Overview of the research programme

The six ESR projects are subdivided into 4 work packages (WPs). These WPs represent the common stages in industrially--oriented mathematical research for financial as well as insurance applications. There is the "Modeling Stage" (WP1) in which stochastic models for the dynamics of the uncertain underlying factors (assets, interest rates, liabilities, etc.) are selected and investigated. The choice of stochastic dynamics has a major impact on the valuation of financial products and insurance portfolios, as well as on the assessment of the risks associated with them. WP2 is therefore devoted to "Risk". Here, we cover the different types of risk that we encounter in the ESR projects, like market risk, credit risk and, in particular, model risk. Specific modeling choices may lead to highly efficient pricing techniques for corresponding financial products and portfolios. However, for involved stochastic dynamics in industrial applications, it is a major challenge to develop efficient pricing and accurate risk measurement. This is the contents of WP3 "Efficient Algorithms". In order to significantly reduce the associated computational costs, as demanded by industry, WP4 "Software and Hardware" is then connected to the implementation of the efficient algorithms on state-­of-­the-­art hardware, like on Graphics Processing Units (GPUs). At the same time, WP4 covers software aspects, like specific programming languages, definition of a suitable software framework, and user interfaces as desired by the industrial beneficiaries. Together, these four WPs will give the ESRs a fairly complete overview of aspects of risk management in the financial and insurance industries. Each ESR will be working on each of the work packages.

Each of the ESR projects contains original and challenging research aspects, on the interface of academic and industrial research. The industrial beneficiaries defined interesting ESR research topics, related to the Basel III and Solvency II accords. The research projects are on highly relevant topics for the industry, and will at the same time lead to PhDs. We would like to emphasize the experience and the strong links of each of the academic beneficiaries with industry and business projects.

ESR1: The use of asset liability management (ALM) for the valuation of large portfolios of insurance companies involves high computational costs and requires sophisticated stochastic modeling and appropriate calibration. Mainly focusing on life insurance portfolios, the innovative aspects will improve the long-term modeling and portfolio optimization, reinsurance and managing strategies. The risk analysis with respect to optimal portfolios represents an important original feature (wrt risk measures). The efficient implementation of numerical algorithms on GPUs will provide a highly competitive software-­hardware toolbox to be transferred and exploited by the insurance sector.

ESR2: Financial model risk, i.e. the risk related to the choice of stochastic model for assets, interest rates, FX, etc., has the potential of severe consequences of banking decisions based on incorrect or misused model outputs and reports. The main aim is the design of a general and theoretical framework for model risk management. Neglecting model uncertainty may lead to large losses and/or reputational costs. Model risk management is a relatively new area of risk management receiving a growing amount of regulatory scrutiny. It will be applied to real data, obtaining an inventory for different settings, computation of credit value adjustment (CVA) and CVA Value at Risk (CVA VaR). Efficient valuation methods will be included in software packages.

ESR3: In the credit risk research, we will develop a digital library for the generation of scenarios for stochastic credit default models. Three main themes are identified (credit risk referring to a single entity, to multiple entities, and sovereign risk). For each subject the aim is too enrich existing models or creating novel models to give more realistic descriptions of markets. Some innovative aspects include:
— Improve models for the correlation between the CDS market (quoted in a foreign currency) and the market for sovereign bonds (quoted in domestic currency).
— Implementation of novel models of dependence between default events of multiple entities.
— Derivation of new closed-­form formulas for the probability of survival of an entity, for the prices of plain vanilla call options on equity and for the prices of a default free zero-­coupon within a single framework capable of reproducing market data.
— Introduction of stochastic dynamics for interest rates in an advanced stochastic model with jumps, which would be a considerable step forward with respect to the modeling of the credit risk of a single entity.

ESR4: In the industrial sector, the belief grows that the current pace of increase in the computations of risk measures by the financial community is not sustainable, as it relies only on fast simulation and information technology. The objective of this research is therefore to move the center of gravity of innovation from computer science aspects towards methodological development, in order to better exploit the computing power already available. For this purpose we look at risk assessment and management for classical engineering applications that may provide useful insight for risk assessment in the financial industry. We think about processes and methodologies for different types of risk, as suggested by experience in different fields (e.g. nuclear industry, oil & gas, clinical, ...). Primarily, the estimation of the safety/reliability performance of technological systems, and more generally the development of reliability/risk-­centered systems for asset management, require the definition and solution of models with involved probability distributions for the aleatory variables, with "low probability" (i.e., "rare" events) and with causes-­effects relationships and dependencies among them. The aim is to develop improved analytic approximations, and at the same time efficient Monte Carlo methods for numerical simulations, also o GPU's to support insurance companies and banks in risk management.

ESR5: To develop a systematic hedge test, in a model setting, for a financial product, on the basis of a stochastic asset model, incorporating counterparty risk, is highly innovative. For this purpose, we require accurate models for all the components involved (like for counterparty risk, etc.), plus algorithms of the highest efficiency, for calibration and for Monte Carlo simulation. Implementation on GPU clusters is mandatory to reach our goal of a hedge test performance within reasonable time. The aim is to place the hedge test in the context of Backward Stochastic Differential Equations. In cooperation with all beneficiaries, we can make huge steps forward.

ESR6: In the market risk research topic, the innovative aspect is to develop accurate and efficient algorithms for estimating market risk in real life portfolios existing at banks. In the academic world highly efficient algorithms have been developed, however, mainly for estimating in the more quiet times that are now behind us. In banks and financial companies, market risk is commonly handled by empirical distributions and Monte Carlo simulations. We aim to combine the best of both worlds, working on hybrid valuation methods. Especially under stressed scenarios, we wish to contribute with academic insight risk measures. The aim is to develop a general software framework, which is in full accordance with the Basel requirements, and in which the industry can assess the quality of their data and policies in the context of the Basel accords.