Modelling of non-bank financial institutions

在2007-2008年金融危机之后，发展用于分析金融稳定性和金融体系系统性风险的数学模型一直是许多研究工作的重点。这些研究大多集中在对银行体系稳定性和银行监管的分析上。然而，银行系统只是金融系统的一个子集，金融系统还包括许多类型的非银行金融机构，最著名的资产管理公司，如共同基金、养老基金、货币市场基金、对冲基金以及保险公司。这些非银行机构持有相当大一部分金融资产，并占交易的很大比例，因此，准确理解它们的角色以及它们与银行体系的相互作用，对于金融体系的现实建模至关重要。该项目旨在通过利用监管文件和市场数据来详细描述金融体系的生态，从而提高我们对非银行金融机构作用的理解。第一步是对监管数据和市场数据进行统计分析，其中我们将利用分类方法和现代数据分析方法，以便按照类似的策略将金融市场参与划分为稳定且可解释的分组，并在微观和宏观层面上表征金融环境。我们的目标是描述金融体系中非银行机构的主要类别，并确定它们的主要特征和相互作用。我们利用机器学习技术，通过利用我们可以从持股和市场数据中提取的所有时空信息，来表征非银行机构的行为。第二步将是在数学框架中对这些非银行机构的行为进行建模，这有助于分析研究或模拟，这可能对压力测试和风险分析有用，并有助于检查系统性风险和金融稳定的后果。该项目属于EPSRC的以下研究领域：人工智能技术、统计学和应用概率。富达投资公司是该项目的工业合作伙伴。

The development of mathematical models for the analysis of financial stability and systemic risk in the financial system has been the focus of a lot of research efforts in the aftermath of the 2007-2008 financial crisis. Much of this research has been focused on the analysis of the stability of the banking system and the regulation of banks. However, the banking system is only a subset of the financial system, which also includes many types of Onon-bankO financial institutions, most notable asset managers such as mutual funds, pension funds, money market funds, hedge funds, as well as insurance companies. These Onon-banksO hold a sizable fraction of financial assets and account for a large proportion of transactions, so an accurate understanding of their role and their interactions with the banking system is crucial for a realistic modelling of the financial system.This project aims at improving our understanding of the role of non-bank financial institutions by exploiting regulatory filings and market data to develop a detailed description of the ecology of the financial system. The first step is a statistical analysis of regulatory data and market data in which we will make use of classification methods and modern methods of data analysis in order to arrive at a stable and interpretable partition of financial market participations into groups following similar strategies and to characterize the financial environment at micro and macro levels. Our aim is to characterize the main categories of non-banks in the financial system and identify their main features and interactions.We take advantage of machine learning techniques to characterize the behavior of non-banks by exploiting all the temporal and spatial information that we can extract from holdings and market data.A second step will be to model the behavior of these non-banks in a mathematical framework which lends itself to analytical study or simulations which may be useful for stress testing and risk analysis, and for examining the consequences for systemic risk and financial stability. This project falls within the following EPSRC research areas: artificial intelligence technologies, statistics and applied probability.Fidelity Investments Inc. is the industrial partner for this project.