A University at Buffalo researcher has developed a framework to help scientists incorporate community input into Earth system models, tools that simulate climate as well as chemical and biological processes.

Abstract

Purpose – Climate change has emerged as one of the new sources of financial risk, but it is still not recognized as a significant influencing factor in existing studies, especially in China. This study aims to investigate how climate policy changes in China affect intersectoral systemic risk from a mixed frequency model perspective.

Design/methodology/approach – We include asymmetric tail long memory for the dependence, which has not been covered by other risk-related literature, in the study of China’s sector risk contribution by proposing the TVM-MIDAS Copula model-based MES approach. Besides, we construct the GARCH-MIDAS-CPU model to investigate the impact of CPU on the contribution of systemic risk in the sector.

Findings – The results show that the real estate sector has the greatest tail dependence on the market, the raw materials sector has the longest memory of upper tail dependence, and the consumer sector has a weaker link to the market. For CPU, when the market falls moderately, CPU amplifies the volatility of the systematic risk contribution of the energy, materials, industrials, and real estate sectors and reduces the volatility of the risk contribution of the consumer, healthcare, and financial sectors. When the market plummets, the CPU amplifies the intensity of the volatility of systemic risk contributions from all sectors except the healthcare sector.

Originality/value – First, this paper analyzes how CPU influences systemic risk within Chinese sectors, offering confident evidence of the link between climate policy changes and sectoral risks. Second, it proposes a TVM-MIDAS copula model to capture dynamic tail dependence with tail memory advantages. Third, it utilizes a GARCH-MIDAS-CPU mixed-frequency model to examine the heterogeneous impact of CPU on systemic risk across sectors, addressing the co-frequency data downsampling issue and providing more precise insights.