Contributions published at Financial Engineering (Markus Leippold)

This thesis investigates the effect of Inflation on corporate cash holdings and the effect of Cash Ratio volatility on stock performance. The dataset consists of the SMI Expanded companies and covers two timeframes from 2020 to 2022 and 2021 to 2022. There is no detectable relationship between Inflation levels and corporate cash holdings on either timeframe. By introducing Delta values, a statistical relationship was detected, but with a low overall significance. Cash Ratio volatility and stock performance have a negative relationship over the short timeframe from 2021 to 2022. Over the long timeframe with two significant market disruptions no relationship could be detected in the sample.

In order to evaluate the social dimension of the ESG domain in Sustainable Finance, this thesis focuses on using deep learning models, specifically the use of a new BERT adaptation model called SocialBERT. By creating a model that is responsive to socially relevant texts, it addresses a significant research gap and makes it more effective at comprehending and assessing these elements. Social aspects are now playing a major part in determining the sustainability of financial investments, revolutionizing the financial industry with the rise of ESG investing. However, because of the richness and diversity of texts that are socially relevant, it is often difficult to quantify these social variables. Therefore, it is noticeable that we require sophisticated technologies to interpret and comprehend these texts. The SocialBERT model has been created to better understand the nuances and context of social aspects than conventional models because it has been pre-trained on a huge corpus of texts with a social focus. The model is assessed for its capabilities and performance, and the results show that it is more effective than conventional models at understanding social texts. Additionally, the thesis emphasizes the shortage of study in this field and the necessity of larger-scale investigations to promote a better understanding and integration of social factors into sustainable finance. This research builds on the development of deep learning techniques, the success of big language models like BERT and GPT, and growing trends in the application of Natural Language Processing (NLP) in finance. In conclusion, the SocialBERT model has the potential to improve sustainable finance decisionmaking by facilitating a more sophisticated understanding of social aspects. The results of this thesis not only to expand the pool of knowledge in the area but also open up fresh possibilities for investigation and advancement in the use of cutting-edge NLP technologies for ESG analysis. II

Discovering the Social in ESG through analyzing linguistic patterns in annual reports reveals the ignored social component within the environmental, social, and governance (ESG) aspects. This thesis uses natural language processing (NLP), a subset of machine learning (ML), to extract social indicators from the ESG framework, thereby addressing the current underemphasis on social variables. The study helps to solve the issue of ESG data quality as a result of non-standardized and selective reporting. This is accomplished by employing ML algorithms that are unconcerned about reporting quality, resulting in uniform ESG data interpretation. Furthermore, the study addresses the underutilization of ML and NLP in the social ESG context by utilizing several sophisticated models such as SocialRoBERTa and SocialDistilRoBERTa. These models, which have been trained to understand the social context, outperform standard models like SVM and RF. Significantly, the study finds an inverse association between a company’s ESG risk score and the frequency of social discourse in its annual reports, with social discourse accounting for only 2.4% of the ESG score variation. This study emphasizes the need for additional research and a holistic approach. The ML models’ performance plateauing after training on 50-60% of the dataset presents an opportunity for optimum resource utilization during training for improved efficiency. In essence, this thesis gives insights into the social elements of ESG while also increasing ESG data recognition and broadening the effective application of ML and NLP in this domain. It emphasizes the great prospect for a more comprehensive approach and additional investigation into the thriving subject of sustainable finance.

This study aims to determine whether it is possible for corporate insiders to generate abnormal returns and to test if the Efficient Market Hypothesis is valid for the Swiss Stock market and to which degree it can be applied. For this purpose, an event study for insider trades between January 2020 and January 2023 was conducted. The results suggest that corporate insiders can generate statistically significant abnormal returns and therefore reject the strong form of the Efficient Market Hypothesis in the Swiss stock market.

This study examines the relationship between a company’s Market Capitalization-to-equity value differential (MED) and its Environmental, Social, and Governance (ESG) scores. Three regression models are employed to analyze the relationship, including linear, multiple linear and quantile regression. Industry and size factors are introduced as control variables. This inclusion enhances the explanatory power of the model and reveals industry specific dynamics and the impact of the company’s size. The findings reveal a significant correlation between the two variables, but also draw attention to the impact of other factors on market valuations. These findings contribute to the understanding of how ESG factors into financial markets and provide information to investors and companies aiming to make sustainable investment strategies. Further investigation is warranted to explore the additional factors influencing this relationship, such as the methodology of ESG scoring, regulatory frameworks or the potential interaction between different ESG criteria.

