|Main index||Financial Markets index||About author|
Julian D. A. Wiseman
Abstract: Reply to the Mexican Ministry of Finance and Public Credit’s questionnaire about a new 30-year Mbono.
Contents: Introduction, pricing, liquidity conflict, maximum size, coupon, cashflow structure: 182 days or six months?, summary.
Publication history: sent to the Ministry in August 2006, and published on jdawiseman.com in September 2006. Usual disclaimer and copyright terms apply.
|This page closely follows the essay submitted to the Mexican Ministry of Finance and Public Credit in reply to its consulation about a new 30-year Mbono. A disclaimer, as well as giving some biographical information, said that the “author works for The Rohatyn Group, a $2bn emerging market hedge fund located in New York. The views are solely those of the author, and not necessarily those of The Rohatyn Group or any of it Principals or employees.”|
The Mexican Ministry of Finance and Public Credit has announced that it will issue a 30-year Mbono, and has circulated a questionnaire. This short essay addresses some of the questions therein, and some other matters which hopefully will be of interest to the authorities.
At what yield would a 30-year Mbono trade? We extrapolate the MXN yield curve using a smooth-forwards model, which prices government bonds and basis-adjusted TIIE swaps, with some sharing of information between the short-dated parts of the curves. This smooth forward model currently has the fair value of a par Nov2036 at +13bp over the 10%Dec2024, and a Nov2039 at +14bp. But long bonds typically trade a little rich to smooth fair value: with the rest of the curve at today’s level, a Nov2036 would probably trade between +5bp and +10bp over the Dec2024.
Long bonds suffer from a liquidity conflict. Investors want liquidity: investors want some reassurance that the bonds can be sold at short notice for modest dealing cost. This effect is seen most strongly in the US, in which the 4½% Feb 2036 yields about a quarter of a percentage point less than implied by a smooth forward curve.
Imagine that a long-term investor turns over the portfolio once a year, and that there are two bonds, similar in every way except that one bond has a dealing bid-ask spread 10¢ wider. Buying and then selling the less liquid bond will cost 10¢ per year more, and so would be worthwhile only if it yields at least 10bp more. Of course, the US 30-year tends to be owned by the market participants with the quickest turnover times and the strongest need for liquidity, the longer-term holders tending to prefer the off-the-runs. But nonetheless the United States Treasury is being well rewarded for the liquidity of the long bond.
But there is a conflict. Although investors want liquidity, they don’t make it. Buyers of long bonds tend to be buy-and-hold type investors, and where data is available, turnover ratios of 30-year paper tend to be lower than that of shorter paper.
So it is particularly important that government issuers concentrate issuance of long paper in few securities. There is no point issuing a 30-year, and a year later issuing a new 30-year. The liabilities that can be hedged with a 2037 can be just as well (or just as imperfectly) hedged with a 2036, and splitting the liquidity over two securities widens both bid-ask spreads, transferring wealth from the taxpayers to the shareholders of the large investment banks.
Hence Mexico is urged to choose one date, and only one date, and having chosen that date, stick with it. It can be 2036, as the US government issues an exactly-30-year security; or it can be 2039 as the UK, Canada, Germany, Netherlands, France, Italy, and Spain have issued their ‘30-year’ with 32 or 33 years to maturity. Investors would be happy with either date. Reopen that security for several years, perhaps even for half a decade, until it is huge and deeply liquid. But once the new Mbono exists, for the next decade there would no need for any other Mbono maturing between the new maturity date and four or five years later.
How large can a new 2036 Mbono be? Two forces push in opposite directions:
Larger — investors pay for liquidity (or at least punish illiquidity);
Smaller — governments dislike having to make a single giant payment, for fear of the budgetary and financial distortions this might impose.
So how large would be safe for the Mexican government? The largest Mbono yet redeemed was the 9% Dec 2005, at MXN 34.34bn; but let us take as the benchmark the 9% Dec 2007, at almost MXN 64bn.
If the Mexican government were to take this as a ceiling in real terms, then a further 29 years of 3% inflation allows a factor of 2.35, permitting an issue size of MXN 149bn.
Or the Mexican government might take this as a ceiling measured as a proportion of GDP. Let’s make very conservative assumptions: inflation is 3% for the next five years and 2% thereafter; and future real growth is a mere 2%. Under this (very pessimistic) scenario, MXN 207bn would be equivalent to MXN 64bn in 2007. (And assuming 3½% real growth allows an issue size of MXN 315bn!)
So the new Mbono can safely be huge relative to the existing debt securities. (A similar argument was made to the UK DMO in reply to its consultation about a 50-year gilt.†)
Most governments issue new securities with a near-par coupon, with the exception of the new South African 6¼% March 2036, first auctioned at a yield of 8.26%! It seems best that Mexico follow general practice by issuing with a near-par coupon. Not doing so might have OID tax implications for some US investors.
Mbonos are different to all other government bonds: they pay every 182 days, not every six months. This system has the advantage of paying on Thursdays. But no other market does this: the UK’s 4¼% 2055 pays every 7th June and 7th December; the French 4% 2055 pays every 25th April; and the US 4½% 2036 pays every 15th February and 15th August. Given that the UK and France have issued 50-year bonds with calendar payments, one can presume that they will not be switching to the Mexican ‘lunar’ system in our lifetimes.
Mexico might wish to maintain its distinctive 182 day system. Or Mexico might wish to adopt others’ general practice. We prefer the switch, but whichever Mexico wants, the decision must be made now. Issuing a 182-day 30-year Mbono will make any such change in the next third of a century much more difficult.
The new 30-year does give the Mexican government the opportunity to harmonize with the rest of the world, but—à la Saint Augustine—not yet. The new Mbono could pay every 182 days until Thursday 5th Dec 2024, and from that date pay every 5th December. This lengthens the maturity by only 15 days, from Thu 20th Nov 2036 to Fri 5th Dec 2036. It also tells the market that Mexico will be adopting the standard form for fixed-income markets from 2024 onwards. It would be a radical step, but a radical step with 18 years notice, and hence Mexico should consult with the market prior to committing to it. Before being locked into 30+ years of being different, such a (delayed) reform should certainly be considered.
Issue only one long Mbono.
Re-issue until it becomes very large.
Consider the merits of switching, eventually, to bonds that pay on the same calendar dates.
Julian D. A. Wiseman, August 2006
† My reply to the UK’s consultation is at www.jdawiseman.com/papers/finmkts/long-consultation.html which contains the observation that a 50-year UK gilt could be £135bn (≈MXN 2.8 trillion) and still be the same proportion of tax revenues as a mid-sized gilt today.
|Main index||Top||About author|