Pension insurance systems can be an important component in maintaining the viability of defined benefit (DB) systems by providing a safety net in cases of sponsor insolvency with underfunded pension obligations. However, creating such a system does not solve all the problems. Such insurance funds already in existence have suffered under recent market conditions. Some of the difficulties these insurance systems face and the necessary features for an insurance system to have any chance of success are here explored.
It is particularly instructive to analyse the US PBGC programme to better understand the challenges being faced by all these major worldwide programmes.
Over the past few years, the PBGC has experienced a rapid and severe deterioration in its net position, from a record surplus of $7.7bn (e6.4bn) in 2001 to a deficit of $11.2bn in 2003, including its largest annual loss since inception of nearly $11.4bn during 2002.1
The immediate cause of the deficits is of course an increase in distressed terminations of underfunded plans. The underlying factors, however, behind the current deficit are:
q Lenient funding rules leading to distressed terminations of sgnificantly underfunded plans:
Existing US pension funding rules, which employ extensive smoothing and deferral mechanisms, often insulate companies from having to fund emerging deficits in a timely fashion. For companies unable to accelerate contributions when finally required, this ultimately results in corporate bankruptcies with massive underfunded pension plans, and thus massive claims for the PBGC;
q Investment policy/hedging:
The typical pension plan insured by the programme has been underexposed to long duration bonds and overexposed to equities. This asset-liability mismatch leaves the plans particularly susceptible to low interest rate and poor equity return environments. Further, those poor economic conditions which lead to a deterioration in funding levels often coincide with those triggering corporate bankruptcies;
q Inadequate premium income:
Over the long-term, premium income should be sufficient to close the gap between assets and liabilities under trusteeship, as well as cover future contingent liabilities. However PBGC premiums have not been calibrated sufficiently to the risks and have thus been disconnected from the benefits claims;
q Understated liabilities:
The PBGC liability exposure only takes into account the present value of benefits for plans already under PBGC trusteeship and plans the agency considers as ‘reasonably possible’ or ‘probable’ near-term distressed terminations (companies that are in or are close to bankruptcy). This approach does not include a contingent liability that would normally be calculated as part of a standard actuarial approach;
q Complex regulatory process:
Changing PBGC policy requires congressional legislation which can be lengthy and bureaucratic.
The strength of funding regulations and the asset allocation decisions of pension funds have a direct impact on the scope and costs of pension insurance systems. Poorly funded schemes provide little security in the case of the corporate insolvency. Similarly, mismatched investments run the risk of solvency shortfalls due to adverse market conditions.
The UK and US funding regulations are examples of funded pension systems where existing funding regulations have been seen not only to be wanting in terms of providing sufficient minimum funding protection, but also to encourage companies (especially weak companies) to take excessive risk.
Pension funds in the UK and the US have taken significant duration risk and equity risk, misaligning their investments with the provision of pension security.
Whilst there is no asset mix that provides a perfect match for pension liabilities, an ALM-driven approach would see pension funds invest high allocations to long duration bonds to meet the duration of their debt-like pension liabilities of about 12-15 years. A small allocation to other (typically higher returning) assets less correlated to the liabilities would then be set to meet other risks;
Instead, in both the UK and the US, we have seen typical equity allocations in the range of 60-80%. Further, the fixed income allocation, broadly making up the balance of the asset allocation at 20-40% has had a typical duration of about four years.
This allocation results in two very large bets. First, the duration mismatch misaligns the move in the assets with the move in the liabilities, meaning, for example, that a 1% fall (increase) in interest rates will increase (decrease) the liabilities by roughly 12-15%, with a corresponding increase (decrease) in the bond assets by only 4%. Secondly, the equity mismatch leads to an investment that does not typically perform in line with the liabilities. Equity performance typically feeds straight through to the funding level, with an increase in equity levels improving the funded status, and a fall in equity levels leading to a deterioration in the funded status.
It is fundamentally important then, that funding regulations and pension insurance systems are not designed separately but in unison. This is highly relevant to the situation in the UK at the moment.
The UK government has, alongside the introduction of the Pension Protection Fund (PPF), announced replacing the current Minimum Funding Requirement (MFR) with new scheme specific funding requirements. It is difficult to envisage how these scheme specific funding regulations can operate alongside a framework of a pension insolvency system whilst maintaining equity amongst the sponsoring corporations.