Diese Arbeit untersucht den Einfluss des Faktors Volatilität auf die Aktienpreisbildung. In einem ersten Teil wird auf etablierte Finanzmarkttheorien und ihre Modelle eingegangen. Der zweite Teil behandelt die dazugehörige Empirie und die aktuellen Untersuchungsergebnisse zum Volatilitätsfaktor. Die Analyse des Volatilitätsfaktors stellt sechs Regressionsmodelle mit unterschiedlicher Parameterzusammenstellung im Kontext des FamaFrench-Dreifaktormodells (Fama und French (1993)) einander gegenüber, wobei die durchschnittlichen adjustierten R2 als Vergleichsmass genutzt werden. Die Ergebnisse zeigen, dass der Einfluss des Volatilitätsfaktors signifikant ist, da das Bestimmtheitsmass des FamaFrench-Dreifaktormodells kleiner ausfällt als dasjenige des Modells mit dem zusätzlichen Volatilitäts-Faktor. Die Untersuchung zeigte zudem eine grössere Relevanz des Faktors Volatilität gegenüber dem Value-Faktor auf. Nichtsdestotrotz ist der Faktor Volatilität kein geeigneter Ersatz weder für das Marktrisikopremium noch den Size-Faktor.

Diese Arbeit untersucht den Effekt von «flight-to-safety» in Gold und Staatsanleihen auf die Portfolioperformance. Dafür wurden Benchmark-Portfolios für unterschiedliche Investortypen und Märkte während verschiedener Krisen gebildet. Diese wurden Vergleichsportfolios mit veränderten Anteilen der Anlagen im Portfolio gegenübergestellt. Dabei wurde der Effekt einer Umschichtung des Portfolios auf die Rendite und die Volatilität untersucht. Die Resultate zeigen, dass die Flucht in Gold bzw. Staatsanleihen nicht immer positiv ausfällt, sondern markt-, krisen-, und portfoliotypspezifisch ist. Allgemein lassen die Ergebnisse darauf schliessen, dass sich Gold besser zur Werterhaltung eignet und Staatsanleihen vorteilhafter zur Reduktion der Volatilität eines Portfolios während Krisen sind.

Diese Arbeit untersucht zwei der meistverbreiteten Aktienmarkt-Anomalien, Value und Momen-tum, im Schweizer Aktienmarkt über den Zeitraum von 1990 bis 2022. Weiter wird untersucht, wie die beiden Strategien miteinander korrelieren. Sowohl Value- wie auch Momentum-Effekt sind in der Schweiz vorhanden. Der Momentum-Effekt ist dabei mit Renditen von jährlich 13.76% deutlich stärker ausgeprägt als der Value-Effekt mit jährlich 2.06%. Value-Strategien scheinen ausserdem von 1990 bis 2010 hervorragend funktioniert zu haben, danach aber nicht mehr. Werden jedoch Faktorrenditen berechnet, die anderen Variablen gegenüber neutral sind, ist nur noch Momentum signifikant. Die Korrelation der beiden Effekte ist mit -0.1 zwar negativ, aber nicht stark.

Um der hohen weltweiten Inflation im Jahr 2022 entgegenzuwirken, waren Zentralbanken dazu gezwun-gen, die Leitzinsen anzuheben. Auf die Renditen und Deckungsgrade der Schweizer Pensionskassen hatte dies einen negativen Einfluss, sodass 2022 alle untersuchten Anleger eine negative Rendite erziel-ten und der technische Deckungsgrad im Median beinahe um 15% abnahm. Positiv wirkten sich unter anderem hohe Strategieanteile in illiquiden Anlagen aus, welche durch verzögerte Bewertungen weniger stark betroffen waren. Langfristig zeigt sich dieser positive Einfluss deutlich schwächer. Es besteht die Gefahr, dass Pensionskassen ihre finanzielle Lage durch die Glättung der illiquiden Anlagen besser dar-stellen, als dies tatsächlich der Fall ist.

This thesis examines the potential impact of social media on the financial markets and associated dangers. The focus lays on the share prices and volume charts of GameStop, AMC Entertainment, Blackberry and Bed Bath & Beyond within the time frame from July 2020 to December 2021. Twitter serves as the social media platform for this analysis. The objective of this analysis is to determine whether the volume and share price movements occur after a corresponding rise or fall in number of tweets. Notably, this investigation reveals a noteworthy correlation and highlights the existence of substantial dangers.

Diese Arbeit untersucht die Auswirkungen von ESG-Exclusions auf die risikoadjustierte Performance eines Portfolios. Dazu werden 22 Exclusions sowohl einzeln als auch in thematischen Gruppen auf das Portfolio der Schweizer Krankenversicherung CSS angewendet. Die daraus entstehenden Exclusion-Portfolios werden anhand der Rendite, Sharpe-Ratio, Tracking-Error, Information-Ratio und in verschiedenen Faktor-Modellen ¨uber den Zeitraum von 2010 bis 2022 verglichen. Die Resultate zeigen, dass die verschiedenen Exclusions unterschiedliche Auswirkungen auf die Portfolios haben. Die Mehrzahl der Exclusions f¨uhrt zu einer besseren absoluten und risikoadjustierten Rendite bzw. zu besseren Performancemassen. Jedoch sind diese Resultate abh¨angig vom gew¨ahlten Zeitraum, wie rollierende Berechnungen zeigen.

