Stablecoins As Pegged Currencies: Standard Models

If we view stablecoins as traditional pegged currencies many tools from finance can be employed.  In particular the description of pretty much every (non-algorithmic) stablecoin looks like a currency board.  Very roughly this is an arrangement where some entity -- usually the Central Bank -- targets an exchange rate, promises to exchange unlimited amounts at that rate, and maintains a pile of foreign currency equal in size to the local money supply.

Mundell-Fleming

The standard model for exchange rate regimes is Mundell-Fleming.  This is, roughly, where the whole "impossible trinity" trilemma thing comes from.  Stablecoins are odd in that they purport to offer a fixed exchange rate with free flow of capital and monetary policy, such as it is, consists of targeting a positive balance of payments within this framework:

The higher the stablecoin rate the less net demand there is for USD: fewer people want to convert out to USD, and more people want to convert in to stable.  If you want to target inflows you pick a rate somewhere above where the lines cross.  Now with our interest rate in hand we can apply the Mundell-Fleming:

This is completely standard theory: the overall economy clears when the demand for money matches the demand for investment (this is a simplistic take, sure).  Recall here we have a balance of payments target met via deposit rates.  Normally this framework is used to analyse government policies.  But here we have only a simple monetary policy to consider.  We can observe in the market that rates are going up and inflows are maintained.  Let's look at the new equilibrium:

There are two new possible equilibrium points.  The upper-left one corresponds to increased demand for money (stablecoin) and a shrinking economy.  That is what naturally happens when rates go up.  Most of economic policy is concerned with how to stimulate the economy into the upper-right equilibrium.  Normally that is done through government spending and/or the creation of base money.

Now it is a bit clearer why this process generates such wild gyrations in crypto prices: there is no government and there is no policy process so there is no way to steer where on the upper green line we land.  Every day we can see base money creation -- backed or otherwise -- and we can see quite high deposit rates.  The higher the jumps in rates needed to maintain the balance of payments the wilder this will get.

One could even argue, cynically, that an exchange or stablecoin issuer buying in the market with freshly-minted unbacked tokens was stimulating things in the precisely the orthodox manner!

Limits

How long can this go on?  Well, as long as the positive balance of payments target is hit basically forever.  So let's revisit that first chart and consider what happens when rates are both extremely high and rising:

The crypto economy is small compared to the rest of the world.  And there is only some fraction of world capital that is going to get converted into stablecoins at any price.  Exxon isn't going to liquidate their capital base for crypto even at 200% deposit rates.  Note that employing a government-backed fiat-linked CBDC is not a conversion into stable as the CB guarantees the peg in fiat.

Rapidly rising rates, dramatically increased volatility, and things that look like traditional Keynesian stimulus are signs we are on the orange part of the curve.

That curve stops somewhere -- there is some point where, at any rate, the balance of payments surplus cannot be maintained.  Historically this happens around when the population riots because the punishing rates have decimated the real economy.  Crypto is a purely financial economy and a balance of payments crisis won't lead to empty supermarket shelves.  It will, however, lead to runs on stable reserves, exchanges and other entities nobody expects to blow up.

Debt Service

So rather than considering the degree of pain that is politically acceptable let's look at the impact of rising rates on the balance of payments demand.  Total stable outstanding is something like $100B.  Let's say the average borrowing rate is 5-10%.  That's $5B-$10B per year in debt service.

What about Italy?  The Italian government has something like EUR50B per year in debt service payments.  Stablecoin net debt service is already on the order of Portugal or Austria.  Except those are compounding at near-0%.  At a 10%+ funding rate the demand on our meagre "monetary policy" to maintain a positive balance of payments in the face of this sort of debt service is going to be quite strong!

If growth and rates continue on their current trajectory the stablecoin space will need more funding in 2023 than Italy.  The math is simple: say the average balance in 2022 is $250B and the average rate is 15%: that's $37.5B.  If 2023 is $400B and 20% that's $80B in debt service.  Simply rolling the $100B outstanding today at 10% would require ~$20B in new funding for 2 years.

One imagines those scenarios -- which will come to pass absent a change in trajectory -- are off the top of the orange line and the game cannot continue.

