Tuesday, 9 August 2016

Birth of a bank

I have been following with some amusement the death and resurrection of the Bitfinex exchange. Bitfinex was forced to suspend trading after losing nearly 120,000 BTC in a hacking on 2nd August. But instead of going into liquidation, as we might expect, it reinvented itself - as a bank.

Bitfinex had been doing bank-like things ever since its creation in 2012. It accepted deposits from some of its customers, and lent out those deposits to other customers - with the depositors' permission, of course. However, as long as customers controlled their own wallets, and default by a borrower rebounded directly back to the lender, Bitfinex could simply call itself an exchange, or a peer-to-peer lending platform.

Bitfinex kept customer money in individual wallets under lock and key - three keys, to be precise. Two of the three keys were needed for funds to be moved from the account (or two of three digital signatures, if you prefer).

One key was held by Bitfinex, and a second was held by a trusted custodian and signatory, Bitgo. For wallets containing borrowed funds and belonging to a US resident, the third key was held by the customer. For all other wallets the third key was held in cold storage under Bitfinex's control.

Clearly, since they did not control the keys, the customers did not have control of their funds. They could only move them by giving instructions to Bitfinex or Bitgo. But Bitgo would only operate its key on the instruction of Bitfinex: it would not accept separate instruction from the customer. So in practice, customers could only move their funds if they instructed Bitfinex to do so. This is already looking much more bank-like, isn't it?

But wait. Bank deposits are loans to banks, and depositors are senior creditors. In the event of insolvency, senior creditors are only entitled to a share of the assets - they don't have an automatic right to recovery at par. This applies even for insured deposits, since insurance is separate from the liability itself. In insolvency, all depositors may take a haircut - but insured depositors are then made good by their insurance. In conventional banking systems, deposit insurance is a public good, though it may be limited to smaller deposits. It exists primarily to prevent bank runs, and secondarily to mitigate the damaging effects of bank failures on more vulnerable people.

In contrast, in a custodian system, customers have not lent their money - they have placed it in a safe deposit box. They have every right to expect that even in the event of insolvency, their money is safe. This is why regulators like the CFTC insist that client money must be segregated from the broker's or exchange's own money. Losses must be born by shareholders and creditors, not by customers.

Bitfinex's system was supposedly custodian. But the fact that customers did not have control of their own funds suggests that it was not. After all, who ever heard of a safe deposit box to which the customer can only gain access with custodian permission, but custodian staff can raid at will?

And raid it they did. No, I don't mean the hacker - though that was a raid on certain customer accounts. I mean Bitfinex management.

Following the theft, Bitfinex management decided unilaterally to "socialise losses". It imposed a 36.067% haircut on all customer balances. It also turned out that Bitgo, the custodian, had no insurance against customer losses. So all customers suffered a loss. In effect, Bitfinex management raided customer deposit boxes to cover the exchange's losses. Importantly, they did so WITHOUT customer agreement. They treated depositors as if they were lenders to the exchange itself, not lenders to other customers.

On the borrowing side, Bitfinex does not allow customers to withdraw funds from their wallets until all borrowings have been cleared. It does not allow them to borrow more than 70% of the value of a short sale. And it will forcibly liquidate balances in order to settle loan obligations.Yet even with such coercive provisions, it seems that Bitfinex was in the habit of absorbing loan defaults rather than passing them on to the lending customers. So it also treated borrowers as if they were borrowers from the exchange itself, not from other customers.

It's all looking highly bank-like, isn't it? But wait. We have credit intermediation and maturity transformation, true. But for Bitfinex to be a bank, not just a peer-to-peer lender, we need a third element. Fractional reserve lending - or as it should be called, money creation.

As long as Bitfinex was only trading existing currencies, there was no fractional reserve element. Deposits were lent out to borrowers, and no money was created. But that has all changed now. As part of the bail-in of depositors, Bitfinex created its own currency.

