No mention of cross-currency payments/settlement which, given how Bitcoin is currently being used, is more relevant than bank transfers within the same country.
Richard Brown here - the author of the linked article.
Very good point re FX and CLS. I wanted to make the article as simple as possible so as to bring out the fundamental concepts (correspondent banking, deferred net settlement, real-time gross settlement, etc).
In so doing, I had to make some gross simplifications and omissions - the most obvious being ignoring FX but I also made a sleight of hand by implying that banks within a country use the correspondent arrangement I outlined.
However, I hope - even with these simplifications - the concepts came across clearly.
Richard, thanks again for the article. I posted a comment there regarding a recent goof up between two banks that resulted in "free money" for me. I'd love to get some insight into what might have really happened, and what, if anything, I might be able to do about it. Please feel free to reply here, there, or directly if you are interested.
I am instinctively against over-simplification of this sort of thing because, in my experience, the devil is in the detail and a little knowledge is a dangerous thing.
Well, I certainly hope nobody tries to build a system off the back of that blogpost!
But I think there is a lot to be said for helping people build useful mental models. My experience is that people either have no mental model for how money moves or have models that are just wrong.
So anything we can do to impart some insight has to be a good thing.... but you're right: a little knowledge can be very dangerous :-)
Not just cross-currency, but also cross-border payments.
I've implemented processing of cross-border payment instructions delivered via SWIFT and there's no way banks will STP a cross-border payment for $150 million. Large payments will always be subject to an authorisation review, at least where I work. That costs money and takes time.
Fundamentally, cash payments will always suffer from counterparty default risk without a central bank. How are you going to build an international central bank or central bank system? The best you'll see are regional efforts like TARGET2.
Digital currencies solve these problems without requiring ridiculous amounts of infrastructure to be created first. If we had to sit around waiting for banks to solve these problems, it would simply never happen.
This is fundamentally why I see digital currencies as the future, and why the hype is indeed warranted (even if perhaps people don't understand why).
Interesting how payments work in different countries. For example, in Australia when dealing in AUD, banks can participate in end-of-day multilateral netting. That is, all the banks' inflows and outflows to all other banks are netted and sent to the Reserve Bank of Australia (RBA).
The RBA is the banks' bank which holds bank's AUD denominated deposits in their separate accounts. Once the bank-to-bank net amounts have been calculated the RBA shuffles around a few numbers in databases to execute all the interbank transfers.
It sounds like in the USA banks have to have accounts with each other in order to affect transfers. The RBA effectively adds a layer of indirection and in some ways ensures banks are meeting their capital requirements.
Edit: I didn't read far enough. The UK has an impressive implementation of a Real-Time Gross Settlement system. My thoughts on the US interbank transfer system were coloured by a recent NPR Planet Money podcast (http://www.npr.org/blogs/money/2013/10/04/229224964/episode-...) where they compare the US and UK systems.
Isn't it the same as the original article description of deferred net settlement? I believe most country-local-currency systems are (or used to) run this way.
> This thought process motivated the creation of deferred net settlement systems. In the UK, BACS is such a system and equivalents exist all over the world.
Also, I can't think of a reason why they wouldn't do it this way.
> Also, I can't think of a reason why they wouldn't do it this way.
Credit risk and liquidity risk, as well as systemic risk exasperated by asynchronous transfers and time lag.
Banks do sometimes go bankrupt. While MFGlobal was not a bank, it did demonstrate difficulty in unwinding client funds. LCFIs [2] usually cross regulatory environments, and failures of financial institutions are numerous, the combination quite unsettling. Herstatt [3] in 1974 led to continuous settlement in FX, having received DEM but not paying out USD. There's quite a reason for sometimes not netting things.
+1 on the coursera classes, Mr. Mehrling is an amazing teacher and really knows the subject well.
I finished part 1 of his course and am close to finishing part 2.
Bitcoin resembles a lot more a one bank operation where accounts are settled immediately and at zero cost. Not an interbank, expensive, RTGS system.
I would also rather ask not how Bitcoin can squeeze into the banking system jigsaw puzzle, but how this puzzle can be optimized to be as effective as Bitcoin.
10 minutes of trust is a lot faster than the days and weeks of trust required for the current payment's systems with chargebacks and so on. I think coffee shops, etc will just accept transfers as soon as they see it on the network, just like how with small food purchases you don't need signatures.
The fraud risk of zero confirmation bitcoin transfers is a lot smaller than the current fraud risk that we have with the current payments system. If it's actually a large purchase, like a car then you can afford to wait until it's confirmed, and with things like laptops, wait a couple minutes for a couple of confirmations.