Tether: The Assets

Tether released their reserves composition:

Let's break that into a simple table:

Component	Fraction	$
Commercial Paper	49.6%	$19.9B
Fiduciary Deposits	18.7%	$7.36B
Secured Loans	12.6%	$5.03B
Bonds	10%	$4B
Cash	2.9%	$1.18B
Reverse Repos	2.7%	$1.1B
T-Bills	2.2%	$895M
Other	1.6%	$657M

Following on to an earlier discussion of the transactions and one about banking data we can now dig in. If this is all held within Deltec then these numbers must be visible in the upcoming Bahamian banking disclosures.  In particular we know that:

• The deposits will appear under assets due from other financials
• The commercial paper holdings will appear under either other financials or investments
• Same for the secured loans and bonds

There is some flexibility in how to carry the various credit instruments there, but the deposits need to appear on a bank balance sheet and as something owed by another financial.  This should be interesting.

Transactions

In that earlier discussion we posited that perhaps the real Tether workflow is:

1. A client takes out a USD loan from Deltec
2. Deltec credits their account and records a corresponding asset for the loan
3. The client transfers those USD to the Tether account
4. Client receives USDT

Given the asset composition we can now conjecture a step 5: Tether transfers USD to client in exchange for commercial paper/loan/bond.

This is not an unbacked issuance!  It is backed by an obligation from whatever client initiated this whole sequence of events in the first place.  In this case Deltec is "ok" and the liquidity problem exists between Tether and whoever is financing USDT issuance.  Deltec is merely greasing the wheels and, likely, keeping all the accounting records.

An Uneducated Guess

If we were forced to guess what happened here: step 5 was added recently to avoid claiming a large USD deposit at Deltec.  It is inconceivable the Bahamian regulator is unaware of these questions.  And they would likely look askance at a $30B+ cash deposit in one of their more heavily-questioned -- if not necessarily questionable -- banks.

It's also possible step 5 was there all along to keep these exposures at arms length from the bank.  Risk-wise they are identical, except in the case where Deltec is putting significant capital behind it's loan book.

From the banking data, if this all shows up, we will also be able to see if the "Fiduciary Deposits" are inside or outside the Bahamas.  And the commercial paper entry alone will be over 10% of total banking system assets!

And just in case anyone is thinking it: no, this is not how fiat currency & TradFi work.  There are some shared words and concepts but no.  

Macroeconomics Of Mined Cryptocurrencies

The "mining" in some cryptos is meant to evoke traditional mineral-extraction mining: digging valuable materials out of the earth.  It is of course not identical but in broad strokes similar models should apply.  Macroeconomics has a lot to say about these sorts of industries.  Let's apply the basics here.

We are not trying to make predictions about the future here.  Instead this is the beginning of a basic framework within which to analyse prices and explore ways in which crypto mining is not like traditional mining.

The Bitcoin Difficulty Adjustment

The Bitcoin blockchain, and many others, incorporates a "difficulty" measure to keep block times consistent.  Roughly: the network adjusts the difficulty of mining new coins to keep new block issuance on a steady schedule.

Every 2016 blocks the network the network checks how long it took to generate the last 2016 blocks and adjusts difficulty up or down to maintain a 10 minute per block target.  A more detailed explanation is here and many other places.

If you go out and spend a bunch of money on mining hardware you will have excess profits until the next difficulty calculation.  But after that the network adjusts and your profits disappear.  Yes you may earn a larger fraction of the coins on offer but, in the long run, you cannot speed up the production of coins no matter what you spend.

Regular old physical mining does not work this way!  This is like trying to build a gold mine where spending more money pushes the gold deposit further underground.

Note that it might be possible to game the system by massively increasing computational power after each 2016-block segment finishes.  But the production of mining hardware is an industrial process that grows in a linear fashion.  And there is no evidence for a cartel in the market hoarding ASICs and GPUs only to release them when the difficulty is low.  Check any chart of historical Bitcoin difficulty and it looks "natural."  Rather we see the price of mining hardware tracking the coins -- evidence for a normal, functioning, secondary market.

