Long Take: The Federal Reserve Payment Processor, its $100B of revenue, and whether blockchains and L2s can match the value it settles

Hi Fintech futurists --

Another chunky week! Just a polite heads up about the Coinbase S-1 dropping (with figures largely similar to our discussion last week) …

… the super interesting developments in the NFT market with Sorare (football NFTs) raising $50 million for a 12 person team

and Grimes launching her own NFTs on Nifty Gateway.

Instead, let’s do some math homework. It’s good for you:

  • The Federal Reserve money movement system broke for several hours. We look deeply into its volumes and transactions, and value it like a Fintech unicorn.

  • The Ethereum ecosystem is throwing around as much volume in settlement as the Fed check processing system. We explore scalability barriers and solutions.

  • Can eCommerce fit into our emerging infrastructure? We anchor to the market numbers in China and the United States.

For more analysis parsing 12 frontier technology developments every week, a podcast conversation on operating fintechs, and novel food-for-thought essays, become a Blueprint member below.


In Partnership with:

FINTECH MEETUP is a new event from the founder/former CEO of Money20/20 and Shoptalk! Get tickets here.


Long Take

Things break.

Sometimes the things that break are the US Federal Reserve ACH service, Check 21, FedCash, Fedwire, and the national settlement service. They were down for a few hours — discovered at 11AM on Feb 24th and still in trouble at 3PM that day. Everything is now up and running again.

There’s the obvious sort of glee about the gods we thought were invincible showing that they bleed. But that’s an animalistic take. Someone might say — look, Bitcoin and Ethereum have had 100% uptime since the launch of their networks. Someone might say that blockchain is the future of payments, and so on and so forth. First, that’s not accurate.

And second, it’s not very interesting, because the comparison is (for now) a strawman. Ethereum, which is the most likely candidate to move around the US dollar, given it already moves around $30 billion of USD, is still several hops away from competing with real time Fed payments.

But what is interesting (or rather educational) is a discussion along two dimensions — (1) what exactly is the Federal Reserve’s money movement infrastructure, and (2) what are the unlocking mechanisms that would allow blockchains to compete with initiatives like FedNow.

The Federal Reserve Payment Processor

Thanks to Rex Salisbury of Andreessen for pointing us to this 2019 annual report from the Fed. In particular, we jump to the quantitative statements to get our hands around the scale of the activities.

The Automated Clearing House is a batch payments engine that allows small-ish payments to come together and be transferred overnight. We see about $28 trillion of ACH volume for commercial entities, and $6 trillion for government, for a total of $34 trillion in this system. That translates to around 16 billion transactions per year, with average sizes ranging between $1,800 to $4,000 per transaction.

There is also still an absurd amount of check processing. Hard to imagine how we are going to be in a digital-first economy when there are still over 4 billion checks handled per year, with an $8 trillion annual volume, most of which are coming from commercial activity. The average check size is similar to the ACH payment — about $2,000 per instrument for commercial sources , $3,000 for government, and $300 for money orders (a check instrument for the unbanked).

The institutional transfer numbers are quite interesting too, especially for their much larger order of magnitude. We see about $1,000 trillion (or a quadrillion) of USD volume for securities and funds transfers, and think that the “Funds transfers” is what maps to the Fedwire service that went down. Wires are the super highway of American payments for banks, with the average transaction at $4 million in value and 170 billion in annual count. The average securities transfer is $18 million, with an annual total of 20 billion transactions. So these numbers must be pooling together all the high-priority assets as institutions send them across.

Hopefully we got our math right. If these numbers are wrong anywhere, send us a note.

Ok, so the Fed is moving all that serious money for the US banking system. Graciously enough, it is trying to do so at cost. All of the different divisions and services have a mandate to do a little better than break even, but not by much. So let’s take a look at how much it costs to operate this government infrastructure.

Checks cost us $120 million per year, so we pay $0.03 on average per check per year.

The ACH costs us $160 million per year, so we pay $0.01 on average per transaction per year.

Fedwires costs us $170 million per year, so we pay $0.80 (give or take) on average per transaction per year. Reminder that these are multi-million dollar transactions.

These numbers give us clarity of thought — the system as it stands is pretty inexpensive. Why then do banks charge $30 to $50 per wire to their retail customers? The issue seems to be not the central bank technology, but the brokerage / intermediation layer of the retail banks.

Since we are here, there is one more chart that is quite amazing for understanding the network of the Federal Reserve banks.

