Felix Mantz explains how employer covenant should be incorporated into defined benefit pension scheme modelling to plan the journey to the long-term end-state
For years, actuaries have perfected techniques to model defined benefit pension scheme assets and liabilities, often using stochastic models to guide useful discussions about journey planning to a scheme’s long-term target (or end-state). Employer covenant is sometimes only given cursory consideration in this modelling, despite being the key driver for a scheme to deliver member benefits in full.
There are three core aspects of UK defined benefit scheme covenant reliance:
- Fund any existing deficit and expenses
- Support unhedged financial, demographic, legal and operational risks
- Allow a scheme to continue as a going concern to reach its end-state
Even for well-funded schemes with ‘slow and steady’ plans to insurance buy-out, understanding covenant is crucial – while further contributions may not be needed, they rely on a solvent sponsor to run on. Should the sponsor go bust, schemes must crystallise their funding level on a solvency basis. In other words, a scheme’s journey may get cut short.
What matters in journey planning
To set a scheme’s journey plan, trustees and sponsors should ensure the tools used can model assets, liabilities and covenant support. Only then can a journey plan be evaluated by its ‘relative risk’ characteristics – the scheme risks relative to the ability of the sponsor to support them, for example:
- How does the chosen end-state compare relative to long-term covenant reliability?
- How does the time to reach this end-state compare relative to covenant visibility and scheme maturity?
- How do contributions compare relative to covenant affordability?
- How does investment risk compare relative to the covenant’s risk capacity?
How might this work in practice?
Say a scheme is well funded and in surplus on its funding (or technical provisions) basis, and not particularly mature. The covenant is rated Strong, with high affordability, but its industry is in structural decline. The trustees consider three different journey plans:
|Plan A||Plan B||Plan C|
|Target returns||Gilts + 1% pa||Gilts + 2% pa||Gilts + 3% pa|
|Years to end-state||6||8||6|
Using an integrated scheme-covenant model, the trustees can evaluate these journey plans based on their relative risks, and get the following results:
Green: Risks readily supportable
Amber: Risks potentially supportable
Red: Risks unsupportable
Plan A can be ruled out, as the target ‘low reliance’ state is inappropriate relative to the uncertainty that remains over the covenant’s long-term future, given its industry. Buy-out may be achieved in the same timeframe under Plan C, but this would require taking a level of investment risk that the covenant does not support. The trustees adopt Plan B, balancing all risks to a manageable level.
Integrating covenant into scheme modelling is key to assess these relative risk characteristics. Modelling covenant support is not straightforward, though, with two key characteristics needing to be captured:
- Single-name default exposure – Schemes are reliant on a single covenant and are fully exposed to its idiosyncratic risks. This makes it difficult to model the covenant stochastically, as these models are usually based on credit ratings and historical default probabilities of a large data set, typically assuming idiosyncratic risks are ‘diversified away’. Pension schemes reliant on a single-name covenant do not have this luxury.
- Non-linear nature of covenant support – The payoff profile of the ‘covenant contract’ is typically negligible once a surplus is reached on a technical provisions basis as deficit repair contributions cease. At the same time, a scheme’s ability to run on is a step function that drops off when a sponsor becomes insolvent. These factors are difficult to reconcile in a single model, where the focus is often on insolvency.
Some market tools opt to model the scheme stochastically, then overlay an expected sponsor default based on credit ratings to estimate the proportion of benefits paid in different journey plans. While the simplicity of a single number is appealing, this does not fully capture the intricacies noted above.
Modelling the covenant deterministically (for example borrowing from the corporate finance world and using a ‘three statement model’ of profit and loss, cash flow and balance sheet) can be more informative than scheme projections. It allows the covenant’s idiosyncratic risks to be assessed and captured, and the different non-linear aspects of covenant support to be fully considered. As illustrated in the example, this is especially the case where a ‘perfect’ journey plan doesn’t exist. It is ultimately a question for trustees to understand the trade-offs between the risks involved and plan accordingly.
Making educated decisions about a scheme’s journey plan requires an understanding of scheme risks relative to covenant support. Stochastic techniques used to model schemes are not always able to fully capture covenant support, so deterministic covenant projections can be more informative.
Felix Mantz is associate director at Cardano Advisory