Basic Macro

We call the "market clearing price" the level where supply equals demand.  In traditional macro the two sides are known as  "aggregate demand" and "aggregate supply."  And the market clears when they are equal.  The standard diagram is:

Demand, in number of units, goes down as price goes up.  And supply, in number of units, goes up as price goes up.  This pair of upward and downward sloping curves tends to yield stable markets.  Suffice it to say there is more detail out there.  This post is not going to try to review all of macroeconomics -- you have Google for that.

Mining

Now let's build a toy model where supply comes only from mining.  This is not entirely dissimilar from real-world precious metals mining.  Thus the term.  So let's consider a world in which miners sell off their production and something like the Bitcoin difficultly adjustment is in play.  We get:

The difficulty adjustment is essential here.  It means you cannot buy more production for money.  We find a fixed quantity coin supply independent of price.

In econ terms "the aggregate supply curve is vertical" or "supply is inelastic."  That AD curve is also a bit simple.  Let's say the market receives a constant amount of inflows every day.  That means: inflow = price * quantity.  This gives us:

As mining gets more difficult the quantity on offer drops.  We can see the impact on price -- again given a fixed cash inflow -- from changing the quantity supplied easily:

This model is not entirely different from the well-known "stock to flow model."  Simply rearranging the equations we get roughly: price = inflow / mining.  Note that mining and inflow need to be measured over the same time period (i.e. daily, weekly, whatever).

Uneven Flows & Smarter Miners

It is unreasonable to expect the exact same inflow every day.  In practice we will see a range of demand curves for different levels of inflow:

The assumption that miners sell their entire production immediately is also too simplistic.  What if instead miners sell enough of their production to cover the cash costs of mining and hold the rest.  This looks a lot like the demand curve: cost = price * quantity.  Recall that a higher price does not lead to more productive mining!  The supply of mined coins is fixed because of the difficultly adjustment.

There is also some variation in the cost and quantum of production.  So we have a range of supply curves:

Under this pair of regimes volatility will be sky-high!  Find the intersections:

In standard macro the curves are pointing in opposite directions.  But in a market where supply comes from miners who must raise cash but are unable to increase their own inventories and demand from buyers injecting a fixed amount of cash there is no obvious equilibrium.  The intersection of the curves is chaotic.

This sort of behaviour is consistent with high volatility, high sensitivity to mining production and high sensitivity to the size of inflows.  With a difficulty adjustment the market does not "clear" in the traditional sense.

Gold Stablecoins: WTF

Gold stablecoins are odd things.  Everyone accepts that, in practice, fiat currency often has no physical form.  Yes the bank will give you paper money but most transfers occur without any objects changing hands.  But the gold market is not only based around yellow bricks held in basements -- it is also jam-packed with people who are worried their bars are missing/fake/lead/stolen/etc.

Further, wholesale markets have had reasonably efficient solutions for trading gold via certificates for decades.  Nothing is perfect, and there are surely limitations, but the gold market functions fine day-to-day.

So let's find two gold stablecoins and see if a little knowledge can light our way.

PAX Gold

Whitepaper

Each token represents 1 oz of LBMA good delivery gold.  These are ERC20 tokens so they are very transferable.  Loco London gold is also quite liquid but fair enough, ERC20 tokens are more mobile.  But digging further into the details these contracts are weird.

To start, there are no storage fees.  These are allocated bars in "Brinks' bullion vaults in London" that surely charge fees.  Someone is paying the fees.  We haven't reviewed the contact between Paxos Trust and Brink's but it's a fair guess that if the fees aren't paid Brink's sells gold to recoup them.  Brink's does not provide storage for free.  Yes Paxos charges transaction fees for trading the tokens which generate revenue -- but if those fees don't cover the storage costs it is unclear what happens.

A traditional Gold ETF charges fees which go to cover storage etc.  For example the prospectus here explains: "[t]his includes (1) selling the Trust’s gold as needed to pay the Trust’s expenses (gold sales are expected to occur approximately monthly in the ordinary course)..."  So everyone pays pro rata whatever the storage costs are.  That is transparent.

