← 論文一覧に戻る

金融市場に関する論考:水産養殖金融、エネルギー転移、無形ガバナンス

Essays on financial markets: aquaculture finance, energy transition, and intangible governance (原題)

Longguang Sun

University of Glasgowジャーナル2026-11-17#気候金融経営インパクト: 資金調達対象セクター: power
DOI: 10.5525/gla.thesis.86362
原典: https://doi.org/10.5525/gla.thesis.86362

🤖 gxceed AI 要約

日本語

本博士論文は水産養殖の最適収穫、バイオマス先物、原子力投資のリアルオプション、組織資本と株価暴落リスクの4論考を統合する。深層強化学習や物理情報を組み込んだリアルオプション、機械学習による非平滑な行使境界の近似など計算手法を金融実務に応用する。気候政策のティッピングポイントが炭素価格関数に構造的断絶をもたらす点を「恐怖プレミアム」として評価する。

English

This dissertation integrates four essays spanning aquaculture harvesting (Deep Q-Learning real options), biomass futures hedging under climate-driven biological risk, a physics-informed real options model for nuclear investment with a climate tipping point in carbon pricing, and organizational capital as an intangible governance mechanism reducing crash risk. It demonstrates how advanced computational methods (DRL, tree ensembles) can price complex climate and resource risks where traditional linear frameworks fail.

Unofficial AI-generated summary based on the public title and abstract. Not an official translation.

📝 gxceed 編集解説 — Why this matters

日本のGX文脈において

炭素価格の構造的断絶や長期脱炭素投資の評価手法は、日本のGX推進・トランジション・ファイナンスや電力分野の長期投資判断に示唆を与える。SSBJ開示の文脈では無形資産ガバナンスとリスク評価の接続が参考になる。

In the global GX context

The paper contributes to transition-finance and climate-risk-pricing scholarship relevant to ISSB/TCFD disclosure of climate-related financial risk, showing how tipping-point uncertainty and non-linear carbon pricing should enter long-horizon valuation and how intangible governance links to risk disclosure.

👥 読者別の含意

🔬研究者:深層強化学習・ツリー系手法をリアルオプションや気候リスク評価に応用する方法論的参照点となる。

🏢実務担当者:長期脱炭素投資や気候関連リスクの定量評価、無形ガバナンスとリスク管理の設計に活用できる。

🏛政策担当者:炭素価格の不確実性とティッピングポイントが投資判断に与える影響を政策設計に反映する示唆。

📄 Abstract(原文)

This doctoral dissertation investigates the pricing, management, and mitigation of risk in modern financial markets, traversing the domains of natural resource economics, energy transition finance, and corporate governance. As the global economy enters an era of polycrisis defined by accelerating climate change, structural policy shifts, and complex agency conflicts, traditional valuation frameworks, which are often predicated on assumptions of linearity, stationarity, and limited dimensions, are increasingly insufficient. This research proposes and validates a unified computational and empirical framework to address these challenges, comprising four independent but thematically integrated essays. The first essay addresses the curse of dimensionality in natural resource valuation by applying Deep Q-Learning to the optimal harvesting problem in aquaculture. Treating the harvesting decision as a real option, the study evaluates Deep Reinforcement Learning as a scalable, nonparametric alternative to regression-based methods such as Least-Squares Monte Carlo. In the baseline setting, the DQN policy produces harvesting times that are broadly comparable with the LSMC benchmark, although the estimated option values remain lower. The main advantage of the framework is therefore its capacity to accommodate richer state spaces without requiring a rapidly expanding set of manually specified basis functions. The second essay bridges biological modeling and financial engineering to value a novel hedging instrument: the biomass futures contract. By integrating a stochastic biological growth model, which accounts for thermal stress, mortality shocks, and growth rates contingent on temperature, with a Schwartz dual factor pricing model, the study quantifies the hedging demand against production risks driven by climate change. The findings reveal that the economic value of biomass futures is highly dependent on the underlying state. These contracts offer substantial downside protection and fundamentally alter optimal harvesting strategies, primarily under scenarios of extreme environmental volatility. The third essay develops a physics-informed real options framework for evaluating nuclear power investment over extended horizons under climate-policy uncertainty. A key innovation is the incorporation of a climate tipping point that introduces a structural break into the natural carbon-decay mechanism and, in turn, the carbon-pricing function. Comparing alternative continuation-value approximators shows that tree-based methods, particularly Random Forest and XGBoost, are well suited to representing non-smooth exercise boundaries and can produce economically meaningful differences at the margin relative to global polynomial smoothing. The model interprets the additional value associated with the anticipated crisis regime as a "fear premium", reflecting ambiguity about a potentially irreversible environmental regime shift before the physical threshold is actually crossed. The fourth essay extends the analysis to corporate finance, examining the role of Organizational Capital as an intangible governance mechanism. Using a comprehensive panel of U.S. firms, the study documents a significant negative association between Organizational Capital and future stock price crash risk. This relation is conditional on the surrounding governance environment and is stronger in several settings characterized by effective monitoring, managerial expertise, and stronger compensation incentives. Together, these essays advance the understanding of how financial agents price and manage complex risks, demonstrating the necessity of integrating advanced computational techniques with contextual physical and organizational insights.

🔗 Provenance — このレコードを発見したソース

🔔 こうした論文の新着を逃したくない方は キーワードアラート に登録(無料・3キーワードまで)。

gxceed は公開メタデータに基づく研究支援データセットです。要約・翻訳・解説は AI 支援で生成されています。 最終的な解釈・検証は利用者が原典資料に基づいて行うことを前提とします。