But transfering bitcoin is not instant and not free is it? Please correct me if I'm wrong but as far as I understood you will have to wait for your transaction to be included in a (few steps of the?) chain, and unless you pay a fee you might not get included for a while?
And is there not possible that in the future, if the amount of transaction increase as bitcoin becomes more popular, and the money you get from mining goes to zero, that transactions with fees will be the only ones accepted (at a reasonable rate)?
The transaction is propagated through the network in a few seconds. The bitcoin wallet will allow you to spend the funds received, even if they are not confirmed by the network[1]. You simply add the risk of a double-spend attack against your balance, if an attacker also sent coins to someone else.
It is up to miner discretion which transactions to include in blocks. If the fee included is zero, then many miners will pass on your transaction, and it may take longer to confirm the original transaction.
> But transfering bitcoin is not instant and not free is it? Please correct me if I'm wrong but as far as I understood you will have to wait for your transaction to be included in a (few steps of the?) chain, and unless you pay a fee you might not get included for a while?
Well, it's free to the transferers, because new coins are issued for work performed, and likewise, the new bitcoin compensates and incentivizes the efforts of the blockchain verification, so that balance should be even as well. Sounds zero-sum to me.
> And is there not possible that in the future, if the amount of transaction increase as bitcoin becomes more popular, and the money you get from mining goes to zero, that transactions with fees will be the only ones accepted (at a reasonable rate)?
Bitcoin is hobbled as a banking reserve system as it is deflation biased and fixed in quantity, making it extremely difficult to accommodate any crisis.
That is rather the point, isn't it? Without the vocal support of libertarians who like it precisely because it can't inflate or be manipulated, would it ever have become his popular?
This puzzle can't be optimised in principle. The whole network of fiat currencies and bank accounts is a web of promises, not hard money. Every transaction is another promise from one banker or person to another. And at the top of that scheme there are central banks printing new promises backed by promise to pay back with more promises. It's easy to abuse and it's done constantly in multiple ways: inflation, capital controls, withdrawal limits, censorship, haircuts, bail ins, bail outs etc.
Bitcoin will be just a place where people will exit en masse and trade in BTC completely, without relying on fragile and corrupted system that we have today.
That's like saying that instead of embracing a new invention called the "automobile", we should rather be optimizing the horse buggy. Why would we optimize an antiquated banking system to maybe/at-some-point match it with an already existing and better system? Where are the gains?
What would also be interesting would be an explanation of how this then ties in to the "headline" short term interest rate that the Fed, ECB, Bank of England etc. set.
All the ground work has been laid - and I think many people would be interested to know the mechanics behind what happens when the Bank of England "cuts rates" or "raises rates".
The only other place I've seen this explained well is at the start of the book "Pricing Money" by J.D.A Wiseman - but that's not available online.
The gist is that even though movements between banks tend to balance fairly well, they don't balance exactly. Therefore, banks can end up with excess reserves or a shortage of reserves, which they are then interested in lending out / borrowing, respectively. Or they might be interested in buying or selling other assets, such as bonds, to bring their excess reserves closer to zero.
When a central bank sets the interest rate, it's talking about the interest rate at which such short term loans happen. It does so by simply participating in the market until the interest rate is where it wants it to be.
I'd avoid setting too much stock in Modern Monetary Theory unless you have a decent grounding in the basic operation of the monetary system already. Whilst their explanations of the banking system are very detailed, very accessible and mostly correct, their monetary theory is very much a view of how a fiat currency regime could work as opposed to how it does work, and they wildly misrepresent basic macroeconomic concepts when it suits them to make a point.
(the most relevant example here is their model's assumption that governments can and do spend money into existence. Whilst this is theoretically possible in a fiat currency system, it's very different from the actual mechanisms in place, and entirely impossible with the Euro)
> the most relevant example here is their model's assumption that governments can and do spend money into existence
There are three points here. First of all, they are very clear that this only applies to monetarily sovereign governments (i.e., not Euro members, not state governments).
Second is their claim that directly spending money into existence is the more natural way of looking at things, and everything involving debt issuance is a voluntary add-on. That's not how things developed historically, but it does make sense when you think about how you would design a state-run currency from first principles.
Third is their claim that, as a first order approximation, even when all sorts of debt issuance dances are in place, the economy behaves as if they weren't and government simply spent money into and taxed money out of existence. [1]
The second point is admittedly a bit philosophical and subjective. You might care more about the historical account than the first principles account.
However, the third point is quite objective, and if you have a concrete example of how debt issuance actually makes a significant macro-economic difference that contradicts MMT, I would be highly interested (and I assume MMT economists would be as well).
