PhD Defences

Daphne Lenders

• Tuesday 3 September 2023, 4.00 p.m.
• Location: University of Antwerp - Stadscampus Room S.B.004 (it will also be possible to attend online)
  
• Supervisors: Prof. Toon Calders and Prof. Sylvie De Raed
• Please contact Daphne Lenders (daphne.lenders@uantwerpen.be) to inform her whether you wish to attend the PhD defence (in person or online) before 20 August 2024
• Title: Targeted Approaches Against Discrimination - New Methods for Bias Detection and Mitigation in Automated Decision Making Systems​

Abstract:

Automated decision-making (ADM) systems used in high-stakes areas such as lending or hiring often perpetuate biases present in their underlying data. Consequently, these systems can adversely impact certain population groups, mirroring the sexist or racist practices of our society.

In this thesis, we inspect current approaches to auditing and mitigating such discriminatory biases in ADM systems. We highlight how these approaches typically centre around single definitions of fairness, that aim to express how (un)fair some system is through a single number and try to optimize for fairness accordingly. We explain how these approaches fall short in adequately understanding and resolving discrimination and argue how better approaches should be driven by more nuanced considerations: rather than having one single fairness measure, auditors should focus on which parts of the data a system behaves discriminatory, so that they then can then address this behaviour in a targeted manner.

To that end, our first two chapters focus on new tools and methods for bias detection in ADM systems. The first inspects the potential of interactive auditing toolkits, while the second improves an existing method for measuring individual fairness, allowing auditors to decide for one decision subject at a time whether they received just treatment.

Our third chapter introduces a human-in-the-loop approach to mitigate bias in ADM systems. We design a selective classifier that refrains from making predictions when they are deemed as discriminatory. These rejected instances, along with an explanation for their rejection, can be passed on to human experts who can make better-informed decisions for them.

The fourth chapter shifts focus from new bias mitigation techniques to evaluating their effectiveness. We emphasize how the traditional evaluation scheme, based on single fairness definitions, is not sufficient and instead introduce a benchmarking-dataset to facilitate the evaluation of bias mitigation strategies. This dataset includes a fair and biased version of its decision labels, allowing precise assessment of how well a model can predict the fair labels after being applied on the biased ones. Our fifth and final chapter zooms out from these specific considerations surrounding bias in ADM systems and provides an overview of the research field in general. By highlighting research gaps, we also conclude this thesis with a discussion and its implications for future work.

​

Michiel Van Roy

• Thursday 19 September 2024 - 5.00 pm
  
• Location: University of Antwerp - Stadscampus Room S.C.101
  
• Supervisors: Prof. Ann Jorissen, Prof. David Martens and Prof. Anne Van de Vijver
  
• Please contact Michiel Van Roy (michiel.vanroy@uantwerpen.be) to inform him whether you wish to attend the PhD defence before Friday 13 September 2024
• Title: Tax Disclosure: Firm’s Governance Characteristics and Utilization by Tax Authorities

Abstract:

Tax fairness of companies has increasingly becomea point of interest for society. According to the report on the “PublicConsultation on the Review of the Non-Financial Reporting Directive” (2020) ofthe EU, tax is the fourth most important non-financial topic on the agenda.There is an increasing amount of pressure on companies to be transparent ontheir tax payments because mandatory disclosure regimes often do not compelcompanies to disclose detailed information on their taxes. Therefore, societycalls companies to voluntarily disclose more information. This research willexamine whether there is an association between tax transparency and corporategovernance characteristics of companies.

Recently, corporate governance codes have beenadapted to a broader, stakeholder-based view. This is in line with theincreasing expectations of companies to serve a multitude of stakeholdersinstead of just shareholders, increasing the importance of Corporate SocialResponsibility (CSR). Therefore, the theoretical framework of this study willbe grounded in the established literature researching the impact of corporategovernance on CSR disclosure and Voluntary Disclosure. Previous research hasshown a positive relation between certain characteristics of corporategovernance and CSR-disclosures, as well as between corporate governancecharacteristics and voluntary disclosures.

