Dynamic Risk Transmission and Dependence Modeling: Copula Studies of China's Real Estate, Carbon Markets, and the AI-Energy Nexus

Abstract

This thesis dissects dynamic risk transmission within and across three critical interconnected markets---real estate and carbon trading in China and the global AI-energy nexus---employing advanced econometric and network methodologies. These markets represent critical sectors undergoing significant structural transitions, where understanding risk propagation mechanisms has become essential for financial stability and policy formulation. Despite extensive research on individual market risks, the complex, state-dependent, and potentially non-linear patterns of risk transmission across these sectors remain inadequately explored.

The first study examines China's real estate market (2006-2023) through a state-dependent vine copula network approach. The analysis reveals a persistent center-periphery structure where top-tier cities function as central risk nodes. During high-risk periods, risk contagion intensifies significantly, with network connectivity increasing by approximately 15. Macroeconomic factors---particularly GDP growth and inflation---substantially influence risk state transitions, with deteriorating fundamentals increasing the probability of entering high-risk regimes and reshaping network topologies.

The second study investigates the impact of China's 2021 national carbon market unification through a multi-layer network framework incorporating both copula-based dependencies and Diebold-Yilmaz spillovers. Following unification, market information efficiency improved substantially (transfer rate increased from 0.432 to 0.516), and the national market emerged as a central coordinator. Regional market roles underwent significant transformations, with Tianjin notably shifting from a primary risk receiver (net spillover 21.5) to a significant transmitter (net spillover 26.1), demonstrating how institutional reforms can fundamentally reshape risk transmission patterns.

The third study examines how the artificial intelligence expansion influences global energy market risk under extreme conditions. Using time-varying copulas and spillover indices, the research finds that AI exhibits relatively weak dependence on fossil fuels and carbon prices but significantly stronger connections with clean energy sectors, particularly photovoltaic. These technological sector linkages intensified markedly following the ChatGPT release (tail dependence increasing from 0.32 to 0.52), highlighting how technological breakthroughs can reshape market interdependencies beyond traditional energy-industrial relationships.