Mark and Mitu teach courses on sovereign debt at the Universities of North Carolina and Virginia respectively, and host the podcast Clauses and Controversies. They want to also thank Leah Berger, Heream Yang and Rishabh Sharma for their research.
Pakistan is in deep debt distress. It was already in trouble as a result of Covid expenditures and political mismanagement. But the recent floods made things dramatically worse. A debt restructuring is therefore likely, if not imperative (protestations of government officials notwithstanding).
Some relief has come in the form of IMF and ADB financing, but, as the market recognises, that additional debt is just making the hole Pakistan will need to dig itself out of even bigger. Pakistan’s credit rating is in junk land and its bonds are trading at deep discounts to their par value.
We’ve been looking at the fine print of Pakistan’s foreign currency bonds. Initially, we were interested in the sukuk bonds, since we’ve never seen a sovereign sukuk restructured. But we stumbled across a plain vanilla sovereign bond that strikes us as anything but vanilla.
Pakistan’s 2024 dollar bond — issued in 2014 — has at least two odd characteristics that could spell trouble in a restructuring.
The first odd aspect of the bond involves the vote required to pass an Extraordinary Resolution (for example, to alter the issuer’s payment obligations).
Before 2003, sovereign bonds often had onerous vote requirements to change payment terms, often requiring creditors’ unanimous approval. Starting in 2003, the market shifted to allow payment terms to be modified by a vote of 75 per cent of each bond series. Then, in 2014, the market shifted again to allow the issuer to hold restructuring vote across multiple series of bonds, with approval requiring an affirmative vote of 66.67 per cent of the entire group, plus 50 per cent of each affected series. All of these changes were intended to facilitate restructuring and to minimise the threat posed by holdouts.
Pakistan’s 2024 bond seems to reveal a total lack of awareness of all of these developments in the wider market.
It defines Extraordinary Resolution as “a resolution duly passed by not less than 90 per cent of the votes cast.” Ninety per cent? In 2014, when this bond was issued, the entire market was shifting to facilitate restructuring, and there was no sign that issuers suffered a pricing penalty for using a lower voting threshold. So why issue what has to be one of the most holdout friendly sovereign bonds in existence?
This is a billion-dollar bond trading at a deep discount. It will not take much to buy a position large enough to block a restructuring of this security. There is a smaller bond, with face value of $300mn due in 2036, that also appears to have a 90 per cent voting threshold (although the Offering Circulars are not quite identical).
But this is not the end of the weirdness.
The Offering Circular to the 2024 bond supposedly reprints the actual text of the underlying Agency Agreement (which we haven’t seen). It defines an Extraordinary Resolution as “a resolution duly passed by not less than 90 per cent of the votes cast.” Every other sovereign bond we have seen defines the voting threshold as a proportion of the “aggregate principal amount” of outstanding notes.
That makes sense. Someone with $100mn in bonds should count more, for voting purposes, than someone holding $1,000. But the 2024 bond just refers to “votes cast.” So does the investor with a $1,000 stake count the same as the investor with $100mn? Weirdly, the Offering Circular for the 2036 bond that we mentioned earlier doesn’t have this “votes cast” language. Perhaps it is in the Agency Agreement, which we don’t have.
If a restructuring requires the support of 90 per cent of votes cast, it introduces even more uncertainty into a restructuring. In theory, a group of investors with small holdings could block a restructuring, even if Pakistan has a bunch of big bondholders in its corner. That is a risk that most sovereign restructurers do not have to consider.
One can also imagine circumstances where the “votes cast” language works to the disadvantage of a big prospective holdout (such as where over 90 per cent of votes cast support the restructuring even though the yes votes together comprise only a small part of the outstanding debt). This uncertainty benefits no one.
At first, we assumed the “votes cast” language was a typo. Literally two sentences above that language, in a provision describing procedures for giving electronic or written consent to a restructuring, the Offering Circular explains that consent by “90% in aggregate principal amount of Notes . . . will take effect as if it were an Extraordinary Resolution”.
Surely, we thought, the method for tallying the vote would be identical across all of these contexts. One of our friends, a veteran of this business, tells us that it might not be so. Apparently, a long time ago, for issuances held by only a few bondholders who would all show up to meetings in person, the agreements would allow for voting either by a show of hands or by a vote according to aggregate principal amount holdings. The vote was presumed to be by show of hands, unless the meeting’s chair or a 5 per cent holder requested a “poll.” And in the poll, $1,000 would represent one vote.
We don’t have the Agency Agreement to see whether it has this antiquated definition of “votes cast.” The Offering Circular does not capitalise “votes cast,” which implies that this is not a defined term. The Offering Circular also purports to reprint “the text of the Terms and Conditions of the 2024 Notes,” so one might assume it includes all important terms that an investor would need to know.
However, the bonds are issued “subject to” the Agency Agreement, so perhaps there are additional details there to clarify the voting process. Whether those provisions should bind investors who were never given access to the Agency Agreement is a separate legal can of worms.
One way or the other, we do not envy the lawyers who may have to figure out how to restructure the 2024 bond. And the 2036 bond doesn’t look much easier.