As a result of an increasingly stringent regulation aimed at monitoring financial risk exposures, nowadays the risk measurement systems play a crucial role in all banks. In this thesis we tackle a variety of problems, related to density forecasting, which are fundamental to market risk managers. The computation of risk measures (e.g. Value-at-Risk) for any portfolio of financial assets requires the generation of density forecasts for the driving risk factors. Appropriate testing procedures must then be identified for an accurate appraisal of these forecasts. We start our research by assessing whether option-implied densities, which constitute the most obvious forecasts of the distribution of the underlying asset at expiry, do actually represent unbiased forecasts. We first extract densities from options on currency and equity index futures, by means of both traditional and original specifications. We then appraise them, via rigorous density forecast evaluation tools, and we find evidence of the presence of biases. In the second part of the thesis, we focus on modelling the dynamics of the volatility curve, in order to measure the vega risk exposure for various delta-hedged option portfolios. We propose to use a linear Kalman filter approach, which gives more precise forecasts of the vega risk exposure than alternative, well-established models. In the third part, we derive a continuous time model for the dynamics of equity index returns from a data set of 5-minute returns. A model inferred from high-frequency typical of risk measures calculations. The last part of our work deals with evaluating density forecasts of the joint distribution of the risk factors. We find that, given certain specifications for the multivariate density forecast, a goodness-of-fit procedure based on the Empirical Characteristic Function displays good statistical properties in detecting misspecifications of different nature in the forecasts.

This thesis investigates two aspects of credit risk measurement in the context of Basel 11: The International Convergence of Capital Measurement and Capital Standards. The first is the problem arising when two credit rating agencies disagree over the rating assigned to an issuer and a split rating arises. The second area is the determination of internal credit rating models for use under the Internal ratings-based approach. This thesis presents a variety of bank rating modes for individual and long term ratings across different agencies and regions. Using an extensive database of credit rating agencies with a sample of over 52,000 split ratings covering a four year period from 1999 - 2004 the first study shows that there is a ranking of agencies from the most to least generous that is stable over time. In most cases, the differences between the mean ratings of the agencies are significantly different from each other at the 1% level. The greatest differences arise between the US and Japanese agencies. When the split ratings are compared in terms of Basel II risk weights the differences between the US and Japanese agencies are still highly significant and the conclusion is that supervisors should alter the mapping of the Japanese agencies to the risk assessments under the provisions of Annex 2 to Basel II. Contrary to earlier research this study does not find that the highest level of split ratings arise for banks. The level of consensus between agencies appears to correspond to the average credit quality of the industry in question. Bank credit ratings are modelled from financial ratios and variables using ordinal logistic regression. Sample sizes exceeded 1,100 banks for the largest agencies.

The public-private partnership (PPP) arrangements require the optimization of risk allocation between the public and private sectors in order to achieve the best net present value (NPV). Many researchers mentioned that the risk events of a PPP infrastructure projects are interdependent over project life cycle. Sterman (1992) stated that a large-scale construction project that is complex and has highly dynamic and interdependent risks and uncertainties over long-term project life cycle. Williams (2002) also mentioned that the risk usually interact each other with nonlinear relationships over time in a complex project. Dey and Ogunlana (2004) contended that there is a need to analyze risk interactions of complex infrastructure projects such as build-operate-transfer (BOT) projects over their long-term project life. In modern approaches to PPP project risk management, experts assume risk factors are independent and ignore the risk interaction effects over project life cycle, so the project risks cannot be effectively managed and controlled. The researcher proposed a modelling approach that used a risk network model applying System Dynamics (SD) techniques to estimate risk interaction effects on project NPV over time. The researcher used another SD model built on the risk network model to estimate the beneficial effects of bidding proposals on project NPV over time and to see how efficiently the risk effects can be reduced and the NPV performance can be improved. Then, the researcher applied appropriate stochastic analyses including mean-variance, mean semi-variance, stochastic dominance and expected-loss ratio to compare range values of NPV among different bidding proposals. A capable PPP concessionaire with the best project NPV performance can hence be selected. An industry case was applied to demonstrate SD decision models. The SD decision models have been validated through the behaviour reproduction test and multivariate sensitivity analysis. This proved that the proposed approach is robust and applicable to address real world problems to evaluate the longterm performance of a PPP project concessionaire

Arguably, the credit risk models reported in the literature for the retail lending sector have so far been less developed than those for the corporate sector, mainly due to the lack of publicly available data. Having been given access to a dataset on defaulted mortgages kindly provided by a major UK bank, this work first investigates the Loss Given Default (LGD) of mortgage loans with the development of two separate component models, the Probability of Repossession (given default) Model and the Haircut (given repossession) Model. They are then combined into an expected loss percentage. Performance-wise, this two-stage LGD model is shown to do better than a single-stage LGD model (which directly models LGD from loan and collateral characteristics), as it achieves a better Rsquare value, and it more accurately matches the distribution of observed LGD. We next investigate the possibility of including macroeconomic variables into either or both component models to improve LGD prediction. Indicators relating to net lending, gross domestic product, national default rates and interest rates are considered and the interest rate is found to be most beneficial to both component models. Finally, we develop a competing risk survival analysis model to predict the time taken for a defaulted mortgage loan to reach some outcome (i.e. repossession or non-repossession). This allows for a more accurate prediction of (discounted) loss as these periods could vary from months to years depending on the health of the economy. Besides loan- or collateral-related characteristics, we incorporate a time-dependent macroeconomic variable based on the house price index (HPI) to investigate its impact on repossession risk. We find that observations of different loan-to-value ratios at default and different security type are affected differently by the economy. This model is then used for stress test purposes by applying a Monte Carlo simulation, and by varying the HPI forecast, to get different loss distributions for different economic outlooks.

