Samuel Sender argues that IAS19 should tolerate funding volatility for DB obligations and that pension funds should improve internal risk models

In a recent study1, EDHEC has analysed the impact of prudential and accounting constraints on the asset-liability management (ALM) of European pension funds in the Netherlands, the UK, Germany, and Switzerland. Defined benefit (DB) plans, in which sponsors provide guarantees to employees, are under intense scrutiny from regulators and the financial community.

Pension funds and their sponsors face two main bodies of regulation: accounting standards and prudential regulations. In this article, we highlight three main challenges for pension funds, as well as the means by which these challenges may be met:
• The trend towards stricter accounting regulations implies careful management of the accounting volatility from pension funds, with particular attention paid to the accounting discount rate.
• Higher prudential funding requirements cannot be avoided. Stricter minimum funding constraints, ie, limited allowances for underfunding, can be managed with modern ALM techniques. These techniques also help to meet the regulatory requirement for better risk management.
• Pension funds are long-term investors subject to short-term regulatory constraints. Because short-sighted strategies are counter-productive, the challenge for both pension funds and their regulator is to take a long-term approach to investing pension fund assets and to regulating pension funds.

Stricter accounting regulations
Worries are that tightening accounting standards will have a considerable impact on the financial information communicated by sponsors. For instance, were IAS19 to be aligned with FRS17 in the UK, smoothing would no longer be allowed, and under current strategies the cost of providing pensions as reported in P&L accounts would become very volatile.

One of the trends observed in the UK market after the implementation of FRS17 (the restrictive version of IAS19) is the closing of defined benefit plans and the opening of defined contribution (DC) plans.

Interestingly, because the IAS19 constraint is mainly a volatility constraint, it can be managed with modern financial tools.
• Prescriptions in the calculation of the liability must be taken into account in the design of the investment strategy. For instance, we show that as IAS19 requires spreading, ie, discounting liabilities at a rate that assumes a credit spread, fixed liability cash flows require a portfolio of forward credit rate agreements as a match.
• Modern ALM, or dynamic liability-driven investment (LDI), involves three building blocks - the performance-seeking portfolio (PSP), the liability-hedging portfolio (LHP), and the cash account - and requires in particular a reduction of the allocation to risk when the funding constraint becomes binding. The LHP, the projection of liabilities over tradable assets, is the risk-free portfolio for the ALM investor - it is the portfolio of assets that generates the least risk for the pension fund, given its liabilities. Modern ALM permits the management of funding constraints; for instance, it is possible to reduce the likelihood that the funding ratio will breach the 90-110% range beyond which surpluses and deficits impact the P&L of the sponsor.
• Because of the costs associated with developing models that allow the implementation and monitoring of these modern ALM approaches, fiduciary management is an advantageous option for small and medium-sized pension funds.
• Actuarial risks cannot be efficiently managed, as effective means of transferring longevity risk to the capital markets have yet to be found.

Stricter prudential regulations
Prudential regulations set the funding (capital) requirements for pension funds. They apply to all pension funds and sponsors, regardless of their status and organisation. Multi-employer pension plans as well as state-driven funded pension plans must comply with prudential regulations.

The IORP directive
European regulations are bound by the 2003 IORP framework directive, which lays down the following four principles: 
• Technical provisions are a prudent valuation of accrued benefits, ie, the value of the benefits available for a participant in the event his employer declares bankruptcy and the pension plan is closed.
• Some flexibility in funding requirements is provided as underfunding is allowed for "a limited period of time".
• Underfunding requires a realisable recovery plan.
• The ‘prudent person' rule, not quantitative restrictions, applies to investments.

A shift towards Solvency II?
As domestic prudential regulations are very diverse, revisions to the IORP directive are being discussed. One of the main outside references is the forthcoming prudential regulation for insurance companies, Solvency II. This is the most accomplished example of risk-based regulation, as capital requirements, defined as the value of assets over that of liabilities, are set according to a 99.5% one-year value-at-risk target and approximation. Applying Solvency II to pension funds would initially raise funding requirements.

Solvency II, however, may provide significant incentives to build internal models, a feature we view as favourable to the development of sound risk-management practices.

Higher funding requirements
Higher requirements involve an immediate increase in funding ratios, and thus require additional contributions. The risk is that these contributions will be perceived as a direct and unwelcome cost and result in knee-jerk closures of DB schemes.

Fully avoiding prudential regulations requires getting out of defined benefits altogether, such as through pension buy-outs, where the pension sponsor transfers the pension fund's balance sheet, together with its commitments, to a third party, generally an insurance company.

Stricter minimum funding constraints
Higher funding requirements, of course, involve immediate additional contributions to all pension funds and cannot be avoided. By contrast, stricter funding requirements involve improved monitoring of minimum funding constraints rather than immediate additional contributions. Stricter funding requirements can thus generally be managed (just as accounting volatility can be) with modern ALM.

The regulatory discount rate must be highlighted in the design of the investment strategy. When the regulatory constraint is binding, pension funds may be forced into prudent or even risk-free investment strategies. Risk-free, that is, from the prudential regulator's point of view. In this case, asset allocation will not merely need to take into account the presence of this constraint; instead, it will be governed entirely by the regulatory LHP.

