Trust is fundamental to managing the uncertainty that characterizes many social interactions, such as those within project teams and between firms and their customers. As these interactions become increasingly mediated by digital technologies, research is needed to understand how trust can develop when technology introduces new uncertainties. This thesis examines extended reality (XR), which includes virtual reality (VR) and augmented reality (AR), as an emerging technology that not only creates novel uncertainties for users but also amplifies existing ones, making it a particularly valuable context for research. First, XR makes avatars the primary mode of human representation across applications and digital platforms. Because facial features are central to social attribution, this shift raises the question of whether users can form trusting relationships when their interactions are mediated through these artificial and anonymous representations. Second, XR intensifies privacy risks because headsets collect highly sensitive behavioral and environmental data, increasing uncertainty about how such data are collected, used, and protected. This development highlights the need to understand how consumers can be supported in their adoption decisions when privacy risks are pronounced. Adopting an interdisciplinary perspective rooted in the Information Systems (IS) discipline, this thesis draws on theories and methods from behavioral economics, social psychology, and communication research to examine these issues through three essays. Essay 1 reports an online lab experiment that examines whether transparency about privacy risks helps consumers make more informed privacy decisions by analyzing their choices between different levels of transparency. Employing an experimental design adapted from a classic behavioral economics paradigm for decision making under uncertainty, the essay shows that individuals may avoid transparency about privacy risks precisely in high-likelihood loss situations, where transparent information would seem particularly valuable. Essay 2 explores social virtual reality as a new form of computer-mediated communication that can serve as the foundation for emerging types of digital platforms. Based on a laboratory experiment and an online experiment, the essay investigates how avatar representation and co-presence, as distinctive properties of social VR, shape the uncertainty inherent in first encounters and whether interpersonal trust can emerge in this context. The findings show that avatar representation in social VR increases perceptions of anonymity, but does not impede trust formation. Instead, information conveyed through the voice channel and individuals’ dispositional trust can offset the trust-reducing effects that avatars might otherwise exert. Essay 3 complements these findings by identifying a key boundary condition for when avatars can support trust at levels comparable to human representations. Across two online experiments, the essay shows that representational context matters: trust in avatar-represented individuals depends on whether avatars appear in a uniform representational environment or alongside human representations. These findings make three distinctive contributions to research. First, they advance privacy research by identifying ambiguity attitudes as an important behavioral driver of privacy decision making and by extending insights from behavioral economics on monetary outcomes to decisions involving personal data. Second, they contribute to research on XR, the metaverse, and avatars by showing that avatar-mediated interaction does not inherently undermine positive social evaluations. Rather, trust depends on both compensatory information channels and the broader representational context. Third, they extend theories of trust by illuminating the mechanisms through which trust forms in anonymous, avatar-mediated interactions. In summary, this thesis therefore advances IS research by contributing to a scientific understanding of XR as a technology that is still emerging but likely to shape organizations and society more profoundly in the near future.

Over the past two decades, mystery marketing, consisting of strategies that intentionally withhold information to stimulate consumer curiosity, has emerged as a powerful tool for capturing and sustaining consumer attention in an increasingly saturated and competitive market. By leveraging suspense and uncertainty through tactics such as mystery deals, surprise discounts, and concealed product details, firms aim to foster deeper consumer engagement, enhance brand loyalty, and drive positive word-of-mouth behavior. Despite the growing popularity and practical relevance of mystery marketing, academic research on how consumers respond to these strategies across the full trajectory of their engagement remains fragmented. In particular, limited insights exist on how consumers emotionally and behaviorally react before, during, and after the resolution of uncertainty and how specific design elements of mystery deals shape these responses. This dissertation addresses this gap by investigating how mystery marketing strategies influence consumer behavior across the entire uncertainty resolution journey. It explores consumer reactions not only in anticipation of a mystery offer but also during and following the disclosure of its outcome. Across three independent essays, this work seeks to answer the overarching research question: How do mystery marketing strategies impact consumer behavior throughout the uncertainty resolution journey? Essay 1 adopts a conceptual approach to synthesize and structure existing findings in mystery marketing. Drawing on literature from marketing, psychology, and behavioral decision-making, it introduces a framework that classifies consumer responses—emotional, cognitive, and behavioral—across