Believe those who are seeking the truth. Doubt those who find it. Andre Gide

Saturday, January 29, 2022

Some Thoughts on the Fed's CBDC Report

Economists within the Federal Reserve System have been musing about central bank digital currency for a while now. For example, yours truly provided a few thoughts on the possibility way back in 2015 (see here) and most recently here. But the views of individual Fed researchers are simply our own personal views. What people are really interested in is an official view--namely the view of the Board of Governors of the Federal Reserve. At long last, their report is now available here). 

The report makes it clear that the Fed has no immediate plans to issue a CBDC. The main purpose of the report is to provide some background information, discuss the potential costs and benefits of a CBDC, and solicit feedback from the general public. It is a very nice educational piece. There was nothing really new in it for people who have been following the discussion over the past few years, but there was one thing I found interesting (and potentially important). The interesting part, I think, has to do with their definition of CBDC:

For the purpose of this paper,  a CBDC is defined as a digital liability of the Federal Reserve that is widely available to the general public.
I used bold font to highlight the important part in the passage above. Most of the digital money Americans use today consists of dollar credits in transaction accounts that reside in the broader bank sector (which includes non-banks providing digital payment services). This form of digital money is technically a liability of private sector agencies, not the government. Is this an important distinction to make? I'm not sure--I suppose it depends on what one thinks is important. 

For one thing, the Board's adopted definition appears to rule out the "wholesale CBDC" proposal that some have advocated for (including myself). Wholesale CBDC (also referred to as "synthetic CBDC" or "sCBDC") is essentially a narrow-banking proposal (which has a long history in the banking policy debate). 

In my writings, I contrasted "wholesale CBDC" with "retail CBDC," the latter of which I interpreted as a product that is both a liability of the Federal Reserve and a payment service managed by the Federal Reserve. I thought this latter property was a bad idea--the Fed should probably not get involved with the challenges of retail banking. I missed the possibility of separating the legal status of digital money from the agencies responsible for managing digital money accounts. As George Selgin explains well here, the Board's definition of CBDC means a privately-intermediated liability of the Federal Reserve (George calls this iCBDC to distinguish it from sCBDC). 

Speaking for myself, I'd be happy with either an sCBDC or an iCBDC (assuming they are properly designed). My main concern was over the prospect of a government agency operating a large retail business. Both sCBDC and iCBDC are intermediated products--we can let the private sector manage the day-to-day operations of processing payments. George, however, believes that sCBDC should be preferred to iCBDC for reasons that he spells out here (I'm not sure I share his concerns).

My own view is that the distinction between sCBDC and iCBDC will be of interest mainly to lawyers and regulators (John Kiff points me to this IMF report). From an economic perspective, the two products appear to be very close--if not perfect--substitutes (again, assuming their design is optimized). The practical difference between a Federal Reserve liability and a private liability fully-insured by the government seems almost non-existent to me. But I'm willing to be persuaded otherwise. 


  1. "My own view is that the distinction between sCBDC and iCBDC will be of interest mainly to lawyers." You err, David, in assuming that safety is the sole issue, so that so long as the two arrangements are similar w.r.t. it, other differences may safely be set aside as of minor significance. But you neglect--as monetary economists are too prone to neglect--the usual reasons why economists prefer competition to monopoly (for sCBDC is actually competitive DC, while iCBDC is monopoly DC). These have to do w/ such things as allowing for product variety, achieving static efficiency (encouraging efficient input use to maximize profit) and encouraging dynamic innovation. These differences between monopoly and competition are surely among those that ought to concern economists, whether they concern lawyers or not! Indeed, they are things only overlooked if one approaches the problem, not as an economist should, but as a mechanic or engineer might, equipped with the convenient assumption that all the "best methods" are known to monopolist and private sector firms alike, instead of having to be discovered and tested; and that there is no need for a profit and loss incentives to encourage efficiency and innovation, because the beneficent technocrat has no need for them. The same assumptions make up what Hayek calls "The Fatal Conceit" underlying plans the wholesale substitution of central planning for competition. Writ small the conceit is not so fatal. But it can still do plenty of harm.