"If you don't pay the storage fees then you loose your gold" is fair enough on a per-account basis.  It is an entirely different proposition to be dependent on aggregate market trading generating sufficient fees to cover collective storage costs.  For a market packed with hodlers and obsessive about ownership rights this is downright shocking.

This is somewhat mitigated by Paxos offering to redeem tokens by delivering physical gold.  Here we move from the shocking to the marketing:

PAX Gold is redeemable in several forms, unlike any gold products available today. Customers can convert their PAX Gold into physical allocated gold, unallocated gold entitlements or fiat.

That may or may not be a useful service but it is not exclusive to Paxos.  These folks all offer it.  Pretty sure those are mostly (all?) regulated institutions.  And nobody does it for free, including Paxos.  So you can withdraw your gold but only through a significantly-more-byzantine process than standard LBMA allocated gold today.  And it's not like this is free: Paxos still charges creation & redemption fees on tokens when you deposit/withdraw 400oz bars.  For smaller amounts the total fees are surely higher.  This is further away from "not your keys, not your coins" than the TradFi alternatives.

This one looks basically like LBMA allocated gold today with a bit more mobility weighted against a more complicated process, less direct ownership and some odd untested quirks.  If this makes sense, depends upon your point of view.

Tether Gold

Whitepaper

A similar setup where each token is 1 oz of LBMA gold, this time in a vault in Switzerland.  This token has no trading fees and only charges on issuance and redemption.  That compounds the problems described above with vaulting costs: it is unclear how Tether can pay storage costs for a significant period given only an issuance fee of 25bp.

This is again an ERC20 token and so it is quite mobile.  Swiss-delivery gold isn't really as large a market as London, nor is it clear precisely which gold Tether is offering.  LBMA good delivery is a standard for bars but requirements to get into vaults are more onerous.  One cannot simply rock up with a bar of unknown provenance and deposit it for good delivery.  The Swiss market is more fragmented than London.

But things start to get really weird the deeper we dig here.  The Tether "parent company" is Tether Holdings in HK.  LEI reference here.  Roughly speaking to be a part of the TradFI system you need a Legal Entity Identifier.  Paxos has one.  But the Tether whitepaper is written in the name of "Tether Operations Limited" which has only a lapsed LEI from 2018 in the BVI.  Tether Holdings is incorporated in HK and shows up on their registry.

Maybe this is just a clerical error by Appleby as the Operations business is registered to their address.  But the whitepaper is quite clear on the legal structure of the tokens:

A new entity, TG Commodities Limited (Tether Gold), digitizes the value of gold using the new token, XAUt.

This company does not have an LEI, nor does it appear on any registry we can find.  Amusingly there is a "T G Commodities Limited" which a space between "T" and "G" registered in the UK.  One of the directors is Sanjeev Gupta (yes, that one) and it was previously called "Liberty NFM Limited."  Here is the official record.  We expect this is just a weird coincidence as that company was first formed in 1998 and the spelling does not match.

Anyway there is no record of the Tether-referenced company which holds the gold existing anywhere.  And before someone suggests these records are not relevant as this is DeFi: gold in a vault in Switzerland is OG-TradFi.  And the grizzled Swiss vault operator surely needs someone to bill for storage.

Maybe none of this matters as long as there are gold bars somewhere, in a vault, with storage costs getting paid.  Tether provides a nice lookup tool to match bars -- well, shares in bars -- to tokens here.  It's all public so we can check a few addresses.  First find large holders at etherscan.  Then paste some addresses into the scanner.  Many wallets don't match.  This is not a high-volume token so it's unlikely these are timing issues.

To given a sense of scale the largest holder at the time of this writing is "Bitfinex 3" with ~42k oz per etherscan and ~36k oz per the "Gold Allocation Lookup."   The second largest holder is "Bitfinex 2" and that is near matching (27,990 vs 27,981 oz).  Choosing random other addresses and you see many mismatches.  Who knows.  Reliable immutable public records these are not.

