Minor Stablecoins & USDT Issuance: Even More Coincidences
Previously we found a pile of unquestionably-real USD that matched USDT issuance remarkably well. And we all seemed to agree these were not the USD inside USDC or BUSD. Those have nice documentation that (now at least) shows the vast majority of the backing is standard money-market instruments rather than cash anyway.
So where else could these USD have come from? This tweet got us thinking how to maybe resolve the mystery.
We are not saying Bennett is right or wrong — we aren’t even addressing that coincidence — but the tweet got us thinking and lead to two observations. First, HUSD has a small market cap but tremendous minting and burning volume. And second, USDP — generally considered super-legit — has periods of similarly large USD “throughput” and it’s issuer (Paxos) even worked with Huobi some of the time and in doing so generated some suspicious data for their own coin. Is there a simple explanation for all of this stuff somehow? Yes actually, though finding it and learning to read the evidence was maybe not that simple.
Here we analyze the flows into a collection of “minor” stablecoins and find a remarkable connection between these tokens and USDT issuance. And — importantly — we find that this pattern does not hold for USDC and BUSD. Those stablecoins are widely used and do not look like this sort of funnel into USDT.
The plan here is as follows:
- Background on a few minor stables
- Present a very nice data set
- Statistics and analysis of said data
- Compelling coincidences
There are zero allegations anyone broke any rules here. None of this requires rulebreaking.
What do we mean by minor stablecoins? HUSD, TUSD, USDP, GUSD. There is also USDK but it’s so tiny the word “minor” doesn’t do it justice.
GUSD & USDP likely everyone will agree are properly USD backed and perform something like the required KYC/AML checks. The other two…let’s just say questions have been raised. As is usual here we are going to assume everyone is telling the truth and try to find a way to fit all the public data together with some newly unearthed facts.
HUSD is an odd stablecoin. First, it’s not particularly large in market cap terms but sure has a lot of minting and burning:
People may associate Huobi with the exchange and an Asian base of operations but HUSD was run by Paxos up until July 2021:
And we can see this change in the payment instructions for sending USD to Huobi:
So it’s operated either by Paxos (a NY company) or Huobi Trust. What’s that? Well:
And they switched over to Stable Universal from Paxos as the stablecoin service provider:
Which seems to be an odd outfit in that they still take payments into Silvergate bank:
Note the account numbers. Stable Universal is taking payments into the same accounts that Huobi referenced above. And they are a strange outfit in that, despite all the US banking stuff, they don’t take US customers:
So Huobi and this HUSD are a bit weird. But there is a plausible route for them to have all the USD. And, at least in part, those USD sit at US banks. Huobi Trust is a licensed (in Nevada) financial services business in the US so they should have no trouble moving money into the US banking system.
This one is also kind of wonky. But these folks tell you where their USD sit:
Recognize some of those bank names? Good. From here on our we are going to assume these things are all backed by real USD and that, to a large extent, those USD migrate to Silvergate and Signature. There is ample reporting that these folks all bank at the same two places. For example:
Rivals Signature Bank and Prime Trust Team to Offer Instant Payments for Institutions
Crypto banking competitors Prime Trust and Signature Bank have partnered in a bid to appeal to institutional clients…
There is also ample public documentation of Prime Trust offering services which involve bank accounts. For example here, here, here, here and here.
First Digital is a bit murkier. They aspire to offering a bank account settlement network. And raised money to this effect. There is a brand new UK sub which can probably open normal bank accounts (now at least). But the only hard info we have concerns a Vietnamese bank call PVcombank.
That last link is an interesting SEC filing where an investment company called Vemanti set up a USDV stablecoin and then gave up because:
Due to the current lack of a clear legal framework specific to stablecoins, the Company currently does not foresee an unhindered path for USDV to become commercially available and widely adopted in any jurisdiction without regulatory and compliance risks and thus has decided to end the USDV project.
This USDV thing does have a whitepaper with the following text and flowchart:
So it’s at least “not crazy” to treat Prime Trust and First Digital as similar sorts of service providers. Vemanti did.
Oh, and First Digital works with Fireblocks:
If you try to check this out yourself be careful. There is a “First Digital” founded by Ran Goldshtein that is now owned by Fireblocks which does not appear to have anything to do with the Hong Kong-based “First Digital Trust” except they share a name and both work with Fireblocks. It’s a small industry and, truthfully, “First Digital” is a reasonable and reasonably-generic name. Fireblocks, as discussed in an earlier post and elsewhere, connects everywhere.
Minting & Burning Data
That HUSD screenshot showed some weird minting and burning behavior: $250 million in market cap with $10.5 billion in mints and $10.25 billion in burns. That is odd. But how can we dig deeper?
One thing we can do is pull every transaction for these 6 stablecoins going back to their beginnings on Ethereum. This analysis aggregates about 172 thousand stablecoin mints and burns on Ethereum mainnet. It ignores other blockchains because, yeah, this is not a trivial amount of work and Ethereum is by far the largest relevant source of transaction data here.
