Contributions published at Finance and Insurance (Pablo Koch Medina)

Predicting equity returns is a complex task in finance. This paper examines the volume profile as a predictive tool using machine learning techniques. We process and summarize the volume profile into features, using an XGBoost classifier to forecast stock return direction. Our approach is validated across two equity sets, demonstrating its capability to identify high-return periods. Based on the probability estimates, trading strategies are created and shown to be able to outperform the benchmark on a risk-adjusted and total return basis. Overall, the results indicate the predictive potential of the volume profile leveraged by the model.

We use the theory of cooperative games for the design of fair insurance contracts. An insurance contract needs to specify the premium to be paid and a possible participation in the benefit (or surplus) of the company. We suppose a convex commonotonic premium functional is used to value the aggregated liability of the insurance company. It results from the analysis that when a contract is exposed to the default risk of the insurance company, ex-ante equilibrium considerations require a certain participation in the benefit of the company to be specified in the contracts. The fair benefit participation of agents appears as an outcome of a game involving the residual risks induced by the default possibility and using fuzzy coalitions.

Given a stock price process, we analyze the potential of arbitrage in a context of short-selling prohibitions. We introduce the notion of minimal supermartingale measure, and we analyze its properties in connection with the minimal martingale measure. This question is more specifically analyzed in the case of an investor having additional, inside information. In particular, we establish conditions when minimal martingale and supermartingale measures both fail to exist. These correspond to the case when the insider information includes some nonnull events that are perceived as having null probabilities by the uninformed market investors, even as they cannot observe them. The results may have different applications, such as in problems related to the local risk minimization for insiders whenever strategies are implemented without short selling.

Reinsurance has traditionally been one of the major risk management tools primary insurers use to manage their exposure to underwriting and solvency risks (Shiu (2020)), effectively acting as contingent capital. As such, reinsurance purchase is a capital structure decision and helps insurers to keep an optimal level of underwriting risk relative to their capitalisation level (Cummins, Dionne, and Gagné (2021)), especially in the presence of regulatory risk-based capital requirements. At the same time, the optimal level of underwriting risk reflects the cost of reinsurance, which contributes to determining the insurer’s profitability. This thesis aims to understand better and empirically analyse the interdependence between reinsurance, capital and performance measures for U.S. property and casualty insurers. The empirical results of a simultaneous equation model confirm the mutual interactions among capital, reinsurance, and performance. Reinsurance purchase is positively related to capitalisation and vice versa, which is inconsistent with the capital buffer hypothesis. Confirming the pecking order theory, Return on Assets has a statistically significant and positive effect on capital, and a higher Return on Assets reduces the reinsurance usage. The analyses shed some light on the relationship between performance, reinsurance usage and capital.

Within the context of expected utility and in a discrete loss setting, we provide a complete account of the demand for insurance by strictly-risk averse agents and risk-neutral firms when they enjoy limited liability. When exposed to a bankrupting, binary loss and under actuarially fair prices, individuals and firms will either fully insure or not insure at all. The decision to insure will depend on whether the benefits the insuree derives from insurance after having compensated the damaged party are sufficiently attractive to justify the premium paid. When the loss is nonbinary, even when prices are actuarially fair, any amount of coinsurance can be optimal depending on the nature of the loss.

We study capital management and investment decisions of a value-maximizing insurance firm with a broad ownership base in a discrete-time setting. We highlight that the valuation measure used to determine the value of the cash flows to shareholders should reflect two economically sound requirements: market-consistency and indifference to idiosyncratic risk. We provide a rigorous construction of this economic valuation measure and use it to derive the optimal capital-management and investment strategies that realize the economic value of the firm. Our objective is to shed light on the controversial question of whether insurers should invest in liquidly-traded risky assets. Decomposing firm value into net tangible value, default option value, and franchise value, we find that whether to take investment risk is optimal or not essentially depends on the tradeoff between the impact of investment risk on the owner’s option to default and on the firm’s franchise value. A variety of numerical examples illustrate how changes in the regulatory and financial environment can result in materially different optimal investment strategies.

We discuss when law-invariant convex functionals "collapse to the mean". More precisely, we show that, in a large class of spaces of random variables and under mild semicontinuity assumptions, the expectation functional is, up to an affine transformation, the only law-invariant convex functional that is linear along the direction of a nonconstant random variable with nonzero expectation. This extends results obtained in the literature in a bounded setting and under additional assumptions on the functionals. We illustrate the implications of our general results for pricing rules and risk measures.

