"An Adaptive Residual-Based Test for Factor Structure", with Yayi Yan, Journal of Business & Economic Statistics (2025), 1-26. [Journal Link] [SSRN Link]
"Impact of Climate Change on Agriculture in Australia: An Interactive Fixed Effects Model Approach", with Natalia Bailey, Zvi Hochman, Mervyn Silvapulle, and Param Silvapulle, Applied Economics (2024), 57(43), 6901-6914. [Journal Link]
AI Technology Diffusion in the Stock Market [SSRN Link]
with Runfeng Yang, Massimiliano Caporin, and Juan‐Angel Jimenez‐Martin, ready for submission
Summary: The AI technology diffusion over the past decade remains less understood. In this paper, we address this question by studying the information content of the ``AI technology premium''. We construct novel, unified measures of firm-level AI exposures through product and labor channels and analyze the US stock market from 2012 to 2024. We document a significant and highly time-varying AI technology risk premium. The product premium is positive and driven primarily by large firms, with positive shocks observed only after 2018. In contrast, the labor premium is weaker and is primarily driven by small firms. The observed premium appears to be driven by a combination of slow realization, investor attention, and the pricing of uncertainties related to AI adoption. Overall, the adoption of AI technology is a complex process characterized by evolving investor perceptions and a gradual realization of its potential benefits.
Presented at: Universidad Complutense de Madrid; University of Pavia; IFABS 2024 Shanghai Conference, Shanghai; 23rd International Conference on Credit Risk Evaluation, Palazzo Franchetti, Venice; the Generative AI in Finance Conference, Montréal.
Sector Structure in Digital Asset Returns [SSRN Link]
with Runfeng Yang and Massimiliano Caporin, to be updated
Summary: We identify a sector structure within the digital asset market, where different types of digital assets (the "digital asset sectors") exhibit distinct risks and returns. We investigate the observed sector variation through two channels: the systematic risk channel and the idiosyncratic risk channel. Our findings indicate that sector variations within the systematic channel are insignificant. In contrast, sector variations within the idiosyncratic channel are substantial, driven by sector-specific events, sentiments, and inter-sector spillovers.
Presented at: University of Verona; 2nd Financial Econometrics Meets Machine Learning Conference, Università della Svizzera Italiana; 2024 Australasian Meeting of the Econometric Society, Monash University.
A Multilevel Latent Factor Model for Longitudinal Network and Relational Data
with Bin Peng, draft available upon request
Summary: This paper develops a methodology for estimating a multilevel latent factor model for longitudinal network and relational data, where the cross-section of the time series is comprised of entities acting as senders and receivers. The model consists of global factors and separate factors for senders and receivers in the network. This class of models are useful in several areas such as communication, economics, and health sciences. The current methodological literature is not sufficiently advanced for this family of models. We show that our proposed method consistently estimates the number of factors and the factor spaces. The asymptotic results are established for the case when the number of senders, receivers, and time periods are large. We illustrate the method by applying it to model volumes of bilateral trade among 23 countries.