The current trend is to use the swap yield curve for discounting. For very long-term maturities, the swap market is more liquid than the market for government debt. As such, swaps are the preferred hedging instrument for very long-term liabilities and the swap curve is becoming the standard for discounting. Two historical standards, still used today, are worth considering as well.
• The fixed discount rate: historically, most countries have used fixed rates to discount liabilities. In this case, the value of liabilities is totally independent of market interest rates. When the bond portfolio is marked to market, the portfolio that immunises the funding ratio against movements in financial markets (mainly interest rate movements) is the cash account or, when regulatory reporting takes place every year, the one-year zero-coupon bond.
• Discounting at an equity risk premium: discounting liabilities at an equity risk premium (ERP) means under-estimating these liabilities. From a regulatory perspective, discounting at an equity return is an unsophisticated way to allow pension funds to take risks.

The traditional practice of discounting at an ERP is pro-cyclical, ie, it amplifies the impact of the business cycle on the financial health of the pension fund. Many financial institutions have been using a historical estimate of the risk premium, computed over a period of say fifteen years (between ten and thirty years, sometimes depending on the maturity of the liabilities). This practice, without any further analysis, may lead to the ‘equity risk premium trap', ie, the risk of being trapped in permanent underfunding after market downturns.

In regulations that tolerate or encourage discounting liabilities at a risk premium, we strongly recommend practitioners to use forward-looking equity risk premiums, based on the valuation of the stock market rather than on historical excess returns. Because the forward-looking ERP increases when past performance falters, it follows that the forward-looking measure is counter-cyclical.

The role of internal models
Risk-based regulations are meant to foster the development of good risk management practices. In the Basel accords, risk management is simply a qualitative obligation. In Solvency II, incentives for good risk management are such that capital requirements can be measured with internal models, so that the risk of insolvency or underfunding is limited to 0.5% per year for insurance companies. Internal models must be used to manage and control risks and are thus best defined not as risk measurement software but as a full risk management system.

The Solvency II framework has been applied to pension funds in the Netherlands2. Other countries may follow. Pension funds facing risk-based prudential regulations will greatly benefit from building internal models and having them approved for the purpose of setting funding requirements. Doing so will enable them to have funding requirements more closely aligned with the nature of their risks and to enjoy reduced quantitative requirements when they have risk-mitigation techniques and instruments unrecognised in the standard formula. (For instance, when risk management relies on these models to reduce the likelihood of underfunding, a smaller capital buffer is required. Pension funds would do well then to develop internal models that rely on modern ALM principles.) In addition, when internal models show that there is significantly less risk over the medium term than over the short term, as with real liabilities, funding requirements will be reduced.

Short-term regulatory constraints
The horizon for risk measurement tends to be short, whereas pension funds are very long-term investors. As it happens, traditional wage-indexed pension plans illustrate perfectly well why short-term regulations, either accounting or prudential, may be counter-productive.
• Pension funds as long-term investors: first, pension funds are not commercial entities so in general they do not go bankrupt. Nor do they face the risks (of client runs or massive surrenders) that shorten the investment horizon in insurance companies and banking corporations and make an assessment of risk over relatively short horizons mandatory.
•  A replicating strategy for long-term, traditional liabilities: for wage-indexed liabilities, we analyse the liability-hedging portfolio (LHP), which is the portfolio of assets that generates the least risk for the pension fund.

In the long run, real assets such as equities are needed to replicate wages. After all, wages, like equities, are linked to overall economic performance. This allows us to run a specific model in which equities help predict wages but wages do not help predict equities.

Risk of short-sighted strategy
In the early years, high equity exposure results in high volatility in the funding ratio and consequently in the sponsor's accounts. Regulations and accounting standards that focus on short-term figures are incentives for short-sighted investments, especially for weaker sponsors or the least well funded pension plans. However, short-sighted strategies in the case considered here imply ever-greater long-term volatility.

Short-sighted strategies involve investing in equities in accordance with their short-term relationship to wages, ignoring the longer-term dependencies. Because wage-indexed liabilities mean considerable exposure to the economy, and because in our model exposure to the economy can be achieved only through equity investments, the exposure to the economy on the asset side is not sufficient to cover the liabilities. As a consequence, risk builds up on the liability side for the short-sighted investor and annual volatility rises. In addition, a careful examination would show that the expected return of bond-like strategies would be insufficient to cover the rise in wage-indexed liabilities.

The ramifications for both accounting and prudential regulations are:
• We strongly recommend that IAS19 tolerate some volatility in the funding ratio for wage-indexed liabilities. In this respect, alignment with FRS17 would be counter-productive.
• Risk-based prudential regulations, which require buffers as a function of risk taken in pension funds, should base their measures of risk on the long-term volatility of ALM strategies - not the short-term volatility.

There are also ramifications for pension funds. As they need real assets to generate real liabilities, they must develop the techniques and models for efficient long-term investing and prove to regulatory bodies that they have mastered a long-term approach to managing their risks. As a result, internal models must be developed.

1 N. Amenc , L. Martellini and S. Sender, Impact of Regulations on the ALM of European Pension Funds, January 2009, EDHEC Publication. Sponsored by AXA Investment Managers.
2 Implementation measures differ from those for insurance companies. Assets must be high enough to cover the liabilities over a one-year horizon with a probability set at 97.5%.

Samuel Sender is applied research manager at EDHEC Risk and Asset Management Research Centre in Nice. This article was based on research carried out within the EDHEC/AXA Investment Managers regulation and institutional investment research chair.