Selected work

Representative Papers

New Confidence Regions for Linear Regression Parameters with Stationary-Ergodic Dependent Errors

May 19, 2026

This study addresses the challenge of constructing valid joint confidence regions for linear regression coefficients when regression errors exhibit unknown serial dependence and are jointly stationary and ergodic with the covariates. The authors propose a novel approach that avoids explicit modeling of the error dependence structure by introducing independent auxiliary samples and applying stochastic smoothing with a decaying bandwidth to both the regression function and second moments. Coupled with data-driven bandwidth selection and mild truncation, this method yields Wald-type confidence regions and simultaneous confidence intervals. It does not rely on long-run variance estimation or parametric assumptions about dependence, achieving coverage probabilities close to nominal levels across diverse dependence structures—including ARMA, ARFIMA, copula-based Markov processes, and fractional Gaussian noise—while producing smaller confidence region volumes than Newey–West HAC and MAC methods. The approach is successfully demonstrated in an analysis of Beijing PM2.5 data.

0 citationsRead paper

Penalized KLIC Model Selection for the Generalized Method of Moments in Longitudinal Data with Time-Dependent Covariates

May 05, 2026

Model selection plays an important role in longitudinal data analysis, especially when models are estimated using the generalized method of moments (GMM) in the presence of time-dependent covariates. In this setting, the number of valid moment conditions can grow quickly and may lead to over-parameterized models. The Kullback--Leibler Information Criterion (KLIC) has been proposed as a model-selection tool for this framework; however, the original KLIC criterion may favor overly complex models when the number of parameters or valid moment conditions increases. To address this limitation, this study proposes two penalized versions of KLIC that incorporate penalties based on both the number of model parameters and the number of valid moment conditions. The proposed criteria are referred to as the Moment--Parameter Product Penalty KLIC (MPPP--KLIC) and the Logarithmic Penalty KLIC (LP--KLIC). These criteria provide a theoretically motivated mechanism for balancing model fit and model complexity in GMM-based longitudinal models. Through an extensive simulation study involving both binary and continuous response settings, the proposed criteria are shown to improve the ability of KLIC to distinguish among competing models and to reduce the selection of over-parameterized models. The performance of the proposed methods is further illustrated using the Filipino Child Morbidity dataset, a longitudinal study of child health in the Philippines. The results show that the proposed penalized criteria provide stable and interpretable model rankings and consistently identify age as the most important predictor of child morbidity. Overall, the proposed penalized KLIC criteria offer practical and theoretically grounded tools for model selection in GMM-based longitudinal data analysis with time-dependent covariates.

0 citationsRead paper

LLM-Based Multi-Agent System for Simulating and Analyzing Marketing and Consumer Behavior

Oct 20, 2025

Traditional consumer behavior modeling relies on post-hoc analysis and rule-based agent models, limiting its capacity to capture cognitive complexity and emergent social dynamics. To address this, we propose a large language model (LLM)-driven generative multi-agent system that abandons predefined rules and instead enables dynamic simulation of consumer decision-making, habit formation, and social diffusion through natural-language interaction, internal cognitive modeling, and co-evolutionary learning. Deployed in a price-promotion sandbox environment, the system autonomously generates interpretable strategic feedback and— for the first time—uncovers latent population-level consumption patterns and social cascade effects. Compared to conventional approaches, our framework achieves higher ecological validity, lower experimental cost, and superior scalability, establishing a novel computational experimentation paradigm for pre-testing marketing strategies.

0 citationsRead paper

An Extension of the d-Variate FGM Copula with Application

Sep 08, 2025

Classical d-variate FGM copulas inadequately capture complex, high-dimensional dependence structures inherent in bearing degradation data. Method: We propose an extended FGM copula model based on Legendre polynomials, enabling explicit, closed-form estimation of high-dimensional FGM parameters for the first time. Under i.i.d. assumptions, we derive a parsimonious, interpretable, and computationally efficient version via simulation studies and model selection criteria (e.g., BIC). Contribution/Results: We rigorously prove that the proposed estimator is unbiased, consistent, and asymptotically normal. Empirical evaluation demonstrates that classical FGM copulas severely underfit bearing data, whereas our model significantly improves dependence modeling accuracy. Crucially, it automatically identifies the optimal low-dimensional parameter structure—balancing theoretical rigor with practical applicability in reliability engineering and prognostics.

