[Discussion] Pause the USDR/am3CRV Gauge on Polygon

Gauge Proposals
#1

Summary:

This is a discussion in connection with the recent Llama Risk report and Tangible’s response. In the report, Llama Risk raised concerns that we believe warrants a pause on the USDR/am3CRV Polygon gauge until certain criteria are met. Tangible is an up-and-coming protocol that recently applied for and received a Curve gauge (proposal here) and they wish to continue gauge emissions.

This discussion is to make a case for or against taking any action on this gauge.

Intro:

Tangible issues the USDR stablecoin backed by a combination of real estate (RE), DAI, POL (USDR/am3CRV Curve LP), its own TNGBL token, and an insurance fund of various assets. Its treasury management strategy make it a unique hybrid RWA/algorithmic stablecoin, and this blend of collateral types was selected as a method to keep the system liquid amidst the challenge of custodying a rather illiquid asset class (RE).

As part of its growth strategy, the team manages the treasury by minting USDR from system gains (example tx), minting USDR from TNGBL (example tx), withdrawing DAI to purchase RE (tx1, tx2), managing POL in the USDR Curve pool (example tx), and managing the protocol insurance fund. These processes involve a high level of trust in the team, and some unusual activity in treasury management has prompted our concern about how user funds are being used.

Curve Pool Bribe Strategy:

Tangible has been offering vote incentives on Warden Quest at 108K DAI/week for 2 weeks, and recently increased it to 162K DAI/week. This is a substantial sum, ranking it among the top 3 Curve pool bribers (currently second only to Convex’s bribe for the cvxCRV/CRV pool, notably made in its native token rather than stablecoin).

Previous round bribe compared to other bribes:
image
Source: https://twitter.com/HubertX13/status/1649465610940588033?s=20

In the latest bribe round, Tangible has sourced the DAI from 2 places:

The source of funding for the bribe manager is all traceable onchain. The bribe manager is controlled by the team and has a team allocation of TNGBL, it regularly mints USDR from TNGBL, and in this instance has swapped USDR to DAI, and bridged to Ethereum for deposit on Quest.

The Binance funds are a question that perhaps the Tangible team can clarify. We noticed large DAI withdrawals from the Treasury sent to Binance around the time that bribe funds began arriving from Binance. Each DAI withdrawal has an NFT minted with the tx that purports to represent a specific property purchased at the value of the DAI withdrawal. Tangible lists the properties with the unique NFT identifier on their website, but proof of purchase documentation appears to be lacking for all properties since the time Binance became a bribe fund source (since end of March). Tangible has rapidly expanded its real estate portfolio in the month of April (adding 32 of its 50 properties just this month), so it may be plausible that the immense workload to manage this expansion has left a backlog of paperwork to process. Tangible team perhaps can clarify on the status of documentation.

For Reference on Treasury withdrawals to Binance vs Binance deposits to bribe wallet:

To be very direct, I fear that user DAI deposits in the treasury may be, at least in part, being diverted from their stated purpose to purchase these particular properties. I am concerned that, instead, the DAI is used as bribes to fuel rapid expansion. Updated documentation for properties purchased in the month of April would alleviate this concern.

RE Purchase from DAI strategy:

I’ll elaborate more on this strategy that uses DAI deposits to purchase additional real estate properties because I did not realize this was a strategy outlined by the team. Perhaps they can point to a source that explains this use of treasury funds. (According to the docs, Tangible would mint USDR from gains in the treasury assets to purchase additional properties. DAI is used in the POL strategy, but I see no reference of using treasury DAI to buy RE).

The image below shows the total DAI withdrawn from the treasury to allegedly purchase additional properties. The InitialReseller handles the exchange of an NFT to the treasury for the value of the underlying property in DAI. Those funds are passed to the Team msig, which make their way to Binance.

Screen Shot 2023-04-26 at 2.49.00 PM
Screen Shot 2023-04-26 at 2.49.00 PM1600×610 117 KB

As an example of this process:

Since minting USDR from TNGBL is limited (and as explained below, is dominated by team minting), depositing DAI is essentially the only way end users can mint USDR. Furthermore, DAI is the only liquid asset that USDR can be redeemed against. Once all DAI is drained from the treasury, the team becomes responsible for liquidating RE in order to pay back the redemptions.

