Research
Theme 1: Systemic Consequence of Climate Change
Spillover Losses in Reinsurance Network under Common Shocks (with Qihe Tang and Hua Chen)
Keywords: Contagion; network effect; reinsurance; systemic risk.
We construct a reinsurance network where (re)insurers engage in reinsurance and retrocession transactions to transfer insured risks from policyholders. We investigate systemic risk within this network framework, measured by spillover loss occurring when liability payoffs decrease due to cascading defaults. Under tail dependence between primary losses, we analyze how integration of the reinsurance network shapes its spillover loss. Our findings reveal that integration exhibits a U-shaped relationship with spillover loss in reinsurance networks. We further implement a leave-one-out (LOO) approach to identify systemically important reinsurers and validate their importance through tests of bailout. Using data from the U.S. Property and Casualty reinsurance market, we identify a core-periphery structure and conduct extensive simulations that support our key findings.
Bayesian Learning of Regional Economic Impacts of Climate Change (with Qihe Tang)
Keywords: Bayesian learning; climate change; regional economic impacts; spatial dependence; uncertainty.
We develop a Bayesian learning framework to analyze the deep uncertainty surrounding the economic impacts of climate change across multiple regions. We disaggregate GDP losses from climate damages to capture regional heterogeneity and incorporate spatial dependence to reflect cross-regional correlations. The framework enables joint learning across regions by leveraging shared information. By characterizing the dynamics of uncertainty reduction, we estimate the years required to reach target confidence levels in damage assessments across countries and sectors. The results indicate that joint learning achieves faster convergence than individual learning when the uncertain parameters follow a multivariate normal distribution. Our empirical analysis is based on the FUND dataset and on Representative Concentration Pathways (RCPs) adopted by the Intergovernmental Panel on Climate Change (IPCC).
Awarded 2025–2026 UNSW Business School Sustainable Development Goals (SDGs) Research Grant
Theme 2: Electricity Price Forecasting with Weather Indices
Forecasting Australian Electricity Spot Prices with Weather Indices (with Katja Ignatieva and Han Li)
Keywords: Electricity price forecasting; interpretable models; spatial-temporal; weather indices.
We improve day-ahead electricity price forecasting in Australia by integrating linear time-series models with external spatial-temporal weather indices and demand factors. The model accommodates multiple seasonalities and employs distributed lag non-linear models (DLNM) to capture delayed and non-linear effects of weather variables (temperature, precipitation, humidity, wind speed) on price formation.
Theme 3: Stochastic Control Problems in Finance
Focus: Deep learning method; high-dimensional control problem; optimal retirement.
Peer-reviewed publications (Master’s):
- Ma, J. and Yang, S (2024). High-dimensional stochastic control models for newsvendor problems and deep learning resolution. Annals of Operations Research, 339, 789–811.
- Ma, J., Xing, J., and Yang, S. (2022). Dual control methods for a mixed control problem with optimal stopping arising in optimal consumption and investment. Numerical Mathematics: Theory, Methods and Applications, 15(3), 641-661.