In this paper, we discuss the developments seen in the interest rate markets over 2022 and how this has affected the level of liability discount rates for insurers.
Background
Since the 2007-2008 global financial crisis (GFC), we have seen an almost continuous decline in interest rates, causing financial institutions to focus on the low interest environment and its implications for areas such as valuation, capital management and risk management. In 2022 this all changed. Rising inflation forced central banks around the world to increase their rates, materially impacting insurers’ balance sheets and risk profiles.
This briefing note is the first of a series of three articles where we investigate potential implications for insurance companies of last year’s increasing interest rate environment.
In this first paper we show the key developments of interest rates in 2022, try to put them into historical perspective and show how recent developments impact the level of liability discount rates, considering different extrapolation techniques.
What happened to the interest rates?
In Figure 1, the development of six-month (6M) European swap rates is shown for various maturities, from full-year (FY) 2003 until now. Over the last year we have seen substantial increases in rates (phase 5), similar in amount to the years preceding the GFC (phase 2). However, the interest rate rises this year were at a significantly more rapid pace. Similar to the GFC, there was a flattening of the curve. But this time it extended into a significant inversion, amplifying the inversion of the curve that had already started halfway through 2019 (phase 4). We also see a clustering of volatility that is very similar to the GFC and its aftermath in 2009. The fundamental difference now is that the increased volatility is mainly driven by rate increases instead of rate decreases.
Figure 1: Development of 6M Euribor swap rates for various maturities
Source: Refinitiv.
Figure 2: Liability discount curves at FY2021 and FY2022 including credit risk adjustment and excluding volatility adjustment
Source: Refinitiv, tooling for interpolation and extrapolation by Milliman.
Impacts on liability discount rates
Figure 2 shows a comparison of various liability discount curves between FY2021 and FY2022. Under Solvency II (SII), the liability discount curve is currently determined using market rates that are extrapolated to an ultimate forward rate (UFR) based on the Smith-Wilson (SW) extrapolation method. This extrapolation starts at the last liquid point (LLP) of year 20. The substantial increase in the SII discount curve has led to a significant decrease of the value of technical provisions for insurers over 2022.
The graph also shows a curve (including credit risk adjustment) which is fully based on (an interpolation of) market observable rates until year 50. The gap between this curve and the SII discount curve has clearly decreased over 2022, leading to a decrease in the so-called “UFR drag” under SII. It should be noted that the gap between the market curve and SII discount curve would have been even lower without the inversion movement over 2022. In Figure 3, this is illustrated by plotting the market curve without the inversion movement over 2022 for maturities after 20 years.1
In the second paper in this series we will show that the flattening and further inversion of the curve in conjunction with the extrapolation of the SII discount curve causes some interesting dynamics with respect to the development of the UFR drag. These dynamics can have consequences on the effectiveness of an insurers’ interest rate hedging policy.
Lastly, the rate increases also have consequences for the curves based on the alternative extrapolation method (AM) found in the opinion of the European Insurance and Occupational Pensions Authority (EIOPA) on the 2020 review of Solvency II,2 which takes into account information from longer-term interest rates. In previous briefing notes, we have described the technicalities of the alternative extrapolation method and the impact of lower interest rates on the curve dynamics.3 In these notes we have, amongst other things, discussed the relevance of the speed of convergence parameter on the level of the yield curve. In Figure 2, the curves are shown for a convergence parameter of 10% and 2%. For both convergence parameters it can be seen that the gap between the alternative extrapolation method and current SW extrapolation method has decreased over 2022. Moreover, the sensitivity of the AM curve towards the convergence parameter has visibly decreased. This will likely lead to a lower transition impact for insurers. This reduced impact and sensitivity could potentially affect the outcomes of the ongoing legislative discussions on the Solvency II reform.4
Figure 3: Liability discount curves at FY2022 with and without inversion movement over 2022 for maturities after 20 years
Implications for liability valuation and UFR drag
In Figure 4, the impacts of the changes in the liability discount curves are illustrated. For this example, we have used a proxy cash flow, internally constructed, representing an average Dutch life insurance company. The cash flow is calibrated to represent a life insurance best estimate liability with duration 16 when applying the Solvency II curve excluding the volatility adjustment (VA).
