Selected work

Representative Papers

Behavioural feasible set: Value alignment constraints on AI decision support

Mar 22, 2026

This study addresses how organizations adopting commercial AI decision-support systems often passively accept vendors’ embedded and non-negotiable value judgments, thereby constraining their own decision flexibility. The paper introduces the concept of the “behaviorally feasible set” to formally characterize the range of recommendations an AI system can generate under value-alignment constraints and identifies critical conditions under which organizational needs exceed the system’s adaptive capacity. Through controlled experiments comparing binary decisions and multi-stakeholder preference rankings, the research demonstrates that value alignment substantially shrinks the behaviorally feasible set, diminishing the system’s responsiveness to legitimate contextual variation. Commercial models exhibit heightened rigidity, and the alignment process systematically shifts—rather than neutralizes—stakeholder priorities, revealing that value alignment functions as a structural mechanism embedding vendor values and narrowing organizational negotiation space.

0 citationsRead paper

Forecasting House Prices

Sep 25, 2025

This study investigates the core drivers of housing price growth across 13 advanced economies from 1988 to 2023. Moving beyond conventional linear assumptions, we develop a cross-national housing price forecasting model based on Breiman’s random forest algorithm, incorporating ten macroeconomic and financial variables—including price momentum, rent-price ratio, and household credit growth—and employ Shapley values for feature importance quantification and partial dependence analysis to uncover nonlinear mechanisms (e.g., inflation’s threshold effects). The model achieves robust out-of-sample generalization across countries without country fixed effects. Relative to an OLS benchmark, it reduces out-of-sample prediction error substantially: RMSE decreases by 44% and MAE by 45%, demonstrating superior accuracy and stability. Our key contributions are (i) identifying critical nonlinear housing price drivers and (ii) empirically validating the cross-country applicability, predictive power, and interpretability advantages of data-driven methods in macro-housing modeling.

0 citationsRead paper

Towards Verifiability of Total Value Locked (TVL) in Decentralized Finance

May 20, 2025

Total Value Locked (TVL) in DeFi lacks standardized, on-chain verifiable computation, relying instead on community-reported and opaque off-chain data—hindering auditability and transparency. Method: We systematically audit 939 Ethereum-based DeFi protocols, identifying 68 non-standard TVL query patterns and finding that 10.5% depend on external servers. Leveraging bytecode and event-log analysis, methodological TVL auditing, and standardized query pattern recognition, we propose “verifiable TVL” (vTVL)—a metric computed exclusively from on-chain state and standard balance queries to ensure full reproducibility and verifiability. Contribution/Results: We introduce the first vTVL benchmarking framework; across 400 protocols, vTVL matches publicly reported values in 46.5% of cases. We further formulate design guidelines for verifiability-aware DeFi metrics, advancing TVL standardization, transparency, and auditability.

0 citationsRead paper

Recent publications

Latest Papers

Behavioural feasible set: Value alignment constraints on AI decision support

Mar 22, 2026

This study addresses how organizations adopting commercial AI decision-support systems often passively accept vendors’ embedded and non-negotiable value judgments, thereby constraining their own decision flexibility. The paper introduces the concept of the “behaviorally feasible set” to formally characterize the range of recommendations an AI system can generate under value-alignment constraints and identifies critical conditions under which organizational needs exceed the system’s adaptive capacity. Through controlled experiments comparing binary decisions and multi-stakeholder preference rankings, the research demonstrates that value alignment substantially shrinks the behaviorally feasible set, diminishing the system’s responsiveness to legitimate contextual variation. Commercial models exhibit heightened rigidity, and the alignment process systematically shifts—rather than neutralizes—stakeholder priorities, revealing that value alignment functions as a structural mechanism embedding vendor values and narrowing organizational negotiation space.

0 citationsRead paper

Forecasting House Prices

Sep 25, 2025

This study investigates the core drivers of housing price growth across 13 advanced economies from 1988 to 2023. Moving beyond conventional linear assumptions, we develop a cross-national housing price forecasting model based on Breiman’s random forest algorithm, incorporating ten macroeconomic and financial variables—including price momentum, rent-price ratio, and household credit growth—and employ Shapley values for feature importance quantification and partial dependence analysis to uncover nonlinear mechanisms (e.g., inflation’s threshold effects). The model achieves robust out-of-sample generalization across countries without country fixed effects. Relative to an OLS benchmark, it reduces out-of-sample prediction error substantially: RMSE decreases by 44% and MAE by 45%, demonstrating superior accuracy and stability. Our key contributions are (i) identifying critical nonlinear housing price drivers and (ii) empirically validating the cross-country applicability, predictive power, and interpretability advantages of data-driven methods in macro-housing modeling.

0 citationsRead paper

Towards Verifiability of Total Value Locked (TVL) in Decentralized Finance

May 20, 2025

Total Value Locked (TVL) in DeFi lacks standardized, on-chain verifiable computation, relying instead on community-reported and opaque off-chain data—hindering auditability and transparency. Method: We systematically audit 939 Ethereum-based DeFi protocols, identifying 68 non-standard TVL query patterns and finding that 10.5% depend on external servers. Leveraging bytecode and event-log analysis, methodological TVL auditing, and standardized query pattern recognition, we propose “verifiable TVL” (vTVL)—a metric computed exclusively from on-chain state and standard balance queries to ensure full reproducibility and verifiability. Contribution/Results: We introduce the first vTVL benchmarking framework; across 400 protocols, vTVL matches publicly reported values in 46.5% of cases. We further formulate design guidelines for verifiability-aware DeFi metrics, advancing TVL standardization, transparency, and auditability.

0 citationsRead paper