18 March Research Day Agenda |
| |
|
| 08.50 - 09.00 |
Welcome and opening remarks |
| Session 1 |
Mechanisms of New Finance Market (session chair: Yi Ding) |
| 09.00 - 09.40 |
Jungpil Hahn - Professor in the Department of Information Systems and Analytics, National University of SingaporeLink opens in a new window Understanding Decentralization in Proof-of-Stake Blockchains: An Agent-Based Simulation Approach - PresentationLink opens in a new windowLink opens in a new window Blockchain systems allow for securely keeping shared records of transactions in a decentralised way. This is enabled by algorithms called consensus mechanisms. Proof-of-work is the most prominent consensus mechanism, but environmentally unsustainable. Here, we focus on proof-of-stake, its best-known alternative. Importantly, decentralised decision-making power is not an inherent feature of blockchain systems but a technological possibility. Numerous security incidents illustrate that decentralised control cannot be taken for granted. We, therefore, study how key parameters affect the degree of decentralisation in proof-of-stake blockchain systems. Based on real-world implementation of a proof-of-stake blockchain system, we conduct agent-based simulations to study how a range of parameters impact decentralisation. The results suggest that high numbers of initial potential validator nodes, large transactions, a high number of transactions, and a very high or very low positive validator network growth rate increase decentralisation. We find weak support for an impact of changes in transaction fees and initial stake distributions. Our study highlights how blockchain challenges our understanding of decentralisation in information systems research and contributes to understanding the governance mechanisms that lead to decentralisation in proof-of-stake blockchain systems as well as to designing proof-of-stake blockchain systems that are prone to decentralisation and therefore, more secure. |
| 09.40 - 10.20 |
Xiao Qiao - Assistant Professor in the School of Data Science, City University of HKLink opens in a new window A Novel Portfolio Optimization Framework for Online Loan Investments and Interest Rate Determination PresentationLink opens in a new window We investigate the suitability of online loans as an investment through the lens of a portfolio optimization framework. We introduce general characteristic-based portfolio policies (GCPP), a framework which overcomes unique challenges associated with building a portfolio of online loans. Under this framework, we propose a nonlinear portfolio policy based on a shallow neural network. Whereas an equal-weight portfolio achieves an average annual internal rate of return of 6.55%, a nonlinear portfolio leads to an improved annual IRR of 13.08%. The nonlinear portfolio also enables more access to credit by investing more in loans with lower credit grades. To assess the attractiveness of online loans, we compare the performance of the nonlinear portfolio to other benchmark assets, including stocks, bonds, and real estate. We find that online loans earn competitive rates of return to the other assets while showing limited comovement. Our results indicate that online loans are an attractive novel asset class for investors, and investors can diversify their holdings by investing in online loans with increased expected returns. The GCPP framework can also help set the interest rate associated with loans, leading to more efficient pricing that benefits both borrowers and lenders. |
| 10.20 - 10.40 |
Coffee Break |
Session 2
|
AI and Finance (session chair Lei Wang) |
| 10.40 - 11.20 |
Xitong Li - HEC ParisLink opens in a new window My Advisor, Her AI and Me: Evidence from a Field Experiment on Human-AI Collaboration and Investment Decisions PresentationLink opens in a new window Contributing to current policy and academic debates about bringing humans in the loop of Artificial intelligence (AI), we explore whether allowing humans to collaborate with AI in the AI-based service production, compared to a pure AI solution, benefits the service production and consumption side. We conduct a field experiment with a large savings bank and produce pure AI-based and human-AI collaborative investment advice to the bank's customers. On the production side, we find that implementing a human-AI collaboration by allowing bankers to have the final say with AI output does not compromise advice quality. More importantly, on the consumption side, we find that the customers are more likely to align their final investment decisions with advice from this human-AI collaboration, compared to pure AI, especially when making more risky investments. The higher reliance on human-AI collaborative advice also translates to higher monetary payoffs. Overall, the results from the field experiment suggest that bringing humans into the AI-based advisory service production is pivotal to allowing AI-enabled efficiency gains to transmit to downstream customers. In a complementary online experiment, we further uncover the mechanism underlying customers' higher reliance on bankers' participation in generating investment advice. We find that the persuasive efficacy of human-AI collaborative advice stems from social influence on the customers. Our findings not only offer new insights for companies contemplating the provision of pure AI-based services, but also enrich policy and regulatory discussions by demonstrating the value of humans in AI-based service production. |
| 11.20 - 12.00 |
