Abstract
dc:description.abstractThe objective of this thesis is to evaluate how today's portfolio decisions perform across the plausible future states of the world generated by climate and policy outcomes. We argue that the core problem for long-horizon institutional investors, such as pension funds, is not merely identifying sustainable assets, but rigorously stress-testing investment choices against forward-looking climate-economic scenarios. We execute this research in the following chapters: We establish in Chapter 1 that climate change poses a material threat to long-term institutional investors. Physical risks and transition risks affect asset values, yet financial markets price these risks only partially. International bodies document these risks and provide scenarios, but none tell investors how to construct portfolios that reflect investors’ risk appetites on climate change risk. We then identify a critical gap in the academic literature that motivates the thesis. To bridge this gap, we propose a framework that evaluates portfolio decisions across plausible climate futures. The framework draws on four strands of literature: 1. Literature on climate and the economy motivates the regime structure. 2. Climate portfolio analysis literature informs the empirical strategies we develop. 3. Divestment research motivates the decarbonization strategies. 4. Climate and bond markets research motivate the fixed income model. Finally, we define the key concepts that underpin the entire thesis. We categorize climate risk into physical and transition components and represent climate uncertainty through three regimes. We thus define the problem, identify the gap, state the contribution, and establish the building blocks for the chapters that follow Chapter 1. In Chapter 2, we evaluate the performance of a representative portfolio under a range of climate-related constraints. Using a geographically diverse dataset, we conduct a series of analyses to evaluate portfolio performance. First, we find the Sharpe-ratio-maximizing portfolio using the in-sample data. We then use a stepwise optimization that adds one climate constraint at a time to this portfolio. We consider five climate metrics: the ESG scores, the environmental component of ESG, the UNPRI threshold, the Disaster Risk Index (DRI), and sector-level Impact Factors. These climate metrics were not originally proposed for investment purposes. Whether any of them work well in portfolio optimization is not known in advance. Second, we assess the out-of-sample performances to evaluate the impact of the climate related constraints. Third, assuming perfect foresight, we identify the best-performing portfolio to explore the potential benefits of optimal climate metric selection. Finally, we compare the climate-smart and traditional portfolios against a robust portfolio optimization strategy. Our findings reveal that historical performance is not a reliable indicator of future outcomes. Metrics that perform well in-sample do not necessarily maintain their superiority out-of-sample. Under the assumption of perfect foresight, we observe that the sequence in which climate metrics are incorporated significantly influences portfolio performance. Among the five climate metrics considered, ESG proves the most beneficial overall, suggesting that social and governance factors play a crucial role in enhancing the effectiveness of environmental initiatives, but the results are sensitive to the in-sample period. In Chapter 3, we shift from a backward-looking empirical analysis to forward-looking scenario modeling. We develop a regime-switching model where climate policy determines the economic environment. Within each regime, we model sector-specific returns, which incorporate sector and regime dependent effects of transition and physical climate risks. Transition risk follows a three-phase cycle of initial decline, plateau, and recovery while physical risk applies a constant penalty over time. These adjustments are educed from historical data and climate science literature. We assume that strategies for transitioning to a low carbon economy are exogenous to the pension plan, and these are modeled using different climate regimes. The endogenous decision variable we examine is the choice between three decarbonization strategies for the pension plans' asset portfolio. This allows us to analyze the efficacy of different portfolio decarbonization pathways under climate scenario uncertainty. Our results show that the decision to decarbonize consistently outperforms inaction across the climate scenarios considered. These findings persist when investor behavioral biases, including recency bias and climate skepticism, are incorporated. However, significant tail risk persists across all strategies, reflecting the potential for systemic market failure under adverse climate outcomes. Overall, the analysis suggests that pension plan fiduciaries can pursue decarbonization not despite their financial obligations but in alignment with them, provided climate risks are properly integrated into long-term investment modeling. In Chapter 4, we develop a climate-augmented Economic Scenario Generator (ESG) to evaluate how a traditional