El Salvador announced plans for a partial buyback of certain of its foreign bonds yesterday, the 7.75% 2023s and 5.875% 2025s, in an effort to stave off default and save some money. The two bonds have a combined amount outstanding of US$1.7bn nominal – comprising US$800mn of the '23s, which mature in January, and US$800mn of the '25s (plus another US$113mn in a different series) – although distressed prices mean their market value is much lower. Crucially, the buyback will take place at market prices at the time, according to President Bukele.
El Salvador bonds rallied across the curve on the news, with the '23s and '25s seeing the biggest gains on the day. The '23s jumped 10pts, from US$76 to US$86 (and are up 20pts over the past two weeks), while the '25s rose by even more, up 15pts from US$36 to US$51 (and have nearly doubled in price over the last two weeks), on an indicative mid-price basis on Bloomberg (based on closing prices), although it is not clear how much traded and in what size. The belly and long end of the curve was up 3.5-6pts.
The news comes after the failure to launch the planned US$1bn "Bitcoin bond", whose proceeds could have been used to repay the looming 2023 maturity, left the country with limited alternatives. The bonds have continued to struggle as a result, in the shadow of default, while economic prospects diminish.
However, details on the buyback are scant. President Bukele said two bills had been sent to Congress, where his party holds a two-thirds majority, to facilitate a “transparent, public and voluntary” purchase, according to media reports. The transaction is expected to start in six weeks. Proceeds would come from El Salvador's SDR general allocation, multilateral funding and a local bond issue (issued to the central bank).
Moreover, the size of the buyback is not yet known. Media reporting suggests the government is talking about all the bonds, although we doubt it has the resources for this, while observers say it will be cUS$500mn-600mn in size (which may only be enough to retire most of, but not all, the '23s, depending on the price; for instance, we think US$600mn cash consideration would have been enough to purchase nearly US$790mn nominal of the '23s at a pre-announcement price of US$76, but it would be lower now given bond prices have risen). Nor is the proposed allocation across the two bonds known. We presume it would make sense to take out as much of the '23s as possible, being the more pressing maturity, although a partial buyback of the '25s would save some two years' interest (at 5.875%).
We make three further observations:
While taking advantage of distressed prices through a buyback may make sense (and other sovereigns may be thinking the same), the very announcement of it, and that it will take place at market prices, is likely to drive prices against it and make it more expensive for the government (ie it will have to pay more than it otherwise would/save less than it could). We reckon it has already cost it an extra US$90mn. This will reduce the operation's effectiveness.
This is the classic "Bolivian buyback boondoggle”, coined by Rogoff and Bulow in the 1980s, based on Bolivia's buyback experience, where the very act of publicising the possibility of a buyback drove prices up and gave Bolivia a poor deal. Better to be discreet. Alternatively, cost-effectiveness could be maintained through an auction design, eg through a modified Dutch auction (à la Ecuador 2009), where bondholders can tender their bonds at a price higher than the minimum clearing price set by the government (and hope their offer is accepted) but all bondholders receive the same final clearing price (ie a uniform price auction).
The other issue is the incentive to participate, and the prisoner's dilemma. If it is voluntary, as the government has said, holders may have the incentive not to participate, hoping that others will, so that they come to represent such a small rump of the remaining bond that it is not worth the government defaulting on them (ie holdout for par). However, recognising the value of non-participation increases, if everyone does this (or most) this could scupper the whole deal and leave everyone worse off (ie heading to default). This means the buyback (here: market) price would need to increase. But if the buyback price is forced up too high, it increases the cost to the government and may make it not viable.
Although a successful buyback may help avert a near-term default (and, in El Salvador's case, would ease the amortisation burden through to 2027), it does not guarantee debt sustainability. This, instead, would depend on a credible medium-term fiscal strategy and structural reform agenda, which is still sadly lacking. Hence, the rest of the sovereign curve may remain under pressure.
We maintain our Hold on El Salvador bonds, with the 2032s indicated at US$38.2 (yield 26%) as of cob 26 July on Bloomberg (mid-price basis), although we recognise they have fallen substantially (-35%) since we last reiterated our Hold in January.