Debt Management Office
20 Jan 2005
Consultation on Issuance of Ultra-Long Gilts and Gilt Annuities
A. Potential demand for new instruments in 2005-06
a) What is the potential scale of demand for ultra-long (circa 50-year) conventional or index-linked gilts?
Pension funds are increasing the proportions of their assets held in bonds for a number of reasons including:
(i) the accounting standard FRS 17 uses the yield on AA corporate bonds to value pension fund liabilities and companies wish to see the results of these valuations being reasonably stable from one year to the next;
(ii) the large majority of defined benefit (DB) schemes are now closed to new members and trustees are paying increasing attention to the end-game position where future liabilities become increasingly more certain. Bonds are more appropriate assets the more certain the amounts of the liabilities.
(iii) an increasing awareness of the benefits of diversification of assets against a background of pension funds still holding in excess of 60% of their assets in equities at the end of 2003.
(iv) a defensive reaction to the large declines in equity indices since 1999.
Although the bonds being bought are considered to be the best available match for pension fund liabilities the very large proportion of those bonds will still be a mis-match because their duration is far too short. Even a fifty year maturity will be shorter than a proportion of the liabilities.
Pension funds have been building up their holdings of corporate bonds in recent years, partly by switching out of gilts, and ideally would continue to buy corporate rather than government bonds. However, while fixed interest corporate bonds are readily available this is not true for index-linked where HMG continues to be the dominant supplier.
The effects of (i), (ii) and (iii) are likely to lead to funds increasing their proportions of assets held in bonds for the foreseeable future. UK pension fund assets are believed to be of the order of £700bn. so a 1% shift in assets represents a movement of about £7bn.
b) If the DMO were to extend the maturity range of gilts it issues, for which maturities would there be most potential demand?
DB schemes with members in their 20’s and 30’s have liabilities extending for another 60 or 70 years. With the longest dated gilts in issue being of the order of 30-33 years to maturity the “rump” liability, beyond that which can be hedged or matched, is not only very large but covers a period of about 35 years. If 50 year gilts were available the rump would not only be reduced to 15 years but, because of the maturity profile of the liabilities, the rump would be considerably smaller in size than the proportional reduction in years would suggest.
While defined contribution (DC) schemes do not have liabilities in the strict sense the demand for annuities from member retirements, and for bonds from active and deferred members, will gradually replace DB liabilities as time passes and demand for ultra-long gilts will continue.
A 50 year maturity can be expected to be very popular.
c) What is the potential demand for gilts structured in annuity format?
The annuity format is very likely to be the pensions industry’s preferred choice as it matches the pensions in payment format. Current index linked issues in particular are unwieldy because the payment profile is dominated by the redemption amount. Although it is possible to use the swaps market to change the profile there are various reasons why this is unsatisfactory. It is difficult to judge whether demand for annuities would be greater or less than that for conventional gilts as investors have had to make do with what is available. Annuities are ideal for pensions in payment but deferred annuities would be more appropriate for deferred pensions.
d) How sustainable would demand be for ultra-long and annuity format gilts over time?
DC scheme members have the same need as DB schemes for duration and increasing longevity is an added reason for wishing to invest in ultra-long maturity or annuity format gilts. It is possible that demand for index-linked annuities would reduce relative to fixed annuities over time as there will be no requirement for members of DC schemes to buy index-linked pensions on retirement. Also, the ability for DB schemes to cap the inflation-proofing of future accrual at 2.5% may have a similar effect while nominal yields exceed real yields by a similar or larger amount than 2.5%.
However demand for ultra-long or annuity format index-linked gilts is likely to remain strong for the foreseeable future given that a very large proportion of current pension fund liabilities are index-linked, very long term or both. The Government would appear to be the obvious issuer of ultra-long debt given that company covenants change over time and it is doubtful whether much 50 year corporate debt is appropriate for investors.
e) In the event that it was decided to issue only a single form of new instrument in 2005-06, which of the instruments outlined above is of most potential interest?
The index-linked annuity format. An annuity because of the correspondence with the format of pensions in payment and index-linked because of the sheer volume of index-linked pension fund liabilities and the fact that only the Government is likely to be an issuer in any meaningful way.
The size of the corporate index-linked market is of the order of £10bn and has never looked likely to be more than a small addition to index-linked gilts. Conversely the UK fixed interest corporate debt market is now bigger than the gilts market.
f) How much, in £ billions should the Government initially plan to supply in one financial year?
In response to question p) we have listed some of the advantages to HMG of issuing new instruments as proposed and, given the weight put on long-dated gilt yields, there would ideally be a sufficient amount made available to avoid there being a false market in the early stages. We note that about 35% by value of the index-linked gilts market matures in the 5-10 year band and, to reduce the strain on new issuance, consideration might be given to providing reinvestment terms into any new long dated issues.
