
AI assistance can change mortgage application incentives and the composition of borrowers reaching a credit decision. We develop a lender-choice and documentation-continuation model with general differentiable documentation and perceived-cost functions. The framework admits privacy concerns and trust improvements, distinguishes participation and funded-volume thresholds, and shows how conditional approval can move in either direction under fixed underwriting. Public HMDA records provide a descriptive application to decision measurement. At a common application mix, Better Mortgage’s 2024–2025 purchase change relative to 30 comparison institutions combines higher recorded administrative resolution and funded conversion with lower conditional approval. Documentation completion and borrower-level AI exposure remain latent. Across 33 selected reporting entities, seven purchase comparisons change sign under a common denominator-classification fraction. Using five- and ten-percentage-point classification-spread benchmarks, chosen for sensitivity rather than calibrated to observed practice, makes 11 and 13 documentation-only comparisons direction-ambiguous. Better’s 32-peer comparison ranges from −3.80 to +5.82 percentage points at the classification endpoints. These are sharp measurement ranges conditional on declared restrictions rather than estimates of the correct latent classification or AI effects. The contribution combines general selection mechanisms with transparent measurement sensitivity in recorded lending outcomes.
Banking automation changes both the work reaching employees and the incentive to retain human-service capacity. We model customer automation, employee assistance and endogenous staffing. A marginal staffing-cost elasticity boundary determines whether routine automation lengthens or shortens retained completion time. Consumer urgency creates a wedge between private adoption incentives and resource-plus-delay objectives. Heterogeneous work and unsuccessful automation separate case counts from workload, while the proposed safeguard combines successful escalation with conditional mean and tail commitments. The model yields operationally defined quantities for evaluating human recourse. Complementary public evidence examines how contact and waiting difficulties are recorded in complaints. Of 15,211 geographically mapped US complaints, 4847 have published narratives. Independent coding and adjudication of 320 probability-sampled narratives yield weighted precision of 0.753 and recall of 0.177 for the literal contact/wait measure. Human-reference corrections reverse the signs of the original branch-scarcity point estimates, and the associated sampling interval includes zero. This result demonstrates the sensitivity of reported geographical patterns to complaint classification. Together, the theory and evidence motivate separate measurement of routine resolution, retained workload, access to human help and end-to-end completion.
This paper examines whether apparent dependence among artificial intelligence (AI) newspaper coverage, cryptocurrencies, financial-market stress, and technology-related equity portfolios reflects distinct intersectoral transmission or common asset-pricing exposures. We analyze 2016 aligned daily changes from January 2018 to April 2026 using generalized and clustered vector-autoregressive forecast-error variance decompositions, common equity-factor conditioning, alternative partitions, rolling estimation, and paired moving-block inference. Baseline clustered connectedness is 18.731% (16.011 percentage points within clusters and 2.719 across clusters), versus 20.685% under generalized identification. Residualizing fintech, cybersecurity, and clean-energy equity returns on SPY and QQQ reduces clustered connectedness to 6.741%. The reduction persists with VAR(2) and VAR(3), alternative forecast horizons, Monday exclusion, and economically plausible cluster structures. With 2000 paired bootstrap draws, the 95% intervals for adjusted-minus-unadjusted total connectedness are [−13.676, −10.207], [−13.777, −10.220], and [−13.815, −10.107] percentage points for VAR(1)–VAR(3). Cross-cluster attribution remains partition-dependent; one data-driven grouping yields essentially no reduction in its small cross-cluster component. HAC-robust tests across alternative AI transformations and lag orders show no incremental AI-news predictability after adjustment for multiple testing. The results favor an identification-aware, factor-conditioned account of digital-financial connectedness rather than a structural claim about technology shocks or economic contagion.
This study examines how banking professionals in Morocco perceive the role of artificial intelligence (AI) in assessing the credit risk of small and medium-sized enterprises (SMEs). Based on a survey of 300 banking professionals, it explores whether perceptions of AI’s analytical capabilities, its limitations, information asymmetry, borrower behavioral uncertainty, and SME financing constraints are related to the perceived improvement in credit risk assessment. The study uses descriptive statistics, factor analysis, and generalized ordered logistic regression. The results show that banking professionals recognize AI’s potential to process information, identify risks, and support credit analysis, but their overall views of its contribution remain cautious. Professionals who have more positive perceptions of AI’s analytical capabilities tend to believe more strongly that it can improve credit risk assessment. Conversely, those who perceive greater limitations tend to be less convinced of its contribution. Both relationships are statistically significant. However, the results do not show a clear relationship between perceived information asymmetry and the perceived contribution of AI. Similarly, perceived borrower behavioral uncertainty and financing constraints are not significantly associated with this assessment. Overall, the findings show that banking professionals recognize the potential benefits of AI while remaining cautious about its contribution to SME credit risk assessment.
Proof-of-Work (PoW) cryptocurrencies create electricity demand globally, but their carbon consequences depend on where the electricity is generated. This study presents CryptoCarbonApp, an open-source R Shiny business intelligence application that expresses network-level energy and emissions estimates for four PoW cryptocurrencies through the electricity system of Jordan, whose generation mix averages about 0.39 kgCO2/kWh. The analysis reports network totals with conditional sensitivity ranges and examines an assumed spatial allocation of mining activity with a 20,000-replication Monte Carlo procedure. On Jordan’s mix, Bitcoin’s network demand corresponds to about 66 MtCO2 per year (50–83 MtCO2 for a ±25% change in network consumption); because this demand is roughly eight times Jordan’s national electricity supply, the figure is an intensity benchmark, not a forecast. A 50 MW facility would emit about 154 ktCO2 per year. The allocation procedure cannot identify actual mining hotspots, and raising the renewable share of supply from 10% to 60% reduces emissions by 55.6% under static accounting. At the September 2025 hashprice, efficient hardware breaks even at about USD 0.12/kWh before other costs, below Jordan’s business tariff of USD 0.169/kWh. Energy labelling at supervised virtual asset service providers is therefore proposed as a conditional, demand-side complement to mining-side instruments.