Post-Crisis Interest Rate Markets: Analysing, Modelling and Stress Testing of Multiple Yield Curves

2007/2008年的金融危机引领了一个利率市场的新时代，利率市场的特征是多条(取决于期限的)收益率曲线。利率期限结构(或收益率曲线)将利率描述为期限的函数，是定价、风险管理和货币政策的重要工具。它是由不同期限的利率产品构成的，这些利率产品与相同的参考利率挂钩，例如6个月期EURIBOR(即期限为6个月)或隔夜指数掉期(OIS)利率(用于无风险贴现曲线)。尽管在2007/2008年危机之前，期限相同但期限不同的利率之间的利差可以忽略不计，但在危机最严重的时候，它们增加到了250个基点以上的水平。最值得注意的是，这些利差也在危机之后保持在显著水平。作为这种结构性变化的结果，需要以全新的方式对待危机后的利率市场。在拟议的项目中，我们面临着这一挑战，我们计划实现以下四个主要贡献：·首先，我们将确定决定整个期限结构和相应利差动态的经济因素。我们的方法将考虑不同基调的收益率曲线之间的相互依赖关系，并允许研究依赖模式如何随时间变化。这对于投资组合配置和政策决策将特别有趣。·第二，我们将开发一类新的多收益率曲线宏观金融模型，以研究宏观经济、贴现曲线和基差期限结构之间的相互作用。此外，我们将分析风险溢价如何依赖于经济因素，以及风险溢价的分解是否随时间变化。我们的结果将对理解各种货币政策规则对收益率差和整个银行间市场的影响具有重要意义。第三，我们将应用机器学习的新技术来预测未来的利率期限结构。通过检测收益率曲线参数中的某些模式，并将这些方法与宏观经济变量的包含相结合，我们预计预测精度将显著提高。通过这种方式，该项目将对收益率曲线的预测做出重大贡献。第四，在以前贡献的基础上，我们将构建适用于危机后利率市场的新压力测试框架，该框架允许研究基调依赖收益率曲线的变化以及交叉基调依赖对市场投资组合和银行股价的影响。这将对银行的内部风险管理以及监管机构产生重要影响。凭借这一丰富的议程，该提案为危机后利率市场的经验和理论分析做出了重要贡献。

The financial crisis of 2007/2008 has led a new era of interest rate markets characterized by multiple (tenor-dependent) yield curves. The term structure of interest rates (or yield curve) describes the interest rate as a function of maturity and is an important tool for pricing, risk management and monetary policy. It is constructed by bootstrapping from interest rate products of different maturities that are linked to the same reference rate as e.g. the 6-month EURIBOR (i.e. for a tenor of 6 months) or the Overnight Indexed Swap (OIS) rate (for the risk-free discount curve). While spreads between interest rates that have the same maturity but are based on different tenors were negligible before the crisis of 2007/2008, they increased to levels above 250 bps at the peak of the crisis. Most notably these spreads remained at significant levels also after the crisis. As a consequence of this structural change, a completely new treatment of post-crisis interest rate markets is required.In the proposed project we face this challenge and we plan to achieve the following four major contributions:• First, we will identify economic factors that determine the dynamics of the whole term-structure of interest rates and of the corresponding spreads. Our approach will take into account interdependencies between yield curves of different tenors and allows to study how dependence patterns vary over time. This will be particularly interesting for portfolio allocation and policy decisions.• Second, we will develop a new class of macro-finance models for multiple yield curves that allows to study interactions between the macroeconomy, the discount curve and the term structure of tenor basis spreads. Moreover, we will analyse how risk premia depend on economic factors and whether the decomposition of risk premia is varying with time. Our results will have important implications for understanding the impact of various monetary policy rules on yield spreads and the whole interbank market. • Third, we will apply new techniques from machine learning to predict future term structures of interest rates. Through detecting certain patterns in yield curve parameters and in combining these methods with the inclusion of macroeconomic variables, we expect a significant improvement of the forecasting precision. In this way, the project will provide a major contribution to the forecasting of yield curves. • Fourth, in building on the previous contributions, we will construct new stress testing frameworks suitable for post-crisis interest rate markets, which allow to study how shifts in tenor-dependent yield curves as well as cross-tenor dependencies affect market portfolios and banks’ share price. This will have important implications for banks’ internal risk management as well as for regulators. With this rich agenda the proposal provides important contributions to both the empirical and theoretical analysis of post-crisis interest rate markets.