Overview of Longevity Swaps
Background
What are longevity swaps and how have they arisen?
Over the past decade, so relatively recently by insurance and pension timing standards, there has been a big push by pension schemes and pension providers to try to minimise their exposure to longevity risk, that is the risk of pensioners living longer than expected and causing the total value of pension pay-outs to exceed the assets or reserves held to cover them.
This trend has been driven by the continuing focus on, and uncertainty around, longevity improvements. To help address the desire to manage longevity risk, reinsurance firms have developed longevity swaps; these swaps are insurance contracts that enable pension schemes and insurers to hedge or reduce their longevity risk.
Structure
(Note that throughout this article we will use the term “cedant” to describe the pension scheme or first level insurer that has taken out the longevity swap, only referring specifically to scheme or insurer where the distinction is relevant.)
The longevity swap has a relatively simple structure which is not dissimilar to the more commonly recognised interest rate swap. (Under an interest rate swap, a series of (fixed) payments, based on a pre-agreed interest rate, are made in exchange for a series of (floating) payments, based on the actual interest rate experienced at each future point in time.)
In a longevity swap, the payments are based on the benefit pay-outs of the underlying pension policies. In this case the “floating” leg payments are the actual pension payments as they become due, and are paid by the reinsurer to the cedant. In exchange, the cedant makes a series of “fixed” payments which are based on the best estimate projection of future payments at the date of signing the contract, calculated using a mutually agreed basis. These payments are confirmed at the beginning of the contract and will not vary from the profile initially established except in very extreme circumstances. In practice only the net difference between the “floating” and “fixed” payments is paid.
The cedant will also pay an additional premium with each fixed payment, usually specified as a percentage of the fixed payment. Quite often this percentage will change over time, although whether it decreases or increases depends on the shape of the deal. In fact, often it will increase; this is because the actual mortality experience is likely to diverge from the level assumed in the fixed leg payments more as the contract proceeds, so the contract will tend to be shaped with a higher premium at later dates, to reflect the increased risk of losses as time progresses. Of course this could result in the contact becoming more profitable over time if it turns out mortality has been underestimated in the pricing.
Alternatives
Longevity swaps are one of a number of different products in the market to hedge or reduce longevity risk. This article deals chiefly with longevity swaps, but as background the two main alternatives for hedging longevity risk are:
- Pension buy-outs – where a pension scheme pays a premium to an insurer, which then takes on the pensioners as insured policyholders and pays their benefits.
- Pension buy-ins – where the pension scheme pays a premium to the insurer, which then makes a series of payments to the pension scheme; the scheme will then in turn pay their members’ benefits and the buy-in is purely an investment of the scheme.
It may seem from the above that a pension buy-in and a longevity swap are similar contracts, as in both cases the members remain as members of the scheme even though the payment risk no longer remains with the scheme. However, there are key differences:
- In a pension buy-in / buy-out the premium is paid up front whereas in a longevity swap it is spread over time.
- In a longevity swap, because the scheme still has a set of fixed leg payments to meet over time, asset risk (ie the investment return required to enable the cedant to meet these payments) still remains with the scheme. In a buy-in, because the insurer is fully liable for all future payments, this asset risk lies with the insurance company.
- For most longevity swaps, where the policies are index-linked, both the floating and fixed leg will be updated with inflation, so any inflation risk associated with these policies remains with the cedant. In most cases a longevity swap should be a more cost-effective way for a scheme to hedge its longevity risk than a full buy-in / buy-out, and it also allows the cedant to maintain control over the assets and policyholders.
Reserving and capital considerations
The purchaser of a longevity swap
Reserves / funding
For the ceding insurer / pension scheme purchasing a longevity swap, the liabilities will now merely be the present value of the future fixed payments and premiums. These payments are now known and fixed, so this reduces any uncertainty in these liabilities.
The only assumption which may be needed to calculate the reserves / value the liabilities, other than the discount rate, is an assumption for future inflation if member benefit increases are linked to inflation. In most longevity swaps a non-escalated premium vector will be agreed with the reinsurer, such that either no allowance will be made for future inflation, or a notional assumption will be used. Then as future inflation becomes known the fixed vector will be updated. This means that any risk associated with inflation is explicitly excluded from a longevity swap and so remains with the pension scheme; there are of course other ways of hedging inflation risks, for example index-linked gilts.
Otherwise, from the perspective of a pension Scheme Funding valuation, the Scheme Actuary will use the mortality basis assumed in the fixed leg in calculating the Technical Provisions, as this is what future cash flows will be based on. Interestingly, because this basis is likely to be stronger than the existing Technical Provisions basis (as the insurer will have allowed some margin for error) this is likely to result in an initial increase to the level of the Technical Provisions and therefore a worsening of the funding position. However, trustees and sponsoring employers may well feel this is a price worth paying for the increased certainty.