All that payments stuff might take a couple of hundred million here and there in cost. That is small potatoes. Peanuts. Tiny little atoms floating about.

The Fed system generates over $100 billion in revenue on about $45 billion of cost. This comes from the financial returns on SOMA (“System Open Market Account”). What’s that, you ask? Well — “the FOMC has authorized and directed the Federal Reserve Bank of New York execute open market transactions to the extent necessary to carry out the domestic policy directive adopted by the FOMC”. This is your government-run financial backstop.

Now, Apple generates about $300 billion of revenues. But it also has a lot more than 300 million customers. If we wanted to be cute, and look at SOMA as equivalent to a B2C fintech startup (i.e., enterprise value per user), $100 billion of revenue for 300 million people is $333 per customer.

At venture capital valuation levels of 20x revenue, the Federal Reserve is a $2 trillion fintech startup.

Unlocking the Crypto-native Payment Machine

Switching back to blockchains, Ethereum is now settling $26 billion per day. Let’s pretend that implies $10 trillion in transaction volume per year, or 400 million transactions ($25,000 average). For Bitcoin, the answer is $4 trillion per year.

If our order-of-magnitude arithmetic is right, then this looks like Ethereum is equivalent to check collection volumes, and Bitcoin is like government ACH.

The problem is that each transaction is taking a lot more than a few pennies to process. Transfer prices are hitting $10 for regular payments, and around $50 to $100 for more complex investment and lending actions (data here and here).

The industry is rallying around this problem, because it is an existential issue. A network is supposed to become better as more people use it, rather than more difficult and expensive with adoption.

In the words of crypto technologists, the main network is “Layer 1”, which retains the record of truth, and the additional scaling solutions are on “Layer 2”. That means that as the underlying programmable blockchain keeps chugging along, new mechanisms are attached to it, all with their own on and off ramps.

There are many hard things about this. One of the hard things is that if you use Ethereum as a computing platform (i.e., where software is written and executed), you need your Layer 2 to be compatible with that computation. There are 4 main paths to scalability in the market today: (1) the roadmap for Eth2, which is likely to respond to the strategic choices of the ecosystem, (2) state channels, which are like warp gates between known participants in the network, (3) side chains, which compute away from Ethereum and then re-anchor to it, and (4) rollups, which move computation and data off the main blockchain.

The names that keep coming up in this space include Optimism (just raised $25MM), Starkware, and Matic/Polygon. We also track SKALE, xDai, OMG Network, and Loopring. There are many other pioneering players making their way through grindy mathematics and game theory to try and make scaling happen. Of course, there is also the long list of Layer 1 competitors, aiming at de-throning Ethereum from its perch, including Solana, Binance Smart Chain, Avalanche, Cardano, and others that shall not be named.

The game theory of this competition is quite rough. The industry must coalesce around a main scaling solution (or a couple of main approaches), or risk splintering into dozens of separate off-chain networks that each have deposit gates with timed delays to be used. That’s an unfortunate user experience, and potentially costly.

Concluding with eCommerce

Let’s turn our final attention on the volumes of commerce. American point-of-sale commerce stands at $8 trillion in value, and $2 trillion in eCommerce (see the 2021 Global Payments Report by Worldpay from FIS). Those flows are fairly evenly distributed between credit cards, debit cards, and mobile wallets. This behavior is reflected in the $500 billion valuations of Visa and Mastercard, the core card networks. Both are trying to connect to, and potentially subsume, crypto-networks.

In China, point of sale commerce turns over at $17 trillion, and eCommerce is around $3 trillion. Over 70% of eCommerce is done through a digital wallet, like Alipay or WeChat Pay. Soon enough, that could be the Chinese DC/EP digital currency.

The Federal Reserve in the US is not standing still. It is hiring all the right people. It is working on the FedNow instant payments project. And it is exploring the preconditions for a general purpose Central Bank Digital Currency.

But is it moving fast enough?

Ethereum generated $760 million in fees over the last 30 days, putting it on a $10 billion annual revenue run rate. As we mention before, this is not helpful for the day-to-day usability of the network. But it is generating re-investment into the protocol’s infrastructure. We definitely need to sharpen our analytical tools for comparing these numbers to government payments software budgets. At the very least, it does feel like the technology can take a meaningful portion of both money movement and eCommerce by value alone.


For more analysis parsing 12 frontier technology developments every week, a podcast conversation on operating fintechs, and novel food-for-thought essays, become a Blueprint member below.