How Terra Recovers: Re-Anchoring Expectations
A flight of unproductive capital is _exactly_ what Terra needed. Terra will emerge stronger, but only if it learns the right lessons
I cross-posted this content at my content partner, the State of Stake newsletter.
The last 72 hours have seen a reckoning of the critical shortcomings of Terra, which, despite its cartoonish flaws, was and is still one of the most uniquely innovative in crypto. A combination of
extreme cross-crypto weekend volatility,
Terra’s shift from the 3pool to the 4pool on Curve (reducing overall UST liquidity on Curve),
a looming $1-2B recapitalization of the Anchor Protocol again (funded by LUNA transfers to the Anchor wallet, which then sells the LUNA in the open market in exchange for UST, to keep Anchor’s unsustainable 23% yield running), and
a quiet relaxation in Terra’s soft capital controls from $100M to $300M of daily outflow (implemented in 1Q22, but never before tested),
… crescendoed on Terra’s Black Monday, as May 9th, 2022 will now be known.
Anchor’s total deposits dropped from $14B to $6B in the span of 2 days. Most of that drop took place yesterday.
Terra’s on-chain data suggest that around $650M of UST fled the ecosystem (was “burned” to mint 18m new LUNA) yesterday. $650M of UST burning for 21m LUNA implies an equivalent LUNA offering price of $31 UST per LUNA. [update: the numbers kept changing throughout the day. Over 40M new LUNA ended up being minted]
Network congestion was also severe. Transactions on Monday afternoon that normally took seconds took 1.5-3 hours to clear. However, network congestion had largely cleared up 12 hours later.
Many will say that Terra will never recover from this loss.
In fact, if Terra learns the right lessons, Terra can and will recover from Black Monday.
Silver lining 1: Terra’s fan club.
The person who most clearly echoed my own feelings was Emin Gun Sirer, head of AVA:
[thread continues; read it all]
Sirer’s sentiments were widely shared across the ecosystem. Terra was the best hope for an unbanked algostable at scale, and probably still is. Other algostables that scale will have different versions of the same problems.
Silver lining 2: Terra kept 50%+ of its BTC reserve intact
After committing $1.5B of UST and BTC to defending the peg yesterday afternoon, Do Kwon seemingly held off on committing the remainder. This was a very good decision, because it accepted the reality that Terra’s ecosystem had grown unproductively large: you can’t stop a forest fire, fueled by an exodus of low-conviction investors, with a finite water tank.
People who mock Kwon’s BTC reserve build (“Did Kwon seriously buy BTC at 47k to dump it during a margin call at 30k?” forget that it was funded with LUNA that was probably sold, on average, in the $80s. In other words, it was a profitable trade.
Silver lining 3: Unflinching loyalty from stakers, as Terra’s broader crisis mechanisms (eg, drastically higher staking income) functioned as intended
Pedro Ojeda, a data scientist who deserves a LFG board seat for his data-driven public services rendered to the Terra ecosystem (he’s a must-follow if you’re a Terra investor), ran a death spiral simulation of Terra 45 days ago which is extremely timely today.
Terra has a unique property for its LUNA stakers.
Stakers are paid in a (UST + LUNA) / LUNA staking yield. The UST comes from transaction fees, which averaged $4.5M / per day during Terra’s “normal” days of consistent inflows. So, a LUNA staker obtains revenue from 3 sources: LUNA staking issuance, UST rewards from transaction taxes, and LUNA token appreciation via LUNA “burning” / UST issuance during periods of rising UST demand.
On abnormal days, as yesterday surely was, the Terra protocol’s transaction fees explode as directional flows exceed Terra’s soft capital control cap (~$300M). Terra normally has a .3% Tobin tax on all Terra inflows & outflows, and extremely low Terra-to-Terra transaction fees. This de facto exit tax causes a surge of UST revenue for LUNA stakers.
Terra’s Tobin taxes, however, do not apply to “off-chain” UST dumps, eg, ETH-wrapped UST on Curve or Binance. This was what I suspected was an explanation for the large spread between on-chain and off-chain UST pegs, seemingly confirmed by this TFL employee.
As UST capital flight (which causes a LUNA supply shock, making the downward volatility of LUNA very large and unpredictable) becomes more acute, LUNA stakers are compensated disproportionately more in UST for absorbing LUNA volatility: not only does their staking yield explode in LUNA terms, the part of it denominated in non-LUNA stablecoin explodes also.
This gives LUNA stakers a degree of conviction — yield in a non-LUNA asset — which assures them of at least decent ROIs if they HODL. (yes, I know, it assumes UST does not go to zero.) This is a key reason why, despite 3 heart attacks in 2.5 years (March 2020, May 2021, and yesterday), Terra’s stakers have never flinched. This is just one of many of Terra’s protocol-saving design ingenuities.
The unforced error: Anchor’s hand in Black Monday
Unfortunately, Terra’s ingenious balance of incentives was never applied to the Anchor Protocol, Terra’s main central bank. Anchor’s key design flaw — The Anchor Protocol’s unsustainable 23% APY deposit rate (not 20%), which far more hot money hot money than Anchor could ever put to work — became something that Kwon defended against the strong urgings of some of his core team. (Or so I heard.)
However, Terra will not survive unless it incorporates the clear lesson: Anchor’s unsustainably high rates attracted unsustainable UST demand, which made the inevitable perfect storm 100x worse than it had to be. Anchor’s model must permanently change. If Anchor’s model doesn’t change, the market will take it as a lack of understanding, an admission that the bears were right all along, and/or a guarantee that the same story will play out again.