We establish dual representations for systemic risk measures based on acceptance sets in a general setting. We deal with systemic risk measures of both "first allocate, then aggregate" and "first aggregate, then allocate" type. In both cases, we provide a detailed analysis of the corresponding systemic acceptance sets and their support functions. The same approach delivers a simple and self-contained proof of the dual representation of utility-based risk measures for univariate positions.

We establish general versions of a variety of results for quasiconvex, lower-semicontinuous, and law-invariant functionals. Our results extend well-known results from the literature to a large class of spaces of random variables. We sometimes obtain sharper versions, even for the well-studied case of bounded random variables. Our approach builds on two fundamental structural results for law-invariant functionals: the equivalence of law invariance and Schur convexity, i.e., monotonicity with respect to the convex stochastic order, and the fact that a law-invariant functional is fully determined by its behavior on bounded random variables. We show how to apply these results to provide a unifying perspective on the literature on law-invariant functionals, with special emphasis on quantile-based representations, including Kusuoka representations, dilatation monotonicity, and infimal convolutions.

Pay-How-You-Drive insurance products are a recent motor insurance innovation, based on telematics technology which enables to monitor and measure individual driver behavior. This innovative approach bears the potential to disrupt the insurance industry by improving individual risk assessment and enhancing driver behavior. While experimental research suggests improved driver behavior under the monitoring of a telematics device there is scarce evidence from real-world settings and almost none of the impact of telematics on motor insurance claims data. Therefore, this thesis aims to examine if and how, monitoring of driver behavior does impact motor insurance claims data. This is approached by analyzing claims data from a Swiss full-service leasing provider who offers company car insurance, including a telematics option, and related claims handling to its customers. In this specific circumstance the decision for the telematics-based insurance product does not lie with the drivers but with their employers, thus, excluding potential biases due to self-selection. The effect of being monitored by a telematics device on number of claims and cost of claims is analyzed by applying a panel regression to the dataset. The dataset in panel format contains 3'695 claims filed in the years 2014 to 2020 coming from 1'410 unique drivers resulting in a total of 2'482 observations across the years. The analysis found significant evidence for the impact of telematics on the cost of claims. Depending on the different fixed effects which were included, the results suggest that drivers who are being monitored have approximately 40% (p<0.01) lower cost of claims than drivers without a telematics device, on average. No significance could be found for the impact of telematics on the number of claims. Assuming that lower cost of claims could be a result of less severe accidents this might be indicative of better driving behavior. These findings are a first attempt to provide evidence of Pay-How-You-Drive on claims data which is of relevance to insurers as well as broader considerations, for example in regard to road safety.

Catastrophe Bonds are securities allowing re/insurers to transfer risk to nancial markets. Due to their connection to natural hazards and their relative independence from other asset classes, CAT Bonds prices obey to dynamics that markedly dier from those of other nancial instruments. Firstly, these securities are investigated from a theoretical perspective by means of the approaches proposed by Froot (2001) and Nell and Richter (2000). Based on such theoretical premises, a comprehensive set of hypothesis is formulated with regards to the main determinants of issuance prices. The hypothesis are then tested on a comprehensive dataset of public Property & Casualty bonds issued between 2008 and 2019. The OLS regressions performed conrm many results reported in previous literature, restating the relevance of transaction-specic features such as the Expected Loss, the territory and peril covered, and the trigger applied. With respect to the latter, the results also indicate that substantial changes recently took place with regards to how the dierent types of triggers are perceived on the market. Moreover, data also suggest that macroeconomic elements like the stage of the reinsurance cycle and that of yields on comparable corporate bonds in uence CAT Bonds prices. Finally, time xed eects are included in the analysis to take into consideration the high volatility that characterized this market over the period considered.

The credit rating of corporations aims to provide insights about their creditworthiness and model the bankruptcy probability. These ratings often consult financial key figures and neglect the macroeconomic perspective. This descriptive thesis investigates the contemporaneous effects of macroeconomic drivers on bankruptcies in Switzerland. Applying a first-difference lagged model I found that the effects of the variables included vary across industries and that these effects tend to rise when considering the seasonally not adjusted GDP.