Inbal Yahav Shenberger- Senior lecturer, Tel Aviv UniversityLink opens in a new window AWER - A Framework for Automated Worker Evaluation Based on Free-Text Responses with No Ground Truth PresentationLink opens in a new window Evaluating workers based on their work quality is an important managerial task. However, this task becomes highly challenging when no ground-truth information is available to compare workers' output. Previous work has addressed this problem in settings where workers produce binary, numerical, or multi-categorical labels. Here, we consider the problem of automatically evaluating workers based on their free-text responses to open-ended questions without ground truth. To address this problem, we propose a new, unsupervised framework for automated worker evaluation. The framework is based on two main ideas: (a) framing the problem of textual response-based worker evaluation as a multidimensional voting problem; and (b) using an iterative reweighting algorithm that benefits from a holistic assessment of workers' inherent capabilities. To evaluate the framework, we empirically test its performance in two separate studies: using a semi-synthetic data-based evaluation and using two datasets of real workers' textual responses. In a third study, we use a pure numerical simulation to explore the method's operating conditions. Overall, we find that, across multiple settings, the framework consistently obtains superior results compared to a baseline approach and that its performance is robust in various challenging conditions. Thus, our framework can serve as a useful benchmark bere research on this problem. Additional benefits of our framework include scalability, modularity, and compatibility with existing, advanced textual representations. |
| 12.00 - 13.00 |
Lunch (Lab demonstrating VR & other features) |
| Session 3 |
Impacts on Digital Finance (session chair: Kalina Staykova) |
| 13.00 - 13.40 |
Amit Mehra - Professor, Information Systems UT DallasLink opens in a new window No Pain, No Gain: Examining the Impact of Lazy Minting on NFT Market Matching PresentationLink opens in a new window A common challenge faced by two-sided marketplaces is to expand the market while ensuring efficient matching of the two sides. We study a new platform growth strategy that increases market thickness with free entry while allowing suppliers to discretionarily signal their quality to buyers, thus forming a tiered market structure that maintains high matching efficiency even as the market grows. We use a real-world example of this growth strategy in the context of non-fungible token (NFT) markets, in which suppliers could choose to delay the payment of NFT creation fees (i.e., market entry costs) until the moment when an NFT is sold (lazy minting), or make this payment up front at the time of NFT creations (gas minting). We use the difference-in-differences strategy to identify the impact of this lazy-minting policy on matching efficiency. We find that, although the introduction of lazy minting reduces the average matching likelihood of an NFT by 45.5%, it increases the gas-minting segment’s matching likelihood by 116.9% and first-sale price by 126.5%. We establish that suppliers opt for gas minting, based on their high-quality creations, and that choosing gas minting enables them to signal high quality to buyers, resulting in improved matching likelihood and higher first-sale prices. Overall, this new growth strategy benefits the platform due to increased sales and higher commission revenues. Our study establishes the value of a new platform growth strategy that allows entrants to self-select their entry costs, thereby shaping a tiered market segmentation with an effective quality-signaling mechanism. |
| 13.40 - 14.20 |
Raj Sharman – Professor in Management Science and Systems, University of BuffaloLink opens in a new window Do natural disasters impact lending behavior in Minority Regions: Evidence from P2P lending
Borrowers seeking credit from banks may encounter higher interest rates, in the months following natural disasters, than in pre-disaster periods. In this paper, we examine the lending behavior of peer-to-peer (P2P) platforms in the wake of disasters. Since P2P platforms involve individual investors in lending decisions, they accommodate both profit motives and empathetic responses. A strong sense of empathy is a natural human response leading to prosocial behavior, especially when sensitivity to others’ distress is paired with a drive toward their welfare. Once a disaster strikes a region, it may invoke empathy for victims living in those areas and encourage individuals to involve in prosocial behavior. Empirical evidence from psychology suggests that empathy can lead to prosocial behavior. Consistent with this assertion, we find that loans originating in minority regions affected by natural disasters have lower interest rates than comparable loans not issued in the wake of disasters. Loan amounts increase after disasters, while loan defaults risk does not, suggesting that lenders’ profits do not decline. These results indicate that banks’ post-disaster credit tightening is overly conservative, while FinTech platforms, by including empathetic lenders, may facilitate the channeling of relief to those in need. To establish causality to our results rigorously, we use a matched Differences-in-Differences approach to compare interest rates of loans issued six months after a disaster to similar loans from borrowers not exposed to such an event. We control for a variety of possible factors and also address other identification issues that lead to biased estimators.
|
| 14.20 - 14.30 |
Coffee Break |
| Session 4 |
FinTech Governance (session chair: Kalina Staykova) |
| 14.30 - 15.10 |
Luigi Cantisani - Lawyer at Futura Law Firm, PGR Law School of the University of Warwick The Legal Challenges of the Decentralised Online Platform Application Presentation This presentation aims at exploring how notion such as Dapps, disintermediation, and decentralization are understood from a legal standpoint, what are the legal challenges to be considered when launching a Dapp, and finally, offer an overview of how the current EU “Act-ification season” might impact this sector. Why does what is happening in the EU matter all over the world? Well, it is widely accepted that the GDPR has considerably impacted legal relationships and negotiations of players from all over the world, not only from the EU because of the cross-border nature of contractual and legal relationships arising from the digital economy. Many scholars and practitioners estimate that the European Union's most recent and stringent regulations, such as Platform-to-Business Regulation, Data Act, Markets in Crypto-Assets Regulation, and Digital Services Act could have a similar impact on the global Dapps market, thus establishing elements of legal compliance that are of interest to all. |
| 15.10 - 15.20 |
Coffee Break |
| 15.20 - 16.30 |
Panel Discussion and Closing Remarks |
