ECB + CACs: Fig Leaf Aflutter

Further to Mitu's post about the European Central Bank's bond-buying bellyache, let us linger on the rationale for the 33.33% limit on the central bank's holdings of a euro area sovereign bond series. 

In the middle of the Greek sovereign debt crisis, euro area policy makers agreed to adopt functionally identical collective action clauses (CACs) in member state sovereign bonds, with two amendment options for important financial terms: (1) series-by-series voting or (2) aggregated voting, pooling two or more series. In a series-by series vote, two-thirds of the principal amount outstanding can override opposition by a third. In an aggregated vote, amendment requires at least two-thirds of the total principal amount outstanding in the voting pool plus at least half of each series in the pool.* 

The ECB has taken the position that "in the context of a restructuring subject to CACs, it will always vote against a full or partial waiver of its claims" to comply with the European treaty framework. This means that, if the ECB held more than one-third of a bond series, it would single-handedly block a single-series restructuring. If the ECB held more than half of a series, it could force it to drop out of an aggregated restructuring as well. To avoid being the holdout, the ECB adopted the 33.33% bond purchase limit in 2015 (see Article 5 of this and this excellent piece by Sebastian Grund). Of course, no matter how little sovereign debt it holds, the ECB makes life easier for other holdouts with its pre-commitment to vote against restructuring -- the others need to buy that much less to free ride.

It is unclear whether the 33.33% series limit was motivated by the mere possibility that a government would restructure its debt by written consent* using series-by-series CACs--or a view that this approach was most likely, or even required. Perhaps 33.33% was the the lowest plausible limit, since it is hard to predict how pooling and aggregated voting would work in any given case. Who knows? The only certainty is that in 2015, the ECB voluntarily subjected itself to some oddly-reasoned bond buying caps, and that today, these could knee-cap Italy.

The limit makes even less sense in light of the 2018-2019 commitments by euro area policy makers to adopt so-called "single limb" aggregated stock-wide voting.  Under the latest market standard (due to take effect in the euro area in 2022), a government need not even poll individual series. Now add the likelihood that a country like Italy would restructure using domestic statutes, rather than CACs, as explained in Mitu's post and his article with Ugo Panizza, and 33.33% begins to look mighty random. 

To be sure, it is politically and maybe legally awkward to say that a government would ignore CACs and use domestic law in a restructuring after pushing those clauses as the be-all for years.  It would be more awkward still to say that a government would ignore contracts, statutes, and any other law that might get in its way in an emergency, because aside from peacetime liability management operations, most sovereign debt restructurings could make out a pretty decent case for emergency/necessity rule.

But even if you assume that all euro area sovereign bonds have CACs (they do not) and that they would use them in the unlikely event of a restructuring (they would not), a 33.33% ex ante purchase cap makes little sense. It does not prevent the ECB from helping holdouts, or even from being the main holdout. It does not protect the ECB from losses in the event it is outvoted or suffers a default. It looks like a skimpy formalistic fig leaf covering up for real problems in the underlying treaty framework and short-sighted pre-commitments, at the cost of potentially impeding the ECB's monetary policy efforts at a critical time. The central bank (and the world) would be better off if it were disenfranchised altogether or at least confined to a separate voting pool. Besides, disenfranchisement and confinement are so au courant.

*The thresholds are different for votes taken in a meeting (25%+) and by written consent (33.33%+).