EIOPA consults on UFR derivation
di Silvia Dell’Acqua

Giu 16 2016
EIOPA consults on UFR derivation  di Silvia Dell’Acqua

On the 6th April 2016 EIOPA published a consultation document to seek for a feedback on the methodology to derive the Ultimate Forward Rate (UFR). The Authority intends to decide on the outcome of the review by September 2016.

Under Solvency II (SII) the insurance and reinsurance companies have to evaluate their assets and liabilities following some harmonized principles. Among those, there is the discounting of the liabilities cash flows through a risk free interest rates yield curve, which EIOPA has been publishing on a monthly basis since February 2015. These risk free yield curves are derived from financial instruments (interest rates swaps and, if not available, government bonds) traded in deep, liquid and transparent markets till the Last Liquid Point (LLP) and then extrapolated to longer maturities. The extrapolation is needed as the insurance liabilities may have durations of several decades, while the LLP are usually shorter – e.g. 20y for EUR. The extrapolation is performed such that the forward rates converge towards the UFR within the convergence period – e.g. 40y for EUR. This means that at the end of the convergence period – 20y+40y for EUR – the forward rates are equal to the UFR, but different from the published risk free interest rates, being those expressed as spot rates. The UFR has to be set appropriately to ensure the protection of policyholders: the liabilities calculated by the insurance companies have to be sufficiently high to cover their obligations.

The methodology to derive the UFR, which was initially described in a background document to the QUI5 study back in 2010, has to be aligned to the Delegated Regulation established in 2014: the derivation approach that will be applied on an ongoing basis has to be clarified, bearing in mind that the UFR should change with long-term expectations. The UFR currently in use is the one derived in 2010 – e.g. 4.20% for EUR – and starts to seem too high when compared to the market rates observable in the low interest rates environment of some economies. On one hand, it needs to be taken into account that the UFR is a long-term forward rate, but on the other hand it needs to be considered that a delay in the change of the UFR may result in more drastic impacts in case the long term expectation will move further away from the current UFR level.

What is clear is that any change in the UFR is expected to have a material impact for the insurance companies with significant long-term liabilities (beyond the LLP). To understand this better, it is worth recalling the results recorded in the 2014 stress test low yield module:

– the interest rate risk was material due to the common duration mismatches between asset and liabilities: the more positive the duration mismatch (liabilities > asset), the more vulnerable insurance companies are to negative interest rate shocks (and low yield curve in general)

– the average guaranteed rates on life insurance business (excluding unit-linked and index linked business), taking into account options and guarantees and surrenders, were between 2% and 4%, although being progressively decreased in the previous 5 year

– considering the application of LTG measures the percentage of the sample that did not meet the SCR requirements increased from 16% (base situation) up to 20/24% (low yield module).

The proposal made by EIOPA for the UFR methodology and its implementation is as follow:

– the UFR will be calculate every year, announced by the end of March and used to calculate the risk free yield curve three months later (the first UFR will be calculated by the end of March 2017 and applied to the risk free term structure for end-June 2017). A more frequent update would be not in line with the objective of stability and the three months period is intended to warn the insurance companies to prepare for the possible change

– the UFR cannot change by more than 20 bps (up or down) per year: a limitation is applied in case. As an example, given the current level of UFR=4.20% for EUR, if the long term expectation was, let’s say, 3.70%, in the next four years the UFR would be 4.00% in 2017, 3.80% in 2018 and 3.70% in 2019.

– the UFR before the limitation of the annual change is the sum of an expected real rate and an expected inflation rate; the former is the same for each currency, while the latter is currency specific. Making use of historical real rates and inflation targets, this approach have similarities with the methods used by both Moody’s Analytics (ex Barrie & Hibbert) and IAIS

– the expected real rate R is the weighted geometric average of the n annual real rates ri from 1960 to the year before the update of the UFR (wider window than the one currently adopted)

  • the annual real rates ri are derived as the simple arithmetic average of the annual real rates of Belgium, Germany, France, Italy, the Netherlands, the United Kingdom and the United States. The current approach, based on the average of 12 economies, was changed not to give too strong rates to certain economies (like Australia or Canada) that are not represented in the asset portfolios of European insurers
  • for each currency the annual real rate is calculated as  . The short term nominal rates are taken from the macroeconomic database of the European Commission (AMECO); the inflation rates are taken from the Main Economic Indicator database of the OECD. These two sources of data have been changed from the current approach to rely on databases provided freely and maintained transparently by public institutions
  • the expected real rates is rounded to full five bps (upward or downward) in the direction of the value registered the previous year (e.g. t-1=420; t=434 → 430; t=411 → 415)
  • differently from the current approach that relies on an arithmetic average, the proposed approach makes use of a geometrical weighted average with a fixed parameters of 0.99 to give less influence to oldest data
  • following the article 47 of the Delegated Regulation, EIOPA intends to exclude the term premium (and the convexity effect embedded in the term premium) that reflect the additional risk of holding long-term investments by using short-term interest rates from the AMECO database

– the expected inflation rate is based on the inflation target if announced by the central bank

  • 4 buckets are considered
    • target  < 1% → expected inflation=1%
    • target ∈ (1%, 3%] → expected inflation=2%
    • target ∈ (3%, 4%] → expected inflation=3%
    • target  > 4%→ expected inflation=4%
  • in case the central bank does not target a specific figure but indicates a corridor, the midpoint of the corridor is considered
  • in case the central bank does not announce a target, the expected inflation is set equal to 2%. The value can be change whether the long term expectation is at least 1% higher/lower than 2% (downward rounded to full percentage points). The past inflation is assessed against the average of 10y data and the projection derived on a basis of an ARMA model.

The impact of a change in the UFR on the present value of guaranteed benefits is not material for most life insurance portfolios, but can be significant for those with long-term guarantees

– considering cash flows data from 2014 stress test for only life insurance with profit participation and fixed guarantees and a decrease of the UFR from 4.20% to 3.50%

– the present value of fixed guarantees would increase by 0.79% and the eligible Own Funds would decrease by 6.7%

– for three quarters of the companies these figures would be +0.68% fixed guarantees and -5% Own Funds, but for few insurers they would be +2.4% and -50%.

To go deeper into these results one could look at

– the differences in the term structures, which differs significantly by currency

  • the impacts depends on the LLP and the speed of convergence. The highest impact can be observed for the Swedish krone (LPP=10y, convergence period=10y), the lowest impact for the pound sterling (LPP=50y, convergence period=40y)
  • the change of the extrapolated risk free interest rates depends approximately linearly on the change of the UFR. For example, for EUR T=30y the impact of increasing/decreasing the UFR by 10, 20 or 30 bps corresponds to an increase/decrease of 2.1, 4.2, 6.3 bps of the spot rates

– the impact on the present value of simple immediate/deferred annuity. The impact is usually small for immediate annuities as the large part of the cash flows correspond to maturities before the LLP, but can be relevant for deferred annuities being the payments deferred in the future. E.g. considering the Dutch life tables and a person aged 65, a decrease of 10bps in the UFR corresponds to an increase of the Technical Provisions up to 0.3% for an immediate annuity and up to 1.8% for a deferred annuity.


I commenti per questo post sono chiusi