A Dynamic Risk-Adjusted Performance Framework for Bank Portfolios

Thesis Abstract

This study addresses the rising need for robust performance evaluation in bank portfolios under dynamic risk conditions, where traditional static metrics fail to capture time-varying risk-return trade-offs and regulatory capital pressures. The problem is twofold first, portfolio performance measures often overlook state-dependent risk, liquidity constraints, and macroprudential spillovers; second, banks face challenges in aligning risk-adjusted returns with capital requirements and long-horizon stability. The aim is to develop and validate a Dynamic Risk-Adjusted Performance Framework (DRAPF) that integrates time-varying risk, transaction costs, liquidity risk, and regime-switching behavior into performance attribution and optimization. Specific objectives include (1) constructing a dynamic, risk-adjusted performance metric that extends standard Sharpe and Information Ratios with macroeconomic regime indicators; (2) modelling portfolio-level risk using a Markov-switching GARCH framework to capture volatility clustering and regime shifts; (3) embedding a capital-aware constraint via a dynamic risk-adjusted efficiency frontier that incorporates Basel III/IV liquidity and capital metrics; (4) assessing out-of-sample predictive performance and stability across banking sector sub-portfolios; and (5) deriving practical decision rules for portfolio rebalancing that balance expected returns, risk, and capital costs. Methodologically, the study adopts a quantitative, explanatory design with a longitudinal panel of 40 mid-sized commercial banks across a four-year horizon (2019–2022) from a diversified set of developed and emerging markets. A stratified random sample yields 320 bank-quarter observations, ensuring representation across lending, trading, and investment portfolios. Data sources include bank financial statements, regulatory reports, market data from Bloomberg, and macroeconomic indicators from the IMF. Primary instruments comprise portfolio return series, risk metrics (value-at-risk, expected shortfall), liquidity proxies (bid-ask spreads, liquid asset ratios), and capital measures (leverage ratio, liquidity coverage ratio). The DRAPF integrates (i) a regime-switching risk model using a Markov-switching GARCH(1,1) to estimate time-varying conditional variances and regime probabilities; (ii) a dynamic risk-adjusted performance metric built upon a state-dependent extension of the Treynor–Spike framework and a Regime-Switching Information Ratio; (iii) a capital-aware frontier derived from a dynamic programming approach that optimizes risk-adjusted returns subject to Basel-compliant liquidity and capital constraints; and (iv) an optimization module implementing a receding-horizon rebalancing policy. Validation employs out-of-sample forecasting tests, Diebold–Mariano comparisons for predictive accuracy, and bootstrap methods to assess parameter stability. Instrument validity is examined via content triangulation with regulatory disclosures, while reliability is ensured through cross-validation and robustness checks across regimes and sub-samples. Expected findings indicate that the DRAPF will outperform conventional risk-adjusted measures in capturing regime-dependent performance and in predicting subsequent bank portfolio returns under stress periods. It is anticipated that portfolios operating in higher-liquidity regimes will exhibit superior risk-adjusted performance when incorporating dynamic capital costs, whereas in stressed regimes, risk-adjusted performance will hinge more on liquidity management and hedging effectiveness. The framework should reveal that the inclusion of regime-dependent risk and capital constraints yields a more stable efficiency frontier over time, reducing tail risk and improving capital efficiency without sacrificing return potential. The study contributes to knowledge by integrating regime-switching risk dynamics with capital-aware performance measurement, extending existing theories on dynamic asset allocation under regulatory constraints, and offering a parsimonious yet implementable framework for banks to evaluate and optimize portfolio performance in a changing macro-financial landscape. Practical implications include guidance for risk-management practitioners on timing and sizing of rebalancing decisions, policy-makers on the interplays between market risk, liquidity risk, and capital requirements, and researchers on the integration of macroprudential considerations into performance analytics. The main conclusion posits that a dynamic, regime-aware, capital-conscious performance framework provides superior explanatory power and decision support for bank portfolio management, and it is recommended that banks adopt DRAPF components in their risk governance and strategic asset allocation processes, with ongoing calibration to evolving regulatory standards and market conditions.