Other respondents are likely to be better placed to be specific about amounts.
B. Issuance procedure
g) If the DMO were to issue any such new instruments, should issuance take place from the outset through auctions?
No view
h) Alternatively, should the DMO consider other means to distribute the bonds when first issued (or perhaps on an ongoing basis) and if so, which?
Consider private placement to large pension funds
C. Instrument design
i) If the DMO were to issue ultra-long and/or annuity gilts, should such issuance take place preferably in conventional format or in index-linked format?
Reasons were given in response to e) for our preferring index-linked and our response to p) reinforces that view.
j) If the DMO were to issue gilts with an annuity structure (in nominal or index-linked format), should the DMO allow for these new gilts to be stripped?
The idea of stripping an annuity contradicts the view that pension funds will need the whole annuity to best match pensions in payment. It may take time for market participants to adjust to dealing with an annuity structure without any added complications.
k) For new index-linked gilts with a three-month lag, and/or for any new gilts issued in annuity format, the DMO would be interested in views on whether such gilts should be held only in CREST. This would allow a zero ex-dividend period to be introduced for those gilts. Respondents may wish to take account of possible liquidity issues arising from the need to make payments into Cash Memorandum Accounts (CMAs).
No view
l) If the DMO were to issue gilts in annuity format, what are stakeholders’ preferences regarding the frequency of cash flows?
No strong view but six-monthly as for gilt interest payments should not present a problem.
m) Do recipients have any other comments or suggestions on the proposed product design in Annex B?
No
D. Timing for introduction of new instruments
n) If the DMO were to issue any such instruments – and gilt annuities in particular – what would be the lead times required by investors, Gilt-edged Market Makers (GEMMs) and other interested stakeholders before such issuance can realistically take place?
No view
o) What are the lead times required by investors, GEMMs and other interested stakeholders before issuance of a three-month lag index-linked gilt can take place?
No view
E. Potential advantages/disadvantages with issuance of new instruments
p) How do respondents assess the benefits, for HM Government and for investors, of any new instrument being created?
HM Government
In the Spring 2004 edition of the Bank of England Quarterly Bulletin the Bank observed that “if long-term nominal forward rates reflect long-term inflation expectations, real interest rates and various risk premia, the long-term sterling forward rates remain puzzlingly low.” The Bank went on to suggest that this might be explained by the high level of demand for long-dated gilts by UK life insurers and particularly by UK pension funds for index-linked gilts.
UK real forward rates at the long end of the curve have since fallen further and as at 10 January 2005 the forward rate between IL 2024 and IL 2035 was 1.22%. It is arguable that the rate of inflation derived from the differences between nominal and real gilt yields is over-stating the outlook for future inflation. If this is correct then pension fund deficits are generally being overstated.
Given that index-linked gilt yields are a major input to actuarial valuation bases (although usually shown as the combination of a long-dated nominal yield and the associated implied inflation rate) it is difficult to believe that a rise in index-linked yields would be considered wholly undesirable. This “overstatement” effect could be very significant both in absolute terms and in terms of the overall funding position of pension schemes.
It is surely preferable for HMG to issue (ultra)long-dated gilts on a real return basis given that the real risk of issuing nominal debt is unknown. The Bank of England is implying that pension fund demand for index-linked gilts makes this a particularly auspicious time to be doing so.
Investors
The inability of funds to invest in assets of sufficient duration is a major source of risk and its removal would allow a considerable improvement in their ability to hedge or match pension fund liabilities.
The main drawback with DC schemes is that investment risk is transferred to the member. Ultra long gilts would allow the duration risk to be considerably reduced or the risk taken to be measured more accurately.
Buy-out market terms should improve markedly as one of the major uncertainties is removed. Actuarial Guidance Note 9 (GN9) recommends a default buyout valuation yield of long gilts minus 50bp: based on a proxy of a long gilt yield for pensioners and long gilts minus 100bp for deferred pensioners. Clearly the life assurers’ premium loadings for duration risk are substantial even allowing for the bases also reflecting mortality risk.
q) Would the adoption of any annuity format detract from the liquidity and benchmark status of such bonds?
Yes as it is very likely that annuities will be very much a “buy and hold” type of investment and more so than index-linked gilts and the longest dated fixed-interest issues are currently. The problem would be greater if annuities were strippable as matching could then be more exact.
r) Would the introduction of an annuity format detract from the liquidity of standard benchmark gilts (particularly as the annuity format bonds age), or otherwise fragment the overall gilts market?
Yes - for reasons given in response to q)
F. Market maker responsibilities
s) If the DMO were to issue any such instrument, should the roles and responsibilities of GEMMs be identical to those applying to other gilts?
No view
Pension Trusteeship
Back to News