Even if you take the view that Do Kwon cynically pumped Anchor with $600M to dump $3B of LUNA at inflated prices onto other LUNA buyers (basically selling LUNA at $40-90 to buy BTC at 45k-50k), the loss of stability caused by Anchor’s capital flight did more damage to Terra than any currency reserve.
Personally, I think Kwon clearly understood the flaws of Anchor. He just thought, despite a mountain of empirical evidence, that the ecosystem could find uses for aUST as fast as hot money flowed in.
aUST is the proof of deposit in the Terran financial system, the equivalent of JP Morgan going to Blackstone and saying, “Hi: We have $10 billion of deposits. (See this $10 billion of aUST as proof.) We would like to invest those deposits with you.” However, in the Anchor model, the depositor, not Anchor, receives the aUST as a certificate of deposit. The depositor can then rehypothecate that deposit to another Terra protocol or use case of his choice by putting the aUST elsewhere while having the UST APY accrue to him, as shown by the Anchor website. Over a 1-year period, the aUST appreciates in value as it captures the interest received on the UST deposit, until it is worth 1.248 UST at the end of a year:
In this way, aUST was meant to be a “Lego block of decentralized leverage” within the Terra financial system. However, in reality, aUST was not composable beyond Anchor and a few select Terra protocols, as the Anchor financial bubble grew much faster than Anchor devs’ ability to build new aUST integrations.
Furthermore, because Anchor’s APY was so high (24.8% apparently, not the oft cited APR of 19.5%), Anchor actually sucked capital away from productive uses to be looped back into Anchor itself, so that parasitic capital could leverage its parasitism up multiple times, instead of routing itself to other, differentiated use cases. This was Anchor’s second unforced error: Not only did Anchor’s yield create an unsustainable cost of capital for the ecosystem, it even fed on itself, at the expense of other protocols, to leverage its insanity up.
As Anchor’s deposits-to-loans imbalance blew out to $14 billion of deposits for $3 billion of loans, Anchor’s need to curb this behavior became very obvious, and was loudly debated on Anchor’s own forums. Anchor did finally act on it, but its action (walking rates down 150bps per month to a 15% floor) was a very weak half-measure, even under much less trying circumstances.
Anchor should cut its deposit rate to 10% immediately, and remove the floor deposit rate completely, to bring Anchor into financial viability ASAP. This would have several immediate positive effects.
(++) Radical times, radical solutions. It would kill the demand for “ponzi” UST at the source, showing that Kwon correctly diagnosed the source of Terra’s hot-money troubles and was implementing a radical long-term solution.
(++) No more Anchor reserve recaps. It would eliminate the LUNA supply overhang of another, much larger Anchor Protocol recapitalization.
(+) No more degens. It would expel most of the remaining unproductive Anchor deposits from the protocol.
(+/-) It would cause some final UST outflow (and pressure on LUNA short-term). However, further capital flight of hot money would indirectly benefit the Terra protocol, as that money would probably pay the punitive exit tax.
Anchor should also expand cashflow-positive, i.e., real-world asset (RWA) integrations, without compromising its KYC. (I outlined a theoretical path to doing so in last week’s newsletter, although that path will be very difficult operationally.)
The Anchor team has avoided RWAs (I suspect for KYC-related reasons), instead placing their hope in expanding the “bonded asset” (bAsset) model beyond LUNA to ETH, AVAX, SOL, and ATOM, which yielded negligible assets. This model is also not economical for assets whose staking yields are 25% or less of the Anchor Protocol deposit rate (eg, SOL), and would be utilized less still if Anchor didn’t subsidize bAsset borrowings with ANC issuance.
RWA integrations are much more challenging than digital asset integrations. However, RWAs promise real world cash flow.
A functional Anchor Protocol would offer an a UST deposit rate of 5-7%, and a range of low-volatility aUST options for users to further enhance their yield — by putting their capital into productive risks. Anchor depositors, by taking on some additional risk, could still achieve 10-15% risk-adjusted deposit returns while the Anchor Protocol is actually profitable.
However, Terra/Anchor has not shown any of the interest in RWAs shown by competitors such as Frax, TrueFi, and especially their arch-nemesis, MakerDAO. With the probable loss of degen interest, my sincere hope is that Anchor takes real-world assets seriously despite the many extra regulatory and tech complexities they add to the equation, to place a greater weight on yield sustainability for the Anchor protocol going forward.
I distinctly remember Do Kwon, with his usual swagger, tweeting in response to someone pessimistically forecasting an imminent $1B Anchor bailout, “Seems low.” I can’t find that tweet now. Did he delete it? Potentially big, if true...
$6 billion of UST exited Anchor in the last 2 days. $375m was dumped onto the 3pool. $1.2B was formally burnt, dumping 40M new LUNA supply onto the market at a $3/UST average price. [edit: I overestimated the 4pool’s size by an order of magnitude, it’s much smaller than I thought]. There is a massive amount of UST still sitting at cexes waiting to be burned.
Do Kwon’s reticence to use his remaining BTC reserve implies that he’s worried that Curve stranded UST will be burned-and-returned as newly issued LUNA supply — which would raise the LUNA issuance quantity as UST redemptions continued.
I speculate that off-chain UST is dumping harder because off-chain UST would rather dump to Ethereum/Avax/etc tokens than pay the (currently astronomical) exit tax incurred when burning UST on-chain. If my speculation is right, this is a systemic issue that Terra needs to fix.