The haircut of customer balances exchanged existing currencies (which Bitfinex can't create, but needs in order to settle its debts) for newly created "tokens" called BFX. These bear no interest, are pegged at par to the US dollar, and will (hopefully) be tradable. They have no maturity date, but they can be redeemed by their originator. So they are zero-coupon, perpetual, callable debt securities. Possibly, in the future, they could be shares.

Now, let's remind ourselves of the characteristics of a fiat currency. Fiat currency consists of "tokens", which usually bear no interest* but are tradable. The tokens have no maturity date, but can be called in by the issuer at any time. So fiat currencies are zero-coupon, perpetual, callable - well, are they debt securities, or are they equity shares?

But, I hear you say, fiat currencies are produced by central banks out of thin air, aren't they?

No. They are created by banks. Any fractional reserve bank can create fiat currency. The vast majority of fiat currency in circulation is created by commercial banks in the course of lending. Most countries choose to restrict the issuance of notes and coins to central banks or government. But in Scotland and Northern Ireland, banks issue their own banknotes, which are pegged at par to sterling.

Bitfinex has been lending on its own account. I'm sorry, but it has. And now it is creating its own IOUs to cover the fact that it doesn't have enough reserves to meet its obligations. These IOUs are currently worthless, of course, since Bitfinex has currently no intention whatsoever of honouring the notional convertibility to US$. But the customers either have to accept them or force Bitfinex into liquidation - which would potentially leave them worse off. The best the customers can hope to do is create some value by sticking with Bitfinex and trying to create a market in BFX.

And of course, if they succeed, there will be absolutely nothing to prevent Bitfinex doing more own-account lending and issuing more BFX to fund it.

Bitfinex has become a bank.

Related reading:

You've been buttfinessed - BitMex Blog
I am a bank
The IMF proposes the death of banking
Of course Scotland can use the (Scottish) pound - Forbes

Image from Positive Money.  

* In fact many forms of fiat currency do bear interest. But of course Bitfinex could pay interest on its tokens. That wouldn't really make any difference to what is going on here.


  1. It might be better to move away from the term "Fractional Reserve Banking" because it does not exist except in economics text books as misinformation, producing even more economists who are clueless about where money comes from.

    They go around with the text book 10% example in their heads and models, with deposits having to be taken in before debt can be handed out.

    1. agreed , fractional reserve banking is a myth. In UK and Australia and Canada for example there are no mandated reserve requirements.

  2. Given the libertarian sentiments underlying the emergence of crypto currencies, might not your analysis gain further bite by developing the distinction between state issued fiat and commercial bank issued fiat. I have a put option on a deposit to redeem it for currency/reserves and I have a put option on currency/reserves to redeem it for my tax obligation. The whole rationale of crypto systems is to abolish that final put but still maintain a stable system. It doesn't work. It collapses back into a chartal base. And there you have the whole (Minskian) history of financial innovation.

    1. Could you maybe repeat all of that in a language and form that is accessible to mere humans?

    2. There is a constantly evolving and shifting hierarchy of monies. At one end is the 'thing' the state accepts to extinguish a tax obligation. At the other is whatever you and I can get other people to accept as payment. Or as Minsky said, anyone can create money, the difficulty is getting other people to accept it. As the mode of production evolves, new forms of speculative 'Finance' are pulled into existence to support it, creating cycles of financial instability. The state shapes those cycles by altering the eligibility criteria for swapping new forms of money for its money at the central bank. The article is describing an example of this process in action for crypto currencies, which is mildly ironic, as they are created to be freestanding systems.

    3. By which process does the state alter the eligibility criteria for swapping new forms of money for its money at the central bank?

      The Central Bank in pursuit of its policies will hold forex and also make currency swaps with other Central Banks to ensure sufficient liquidity in international money flows.

      I must have missed something big if the Central Bank is also involved in exchanging new forms of money into the sovereign currency. This surely only happens outside the banking system through private companies prepared to deal in the market and make an exchange rate.