the three key phases o the uncertainty resolution journey: pre-uncertainty resolution, uncertainty resolution, and post-uncertainty resolution. This essay highlights critical gaps in the current literature and outlines promising directions for future research that aim to build a more integrated understanding of the consumer experience in mystery marketing. Essays 2 and 3 delve deeper into mystery deals, a particularly prevalent form of mystery marketing in which firms conceal key product attributes until after purchase. Essay 2 investigates how the design elements of transparency and control interact with consumers’ prior product knowledge to influence interest and engagement in the pre-uncertainty-resolution phase. The findings indicate that the effectiveness of these design elements varies based on consumer expertise and psychological distance, offering strategic guidance for tailoring mystery deals to different consumer segments. Essay 3 focuses on the post-uncertainty-resolution phase and examines how consumers emotionally and behaviorally respond when the mystery is revealed. Through six empirical studies, this essay reveals that disconfirmation of expectations can negatively affect consumer loyalty. However, positive affective spillover from the pre-purchase phase can mitigate these effects. These findings contribute to understanding how emotional dynamics unfold after mystery resolution and how firms can manage consumer experiences when outcomes do not align with consumers’ preferences. Together, the three essays of this dissertation yield four key insights. First, mystery marketing influences consumer responses dynamically across multiple touchpoints, and these effects are highly context- and design-dependent. Second, consumer reactions differ substantially across the phases of the uncertainty resolution journey, underlining the need for a temporal and process-based perspective. Third, psychological mechanisms such as information gaps, control perceptions, construal levels, and affective spillover play central roles in shaping consumer experiences with mystery deals. Finally, this dissertation offers actionable implications for firms aiming to design mystery marketing strategies that maximize engagement while minimizing post-purchase dissatisfaction. In sum, this dissertation advances academic understanding of mystery marketing by systematically mapping consumers’ uncertainty resolution journey from curiosity to resolution, and by uncovering the drivers and consequences of consumer responses at each stage. It provides a solid theoretical and empirical foundation for future research and offers practical guidance for marketers seeking to leverage curiosity in a meaningful and consumer-centric way.

The first paper (“Natural Language Processing und Publikationsschwerpunkte - Eine (netzwerkgrafische) Darstellung der Steuerliteratur in der DStR”) provides an (automated) overview of the main focus of publications within German tax literature by employing NLP techniques. The paper employs network graphs to visualize the evolution of publication foci over time. While bibliometric analyses of authors and citations have already been conducted in various contexts and specifically in tax literature (e.g., Ya’u and Saad, 2021; Bashir, Ma, Bilal, Komal, and Bashir, 2021), comparable studies of German tax literature are still lacking. The paper closes the gap by tracing the thematic development of German-language tax scholarship. Its empirical strategy involves analyzing all articles published between 1991 and 2021 in the journal “Deutsches Steuerrecht”. To enhance the quality of the analysis, the authors implemented a preprocessing step wherein common stop words were systematically removed from the articles, along with generic terms explicitly defined by the authors. To identify key terms, the study calculates the Term Frequency-Inverse Document Frequency (TF-IDF) across the entire corpus of articles (Salton and Yu, 1973; Jones, 2021). The ten terms with the highest TF-IDF values per article are defined as key terms (Yao, Pengzhou, and Chi, 2019). A co-occurrence matrix is then constructed to identify relationships between key terms, where the terms serve as nodes and their co-occurrences form the edges in network graphs. Utilizing this methodology, the study conducts yearly moving network graphs, with each graph representing a five-year period, to illustrate the evolving relationships between thematic foci in the field. The approach successfully identifies key focal points in the academic discourse, such as the discussions surrounding the “Unternehmenssteuergesetz 2008” (Corporate Tax Reform Act 2008). The second paper (“(Chat)GPT und Steuern - Die Erweiterung der Wissensbasis gen erativer KI am Beispiel der steuerlichen Behandlung von Photovoltaikanlagen”) examines the integration of tax-related domain knowledge for Large Language Models (LLMs), such as GPT. The primary objective of the study is to develop a process that facilitates the conversion of tax knowledge into a machine-readable format, with a subsequent utilization of this knowledge by LLMs. The paper explores the potential of LLMs for application in tax advisory practices while also addressing potential challenges. One key limitation of LLMs is the constrained training period, which conflicts with frequent changes in tax regulations that are not reflected in the model. Another challenge lies in the inaccessibility of proprietary tax knowledge, which is unavailable to the LLM. To explore the potential of LLMs in the context of tax advisory, the capabilities of GPT-3.5 and GPT-4 were evaluated using a custom questionnaire addressing the income and value-added tax treatment of photovoltaic systems under the Annual Tax Act 2022. The questionnaire is divided into two sections: the first part focuses on theoretical questions, while the second part