Charles Rodgers, a London-based pensions consultant at Watson Wyatt, has a problem: the long-awaited International Accounting Standards Board (IASB) due process discussion document on pensions accounting. His issue? The board’s new contribution-based promise definition and its proposal that the world’s bean counters should account for pension promises according to how they accumulate.

“There is no denying that a typical contribution-based promise will give rise to different benefits in the accumulation phase,” he says. “The issue is that once the benefit has crystallised and the pension is in payment, the proposed valuation method can give rise to a different valuation for the same promise. This seems difficult to square.”

And his solution? “To remove the inconsistency, you would have to move - back - to a single measurement approach. Going forward, this would suggest the wider application of fair value and having to value final-salary pensioners using the same measure.” It is an implication that the paper stops short of spelling out.

The document in question sets out the IASB’s preliminary views for the interim future of pensions accounting. The project splits into two phases - the short-term and a longer-term broad reappraisal. The discussion paper is the first milestone in the short-term phase, which should see an interim accounting standard by 2013. Then, in the second phase of the project, the IASB will work alongside the US Financial Accounting Standards Board to develop a fully converged, single, employee-benefits standard.

The key changes proposed by the discussion paper over existing IAS19 accounting include:

 Removing the optional corridor method for deferring recognition of actuarial gains and losses; Replacing the existing plan definitions of defined benefit and defined contribution with two new definitions comprising defined benefit and contribution based; Presenting actuarial gains and losses in a statement of comprehensive income, using a format compatible with revised IAS1, Presentation of Financial Statements; and Changing the unit of account measure from the plan to the individual benefit promise or promises.

The clearest consequence of these changes in terms of the discussion paper’s approach is the elimination of the defined contribution plan definition. This means that in future a plan might be assessed in terms of whether it contains either or both defined benefit of contribution-based promises. Defined benefit promises, in common with existing IAS 19 thinking, remain the default category.

Despite the clarity the IASB staff has managed to achieve in presenting the issues, the picture on financial statement presentation is to some extent as difficult to grasp as the revised plan definitions.

The decision to ditch deferral mechanisms in favour of immediate recognition of actuarial gains and losses in income will come as little surprise - in fact, IPE has routinely warned of the likelihood over the past year. Indeed, IASB has proposed this same presentational treatment - the only one mooted - for contribution-based promises.

Two further approaches are proposed for defined benefit promises: separating out service costs and presenting them in profit or loss, with all other costs going to comprehensive income; and reporting all changes in profit or loss, with remeasurements arising from changes in financial assumptions in other comprehensive income.

Although it is the interplay between the two promise definitions - not to mention their diverse measurement and presentation possibilities - that give rise to Charles Rodger’s concerns: “A number of organisations will have both final salary and contribution-based types of benefit in the same plan and I expect this to mean that they will have some problems trying to comply with the proposed approach,” he explains. “You will be asked to treat people differently depending on how their benefits arose. This can cause problems if your historical information about each pensioner is not up to speed or if you have picked up or bought plans in the past and you’ve just amalgamated the pension populations.”

In short, he warns, it is fairy easy to envisage a situation where a company has made benefit changes - as part of a derisking exercise - yet not tracked this going forward. This could leave a company unable to identify both how pensions have accumulated and their value.

“I would say it could well be a big problem for a number of people,” Rodgers argues. “Traditionally, record-keeping has not been strong for good or ill. At the heart of the issue is that companies would not, historically, have envisaged having to treat one class of pensioner differently from another.”

And, he says, it is hard to make an economic argument for treating the two classes of pensioner differently, given that you wouldn’t value them differently from a funding perspective - certainly in the UK regulatory context: “You would have valued benefits differently depending on what they were in the accumulation phase. But once the pension is determined, the liability that you have to meet is the same, independent of how it accumulated.”

Watson Wyatt’s concerns are also shared by Mercer’s Phil Turner - particularly in respect of the UK’s plans: “To value a pension you can get away with having some fairly rudimentary information,” says Turner. “The changes that the IASB has mooted in the discussion paper mean that plan sponsors will now need to look at how the benefit in payment was arrived at and ask themselves ‘was it contribution-based or final salary?’ The board has floated the notion that two people with identical benefits can be measured differently solely because of the way the benefit accrued.”

In fact, one could be forgiven for thinking that some would prefer to the IASB to drop the whole interim-fix approach and move straight on to a root-and-branch reappraisal. “They are trying to find a single number that captures everything,” he says. “It is an accounting necessity but it is always going to be a simplification of a complex situation.

“If a fuller review isn’t to be done then perhaps some more guidance around dealing with the genuinely ‘troublesome’ plans and some work on the disclosure of cost and spreading aspects would have been a better stopgap position than the situation where we will have this discontinuity between very similar types of plan.”

In fact, the IASB’s ‘disconnected’ approach to presentation - three options for defined benefit promises but just one for the contribution-based variety - led Steedman to wonder whether the proposals are workable in practice: “The IASB has asked for comment as to whether it is feasible to have different presentational approaches for final salary plans and contribution-based plans. I think it will be practically difficult to present these things in different ways, especially for a pensions plan that has both elements.”

The discussion paper opens the door to disaggregating contribution-based plans in the same way as the defined benefit plans.

All this discussion of the potential issues for the UK’s pensions population pushes the pension plans that gave the project its raison d’être to the sidelines. Which is a pity, because if the project is intended to help anyone, it should be the Swiss.

The general view in Switzerland is that their account-balance plans should be treated as defined contribution plans in the purest sense of the term - IAS19 currently treats them as defined benefit. Under the proposals put forward by the IASB, the Swiss look set to be disappointed.

“It is an improvement, but we are moving away from the projected unit credit approach to a cost-plus-option-pricing approach,” says David Pauls, a senior consultant with Watson Wyatt in Zurich. “In short, we Swiss are not going to get pure defined contribution out of this.”

There are also a lot of questions, he says. Like his UK-based colleagues, he points with scepticism to the notion that a defined benefit plan and a contribution-based promise should be valued differently in the pension phase - especially given that a number of plans here have pensions that combine elements of both types of plan design. And like his UK colleagues he also points to difficulties with historic record keeping.

What is more, Pauls also argues that the measurement proposals in the discussion paper could prove troublesome for the very “troublesome plans”, pace IASB, they are intended to cater for: “The discussion paper talks about probability weighting but it leaves more questions than it really answers,” he says. Somewhat disappointingly for the IASB, he adds that: “It doesn’t get us to any less of a complicated place than we are in right now.”

For the uninitiated, interest crediting is how the Swiss define the allocation of interest returns to a pension plan member’s individual plan account. The pension plan trustees’ discretion over the level of any interest credit is subject to a legally fixed minimum. Trustees set the amount of any credit on a year-by-year basis either prospectively or retrospectively.

It is this feature that creates the problem of modelling behaviour, says Pauls. “Because the amount is fixed annually, there is no certainty, beyond the legal minimum, of the amount of future increases. So how you project forward remains the big unanswered question.”

The issue facing the Swiss also extends further than looking to one or more equity indices and modelling future asset returns. Swiss pension plans reach their decisions on interest credits only once the funds management has been able to evaluation the level of income available for distribution. The trustees will also retain some fund income as a reserve. Just try fair valuing that.