Tether Transparency

The total at etherscan.io does match, exactly, the total issuance per the transparency report.  Now this is interesting.  The transparency report breaks out the "ready for sale" tokens:

We can compare this against the largest holders:

And now we know the "ready for sale" wallet address.  However if you check for the gold behind that address it shows:

There is no gold listed.  Before someone says "those aren't issued yet" note that "total assets" matches  "total liabilities" -- those 5,769 oz of gold are listed on the transparency page as assets of "Tether Gold" or whatever it's called.  Mechanically it's possible Tether pre-buys gold, issues unbacked tokens, and only attaches the tokens to bars when someone else pays them USD.  But let's check the whitepaper:

Specific gold bars will be associated with each on-chain address that holds XAUt tokens. Tether Gold will make a “Look-up Website” available where anyone can identify the specific gold bar(s) associated with each on-chain address where XAUt tokens are held. 

When an on-chain transaction transfers XAUt from its initial on-chain address (A) to a second on-chain address (B), the Tether website will reflect that some of the physical gold previously associated with A moved to B. Gold associated with other on-chain addresses won’t be impacted by this transaction. Where needed (e.g., to support a redemption or to mitigate fractionalization), the physical gold associated with the XAUt held at on-chain addresses can be consolidated. This reallocation will happen instantly so that each XAUt token will always represent ownership of physical gold on a particular gold bar.

Maybe this is a bit old-fashioned but if a token shows up on etherscan.io it's on-chain right?

This is not the only address where we observed a non-zero balance and no physical gold listed on the "Look-up Website."  Confused reporting clouds everything, impossible to see, the correct balances are.

Thoughts

These two coins are not quite the same.  But both share weird features that no hodling goldbug would want!  Yes, the ability to transfer tokens easily is an improvement over physical bars...but this is a market where all manner of conspiracy theories run wild.  If you believe the Fed -- or ECB or Trilateral Commission or whoever -- is faking gold records why would you want anything to do with this?  If you believe JPM London is selling unbacked futures to push down the price of gold why would you want to mutualize your storage costs with anonymous other hodlers?

If the US treasury put out an errant press release missing a 0 in total precious metals holdings the community would go wild and never let it go.  But the documentation associated with these gold stablecoins -- and XAUt in particular -- is riddled with oddness.  Oddness leads to confusion.  Confusion leads to losses.  Losses lead to suffering.  This market badly needs some light.

Tether: The Bahamian Data

Following the first dig into Tether several people directed me to this medium post.  As there are significant errors in that analysis the plan was to ignore it.  Having been asked a few times now it's worth diving in with clarifications.  Doubly so as this gives us a new data point to watch published in next few weeks.

Read that post first, particularly the parts about The Bahamas.  Then continue.

The Bahamas does publish aggregate banking system statistics (which are also referenced earlier on this blog).  But its' important to check the right numbers.  First we need to identify how Deltec is regulated and then chase down the correct reports.

There is a list of regulated institutions -- the ones that roll up into the aggregates -- here.  Deltec is regulated as a "Authorised Agent" in the "Bank & Trust Companies" category and it is both "Public" and "Resident."  Those are specific terms defined in the document as:

Authorised Agent:

An authorised agent is a trust company authorised by the Central Bank to deal in Bahamian and foreign currency securities and to receive securities into deposits (i.e. to act as custodians) in accordance with the terms of Exchange Control Regulations Act and Exchange Control Notices issued by the Central Bank.

Resident:

A resident status allows a bank and/or trust company to deal only in Bahamian Dollars, but operations in foreign currencies require Exchange Control authorisation. This does not apply to an Authorised Agent or Authorised Dealer which is a designated Resident.

Public License:

A public bank and/or trust company is one which is permitted to carry on banking and/or trust business with members of the public.

The Bahamian Dollar is restricted and that's what the "Exchange Control" stuff is about.  A useful clarification is given here:

Trust companies with resident status are allowed to deal in foreign securities on behalf of nonresident customers.  A bank which is designated as 'Nonresident' permits a bank and/or trust company to operate freely in foreign currencies, however, Exchange Control approval is necessary to operate a Bahamian dollar account to pay local expenses.

So the Bahamian reporting regime is aware of, and tries to monitor, both local and foreign currency activities within the banking system across a range of licenses.

And, unsurprisingly, they publish data on a regular schedule.  The Quarterly Statistical Digest gives us lots of data.  The linked medium post references the "Domestic Banks: Summary of Foreign Liabilities" page in that report.  A lot of discussion -- including a podcast linked earlier on this blog -- asserts that is the wrong category.  It is not 100% clear from the definitions given above and time spent reading the central bank's website what the right answer is.  So let's just look at the entire banking system:

Note the comment at the bottom: The data represents Resident and Non-resident Banks &/or Trust Companies.  The entire banking system's balance sheet as of end-2020 is $173B.  That number is pretty stable, and the composition is pretty stable.  Any USD backing USDT held by Tether at Deltec will show up on both sides of the aggregate balance sheet.

Liability Side

First let's look at the liability side.  Tether isn't resident*, it isn't an SFI in The Bahamas and it isn't a bank.  But it might be a "financial institution."  So their USD deposit holdings appear either under "customer deposits" as "non-resident" or perhaps in the "head office or branches outside The Bahamas" column under "due to financial institutions."  

In calendar 2020 there was about $15B of Tether issuance.  It is certainly possible that sits inside one of those columns.  One went from $57.8B to $41B while the other started at $43.2B and ended at $49.8B.  The end-2020 issuance was $20.9B.  Wherever it sits, the $20.9B of cash behind Tether at end-2020 would up a lot of one of those entries.  If it's under deposits there were some pretty substantial outflows by other customers!

This gives us something to watch when the 2021Q1 report comes out.  With an end-Q1 total issuance of $40.7B something should go up by $20B.  Given the overall stability of these numbers that should be visible.

Asset Side

Now the asset side.  Here we should find Deltec's USD holding, in their Citi UK account or elsewhere.  This is a problem.  Let's take each possible column in turn:

• "Notes & Coins" is both too small and it's pretty unlikely the USD are in physical cash. Though it would be hilarious.
• "Balance with the Central Bank" and "The Bahamas Government" can't be it*
• It's not from a Multilateral Development Bank
• It's not in local currency "Loans & Advances"*
• "SFIs in The Bahamas" aren't holding it*
• "Head Office or Branches Outside The Bahamas" cannot be it as Deltec doesn't have such branches*.  Recall their account at Citi UK is in the name of the Bahamas entity.

This leaves 4 possibilities:

• "Other Governments" could hold some UST if that is part of the reserve.
• "Other Investments" could maybe be it.
• "Loans & Advances" in "Foreign Currency" could be part of it.
• "Other Banks Outside The Bahamas" could be part of it, but it's quite small.

Those columns are remarkably stable for several years.  It is possible the $20B as of end-2020 is hiding in there somewhere -- but the total across all 4 possibilities is only $64.8B.  An incremental $20B in Q1 will be very visible.

This is a key point: while that medium post is incorrect about the Bahamian reporting there is useful data nearby and it gets published soon.  An incremental $20B must be visible in these reports.

What If It's Partially Backed?

The total amount of Tether is now so large relative to these aggregate banking system reports that it doesn't really matter if they are only 80% backed.  80% of $40B will show up in the reports.  We can't tell precisely what is going on.  But we can establish if there is a reasonable quantum of backing or not.

At $40B end-Q1 any substantial fraction held as USD at Deltec will be visible in the regulatory reports.

Getting Real

There is a reason the "Head Office or Branches..." column is the largest: The Bahamas is expecting banks like Citi to open branches in Nassau and deal a lot with their own other branches around the world.  This column kind of doesn't matter to the local authorities.  The banking system ex these inter-company assets and liabilities is on the order of $100B in total.

These 4 possible places to find the money are small, stable, and long-used integral parts of an offshore banking center.  At $50B Tether would represent something like 1/3  to 1/2 of the "real" banking system.  It could be effectively all of the locally-held foreign government securities.  Or maybe the "Other Investments" across the banking system are dominated by shares in cryptocurrency businesses.  Those are all problems. 

(*) Possible Errors In Assumptions

This analysis makes a few assumptions that could be wrong.  Quick comments about them are included here to avoid mucking up the narrative.

"Tether isn't resident in The Bahamas" could be incorrect.  Similarly some assets could be held in local currency.  But these seem exceedingly unlikely as the total amount of money in those categories is small and stable.  And everyone who lives and works in The Bahamas needs those balances to live so they aren't getting transferred to another banking center.

"SFI aren't holding it" is roughly the assumption that Deltec, within The Bahamas, is holding it's own USD.  It's a bit ridiculous to take wires via Citi UK only to then move that money to an account at RBC  or Citi in The Bahamas!  If Tether's reserve is a cash or securities account at JP Morgan Trust Co (Bahamas) Ltd you'd have thought somebody would have mentioned that by now.  Anyway the internal numbers are too small as, again, the main use of an offshore financial center brnach is transactions with your own head office in an onshore money center.

"Deltec doesn't have other branches" is based on the assumption that, if Deltec had other branches, there would be evidence on the internet somewhere.  Recall their UK account is in the name of the Bahamian entity and nobody has mentioned a "Deltec Zurich" or anything like that.  There is an asset management operation in the US but if this money is sitting in a custody account in NY under their asset management arm you'd think somebody would have mentioned that to the NY Attorney General.

Tether: Maybe Everyone Is (Partially) Telling The Truth

Lots of things about Tether are controversial.  A few things about Tether aren't:

1. That it is controversial.
2. Issued during holidays & weekends.
3. Issued in round amounts.
4. The disclosures are non-standard and not maximally transparent.
5. Tether uses some smaller offshore banks and there are reasonable questions about exactly what is going on.

Background

The original Tether design was for a currency board setup.  Setting aside what was really happening or what is now claimed: a currency board is simple.  There are myriad examples of these and similar arrangements around the world.  And we know what makes them work: a transparent and credible monetary authority that continually discloses adequate reserves.  Among countries currently operating credible regimes similar to this are Denmark, Hong Kong and The Bahamas.  Yes, The Bahamas, where Tether's bank sits.  The Bahamas has a currency that is pegged to the USD.

The Bahamas discloses the composition of foreign reserves including USD cash and government securities.  Denmark and HK do something similar (and more frequently).

Whatever is really going on this all could have been much simpler if someone just asked the Central Bank of The Bahamas for a meeting to explain how a currency board works.  Google maps tells me it's a 27 minute drive from the Deltec head office.  That's quite far on Nassau.  The Central Bank is on Market Street in the financial district while Deltec is, somewhat famously, in a more residential part of the island.  Anyway it's a small island.

Deltec's Accounts

We know Tether uses Deltec.  We know Deltec has various currency accounts around the world including, at least at one time, Citibank in the UK.  There are settlement instructions floating all over the web for various cryptocurrency-related businesses going through those accounts such as this.  Whatever is going on there -- that is not how a "correspondent bank" works -- Deltec has various foreign currency accounts around the world.

Normal accounting for these accounts would consolidate them onto Deltec's balance sheet and they would show up in any reporting that aggregates bank balance sheets including Deltec.  The accounts are clearly in the name of the Bahamanian entity.  Maybe there is some unknown Deltec Bank UK related entity and the assets don't roll up to the Bahamas.  Who knows.  Anyway Deltec has hard currency in accounts with other banks.

One imagines Deltec itself runs USD and other accounts internally.  Those wire instructions include a comment "23336 Deltec FFC Alameda" which is likely intended to route the payments to the correct internal account.

This leaves us with two choices:

1. Deltec comingles all client assets in it's own external bank accounts and tracks the split somewhere (Excel?).
2. Deltec runs foreign currency accounts internally.

Scenario 1 is just ridiculous (and quite Madoff-esque).  Let's go with 2.  This somewhat contradicts this podcast interview but then the very existence of non-GBP accounts at Citibank UK contradicts that discussion.  Let's call that a good faith misunderstanding and assume Deltec holds normal corporate bank accounts with normal banks in major money centers.

Tether's Accounts

When Tether says they have an account at Deltec the assumption is an account of type 2 as described above.  This is 100% normal stuff.  Per the currency board discussion above the simplest thing to do is to provide frequent disclosure of the balance of that account.  For reference the HKMA publishes pretty much exactly that information here every day and for many years.  Whatever is happening in HK now that report has been around and trusted "forever."

Various entities in the cryptocurrency world claim they can send USD to Tether to create USDT.  One imagines what they mean is transferring into these accounts, having them moved over to Tether's Deltec internal account, and then receiving tokens.  This Twitter thread suggests this is exactly right.  Let's not worry which banks they are, or that "correspondent bank" is an abuse of notation here, and just go with this flow of funds.

This would explain issuance on weekends.  Citibank UK doesn't process USD wires on weekends.  But it does allow some forms of internal transfer on weekends and holidays.  What if Deltec is just doing the same thing?  Deltec can surely shuffle it's internal records to transfer USD among its clients on weekends so long as Deltec's office is open.  These are just journal entries in an internal system.  FTX wants some USDT so it instructs Deltec to transfer USD to the Tether account and get some USDT out.  This is 100% normal.

A Theory

If this was the entire story it would be trivial to produce a series of bank account statements and transfer records and put this to bed.  Let's assume there is more to the story than some pretty epic trolling.  What if instead sometimes:

1. A client takes out a USD loan from Deltec
2. Deltec credits their account and records a corresponding asset for the loan
3. The client transfers those USD to the Tether account
4. Client receives USDT

USDT are not "unbacked" from Tether's perspective.  Tether's USD bank account has the right balance.  But it is funded by the USD loan from another Deltec client.  Deltec's net USD assets don't change here and consequently any reporting they roll into would be unchanged.  These USD wouldn't show up in aggregate Bahamas reporting because the net asset impact is 0.

In this scenario Tether is telling the truth -- they got USD before any USDT issuance.  Deltec and the auditors are also telling the truth as USD are in their account.

But what are those USD really worth?  Deltec's balance sheet has that deposit as a liability with the USD loan as the corresponding asset.  The backing, in practice, is a loan to some other cryptocurrency business.  It might even be a collateralised loan (BTC, shares, whatever).

Implications If Right

These gigantic USD deposits -- USDT is over 50B as of this writing -- are unsecured liabilities of a small offshore bank.  There is a world of difference between having 50B in a UST money market fund, JPM NY or a small offshore bank.

What we have here looks a lot like the 2008 money market fund crisis.  Everyone involved is marking these USD at par.  But there is essentially 0 chance Deltec can send an outgoing wire for all of this money to the traditional banking system.  Deltec's net USD assets are unchanged by this operation: they are the same before step 1 and after step 4.  Properly marked the USDT are worth whatever the loan is.  This is likely well below par.

We are wholly ignorant of the Bahamanian banking rules.  But it's a reasonable working assumption that if Deltec is a privately owned bank without any domestic presence they can do pretty much whatever they want.  That's what offshore financial centers do and the shareholders are supposed to police this.

It is likely the median Bitcoin Maximalist thinks this is exactly how fiat currency works: banks just make it up and credit their clients.

But trying to redeem the USDT poses a problem.  There is little chance the borrower in step 1 has the cash.  And even if the loan is collateralised it seems pretty unlikely the collateral can be quickly turned into cash at par.

Terminology

If the loans are uncollateralised any end-user redemptions require a fresh injection of USD.  They are "tak[ing] money from new investors to satisfy the [liquidity] requirements of the existing investors.  There is a name for that." From 9:30 in a talk everyone should watch.

If the loans are inadequately collateralised this is an accounting issue.  Watch the rest of the linked talk.

In either of those cases Deltec is balance sheet insolvent.  Were it operating in a regulated space it would be shut down.

If the loans are adequately collateralised all is fine.  Deltec may have a bit of a maturity mismatch but it's a bank so that isn't shocking.  If it was regulated maybe there would be a capital raise or something.

Maybe Nobody Is Lying

To the extent there is any issue in this scneario it's an accounting problem.  And maybe a future liquidity problem.

It is entirely believable the entities involved have clear audit opinions and approved accounting policies that permit these activities.

There are obvious-but-confusing inconsistencies all over the Tether-Deltec story.  Wouldn't it be supremely ironic if the real story was simply a case of turbo-charged fiat financial engineering.