Tractable bank capital structure: optimal control under Basel III constraints

By Erhan Bayraktar, Etienne Chevalier, Vathana Ly Vath, Yuqiong Wang

Rating

1850
Battle Count: 58

Relevance

3/10
The paper is primarily about bank capital regulation and corporate finance rather than trading strategies. However, it has indirect relevance: (1) understanding how Basel III constraints affect bank risk-taking and portfolio allocation informs market microstructure and credit risk models; (2) the stochastic control framework (singular-impulse control, viscosity solutions) is methodologically relevant to optimal execution and portfolio management; (3) the safety-profitability frontier analysis is relevant for risk management in financial institutions; (4) the state-dependent investment constraint structure could inform constrained portfolio optimization. The paper does not directly address trading signals, alpha generation, or market prediction.

Implementation Complexity

8/10
High complexity due to: (1) solving variational inequalities via viscosity solution theory requires advanced PDE/numerical analysis; (2) the combined singular-impulse control structure demands careful treatment of reflection barriers and jump conditions; (3) the state-dependent kinked investment constraint pi(y) adds nonlinearity; (4) Monte Carlo simulation for the outer regulatory optimization over a 3D parameter grid (a1, a2, a3) with 2000-5000 paths per configuration; (5) monotone finite-difference schemes with convergence tolerance 10^-6; (6) the delegation analysis requires nested optimization (regulator sets a, bank chooses y); (7) confidence-adjusted Pareto frontier construction. The GitHub repository provides solver scripts, but understanding and extending the model requires expertise in stochastic control, viscosity solutions, and numerical PDE methods.

Reproducibility

4/5
The paper provides a GitHub repository with solver scripts, simulation code, and parameter checks. Baseline parameter sets are fully specified (r=0.02, mu=0.04, mu_L=0.03, rho=0.12, gamma=0.02, sigma=0.08, sigma_L=0.03, c=0.20, kappa=0.01, kappa'=0.02, a1=0.045, a2=0.05, a3=0.30, y=1.02). Numerical methods are described (monotone finite-difference scheme, 301 state points, 151 control points, Euler-Maruyama with step 0.001, 2000-5000 MC paths). However, the parameters are illustrative rather than empirically calibrated, and the model is theoretical. The delegation analysis grids and full sensitivity experiments are documented in the archive.

About this paper

Methodology: Combined Singular-Impulse Stochastic Control with Viscosity Solutions. Problem types: Optimization, Risk Management, Portfolio Optimization, Stochastic Control.

The interactive Everscope explorer (charts, battles, favorites) loads below.