Dynamic Carbon-Budget Black--Litterman Portfolio Optimization under ETS Transition Risk: EU and UK Historical Learning Cases and a Thai Pre-ETS Application

Technical Summary: Dynamic Carbon-Budget Black–Litterman Portfolio Optimization under ETS Transition Risk

Problem Statement
The paper addresses a critical gap in sustainable portfolio construction for emerging markets transitioning toward carbon pricing. While institutional investors increasingly recognize carbon transition risk as a material financial factor, existing literature offers limited guidance for markets that lack a liquid domestic emissions trading system (ETS). Specifically, there is no established allocation rule for "pre-ETS" environments where investors must manage transition risk using firm-level carbon data and external carbon price signals (e.g., EU Emissions Trading System allowances) without a domestic compliance price. The core challenge is to design a portfolio framework that controls financed carbon intensity and hedges against transition shocks without resorting to mechanical exclusion rules or assuming the existence of a domestic allowance market.

Methodology
The authors propose a Dynamic Carbon-Budget Black–Litterman (BL) framework that integrates carbon transition views and a state-dependent carbon budget into a standard mean-risk optimization. The empirical design is bifurcated into two distinct blocks to ensure methodological rigor:

1. Operating-Market Learning Cases (EU & UK):

   • Data: Uses historical equity data (2000–2026) and a recent matched panel (2021–2026) for EU and UK firms in ETS-sensitive sectors.
   • Purpose: To validate the premise that high-transition firms are vulnerable during carbon-price stress and to test the carbon-budget BL logic in jurisdictions with active allowance markets.
   • Mechanism: Compares standard BL portfolios against dynamic carbon-budget portfolios that include a capped EUA futures sleeve. The dynamic strategy adjusts transition views and high-transition exposure caps based on EUA shock states.

2. Pre-ETS Application (Thailand):

   • Data: A historical backtest (Feb 2021–May 2026) on the Thai SET100 universe (91 equities).
   • Constraint: No domestic Thai allowance market exists; therefore, EUA futures are treated strictly as an external transition-risk signal and hedge, not a compliance asset.
   • Optimization Model:
     • Prior: Market-implied equilibrium returns derived from market capitalization weights.
     • Views: Transition views are constructed by comparing baskets of low-transition vs. high-transition firms. The strength of these views is modulated by a risk-adjusted EUA momentum signal ($z_{EUA}$).
     • Carbon Budget: A binding constraint limits the portfolio's weighted carbon intensity relative to a reference baseline. The budget is dynamic: $b_t = \text{clip}(0.90 - 0.08z_{EUA}, 0.72, 0.95)$. When external carbon prices rise (positive $z_{EUA}$), the budget tightens, forcing lower carbon intensity.
     • Hedge: A capped (10%) allocation to EUA futures is permitted only in the dynamic strategy to hedge transition risk.
     • Objective: A concave mean-risk-downside-turnover problem incorporating Black–Litterman posterior returns, a lower-partial-moment downside penalty, and turnover constraints.

Key Contributions

• Framework Innovation: The paper develops a transparent, state-contingent allocation rule that treats carbon exposure as a binding input rather than an ex-post label. It bridges the gap between operating-market evidence and pre-ETS portfolio construction.
• Empirical Separation: By strictly separating EU/UK operating-market data from the Thai pre-ETS application, the study avoids the identification error of treating foreign allowance prices as domestic compliance prices.
• Dynamic Risk Control: The methodology demonstrates how to integrate external carbon signals (EUA futures) to dynamically adjust carbon budgets and transition views, rather than applying static exclusion screens.
• Asymmetric Insurance Logic: The framework is explicitly designed as a risk-control overlay that performs well during transition tightening but accepts costs during policy relaxation or carbon-market liquidity shocks.

Results

• EU/UK Operating-Market Evidence:

  • Long-history diagnostics (2000–2026) show that high-transition firms underperform the local market during specific ETS phases, particularly in the UK (Phase IV) and EU (Phase I–II), though the effect is not universal across all periods.
  • Recent EUA-shock diagnostics reveal that high-transition firms underperform significantly on top EUA shock days in specific jurisdictions (e.g., Italy, Germany), but not universally.
  • Portfolio Backtests: The dynamic carbon-budget BL strategy successfully reduced high-transition exposure (e.g., from ~69% to ~27% in the UK) and improved returns on top EUA shock days compared to standard BL. However, this came at the cost of underperformance during "relaxation" states (bottom EUA days), confirming the insurance trade-off.

• Thai Pre-ETS Application:

  • Performance: The dynamic carbon-budget strategy achieved a Sharpe ratio of 0.467 (full walk-forward) and 0.786 (24-month forward test), outperforming both standard BL (0.441/0.723) and screened BL (0.450/0.729).
  • Risk Metrics: The dynamic strategy reduced annualized volatility (13.35% vs. 13.71%), maximum drawdown (-25.22% vs. -25.82%), and expected shortfall.
  • Carbon Outcomes: The strategy reduced financed carbon intensity by 40.30% relative to the reference baseline, compared to a 31.52% reduction for the static screened strategy. This was achieved without excessive turnover (avg. monthly turnover ~9.8%).
  • Stress Testing: Under modeled ETS tightening and CBAM-style repricing scenarios, the dynamic strategy incurred significantly lower expected 20-day losses (2.25% and 3.44%, respectively) compared to standard BL (4.54% and 4.87%). Conversely, under policy relaxation and carbon-liquidity stress, the dynamic strategy incurred higher losses, validating its asymmetric risk profile.

Significance and Claims
The paper modestly claims that the proposed method serves as a transparent transition-risk-control overlay for pre-ETS investors, rather than an unconditional alpha generator.

• Not a Universal Alpha: The authors emphasize that the strategy is not evidence of unconditional outperformance; it is state-dependent, excelling when transition risk tightens but underperforming when policy relaxes or external carbon instruments sell off.
• Pre-ETS Preparedness: The study provides a practical workflow for investors in markets like Thailand to prepare for carbon pricing by decoupling internal carbon control (via firm-level intensity and BL views) from external hedging (via EUA futures), without falsely assuming a domestic allowance market exists.
• Risk Budgeting Interpretation: The results support interpreting carbon transition risk through a risk-budgeting lens—where carbon exposure is actively managed and hedged—rather than through mechanical divestment.
• Limitations: The authors acknowledge limitations including the use of current-investable-universe data (potential survivorship bias), the reliance on sector-median imputation for missing revenue data, and the fact that EUA futures are an external proxy rather than a domestic Thai instrument. They note that the framework requires further production-level stress testing regarding FX hedging, collateral, and transaction costs before full institutional deployment.