Coinbase

So Coinbase has not done a PoR nor do they seem likely to (despite my many entreaties). The TX PoR legislation HB1666 (https://legiscan.com/TX/bill/HB1666/2023) which would ordinarily cover them (since they meet the criteria for having operations in the state) gave them a carveout since they are publicly traded company that adheres to a high standard of audits. So I can’t apply my PoR methodology because they don’t do a PoR.

Coinbase did write a blog post a while back about why their approach of having formal audited financial statements surpasses the assurances you get from a PoR (https://www.coinbase.com/blog/how-crypto-companies-can-provide-proof-of-reserves). Interestingly, the SEC’s much-maligned rule change, SAB121 (https://www.sec.gov/oca/staff-accounting-bulletin-121), which forces custodial firms to account for customer cryptoassets ON balance sheet (as opposed to off balance sheet, like Coinbase used to), means that Coinbase is now listing customer assets as a balance sheet item in their quarterly filings. [Note: SAB121 is a horrible policy, because it means banks have to take a capital charge against cryptoassets they would seek to hold for clients, making it completely uneconomical for banks to custody crypto. But in this case it’s helpful from an accounting perspective]

So Coinbase won’t provide a PoR, but they have to provide audited financials (Deloitte is their auditor) on a quarterly basis and as a public company they are held to a high standard. If you read their Q3 filing (https://d18rn0p25nwr6d.cloudfront.net/CIK-0001679788/9dcd01f7-3e99-4ad7-8d9c-fb39e00db165.pdf), as of eom June 2023, they had $124.3b worth of crypto assets held on behalf of their customers. This is a sufficient “proof of reserve” in my opinion as it gives, in my view, even stronger assurances than a conventional cryptographic PoR, since Coinbase is legally compelled to tell the truth here, and these numbers are audited. I still think they should do a higher frequency cryptographic PoR though. They don’t really provide a lot of detail here, no breakout by asset type. Even though it’s highly credible, it’s unspecific and low-frequency.

Coinbase also has a balance sheet they can use to remediate a loss they suffer if they are hacked (for instance in their Q3 filing they hold $760m of non-customer cryptoassets on their balance sheet, and $5.2b of cash), and they have the ability to obtain debt or equity financing if they run into trouble. From what I understand, Coinbase has strong controls in terms of security and IT, so I don’t expect them to be hacked. They also have a 10-year track record of not being hacked.

However, the risk of a hack is always unavoidable, and can never be fully mitigated. Given all of the above, I give Coinbase a 98% chance of being able to safely custody all customer deposits for the next four years.

Circle

As a stablecoin issuer, Circle doesn’t do a PoR either. A proof of reserve is applicable when a custodial firm isholding cryptoassets on the asset side, with client liabilities on their own database. The PoR endeavors to reconcile the two in a credible way. A fiat-backed stablecoin is the opposite - the liabilities are on the blockchain, and the assets are generally fiat assets in the banking system.

Circle has improved their disclosure framework. They provide monthly attestations signed by their auditor, Deloitte. This is not a “financial statement audit” the likes of Coinbase is required to procure though, it is an “examination” of management’s “assertion”. Still, Deloitte has a reputation to uphold so they have a very strong incentive to provide a reliable examination. Circle’s July report lists $26.4b in USDC liabilities, and $26.4b in assets held in reserve (https://www.circle.com/hubfs/USDCAttestationReports/2023/2023 USDC_Circle Examination Report July 2023.pdf). The reserves consist of $8.3b worth of short dated US Treasuries, $15.9b worth of reverse repos, $1b of cash in their reserve fund, and $1.9b of cash held in banks. The Treasuries are listed individually by CUSIP. The reverse repos are overnight cash lent to financial institutions, backed by Treasuries held by those borrowers. The cash is mostly held at BNY Mellon, the largest custodian in the world. The reserve assets, aside from the cash position, is bundled as the “circle reserve fund”, which is managed by Blackrock. Because each of the service providers involved is highly credible - BNY Mellon, Blackrock, and Deloitte – I have no concern about the asset quality here. Even if 3-month treasuries sell off (i.e. yields rise), and decline in face value, Circle can simply wait to collect the coupons and receive the full nominal payout. So even if their Treasuries decline in value in real terms, all they care about is the nominal payout, since their liabilities are also denominated in USD.

Although Circle wants us to think the assets are bankruptcy remote, in my opinion they aren’t really. They operate under state by state MTL licenses, rather than a state trust, like Paxos does with PYSUSD. It’s not clear that client assets would be protected in the case of bankruptcy or liquidation. Paxos chirped Circle for this a while back (https://paxos.com/2021/07/21/a-regulated-stablecoin-means-having-a-regulator/). Circle isn’t really regulated as a custodian of assets by a state regulator like NYDFS. The USDC terms don’t say anything about bankruptcy remoteness or treatment of assets in the case of Circle’s liquidation (https://www.circle.com/en/legal/usdc-terms). They do mention that USDC assets are segregated from Circle corporate assets though.