First up, some aggregate stats:
So right off the bat we can see a lot more cash looks to be churning through the minor coins (ex USDK) than BUSD and USDC. So let’s start by looking at the ratio of daily burn to market cap. And because it’s noisy — this is on-chain data after all — let’s take 30 day windows and run percentiles.
Here we see under 1% of the outstanding market cap getting burned daily, on average. Occasionally there is a big day >5% — but that’s pretty rare and it gets rarer as the market cap grows over time. Note every chart here is to the same scale.
BUSD is similar if a little noisier:
Things are quite different for some of these minor stables. Look at HUSD:
Also note the minimum is well above 0 a lot of the time. That means there are many 30 day periods where at least 1% of the market cap was burned every day. This is not a place to store funds — it’s a thing funds pass through.
TUSD is similar:
As are, somewhat surprisingly, USDP:
These 4 do not look like places where anyone keeps funds. BUSD maybe had a bit of this “funds mainly pass through” oddness years ago. But until mid 2020 market cap was well under $200 million; for the big move up >$10 billion it got as clean as USDC. We’re going to ignore USDK as it’s just too small in all senses to matter.
As we did previously when we just assumed all the money was at Signature and Silvergate here we are going to do something similar. There is probably some amount of weirdness at BUSD (it’s huge!). And not everything in these minor stables is weird. Let’s just guess they give-or-take even out. As we will see later things fit together pretty well.
You Said USDT Right?
Indeed we did. We know these stablecoins interact with the same banks as the USDT-sized pile of dollars we found last time. Now let’s compare USDT market cap against the cumulative burning of these stables:
We can see based on the shape of the combined minor stable, USDC and BUSD burns that the two “more normal looking” stables take off much later and bend more aggressively. All three are on the same secondary axis at the same scale. There is an uncanny correspondence between USDT growth and the amount of funds being burned through these minor stablecoins.
What Is Happening?
This is yet another weird coincidence. But if these are all real USD — and we have no reason to believe they are not — then we have three possible explanations:
- The USD flowing through these minor stables end up at Silvergate and Signature, somehow, as some kind of (maybe oursourced) backing for USDT. They pass through these minor stables as a sort of funnel into the “Tether complex.”
- There is another $20 billion in crypto somewhere.
- These burns mean something else.
Let’s take the last one first. What else could these burns be? These tokens aren’t exactly huge in DeFi. Quite a few people we discussed these results with struggled to believe USDP had over $5 billlon in burns or that HUSD had over $10 billion. Bluntly: nobody uses these things. People also saw flat lines on market-cap charts and incorrectly inferred no activity. USDP spends much of 2020 with what looks like an unchanged market cap — but hundreds of millions of tokens were created and burned along this line:
Anyone with a good justification for #3 should reach out. For example: if your trading firm borrows industrial amounts of HUSD let us know — along with, you know, some evidence — and we will amend this post.
As for 1 and 2…Occam’s Razor tells us to assume these are the same dollars until some evidence pops up to show otherwise. We have good reasons to believe they are the very same dollars. And there is one place where they match even better.
Some folks may be thinking “yeah that’s just washing money.” Maybe, sure. But the money has to end up somewhere. It is theoretically possible these funds were washed and ended up somewhere else. Burned HUSD might get withdrawn in cash somehow. But it’s a lot simpler to suppose they ended up in the pile we already know about with a similar size and shape.
Oh, and similar shape? Here is burning of the minor stables vs issuance of USDT on Ethereum and TRON:
Pretty tight connection during 2019-2020 no? These are all on the same axis.
Another coincidence: Huobi announced support for USDT on TRON in March 2019 and Poloniex in April 2019. Those are right after Tether started supporting TRON. It starts to diverge in late 2020 and there is of course more but this post is already too long.
To be completely clear we are not suggesting any party referenced here did anything illegal or even “wrong.” The key takeaway is that it looks like a large chunk of real USD travelled through a few minor stablecoins and ended up in one of the larger US-based crypto-serving banks as some kind of backing for USDT on TRON. And this funnelling process may well have been accomplished without any out-of-bounds activities.
Statistics are a bit difficult here as total burn can only ever go up while market cap goes up and down. And USDT does weird stuff with their authorized vs issued processes. We would like some tighter methods. And rest assured better methods with clearer charts are coming in a later instalment. This is intended as a (somewhat) accessible sketch rather than a rigorous treatment. Also, yeah, this thing is pretty far from normal finance and needs some novel-ish methods. Ideas for tests would also be appreciated.
While preparing this HUSD started to loose it’s peg. For all of the stablecoin work we have presented so far the thesis is that enough real USD are there but the pegging mechanisms are informal and subject to the whims of the operators. If the operators want to take the money and run they probably can. If the inflows originated from questionable sources they also might get siezed. We are not defending these stablecoins — we are just trying to explain what happened.