With the IASB about to embark on an overhaul of pensions accounting, Stephen Bouvier looks at what it might mean to apply the board’s insurance liability model to a pension promise
The decision by the International Accounting Standards Board (IASB) to embark on a potentially wide-ranging reconsideration of its pensions accounting standard IAS19, Employee Benefits, certainly raises questions. And nowhere has this been more apparent in recent weeks than in relation to suggestions from senior board figures that the IASB might look at applying to pensions some of the lessons learned on its insurance-accounting project.
The origins of the IASB’s insurance project can probably be traced back to 2004 with the decision to set up a working group comprised of senior industry experts. The board followed that move with a discussion paper in 2007 and two exposure drafts (EDs). Aspects of the most recent ED, issued in 2013, are being discussed by the board.
One of the first hints that the insurance accounting model might hold some relevance for pensions appeared in September 2013 when staff at the IASB’s IFRS interpretations committee presented a broad analysis of the impact of applying the model to the committee’s September 2013 meeting. At the time, the committee was considering the measurement challenges thrown up by so-called intermediate-risk plans.
Applying the insurance approach to a pension promise, staff noted, “would require the benefit promise to be measured using an expected value approach that would be consistent with observable market information”. This would, they added, “also require an entity to use a discount rate that reflects the dependence of estimated future cash flows on returns on underlying items”.
The approach nonetheless brought with it several trade-offs. For example, although it would “faithfully represent the benefit promises within the agreed scope”, staff also warned that it brought with it a high cost penalty. Similarly, despite doing a better job of addressing the wide variety of modern plan designs, the model also produces a similar effect to the board’s ill-fated 2008 discussion paper, which exceeded its anticipated scope.
In many ways, the concerns about scope fall away, if only because the IASB is talking about applying the new measurement model to all types of plan design. But back in 2008, the board wanted to focus only on so-called ‘troublesome’ plans and not traditional DB and DC ones. But with the project well under way, the IASB discovered to its cost that it was impossible to rein in the scope of its new fair-value model.
The IASB’s putative insurance-accounting model rests on four building blocks:
• A probability weighted calculation of expected future cashflows;
• A risk adjustment; and
• A contractual service margin (CSM).
Most analyses of the model aimed at insurance preparers start with the contractual service margin. However, for reasons that will become apparent, the CSM is perhaps the least relevant for the three building blocks from a pensions perspective.
The first building block requires preparers to arrive at a probability-weighted calculation of expected future cashflows. The obvious lack of certainty here means that they must look at the range of possibilities and assign probabilities to them. This is not just a single best estimate, although if your expected outcomes follow a normal distribution curve, you would come up with more or less the same answer.
“I have no idea about the extent to which preparers will use a sophisticated approach other than where they already use an option-pricing approach,” says David Holliday, a senior insurance specialist at KPMG.
“There is a similar approach in Solvency II, and I suspect they will do that calculation and then worry about how they apply IFRS 4, although there are differences between IFRS 4 and Solvency II in terms of what you include in your cashflows, particularly around expenses.”
Next is discounting. One of the questions raised by commentators since the IASB first
made noises about looking at pensions is whether this will lead to a Solvency II for pensions through the back door. Another question is whether the IASB will eventually move to a risk-free discount rate.
“Because your first three building blocks are a measure of the insurance liability, the CSM is really a sort of plug or a balancing figure defined as ‘unrecognised profits’”
First, Solvency II is prescriptive about discounting, although it does allow the use of higher discount rates to cater for the fact that insurers are not going to trade and sell cashflows where they hold swaps. Both Solvency II and embedded-value accounting start with the premise that you are valuing liabilities in the same way as markets would – albeit with no allowance for own credit.
“So an insurer might start with a reference base-rate such as government securities, although recent market upheaval means that some are now looking at swap rates as the base reference rate with, perhaps, a matching or volatility adjustment that ultimately gives a lower liability,” says Holliday. “The IASB insurance model is much more open and allows you to factor in issues such as the fact that you will not be trading the swaps.”
For its third building block, the IASB has proposed a risk adjustment. Again, whereas Solvency II has a specific risk-adjustment methodology, the cost of capital approach, IFRS 4 is more relaxed. For example, it allows you to use virtually any approach you want to use – cost of capital or confidence limits – as long as it has certain characteristics and you explain it. Different offsetting requirements across the two models can also lead to different outcomes.
The fourth building block is the CSM. Ironically, although this component is a key performance metric for insurers, it has little relevance for pensions. At contract inception, an insurer will work out its expected cashflows, discount them and make a risk adjustment. A contract that looks profitable at inception should produce an asset. But recognising that asset would mean recognising a day-one profit, which is something that IFRS 4 prohibits. As a result, the IASB has come up the contractual service margin.
“Because your first three building blocks are a measure of the insurance liability,” Holliday says, “the CSM is really a sort of plug or a balancing figure defined as ‘unrecognised profits’. That is not the full picture, however, because there are restrictions on what you can and cannot include in your cashflows. In essence you defer and spread profits – but not losses – by recognising the CSM in P&L over the contract term.
“I cannot see how you would have a CSM with a pension obligation. It is more about how you would recognise profit over the term of a contract and not really part of the liability, it is an extra bit to avoid recognising day-one profits.”
So would the IFRS 4 model mean a shift to a fair-value model? Not really, Holliday says, who notes that there are express points of difference between IFRS 4 and fair value. “First, it doesn’t take into account the credit standing of the insurer, whereas a pure fair-value model would do,” he says.
“The other is that IFRS 4 allows the insurer to measure the liability using the fulfilment value approach, which is basically taking into account the cashflows that it expects to see under the contract – particularly expenses. So this is more of an own-entity approach than an exit price.” It’s also a lot more work.