DB Advisors: IAS 19 Discount Rate - A Moving Target? - Martin Thiesen
The year-end 2008 discount rate assumptions regarding the IAS 19 valuation of defined benefit obligations (DBOs) showed a range of accepted IAS 19-compliant discount rates. Martin Thiesen, head of investment and pension solutions at DB Advisors, reveals the asset management view on the volatility of the discount rate assumptions.
Over the last year the financial crisis and its tremendous effects on equity and credit markets, and the volatility of the International Financial Reporting Standard (IFRS) valuation of pension liabilities were the issues that frequently kept me busy.
As an actuary working for an asset management company, several portfolio managers of plan assets asked for an explanation regarding the fluctuation of their liability benchmarks. Familiar with the concept of using the swap rate’s development read-off at the liability duration point as a proxy for the liability performance, they wondered why in the third quarter of 2008 this simplification was no longer valid.
In the following months the discussion with respect to the appropriate benchmark for the asset management of plan assets has engaged clients, actuaries, investment consultants and asset managers. The following article tries to highlight some of the main issues and potential solutions.
Discount rate volatility
Regarding IFRS, the valuation of DBO is based on IAS 19 ‘Employee Benefits’. This valuation requires several actuarial assumptions such as mortality, salaries, benefits and medical costs (i.e. inflation) and, last but not least, the discount rate.
The discount rate is commonly considered the actuarial assumption, which causes most of the volatility in the valuation of the present value of future pension payments. IAS 19.78 defines the assumption as such: ‘The rate to discount post-employment benefit obligations (both funded and unfunded) shall be determined by reference to market yields at the end of the reporting period on high-quality corporate bonds. In countries where there is no deep market in such bonds, the market yields (at the end of the reporting period) on government bonds shall be used. The currency and term of the corporate bonds or government bonds shall be consistent with the currency and estimated term of the post-employment benefit obligations.’
Due to the financial crisis, one phrase of paragraph 19.78 became more important: ‘…reference to market yields at the end of the reporting period on high-quality corporate bonds.’ Once interpreted as market yields of AA Corporate bonds, the discount rate assumption is for good or for evil bound to the development of duration-adjusted AA Corporate bond yields. This is because the interpretation of most of the auditors states that each bond or at least the average yield of the underlying bond universe (included in the discount rate’s determination) has to reflect the market yield of an AA-rated corporate bond. This interpretation made a significant impact on the valuation of the DBO and, afterwards, the volatility of the balance sheet due to the massive AA corporate spread widening at the end of 2008 and the shrinking of the AA Corporate universe at the beginning of 2009.
Effects on equity and credit losses
Although some actuarial companies discussed and applied the smoothing effect in order to mitigate the consequences of this massive spread widening alongside a significant decrease of the pension liabilities, it led to a partial mitigation of the deterioration on equity and credit markets. Without this positive outcome of the credit crunch the global pension funding status would have been even worse than the outcome predicted by the pension indicator report of Watson Wyatt (see diagram). Pension schemes that implemented an interest rate hedge via a swap overlay (extending the asset duration towards the liability duration – Liability Driven Investment strategy) before the financial crisis benefit twice from this spread widening.
On the one hand the AA Corporate yield’s increase decreased the value of their liabilities. On the other hand the plan assets earned money through decreasing swap rates. The first reduced the denominator and the second increased the numerator of the funding status ratio, both resulting in an overall improvement of pension scheme solvability. Though the swap overlay was implemented as an interest rate sensitivity hedge it contributed positively to the mitigation of the equities and credit spread losses in the plan assets.
High assumption risks
Recognising the insufficient current replication of the implemented liability benchmarks, a better match of the discount rate methodology was requested by several pension scheme sponsors. A survey among different discount rate assumptions of pension schemes throughout Europe resulted in an obvious wide range of approximately 100 bps with the highest discount rates even above 7.0%.
As mentioned before, the discount rate assumptions were significantly biased by the AA Corporate spread widening. The lower end of the range represents the countries or actuaries that considered the AA Corporate market not deep enough and/or distorted by the credit crunch and therefore applied government bond yields, for instance, as an appropriate assumption. Their objective was not to reward pension schemes in trouble by underestimating their pension liabilities due to, in their opinion, dislocations in the markets. To avoid these ‘unacceptable’ high discount rates these countries or companies eliminated outlier bond yields based on a predefined algorithm.
The upper end of the range more or less reflecting the development of an AA Corporate 10+ index (maturities of ten years or longer) underlined the IAS 19.78 phrase ‘… reference to market yields.’ This phrase was their justification for applying unadjusted market data to value the DBOs.
In my opinion the truth lies somewhere in between. From an asset manager’s point of view, smoothing effects aren’t acceptable because it’s impossible to replicate their impacts on a benchmark. Excluded in the previous month a bond yield could be included in the current calculation and excluded in the next month once more, which is an unpredictable movement of the index.
In order to minimise the influence of outlying bond yield, upfront minimum issue amounts should be determined. Liquidity ensures to a certain extent the stability in pricing and minimises the occurrence of extreme bond yields.
On the other hand, applying an unadjusted AA Corporate 10+ yield is too simplified an approach. In my opinion it would be more appropriate to define a liquid AA Corporate universe and derive a yield curve out of this universe. Due to liquidity issues one is definitely limited to a maturity of 25 years within this curve approach. But in most cases the projected liability cashflows have to be discounted up to a maturity of 50 years or even longer. To address this issue the plan sponsor could apply a curve extrapolation via an appropriate market curve (for instance: adjusted government bond yields or swap rates). The ultimate objective is to define the discount rate assumption in a way that can be replicated (with minimised tracking error) using market instruments (i.e. plain vanilla interest rate swaps).
But even setting up such a sophisticated liability benchmark doesn’t ensure the full replication of the liabilities on the assets’ side. So why are there remaining gaps?
Firstly, even the high liquid AA Corporate bond issues, especially for the longer maturities, are unavailable. Therefore, the asset manager has to find an appropriate proxy bond reflecting the highest correlation with the benchmark bond or build up the AA Corporate spread exposure synthetically.
Secondly, earning an AA Corporate bond yield is limited to a maturity of up to 20 years. Above that maturity point the asset manager falls back to government bonds and for maturities higher than 30 years to interest rate swaps. Earning the gap would require additional risk capital.
Thirdly, the impact of defaults on both sides of the balance sheet is different. Even though a direct profit and loss effect is recognised on the asset side the effect on the liability side is quite complex. A near-to-default bond could cause a higher discount rate, which leads to a lower valuation of the liabilities and potentially a gain in the pension scheme. After the default, the bond is excluded from the bond universe, which, ceteris paribus, leads to a lower discount rate and an accounting loss in the overall funding status linked with the direct loss on the asset side.
Earlier on I mentioned the intrinsic risks of high discount rates, especially biased ones at the end of 2008 by the AA Corporate spread widening. The most obvious and important one is the AA Corporate spread tightening, which the pension schemes partially faced at the beginning of 2009. Without reacting on this spread tightening the effect could be even worse than the direct impacts of the financial crisis.
A joint effort of clients, investment consultants, actuaries and asset managers is necessary to move both the asset side via asset allocation decisions and hedging strategies towards the best possible replicated liability side.