tests the models’ ability to independently resolve tax-related case studies (transfer tasks). Since the necessary knowledge is not inherently present in the models, they were supplemented with external sources, including literature and updated legal texts. These resources were embedded and stored in a vector database, enabling a semantic and syntactic comparison between the embedded questions and the database to identify relevant text passages for answering the questions. Subsequently, both models generated responses, which were manually evaluated by the authors. The findings demonstrate that enriching LLMs with additional, up-to-date (tax) information is feasible and yields promising results. However, errors in the responses persist, and their identification requires tax expertise. Consequently, LLMs should be regarded as a supportive tool for tax professionals rather than a replacement for their expertise. The third paper (“Adherence to the OECD Model Tax Convention: A Textual Analysis of Member and Non-Member Countries’ Double Taxation Agreements”) examines the textual design of DTAs and the factors influencing their alignment with the OECD Model Tax Convention, a widely recognized guideline for DTA negotiations among both OECD member and non-member countries (OECD, 2011; OECD, 2017). While the OECD Model Convention provides a standardized framework, it is subject to criticism in the literature due to its bias toward reallocating taxing rights from the source state to the resident state (Neumayer, 2007; Braun and Zagler, 2018). This redistribution often disadvantages developing countries by reducing their tax revenue, which is particularly problematic given that nearly half of all DTAs are negotiated between developing and developed nations (Braun and Zagler, 2014; Hearson, 2018). Despite the prevalence of this phenomenon, existing research on DTA texts is primarily focused on single countries (e.g., Smith and Sawyer, 2011; Braun and Fuentes, 2016; Steenkamp et al., 2014) or isolated aspects such as withholding tax rates and double taxation relief methods (e.g., Rixen and Schwarz, 2009; Marques and Pinho, 2014). Only the study by Ash et al. (2020) adopts a comprehensive analysis of DTA texts to investigate the convergence of these agreements with the Model Convention over time. However, deviations from the OECD consensus remain unexamined. The paper seeks to address the existing research gap by proposing a novel approach to measuring the heterogeneity of a country’s DTA set using NLP for six articles (“Dividends”, “Interest”, “Royalties”, “Business Profits”, “Permanent Establishment”, and “Associated Enterprises”) that significantly influence the extent of source taxation and are central to scholarly interest. Utilizing the TF-IDF vectorization method, the study transforms the text of all DTAs pertaining to a common article into a unified vector space. This enables the construction of a “country vector”, representing an average DTA vector for each country. Based on these vectors, the heterogeneity of a country’s DTAs for each of the six articles is quantified, providing the foundation for an exploratory analysis. Subsequently, the study calculates the cosine similarity between each article of a DTA and the timely matched article of the respective OECD Model Convention. These similarity scores serve as the dependent variables in regression analyses to identify the socioeconomic, institutional, and geopolitical factors influencing a country’s adherence to the OECD Model Convention across the six articles. The exploratory analysis utilizes a dataset of 2,922 DTAs concluded between 1963 and 2021, with each DTA representing the most recent amendment. To guarantee that the dataset exclusively comprises the outcomes of bilateral negotiations, 202 DTAs modified through the Multilateral Instrument were excluded. For the regression analyses, covariates from various sources were incorporated, with each variable temporally matched to the DTA’s last amendment date, resulting in a dataset of 2,211 observations. After subjecting both analyses to extensive robustness tests, the study reveals several key findings. Economically powerful countries maintain a more heterogeneous DTA set compared to poorer nations. Furthermore, the results indicate that asymmetries in country size and wealth drive greater adherence to the OECD Model Convention, with OECD and EU membership further contributing to this trend. Consistent with literature, this reflects a shift of taxing rights to residence countries in treaties between economically asymmetric countries. In contrast, DTAs between two wealthier countries exhibit greater deviation from the Model Convention, especially when economic ties are strong. Geopolitical factors, such as a shared colonial history, have also been identified as contributing elements to non-adherence. The study highlights how NLP techniques can enhance the analysis of DTA texts, emphasizing the critical role textual elements play in shaping taxing rights distribution. Policymakers must account for these complex dynamics, extending their focus beyond simple tax rate effects to consider the broader implications of DTA text alterations. The fourth paper (“Birds of a Feather Tax Together? Economic and Geographic Mimicry in German Tax Competition”) investigates the spatial-econometrics of trade and property tax multipliers among German municipalities, focusing on how urban and rural tax rates are influenced by neighboring municipalities — a phenomenon termed the “mimicking effect”. To attribute this effect specifically to tax competition, the paper develops a research design that distinguishes tax competition from expenditure spillovers and political yardstick competition, two alternative explanations for spatial tax rate correlation. Expenditure spillovers emerge when public goods and services funded by one jurisdiction also benefit neighboring areas. In response, nearby municipalities may adjust their tax and spending policies — either by reallocating resources, enhancing services, or changing tax rates — to remain attractive to residents and businesses (Ferraresi, Migali, and Rizzo, 2018; Costa, Veiga, and Portela, 2015; Kelejian and Robinson, 1993). Political yardstick competition (Salmon, 1987), on the other hand, assumes that voters, constrained by information asymmetry, compare local tax rates and public goods provision with those in neighboring municipalities to evaluate policymakers’ performance (Besley and Case, 1992). Policymakers, aware of this dynamic, may align their tax policies with those of neighboring jurisdictions to meet voter expectations (Bosch and Solé-Ollé, 2007). Finally, tax competition theories attribute the phenomenon of tax mimicking to the mobility of capital. In this theoretical framework, municipalities lower tax rates in order to attract businesses, thereby incentivizing neighboring jurisdictions to adopt similar tax policies (Tiebout, 1956; Oates, 1972; Wilson, 1999). Factors such as municipal size (Buettner, 2001) and the urban-rural divide (Janeba and Osterloh, 2013) further influence this dynamic. Given the empirical complexity of disentangling these mechanisms (Brueckner, 2003), the paper’s methodological approach seeks to provide a contribution to isolating tax competition effects within the broader framework of inter-municipal tax rate mimicking. The methodology employed in the study introduces the concept of “economic neighbors” to analyze tax mimicking, defining socioeconomically similar jurisdictions that may be geographically distant. Similarities are quantified using a set of socioeconomic variables and measured through cosine similarity. It is hypothesized that economic neighbors exert a greater influence on tax mimicking among cities than among rural municipalities, as prior research suggests that cities engage in competition with (more distant) cities (Janeba et al., 2013). To further trace the mimicking effect to tax competition over mobile capital, the study separately examines taxes on mobile (trade taxes) and immobile capital (real estate taxes). These tax variables are incorporated into two regression models: a spatial lag regression model based on Lyytikäinen (2012) and a spatial instrumental variables model estimated via two-stage least squares following Baskaran (2014). Both models are extended to include the influence of economic neighbors, providing a framework for disentangling the drivers of tax mimicking across municipalities and the effect of mobile and immobile tax bases. The empirical analysis is based on a dataset covering 47,251 observations from 2011 to 2021, revealing several key findings through robustness testing. For rural municipalities, the models consistently demonstrate a strong tendency to align their tax rates with those of geographically proximate neighbors. This pattern is evident in both taxes on mobile and immobile capital, suggesting a broad application of tax mimicking across different types of taxation. In contrast, cities exhibit more nuanced mimicking behaviors driven primarily by economic similarity rather than geographic proximity, underscoring the critical role of tax competition in their fiscal decision-making. The divergence between rural and urban mimicking behaviors can be attributed to the overlap between geographic and economic neighbors in rural areas, where socioeconomic similarities simplify tax policy alignment. Conversely, cities, characterized by diverse and complex economic structures, are less influenced by immediate geographic neighbors. Instead, their tax policies are shaped by competition with other cities possessing similar economic profiles. These findings emphasize the need for future research to differentiate between urban and rural contexts to better isolate the effects of tax competition within the broader mimicking framework. For policymakers, particularly those operating at supra municipal levels, the results highlight the importance of addressing competitive dynamics in urban areas. Implementing strategies such as raising minimum tax multipliers could help mitigate the risks of a tax competition-induced race to the bottom. The fifth paper (“Vorweggenommene Grundsteuerhebesatzerh¨ohungen der Kommunen im Zusammenhang mit der Grundsteuerreform – eine empirische Analyse”) examines the potential for municipalities in Germany to disproportionately adjust property tax multipliers in anticipation of the 2025 property tax reform, aiming to capitalize on potential windfall effects. Speculation in the literature (Hey, 2017) and among industry representatives (IHK Niedersachsen, 2024) suggests that municipalities may exploit the reform as an opportunity to raise property tax multipliers, regardless of actual revenue needs. As the 2025 property tax reform approaches, it can be observed that municipalities in North Rhine-Westphalia have raised their property tax multipliers in 2024. This suggests a response to the reform, as municipalities may seek to secure additional revenue. While existing research focuses on the reform’s impact on the tax burden through changes in valuation parameters and the suitability of different property tax models (e.g., Maiterth and Lutz, 2019), it overlooks the critical role of multiplier adjustments in determining taxpayers’ actual liabilities. Although the 2018 Federal Constitutional Court’s ruling clearly mandated the implementation of the reform by 2025, the assessment bases for future taxation remained uncertain until the legislation was finalized in November 2019. This provided municipalities with binding information about the redefined framework of the property tax system, suggesting that multiplier-setting policies have likely been influenced by the reform since 2020. By using neutral multipliers published by the fiscal administration — the multipliers required to maintain revenue neutrality at the municipal level in the reform year 2025 — it is possible to identify which municipalities are most affected by the reform. This enables an analysis of whether only municipalities that can justify increases based on their impact from the reform raise their multipliers. The empirical strategy relies on a panel of 5,940 observations drawn from municipal financial records covering the years 2010 to 2024. Employing a difference-in-differences framework, municipalities that are required to raise their property tax multipliers in 2025 to maintain revenue neutrality are classified as the treated group, while those that do not face such adjustment serve as the control group. Contrary to expectations, the primary estimates reveal no statistically significant divergence in the evolution of property tax rates between treated and untreated municipalities. To further probe underlying heterogeneity, an alternative identification strategy classifies municipalities according to their per capita debt levels. This stratification indicates that former increases in property tax multipliers can be partially attributed to municipalities’ financial constraints. However, the uniform elevation of multipliers observed in 2024 cannot be explained solely by indebtedness, suggesting that even jurisdictions with stable or expanding tax bases have elected to raise their rates. Overall, these findings cast doubt on the reform’s revenue-neutral mandate: rather than merely offsetting tax base changes, municipalities appear to have capitalized on the reform to secure additional — and potentially enduring — revenue increments.

The predominant approach of various data privacy regulations around the world is to stop or limit organizational arbitrariness in collecting personal data on the internet. The aim of these data privacy regulations is to address the privacy concerns of users and enable a safe online space where they can disclose their data. Some studies find the intended effect, i.e. more regulation leads to a reduction in privacy concerns of users. However, more regulation does not simply mean that people are less concerned. Research also shows the effect of regulation to be dependent on user characteristics or the strength of the regulation determines whether the effect is positive or negative. Further, there is not the one regulation but depending on what aspect of regulation we look at, for example degree of governmental involvement or restrictiveness of the laws, regulation can have a different form. Due to this variety of regulation and its effects it is necessary to take a closer look at the multiple forms of regulation and their specific interplay with individual behavior to gain a better understanding of the underlying mechanisms that provoke the differences in previous findings. This will then help to adjust regulations according to their intended effects on user behavior, i.e. disclosure of data. To achieve this, the thesis examines the different types of regulatory measures in data privacy regulations and their individual effect on user behavior. This thesis consists of five essays that make use of multiple research methods, i.e. a structured literature review, taxonomy development and quantitative studies. The findings of the thesis contribute to the understanding of the impact of data privacy regulations on individual disclosure behavior by 1) identifying the construct of regulation to be multifaceted with each type of regulation impacting individual behavior differently, 2) ascribing contextual as well as personality based preconditions to the different types of regulatory measures that determine their effectiveness to provide privacy to users, and 3) relating regulatory impact on users with cultural influences, i.e. variations in the impact of the regulatory types depending on the cultural believes of an individual.

Mobile technologies and digital platforms have expanded rapidly across sub-Saharan Africa, creating opportunities to strengthen state capacity, broaden financial inclusion, and build agricultural resilience. This dissertation examines their development impacts across three settings. Chapter 1 combines novel administrative tax records with high resolution data on mobile network rollout in Uganda to estimate the effects of mobile internet access on firm tax behavior and public revenue. Exploiting plausibly exogenous rollout timing, we find that improved access increases firm formalization and expands the tax base, strengthening revenue collection. Chapter 2 presents a randomized controlled trial in Niger—the world’s most financially excluded country—to identify barriers to adoption of a mobile money platform. Information provision raises awareness but not use, consistent with information being necessary but insufficient for diffusion. By contrast, a modest financial incentive significantly increases both adoption and usage. Chapter 3 uses a household panel collected before and after a severe drought in northern Ghana and exploits variation in rainfall in a differences-in-differences framework to estimate effects on production, income sources and adaptation plans. We find that increasing drought severity lowers soybean yields and revenues and decreases reliance on own-business income and remittances, reflecting broader livelihood impacts. Farmers who use mobile phones to access agricultural information make different input choices and adaptation plans. Taken together, these findings highlight the promise of mobile technologies and digital platforms for development. Their effectiveness, however, depends on complementary infrastructure and local capacity, underscoring the need for scalable, context-specific strategies to harness these tools for inclusive growth and resilience.