Proposal to add a Paxos Dollar (USDP) PSM

Proposal to add a Paxos Dollar (USDP) PSM


Paxos Dollar is a USD-pegged, fiat-backed stablecoin created by Paxos. Paxos Dollar is 100% backed by cash and U.S. treasury securities. Paxos publishes a monthly report detailing their asset allocations. MakerDAO’s DAI is currently backed by 7.8% USDP. Adding USDP as a PSM asset for USDi can help increase demand to hold USDi, as MakerDAO’s case proves that large actors are using USDP to acquire other decentralized stablecoins. USDP’s backing is similar to USDC, so the risk of adding USDP is not significantly higher than that of USDC.


Token Address: 0x8E870D67F660D95d5be530380D0eC0bd388289E1


MCAP: $896M
Uniswap v3 liquidity: $7.53m

The reason why on-chain liquidity is thin is due to MakerDAO’s USDP PSM, which allows USDP to “piggy-back” off of USDC and DAI’s deep liquidity. This PSM also allows Interest Protocol to quickly redeem all USDP for USDC or DAI if necessary.

Technical risks

  1. Type of contract: Stablecoin
  2. Underlying asset: Cash, U.S. Treasury Securities
  3. Time: 1499 days
  4. Value: Claim over $1 of assets
  5. Privileges:
  • The controller can mint & blacklist USDP tokens
  • The transfer of USDP tokens can be disabled temporarily on a per-account basis, as well as for the entire smart contract
  1. Upgradability: Yes

Hi Fishy,

Generally, I think adding high-quality stablecoins such as USDP to the PSM makes sense. In the long run, it would be cool if the PSM was a melting pot of the major stablecoins allowing anyone to deposit a stablecoin for USDi and any USDi holder to redeem for a stablecoin. The protocol is designed so an upgrade could introduce this functionality if IPT holders approve it.

The main question I have been trying to answer is what mechanism might be good for minting & burning? MakerDao’s model of having separate PSMs for each stablecoin with individual settings makes sense. Then there is mStable’s basket model, where anything can come in and out without cost but there are limits to each coin’s share of the basket. Finally, the Curve AMM model, where the protocol tries to target some balance of stables.

Each has its upsides and downsides. I think the MakerDao & mStable mechanisms make the most sense, but I’m curious about what you (and everyone) think.

If MakerDAO has an emergency shutdown, DAI holders redeem a portion of each collateral type.

If we let users mint USDi with depegged stables, and redeem pegged stables, then the protocol can have a run.

Anatomy of a Vault, showing its three sets of rules.

Consider a simple example, a Vault will mint 1 NOTE token whenever 1 DAI or 1 USDT is deposited, and will burn 1 NOTE and redeem it for 1 DAI or 1 USDT, as the user chooses. What is the value of 1 NOTE? As should be obvious, these deposit and minting rules create a constant-sum condition. Thus, if the price of DAI is less than USDT, users will use DAI to mint NOTE and redeem them for the more valuable USDT, and vice versa. This Vault’s collateral will consist almost entirely of the least valuable token.

Even if we have separate PSMs, the protocol as a whole acts as one big PSM for the USDi note.


If we were to go with the mStable model, we’d most likely want flat maximums on the quantity of a given stablecoin in the pool, as those trying to flood the pool with depegged stables could mint using on-peg stables to modify the pool’s allocation ratio.

Curve pools don’t work super well either as even a very low A factor can’t do much if there is a serious depeg.

Traditional PSMs only protect against a depeg if there are minting limiters such as a price oracle (refusing minting if stablecoin is depegged) or quantity limit (e.g. no more than 1M USDi minted using USDP daily). It seems like traditional PSMs are easier to secure by adding these components. We should seriously consider a method faster than token-governance for freezing the PSM during a depeg event, such as an Orca Pod (manual) or Sentinel (automatic).

Ultimately it seems like a rate-limited PSM approach is the best one given the high security comparative to the mStable/Curve approaches.

1 Like

You’re right. The PSM model only works if the value of each asset is approximately the same. For example, USDC, USDT, USDP, BUSD, TUSD, and GUSD are all traditional dollar stablecoins. However, it is a different story if we mix in DAI, FRAX, LUSD, and sUSD.

Similar to MakerDao, setting individual caps on the number of stablecoins to hold in the PSM makes sense to me.