Pandemic bonds, as a subtype of catastrophe (CAT) bonds, are designed to transfer public health risks to capital markets, yet they face substantial liquidity risk due to their nascent market structure and limited risk design. Using a relatively small sample from 2020 to 2024, we apply Bayesian model averaging (BMA) and interpretable machine learning (IML) to examine the associations of macroeconomic and bond-specific characteristics with liquidity premia across market phases. We find that the average liquidity premium of natural catastrophe bonds (1364 bps) is substantially higher than that of pandemic bonds (452 bps), potentially reflecting differences in bond characteristics and contractual structures. The estimates suggest that higher ESG scores and higher disclosed health scores are associated with lower estimated pandemic-bond premia, with health-score categories receiving slightly greater posterior inclusion support than ESG in the joint specification. Pandemic bond premia are higher during hard market phases, while several predictors receive greater posterior inclusion support during soft markets. Overall, the results suggest that incorporating a health-specific index, alongside insights from traditional CAT and green bond design, could enhance the resilience of pandemic bonds and strengthen their role in sustainable development.
This study focuses on the spatiotemporal prediction of supply chain finance risk contagion and constructs a model based on the dynamic graph attention diffusion mechanism for in-depth exploration. The risk-aware dynamic graph attention network (DyGAT) is used to model the time-varying network structure of supply chain finance, where the spatial attention weight α_ij(t) is updated adaptively at each time step by integrating real-time risk intensity signals to realize real spatiotemporal coupling modeling. A bounded diffusion mechanism with natural risk attenuation is introduced to characterize the propagation process of risks in spatial and temporal dimensions, avoiding unbounded cumulative defects in traditional formulas. Methodologically, a supply chain finance dynamic graph structure is constructed by integrating multisource data such as historical transaction data and corporate financial information, with clear data cleaning, fusion and feature engineering processes, and the model is trained, validated and ablated strictly. The results show that the model can effectively predict the spatiotemporal paths and topological characteristics of supply chain finance risk contagion and identify key nodes and critical periods of risk propagation, and the ablation experiment verifies that the dynamic attention module and bounded diffusion module bring 18.2
Cryptocurrency adoption for everyday transactions remains limited, particularly among nonusers, due to concerns about transparency and trust in decentralized systems. Drawing on trust theory, this study examines how initial trust shapes willingness to use cryptocurrencies for transactions in the UK. Survey data were collected from 402 nonusers between 14 December 2023 and 7 February 2024 and analyzed using partial least squares structural equation modeling (PLS-SEM). The proposed model demonstrates strong explanatory power (R2 = 0.80 for initial trust, 0.45 for perceived risk, and 0.52 for willingness to use). The results revealed that transparency, anonymity, ease of use, security, privacy, and knowledge significantly influence initial trust, which in turn strongly predicts willingness to use cryptocurrencies. Contrary to traditional financial adoption models, perceived risk does not significantly affect willingness to use, suggesting that initial trust can override risk concerns in decentralized finance. Theoretically, this study extends trust theory by conceptualizing initial trust in contexts with minimal institutional assurances. Practically, the findings highlight the need for initial trust-building strategies, such as transparent governance, clear communication, and user education, to foster broader cryptocurrency adoption and digital financial inclusion.
This study aims to examine the role of sustainable finance vs. nonsustainable finance in environmental sustainability using OECD nations from 2012 to 2023. Surprisingly, the data characteristics provide a clear guide to choose an appropriate set of estimators. For instance, all variables are integrated at first difference, which compels this study to adopt the Johnson cointegration approach. In addition, to address cross-sectional dependence and serial correlation problems, this study employs the Driscoll-Kraay Standard Error estimator. To break the suspense, this study considers total factor productivity, urbanization, technology (including information communication technology and artificial intelligence), green innovation, public‒private partnerships (PPPs), sustainability, and nonsustainable finance as environmental determinants. The obtained results remain fascinating for the specified nations. For instance, outcomes reveal an inverse relationship between total factor productivity, urbanization, and green innovations with carbon emissions. In addition, information and communication technology, artificial intelligence, and public‒private partnerships contribute to emissions. Similarly, sustainable finance (SF), including green bonds and green foreign direct investment, plays a significant role in addressing environmental issues. In contrast, nonsustainable finance increases emissions, except for financial stability and financial efficiency. Additionally, this study offers a moderate effect of SF on PPPs and green innovations, which does not support the sustainability theme. A heterogeneity view yields awe-inspiring results concerning environmental proxies, including CO2 emissions, ecological footprint, and load capacity factor. Finally, cross-country analysis describes fascinating outcomes under the proposed theme. Regarding outcomes, the present study offers imperative implications for becoming clean and green shortly.
Abstract Artificial intelligence (AI) claims in public markets increasingly raise disclosure, comparability, and investor-protection questions, yet evidence on how listed financial firms publicly describe implementation remains fragmented and difficult to compare. This study examines public technical disclosure rather than internal adoption by constructing a screened, exchange-bounded corpus for 253 NYSE-listed financial companies. The source intake contained 554 rows; six exact repeated rows were removed before screening, leaving 548 screened rows. The 505-sentence extended analysis corpus combines a 253-sentence firm-anchor layer with a 252-sentence direct-verification layer; the remaining issuer is retained in the direct-verification audit as a closed unit with no strict direct match. Firms are coded into three disclosure states: explicit Python, indirect AI, and none, and the corpus is used to map subsector profiles, channel location, and the rarity of quantified public outcome claims through an outcome claims index (OCI). The results show that 34 firms (13.4%) are classified as Explicit Python, 216 (85.4%) as Indirect AI, and 3 (1.2%) as None. Explicit Python naming is uneven across subsectors and is concentrated mainly in employment materials and technical documentation rather than in filings or investor-facing channels. The public technical surface is therefore real but thin: firms more often signal implementation orientation than disclose a detailed tool stack. Quantified public outcome claims are rarer still, with five OCI-positive firms and five OCI-positive sentences (1.98% of firms). These findings indicate that public financial-sector communication is much stronger on AI-related implementation signals than on measurable outcomes. The paper offers a reproducible disclosure map for a bounded listed firm population and a framework for distinguishing general AI language, implementation-specific technical naming, and quantified public claims in regulated capital market settings.