The direction the UK government appears to be heading in is to weaken existing minimum funding regulations, to maintain an environment that encourages the mismatching of assets and liabilities and to rely on the PPF to take up the slack. This approach is remarkably similar to that in the US, under which the PBGC has operated over the past years.
Creating a better system
Macro-considerations: There are two major inherently uninsurable risks that cause pension insurance systems to fail.
q Moral hazard:
A poorly designed system can be gamed. A weak company could eg, grant benefit improvements in lieu of pay, defer funding, or increase asset risk when funded status is low. In all instances, the downside result will inevitably be financed by the insurance fund, with all of the upside results benefitting stockholders;
q Systematic risk:
Equity exposure and other systematic risks cannot be diversified away or insured at a price lower than that achievable by capital market hedges (or investing in liability matching bonds).
The basic principle underlying any insurance system is sharing of non-systematic risk across a broad group of policyholders. The incidence (and severity) of these events should be relatively independent across the insured population. For a pension insurance scheme, insolvencies brought about by events such as bad management choices, fraud, error, and plain old ‘bad luck’ are non-systematic. But what about insolvencies brought about by macroeconomic and associated capital market valuations?
In a classic recession funded status levels drop precipitously and simultaneously for all plans invested in equities (versus bond-like liabilities), and the abilities of many companies to finance the shortfall also deteriorates, resulting in a highly clustered insolvencies. Further, as pensions become a larger portion of corporate capital structure, systematic (and systemic) risk increases.
The principles guiding development of a sound pension insurance system are similar to those that should guide any insurance system (or individual insurance company).
q Reasonable benefit levels:
In order to provide meaningful financial security to participants, a pension insurance system needs to protect a substantial portion of promised benefits (of course the promised benefits should be reasonable). But at the same time, the amount covered must be limited protecting against moral hazards.
q Pricing and premium structure: Premiums for pension insurance should reflect risk of incurring a claim, a risk that increases as the financial strength of covered companies decreases. A number of proxy measures of potential claim incidence are possible (eg, rating agency classification, credit default swap levels). The premium structure should also reflect the potential severity of the claims, in particular, the size of the plan and current level of underfunding.
An additional factor that could be used to classify risk is the degree it is inherent in the asset-liability strategy of the covered plan. A system that rewards a corporation for better matching assets and liabilities would provide incentive for corporations to adopt risk minimising behaviour.
As in any insurance system, an appropriate pension benefit insurance system must spread risk to a certain degree. An obligatory monopolistic system will in general allow more risk sharing, as employers would not have the option to seek an insurer with the most favorable premium. But if the system has too little classification (too much spreading), employers may choose to ‘opt-out’ via plan termination, which is also a poor outcome.
Over a long horizon, the aggregate level of premiums, together with investment returns, should approximately reflect aggregate claim levels, perhaps with a margin for building a surplus.
q Investment strategy:
Like the plans it insures, a pension insurance programme needs an investment strategy that is appropriate for both its current liabilities (as described above) and its contingent liabilities (which should be invested counter cyclically by, for example, employing instruments ranging from relatively simple derivative strategies to more complex contingent swap arrangements).
q Measurement and disclosure:
Systems that offer little to no disclosure can undermine their strength simply by keeping their finances hidden. Financial measurement and disclosure should be substantial and done consistently. Asset and liabilities in respect of existing claims should be measured on a marked-to-market basis, with liabilities discounted at a rate consistent with high-quality market rates of similar duration. Future premiums and contingent future liabilities are harder to measure – nonetheless, a projection methodology (a stochastic or multiple scenario framework) should be employed to estimate ranges for these values.
A pension insurance programme should also strongly encourage complete disclosure from the plans it covers, so there is clarity on the nature of the insured risks.
Nigel Cresswell and Aurelie Rabou are with Morgan Stanley’s Global Pensions Group, based in Frankfurt and London respectively
Derived from “Insuring the Uninsurable?” published in Morgan Stanley’s March 2004 Global Pensions Quarterly. Contact: email@example.com, or firstname.lastname@example.org for the full copy.
1Data from the PBGC Annual Reports