SAFE working paper
https://safe-frankfurt.de/de/publikationen/working-papers.html
Refine
Year of publication
Document Type
- Working Paper (429) (remove)
Language
- English (429)
Has Fulltext
- yes (429)
Is part of the Bibliography
- no (429)
Keywords
- Liquidity (9)
- COVID-19 (8)
- General Equilibrium (7)
- welfare (7)
- Asset Pricing (6)
- ambiguity (6)
- climate change (6)
- competition (6)
- liquidity (6)
- Contagion (5)
Institute
- Wirtschaftswissenschaften (428)
- Center for Financial Studies (CFS) (409)
- House of Finance (HoF) (408)
- Sustainable Architecture for Finance in Europe (SAFE) (397)
- Rechtswissenschaft (8)
- Institute for Law and Finance (ILF) (7)
- Foundation of Law and Finance (6)
- Frankfurt MathFinance Institute (FMFI) (3)
- E-Finance Lab e.V. (1)
- Exzellenzcluster Die Herausbildung normativer Ordnungen (1)
270
We show that High Frequency Traders (HFTs) are not beneficial to the stock market during flash crashes. They actually consume liquidity when it is most needed, even when they are rewarded by the exchange to provide immediacy. The behavior of HFTs exacerbate the transient price impact, unrelated to fundamentals, typically observed during a flash crash. Slow traders provide liquidity instead of HFTs, taking advantage of the discounted price. We thus uncover a trade-o↵ between the greater liquidity and efficiency provided by HFTs in normal times, and the disruptive consequences of their trading activity during distressed times.
62
In this paper we investigate the implications of providing loan officers with a compensation structure that rewards loan volume and penalizes poor performance versus a fixed wage unrelated to performance. We study detailed transaction information for more than 45,000 loans issued by 240 loan officers of a large commercial bank in Europe. We examine the three main activities that loan officers perform: monitoring, originating, and screening. We find that when the performance of their portfolio deteriorates, loan officers increase their effort to monitor existing borrowers, reduce loan origination, and approve a higher fraction of loan applications. These loans, however, are of above-average quality. Consistent with the theoretical literature on multitasking in incomplete contracts, we show that loan officers neglect activities that are not directly rewarded under the contract, but are in the interest of the bank. In addition, while the response by loan officers constitutes a rational response to a time allocation problem, their reaction to incentives appears myopic in other dimensions.
75
In this paper, we investigate how the introduction of complex, model-based capital regulation affected credit risk of financial institutions. Model-based regulation was meant to enhance the stability of the financial sector by making capital charges more sensitive to risk. Exploiting the staggered introduction of the model-based approach in Germany and the richness of our loan-level data set, we show that (1) internal risk estimates employed for regulatory purposes systematically underpredict actual default rates by 0.5 to 1 percentage points; (2) both default rates and loss rates are higher for loans that were originated under the model-based approach, while corresponding risk-weights are significantly lower; and (3) interest rates are higher for loans originated under the model-based approach, suggesting that banks were aware of the higher risk associated with these loans and priced them accordingly. Further, we document that large banks benefited from the reform as they experienced a reduction in capital charges and consequently expanded their lending at the expense of smaller banks that did not introduce the model-based approach. Counter to the stated objectives, the introduction of complex regulation adversely affected the credit risk of financial institutions. Overall, our results highlight the pitfalls of complex regulation and suggest that simpler rules may increase the efficacy of financial regulation.
133
In this paper, we examine how the institutional design affects the outcome of bank bailout decisions. In the German savings bank sector, distress events can be resolved by local politicians or a state-level association. We show that decisions by local politicians with close links to the bank are distorted by personal considerations: While distress events per se are not related to the electoral cycle, the probability of local politicians injecting taxpayers’ money into a bank in distress is 30 percent lower in the year directly preceding an election. Using the electoral cycle as an instrument, we show that banks that are bailed out by local politicians experience less restructuring and perform considerably worse than banks that are supported by the savings bank association. Our findings illustrate that larger distance between banks and decision makers reduces distortions in the decision making process, which has implications for the design of bank regulation and supervision.
262
We investigate the default probability, recovery rates and loss distribution of a portfolio of securitised loans granted to Italian small and medium enterprises (SMEs). To this end, we use loan level data information provided by the European DataWarehouse platform and employ a logistic regression to estimate the company default probability. We include loan-level default probabilities and recovery rates to estimate the loss distribution of the underlying assets. We find that bank securitised loans are less risky, compared to the average bank lending to small and medium enterprises.