Mint USDR from TNGBL strategy:

Minting USDR from TNGBL is limited based on the proportion of assets in the treasury. The price of TNGBL takes the Uni TNGBL/DAI pool as a pricefeed with an upper bound set by the team treasury manager multisig. I haven’t seen anything that leads me to think the team is circumventing any of the limits they impose, but I have determined that the team accounts for >92% of TNGBL deposits in the treasury, mostly from their bribe manager and gov manager wallets (the former is used to fund liquidity incentives for the protocol, the latter is used to buy gov tokens such as CVX and VELO). While I can’t say for certain why the team accounts for such a high % of TNGBL in the treasury, I would speculate that it’s due to a high concentration of circulating supply in possession of the team and perhaps stronger incentives for average users to LP or lock their TNGBL rather than mint USDR.

This analysis shows flow of TNGBL to the USDRTreasury (note there have been 3 treasury implementations and the contents of the treasury are transferred from treasury1->treasury2->treasury3): Breadcrumbs - Blockchain Investigation Tool

Between the bribe manager and gov manager, they have minted 1.7M USDR from TNGBL, ~8% of USDR supply. The treasury manager can choose to mint additional USDR from the treasury in case of TNGBL price appreciation. The team say in their response to the Llama Risk report “As the report notes, the price of TNGBL went up which resulted in an expansion to this component of the backing. While new USDR was minted against these gains, the team determined that the most responsible action was to burn this new USDR, preventing future undercollaterlization if TNGBL retraced. Minting against gains to TNGBL will be deprecated in USDR v3.”

TNGBL is essentially a one-way street in the tangible treasury. While it can be used to mint USDR, it cannot be redeemed except as a last resort after all DAI and RE has been liquidated and redeemed.

The Univ3 pool used as a price reference has very low liquidity, only having $264K TVL. Most TNGBL liquidity is in the Balancer pool, which is weighted 80%TNGBL/20%USDC.

TNGBL’s inclusion in the treasury amounts to a “growth hack”. It would not take much to push the price of TNGBL up and increase USDR minting potential. The team claim themselves that the reflexive nature of this collateral type is high risk and they don’t plan on using this functionality to mint from TNGBL system gains in the future. I think that’s a great decision for the future stability of their system. I’d go further and say the inclusion of TNGBL as a collateral type AT ALL presents an undue risk to maintaining system solvency, given the illiquidity of its primary collateral type (RE).

Conclusion:

The Llama Risk report says we believe the gauge for this pool should be paused until the following criteria are met:

  1. Integrate the Proof of Reserve and RE valuation reports from a trusted third party auditor and oracle service
  2. Remove TNGBL as a collateral type

The team have said they are working with Chainlink and an independent auditor for the first point. This would alleviate concerns raised in the Curve Pool Bribe Strategy and RE Purchase from DAI strategy sections.

The team have said they plan to eliminate TNGBL mint from system gains in the future. There’s no need to include this collateral type. RE + DAI and the POL strategy are sufficient to build Tangible into a successful protocol. Endogenous collateral types are highly risky and likely not regulatory compliant. The system would be much more sustainable by eliminating this collateral type.

I stand by the recommendation in the Llama Risk report that we need to pause the USDR/am3CRV gauge until these two points have been satisfied. I would further urge Tangible to refrain from bribing the Curve pool until they have provided reasonable assurance about the use of treasury funds.

Final Note:

As a final note, I want to say that Llama Risk has been in contact with the Tangible team while researching for our report, and they have been forthright in all our dealings with them. They appear to be a team that believe strongly in the viability of their product and have clearly put a great deal of work into building Tangible into what it is today. I don’t have any reason to believe, based on my dealing with them, that there is any intent of wrong doing. It’s my hope this post is received as an objective analysis; I’m not looking to FUD this team or their project, I’m trying to lay out the facts and the relevant questions that ought to be answered. Ultimately, I stand with team Curve and all the stablecoin protocols building with it- I want to see Tangible grow to be a successful protocol for the long term.

3 Likes
#2

Thank you for taking the time to clearly present your views. We’ll attempt to share the facts which we hope will assuage your concerns and eliminate the need for a vote to remove the previously approved USDR gauge.

Summary Response:

  • Minting on gains is a collateralization strategy that has no impact on bribes. USDR minted from gains to TNGBL has only ended up in two places 1) burned 2) spent on real estate to collateralize USDR.
  • Purchasing real estate is an off-chain activity which explains the flow of funds from the Treasury to the Tangible Labs multisig to Binance. ~98.7% of all real estate backing USDR has been purchased with the DAI used to mint Real USD. Only 1.3% of the real estate has been added through minting on gains.
  • The process to purchase real estate (transfering funds off-chain) has been clearly communicated in our docs and relevant timelines have been shared with our community on an ongoing basis.
  • Chainlink Proof of Reserves (and pricing oracles) should clear up any concerns on the utilization of treasury funds that have moved off-chain, proving ownership and asset values.
  • We firmly believe that a stablecoin backed by real estate must also include a token like TNGBL to remain capital efficient. Clear limits have been placed on minting with TNGBL; there’s no ability for this to run away disconnected from a correlated organic growth to USDR market cap, 90%+ of that being backed by “real” assets.
  • TNGBL should not be treated differently than any other endogenous token used in stablecoin design. We’d argue TNGBL is good collateral, however if it’s considered insufficient, does that also apply to the other stables backed by a percentage of endogenous collateral?
  • We are transparent in our approach to supplementing bribes. However, we are also farming the gauge with POL which makes our efforts sustainable going forward. There is not enough liquidity on Polygon to create a large Curve pool without bribing.