The Basel II accord regulates risk and capital management requirements to ensure that a bank holds enough capital proportional to the exposed risk of its lending practices. Under the advanced internal ratings based (IRB) approach, Basel II allows banks to develop their own empirical models based on historical data for probability of default (PD), loss given default (LGD) and exposure at default (EAD). This thesis looks at some examples of modelling LGD and PD. One part of this thesis investigates modelling LGD for unsecured personal loans. LGD is estimated through estimating Recovery Rate (RR, RR=1-LGD). Firstly, the research examines whether it is better to estimate RR or Recovery Amounts. Linear regression and survival analysis models are built and compared when modelling RR and Recovery Amount, so as to predict LGD. Secondly, mixture distribution models are developed based on linear regression and survival analysis approaches. A comparison between single distribution models and mixture distribution models is made and their advantages and disadvantages are discussed. Thirdly, it is examined whether short-term recovery information is helpful in modelling final RR. It is found that early payment patterns and short-term RR after default are very significant variables in final RR prediction models. Thus, two-stage models are built. In the stage-one model short-term Recoveries are predicted, and then the predicted short-term Recoveries are used in the final RR prediction models. Fourthly, macroeconomic variables are added in both the short-term Recoveries models and final RR models, and the influences of macroeconomic environment on estimating RR are looked at. The other part of this thesis looks at PD modelling. One area where there is little literature of PD modelling is in invoice discounting, where a bank lends a company a proportion of the amount it has invoiced its customers in exchange for receiving the cash flow from these invoices. Default here means that the invoicing company defaults, at which point the bank cannot collect on the invoices. Like other small firms, the economic conditions affect the default risk of invoicing companies. The aim of this research is to develop estimates of default that incorporate the details of the firm, the current and past position concerning the invoices, and also economic variables.

High demand of Liquefied Natural Gas (LNG) in recent time requires LNG carriers in more frequent operations in order to meet customers' needs. To ensure that the LNG carriers are always reliable in service, it has become necessary to adopt various advanced modelling techniques such as Genetic Algorithm (GA), fuzzy logic and Evidential Reasoning (ER) for risk/safety assessment and maintenance modelling of LNG carrier operations. These advanced computational techniques can help to overcome challenges posed by uncertainties associated with the LNG carrier operations. Their usefulness is demonstrated using case studies in this research. Firstly, two major hazards of LNG carrier operations such as "failure of LNG containment system" and "LNG spill from transfer arm" are identified and estimated as high risk ones using a risk matrix technique and expert judgement. The causes (failure modeslbasic events) of these high risk hazards are analysed using a Fault Tree Analysis (FTA). The failure logics of their failure modes are established and Boolean algebra is applied to facilitate the evaluation of the failure probabilities and frequencies. Secondly, a GA model is developed to improve the safety levels of the LNG containment system and transfer arm, to minimise their maintenance costs and to realise optimal resource management. The GA is used to optimise a risk model that is developed with exponential distribution and parameters such as failure frequencies, unit costs of maintenance and new maintenance costs of the LNG containment system and transfer arm. Thirdly, the uncertainties of some parameters in the GA model such as unit costs of maintenance are subdued using the strength of Fuzzy Rule Base (FRB) in combination with GA. 125 fuzzy rules of LNG carrier system maintenance cost are developed, which makes it possible to facilitate the evaluation of maintenance cost in any specific LNG risk-based operation. The outcomes of unit costs of maintenance are used in the GA based risk model to update the optimal management of maintenance cost. Finally, the uncertainties of failure modes of the LNG containment system and transfer arm are investigated and treated based on the Formal Safety Assessment (FSA) principle using a Fuzzy ER (FER) approach. The fuzzy logic is used to estimate the safety/risk levels of those failure modes while the ER is used to synthesise them to facilitate the estimation of safety/risk levels of the top events. Risk Control Options (RCOs) are developed to manage high level risks. The costs for each of the RCOs are estimated and synthesised using ER, which facilitated the investigation of the best RCOs in risk-based decision making. There is no doubt that the methodologies proposed possess significant potential for use in improving safety and maintenance of LNG carrier operations based on the verifications of their corresponding test cases. Accordingly, the developed models can be integrated to formulate a platform to facilitate risk assessment and maintenance management of LNG carrier systems in situations where traditional techniques cannot be applied with confidence.