SCR and Risk margin
From a Solvency Capital Requirement (“SCR”) and Risk Margin viewpoint where the cedant is itself an insurance company, a longevity swap should also remove all insurance components from the SCR and Risk Margin calculations for the members covered by the longevity swap agreement. Again, the exception to this may be inflation, if the deal is agreed on a non-escalated fixed leg. It is also worth noting that a longevity swap will introduce counterparty risk and so the insurer will face higher capital requirements in the counterparty default risk SCR. In most cases, however, this will be much lower than the longevity component that is removed (unless the reinsurer has a very bad credit rating).
The issuer of a longevity swap
Reserves
From the point of view of a reinsurer issuing a longevity swap the reserves can be calculated as:
Net Present Value (Fixed leg payments + Premiums – Floating leg payments – expenses)
At the outset of a contract, because the fixed leg is agreed on a best estimate basis, the floating and fixed leg payments should be equal, and so the best estimate reserve calculation simply becomes:
NPV(Premiums – expenses)
As the contract is normally written to be profitable this will in most cases yield a negative reserve, which should remain negative but run off towards zero as the contract matures. This is because reserves can basically be thought of as the NPV of all future premiums less the NPV of all future outgoings; given that the contracts are profitable at outset and that the premium is paid over time, the future premium payments should generally be worth more than any future payments made (and expenses).
Obviously as experience emerges and best estimate assumptions are updated, the floating and fixed legs may begin to diverge, which could affect reserves over time. The fixed leg will not change unless information comes to light that should have been available at the time of pricing the contract, and even then it may not be changed unless the impact would have been deemed material.
Under other bases the reserves may follow a different shape – a typical reserve shape may look as follows:
As explained above, the Best Estimate Reserves (ie the reserve under our most realistic assumptions of the future) are negative. US STAT reserves are even more negative, as the assumptions used are prescribed and rather out of date, making products seem more profitable than they are in reality. Both USGAAP and IFRS reserves are based on assumptions that include margins in case future experience is worse than expected. These margins lead to more prudent (higher) reserves.
SCR and Risk Margin
For a reinsurer, the biggest risk in taking on a longevity swap is longevity trend, ie the risk that longevity improvements are consistently underestimated over time, which means that mortality rates will have been increasingly mis-estimated as time goes on. Given this is the risk these contracts are designed to hedge this makes sense. Clearly if longevity is overestimated this will have a positive impact for the reinsurance firm as there will be lower actual benefit payments, so the floating leg payments the insurer has to make will be lower than the fixed leg payments it receives from the cedant.
There are two main factors that will affect the relative amount of longevity trend SCR capital that needs to be held for a longevity swap:
- Average age of the underlying members: As longevity trend capital arises from mis-estimation over time, the longer that members can potentially be in force for will increase the cost of mis-estimation owing to compounding. Hence younger members will tend to have higher longevity capital requirements.
- Average escalation rate on underlying member benefits: The higher the average escalation rate the higher the capital requirement for longevity trend will be. The logic for this is similar to the average age logic above, in that a higher escalation rate on benefits will mean that future benefit payments are higher and so any mis-estimation will be on increasingly larger amounts – hence the longevity trend capital will also be higher.
Most of the other capital requirements are as you would expect on long-term pensions business. However, the SCR and Risk Margin will usually be significantly higher than the (negative) reserves, and so more than sufficient to push the total capital required to issue a longevity swap into the positive.
That said, the capital required in comparison to the NPV of benefits will be very small, often in the region of a few percentage points, meaning that longevity swaps are very capital efficient compared with any other insurance contract with similar claims value.
A typical undiversified SCR and risk margin run off of a longevity swap, to the reinsurer, will be of a similar shape to the graph below.
So, for example, for a typical longevity swap hedging a value of c£500m of benefits in payment, reserves and capital requirements may be of an order indicated below (note: these are indicative figures only but will serve to demonstrate how low reserves and capital are in relation to the total benefits covered by a longevity swap).
Metric | Value at inception (£m) |
Total Benefits | 500.0 |
Best Estimate Reserves | -20.0 |
SCR | 30.0 |
Risk Margin | 20.0 |
In this case, required capital is only 6% of total benefits insured, somewhat lower than other products such as, for example, individual annuities.
Current Issues
We will now examine some of the key issues surrounding longevity swaps at present.
Is longevity improvement slowing?
One of the key assumptions in a longevity swap is around future longevity improvements. As outlined above, this assumption leads to the main source of SCR capital. Up until the last couple of years, in recent history longevity improvement rates have been universally positive (ie mortality rates have been decreasing year-on-year), and an assumption for this trend to continue in future is almost certain to have been included in longevity swaps already priced.
