Trevor Fitzpatrick | Searcharxiv

arXiv2025

Assessing Generative AI value in a public sector context: evidence from a field experiment

The emergence of Generative AI (Gen AI) has motivated an interest in understanding how it could be used to enhance productivity across various tasks. We add to research results for the performance impact of Gen AI on complex knowledge-based tasks in a public sector setting. In a pre-registered experiment, after establishing a baseline level of performance, we find mixed evidence for two types of composite tasks related to document understanding and data analysis. For the Documents task, the treatment group using Gen AI had a 17% improvement in answer quality scores (as judged by human evaluators) and a 34% improvement in task completion time compared to a control group. For the Data task, we find the Gen AI treatment group experienced a 12% reduction in quality scores and no significant difference in mean completion time compared to the control group. These results suggest that the benefits of Gen AI may be task and potentially respondent dependent. We also discuss field notes and lessons learned, as well as supplementary insights from a post-trial survey and feedback workshop with participants.

q-fin.GN↗

arXiv2019

The Network Effect in Credit Concentration Risk

Measurement and management of credit concentration risk is critical for banks and relevant for micro-prudential requirements. While several methods exist for measuring credit concentration risk within institutions, the systemic effect of different institutions' exposures to the same counterparties has been less explored so far. In this paper, we propose a measure of the systemic credit concentration risk that arises because of common exposures between different institutions within a financial system. This approach is based on a network model that describes the effect of overlapping portfolios. This network metric is applied to synthetic and real world data to illustrate that the effect of common exposures is not fully reflected in single portfolio concentration measures. It also allows to quantify several aspects of the interplay between interconnectedness and credit risk. Using this network measure, we formulate an analytical approximation for the additional capital requirement corresponding to the systemic risk arising from credit concentration interconnectedness. Our methodology also avoids double counting between the granularity adjustment and the common exposure adjustment. Although approximated, this common exposure adjustment is able to capture, with only two parameters, an aspect of systemic risk that can extend single portfolios view to a system-wide one.

q-fin.GN↗