none

One of the most important government policies to support and satisfy financing needs for marginal enterprises or special sectors in economic system is to provide credit guarantee. In Taiwan, while Small and Medium Enterprise Credit Guarantee Fund of Taiwan (SMEG) had been established to help small and medium enterprises (SMEs) acquiring bank loans successfully by providing credit guarantee, there is still a need to set up an appropriate credit rating systems for SMEs. This research proposes three kinds of assessment models to the credit risks of SMEG. While Model one employs a firm¡¦s financial performance, substituting debt level and estimated asset value and volatility into the model to derive probability of default (PD). Model two and three utilize a firm¡¦s risk premium observed from the loan rate to estimate credit level. The former belongs to the application of structure-form approach in the credit risk management model, on the other hand, the latter is the reduced-form approach. On the structure-form approach, due to the difficulties in accessing SMEs¡¦ public trade information in Taiwan, we adopt the Private Firm Model developed by Moody's KMV Company. We had also improved this PD evaluation model by taking some peculiar operating characteristics of Taiwan¡¦s SMEs into consideration. On the reduced-form approach, we apply risk-neutral model to estimate a firm¡¦s PD, which then been utilizing to evaluate the expected value of subrogation payment in the case of default. This can further go deeper to calculate the guarantee fee of a loan. The processes used in this model is same as that of actuarial methodology being used to determine the premium of a term insurance. The three credit risk management models proposed in this research are designed to reflect the market information of a SME, and to the applicability of operating in real world case. The empirical results indicate they could adequately reflect the risk levels of the SMEs to a certain extent. We hope to provide the SMEG with a method of evaluating credit risk of SMEs to establish a fairer and more reasonable guarantee fee, and contribute in enhancing and managing credit guarantee mechanism in Taiwan.

Master of Agribusiness / Department of Agricultural Economics / Christine Wilson / Credit scoring is a tool used to make lending decisions. AgChoice Farm Credit has a dealer financing program called Farm Credit Express that makes lending decisions based on a scoring model. Farm Credit Express is a dealer financing option for farm equipment purchases. AgChoice has generated significant loan volume with this program but has also experienced challenges with loan delinquencies as field staff must service loans that they did not originate. This thesis evaluates loan delinquencies within AgChoice Farm Credit’s Farm Credit Express (“FCE”) program. The thesis develops a regression model that includes delinquencies as the dependent variable and Total AgChoice Borrowing, Original Loan Amount, Farming Segment, CBI Score, AgScore, and FCE Only as the independent variables. The model provides an examination of AgChoice’s Farm Credit Express delinquencies and evaluates the variables mentioned above and their ability to predict delinquencies. The results showed that Total AgChoice Borrowing, Original Loan Amount, CBI Score and FCE only were statistically significant independent variables. Based on results of the model, recommendations were made to potentially reduce future delinquencies in the Farm Credit Express loan portfolio.

An important part when managing credit risk is to assess the probability of default of different counterparties. Increases and decreases in such probabil- ities are central components in the assessment, and this is where transition matrices become useful. These matrices are commonly used tools when as- sessing counterparty credit risk, and contain the probability of default, as well as the probability to migrate between different predefined rating classifica- tions. These rating classifications are used to reflect the risk taken towards different counterparties. Therefore, it is important for financial institutions to develop accurate transition matrix models to manage predicted changes in credit risk exposure. This is because counterparty creditworthiness and prob- ability of default indirectly affect expected loss and the capital requirement of held capital. This thesis will analyze how two specific models perform when used for generating transition matrices. These models will be tested to investigate their performance when predicting rating transitions, including probability of default. / En viktig del vid hanteringen av kreditrisk är att bedöma sannolikheten för fallissemang för olika motparter. Ökningar och minskningar i dessa sanno- likheter är centrala komponenter i bedömningen, och det är här migrations- matriser blir användbara. Dessa matriser är vanligt förekommande verktyg vid bedömning av kreditrisk mot olika motparter och innehåller sannolikheten för fallissemang samt sannolikheten att migrera mellan olika fördefinierade be- tygsklassificeringar. Dessa betygsklassificeringar används för att återspegla den risk som tas mot olika motparter. Det är därför viktigt för finansinstitut att utveckla träffsäkra migrationsmatris modeller för att hantera förväntade förändringar i kreditriskexponering. Detta beror på att kreditvärdigheten hos motparter samt sannolikheten för fallissemang indirekt påverkar expected loss och kapitalkrav. Detta examensarbete kommer att analysera hur två specifika modeller presterar när de används för att generera migrationsmatriser. Dessa mod- eller kommer att testas för att undersöka hur de presterar när de används för att förutsäga övergångar inom betygsklassificering, inklusive sannolikheten för fallissemang.

