How to Destroy the Collective Action Clause

Mark Weidemaier & Mitu Gulati

We almost hate to post this, because it is so simple, and so fundamental, that it seems almost surely wrong. But if it’s wrong, we can’t see why. Maybe a reader can explain? Here goes.

For at least 20 years, reform efforts in sovereign debt markets have promoted collective action clauses (CACs) (here and here). The current version of the clause was drafted by a super-committee of senior lawyers, investors, and finance ministers – many of them people for whom we have enormous respect. It lets the sovereign bond issuer hold a restructuring vote across multiple series of bonds in a so-called aggregated vote. Before, most CACs in the market required a vote for each series of bonds. The point of the reform was to make it impossible for litigious holdouts to exclude one or more individual series of bonds from a restructuring that had garnered the support of a creditor supermajority. But—and here’s the important point—outside of the euro area, these aggregated CACs are reserved for bonds issued under foreign law. They don’t have to be. But contract reform to solve the holdout problem hasn’t seemed important for bonds governed by local law, which the sovereign can already restructure just by changing its law.

Most sovereigns issue most debt under local law. So, here’s the CAC destroying idea:

Phase 1, the sovereign restructures its local law debt (either by passing legislation or by asking bondholders to tender). The restructured bonds might or might not include new financial terms. What they definitely will now include is a modification provision substantially similar to the one that appears in its foreign law debt. However, the restructured bonds are still governed by local law.

Phase 2, the sovereign proposes a restructuring of the entire debt stock, aggregating the vote of local and foreign law bonds together.

Holders of local law bonds are pliable. Many will be local institutions subject to the government’s sway. Even if not, they will be pliable because they hold local law debt, and the sovereign can put the screws to them whether they consent or not (the so-called “local law advantage,” see here, here, and here). So they will vote in favor of the restructuring. If there is any doubt, the terms can be sweetened slightly, relative to the new financial terms, if any, included in the phase 1 restructuring. Holders of foreign law debt will have their vote diluted—in some cases, utterly swamped—by the favorable local-law vote, and the issuer can impose restructuring terms on everyone.

Best we can tell, nothing in the new aggregated CACs prevents this. It seems the only requirement for a bond to be included in the aggregated vote is that it satisfy the definition of “Debt Securities Capable of Aggregation.” For instance, here’s the provision in Sri Lanka’s newer bonds allowing it to conduct a fully aggregated (“single limb”) vote:

In relation to a proposal that includes a Reserved Matter, any modification to the terms and conditions of, or any action with respect to, two or more series of Debt Securities Capable of Aggregation may be made or taken if approved by a Multiple Series Single Limb Extraordinary Resolution or by a Multiple Series Single Limb Written Resolution...

...

A “Multiple Series Single Limb Written Resolution” means each resolution in writing (with a separate resolution in writing or multiple separate resolutions in writing distributed to the holders of each affected series of Debt Securities Capable of Aggregation, in accordance with the applicable bond documentation) which, when taken together, has been signed or confirmed in writing by or on behalf of the holders of at least 75% of the aggregate principal amount of the outstanding debt securities of all affected series of Debt Securities Capable of Aggregation (taken in aggregate).

And, crucially, here’s the definition of Debt Securities Capable of Aggregation:

“Debt Securities Capable of Aggregation” means those debt securities which include or incorporate by reference the provisions described in “Modifications and Amendments; Meetings of Holders” and “Aggregation Agent; Aggregation Procedures” or provisions substantially in these terms which provide for the debt securities which include such provisions to be capable of being aggregated for voting purposes with other series of debt securities.

We don’t see how these provisions prevent a restructuring like the one we have described above. A local-law bond retrofitted with a substantially similar modification clause seems to be a Debt Security Capable of Aggregation. (To be safe—since the security must “include or incorporate by reference” the voting provision—the issuer might want to do a tender offer rather than simply pass a law inserting the CAC into its local-law debt.) Nor can we find anything else in the contract to prevent this. But the new CACs are so looooooooooooong. And complicated. Maybe we are missing something obvious. It would not be the first time.