However it’s pretty hard to imagine a situation where USDC goes bust or takes on a ton of superseding liabilities. Circle itself also maintains a balance sheet from capital raised, although we don’t know its exact size. Circle had the snafu with their cash being held at SVB back in March 2023, but they have since changed the configuration. As of their most recent report, 88% of their cash was held at BNY Mellon. They have 12% at “other regulated US financial institutions” which they don’t name, but I expect they are using a more prudent model of using multiple banks to get “super FDIC” insurance (although we don’t know this for sure).

Overall, Circle assets can be considered highly secure, despite the lingering issue of questionable bankruptcy remoteness. I don’t expect any other SVB type situations to develop now that the Fed has reacted and created liquidity facilities like the BTFP.

Since the USDC asset side is in regulated traditional finance instruments, there’s no real risk of hacks here either. I give Circle a 95% of safely custodying all customer assets for the next 4 years.

Tether

Tether lists on their USDT transparency page $86.4b in assets corresponding to $83.1b in liabilities, with a $3.2b in listed shareholder capital as a cushion. They do quarterly reports, currently undertaken by BDO Italia (not a top tier auditor). Their latest is June 30, 2023 (https://assets.ctfassets.net/vyse88cgwfbl/63oJePOHqIvrcnXWMPZ1M0/4cfaf2e7cdf80c30b17fdc70faaf741f/ESO.03.01_Std_ISAE_3000R_Opinion_30-06-2023_BDO_Tether_CRR.pdf). Their reserves consist of $55b in 3-month Treasuries, $8.8b of reserve repos, $8.1b of MMMFs. They don’t list the Treasury CUSIPs and they don’t disclose where the assets are custodied. The weirder stuff is the following: 115m in corporate bonds, $3.2b in precious metals, $1.6b in BTC, $2.3b in unknown ‘other investments’, and $5.5b in secured loans (loans to crypto entities, if you believe press coverage, overcollateralized by liquid assets).

Each of these is problematic - the corporate bonds may not be liquid (we don’t know), the precious metals don’t correspond to the liability (gold could sell off against USD), BTC is volatile and clearly doesn’t correspond to the USD liability, there’s $2.3b of complete unknowns, and there’s $5.5b in loans to unknown third parties of unclear creditworthiness. On the face of it, this is pretty weak - we have to believe Tether’s relatively unknown auditor, they are doing quarterly and not monthly reports, they aren’t listing counterparties, custodians, intermediaries, and we don’t know where they are regulated or what their domicile is at all. The charitable interpretation is that Tether is opaque by design since they want to be as offshore as possible and they would expose their counterparties to governmental pressure if they named them. With the auditor, it’s likely that no big 4 auditor would agree to audit them since they have had issues in the past. And Tether maintains an asset liability mismatch on a liquidity and a returns basis.

However, Tether also makes a ton of money. $4.7b annualized at a 550 bps rate if they put everything into short term treasuries (and for their loans they are probably charging more). With this, they are able to build a reserve cushion. They also have a ton of breakage in their liabilities - likely a decent chunk of Tethers are simply lost on chain and will never be redeemed. (This is similar to how gift cards work, where some just get lost and never get redeemed). And Tether liabilities are very sticky, because a lot of their holders may not have access to the traditional finance system, so in a sense they are trapped in crypto. If you went into Tether specifically to move your funds away from tradfi, then you may not ever want to redeem. So Tether’s ersatz portion of their portfolio only becomes an issue if they have massive redemptions, which seems pretty unlikely based on history. Even throughout a brutal bear market USDT hasn’t seen that many redemptions, in fact they’ve been growing and their supply is at ATH.

Obviously we know nothing about the bankruptcy remoteness, so we have to assume USDT holders are unsecured creditors of Tether limited, which is registered in the BVI (not exactly a high quality jurisdiction). The main risk that I would be concerned about would be some kind of fine or enforcement action which impairs their ability to operate. If somehow a court decided they had to be shut down and unwound, likely there would be a long settlement process where USDT holders would be able to claim their pro rata share of the reserve.

Unlike the Tether truthers, I don’t think Tether literally doesn’t have the funds, but I can also acknowledge that they have clear and apparent deficiencies in their setup. I don’t worry too much about their lending activity, because that’s a relatively small portion of their reserves and they can handle a ton of redemptions before they get to that. The main concern would be enforcement, especially given their very patchy track record. It’s very unclear what is going to happen here, and it mostly depends on whether US enforcement agencies gain the political will to truly go after Tether. In the case of an enforcement action I think the settlement process will siphon some portion of client funds, similar to how FTX is being drained of its reserve by fees taken by the administrators.

Because of the poor quality of auditors, unclear legal domicile and legal status, questional reserve composition, and patchy track record, I give Tether a 60% chance of being able to fully honor client redemptions over the next 4 years.