      Or am I missing something else?

    4. there is no such thing as an "eligibility criteria for swapping new forms of money for its money at the central bank " That exchange occurs between private sector agents or small government bodies.

    5. Quite agree Dinero. Take TARP for instance, trivial really, nothing to see here -move along move along

    6. They were not swapping liabilities they were swapping assets.

    7. In TARP the central bank did not swap "new forms of money for its money at the central bank " what it did was it swap commercial bank assets for its money. There was no "swapping of money"

    8. Who is 'they'? One side or both?
      How were the assets originated?
      Money is relation not a substance.

    9. "They" is the commercial banks and the central bank

      The assets were originated by contracts with the customers of the commercial banks

      "money is a relation not substance" is that a Haiku poem , whether it is or it is not, its not par of the discussion.

  3. In no particular order:
    1.The CB scheme to which commercial and universal banks must belong. The 'shadow sector' joins in crisis moments (accepts regulation)
    2.The window/OMO/Repo system. Discount rates and eligibility criteria are published setting up schedules of liquidity across asset classes (private and public)
    3. 'Unconventional' balance sheet ops with fiscal effects via levels of remuneration to treasury. E.g TARP, QE tranches aimed at private paper, CB lending support schemes aimed at lowering borrowing rates etc

    1. Forgot to mention deposit insurance under 1

    2. In a nut shell they make their thing more or less equivalent to your thing for a price, whether over a fixed or rolling term or for good.

    3. Or they watch and let you die

    4. Heh Hugo, I'm shocked - how could you forget deposit insurance?!

      Being serious, though, your point 1 contains a really important insight which was played out in 2008 and is just being played out again in Bitfinex. That is the fact that shadow institutions voluntarily become regulated institutions in crises. In 2008, saw unregulated SIFIs such as Goldman becoming BHCs in order to qualify for (then unlimited) Fed support and deposit insurance. Now, we see Bitfinex issuing its own currency but pegging it to the US$ - which creates an implied Fed guarantee for BFX, assuming that BFX will eventually be convertible as Bitfinex say. Someone has just pointed out on Reddit why this might be attractive to investors. The US$ peg is crucial.

      Reddit thread is here (sorry, links not live in comments, you will have to cut and paste): https://www.reddit.com/r/Bitcoin/comments/4x2tfa/possible_reason_for_bfx_tokens/

    5. Oh, and of course Bitfinex is now compliant with the Dodd-Frank act, which it wasn't before.....

    6. Thanks Francis
      It'll be interesting to see if Bitfinex's response to its misfortune sows the seeds for its success. Many great stories start with a theft.

    7. Francis.

      Defending currency pegs requires enormous firepower and is a road that has seen many crashes. Who would/could ultimately defend any peg to the USD?

    8. Ultimately, Bitfinex must defend that peg itself, unless it becomes a regulated bank. However, as these tokens are perpetual it won't have to defend the peg for an infinitely long time. These tokens will trade at a hefty discount. Personally I think the discount should be 100%, because I don't think Bitfinex should get away with this theft. But I suppose enough people will believe their promise of eventual par convertibility to give the tokens a market value. There's always a market for lemons.

    9. Hi Francis,

      Thanks for the answer.

      That's what I thought. I was a bit confused by your reply above mentioning an implied Fed guarantee. I suppose it can be implied but doesn't mean it exists.

  4. I enjoyed the sarcasm.

    "Birth of a bank"

    "Bitfinex has been lending on its own account. I'm sorry, but it has. And now it is creating its own IOUs to cover the fact that it doesn't have enough reserves to meet its obligations. These IOUs are currently worthless, of course, since Bitfinex has currently no intention whatsoever of honouring the notional convertibility to US$."

    "These tokens will trade at a hefty discount. Personally I think the discount should be 100%, because I don't think Bitfinex should get away with this theft."

  5. Did any one speculatively trade Brexit? How did it go?