Treasuries and bonds

I wanted to raise the topic of whether Nexus Mutual should use some of its capital to invest into US treasuries and traditional corporate bonds.

To start, let’s segment out the discussion of selling ETH to stablecoins. I’m thinking of this as an opportunity to get yield on stables, and the amount of stables we should be holding via selling ETH is a separate discussion. I’m bringing this topic up because there’s currently discussions on-going about whether to sell ETH to DAI, and it might be helpful to note that there are options to earn yield on DAI so that it’s not completely stagnant.

This post exists as a forum for discussing the high-level question of, would we want to hold treasuries or bonds? Any vote related to this would be purely a signaling vote.

When it comes to buying US treasuries and bonds, you need a traditional broker and custodian. There is no legitimate way to get around this step. So, ultimately, there will be centralization off-chain with both of those parties (typically the same entity).

There are 2 major options:

  1. Utilize an existing on-chain tokenized bond protocol. For example, Ondo is tokenizing US treasuring with their OUSG token. You send funds to their smart contract to mint the token, they send those funds to Coinbase to offramp to fiat, sending it to Clear Street, a reputable broker, who purchases and holds a Blackrock ETF (AAA) of 1-year or shorter US treasuries. They do the same for a corporate bond ETF (BBB-), and a higher risk corporate bond ETF (BB-). All entities in this chain are USA located.

  2. You offramp funds from Nexus Mutual capital pool to the Panama foundation entity, and contract a broker to purchase whatever treasuries or bonds you want, and to hold them. All entities in this chain would be Panama located.

There’s a third option which is a long-shot, which is that we try to work with MakerDAO to utilize their existing infrastructure that they created as part of MIP-65. This is a similar infrastructure to Ondo, but gives a more diverse exposure to real world assets. I’m particularly impressed by this, and the legal work that went into bringing it to life.

I had a conversation with some of the MakerDAO team about the creation of a vault like this, which would split out a bond essentially, to give more concentrated exposure to their RWA. Ultimately, you’re still carrying the DAI exposure, but you’re getting much more exposure to the RWA piece via the bond. So, it’s slightly more risky than a straight bond / treasury approach, because you hold exposure to something going wrong with DAI.

But, the argument here is whether to hold DAI, or to buy treasuries / bonds. So, to me, you’re getting the MakerDAO / DAI risk exposure anyways if you choose not to invest. I also love how you get such diverse RWA exposure here, and can hop onto the existing very strong railroads they’ve laid down. Long term, this RWA yield should flow through their DSR, but until the RWA are 80-90% of their capital, DSR will always be much lower than the yield on this bond, hence the logic in separating out the bond and buying it directly.

I want to also note that there are a few other protocols aiming to do what Ondo is doing. Backed Finance is the strongest alternative, but has not launched a fixed-income product yet. They are weeks / months away from doing so, and so should be considered. They are Swiss based currently, but I think it’s likely they will ultimately be running this through Lichtenstein. Switzerland is a very friendly entity for this because they allow raw tokenization of treasuries and bonds, will transferred ownership via the token. So, you can buy and hold the actual treasury rather than being required to buy an ETF.

Theoretically, it makes a lot of sense to get exposure to treasuries and bonds, especially given the high rate environment. We’re currently limited on utilizing stables to earn yield on-chain due to technical requirements, the low yields available, smart contract risk and overlapping exposure on the cover side.

Most insurers will hold treasuries and bonds for a very significant percentage of their capital, with good reason. If we can find better yield sources on-chain with a similar risk profile, we should consider those equally, but that doesn’t seem to exist right now.



Thanks for putting this together!

All thoughts theoretical - as you said the ETH vs stables discussion would need to come to a conclusion in favour of stables before this is possible to implement.

Two sides to this in my mind:

  1. From a purely pragmatic, financial point of view, it seems to be a simple answer: if you’re getting 1-2% on supplying to ‘safe’ on-chain protocols like Aave/Compound and 4% on government instruments, the risk-reward balance makes setting up the infrastructure a no-brainer. A few barriers when it comes to liquidity, trusting the centralised parties in the chain and having reliable tokenised representations and oracle for the capital pool.

