Quantitative tools for energy risk management

在能源生产的环境成本日益得到承认和考虑的世界上，鼓励对清洁和有效的生产技术进行投资变得越来越重要。能源相关市场——电力市场、排放市场、天然气市场等——的作用至关重要。我的研究项目涉及数学和计算模型的发展，这些模型具有良好的经济基础，旨在为研究这些市场提供工具。这些工具可以用来帮助理解和管理未来市场行为固有的不确定性所带来的风险。在建模和计算方面有重大的挑战需要面对。许多能源和大宗商品市场相对较新，与更成熟的金融市场有着根本的不同，主要是因为它们关注的是实物的生产和消费。模型必须考虑到许多相互关联的不确定性来源，这可能导致高维性。开发能够处理这种高维的模型是本建议的关键主题。决定使用什么模型是好是坏，需要了解模型的用途，模型对可用数据的适应程度，以及从模型中提取有用的定量信息的容易程度。与最后一个问题相关的是我研究的另一个方面：开发高效可靠的计算算法，用于处理能源市场模型，包括蒙特卡罗、有限差分、傅立叶和自适应小波方法。最后，能够评估在这个过程中出现的数字上应该赋予多少权重依赖于评估模型风险的工具，这是建议的第三个组成部分。

In a world where the environmental costs of energy production are increasingly being recognised and accounted for, it is becoming more and more important to encourage investment in clean and efficient production technologies. The role of energy-related markets - power markets, emissions markets, gas markets and others - is key. My research programme is concerned with the development of mathematical and computational models - with a sound economic basis - that seek to provide tools for studying these markets. Such tools can be used to help understand and manage the risks arising from the uncertainty inherent in future market behaviour.There are significant modelling and computational challenges to be faced. Many energy and commodity markets are relatively new and differ in fundamental ways from more established financial markets, mostly because they are concerned with the production and consumption of real physical things. Models have to take into account many interrelated sources of uncertainty, and this can lead to high dimensionality. Developing models that can deal with this high dimensionality is the key theme of this proposal. Deciding what is a good or a bad model to use involves understanding the use to which the model will be put, how well-suited the model is to the available data, and how easy it will be to extract useful quantitative information from the model. Related to this last question is another aspect of my research: the development of efficient and reliable computational algorithms for working with energy market models, including Monte Carlo, finite difference, Fourier and adaptive wavelet methods. Finally, being able to assess how much weight one should put on the numbers that emerge from this process relies on tools for assessing model risk, a third component of the proposal.