A chemical supply chain (CSC) presents a network that integrates suppliers, manufacturers, distributors, retailers and customers into one system. The hazards arising from the internal system and the surrounding environment may cause disturbances to material, information and financial flows. Therefore, supply chain members have to implement a variety of methods to prepare for, respond to and recover from potential damages caused by different kinds of hazards. A large number of studies have been devoted to extending the current knowledge and enhancing the implementation of chemical supply chain risk management (CSCRM), to improve both safety and reliability of the CSCRM systems. However, the majority of existing risk management methods fail to address the complex interactions and dynamic feedback effects in the systems, which could significantly affect the risk management outcomes. In order to bridge the gaps, a new CSCRM method based on System Dynamics (SD) is proposed to accommodate the need to describe the connections between risks and their associated changes of system behaviour. The novelty of this method lies not only on providing a valid description of a real system, but also on addressing the interactions of the hazardous events and managerial activities in the systems. In doing so, the risk effects are quantified and assessed in different supply chain levels. Based upon the flexibility of SD modelling processes, the model developer can modify the developed model throughout the model life cycle. Instead of directly assessing different risks and providing arbitrary decisions, the obtained numerical results can offer supportive information for assessing potential risk reduction measures and continuously improving the CSC system performance. To demonstrate the applicability of the newly proposed method, a reputed specialty chemical transportation service provider in China is used and analysed through modelling and simulating the chemical supply chain transportation (CSCT) operations in various scenarios. It offers policy makers and operators insights into the risk-affected CSC operations and CSCRM decision-making processes, thus helping them develop rational risk reduction decisions in a dynamic environment.

This study examines the notion that the term “Industrialisation” within the banking system is not clearly understood, nor its impact on the whole value added chain. The goal is to establish a clear definition of the term “Industrialisation” in an international context and study the manifestation and impact of Industrialisation across the length of the banking value added chain. Four indicators of Industrialisation (standardisation, automation, specialisation, quality management) were identified through a systematic literature review. The work focuses on one of the ‘three pillars’ of the German banking system: the East German Savings Banks Group. The research uses a homogenous multi method approach utilizing statistical financial information, existing documentary evidence and questionnaires. The data (quantitative and qualitative) was derived from files held by the national association on the 48 savings banks, and from 36 quantitative questionnaires returned by respondent banks. The 36 complete data sets were systematically combined using a comprehensive regression approach. The data was used to test three over-arching hypotheses, each relating to connections between the (generally understood) four stages of the value-added chain, activities related to each stage and indicators of banking success. The research clearly identified that: 1) Industrialisation dominates the savings banks value added chain. 2) Industrialisation augments financial outcomes and ‘perceived success’ in product development, marketing, settlement and transactions. 3) Outsourcing functions are negatively correlated to banking success in these value added stages. 4) Success in risk management was shown to be contingent on settlement and transactions, but no other activities. Automated services, such as self-service terminals and internet banking, are successful in the areas of settlements, transactions, marketing and customer relations. Increasing automation and standardisation can increase the perceived and quantitative measured success within the value added chain. Conclusions & Implications: The developed model extends knowledge in the area of banking and Industrialisation, showing increasing interaction between stages along the value-added chain. The closer the stages, the stronger the effects. The model provides a guide for managerial attention in adding value through Industrialisation techniques in the industry. The management implications of the study are that the savings banks should focus on their core competencies in providing a holistic in-house service in routine transactions, as well as supporting exceptional financing and investment tasks for their clients. To enhance the efficiency of Industrialisation across the value added chain, savings banks should find standards and routines contributing to Industrialisation success in risk management, and seek to comprehensively link the function of risk management to the value added chain stages.

No description available.

This thesis presents Organisational Sustainability Modelling (OSM), a new method to model and analyse risk and return systematically for the adoption of large systems such as Cloud Computing. Return includes improvements in technical efficiency, profitability and service. Risk includes controlled risk (risk-control rate) and uncontrolled risk (beta), although uncontrolled risk cannot be evaluated directly. Three OSM metrics, actual return value, expected return value and risk-control rate are used to calculate uncontrolled risk. The OSM data collection process in which hundreds of datasets (rows of data containing three OSM metrics in each row) are used as inputs is explained. Outputs including standard error, mean squared error, Durbin-Watson, p-value and R-squared value are calculated. Visualisation is used to illustrate quality and accuracy of data analysis. The metrics, process and interpretation of data analysis is presented and the rationale is explained in the review of the OSM method. Three case studies are used to illustrate the validity of OSM: • National Health Service (NHS) is a technical application concerned with backing up data files and focuses on improvement in efficiency. • Vodafone/Apple is a cost application and focuses on profitability. • The iSolutions Group, University of Southampton focuses on service improvement using user feedback. The NHS case study is explained in detail. The expected execution time calculated by OSM to complete all backup activity in Cloud-based systems matches actual execution time to within 0.01%. The Cloud system shows improved efficiency in both sets of comparisons. All three case studies confirm there are benefits for the adoption of a large computer system such as the Cloud. Together these demonstrations answer the two research questions for this thesis: 1. How do you model and analyse risk and return on adoption of large computing systems systematically and coherently? 2. Can the same method be used in risk mitigation of system adoption? Limitations of this study, a reproducibility case, comparisons with similar approaches, research contributions and future work are also presented.