Now, is this plausible? Our initial thought is that an issuer would have to be feeling pretty frisky to try something like this. Wouldn’t investors go beserk? On the other hand, why should they? This isn’t a case of finding some ambiguous clause and asserting a new meaning for it (remember pari passu?). This clause is in plain sight. Hard to see any court finding what we have described in phases 1 and 2 as a violation of good faith. Do we think it is sneaky? Maybe a little. But we are also a bit embarrassed that we didn’t see the implication of this language till now. It is staring at us in black and white.

Bottom line, if we are reading the language right, the new CACs have a gigantic exploit built in, one that basically swallows the CAC whole. Issuers like Sri Lanka (or Russia), with large local-law debt stocks, may have a LOT more leverage than they think. Indeed, given this clause, one might ask: Would a debt restructurer be duty bound to at least advise their client of this option?

edit: We should point out that Sri Lanka used the template associated with English law bonds. The aggregated CAC in NY law bonds is a bit different, but still doesn't seem to forbid the method proposed above. Here are relevant parts of a 2018 NY-law bond issued by Uruguay:

A change to a reserve matter, including the payment terms of the debt securities of any series, can be made without your consent as long as the change is approved, pursuant to one of the three following modification methods by vote or consent by:

• • in the case of a proposed modification to a single series of debt securities, the holders of more than 75% of the aggregate principal amount of the outstanding debt securities of that series;
  • where such proposed modification would affect the outstanding debt securities of any two or more series issued under the indenture, the holders of more than 75% of the aggregate principal amount of the outstanding debt securities of all series affected by the proposed modification, taken in the aggregate, if certain “uniformly applicable” requirements are met; or
  • where such proposed modification would affect the outstanding debt securities of any two or more series issued under the indenture, whether or not the “uniformly applicable” requirements are met, the holders of more than 662/3% of the aggregate principal amount of the outstanding debt securities of all series affected by the proposed modification, taken in the aggregate, and the holders of more than 50% of the aggregate principal amount of the outstanding debt securities of each series affected by the modification, taken individually.

Any modification consented to or approved by the holders of debt securities pursuant to the above provisions will be conclusive and binding on all holders of the relevant series of debt securities or all holders of all series of debt securities affected by a cross-series modification, as the case may be, whether or not they have given such consent or approval, and on all future holders of those debt securities whether or not notation of such modification is made upon the debt securities. Any instrument given by or on behalf of any holder of a debt security in connection with any consent to or approval of any such modification will be conclusive and binding on all subsequent holders of that debt security.

Now, there is also language in the indenture itself that arguably limits the bonds eligible for aggregation to bonds issued under that indenture. In that case, the NY law bonds offer investors some protection against this strategy. But even that is unclear, and we haven't yet located similar language in the Sri Lankan bonds.

Comments

[Tasos Kanopoulos]

Dear Professor Gulati and Professor Weidemaier,

I am a third-year student at Fordham University School of Law and a big fan of your Clauses & Controversies podcast!

I revisited the ICMA CACs after reading this blog post and listening to Episode 70 (dated April 18). Although I have not managed to find any provisions that would prevent such use of the ICMA CACs altogether, there appears to be some language that may limit its practical application considerably.

Specifically, the definition of "debt securities" only covers "notes, bonds, debentures or other debt securities...with an original stated maturity of more than one year." Using Sri Lanka as an example, all treasury bills would thus be excluded from any aggregation - notwithstanding a retroactive insertion of such provisions into the contract. According to the Central Bank of Sri Lanka, as of March 31, 2022, the aggregate face value of those T-bills was approximately 2.7 trillion rupees (LKR), or $7.7 billion.

As of March 31, 2022, the remaining debt stock (excluding a 'de minimis' amount of rupee loans) had an aggregate face value of approximately $34.3 billion. Of the foregoing amount, the aggregate face value of debt securities governed by local law was approximately $23.8 billion (i.e., just over $22 billion, or 7.7 trillion LKR, in T-bonds, and just under $1.8 billion in SLDBs).

In other words, even if one assumes that every dollar of "aggregate principal amount" of local-law bonds agreed to vote in favor of an 'abusive' single-limb cross-series modification, at least in the Sri Lankan example, this would still fall short of the 75% supermajority required for such a modification to be adopted. That having been said, it is still somewhat of a close call (i.e., if the eligible debt securities have an aggregate face value of $34.3 billion, 75% of that is $25.7 billion - a mere $2-billion cushion in Sri Lanka's case).

In any event, it all makes for very interesting discussion!

Regards,
Tasos