This thesis presents the different kinds of market equilibria in a theoretical framework, based on the most used models in the context of quantitative insurance modelling. Start-ing from a basic model with perfect information, we extend it by allowing for asymmetric information between the insurer and the insurance seekers with respect to an individual’s risk type (high risk or low risk). This makes insurance contracts more realistic and thus easier to study empirically. With the introduction of screening contracts, we further differ-entiate between low risks with high disutility from revealing private information and those with corresponding low disutility. The welfare in terms of an individual’s utility buying this contract can be increased, while those not willing to share private information by choosing the screening contract are faced with a potential loss in utility, whose effect is based on the market situation. Since insurers have been starting to introduce and extend such screening contracts, they have also become relevant for real insurance markets in the last years. Thus, they are now of a likewise importance in theoretical insurance modelling.

We describe a framework for the valuation of insurance liabilities that relies on first principles in finance theory. Key features of the economic value of liabilities are its market-consistency and the inclusion of the costs of financial frictions. We compare this framework to the Solvency II approach and highlight the differences.

Given a stock price process, we analyse the potential of arbitrage by insiders in a context of short-selling prohibitions. We introduce the notion of minimal supermartingale measure, and we analyse its properties in connection to the minimal martingale measure. In particular, we establish conditions when both fail to exist. These correspond to the case when the insider's information set includes some non null events that are perceived as having null probabilities by the uninformed market investors. These results may have different applications, such as in problems related to the local risk-minimisation for insiders whenever strategies are implemented without short selling.

In a capital adequacy framework, risk measures are used to determine the minimal amount of capital that a financial institution has to raise and invest in a portfolio of prespecified eligible assets in order to pass a given capital adequacy test. From a capital efficiency perspective, it is important to be able to do so at the lowest possible cost and to identify the corresponding portfolios, or, equivalently, their payoffs. We study the existence and uniqueness of such optimal payoffs as well as their behavior under a perturbation or an approximation of the underlying capital position. This behavior is naturally linked to the continuity properties of the set‐valued map that associates to each capital position the corresponding set of optimal eligible payoffs. Upper continuity can be ensured under fairly natural assumptions. Lower continuity is typically less easy to establish. While it is always satisfied in a polyhedral setting, it generally fails otherwise, even when the reference risk measure is convex. However, lower continuity can often be established for eligible payoffs that are close to being optimal. Besides capital adequacy, our results have a variety of natural applications to pricing, hedging, and capital allocation problems.

This article summarizes the main topics and findings from the Swiss Risk and Insurance Forum 2018. That event gathered experts from academia, insurance industry, regulatory bodies, and consulting companies to discuss the challenges arising from the impact of data science and, more generally, of digitalization to the insurance sector.

Switzerland is amongst the most developed countries in the world. This is re- ected in its high quality of healthcare. However, if healthcare costs continue to increase the costs supported by the inhabitant may become unsustainable. Consequently, the system must be reformed but how? Some countries ap- pear to be more ecient than Switzerland, one of them is Sweden. Therefore, mimicking some ideas from Sweden could oer good ideas on how best to re- form the Swiss health system. First, we had identied whether the countries were comparable. A PESTEL analysis was used to highlight common points and the dierences. It helped to make a thorough description of the two countries and their health systems. Furthermore, it provided some dier- ences between the two health systems that were used to assess the potential improvements that could possibly be implemented in Switzerland. The health system has been the topic of many studies. Therefore, the anal- ysis and the adaptation of their results to the Swiss system was the second step of this thesis. This identied some key areas in which the Swiss health system could be improved. Switzerland is still based on a stationary care concept because of the low incentives to turn to ambulatory care. Introducing more ambulatory care could save 1 billion CHF per year according to a PWC study. However, those savings should be used for the change of cost gravity to primary care. Indeed, the development of primary care is almost an obligation nowadays with the ageing population and the growth of chronic diseases. Primary care is more ecient in dealing with chronic diseases and multi-comorbidity fac- tors. By identifying patients at an early stage, it could prevent unnecessary hospitalisation and therefore decrease the numbers of beds by inhabitants. Reaching the same eciency as Sweden in terms of cost per bed per inhab- itant could save about 2.7 billion CHF per year. Primary care also focuses on better coordination between the dierent stakeholders in the health sys- tem. The development of new technologies allowing the digitalisations of the health system would empower patients and highlight any ineciencies in the system. It would also enable the implementation of a pay-for-performance pricing system that could persuade stakeholders to choose the most ecient solutions in term of cost and quality of healthcare. Furthermore, a more ecient generic penetration with compulsory usage would save about 1.7 billion CHF if Switzerland reaches the same market penetration as Sweden. One billion more could be saved if the price of generic medicines was the same as bordering countries. Finally, amalgamating all of these savings could allow Switzerland to save more than 5 billion CHF per year. In conclusion, optimistic projections made 1 by the Federal Oce of Statistic in 2007 predicted that the health expendi- ture would be between 111 billion and 128 billion CHF in 2030, in the best scenarios. Unfortunately, this best scenario now seems dicult to attain if the health expenditures keeps increasing at the same rate. It would probably be higher than 130 billion CHF without considering the ageing population and their associated costs. However, several measures can be implemented to slow the pace of the increasing costs. We have seen that by mimicking Swedish ideas, Switzerland could potentially save more than 5 billion per year and that the development of primary care and the use of digitalisation could substantially increase those savings in the future. Politicians are aware of the future challenges to the health system. Therefore, they need to re- form the health system in order to make it sustainable and ecient in the long-term.

