none

none

In this paper, we try to compare the performance of two feature dimension reduction methods, the LASSO and PCA. Both simulation study and empirical study show that the LASSO is superior to PCA when selecting significant variables. We apply Logistics Regression (LR), Artificial Neural Network (ANN), Support Vector Machine (SVM), Decision Tree (DT) and their corresponding ensemble machines constructed by bagging and adaptive boosting (adaboost) in our study. Three experiments are conducted to explore the impact of class-unbalanced data set on all models. Empirical study indicates that when the percentage of performing loans exceeds 83.3%, the training models shall be carefully applied. When we have class-balanced data set, ensemble machines indeed have a better performance over single machines. The weaker the single machine, the more obvious the improvement we can observe.

Abstract With the liberalization and globalization, the scope for banks is much more sophisticated in decade. Accompanied with Asian Financial Crisis, the Non-Performing Loan (NPL) ratio of domestic financial institutions has increased significantly. Consequently, this research adopts 2SLS to estimate simultaneous equations and examines the impact of risk-based capital regulation on NPL ratio and operating performance. The empirical results are summarily as following: 1. There exists a negative relationship between capital adequacy ratio and NPL ratio, which means that the higher the capital adequacy ratio is, the lower the NPL loan will be. Therefore, adopting the capital regulation will be helpful to improve the balance-sheet structure. 2. With the change of capital adequacy ratio, ROE moves in the same direction; that is, there is a positive relationship between capital adequacy ratio and ROE. Usually, the high-profit companies have more cash flow to support the capital requirement. 3. Examining the effectiveness of ¡§First-Stage Financial Reform Policy¡¨, we find that only 12% and 26% of the banks are not qualified for capital adequacy ratio and NPL ratio, respectively.. Thus, the ¡§First-Stage Financial Reform Policy¡¨ has achieved the preliminary goal. 4. We adopt t-test to distinguish whether the difference between actual and standard figures is significantly large for those disqualified banks. It shows that the main factor might be because of the essential problem of bad asset-debt structure, and not the lack of time to adapt themselves to the new regulation.

Este trabalho apresenta o desenvolvimento e aplicação de espécimes SE(T) em aço API X60 para avaliação de integridade estrutural em dutos condutores de gás e petróleo, o qual consiste na implantação de três técnicas que forneçam as curvas de resistência à propagação dútil de trincas, curvas J-Rs. As metodologias utilizadas foram a da variação da flexibilidade elástica, da queda de potencial elétrico e da normalização linear da carga, nas quais duas análises são verificadas para obtenção da área plástica abaixo da curva de carregamento. A primeira, através da medida do deslocamento da linha de carga e a segunda, pelo deslocamento do extensômetro colocado na boca da trinca. Foram ensaiados corpos de prova SE(T) com trincas rasa (a/W = 0,2) e profunda (a/W = 0,5) e os resultados obtidos permitiram observar a iniciação da propagação por rasgamento dúctil e discutir os limites das metodologias aplicadas. Os resultados mostraram que é possível obter as curvas J-Rs medindo o deslocamento na boca da trinca, podendo-se converter os deslocamentos de CMOD em deslocamentos do LLD. Foi constatada também que a técnica de normalização linear da carga é a que possibilitou o menor tempo de preparação e execução do ensaio, por utilizar apenas a medição do CMOD e não necessitar da realização dos descarregamentos. A variação da flexibilidade elástica é uma técnica rápida e confiável e não sofre alterações devido às delaminações. Já a técnica de queda do potencial elétrico se mostrou a mais consistente para o caso de a/W = 0,5, porém no caso de trincas curtas, a plasticidade desenvolvida na frente da trinca influencia nas medições causando imprecisões. As curvas J-Rs são deslocadas para a esquerda quando o tamanho inicial de trinca é reduzido, isto é, ocorre um aumento na energia de iniciação da propagação, JIC. / This work presents the development and application of SE(T) type specimen for structural integrity evaluation of pipeline steels used for gas and oil transportation, and it consists of the implementation of different techniques to provide the crack growth resistance curves, J-R curves. The methodologies used for crack growth measurement were: the elastic compliance, the electrical potential drop and the linear normalization techniques. Two techniques to provide the plastic area, Apl under the load versus displacement curve were used to evaluate Jpl, one based on the load line displacement, LLD, and the other based on the crack mouth opening displacement, CMOD. SE(T) specimens with two different initial crack size, denominate as shallow crack (a0/W = 0.2) and deep crack (a0/W = 0.5). The obtained results allowed the determination of the ductile tearing initiation point and to discuss the limits of the methodologies applied. It was proved be possible to get J-R curves using only one extensometer to measure the CMOD, which provides both crack size and Apl. The linear normalization technique is the one that provided the lower preparation and testing time, since there is no need of multiples unloading for crack growth measurement. The elastic compliance is a very easy, fast and reliable technique, and it is not sensible to the effect of the occurrence of delaminations, as it was the linear normalization technique. The electrical potential drop technique proved to very consistent for a0/W = 0.5, however in the case of shallow cracks the developed plasticity at the crack tip affected the measurements. The J-R curves were displaced to the left as the initial crack size is reduced, this is, there is an increase in the energy for crack tearing initiation, JIC.

