|Main index||Financial Markets index||About author|
Julian D. A. Wiseman
Abstract: On Friday 7th September 2001 the UK’s Debt Management Office published a paper (local copy), consulting “on whether to adopt a new design for new issues of index-linked gilts”. A copy of the reply sent to the DMO is below, and is also available as a PDF file.
Contents: Q1 new design, Q3 Canadian design, Q4 deflation floor, Q5 strips, Q7-9 trading basis, Q11 converting, early redemption clause, frequency, rounding, indices.
Publication history: sent to the DMO, and here. Usual disclaimer and copyright terms apply.
1. No. Or at a minimum, the DMO should ensure that any new design of ILG has a lag at least as long as the average lag in the gathering of taxation revenue.
2. Broadly speaking, the DMO’s objective is to minimise the long-term cost of borrowing, whilst taking account of risk. Shortening the ILG lag would not lower the long-term cost of borrowing, but would increase the risk taken by the government.
3. The marginal investor in all the main government bond markets, and in the UK’s index-linked gilt market in particular, is the institutional investor.
4. Many UK institutional investors have written inflation-linked policies to their end investors. But most of these policies use a constant level of inflation for each calendar year, a level fixed a couple months before the start of each year. Thus the average lag in the investors’ liabilities is also about eight months. (This was discussed at the Bank of England’s 1995 conference on index-linked gilts.)
5. Hence a significant change in the lag would not make the DMO’s product more attractive to the relevant investors.
6. However, a shortening of the lag would increase the riskiness of the government’s liability portfolio during times of financial stress.
7. Imagine that an external shock causes inflation to increase sharply. This shock might be a war, or a contraction in the supply of a material that has become vital to the economy.
7.1. In such a situation, it would benefit the government to be able to ‘default’ on its debts. Of course, recalling Lord North’s “sacred duty”, the UK government would not default, but it would be helpful if the terms of the debt allowed reduced- or non-payment.
7.2. Indeed, at the 1995 conference academic speakers suggested some form of ‘War Bond’, with higher payments during times of peace than of war.
7.3. But an ILG with a long lag functions in this manner. If inflation is large, the 8-month lag lowers the real size of the government’s payments. If inflation is modest, the effect is small.
8. Hence a long lag also has a stabilising effect on the economy. If inflation is low, the government is making high real payments to the private sector; and if inflation is high the government is imposing an inflation tax on the private sector. This is pleasingly counter-cyclical.
9. The government should definitely not allow the lag to become shorter than the average lag in the gathering of tax revenue. If the lag were shorter, a large increase in inflation becomes self-fuelling: real sums are added to the economy, and the same nominal sums are only later taken away.
10. Hence a shortening of the lag would bring no benefits, and under some (hopefully unlikely) scenarios, could cause damage.
11. For the reasons explained above, the shorter lag of the Canadian design has no benefits but does have potential disadvantages.
12. However, if there must be a new design of ILG, and it must have a lag as short as that of the Canadian design, then the DMO should choose the Canadian design.
13. Usage of an already-existing design would minimise system costs.
14. And further, investors are already familiar with the Canadian design.
15. If there must be a new design of ILG, and if the UK is not in EMU, then there would be merit in incorporating a deflation floor.
16. A par floor in ILGs would have a stabilising effect on the economy in the event of sustained deflation, by causing large real payments to the private sector from the government (which should print the money).
17. However, if the UK were in EMU then the stabilising benefit of that floor would be spread over the whole eurozone, but the UK alone would be paying. Further, it would not be possible for the UK government to pay for the floor by printing money.
18. The above reasoning holds even if the markets were now to value that floor at zero (and this is the most likely market valuation).
19. Given the risk that the UK might join EMU during the life of an ILG with such a floor, the balance probably tilts against such a floor.
20. The gilt strips facility is little used, and the TIPS strip facility has been almost entirely dormant.
21. Hence an ILG strips facility is extremely unlikely to work.
22. If it is possible to accommodate an ILG strips facility at no cost, then a new ILG design should be made compatible with subsequent strippability. However, the facility should not be activated until the market demand is clear.
23. The current design of ILGs is not readily compatible with a strips facility, and hence currently existing ILGs should not be strippable.
24. If the DMO chooses to adopt the Canadian design, secondary-market trading should be in terms of real price, as the DMO proposes.
25. Because the DMO proposes to choose the coupon before an auction, the auction should also be in terms of real price.
26. If the DMO retains the current design, changing the trading basis would have no advantage and some systems cost. If the design remains unchanged, so should the trading basis.
27. Further, without a redesign, changing to a real basis would cause considerable complication on days on which RPI is released, as the settlement instructions would then depend on the time of the trade.
28. This risks further fragmentation of the market: such a programme should only be attempted at the express wish of the institutional investors.
29. The DMO is right to advocate replacing the early-redemption clause with an equivalent-index clause. However, I prefer a slightly different wording:
30. In the event that the General Index of Retail Prices (RPI) shall cease to be published, or in the event that it is likely to cease to be published, an appropriately rebased successor index shall be used. This successor index shall be an index as similar in nature to the RPI as is reasonably practicable, and shall be chosen by the Chancellor after consultation with a body that the Chancellor considers to be independent and to have relevant and recognised expertise. The selection of the successor index by the Chancellor shall be binding.
31. The DMO’s wording included the phrase “continues the function of measuring changes in the level of retail prices”, which is unnecessarily broad, and in my opinion provides too little guidance to the Chancellor or to investors.
32. The above phrasing also allows the switchover to take place before the RPI is actually switched off: if the switchover is happening because of wayward behaviour of the RPI, the Chancellor may welcome the added flexibility.
33. The nature of the body’s expertise is slightly altered, becoming the more general “relevant”.
34. The Chancellor’s decision must be binding on everyone, stockholder and derivative trader alike.
35. The DMO is right to continue issuing semi-annual securities, for the reasons in the consultation paper.
36. Fixed-income systems already have a field for the rounding precision of ILGs, so there will not be any difficulty with six decimal places of rounding, with or without a redesign.
37. ILGs of a new design should have their own family of indices. However, FTSE should create three families of index-linked indices: Old-Style ILG, New-Style ILG, and All ILG. The All ILG index should be deemed the successor to the existing ILG indices.
Julian D. A. Wiseman, October 2001
Postscript. The UK DMO announced on 15 January 2002 (local copy, press release, of which local copy) that it had “decided not to introduce a new design of index-linked gilt in respect of the main issue of whether to move from the current eight-month indexation lag to a three-month one”.
|Main index||Top||About author|