Flaws in the holistic balance sheet
On 13 October 2014, the European Insurance and Occupational Pensions Authority (EIOPA) launched a consultation paper – Further Work on Solvency of IORPs – with a deadline of 13 January 2015 for responses. The paper outlines proposals for six different models of a future supervisory framework in which a holistic balance sheet (HBS) could play a role. Furthermore, EIOPA is consulting on key definitions in relation to the HBS and explains how it could assess the support of a sponsoring company (sponsor support) or account for variable benefits in this balance sheet.
We expect that it will not be possible – despite the current consultation paper and the responses that EIOPA might receive – for the HBS to develop into a suitable tool for financial supervision (capital requirements, pillar 1). The reason is that the HBS approach contains serious fundamental shortcomings which cannot be adequately solved.
A call for advice from the European Commission to EIOPA in 2011 on a revised IORP Directive was the starting point for the HBS approach.
The Commission requested EIOPA to investigate whether the quantitative rules of the Solvency II Directive for insurance companies could also be applied in the IORP Directive for occupational pension funds. In its response, EIOPA stated that various provisions of Solvency II could, in theory, be applicable to pension funds, but that the specific characteristics that make pension funds different from insurers should be taken into account. In EIOPA’s view this could be done by means of the HBS approach.
Under the HBS, IORPs domiciled in EU member states should not only value their normal (financial) assets and conditional/unconditional liabilities, but also their adjustment and steering mechanisms, such as increases of contributions, discretionary decision-making processes such as not (fully) granting indexation, sponsor support and reduction of benefits. The aim was to accommodate the principal differences between pension funds and insurance companies. Furthermore, EIOPA proposed to combine the HBS with the calculation of a solvency capital requirement (SCR).
The next step in the HBS process was a quantitative impact study for IORPs (QIS 1) – carried out by EIOPA in 2012 at the request of the Commission – of the potential consequences for European pension funds of an HBS approach.
This study was conducted in eight EU member states where several pension funds in these states participated. Principally, there were numerous unclear elements and criticisms of the HBS approach, along with serious doubts about whether it would ever have the capacity to develop into an adequate tool for financial supervision. The consultation paper has now been published and aims to meet these doubts and criticisms. It proposes improved definitions and methodologies to value the HBS, as well as six potential models for HBS supervisory frameworks.
EIOPA believes that the HBS could be introduced in a future revised IORP Directive and implemented by member states.
However, attention should be paid to the following:
• The HBS proposals are not included in the proposed revision of the IORP Directive, published by the Commission on 27 March 2014;
• EIOPA is undertaking its work on the HBS on its own initiative in its role as independent adviser to the European political institutions, not at the request of the Commission.
Fundamental shortcomings of the HBS approach
Prudent supervision should focus on the question of whether past promises made by the pension fund can be realised. In our view, the HBS approach contains several serious fundamental shortcomings, and as a consequence cannot meet this goal.
Error in Solvency II replicated in the HBS approach. The first problem with the HBS approach is directly derived from Solvency II, where capital is erroneously required for conditional promises, even if these are conditional on the future financial health of the insurance company (profit sharing). As a consequence, there is a double charge for risk taking. First, greater risk increases downside risk and therefore a higher SCR is required.
Second, it also increases upside potential and thereby the profit-sharing option. Where the first charge makes perfect sense, the second does not. Either a company makes a profit and so will have the funds to share a part of this, or it makes no profit and makes no promise to pay anything. Insurers can circumvent this by formulating contracts appropriately or by not promising profit sharing.
For pension funds – at least those in the Netherlands – conditional indexation is an important aspect of the pension contract, which makes this error more binding. The double-capital charge for risk taking gives pension funds an implicit incentive to avoid risks. This can be both bad for the participants as returns are likely to be lower, and for society as a whole as it reduces the ability of pension funds to finance long-term investments.
Apart from introducing the wrong incentives, the requirement also leads to a dynamic inconsistency. Suppose a pension fund has a strong sponsor that is willing to finance a deficit that might arise due to the value of the indexation option; additional sponsor contributions will increase the value of the indexation option, thereby creating a new deficit. This process can continue until indexation is almost fully guaranteed but would be clearly at odds with the agreed conditional character of indexation in the contract.
In the HBS, EIOPA tries to correct this conceptual mistake by adding compensating options available for pension funds to finance future deficits, namely increases of contributions, benefit reductions and sponsor support. As such, EIOPA is compensating one mistake with another. For example, in respect of benefit reductions, it is doubtful whether a supervisor – which ought to protect pensions of current and future participants – should disregard pension promises simply because they are no longer likely to be met.