Articial intelligence is a tool that is becoming so useful and ubiquitous that it has been described as a kind of sixth sense. It is present in our everyday lives in ways that we do not always realize, making us not perceive the speed with which it aects us. Technological instruments, able to handle huge amounts of data and to simulate human behaviours, are expected to be diused in the near future aecting all industries' business models signicantly; this is what gives companies a clear advantage in amplifying its products and services. For many years we have been hearing about technologies that have entered the banking sector by making us familiar with the term ntech; a similar pattern is also being seen in the insurance industry. As new technologies are introduced into the insurance industry, people will have to reimagine it and the way in which the insurance system will work in the future, but the speed with which the market is developing and the complexity of changes that articial intelligence will bring is not completely understood. This Thesis provides a better comprehension of what currently exists and is thought to get an insight into how the insurance market is being aected by AI giving a deeper insight on real-time analysis and how the system we are used to will be transformed. The key ndings presented in this Thesis are that the insurance sector is being aected by technology since it is a data-based sector and vulnerable to changes. Insurance companies have to digitizing existing business models through real-time processes to maintain a competitive edge. The revolution in business activities could be benecial to the market development; new innovations allow for an increased individualized risk pricing that will make fairer premiums and more re ective of that risk, leading to an improvement of customer experience. Although customers will be charged with their actual risk level, avoiding assessments based on group characteristics, risk pooling will still be relevant in the insurance sector: if the risk is too high, an insured would not be able to aord the insurance if it were not for risk pooling, he would be fully responsible for any loss, so the principle of paying claims with the premiums of the entire pool will not change. Insurance technologies has not to be seen as a problem but as an opportunity as it allows to keep up with constant changes, reduces information asymmetries and enable insurers to take a proactive role in reducing risks rather than compensating costs after the occurrence of a loss event. This proactive approach increases the social capital of the insurance providers, mitigates preventable claims and customers' frauds and allows for cross-selling of multiple services building an insurance ecosystem of dierent companies. However, it should be noted that AI not only brings benets but also raises some concerns. The impact of articial intelligence is twofold: on one hand, it impacts the employment levels and introduces new risks into the market, on the other hand, it helps the insurance processes and oers the possibility to develop potential new insurance products. This study is presented in six chapters. The rst chapter is an introductory part that contains an overview of the disruptions the insurance industry is facing. The second chapter provides a background on the insurance system and the principles on which it has been based so far. The third chapter explains how articial intelligence will in uence the insurance industry and how it should respond to changes. The fourth chapter is devoted to dynamic insurance based on real-time analysis, with a focus on risk assessment and underwriting and claim management processes. The fth chapter deals with two important limitations that AI brings and the related opportunities. The last and sixth chapter includes the conclusion of the study. This Thesis is supported by gures, tables and examples whenever necessary to assist the reader in developing a clear understanding of the topic.

We use a model with agency frictions to analyze the structure of a dealer market that faces competition from a crossing network. Traders are privately informed about their types (e.g. their portfolios), which is something the dealer must take into account when engaging his counterparties. Instead of participating in the dealer market, the traders may take their business to a crossing network. The dealer must take into consideration that traders have this alternative when choosing a pricing schedule. We show that the presence of a crossing network may benefit traders even if they do not trade in it. Furthermore, it results in more traders being serviced by the dealer and the book’s spread shrinking (under certain conditions). We allow for the pricing on the dealer market to determine the structure of the crossing network, which itself influences the structure of the dealer market. This results in a feedback look that, under the same conditions that lead to a reduction of the spread, yields an equilibrium book/crossing network pair.