However, both 2015 and 2016 actually saw a small increase in raw mortality rates in certain territories (ie life expectancy dropped marginally), and this has led to uncertainty about future longevity improvements. The issue for reinsurers is whether this is the beginning of a new trend of lower mortality improvements or just volatility around the current trend.
If it is a new trend, then for future contracts this should mean that best estimate assumptions on longevity improvements should be lowered, which should lead to the expectation of lower floating leg payments in future and therefore lower fixed legs priced into the contract. This will also lead to a lower premium, as premium is generally a fixed proportion of the fixed leg. Both of these factors will make longevity swaps more attractive to buyers.
The first reinsurer to price this in would therefore have a competitive edge. Obviously if this is not a new trend then adjusting down the assumption could lead to losses for the reinsurer if they undercharge (ie it would mean that the future assumption of longevity improvement is below the actual rate and so the floating leg would be higher than the fixed leg over time, leading to losses).
Of course, for most contracts already in force, as the fixed legs and premiums have already been set it is unlikely that they will be restated, so this will lead to higher profits for the reinsurer if the floating leg they need to pay is below that expected at the time of writing the contract.
How to deal with smaller schemes
Currently the longevity swap market is aimed at larger pension schemes, c£1bn+ of liabilities, and there is less appetite from both pension schemes and reinsurers to transact on small schemes.
This largely stems from issues surrounding data. A small scheme will have less data available for experience analysis and producing a tailored mortality basis. Using a standardised basis will inevitably introduce more basis risk which reinsurers will either be unwilling to take on or will only take on at an increased price. Alternatively, a reinsurer may decide because of lack of data that they would wish to exclude certain member groups who will have the greatest impact on the risk, eg the company executives who have the largest pension pots.
This price increase or exclusion of the riskiest members will then make this longevity transfer less attractive to pension schemes, lowering their appetite to transact. Small pension schemes may also be unwilling to pay the fees to external consultants for advice on longevity swaps which could also make them less attractive, particularly if the highest-risk members are excluded.
There have been some innovations in standardising the market to make transaction costs less prohibitive for small schemes, such as the smartDB solution offered by Zurich and Mercer, but these have been slow to come to market and so longevity swaps are still predominantly sought by the largest schemes.
Deferred pensions
Deferred pension scheme members are individuals who were once a member of the pension scheme, but ceased active membership before retirement age (either because they left employment, opted out or because the scheme closed to future accrual). They have retained their pension with the scheme and will start receiving benefits once they reach retirement. Sometimes this is called a ‘preserved benefit’ or a ‘frozen pension’.
The longevity swap market generally only hedges in-payment pension liabilities (ie pensioners). There have been some swaps transacted which include deferred members but these are small in number. As for the issues over small schemes, this is largely down to data, which for deferred members tends to be of lower quality, again leading to pricing or exclusion issues. The less data available on deferred members the less confident a reinsurer will be in the basis they derive for them; this will lead reinsurers to seek a higher premium to account for the risk, because lack of confidence in the pricing basis, making the swap less attractive to pension schemes and insurers.
Another reason for reinsurers being reluctant to include deferred members is the effect this will have on the scheme duration. As deferred members are generally longer in duration (because they haven’t yet reached retirement), this leads to longer swap duration, and can lead to issues with finding assets of a long enough duration to back these members, whom reinsurers may therefore be reluctant to take on. It also affects the longevity trend, as the lower age of members increases the scope for and potential impact of mis-estimation of longevity in the pricing.
A further complication with deferred members is around commutation (where the member gives up part or all of the pension payable from retirement in return for an immediate lump sum). While assumptions can be made when pricing the contract as to the level of commutation likely to be experienced, if actual experience differs from this it could affect profitability. While insurers may take the view that this is just part of the risk associated with the contract, some may seek to treat this as an exceptional change that should be reflected in future fixed leg payments, but either way it makes the establishment of longevity swaps for deferred members more complex.
James Nicholl
January 2018
Appendix: Useful Links
The following links contain some useful information and further articles about longevity swaps:
- Article in the Actuary giving a high-level overview of longevity swaps: http://www.theactuary.com/archive/old-articles/part-6/longevity-swaps-3A-fad-or-saviour-3F/
- Database of some of the more recent longevity swaps: http://www.artemis.bm/library/longevity_swaps_risk_transfers.html
- Article exploring the decline in longevity improvements: https://www.willistowerswatson.com/en/insights/2016/06/new-mortality-data-show-declining-longevity-improvements
- Most recent CMI papers: https://www.actuaries.org.uk/learn-and-develop/continuous-mortality-investigation/cmi-working-papers/mortality-projections/cmi-working-papers-90-91-and-93
- Mercer smartDB proposition: https://www.uk.mercer.com/what-we-do/wealth-and-investments/mercer-smartdb.html