Minting on Gains

Minting on gains comes up several times in the post above and we wanted to clarify how this mechanism works and what its purpose is.

As part of its growth strategy, the team manages the treasury by minting USDR from system gains.

Minting on gains is a collateralization strategy. It is not a growth strategy, it’s never been used to supplement growth. Nor is it the primary strategy for adding real estate to the USDR treasury (more on this below.)

Its role is, over time, to increase the percentage of the real estate backing to USDR and consequently increase the native yield as a larger percentage of the token is backed by yield-generating properties. We do this by leveraging appreciation to the volatile assets in the treasury (TNGBL and real estate), minting on those gains, and purchasing new real estate with those funds.

We view minting against gains as a relatively low risk approach to improving real estate collateralization. Here’s a quick example:

Starting at 100% collateralization with a 10% TNGBL backing, a doubling in the price of TNGBL leaves USDR 110% collateralized, 20% of that being TNGBL. We could do nothing with this, letting the price fall back and returning to 100% CR if TNGBL retraces. Or we can mint on those gains and lock the value accrual into real estate. We’ve now added 10% more real estate to the backing and even with a subsequent 50% crash to the price of TNGBL, we’re left 105% collateralized, well ahead of our initial starting point. The ability to move value accrual from TNGBL to real estate is a key design feature and user benefit. The chart below demonstrates this rough example.

image
image872×328 39.8 KB

There have been two key instances where TNGBL price appreciation has resulted in a significant amount of USDR to be minted against those gains.

March 8th:

Buzz around USDR growth drove a spike in the price of TNGBL, with ~$611k of new USDR minted against those gains. Given the scale and speed of the price increase, and the unpredictability in the stability of that price, the team decided to burn (tx) the entirety of the USDR created in that event. We shared that decision with our community at that time.

image
image1706×960 183 KB

April 27th

As referenced above, the launch of the TNGBL-USDR gauge on Balancer resulted in an increase to the price of TNGBL and $190,277.55 in gains were minted as a result of this appreciation (tx). Given the already substantial TNGBL overcollateralization, these gains were used to purchase $138,889.95 in new real estate (tx) and the remaining $86,000 of USDR was burned (tx).

As designed, an appreciation to a volatile asset (TNGBL) was “locked in,” and converted to yield-generating asset (real estate) with a predictable history of appreciation vs fiat. In no instance were any of these funds used for bribes or distributed to any additional wallets owned by Real USD or Tangible.

Purchasing Real Estate

There appears to be a misunderstanding around the real estate purchase process, the flow of funds, and how off-chain real estate purchases are funded via DAI deposits.

“According to the docs, Tangible would mint USDR from gains in the treasury assets to purchase additional properties. DAI is used in the POL strategy, but I see no reference of using treasury DAI to buy RE).”

“I’ll elaborate more on this strategy that uses DAI deposits to purchase additional real estate properties because I did not realize this was a strategy outlined by the team. Perhaps they can point to a source that explains this use of treasury funds.”

The primary source of funds for new real estate is DAI that’s ingested through new minting of USDR. Minting on gains will be an important supplement over time, however, the vast majority of new real estate is purchased with new user funds. The process for purchasing real estate with DAI has always been clearly explained in our docs.

Real estate purchases happen in the legacy financial system and as such, funds must move off-chain in order to complete this part of the process. We’ve always been completely transparent about this and the associated timelines. In the docs, we explain how funds are moved from the USDR treasury to the Tangible Labs multisig where they’re held prior to offramp to fund the eventual RWA purchase. We also regularly share the expected timelines to complete these purchases and update the property pages with the finalized documentation. (Ex1, Ex2) With the exception of the properties on West St. in The Rockingham complex, all property docs are up to date on the site.

