Quantitative Finance

This paper develops a dynamic internal fraud model for operational losses in retail banking. It considers public operational losses arising from internal fraud in retail banking within a group of international banks. Additionally, the model takes into account internal factors such as the ethical quality of workers and the risk controls set by bank managers. The model is validated by measuring the impact of macroeconomic indicators such as GDP growth and the corruption perception upon the severity and frequency of losses implied by the model. In general,results show that internal fraud losses are pro-cyclical, and that country specific corruption perceptions positively affects internal fraud losses. Namely, when a country is perceived to be more corrupt, retail banking in that country will feature more severe internal fraud losses.

We show that, under mild assumptions, some unimaginable events - which we refer to as Black Swan events - must necessarily occur. It follows as a corollary of our theorem that any computational model of decision-making under uncertainty is incomplete in the sense that not all events that occur can be taken into account. In the context of decision theory we argue that this constitutes a stronger sense of uncertainty than Knightian uncertainty.

Firms should keep capital to offer sufficient protection against the risks they are facing. In the insurance context methods have been developed to determine the minimum capital level required, but less so in the context of firms with multiple business lines including allocation. The individual capital reserve of each line can be represented by means of classical models, such as the conventional Cramér-Lundberg model, but the challenge lies in soundly modelling the correlations between the business lines. We propose a simple yet versatile approach that allows for dependence by introducing a common environmental factor. We present a novel Bayesian approach to calibrate the latent environmental state distribution based on observations concerning the claim processes. The calibration approach is adjusted for an environmental factor that changes over time. The convergence of the calibration procedure towards the true environmental state is deduced. We then point out how to determine the optimal initial capital of the different business lines under specific constraints on the ruin probability of subsets of business lines. Upon combining the above findings, we have developed an easy-to-implement approach to capital risk management in a multi-dimensional insurance risk model.

In this paper we provide an alternative framework to tackle the first-best Principal-Agent problem under CARA utilities. This framework leads to both a proof of existence and uniqueness of the solution to the Risk-Sharing problem under very general assumptions on the underlying contract space. Our analysis relies on an optimal decomposition of the expected utility of the Principal in terms of the reservation utility of the Agent and works both in a discrete time and continuous time setting. As a by-product this approach provides a novel way of characterizing the optimal contract in the CARA setting, which is as an alternative to the widely used Lagrangian method, and some analysis of the optimum.

Within the context of traditional life insurance, a model-independent relationship about how the market value of assets is attributed to the best estimate, the value of in-force business and tax is established. This relationship holds true for any portfolio under run-off assumptions and can be used for the validation of models set up for Solvency~II best estimate calculation. Furthermore, we derive a lower bound for the value of future discretionary benefits. This lower bound formula is applied to publicly available insurance data to show how it can be used for practical validation purposes.

In this work, inspired by the Archer-Mouy-Selmi approach, we present two methodologies for scoring the stress test scenarios used by CCPs for sizing their Default Funds. These methodologies can be used by risk managers to compare different sets of scenarios and could be particularly useful when evaluating the relevance of adding new scenarios to a pre-existing set.

Recent literature implements machine learning techniques to assess corporate credit rating based on financial statement reports. In this work, we analyze the performance of four neural network architectures (MLP, CNN, CNN2D, LSTM) in predicting corporate credit rating as issued by Standard and Poor's. We analyze companies from the energy, financial and healthcare sectors in US. The goal of the analysis is to improve application of machine learning algorithms to credit assessment. To this end, we focus on three questions. First, we investigate if the algorithms perform better when using a selected subset of features, or if it is better to allow the algorithms to select features themselves. Second, is the temporal aspect inherent in financial data important for the results obtained by a machine learning algorithm? Third, is there a particular neural network architecture that consistently outperforms others with respect to input features, sectors and holdout set? We create several case studies to answer these questions and analyze the results using ANOVA and multiple comparison testing procedure.

Gaussian random vectors exhibit the loss of dimension phenomena, which relate to their joint survival tail behaviour. Besides, the fact that the components of such vectors are light-tailed complicates the approximations of various multivariate risk measures significantly. In this contribution we derive precise approximations of marginal mean excess, marginal expected shortfall and multivariate conditional tail expectation of Gaussian random vectors and highlight links with conditional limit theorems. Our study indicates that similar results hold for elliptical and Gaussian like multivariate risks.

We study arbitrage opportunities, market viability and utility maximization in market models with an insider. Assuming that an economic agent possesses from the beginning an additional information in the form of a random variable G, which only becomes known to the ordinary agents at date T, we give criteria for the No Unbounded Profits with Bounded Risk property to hold, characterize optimal arbitrage strategies, and prove duality results for the utility maximization problem faced by the insider. Examples of markets satisfying NUPBR yet admitting arbitrage opportunities are provided for both atomic and continuous random variables G.

We propose a mathematical model of momentum risk-taking, which is essentially real-time risk management focused on short-term volatility of stock markets. Its implementation, our fully automated momentum equity trading system presented systematically, proved to be successful in extensive historical and real-time experiments. Momentum risk-taking is one of the key components of general decision-making, a challenge for artificial intelligence and machine learning with deep roots in cognitive science; its variants beyond stock markets are discussed. We begin with a new algebraic-type theory of news impact on share-prices, which describes well their power growth, periodicity, and the market phenomena like price targets and profit-taking. This theory generally requires Bessel and hypergeometric functions. Its discretization results in some tables of bids, which are basically expected returns for main investment horizons, the key in our trading system. The ML procedures we use are similar to those in neural networking. A preimage of our approach is the new contract card game provided at the end, a combination of bridge and poker. Relations to random processes and the fractional Brownian motion are outlined.