Having just survived what is arguably the worst financial crisis in time, it is expected that the focus on regulatory capital held by financial institutions such as banks will increase significantly over the next few years. The probability of default is an important determinant of the amount of regulatory capital to be held, and the accurate calibration of this measure is vital. The purpose of this study is to propose the use of the Shannon entropy when determining the parameters of the prior bivariate beta distribution as part of a Bayesian calibration methodology. Various bivariate beta distributions will be considered as priors to the multinomial distribution associated with rating categories, and the appropriateness of these bivariate beta distributions will be tested on default data. The formulae derived for the Bayesian estimation of Shannon entropy will be used to measure the certainty obtained when selecting the prior parameters. / Dissertation (MSc)--University of Pretoria, 2010. / Statistics / unrestricted

The main purpose of this paper is to provide evidence of financial leverage, liquidity, profitability, and firm size ability to predict bankruptcy of Swedish small and medium-sized enterprises (SMEs), and to create a bankruptcy prediction model for Swedish SMEs. The sample consists of 1086 Swedish SMEs, among which 543 did go bankrupt between 2015 and 2019. The paper employs logistic regression and Mann-Whitney U-test to test the hypotheses. The independent variables are derived from previous research and further filtered in a selection process, resulting in a final set of six variables. Financial leverage, liquidity, profitability, and firm size is found to have significantly predictive abilities to determine SME bankruptcy. The model has an overall classification accuracy of 77.6% out-of-sample and is able to classify 82.2% of the bankruptcies correctly out-of-sample.

Master of Agribusiness / Department of Agricultural Economics / Allen M. Featherstone / Institutions of the Farm Credit System (FCS) focus on risk-based lending in accordance with regulatory direction. The rating of risk also assists retail staff in loan approval, risk-based pricing, and allowance decisions. FCS institutions have developed models to analyze financial and related customer information in determining qualitative and quantitative risk measures. The objective of this thesis is to examine empirical account data from 2006-2012 to review the probability of default (PD) rating methodology within the overall risk rating system implemented by a Farm Credit System association. This analysis provides insight into the effectiveness of this methodology in predicting the migration of accounts across the association’s currently-established PD ratings where negative migration may be an apparent precursor to actual loan default. The analysis indicates that average PD ratings hold relatively consistent over the years, though the distribution of the majority of PD ratings shifted to higher quality by two rating categories over the time period. Various regressions run in the analysis indicate that the debt to asset ratio is most consistently statistically significant in estimating future PD ratings. The current ratio appears to be superior to working capital to gross profit as a liquidity measure in predicting PD rating migration. Funded debt to EBITDA is more effective in predicting PD rating movement as a measure of earnings to debt than gross profit to total liabilities, although the change of these ratios over time appear to be weaker indicators of the change in PD rating potentially due to the variable nature of annual earnings of production agriculture operations due to commodity price volatility. The debt coverage ratio is important as it relates to future PD migration, though the same variability in commodity price volatility suggests the need implement multi-year averaging for calculation of earnings-based ratios. These ratios were important in predicting the PD rating of observations one year into the future for production agriculture operations. To further test the predictive ability of the PD ratings, similar regression analyses were completed comparing current year rating and ratios to future PD ratings beyond one year, specifically for three and five years. Results from these regression models indicate that current year PD rating and ratios are less effective in predicting future PD ratings beyond one year. Furthermore, because of the variation in regression results between the analyses completed for one, three and five years into the future, it is important to regularly capture ratio and rating information, at least annually.