There are three transactions listed in the proposal above that are noted as Treasury withdrawals to Binance. These are withdrawals from the Labs multisig to Binance to pay for properties. Similarly timed deposits from Binance to the bribe wallet are merely coincidental. As noted above, “Tangible has rapidly expanded its real estate portfolio in the month of April (adding 32 of its 50 properties just this month.)” This has required a regular off-ramp of funds to pay for the properties. At the same time we’ve been regular, weekly bribers on multiple protocols across multiple chains. It’s logical that funds used to supplement our bribes would be moving on-chain as frequently as other funds were moving off chain.

As noted in our original response, we’ve self-funded/bootstrapped our growth to date. At no point did we have the funds to safely pre-purchase real estate and tokenize prior to its addition to the USDR treasury. Over time we hope to be in a position of financial stability and safe, projectable growth where Tangible Labs can pre-purchase some units, so that the legacy transaction has already been completed by the time the property is added to the treasury. However, even in this situation, there will still be a flow of funds from the treasury to the Labs multisig to compensate Tangible for the assets pre-purchased for the Real USD treasury.

Ultimately, our forthcoming integration with Chainlink, including Proof of Reserves, should put to bed any lingering concerns as to whether the money moved off chain for the purpose of purchasing real estate has been used for that purpose.

Endogenous Backing

As a reminder, TNGBL has its own value accrual system separate from the yield generation in USDR. It’s not part of a circular system with USDR. Tangible marketplace fees, and soon 10% of the rental yield in Tangible Baskets, accrue to locked TNGBL 3,3+. As noted in the initial report, nearly 82% of the tokens are held in locked positions, so there’s little reason to expect a mass sell-off of TNGBL. While TNGBL is an endogenous asset, we’d argue it’s not without clear intrinsic value, it’s different than a purely algorithmic solution. TNGBL currently sits at ~1% of the backing for USDR, the rest is overcollateralization. If TNGBL went to zero right now, USDR would only be 1% unbacked, with mechanisms in place to re-collateralize (50% of the rental yield is retained until full collateralization is reached.)

As explained in our original response, we firmly believe that a stablecoin backed by real estate needs a token like TNGBL to remain capital efficient. Overcollateralizing a real estate backed stablecoin with an asset like TNGBL is critical to success. There must be a cushion to cover any fluctuations to the price of real estate while that component of the backing has yet to appreciate and provide additional collateralization. Once that point in time has been reached, TNGBL is still critical to filling any loss of value resulting from temporary market corrections in real estate.

Further, in the case of USDR, as we aim to purchase real estate in key markets across the globe, TNGBL is a critical backstop to prevent temporary under collateralization due to FX fluctuations i.e. properties purchased in Pounds and priced in Dollars are susceptible to devaluation in dollar terms if the Pound were to lose value vs the Dollar. TNGBL solves for this as well.

While some will view minting TNGBL into USDR as a growth hack, we believe it’s a critical component to maintaining protocol stability and have implemented key design limitations to prevent abuse. We’ve also established an Insurance Fund to capture value minted by TNGBL and save it in other digital assets that can be redeemed as a last resort, if 1:1 redemptions cannot be fulfilled due to a decrease in the price of TNGBL.

The argument above notes, “It would not take much to push the price of TNGBL up and increase USDR minting potential.” This is only true in the sense that Tangible has more funds to mint with over time, however it doesn’t change the fact that the TNGBL mint percentage is fixed based on market cap (10% of minted minus redeemed.) We cannot mint more USDR with TNGBL without correlated product adoption and growth. And while the liquidity in the Uni pool is viewed as somewhat thin, it’s constantly balanced against the Balancer pool, building a total of +$1MM in effective liquidity. If minting on gains were used to fund growth, this might matter more, but as it stands this automated minting from price appreciation is only used to responsibly increase the real estate backing of USDR.

TNGBL should not be treated differently than any other endogenous tokens used in stablecoin design. At the top we’ve argued why we think it’s actually reliable collateral with limited risk, much better than a purely algo approach. If TNGBL is considered insufficient collateral, does that also apply to the other stables backed by a percentage of endogenous collateral? Has there been a conversation around removing the gauges for the Frax BP, the other Frax pools and USDD?

When it comes to Real USD token backing and the risk team’s POV that “RE + DAI and the POL strategy are sufficient,” this really just boils down to a design decision where we fundamentally disagree. And a larger question on fairness, endogenous collateral being treated equally across the board.

Bribe Sustainability

It’s clear that the Llama’s goal is to ensure projects receiving a gauge are built to survive over the long term. As such USDR and Tangible protocol growth and sustainability are germain to this conversation.

USDR bribes on Curve come from two sources:

  1. Farming POL
  2. Protocol supplements

We make no secret of the fact that we are supplementing bribes. However, we are also farming the gauge with POL which makes our efforts sustainable going forward.

