| SEMINAR ŞTIINŢIFIC MONEDA, FINANŢE, BĂNCI – ARHIVĂ |
Quantum Computing, Deep Learning and Financial Stability in Good Times
Marți, 4 noiembrie 2025 – ora 16:30 – sala 3M4 (etaj 1, clădirea Moxa)
Alexie Alupoaiei
National Bank of Romania; alexie.alupoaiei@bnro.ro
Abstract:
In this paper, we investigate how the intensive use of modern technologies for credit risk and loan pricing will affect financial stability. Here we focus exclusively on the pricing channel during favorable macroeconomic periods. Our investigation is a normative one. Secondary we are interested: i) to understand how this emergent tool, quantum computing, can be used in practical credit risk assessment, ii) to explore how the quantum computing can be integrated efficiently within deep learning neural networks models and iii) to navigate new opportunities of using a micro-macro approach for financial stability purposes. Our preliminary results indicate that the widespread use of modern technologies for credit classification will potentially lower the loan interest rates during favorable economic periods. This fact could generate some important spillovers for monetary and macroprudential policy.
Zoom:
https://ase.zoom.us/j/88975840175?pwd=Q0ZnbHl0TjhqcEtzVnhMUWRmdWdPQT09
Activitate de socializare: J’ai Bistrot
The real effects of insurers’ financial constraints: Evidence from marine insurance during the first globalization (1874-1913)
Marți, 28 octombrie 2025 – ora 16:30 – sala 3M4 (etaj 1, clădirea Moxa)
Guillaume Vuillemey
HEC Paris; vuillemey@hec.fr
Abstract:
I show that insurers’ financial constraints have real effects on trade flows and technology adoption. During the first globalization (1874-1913), booming international trade and the shift from sailing to steam ships created large demand for marine insurance. Using archive data on French insurers, I show that unexpectedly large insurance payouts (due to wrecks or accidents) lead insurers to subsequently reduce the supply of insurance. Exploiting the presence of the same insurer in multiple harbors, and data on wrecks and accidents, I construct harbor-level shocks to insurance supply. These shocks reduce trade flows and delay the adoption of the steam ship.
Zoom:
https://ase.zoom.us/j/88975840175?pwd=Q0ZnbHl0TjhqcEtzVnhMUWRmdWdPQT09
Activitate de socializare: TBD
Kernel Conditional Factor Models
Marți, 21 octombrie 2025 – ora 16:30 – sala 3M4 (etaj 1, clădirea Moxa)
Urban Ulrich
ETH Zurich; urban.ulrych@math.ethz.ch
Abstract:
Factor models are widely employed in finance to capture the relationship between asset returns and their underlying factors. Traditionally, these models assume a linear relationship in learning factor loadings. This paper enhances factor models by introducing non-linearity through low-rank kernel functions, offering a flexible, non-parametric representation of complex, non-linear relationships between factors, returns, and asset characteristics. We utilize a reproducing kernel Hilbert space (RKHS) with the associated reproducing kernel as a hypothesis space for modeling the factor loadings, while cross-sectional ridge regression is used to directly learn the factor portfolios. This approach extends existing methods by incorporating non-linear dependence on characteristics, regularization for more factors, and additional characteristics such as industries. Empirical analysis shows that the proposed non-linear learning framework significantly outperforms traditional linear models in terms of out-of-sample performance, as measured by explained variation and optimal factor portfolio performance.
Zoom:
https://ase.zoom.us/j/88975840175?pwd=Q0ZnbHl0TjhqcEtzVnhMUWRmdWdPQT09
Activitate de socializare: Green Hours Jazz Cafe
Nowcasting Models for the Early Monitoring of Risks to Financial Stability
Marți, 14 octombrie 2025 – ora 16:30 – sala 3M4 (etaj 1, clădirea Moxa)
Grigore Ivan and Ștefania Stancu
National Bank of Romania and Bucharest University of Economic Studies; grigore.ivan@bnro.ro
Abstract:
This paper proposes a framework for the rapid assessment of the business cycle in Romania, based on the integration of economic indicators with different frequencies. Beyond using traditional data, the framework thus exploits a diverse set of high-frequency indicators, available with reduced lag, which provide a more detailed and up-to-date picture of economic dynamics. By combining these information sources, the model enables earlier identification of changes in economic activity, compared to official monthly or quarterly statistics. In turbulent periods, marked by many sources of uncertainty, such a tool offers a prompt perspective on trends, contributing to real-time conjunctural analysis.