Ajla Nurkanovic

• This thesis explores the integration of sustainability into continuous-time portfolio optimization from both theoretical and practical perspectives. Motivated by increasing regulatory and societal pressure to align investments with environmental goals, we examine how sustainability can be systematically embedded in portfolio decision-making through utility-based optimization frameworks. We begin by formulating a foundational model the so-called sustainable portfolio optimization, introducing a novel approach in which investors maximize expected utility subject to a sustainability constraint based on scaled ESG scores. Sustainability constraint means the sustainability of the portfolio rating has to be at least as high as the predefined sustainability demand. Building on this, we develop an equilibrium perspective in which sustainability goals are met not through explicit constraints but by modifying market dynamics, mimicking real-world taxation/subsidy mechanisms that steer investment flows toward sustainable assets. Thus, we name this sustainable taxation. In response to challenges arising from the use of ESG scores, we propose a refined scoring framework that better distinguishes between environmentally beneficial (``green'') and harmful investments (``brown''). To align sustainability scoring with real environmental goals, we introduce a penalty on brown assets, i.e., a no-short-selling constraint on brown assets. We develop a dedicated algorithm capable of efficiently handling the resulting constrained optimization problem that we call green portfolio optimization. We then apply our framework to the insurance sector where we develop the so-called rebuilding of actuarial reserve funds (ARFs) to increase the sustainability of the ARF to be able to meet the increased sustainability demand. A simulation-based method is proposed to generate future declaration paths, which can then be used for application in the proposed rebuilding framework. Finally, we present a dynamic optimization setting in which investors' demand directly influences asset returns via a bounded, mean-reverting process. This framework captures a feedback loop between market interest and asset performance, and allows us to reinterpret demand as a sustainability requirement. Thereby, it provides further insight into how sustainable preferences shape optimal investment behavior. In total, the thesis contributes a set of coherent models and computational tools for integrating sustainability into long-term portfolio decisions, offering novel insights for institutional investors, insurers, and regulators.