My research develops econometric and statistical methods for financial, economic, and environmental data. A central theme is robust inference and real-time monitoring for dependent data, including Adjusted-Range Self-Normalisation, long-run variance inference, structural-change detection, functional time series, factor models, and distributional Granger causality.
On the applied side, my work connects these methods to risk monitoring in financial derivatives and cryptocurrency markets, machine-learning-based measurement of digital-finance competitiveness, and environmental management and ecological economics, including virtual water in global supply chains. Across these projects, I aim to link econometric theory with practical tools for empirical research, regulation, and teaching.
Selected working papers and ongoing research projects are listed below.
This working paper proposes an adjusted-range-based self-normalized sequential monitoring scheme (RSMS) for online change-point detection, offering significant improvements in performance, especially in the presence of structural shifts. The method avoids tuning parameter choices, making it more robust and efficient for high-frequency data.
Using city-specific synthetic controls for 99 prefectures (2004–2017), this paper studies the 2013 Yangtze River Delta enlargement and documents a ‘Winner-Takes-More’ pattern with limited cross-boundary spillovers.
This paper models Hukou as a residency-contingent membership rule that prices access to local public goods, and uses an original survey from Beijing, Shanghai, and Guangzhou to quantify impacts on spending, subsidies, and satisfaction.
This working paper develops a unified framework for online monitoring of distributional Granger causality using self-normalized score detectors and anytime-valid quantile e-processes, with an empirical application to BTC downside-risk monitoring using Deribit option signals.
This working paper develops tuning-lean, self-normalized sequential monitors based on discrete Fourier transform (DFT) spectral scores to detect factor-loading instability in large approximate factor models, delivering stable false-alarm control and faster detection of structural change.
This working paper develops adjusted-range self-normalized inference for relevant change-point detection in weakly dependent functional time series, delivering tuning-free procedures and an empirical application to Bitcoin options implied-volatility smiles.
This working paper develops a consistent nonparametric test for whether a subset of conditioning variables can be omitted from the conditional variance function in a heteroscedastic regression model, with asymptotic theory, Monte Carlo evidence, and a Canadian interest-rate illustration.
This working paper reviews how machine learning contributes to economics across prediction, measurement, causal inference, policy learning, structural computation, finance, and operational deployment.
This working paper studies online model-maintenance rules for functional representations, comparing HAC, Shao-type self-normalization, and adjusted-range self-normalization after training-anchored FPCA compression.
This working paper applies GLMY path homology to transaction-level Bitcoin and Ethereum options data, measuring higher-order directed topology in implied-volatility spillovers and signed trading-pressure networks.
This working paper develops a mathematical-sciences framework for trustworthy AI corpora, organizing corpus trustworthiness around representativeness, consistency, fairness, privacy, and robustness across the lifecycle of collection, cleaning, augmentation, and auditing.
Revise and resubmit at the Journal of Financial Econometrics. This working paper develops tuning-free, self-normalized tests for economically relevant functional Granger causality in Hilbert-space time series, with an application to volatility spillovers between Bitcoin options and spot markets.