Este trabalho apresenta o desenvolvimento e aplicação de espécimes SE(T) em aço API X60 para avaliação de integridade estrutural em dutos condutores de gás e petróleo, o qual consiste na implantação de três técnicas que forneçam as curvas de resistência à propagação dútil de trincas, curvas J-Rs. As metodologias utilizadas foram a da variação da flexibilidade elástica, da queda de potencial elétrico e da normalização linear da carga, nas quais duas análises são verificadas para obtenção da área plástica abaixo da curva de carregamento. A primeira, através da medida do deslocamento da linha de carga e a segunda, pelo deslocamento do extensômetro colocado na boca da trinca. Foram ensaiados corpos de prova SE(T) com trincas rasa (a/W = 0,2) e profunda (a/W = 0,5) e os resultados obtidos permitiram observar a iniciação da propagação por rasgamento dúctil e discutir os limites das metodologias aplicadas. Os resultados mostraram que é possível obter as curvas J-Rs medindo o deslocamento na boca da trinca, podendo-se converter os deslocamentos de CMOD em deslocamentos do LLD. Foi constatada também que a técnica de normalização linear da carga é a que possibilitou o menor tempo de preparação e execução do ensaio, por utilizar apenas a medição do CMOD e não necessitar da realização dos descarregamentos. A variação da flexibilidade elástica é uma técnica rápida e confiável e não sofre alterações devido às delaminações. Já a técnica de queda do potencial elétrico se mostrou a mais consistente para o caso de a/W = 0,5, porém no caso de trincas curtas, a plasticidade desenvolvida na frente da trinca influencia nas medições causando imprecisões. As curvas J-Rs são deslocadas para a esquerda quando o tamanho inicial de trinca é reduzido, isto é, ocorre um aumento na energia de iniciação da propagação, JIC. / This work presents the development and application of SE(T) type specimen for structural integrity evaluation of pipeline steels used for gas and oil transportation, and it consists of the implementation of different techniques to provide the crack growth resistance curves, J-R curves. The methodologies used for crack growth measurement were: the elastic compliance, the electrical potential drop and the linear normalization techniques. Two techniques to provide the plastic area, Apl under the load versus displacement curve were used to evaluate Jpl, one based on the load line displacement, LLD, and the other based on the crack mouth opening displacement, CMOD. SE(T) specimens with two different initial crack size, denominate as shallow crack (a0/W = 0.2) and deep crack (a0/W = 0.5). The obtained results allowed the determination of the ductile tearing initiation point and to discuss the limits of the methodologies applied. It was proved be possible to get J-R curves using only one extensometer to measure the CMOD, which provides both crack size and Apl. The linear normalization technique is the one that provided the lower preparation and testing time, since there is no need of multiples unloading for crack growth measurement. The elastic compliance is a very easy, fast and reliable technique, and it is not sensible to the effect of the occurrence of delaminations, as it was the linear normalization technique. The electrical potential drop technique proved to very consistent for a0/W = 0.5, however in the case of shallow cracks the developed plasticity at the crack tip affected the measurements. The J-R curves were displaced to the left as the initial crack size is reduced, this is, there is an increase in the energy for crack tearing initiation, JIC.

