|Main index||Financial Markets index||About author|
Julian D. A. Wiseman
Abstract: Individual government debt securities are sometimes subject to a ‘squeeze’, in which the security becomes very expensive to borrow. This can be caused by one or some market participants gaining control of most of one security, allowing these participants to extract a ‘monopoly rent’ from others. This is a description of a repo-switch mechanism which would enhance the efficiency of government debt markets by deterring would-be squeezers.
Publication history: This document closely follows the text of a letter sent to the Bank of England on 15th August 1996. This letter was sent in the professional capacity of the author, as an employee of a Gilt-Edged Market Maker. Variants of this letter were subsequently discussed with other government issuers. Usual disclaimer and copyright terms apply.
[The letter opened by saying that “Recent market behaviour has re-raised the question of the possibility of official intervention to alleviate tightness in the repo market in some specific stocks” — no less true in early 1999 than in Autumn 1996. It continued by discussing my then employer’s position in one of Europe’s less liquid strip markets. My then employer was long a particular long-dated principal strip. The most obvious hedge would have been to sell this holding — but that strip market was so illiquid that this was near-impossible for a price-maker. An alternative would have been to borrow the coupon strips, reconstitute and sell the whole bond; separately hedging the short position in the circa 10-year duration coupon stream. But the repo market in the coupon strips was so thin as to make this impossible. So instead the long position was hedged with a short position in the bond. The long of the principal was funded at GC, and the short of the bond between 60bp and 1% special, costing a fortune in carry every month, and deterring my then employer from offering to an investor any coupon strip which required stripping more of that bond. Indeed, my then employer suspected that the one other price-maker in that strip market was in a similar position. The introduction concluded by saying that the absence of reliable repo in strips explained — at least in part — why the strip market in that country had failed.]
… Hence we urge the Bank of England to consider a mechanism to prevent excessive squeezes, corners or stock-specific tightness in the gilt repo market. To which issues should this apply? It must apply to coupon strips because of their small size. Currently plans [subsequently realised] envisage coupon strips being identical to principal strips in tax treatment and creditworthiness, and there seems to be no advantage in breaking symmetry here. So such an intervention mechanism would also apply to principals, and therefore, to reduce the administrative burden, to strippable gilts. Given a suitable mechanism — which is given below — we see only advantages in extending the facility to non-strippables and non-conventionals. The Bank might wish to exclude those stocks which are so small that the Gilt-Edged Market Makers are not obliged to make markets in them. We reserve judgment on whether this facility should be extended to T-bills, at least until after the Bank has published its forthcoming consultative document on money-market reform. [Having seen both the Bank’s plans, and the modifications thereto to be implemented by the UK Debt Management Office, the facility should be so extended.]
Currently [in August 1996], with many domestic investors not geared up for repo, such a facility might be used extensively. But over time use of this facility will become less frequent. One should note that even if it is only rarely used, the very presence of the facility will help ensure an orderly repo market.
What properties should an intervention mechanism have?
The Bank should not make any discretionary choices. Any discretionary system would be seen to have wider consequences. Bond analysts would carefully correlate how special issues were allowed to become, with the choice of stock and size at subsequent auctions. Even if the Bank intended its intervention and auction decisions to be independent, the Bank would still be making intervention decisions in the context of GEMM and investor discussion of their meaning. So a discretionary system introduces wider complications which are unnecessary, and which have no obvious benefit. Given the very favourable market reaction to the transparency-increasing structural reforms of the past two years, it would be retrograde for the Bank to appear to exercise significant discretion in this form of intervention.
There should be a published absolute floor for the extent of specialness. We favour about 100bp, but the Bank might wish to be less aggressive and choose 200bp, and we assume this below.
All intervention should involve lending special stocks in return for other government collateral, not for money. This has two benefits. First, it obviates the need for a cash fixing of GC rates on which to base intervention rates. Second, it prevents this mechanism being used as a back-door to the money-markets, rather than to reduce stock-specific difficulties.
It might be that the Bank would like the market to retain some uncertainty as to the price and size of intervention. Indeed, it is with this objective in mind that some central banks retain discretion in some of the details of money-market intervention. This can be done by ensuring that the GEMMs collectively determine the size and price of intervention, rather than this being determined by any GEMM individually. Any GEMM not privy to all the other GEMMs’ positions will not know what is to happen, yet the Bank will not need to take a discretionary decision.
So how can this be done? We favour the following:
The Bank would be willing to repo out stock according to a pre-announced price formula. For each individual stock the formula would convert a requested proportion of issue size into an amount of specialness;
By 12:15 on any business day any GEMM can submit a request of how much of each stock that GEMM wants. The GEMMs requests would not mention prices, and the Bank would not accept conditional requests;
The Bank then totals up the amount of each stock that is required, as a proportion of its issue size. (To calculate the issue size of a strip assume all strippables are fully stripped, and to prevent double-counting of the issue size of the strippable stock and that of its components a request for a strippable is treated as a request for the components);
The Bank then uses its published formula to calculate the price at which the paper is lent to the market. Let x be the proportion of some particular bonds issue size that was requested, and s(x) the degree of specialness at which that stock is supplied. One possible formula goes as follows:
- Up to one issue size, the marginal cost would vary linearly from 100 to 200, and beyond one issue size the marginal cost would be 200;
- Then x<=1 implies s(x)=100(1+x/2), and x>=1 implies s(x)=100(2-1/(2x));
- This means that small quantities of stock are provided at just over 100 special, one issue size of stock (not that much for a coupon strip) at 150 special, and unlimited quantities at just under 200 special;
Alternative formulae are possible. For example, a purist mathematician might argue for a smoother equation such as s(x) = 100(2x+(e^-x)-1)/x, this having a marginal specialness of 100(2-(e^-x)). But the end-effect in terms of liquidity of the strip market would probably be the same.
The Bank then fills all requests in full, subject only to credit constraints. GEMMs who requested stock have no choice but to accept the calculated specialness, and do not have an option to refuse the stock;
The stock would be supplied overnight. An exception might be made just before the Christmas and New Year holidays — the Bank would announce this in advance;
The amount of each stock provided, and the specialness at which this was done, are then published on the newswire services and on the Bank’s internet server;
Later in the day each GEMM would non-publicly send the Bank details of the collateral to supplied in return.
This does assume, hopefully not too optimistically, that the Treasury would be willing to give the Bank carte blanche to create temporarily unlimited amounts of stocks and strips for market management purposes. We defer to the Bank’s expertise as to whether or not this would cause accounting or legal complications.
This facility provides a transparent combination of uncertainty and reassurance. During the morning market-makers know that they can access stock between 100bp and 200bp special, but do not know the exact price at which the assistance will be provided. Thus repo-trading in near-squeeze issues will continue. But because the price is decided by the street rather than the Bank, it is signal-free. Further, the Bank will run this facility at a profit!
This facility would significantly increase the probability that the gilt strip market achieves sustainable liquidity.
[Excepting the substitution of “the Debt Management Office” for “the Bank”, the author still believes this to be an excellent recommendation.]
Julian D. A. Wiseman, August 1996 and April 1999
[Postscript: On Friday 17th September 1999 the UK’s Debt Management Office published a paper (local copy) seeking comment from market participants on a possible non-discretionary special repo facility, to which a reply was sent on 20th September 1999. On 22nd February 2000 the DMO published a “Response to DMO Consultation Document on ‘Special’ Gilt Repo Operations” (local copy), announcing the introduction of such a non-discretionary repo switch facility; see also this author's commentary on the DMO's February 2000 decision.]
|Main index||Top||About author|