We combine Tyler's robust estimator of the dispersion matrix with nonlinear shrinkage. This approach delivers a simple and fast estimator of the dispersion matrix in elliptical models that is robust against both heavy tails and high dimensions. We prove convergence of the iterative part of our algorithm and demonstrate the favorable performance of the estimator in a wide range of simulation scenarios. Finally, an empirical application demonstrates its state-of-the-art performance on real data.

Inspired by the celebrated Joint SPX & VIX Calibration problem, this study delves into the technical derivation of two prominent rough stochastic volatility models from the ground up, with the aim of consolidating and extending previous empirical findings as well as shining a new light on critical details that might have gone unnoticed. A systematic analysis of the latest Deep Learning interpretations of the rough Bergomi and rough Heston models reveals a series of eloquent properties with regards to stylised facts observed in SPX and VIX volatility time series.

In 2021 more than half of U.S. IPOs were SPACs, and more than a third of the companies that went public had chosen a de-SPAC transaction over a traditional IPO. However, the hot Wall Street trend ended soon after its emergence. Investors burnt their fingers, and sponsors walked away with huge profits. The SEC intervened with restrictive regulatory scrutiny to protect the investors. However, increasing the liability, improving disclosure, and ultimately enhancing the SPAC’s structure will not eliminate the risk of SPACs as an investment vehicle.

If behavioral biases explain asset pricing anomalies, they should also materialize in cryptocurrency markets. I test more than 20 stock return anomalies based on daily cryptocurrency data, and document strong evidence of price momentum. Controlling for market and size, price momentum remains statistically significant, whereas price reversal and risk-based anomalies are weak. Cryptocurrency anomalies can be explained by behavioral theories that emphasize noise trader risks than fundamental risks.

Financial research recently started using the latest breakthroughs in natural language processing. This thesis develops an innovative framework for systematic exposure assessment: it measures the notion of exposure by leveraging language models and gathering a priori expert expressions to quantify the use of corpus- and domain-specific vocabulary. Focusing on risks and emerging technologies in earnings calls, we apply our method to reveal insightful market behaviors. A case study then shows that exposures from earnings conferences reflect the content from corresponding 10-Ks. Since our approach solves major issues of other NLP techniques, we strongly encourage other researchers to explore further applications.

In dieser Arbeit wird die Markteffizienz mithilfe einer Momentum-Strategie anhand verschiedener Marktkapitalisierungs-Indizes getestet. Dabei basiert die Strategie auf dem Aspekt, dass Gewinneraktien Gewinner bleiben und Verliereraktien Verlierer bleiben. Durch die Anwendung dieser Strategie anhand eines Small-, Mid- und Large-Cap-Index soll zudem ein möglicher Unternehmensgrösseneffekt erkannt werden. Die empirische Studie der Arbeit konnte anhand des amerikanischen Aktienmarktes für den Zeitraum zwischen 2002 bis 2022 zeigen, dass es durch die Implementierung von auf technischer Analyse basierenden Strategien möglich ist, Überrenditen zu erzielen. Darüber hinaus konnte mit der Differenzierung von verschiedenen Marktkapitalisierungs-Indizes keine Evidenz für den Unternehmensgrösseneffekt bei Small-Cap-Unternehmen aufgebracht werden.

This thesis analyzes the various hedging strategies against currency and inflation risk. The research on the theoretical concepts show that hedging can reduce the volatility of the cash flows and the earnings if applied correctly. Based on the theory, two flowcharts were developed to help decide which project level hedging strategies are most appropriate. To find suitable hedging strategies in practice, a long-term service contract at Stadler Rail AG was analyzed. The finding is that in practice as many risks as possible should already be covered in the contract, while in theory more internal and external strategies are analyzed at the corporate level.

Existing factor models struggle to model the covariance matrix for a large number of stocks and factors. Therefore, we introduce a new covariance matrix estimator that first shrinks the factor model coefficients and then applies nonlinear shrinkage to the residuals and factors. The estimator blends a regularized factor structure with conditional heteroskedasticity of residuals and factors and displays superior all-around performance against various competitors. We show that for the proposed double- shrinkage estimator, it is enough to use only the market factor or the most important latent factor(s). Thus there is no need for laboriously taking into account the factor zoo. Supplementary material for this article is available online.

The aim of this thesis is to develop an automated correction tool using the Python programming language to efficiently correct the Involving Activity 3 in the course Asset Management: Investments. The exercise requires students to create two momentum strategies based on historical stock prices of 18 stocks using varying look-back and holding periods. The tool is designed to be highly flexible in terms of input data, loock-back, and holding periods, enabling the momentum strategies to be effectively tested and compared to a buy-and-hold strategy. The tool offers a powerful approach for correcting the Involving Activity 3 leading to faster processing times and minimized errors compared to manual correction methods.