0 citationsRead paper

Recent publications

Latest Papers

New Confidence Regions for Linear Regression Parameters with Stationary-Ergodic Dependent Errors

May 19, 2026

This study addresses the challenge of constructing valid joint confidence regions for linear regression coefficients when regression errors exhibit unknown serial dependence and are jointly stationary and ergodic with the covariates. The authors propose a novel approach that avoids explicit modeling of the error dependence structure by introducing independent auxiliary samples and applying stochastic smoothing with a decaying bandwidth to both the regression function and second moments. Coupled with data-driven bandwidth selection and mild truncation, this method yields Wald-type confidence regions and simultaneous confidence intervals. It does not rely on long-run variance estimation or parametric assumptions about dependence, achieving coverage probabilities close to nominal levels across diverse dependence structures—including ARMA, ARFIMA, copula-based Markov processes, and fractional Gaussian noise—while producing smaller confidence region volumes than Newey–West HAC and MAC methods. The approach is successfully demonstrated in an analysis of Beijing PM2.5 data.

0 citationsRead paper

Penalized KLIC Model Selection for the Generalized Method of Moments in Longitudinal Data with Time-Dependent Covariates

May 05, 2026

Model selection plays an important role in longitudinal data analysis, especially when models are estimated using the generalized method of moments (GMM) in the presence of time-dependent covariates. In this setting, the number of valid moment conditions can grow quickly and may lead to over-parameterized models. The Kullback--Leibler Information Criterion (KLIC) has been proposed as a model-selection tool for this framework; however, the original KLIC criterion may favor overly complex models when the number of parameters or valid moment conditions increases. To address this limitation, this study proposes two penalized versions of KLIC that incorporate penalties based on both the number of model parameters and the number of valid moment conditions. The proposed criteria are referred to as the Moment--Parameter Product Penalty KLIC (MPPP--KLIC) and the Logarithmic Penalty KLIC (LP--KLIC). These criteria provide a theoretically motivated mechanism for balancing model fit and model complexity in GMM-based longitudinal models. Through an extensive simulation study involving both binary and continuous response settings, the proposed criteria are shown to improve the ability of KLIC to distinguish among competing models and to reduce the selection of over-parameterized models. The performance of the proposed methods is further illustrated using the Filipino Child Morbidity dataset, a longitudinal study of child health in the Philippines. The results show that the proposed penalized criteria provide stable and interpretable model rankings and consistently identify age as the most important predictor of child morbidity. Overall, the proposed penalized KLIC criteria offer practical and theoretically grounded tools for model selection in GMM-based longitudinal data analysis with time-dependent covariates.

0 citationsRead paper

LLM-Based Multi-Agent System for Simulating and Analyzing Marketing and Consumer Behavior

Oct 20, 2025

Traditional consumer behavior modeling relies on post-hoc analysis and rule-based agent models, limiting its capacity to capture cognitive complexity and emergent social dynamics. To address this, we propose a large language model (LLM)-driven generative multi-agent system that abandons predefined rules and instead enables dynamic simulation of consumer decision-making, habit formation, and social diffusion through natural-language interaction, internal cognitive modeling, and co-evolutionary learning. Deployed in a price-promotion sandbox environment, the system autonomously generates interpretable strategic feedback and— for the first time—uncovers latent population-level consumption patterns and social cascade effects. Compared to conventional approaches, our framework achieves higher ecological validity, lower experimental cost, and superior scalability, establishing a novel computational experimentation paradigm for pre-testing marketing strategies.

0 citationsRead paper

An Extension of the d-Variate FGM Copula with Application

Sep 08, 2025

Classical d-variate FGM copulas inadequately capture complex, high-dimensional dependence structures inherent in bearing degradation data. Method: We propose an extended FGM copula model based on Legendre polynomials, enabling explicit, closed-form estimation of high-dimensional FGM parameters for the first time. Under i.i.d. assumptions, we derive a parsimonious, interpretable, and computationally efficient version via simulation studies and model selection criteria (e.g., BIC). Contribution/Results: We rigorously prove that the proposed estimator is unbiased, consistent, and asymptotically normal. Empirical evaluation demonstrates that classical FGM copulas severely underfit bearing data, whereas our model significantly improves dependence modeling accuracy. Crucially, it automatically identifies the optimal low-dimensional parameter structure—balancing theoretical rigor with practical applicability in reliability engineering and prognostics.

0 citationsRead paper