Tax information can be argued to form a part ofCSR disclosures or can be disclosed voluntarily. Taxation forms an interestingtopic however, since many companies are often hesitant to disclose informationon their taxes paid due to concerns like loss of competitiveness. Therefore, itis interesting to examine whether the previously established relations in thedisclosure literature hold when the level of tax information disclosed is beingexamined.

In an additional analysis, the study willexamine whether tax transparency is related with tax avoidance/evasion. In thisregard, this study will contribute to the debate on whether public disclosureof tax information is associated with lower tax avoidance, on which previousresearch has found mixed evidence.

This study is unique in the regard that it is,to the best of our knowledge, the first study to examine whether there is arelation between distinct corporate governance characteristics and the level oftax disclosure. The recent publication of GRI 207-standard on taxes provides anopportunity to improve existing tax disclosure indices used in previousliterature. 

Wouter Dister

• Wednesday 15 November 2023 - 5.00 p.m.
• Supervisors: prof. dr. Anne Van de Vijver (University of Antwerp) & prof. dr. Miet Vanderhallen (University of Antwerp and Maastricht University)
  
• Title: The Belgian Co-operative Tax Compliance Pilot ProjectKey Succes Indicators
• Abstract

Dieter Brughmans

• Friday 8 September 2023 4.00 p.m.
• Supervisors: David Martens & Bruno Peeters
• Title: Counterfactual explanations for real-world applications​
  

Abstract:

We first introduce NICE, a novel algorithm for tabular data that specifically takes into account algorithmic requirements that are often overlooked. NICE exploits information from the training data to speed up the search process of finding an explanation. Our extensive benchmarking study indicates that NICE’s explanations have desirable properties compared to the current state-of-the-art. Next, we study the implementation of NICE within the risk management system at the Belgian Customs Agency. Our research identified four stakeholders: data scientists, targeting officers, domain experts, and decision subjects. For each of the stakeholders, we investigate how they can benefit from local explanationmethods and what their requirements are regarding the form of these explanations. Broadening the scope of our research beyond tabular data, we introduce SEDC and SEDC-T, two counterfactual algorithms for image classification. We experiment with various segmentation methods to find meaningful counterfactual explanations. Finally, this work tackles the ethical issue of disagreement among counterfactual explanations, revealing how their diversity can potentially be exploited to fairwash ML models by hiding sensitive features. Our study finds alarmingly high disagreement levels between the tested methods.

Pedro Moraya Barros

• Monday 11 September 2023
• Supervisors: Ann Jorissen & Anne Van de Vijver
• Title: International Personal Taxation and Country Characteristics​
  

Abstract:

The increase in capital mobility in the last couple of decades, along with the rise in popularity of offshore investments, tax havens, and foreign taxation strategies, gave prominence to the importance of international taxation in domestic politics. Although income taxation often represents more than half of states’ revenue and has been positively correlated with economic development, 51 fiscally sovereign jurisdictions (which represent about 25% of the countries, 15% of the FDI, and 5% of the GDP of the world) chose to not tax their residents in foreign and/or domestic source income. The large divergence between personal income tax systems that exist today leads significant amounts of capital to flow from high to low tax jurisdictions, while potentially increasing wealth inequality. Thus, what are the country's characteristics related to the choice of personal income tax and related policies? To contribute to the answer to this question, I develop a 3-article dissertation divided between: (Chapter 2) What are the characteristics related to a country (not) having personal income taxation? (Chapter 3) What are the characteristics related to a country taxing residents on domestic-source income and not on foreign-source (i.e. personal territorial taxation)? (Chapter 4) What are the characteristics related to a country having policies potentially used by individuals to avoid or evade tax duties, i.e. the official selling of passports and visas? To test the theoretical models underlining these relationships based on Institutional Theory, Median Voter Theorem, Modernization Theory, and Bellicist Theories, I created several novel datasets that cover all sovereign jurisdictions in the world often through a long time series. Using these datasets, I employ a series of quantitative methods (e.g. ordinary least squares, maximum likelihood estimations, event analysis, error correction models, and instrumental variable approaches) that find significant empirical results for the fields of personal taxation, institutions, and foreign investment. This dissertation also has implications for broader questions in the literature.