394
Recent regulatory measures such as the European Union’s AI Act re-quire artificial intelligence (AI) systems to be explainable. As such, under-standing how explainability impacts human-AI interaction and pinpoint-ing the specific circumstances and groups affected, is imperative. In this study, we devise a formal framework and conduct an empirical investiga-tion involving real estate agents to explore the complex interplay between explainability of and delegation to AI systems. On an aggregate level, our findings indicate that real estate agents display a higher propensity to delegate apartment evaluations to an AI system when its workings are explainable, thereby surrendering control to the machine. However, at an individual level, we detect considerable heterogeneity. Agents possess-ing extensive domain knowledge are generally more inclined to delegate decisions to AI and minimize their effort when provided with explana-tions. Conversely, agents with limited domain knowledge only exhibit this behavior when explanations correspond with their preconceived no-tions regarding the relationship between apartment features and listing prices. Our results illustrate that the introduction of explainability in AI systems may transfer the decision-making control from humans to AI under the veil of transparency, which has notable implications for policy makers and practitioners that we discuss.
315
This paper explores the interplay of feature-based explainable AI (XAI) tech- niques, information processing, and human beliefs. Using a novel experimental protocol, we study the impact of providing users with explanations about how an AI system weighs inputted information to produce individual predictions (LIME) on users’ weighting of information and beliefs about the task-relevance of information. On the one hand, we find that feature-based explanations cause users to alter their mental weighting of available information according to observed explanations. On the other hand, explanations lead to asymmetric belief adjustments that we inter- pret as a manifestation of the confirmation bias. Trust in the prediction accuracy plays an important moderating role for XAI-enabled belief adjustments. Our results show that feature-based XAI does not only superficially influence decisions but re- ally change internal cognitive processes, bearing the potential to manipulate human beliefs and reinforce stereotypes. Hence, the current regulatory efforts that aim at enhancing algorithmic transparency may benefit from going hand in hand with measures ensuring the exclusion of sensitive personal information in XAI systems. Overall, our findings put assertions that XAI is the silver bullet solving all of AI systems’ (black box) problems into perspective.
287
Using experimental data from a comprehensive field study, we explore the causal effects of algorithmic discrimination on economic efficiency and social welfare. We harness economic, game-theoretic, and state-of-the-art machine learning concepts allowing us to overcome the central challenge of missing counterfactuals, which generally impedes assessing economic downstream consequences of algorithmic discrimination. This way, we are able to precisely quantify downstream efficiency and welfare ramifications, which provides us a unique opportunity to assess whether the introduction of an AI system is actually desirable. Our results highlight that AI systems’ capabilities in enhancing welfare critically depends on the degree of inherent algorithmic biases. While an unbiased system in our setting outperforms humans and creates substantial welfare gains, the positive impact steadily decreases and ultimately reverses the more biased an AI system becomes. We show that this relation is particularly concerning in selective-labels environments, i.e., settings where outcomes are only observed if decision-makers take a particular action so that the data is selectively labeled, because commonly used technical performance metrics like the precision measure are prone to be deceptive. Finally, our results depict that continued learning, by creating feedback loops, can remedy algorithmic discrimination and associated negative effects over time.
370
Advances in Machine Learning (ML) led organizations to increasingly implement predictive decision aids intended to improve employees’ decision-making performance. While such systems improve organizational efficiency in many contexts, they might be a double-edged sword when there is the danger of a system discontinuance. Following cognitive theories, the provision of ML-based predictions can adversely affect the development of decision-making skills that come to light when people lose access to the system. The purpose of this study is to put this assertion to the test. Using a novel experiment specifically tailored to deal with organizational obstacles and endogeneity concerns, we show that the initial provision of ML decision aids can latently prevent the development of decision-making skills which later becomes apparent when the system gets discontinued. We also find that the degree to which individuals 'blindly' trust observed predictions determines the ultimate performance drop in the post-discontinuance phase. Our results suggest that making it clear to people that ML decision aids are imperfect can have its benefits especially if there is a reasonable danger of (temporary) system discontinuances.
401
In current discussions on large language models (LLMs) such as GPT, understanding their ability to emulate facets of human intelligence stands central. Using behavioral economic paradigms and structural models, we investigate GPT’s cooperativeness in human interactions and assess its rational goal-oriented behavior. We discover that GPT cooperates more than humans and has overly optimistic expectations about human cooperation. Intriguingly, additional analyses reveal that GPT’s behavior isn’t random; it displays a level of goal-oriented rationality surpassing human counterparts. Our findings suggest that GPT hyper-rationally aims to maximize social welfare, coupled with a strive of self-preservation. Methodologically, our esearch highlights how structural models, typically employed to decipher human behavior, can illuminate the rationality and goal-orientation of LLMs. This opens a compelling path for future research into the intricate rationality of sophisticated, yet enigmatic artificial agents.