Markov chains have been widely used to the credit risk measurement in the last years. Using these chains we can model movements and distribution of clients within rating grades. However, various types of markov chains could be used. The goal of the theses is to present these types together with their advan- tages and disadvantages. We focus our attention primarily on various parameter estimation methods and hypotheses testing about the parameters. The theses should help the reader with a decision, which model of a markov chain and which method of estimation should be used for him observed data. We focus our attention primarily on the following models: a discrete-time markov chain, a continuous-time markov chain (we estimate based on continuous- time observations even discrete-time observations), moreover we present an even- tuality of using semi-markov chains and semiparametric multiplicative hazard model applied on transition intensities. We illustrate the presented methods on simulation experiments and simu- lation studies in the concluding part. Keywords: credit risk, markov chain, estimates in markov chains, probability of default 1

In many applications, the success probability 𝜋 of a Bernoulli distributed variable 𝑌 is influenced by another variable 𝑋. For example for loans granted, it is necessary to rate debtors in different rating classes, where the probability of default (PD) 𝜋 of 𝑌 is assumed to be homogeneous within and heterogeneous between the rating classes. The PD of a debtor is largely influenced by macroeconomic and individual variables (𝑋). In this work, we study a Bernoulli mixture model for 𝑌, where the success probability of 𝑌 changes systematically at the breakpoints. We focus on cross-sectional data and our main objective is to estimate all 𝑘 breakpoints with 𝑘 either known or unknown and their corresponding success probabilities between each pair of neighbouring breakpoints. To the best of our knowledge, an estimator for estimating multiple breakpoints has not yet been developed in this context. Thus, we develop an approach with a view to closing this research gap. We show that our estimator works for independent and identically distributed (i.i.d.) 𝑋 as well as for a linear one-factor model for 𝑋. A theoretical foundation for this estimator is also presented. In practice, the number of breakpoints 𝑘 is often unknown a priori. As the multiple estimator is based on an iterative procedure, we propose stopping criteria for estimating 𝑘 correctly. We conduct a simulation study in the context of credit rating to demonstrate the performance of the developed estimator. Furthermore, we apply the new estimator on credit risk data from the Sächsische Aufbaubank, the Development Bank of Saxony. To simplify the use of the new estimator, we also develop an R package called MultipleBreakpoints.

This thesis aims to assist in the development of machine learning models tailored for stress testing. The main objective is to create models that can predict loan defaults while considering the impact of macroeconomic stress. By achieving this, Nordea can continue the development of machine learning models for stress testing by utilizing the models as a basis for further advancement. The research begins with an analysis of historical loan data, encompassing diverse customer and macroeconomic variables that influence loan default rates. Leveraging machine learning algorithms, feature selection methods, data imbalance management and model training techniques, a set of predictive models is constructed. These models aim to capture the intricate relationships between the identified variables and loan defaults, ensuring their suitability for stress testing purposes. The subsequent phase of the research focuses on subjecting the developed models to simulated adverse economic conditions during stress testing. By evaluating the models’ performance under various stressed scenarios, their ability to provide predictions is assessed. This stress testing process allows us to analyse the models’ capabilities of incorporating a stressed scenario in their predictions. The thesis concludes with an evaluation of the developed machine learning models and their abilities to identify defaulted loans in a stressed macroeconomy. By creating these models specifically tailored for stress testing loans, we will provide a basis for further development within the area of stress testing modeling. / Denna uppsats syftar till att bidra till utvecklingen av maskininlärningsmodeller lämpade för stress testing. Det främsta målet är att skapa modeller som kan förutsäga lån som kommer att misslyckas samtidigt som de beaktar påverkan av makroekonomisk stress. Genom att uppnå detta kan Nordea fortsätta utvecklingen av maskininlärningsmodeller för stress testning genom att använda modellerna som grund för ytterligare utveckling. Arbetet inleds med en analys av historisk lånedata, som omfattar olika kund- och makroekonomiska variabler som påverkar lån. Genom att använda oss av maskininlärningsalgoritmer, metoder för urval av förklarande variabler, hantering av dataobalans och tekniker för modellträning konstrueras en uppsättning prediktiva modeller. Dessa modeller syftar till att fånga de komplexa relationerna mellan de identifierade variablerna och låneavvikelser och säkerställa deras lämplighet för stress testning. Den efterföljande fasen av arbetet fokuserar på att utsätta de utvecklade modellerna för simulerade stressade ekonomiska förhållanden. Genom att utvärdera modellernas prestanda under olika stressade förhållanden bedöms deras förmåga att prediktera uteblivna lån. Denna process för stress testning gör det möjligt för oss att analysera modellernas förmåga att inkludera stressade förhållanden i sina prediktioner. Uppsatsen avslutas med en utvärdering av de utvecklade maskininlärningsmodellerna och deras förmåga att identifiera uteblivna lån i en stressad makroekonomi. Genom att skapa dessa modeller specifikt anpassade för stresstestning av lån kommer vi att ge en grund för ytterligare utveckling inom området.