The accuracy of interest rate shocks in different interest rate environments

By Marta Swiderska & Amba Zeggen, February 2024

Executive Summary

In the past two years, the key ECB rates have been rising after a decade of central bank policy to stimulate the economy by decreasing interest rates. It influenced an increase in interest rate levels, which can be observed for all maturities up to 30 years. However, as of 2024, it is expected that the ECB will change its interest rate policy since inflation in the euro area has decreased significantly. This makes the market more volatile than ever before. The Solvency II (hereafter SII) regulation, which is the regulation for insurers in the European Union, was introduced in 2016 in a low-interest-rate environment. This paper focuses on assessing the adequacy of the prescribed SII interest rate shocks. It is important since it influences the amount of capital reserves held by the insurance company. SII-prescribed shock includes a percentage point change in interest rates, which is supposed to describe a 1-in-200-year event. As a part of EIOPA’s second set of advice to the European Commission on specific items in the SII Delegated Regulations, EIOPA suggested using an alternative method to derive the SII-prescribed interest shocks for 1-in-200-year events, called the shifted approach. In this paper, the current SII-prescribed shocks and the shifted approach are compared to the predicted spot rate range using Monte Carlo simulation with three different underlying methods: principal component analysis (PCA), Hull-White, and Nelson-Siegel. Those ranges correspond to 99.5% confidence intervals, which are identical to 1-in-200-year events. Those theoretical methods were chosen in order to capture both forward- and backward-looking perspectives on interest rates. Additionally, models have been chosen to maintain a balance between model sophistication and simplicity of calibration. The SII-prescribed shocks and shifted approach adequacies are evaluated based on the following criteria: performance, difficulty to use, and stability over time in low and high interest rate environments.

Key Takeaways:

  • Takeaway 1: The predicted spot rate ranges based on the Hull-White model are very narrow, especially in low interest rate environments, resulting in the actual data being more volatile than predicted. Therefore, the confidence intervals underestimate possible movements of the spot rate curve. The simulated ranges with the Nelson-Siegel model perform considerably better than those derived with the Hull-White model because the Nelson-Siegel ranges are wider. However, Nelson-Siegel sometimes overestimates possible movements of the spot rate curve, and the width of the range can be influenced by the chosen in-sample period. The PCA method, which is similar to the Nelson-Siegel, has narrower confidence intervals, resulting in less overestimation of the reserves. Hence, from the compared methods, the best predictions are made with the PCA.
  • Takeaway 2: The SII-prescribed interest rate shocks perform poorly in the low interest rate environment. The underlying problem is related to the fact that the method neglects the possibility of the downward movement of the spot rate curve in a negative interest rate environment. Therefore, adjusting the prescribed values of the shock will not improve this method.
  • Takeaway 3: The shifted approach method presented in the EIOPA’s advice to the European Commission results in a significant improvement in performance and is of comparable complexity to the current approach. However, the parameters values are not always adequate because they overestimate ranges for short-term rates and underestimate ranges for long-term rates and therefore should be reassessed.