demand for more complex products greatly decreased.