In January I had the pleasure of participating on a panel on "The New Financial System: Decentralized?" at the Blockchain Economic Forum in Davos, Switzerland. (Video of the panel is here.) The panel was preceded by a talk by Prof. Nouriel Roubini of the NYU Stern School of Business (video here; all quotes below are my own transcriptions). In his talk Roubini made remarkable claims for what he called a "central bank digital currency."
A terminological clarification is immediately needed: the "central bank digital currency" that Roubini and other economists advocate is not in fact a currency. Their proposals are for transferable account balances, not for media that circulate peer-to-peer like coins or paper notes without requiring any new ledger entries. Roubini himself points out, no doubt correctly, that central banks have no incentive to issue a digital cryptocurrency or a token that would pass anonymously or pseudonymously in peer-to-peer transactions, not going through the interbank clearing system but rather validated by a distributed-ledger system.
Because the proposals for transferable account balances make "digital currency" a misnomer, I suggest that we more accurately call them proposals for "central bank retail accounts" or CBRA.
The primary version of CBRA that Roubini discusses would allow ordinary households and businesses to open checking accounts on the central bank's balance sheet. (At present only approved financial institutions and the US Treasury have accounts on the books of the Federal Reserve System.) In his words:
Suppose you create a digital central bank currency. Effectively what does it mean? … It's a centralized ledger where all the transactions occur. … [E]very individual has also an account with the central bank and therefore access to the balance sheet of the central bank the same way Citibank or JP Morgan or any other commercial bank does. … So if I had to give you money, I could do it by transferring through the balance sheet of the central bank … the same way the banks do in the interbank system."
I put aside for now one objection to CBRA: If checkable deposits become central bank liabilities, the loanable funds they provide will no longer be allocated by the competitive private banking system, but rather by the government, which is a recipe for inefficiency and cronyism. I will assume, as Roubini at one point suggests, that the Federal Reserve will conduct wholesale auctions to lend the funds back to the commercial banks. We can then focus only on the provision of checkable deposit services.
What is the case for having the central bank become a retail payment service provider? Roubini continues:
It would be a completely efficient system. It would be almost virtually costless. It would be safe. … It would reduce a lot the transaction costs. … [Y]ou can do it for free with the central bank. And of course it will dominate the deposits that individuals have in their private commercial banks.
He adds that the central bank would provide "much more efficient, zero cost, service." Michael Bordo and Andrew Levin have elsewhere similarly asserted that CBRA would be "a practically costless medium of exchange."
But this case rests on a rather obvious non sequitur: Central banks are efficient at wholesale payments, therefore they would be efficient at retail payments.
Even if the major premise is true (it is actually doubtful that nationalizing the interbank settlement system improves its efficiency), it doesn't follow that central banks have a comparative advantage in retail payment services, i.e. can provide transferable deposit accounts to individuals and businesses at lower cost than commercial banks do, much less that they can provide such services at near-zero cost. Zero-cost service is simply a fantasy.
Managing retail accounts in a way that satisfies users, as we see in a competitive environment where providers have every incentive to be cost-efficient, is far from zero-cost. Commercial banks invest capital in branch offices and ATMs. They employ tellers, customer service representatives, and "phone bankers" for such tasks as processing account applications (including identity verification), dealing with cash and foreign exchange, answering customer questions, and resolving disputes. Central banks have little experience at providing such consumer services. According to the Bureau of Labor Statistics, depository institutions in 2016 employed 444,000 tellers and 120,700 customer service representatives. Meanwhile the BLS estimates that the Federal Reserve in 2016 employed about 100 tellers and 300 customer service representatives. If we assign half of the commercial bank customer service employees to the deposit-gathering side, and then conservatively multiply by 0.165 because checkable deposits were 16.5% of the US banking system's total deposits (M2-M1+checkable deposits) in December 2016, we get 19,915 customer service reps dedicated to servicing demand deposits. Add the tellers, and we get an estimate of 463,915 US commercial bank employees dedicated to handling retail payment accounts in all banks nationwide.
Implication: if the Federal Reserve were to take over the provision of retail demand deposits, it would have to hire more than 400,000 employees away from commercial banks to give the same level of consumer service as the banks currently do. Note that the profit motive, together with the weeding-out of loss-making firms, impels commercial banks not to add employees inefficiently, beyond the point where the additional services they provide are as valuable to consumers as they are costly to the bank. The Federal Reserve faces no such motive or discipline. Presently the Federal Reserve System has around 20,000 employees. If my estimate is in the ballpark, taking over retail payments would be more than a twenty-fold expansion of its workforce. Even if the estimate were only five-fold, the view that the nationalization of retail demand deposits would result in a "virtually costless" system that could readily provide the equivalent or better services "for free" is a fantasy.
Not all commercial banks incur equal costs per checking account dollar, of course. Online banks presumably have lower costs than brick-and-mortar banks. But online banks also provide less service. Customers sort themselves into the arrangements they prefer: more service at higher cost, or less service at lower cost. If central bank retail accounts are online only to reduce expenses, they will not be able to provide the level of service for which many customers are demonstrably willing to pay. Roubini's prediction that CBRA will "dominate" all other payment options – that all customers will voluntarily move from commercial to central bank accounts – seems inconsistent with the relatively small market share of online checking accounts.
The most charitable reading that I can make of the claims about retail payments by central banks being "virtually costless" or "practically costless" is that these claims apply to the processing of the marginal deposit transfer. But when considering a switch from one system to another, of course, we need to compare the total costs of the two systems. Infra-marginal costs of setting up and maintaining accounts cannot be ignored. The marginal deposit transfer in our current system is also nearly costless. That does not obviate the fact that capital- and labor-consuming services are required to attract and keep checking account customers.
For those who are not ready (due e.g. to efficiency or privacy concerns) to nationalize retail payments, Roubini suggests an alternative model, which he attributes to Christine Lagarde of the IMF: retail demand deposits would remain at the commercial banks, but transactions involving special CBRA deposits (presumably requiring something like 100 percent reserve backing on the central bank's balances sheet) would enjoy immediate settlement. Like Bordo and Levin, Lagarde calls this a "public-private partnership." Lagarde concedes that full nationalization "might ironically stifle innovation" by eliminating competition, so that the best system might be one in which "financial institutions and start-ups are free to focus on what they do best—client interface and innovation."
This form of "central bank digital currency" – one which is neither a currency nor issued by a central bank – is less of a threat to efficiency and privacy, provided that consumers are free to remain with commercial banks. Federal Reserve CBRA deposits would have to offer an attractive mix of services (no branch offices, no ATMs, but rapid payment clearing) to attract consumers. In principle the Fed would only compete fairly, not compel. It would be crucial but difficult to ensure, in such a system involving an entity not disciplined by profit and loss, that the prices the Fed charges – and the interest rate it pays to retail account-holders – fully reflect its costs of providing retail services, and are not subsidized.