In the docs we state that the Insurance Fund will return to Tangible Labs once USDR is overcollateralized, 100%+ backing excluding TNGBL and the Insurance Fund. While we’re currently bribing slightly above what we’re able to farm each epoch, the sustainability question is answered by the fact that we’re also minting USDR with TNGBL to account for that difference and allocating that to Insurance Fund assets. Eventually, Tangible Labs will get that money back and any difference in bribes paid vs bribes earned will be closed, making the process profitable for us. Even if we lose a small percentage of funds on bribes at this point in time, this is offset by the eventual net gain in dollar terms of the Insurance Fund.

Lastly, a practical point on the need for bribing, there simply is not enough liquidity on Polygon to create a large Curve pool without bribing. USDR-3crv is currently 20% of all Polygon Curve liquidity and bribing is necessary for our growth, with positive consequences for the Curve Finance DAO members and Polygon liquidity overall.

Chainlink Integrations

This has been covered repeatedly over the course of this ongoing conversation. Chainlink pricing oracles and Proof of Reserves are currently in development and we hope to deploy them within the next 14 days.

As the Llamas state, “This would alleviate concerns raised in the Curve Pool Bribe Strategy and RE Purchase from DAI strategy sections.” We agree.

We’d also like to note that our work integrating with Chainlink started well before any suggestions to do so by LlamaRisk. Our original gauge proposal noted that this work was already underway. It’s always been our intent to reduce risks by decentralizing as much of the protocol as we’re able to, as part of an ongoing effort to decentralize Real USD.

Conclusion

Misunderstandings in the process used to purchase real estate, and the need for an independent third party to verify that these purchases have been made/correctly priced, should be cleared up within the following weeks if they haven’t been already. We’ve also recently added three members to the team who will optimize our approach to real estate acquisition, risk modeling and overall operations efficiency.

The outstanding issue is endogenous collateral, where Tangible isn’t the only protocol to use this type of asset in the design of a stablecoin. Our position is that it’s been done out of need, a conscious choice that we’ve implemented responsibly, with various safeguards in place to clearly cap the minting amount and provide additional backing in the event TNGBL loses value. We’re asking for fairness across the board when considering the permissibility of endogenous backing to stablecoins with approved Curve gauges.

2 Likes
#3

Thanks for the thorough response. I see that documentation for the West St. property is being uploaded today (I see the documentation for units 001-004 added so far). This building alone accounts for $3.6M, of which $550K has documentation (and I’m sure given some time the rest will soon be publically available). Along with the 5 properties added May 1st, there is ~$4.25M worth of properties that we are awaiting documentation for. As you’ve said, this is your normal process that will take some time to provide the documentation.

A robust method to provide proof of reserves and valuation will go a long way. Indeed, if this integration moves forward as you say, I expect that will alleviate the vast majority of concerns I have.

As for Endogenous collateral, you mention Frax and USDD. Frax recently voted to permanently increase protocol CR to 100% from protocol earnings. Since the proposal passed, CR has increased from 92% to 94.75%. This isn’t to say that Frax is undeserving of criticism, but their community has correctly identified that the algorithmic backing is a problem for the primary reason that it undermines the perceived safety of FRAX.

USDD launched as an algo-stable inspired by UST, and quickly pivoted to an “overcollateralized” model after it collapsed shortly thereafter. USDD is a different story in that its backing is in the custody of its Tron DAO Reserve, a group of 7 companies/organizations whitelisted by the protocol. It’s not really a decentralized stablecoin, but it’s also not really an algo-stable. It’s mostly backed by TRX and also BTC (and is overcollateralized).

The OG endogenous-backed stable that comes to mind is sUSD, which predated the aggressively capital-efficient algo stables like UST. I believe its target C-ratio is 425% right now and has always been multiples greater than the SNX value staked. USDR can be minted from TNGBL at 1 USDR: $1 worth of TNGBL. This more closely resembles the algo-stable mechanism popularized by UST, which is widely considered an unsustainable model. USDR has important differences in its mechanics, so it can’t go the way of UST, but insofar as it allows aggressive capital efficiency with an endogenous collateral type, there is an important and potentially destabilizing similarity.

I call TNGBL backing a “growth hack” because the vast majority of TNGBL in the treasury is team deposits, which are used for bribes and to build governance stake (in protocols where the team can further direct emissions to its pools). You’re walking a fine line trying to target 100% CR with a highly volatile and low liquidity asset, in addition to another illiquid asset class that you also custody (RE). It does not appear the solvency of the system is a priority. It appears that rapid growth is the priority, at the expense of assurances to your users.

1 Like