It can be seen that the liability value has substantially decreased over 2022, and that the valuation gap between the various forms of extrapolation has decreased, in line with the observations from Figure 2 above. This decreased gap is driven by the market rate increasing beyond year 20. The substantial decrease in UFR drag (almost halved in the example from Figure 4) will have a significant beneficial impact on capital generation for insurers with long-dated liabilities. In this example, the overall impact of 4.9% would have decreased even further to 1.0%, if the inversion of interest rate curve beyond year 20 had not taken place during 2022.
Moreover, the transition towards a new extrapolation method in the SII reform will likely have considerably less impact than before. If at previously low rates the impact of a convergence parameter of 10% were considered acceptable by most insurers, then at current rate levels a convergence parameter of 2% might now even be acceptable. Consistent with this point, a convergence parameter of 2% has already been used in the liability discount curve for Dutch pension funds, prior to this curve being replaced by a market curve in 2022. For higher convergence parameters, the need for a transitional mechanism5 as proposed by the European Commission might also disappear.
Figure 4: Proxy insurer liability valuation for different liability discount curves (SW LLP 20 at FY2021 = 100), as of FY2021 and FY2022
| SW LLP 20 | AM 10% SoC | AM 2% SoC | Market | |
|---|---|---|---|---|
| FY2021 | 100,0 | 102,2 | 106,4 | 109,7 |
| Impact (abs.) | 2,2 | 6,4 | 9,7 | |
| Impact (%) | 2,2% | 6,4% | 9,7% | |
| FY2022 | 72,9 | 73,6 | 75,0 | 76,4 |
| Impact (abs.) | 0,7 | 2,1 | 3,6 | |
| Impact (%) | 1,0% | 2,9% | 4,9% | |
| Impact 2022 (abs.) | -27,1 | -28,6 | -31,4 | -33,2 |
| Impact 2022 (%) | -27,1% | -28,0% | -29,5% | -30,3% |
Conclusions
The increases in interest rates witnessed during 2022 can be considered a historic event. In some respects, the curve movements are even more extreme than those experienced during the GFC. The increases in interest rates have led to significant decreases in the value of insurance liabilities under SII. The expected interest rate losses in the long term from the UFR drag have also decreased, but this reduction in expected future losses has been limited to some extent by the inversion of the curve that took place for maturities beyond 20 years. Under SII, there will also likely be a lower transition impact of the new extrapolation method from the forthcoming SII reform.
How these movements impact the solvency coverage ratio of an insurer will be highly dependent on the insurer’s hedging policy. In our second paper we will zoom in on the effects that increased interest rates have had on the UFR drag. We will also discuss the potential implications the current economic environment can have on interest rate sensitivities and hedging strategies.
Lastly, the low (and sometimes even negative) rates that were experienced prior to 2022 are (based upon the current economic outlook) expected to be absent in the foreseeable future. The question is how this new reality—and the speed at which the transition towards this new reality took place—will affect insurers’ understanding of interest rate risk in the broader sense. In our third paper, we will also explore the potential impacts of this new environment on economic scenarios, the appropriateness of the Solvency Capital Requirement (SCR) and stress and sensitivity testing.
1 Without this movement over 2022 all market observable rates beyond year 20 would have increased by the same amount as the 20-year rate. Such a movement would have led to a market curve per FY2022 with the same shape beyond year 20 as the market curve per FY2021.
2 EIOPA’s opinion on the 2020 review of Solvency II is available at https://www.eiopa.europa.eu/document-library/opinion/opinion-2020-review-of-solvency-ii_en.
3 See https://www.milliman.com/-/media/milliman/pdfs/2020-articles/articles/11-5-20-the_impact_of_alternative_extrapolation_methods-v1.ashx.
4 For a summary of views known so far of parties involved, please see https://www.milliman.com/-/media/milliman/pdfs/2022-articles/8-17-22_interim-score-of-the-solvency-ii-reforms.ashx.
5 The transitional mechanism has been discussed in a previous briefing note, which can be found at https://www.milliman.com/en/insight/the-transitional-mechanism-for-the-alternative-extrapolation.