[1] One way to look at it is this: Debt issuance is a red herring; what matters is the central bank policy of (usually) maintaining a positive interest rate. Debt issuance is only a way of shifting the cost of this policy from the central bank's books into the general government budget, i.e. it is a political that hides the true cost of monetary policy so that the public won't be as outraged about it.
In summary
- banks' liabilities to you (your deposits) are backed by assets. One of these is cash, but this doesn't earn the bank any return. Banks therefore prefer to hold much larger quantities of interest-earning assets like loans made to other parties, or government bonds (these are assets to a bank because they are promises by third parties to pay the bank money in the long term).
- For the same reasons as the payment clearing system is "netted-out" at the end of the day, the balancing of these banks' assets and liabilities is netted-out at the end of the day. If the risk-weighted assets of one bank doesn't match its liabilities - particularly if it doesn't have enough cash - it needs to borrow from another bank. If a bank has too much non-interest earning cash at the end of the day, on the other hand it will prefer to earn a low risk return by lending to another bank. This is the "base interest rate" we're talking about; the minimum amount of return a bank expects from making a very safe loan.
- Banks could earn a similarly low-risk and low-return by buying government bonds with their spare cash instead, so the interbank lending rate is closely linked to bond prices. The Fed, ECB or Bank of England can take advantage of this by buying and selling government bonds to keep the interbank lending rate at a target level.
- Why would they want to do this? Because the interbank lending rate (and bond prices) affects the rates for all other types of lending that banks do (the interbank lending rate plus a usually relatively large risk premium, plus a bit on top for profit).
- Why would they care so much about the price of lending? Because as the banking system's balance between assets and liabilities is only required to be eventually consistent, banks are given the right to create loans that far exceed the cash (and other banks' liabilities) previously deposited with them, provided they can borrow the funds at the end of the day. And because this credit creation is legal - and essential to financial markets' efficient operation - the only way the government can influence the total quantity of money circulating in the economy is by manipulating the minimum price of credit. As the total amount of credit created by the banking system far exceeds the amount of cash created by the central bank, the economy is very sensitive to the effect of small interest rate changes on demand for credit.
Minor quibble: banks do earn interest on reserve balances since 2008. The rate maintenance strategy was changed at that time to address cost instead of quantity.
"How the Economic Machine Works" was a really great explanation of the different ways a government can influence inflation and deflation: http://www.youtube.com/watch?v=PHe0bXAIuk0
Using todays technology SWIFT (or any other online messaging system for the matter) does not have to be expensive.
I did not study extensively the banking system, but I figured that the central bank was the missing part of the post right away. How did I do it? Because the central bank should do exactly that: CONTROL THE BANKS IN ORDER TO AVOID THEM GOING BURST without anyone noticing (rings a bell?). Now the fact that they NEVER do, is a nice topic for discussion, why exactly to we pay them? Just to guess what the right monetary policy for the next 6 months will be?
This post explains why sending money from a bank’s perspective might be expensive. But what it really shows is that it does not have to be if everyone was doing their job right (central bank included) and be held accountable when shit hits the fan.
As of today exchanging BTC (an asset anyway) to currency is expensive and not straight-forward. How exactly is someone going to exchange 150m BTC in USD/EUR without getting noticed? In what amount of time???
BTC is not optimal for this kind of transactions. An inflationary e-currency, widely acceptable and easily exchangeable would be one hell of an option, but good luck persuading people to use it if you re not a government.
Very interesting article. I work in writing some financial software (consumer facing) and I have never actually read about how the banking system works. Just the other day I was writing some code to make SWIFT transfers.
You're right - sorry. I drew the whole slide first and then subtracted pieces to support the flow of the story. It meant some parts were a bit squashed.... and you're right that, in isolation, the early diagrams look a bit confusing.
I doubt if bitcoin really resembles RTGS; for one thing there is no central agency involved. Or may be it does if you consider distributed network of nodes which does "proof of work" for your bitcoin transaction as a Central Bank
Exactly. My contention is that you can perhaps regard the entire Bitcoin network (of full nodes) as providing RTGS services for the Bitcoin currency.
The question, then, is: "is this sufficient"? Do we believe all Bitcoin transactions would be transacted over this system (i.e. over the blockchain) in the future or will we see aggregators akin to DNSs?
I see the work of Mike Hearn and others on micropayment channels as innovative examples of what a DNS for Bitcoin might look like.
This was a fantastic read, and very educational. I was most surprised to see it put as we are lending money to our banks. Thought there might be mention of FDIC but didn't see any.
Looks like a site with an Agenda (with a capital A) rather than something more neutral/academic/created to inform. I'm always wary of those kinds of groups if I'm trying to learn about something.
An overview of CLS is a good starting point: http://www.snb.ch/en/mmr/reference/continuous_linked_settlem...