  2. However, the main debate imo is whether we want to be lending money to the US government over promoting the on-chain system at this stage. Is it worth the extra centralisation from involving brokers, off-ramping and the nature of ultimate recipient of the lending for a few more percentage points of yield during certain market cycles?


It may be helpful to some of us to recount what accounts for the Mutual’s inevitable exposure to DAI, which is the backdrop to this? What sets the minimum DAI holding? What causes it to change? Why is holding DAI preferred to a pure ETH canonical over-collaterised stable like $LUSD? or a synthetic stable like sUSD? Once we understand the source of this DAI exposure, it is easier to engage in a discussion of whether/how to seek yield on this DAI exposure, as opposed to switching everything to ETH and holding a broadly diversified basket of LSDs with intrinsic yield of about 5% nominal against a deflationary ETH supply raising the yield to about 7% real in ETH terms. Add auto-harvesting of broadly diversified extrinsic yield options on the portfolio LSDs and the option of staying 100% ETH LSDs based starts to look very attractive? Most of us have invested in the Mutual precisely because we are looking to hedge our personal OVER exposure to fiat. Why does the Mutual itself have to have fiat exposure?

Fiat exposure is mainly driven by members buying cover.

Today, members who buy cover are making their own choice to denominate their cover in DAI. The reason is because most of the protocols that we’re covering denominate in stables. So, if you have $1600 USDT in a protocol earning yield, they prefer to buy $1600 DAI cover rather than 1 ETH cover. Why? Because if the price of ETH drops, they still have $1600 USDT at risk, but would have less than that in coverage. So, they don’t want to take on the currency risk.

If we force members to denominate covers in ETH, our offering becomes uncompelling, because of this currency risk and they will likely just use an alternative cover protocol.

If we let members denominate cover in DAI, but only hold ETH, we hold that currency risk ourselves. Meaning that a users $1600 DAI policy is 1 ETH today, but if ETH goes to $200, then we are paying out a much larger % of the capital pool on a claim than we expected when we underwrite the policy. As a result, the policy becomes highly unprofitable.

The potential impact of claims when you have a currency mismatch will overshadow any difference in yield between opportunities like treasuries vs staking ETH.

Many of us have a bias that ETH will be higher next year than today, which is fine. But think about what happens in the worst case scenario. Say Convex has an exploit, and we end up paying out the full $18m of outstanding cover, the extreme majority of which is denominated in DAI.

It’s possible to imagine that an exploit on one of the biggest protocols in the space would also cause the ETH price to move downwards, because of fear. It could also cause it to move upwards as people move away from stable yield farming and towards other opportunities. We don’t know. But there’s a reasonable chance that a Convex hack could cause ETH price to drop, even if only for a couple of weeks.

Then, you’re paying out enormous DAI denominated claims with a devalued ETH. Meaning we’d have to sell more ETH to DAI to pay the claims than we expected. This throws off all of our underwriting risk parameters like the 20% limit on any individual risk / capital pool size, because that 20% could become 25% as an example, as ETH price dropped.

In a nutshell, those are the concerns with holding mismatched currencies to your exposure.

Thanks for clarifying. I think you are right to highlight the mismatch risk, but should ask the Mutual if they are prepared to wear it (I am). Regarding yield on DAI, Maker has announced Spark, a new DAI yield-generation platform on chain. Suggest we park this discussion until post-Shanghai and having evaluated Spark, all the while making clear to the Mutual that we need to express willingness to wear the currency mismatch risk until then.

Great Post @Dopeee . With the difference in rates on chain vs. US treasuries, I would highly encourage The Mutual to consider this path moving forward. The legal framework for something like this would be my only concern and The Mutual is one of the few kyc’d protocols out there that adds a huge step in the right direction for getting this done right. As far as centralization goes, we’ve already entrusted groups to handle these kinds of transactions in the past, and with the kyc nature of this protocol, it makes it that much more difficult for a bad actor to not get the job done. There will never be a sure fire way to avoid that fear, but I’m not of the opinion that it is a good enough reason to not give The Mutual some exposure to this system. The risk averse aspect of partaking in the US Treasury market seems like a perfect fit for this protocol. Would love to see more action here.

1 Like