台灣政府為推動金融自由化，於1991年開放民營銀行之設立，另輔導信用投資公司及信用合作社轉制為商業銀行，公營銀行釋股轉為民營銀行，期待台灣金融業與國際接軌，提高全球的競爭力。因短時間內大量銀行的設立，相互競爭激烈，導致獲利下降，而競爭衍生不當放款，及財團惡意掏空銀行，種下未來之惡果，銀行逾期放款大幅增加，資本適足率不足，政府提出金融改革，展開台灣不良債權市場新的一頁。 台灣不良債權市場自2001年開始至今已邁入第13年，共出售新台幣1兆5,513億元，而不良債權市場高峰於2002年有2,500億，到2013年僅有新台幣93億，何種原因造成市場萎縮，經本研究蒐集彙整不良債權發展過程、相關資訊及法規，從不良債權相關法令與管制、金融機構授信品質提高及台灣經濟景氣與不動產景氣趨勢三項來探討，驗證不良債權市場的波動與金融機構授信品質及台灣總體經濟與不動產景氣呈反向關係，是造成台灣不良債權市場改變之原因，而相關法令之制訂與修改，促使不良債權市場成為健全台灣金融體系重要的一環。

This thesis brings a new approach to research of non-performing loans (NPLs), stepping away from the traditional macroeconomic and bank-specific focused literature by exploring the relationship between NPLs and natural disasters across all 50 US states, aiming to understand to what extent and how natural disasters affect financial instability as reflected by NPLs. Using fixed effects models with Beck and Katz robust standard errors the study analyses annual state-level NPL-ratio, a measure of NPLs to total loan portfolio value with natural disasters measured as the number of disasters, cost of disasters, and deaths from disasters. Macroeconomic variables are included as controls, the included variables are by personal income, unemployment rate, and tax collection. The studied period is 1984 to 2019. Support for the notion that the NPL ratio is affected by natural disasters is found, however, the results are conflicting as the number of disasters is found to increase the NPL ratio according to the theoretical expectation, however, not when interactions are added to the model. Whilst, contrary to previous research the number of deaths from disasters is found to decrease the NPL ratio. The results are thus deemed ambiguous. However, the thesis contributes to the field by highlighting the need for further research into similar research questions and providing a new approach to NPLs in the US context.

Studien undersöker sambandet mellan NPL och lönsamheten för de svenska bankerna under och efter Covid-19. Genom tillämpning av en kvantitativ metod och teoretisk ramverket gällande riskhantering och finansiella metoder ger studien insikter om NPL påverkan en banks lönsamhet. Teorier och finansiella modeller togs upp i den teoretiska referensramen för att bygga en grund för tidigare forskning som användes i senare tillfälle i studien. Teorierna berör kreditrisker samt likviditetsrisker där kreditrisk som teori användes som huvudfokus i studiens senare stadie. Anledningen till detta låg i att tidigare forskningen kopplade NPL och lönsamhet till kredit risker, därav blev den teorin mest relevant. Finansiella modellerna ROA, NPL, ROE, NIM, CAR samt Bank Size togs upp som är även de oberoende samt beroende variablerna som ställs mot varandra i metodens genomförande av beräknings och statistiska analyser. Studiens insamling av data på svenska bankers årsredovisningar mellan 2020 till 2022 skedde via databasen Retriever Business. Tillvägagångssättet skedde genom en deskriptiv analys med korrelationsanalys samt linjära regressioner för att identifiera de samband som kan finnas. Empirin visade på ett generellt svagt samband mellan NPL och lönsamhet. Däremot visade resultatet på andra faktorer som har en betydande roll i lönsamhetens påverkan vilket tyder på behov av en bredare analys. Resultatet visade överlag att det fanns en tydlig variation mellan 2020 till 2022 för både korrelation och linjära regressioner.

During the last decade, many European banks have been troubled with low profitability, while the amount of non-performing loans (NPLs) has increased. This thesis investigates and analyses how the increasing amount of NPL affects banks profitability and the financial system. With econometric models using panel data we examined the relationship between NPL, banks profitability and the economic cycle (GDP-growth). This combined with qualitative economic theories provided a solid analysis of this relationship. We found strong evidence the NPL-ratio has a negative correlation with both the profitability of banks and the economic cycle. With these results in mind we think the NPLs need to be dealt with by the banks and authorities soon. In accordance with our result and analysis we came up with recommendations for the banks and authorities to deal with the issue. We recognize they need to improve the secondary markets for non-performing loans, lifting the loans from their balance sheets, increase the use of Asset Management Companies and improve the NPL-management within banks.