How corporate pension liabilities could vary by 10% or more, even on an agreed set of assumptions

Why is there a range of answers, even using a given set of assumptions? Are these differences real, or artificial?



(Blog post by Patrick Lee)

It would clearly make a difference whether a company's pension liabilities were £475m, £500m or £525m ... 

The value of an organisation's defined benefit (final salary or CARE - career average revalued) pension plan promises normally depends on many uncertainties, including:
  • how long the plan members and their partners are expected to live
  • what proportions of active members will leave service, or retire on ill health, before reaching normal retirement age
  • what the future rates of salary growth and price inflation (and hence pension increases) will be
  • assuming that a perfectly matching asset portfolio can't be found (normally such a portfolio doesn't exist), then what the future rates of reinvestment (for cashflow mismatches) will be.

Defining the liabilities uniquely and making all assumptions explicit

But let's assume an idealised world in which all of the above are known.  Even then, different actuaries (or indeed any other financial professional - the problem is related to defining the liabilities uniquely and making all assumptions explicit), are likely to arrive at different answers (sometimes markedly different) unless they all assume the same with regard to the following:
  • what proportion of pension will members convert to lump sum when they retire, and on what future terms (because those terms can and probably would change under different market conditions)
  • what proportion of active members will retire early, and on what future terms (again these terms are likely to change under different market conditions). 
But let's assume that all the experts can agree on the above. Will the experts all now agree on the value of the liabilities? 

Not necessarily: the different software tools used by actuaries and others may use different calculation methods, some of which may (due to approximations, or yet more assumptions - let's call them secondary assumptions, although another way to describe them would be assumptions that are not normally disclosed - than the ones tied down above) give different answers:

In the UK, factors including the following are likely to be relevant:
  • the treatment of any salary cap (a limit on pensionable salary)
  • the assumed timing of salary, pension and other increases (e.g. if salary increases are around 3% per year, the value of accrued liabilities for active members can be increased or reduced at a stroke by 3% depending on whether the next increase is assumed to occur immediately on in a year's time)
  • differences in the assumed value of cap and collar formulas applied to pension increases. 
  • differences in the assumed treatment of GMPs (Guaranteed Minimum Pensions - these are reducing in significance as they stopped accruing 20 years ago, but could be significant for some pension plans).

Does this matter?

In a situation where the same scheme actuary (or actuarial firm) carries out successive valuations these differences are less relevant (or another way of looking at this is to say that they are more hidden), because the same calculation method and secondary assumptions are likely to be used, so differences in liabilities values 3 years apart are likely to reflect genuine differences in the true value of the scheme (after allowing for any changes in assumptions).

Where it really does matter is when money is changing hands, or when analysts are trying to compare/value different companies allowing accurately on a like for like basis for corporate debt including pensions

In a situation where there is a change of advisors, or a business transaction is taking place (e.g. a merger or acquisition, or a buyout), then it is important to be aware of such factors in order to make sure that value is not being artificially created or destroyed simply by a change of calculation method or of secondary assumptions. It also matters for securities analysts who are trying to compare the value of pension plan liabilities across different companies, unless they can be sure that the figures they are using really have been calculated (often by different advisors) on a comparable basis.


Author: Patrick Lee, InQA Limited, principal author of InQA's Pensions Concerto valuation and cashflow analysis tool.

Comments

Popular posts from this blog

A (UK) Pensions Data Format Standard - let's make it happen!

Welcome!

Creating a corporate dashboard (using OData and Microsoft PowerBI)