investor accumulates wealth under climate uncertainty. The investor is an individual saver, such as a defined contribution pension plan member, who makes fixed annual contributions over a 30-year horizon and follows a simple traditional glidepath. We use this baseline to show how such a strategy performs under different climate regimes. We build the ESG by extending the classical Wilkie cascade. Climate regimes sit at the top of the cascade. They influence inflation through a regime-specific AR(1) process, which feeds into nominal interest rates via smoothed inflation and a regime-specific Vasicek real rate. Long-term rates link to short-term rates through the yield curve. For equities, we adapt the regime-switching model from Chapter 3. For bonds, we introduce a climate risk premium that combines a structural component reflecting sovereign vulnerability and a transitory component capturing policy shocks that decay exponentially. We calibrate the model using historical data from 2005 to 2025. Using the estimated parameters, we simulate 10,000 regime paths using a Markov chain with transition probabilities estimated from history. From these, we select three representative paths to illustrate how inflation, real rates, bond yields, and equity returns behave under different climate futures. We find that the steady decline of brown-energy related assets assumed to occur in orderly transitions compounds over time, while the volatility of disorderly transitions benefits brown energy assets. This baseline establishes the vulnerability of traditional portfolio strategies and provides a benchmark for evaluating climate-smart alternatives in Chapter 5. In Chapter 5, we evaluate whether climate-smart portfolio strategies can improve retirement outcomes for individual investors based on the model developed in Chapter 4. We assume an investor makes fixed annual contributions over a 30-year horizon and can choose among alternative strategies rather than following the traditional glidepath in Chapter 4. We test two broad classes of strategies. Decarbonized glidepaths shift from traditional to green equity at different speeds before de-risking into bonds. Dynamic optimization strategies include mean-variance optimization (MVO) with fixed and time-varying risk aversion, and the minimax portfolio, which selects allocations that perform best under the worst-case scenario. We evaluate these strategies using the climate-augmented ESG developed in Chapter 4. We compare terminal wealth distributions, shortfall probabilities, and drawdown metrics across climate regimes. We then extend the analysis to decumulation by converting terminal wealth into annuity income over a 30-year retirement horizon. We find that all decarbonized strategies outperform the traditional baseline, with immediate decarbonization delivering the highest median wealth among glidepaths. We find that the decarbonized glidepaths offer the best balance of consistent performance, downside protection, and regime robustness, while the dynamic strategies offer greater upside opportunities. For investors who cannot predict the transition path, the presence of green assets in the portfolio matters more than how quickly they get there. In Chapter 6 we outline three directions for extending the thesis. First, we propose a multi-market model where different geographic regions can be in different climate regimes simultaneously. A global climate state influences regional transition probabilities, but regions can diverge. This framework allows us to quantify the value of geographic diversification when climate regimes evolve asynchronously across regions, addressing the home bias constraints faced by pension funds. Second, we extend the analysis from assets to the full balance sheet by integrating longevity risk. Climate change affects both sides of the pension fund. Heat waves and cold spells influence mortality patterns, which in turn affect liabilities. The funding ratio becomes the central state variable. The fund manager controls investment policy, reinsurance policy, and contribution policy. This framework allows us to ask whether natural hedges exist when assets and liabilities move together under climate shocks. Third, we introduce two processes that the current regime-switching models do not capture: geopolitical shocks that temporarily override climate-driven return dynamics, and climate tipping points that permanently alter the economy through self-exciting cascades. Geopolitical shocks, such as the 2026 Iran War, can make brown energy temporarily profitable, raising the question of whether maintaining some brown exposure hedges against missing short-term gains. Tipping points, such as permafrost thaw, trigger cascading physical risks that affect not just brown and green equities but also financials, real estate, and sovereign bonds — sectors that decarbonized portfolios remain exposed to.
Degree
thesis:*- Grantor dc:publisher
- University of Waterloo
- Year dc:date.issued
- 2026
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Dadzie-Dennis, Rhoda
Subjects
dc:subject × 7Rights
- Language dc:language.iso
- en
Identifiers
dc:identifier.*- Handle dc:identifier.uri
- https://hdl.handle.net/10012/23406