Moreover, as indexation options are less sensitive to changes in volatility when they are far out of the money, the relative attractiveness of risk taking as a function of the financial health of the fund is contrary to the desired situation from a supervisory point of view. If a fund is highly underfunded, the indexation option will be far out of the money and extra risk taking will hardly affect this value. But, on the other hand, the benefit reduction option will increase with risk taking as this option is more in the money.
For a fund with a large surplus, the opposite applies – the indexation option is in the money and the benefit reduction option is hardly affected. Consequently, risk taking will be less attractive for a rich fund than for a poor one. Again, this seems contrary to a good policy for the protection of participants.
To conclude, the best solution to circumvent this error of Solvency II is not to enlarge the balance sheet further, but to limit it to unconditional assets and promises.
Combination of HBS and SCR is inconsistent. The second problem with EIOPA’s view on the HBS is the inclusion of an SCR. The HBS shows the current market value of all conditional and unconditional pension promises (assuming there is a complete market, which is not the case), and the way in which these promises are backed by current assets and conditional future payments (or benefit reductions).
As capital requirements are neither part of the pension promise nor of the financing of this promise, there is no place for an SCR on the HBS. This can be illustrated for a simple (complete) contract with a finite horizon where the participants will receive all revenues of the fund when this will close. If the stochastic simulations for the HBS are conducted over the full (finite) lifetime of the pension contract, the HBS will exactly balance.
The current value of assets is exactly balanced by the current value of unconditional liabilities plus the profit-sharing option (indexation option) minus the loss-sharing option (benefit reductions). If the simulation horizon ends before the end of the contract, conditional pension rights after the simulation horizon will not be valued, and consequently there will generally be a residual (positive or negative).
This residual represents transfers to or from the generations that will still be in the fund after the simulation horizon. In the view of EIOPA the pension fund only disposes of sufficient capital when this residual will exceed the SCR. This would then in practice mean that, irrespective of the starting financial situation of the fund, current members should always have to make transfers to future generations.
This cannot be regarded as beneficial for the participants. In addition, the longer the simulation horizon, the smaller the value of the residual (as the transfers are discounted) will be, and therefore the less likely that the HBS (including the SCR) will balance.
Supervisory response problematic. A third problem with the use of the HBS concerns the supervisory response, given that the HBS can only be calculated assuming a complete contract (including an agreement beforehand on the sharing of surpluses and deficits between the different stakeholders and all recovery mechanisms). This can be demonstrated for an IORP with an insufficient ‘holistic funding ratio’ and a deficit of €100m in order to comply with the SCR.
In this situation, for example, an additional payment by the sponsor of €100m will be impossible, because this future security mechanism has already been valued in the HBS. Therefore, the outcome of the HBS is a take-it-or-leave-it deal. If the supervisor does not like the outcome, it might only suggest adjustments in the contract or in the recovery mechanisms; but the resulting HBS outcome will be highly unpredictable as all HBS items are interrelated. Consequently, the approach is not suitable for Prudential supervision.
Technical and practical problems. In addition to these fundamental shortcomings, the HBS approach implies various technical and practical problems. There is, for example, a serious model risk because the outcomes of the valuations are very sensitive to the assumptions made. These assumptions may relate to various elements of the HBS; for example, the investment horizon which is taken into account by the pension fund involved, the parameters to be chosen in the (risk neutral) scenario set and the maximum value of the sponsor support (which may in practice vary between zero and the full value of the technical provisions). IORPs could change assumptions in order to have other, more desirable, outcomes.
And last, but not least, the calculation of the options included in the HBS is highly complex and therefore very costly. In order to mitigate this, the consultation paper offers numerous simplifications to calculate an HBS, but this results primarily in a balance sheet without any logical economic interpretation. This is because the interaction between the various balance sheet items gets lost and because the assumed option prices disregard the price of risk (the expected payout on an insurance policy is not equal to its premium).
Conclusion and recommendations
Although we generally support EIOPA’s statement that specific characteristics making pension funds different from insurers should be taken into account, the HBS approach will in our view never be able to develop into an adequate tool for financial supervision.
This is not only the result of its practical and technical problems and complexity (which will make it very difficult or even impossible for small and medium-sized pension funds to use), but also the serious fundamental shortcomings in the overall approach.
We therefore recommend that EIOPA – at least for the time being – refrains from its intended next steps in the HBS process. More specifically, EIOPA should postpone its plans for a next QIS in 2015. The fact that the Commission has not requested the exercise can be considered an additional reason for this course of action.
Wilfried Mulder is senior policy adviser, group strategy and policy and Peter Vlaar is senior strategist, asset and liability management at APG