    1. Thanks, George.

      I'm not sure I understand your "monopoly vs competition" distinction. Monopoly over what? In either case, the Fed maintains a monopoly over settlement balances. So you can't be claiming that sCBDC avoids this dimension of monopoly. All that's left are the protocols that revolve around secure messaging, privacy, etc. But this will be true of both types of CBDC as well. Government can let innovation take place in the space whether we have sCBDC or iCBDC. Why does one form *necessarily* inhibit innovation (promote "monopoly"). I think you tried to explain this in your article, but it didn't sink in for me. Can you elaborate here or, if not, give me a little while to re-read your piece.

    2. In the sCBDC, and unlike the case of iCBDC, the retail DC media are distinct from the settlement balances themselves: think of old-fashioned bank notes backed by 100% gold reserves, or, for that matter, of extant stablecoins: although the last are in fact backed by various assets today, even if all were backed by fiat reserves, they would be distinct coins with distinct features, with each issuer capable of improving its distinct product along dimensions apart from the reserve backing itself.

    3. Let me elaborate a bit on the last example. Take all the USD stablecloins in existence, and imagine all agree to abide by, and that the Fed allows them to abide by, a 100% master reserve balance backing requirement. As most are presently fully backed by a combination of ordinary bank deposits, Treasury's, and commercial paper, there is no reason why the new rule should clash with their basic business models, though of course it is likely to alter their earnings. (If we assume further that Fed balances earn the IOR rate, even this change needn't be all that substantial for many). Nothing would stop them from continuing to innovate, and from having to do so to remain competitive.

      Now imagine that the Fed instead abolishes all the stablecoins in favor of having an iCBDC monopoly. Bear in mind that, though private firms manage customers' holdings of iCBDC, it remains a Fed product, not their own, over which it presumably would exercise quality control.Hence my comparing the agent banks with "franchisees," though with the nontrivial difference that in this case the franchiser cannot possibly fail, and so has that much less reason to be efficient or to innovate aggressively. This is a very different market structure. My position is that it will different from the competitive case in the usual ways.

  2. Hi David,

    I think the main difference between synthetic CBDC and the intermediated CBDC lies in the technical level. The synthetic CBDC requires two layers: The central bank money layer and the private lability layer. The intermediated CBDC only requires the central bank money layer which increases efficiency and decreases database and information transmission costs. A transfer of money between two customers who use different intermediaries requires four transactions with synthetic CBDC: One in each of the intermediary’s private ledgers and two in the central bank ledger. The intermediated one needs only the two transactions in the central bank ledger. This would also decrease the exchange of communication and transaction time because there are only three parties involved while synthetic CBDC involves four parties.

    Overall, intermediated CBDC is more efficient and therefore preferable to me. However, you can argue that synthetic CBDC can be implemented more easily. The system can be implemented with the existing technical infrastructure of central banks and e-money providers while the intermediated CBDC needs at least a new interface to the central banks plus a technical backend and frontend solution for the intermediary. This could be an option if this weights out the gains of the increased efficiency of an intermediated CBDC.

    Furthermore, I’m a bit confused how the usage of CBDC terms here differs from the usage in the broader CBDC cosmos. Isn’t iCBDC the abbreviation of the “indirect CBDC” approach (= synthetic CBDC) introduced by Kumhof and Noone in 2018? And isn’t what George Selgin means by iCBDC the same as the “Hybrid CBDC” introduced by the BIS?

    From a legal perspective I’m more or less satisfied with the following classification of CBDCs:

    Direct CBDC = Customer has access to a liability of the central bank + Customer has a legal relationship with the central bank to access its funds (e.g. bank account contract) which also regulates the issuance of the central bank money.

    Hybrid CBDC = Customer has access to a liability of the central bank + Customer has a legal relationship with a third party to access its funds (e.g. payment service contract) and a legal relationship with the central bank on the issuance of the central bank money.

    Indirect CBDC = Customer has access to a liability of an intermediary which is fully backed by central bank money (e.g. Synthetic CBDC/ Fully backed e-money/ iCBDC) OR Customer has access to a liability of the central bank + a custodian provides access to its funds (custodian contract).

    I wrote a post and paper where I go into more detail: