The World Economic Forum, which hosts the annual Davos meeting in Switzerland, recently published a survey about central bank experimentation with blockchain technology. Dozens of central banks around the world are investigating whether distributed ledger technology has a role to play in filling central banking functions.

The most interesting application that the WEF report mentions is central bank digital currency, or CBDC. Right now, the only type of money that central banks provide for us regular folks is cash. A CBDC would allow us all to get our hands on a digital version of central bank money.

So, will we one day be making online purchases with a digital version of central bank-issued cash? Some central banks are pushing hard in that direction, but blockchain may not be the technology that gets them there.

The idea for CBDC was inspired by the sudden emergence of bitcoin. This link may be surprising. After all, bitcoin is anarchical whereas a central bank is the opposite of that—staid, bureaucratic, and centralized. But there are some elements of the technology undergirding bitcoin that naturally transfer over to the day-to-day business of central banking.

Bitcoin is decentralized and censorship resistant. But so is one of the key products provided by central banks—cash. Anyone can use a banknote. When I pay you with a $20 note, we don’t have to ask the central bank for permission prior to transacting. Decentralized systems like cash and bitcoin are resilient. A centralized payments system may stop working because of natural disasters or glitches, but a $20 bill still works fine.

So, by picking and choosing various elements of bitcoin technology, it might be possible for central banks to provide a digital replica of cash—decentralized, robust, and available to all. But unlike bitcoin, which is notorious for its price volatility, the value of these tokens would be stable since they are backed by a central bank.

Anyways, that was my thinking back in 2013 and 2014 when I wrote a few early blog posts about a central bank-managed blockchain. Back then I called it Fedcoin. Other bloggers including Sina Motamedi and the Federal Reserve Bank of St. Louis’s David Andolfatto also tried to flesh out the workings of a central bank version of bitcoin.

Central bank research departments converged on the idea in early 2016. In a speech early that year, the Bank of England’s Ben Broadbent talked about putting “reserve deposits on a distributed ledger,” thus coining the term CBDC. Later on the Bank for International Settlement (BIS) and the International Monetary Fund (IMF) would wade into the debate, as would central banks such as Sweden’s Riksbank, the Bank of Canada, and the Monetary Authority of Singapore (MAS).

By 2017 a conceptual distinction had emerged between two different types of CBDC. Wholesale CBDC would only be used by banks to make payments to other banks. Retail CBDC would be more like cash, available to all.

Wholesale payments systems are foreign to most of us since we don’t directly participate in these mechanisms. Commercial banks make millions of payments each day amongst each other using the central bank’s wholesale payments systems. These systems are run from a single hub, typically with a physical backup located elsewhere. For instance, the Federal Reserve’s wholesale payments system is maintained in a data warehouse in New Jersey. Should the New Jersey site go down the Fed has backups in North Carolina and Texas.

By 2017, a number of central banks had begun to pilot wholesale CBDC using blockchains. These projects included the Bank of Canada’s Jasper, MAS’s Ubin, and the South African Reserve Bank’s Khokha. These pilots have relied on permissioned blockchains. Whereas bitcoin is permissionless—anyone can sign up to validate bitcoin transactions—the various pilots have restricted participation to vetted members.

Vetting means that these pilots haven’t had to rely on proof-of work to verify transactions. Proof of work is a slow and expensive way of achieving consensus. By going the permissioned route, the South African Reserve Bank’s CBDC Khohka pilot was able to process 70,000 transactions per day, or 9.72 transactions per second, matching the daily throughput of its centralized wholesale system SAMOS. This throughput was achieved while also bringing a “higher degree of resilience” given that dependence on a “single point of failure is removed and each node additionally acts as a backup of the ledger.”

The retail version of CBDC hasn’t attracted as much attention as the wholesale version. Sweden’s Riksbank and Uruguay’s Banco Central del Uruguay (BCU) have been at the forefront of investigating the potential for retail CBDC. But neither is too keen to implement a digital currency using a blockchain. The BCU piloted the e-Peso between 2017-18. Over the course of the pilot, Uruguayans made around 23,700 person-to-person transactions with e-Peso without experiencing any major down time. But the e-Peso pilot relied on a centralized set of service providers to manage the program, not a distributed ledger.

The Riksbank is pushing forward on a potential e-Krona, but in its most recent report it noted that it is “not currently appropriate” to use a distributed ledger, given that this technology cannot “sufficiently handle large transaction volumes in an adequately efficient manner.”

So retail CBDC, an idea originally inspired by bitcoin, couldn’t look any more different from bitcoin. With the decentralization stripped out, e-Peso and a potential e-Krona are just government versions of PayPal.

And maybe government PayPal is what we need. I recently caught up with David Andolfatto, at the Federal Reserve Bank of St. Louis, to chat about some of the developments in CBDC since we first blogged about the idea five years ago. “I now don’t see any point at all in providing central bank money in a decentralized manner,” he says. More important to him is the potential for a new CBDC to shake up oligopolistic banking systems. Based on his research, Andolfatto finds that by providing the public with a credible outside option, the presence of a CBDC could force banks to increase the interest rates they offer bank account customers. These improved rates would coax the unbanked out of cash and into the banking system. None of this requires a blockchain.

Not all central banks have gone the PayPal route. Two tiny central banks in the Caribbean have decided to test a blockchain-based retail CBDC. The Eastern Caribbean Central Bank (ECCB), which serves as the currency issuer for eight island nations including Grenada, Saint Kitts and Nevis, and Saint Lucia, plans to pilot the digital version of the East Caribbean dollar, or DXCD, on a permissioned blockchain. The Central Bank of the Bahamas, which currently issues the Bahamas dollar, is running the second pilot, Project Sand Dollar. It will likely rely on a permissioned blockchain as well.

A pattern pops out right off the bat. The Bahamas has some 700 hundred islands, thirty of which are inhabited. The ECCB governs eight island economies, some of which are in turn collections of smaller islands. Given the geographical fragmentation of these places, and their susceptibility to being disrupted by weather, it may make sense to use a distributed technology. That way, if one island is knocked out by a storm, the system can continue to work.

Meanwhile, on the opposite side of the planet, the Marshall Islands is developing its own blockchain-based national currency, the SOV. The Marshall Islands doesn’t have its own currency. It depends on the U.S. dollar. Unlike the Central Bank of the Bahamas and the ECCB, it doesn’t intend to link the SOV to traditional currency. Rather, the price of SOV will float according to the whims of the market, much like bitcoin. So unlike the DXCD or B$, the purchasing power of an SOV is likely to fluctuate wildly.

We don’t yet know whether decentralized CBDCs are a technological dead-end. Wholesale CBDC is being rigorously tested by some of the world’s most reputable monetary authorities. But as the WEF notes in its report, none have signed off on blockchain-based infrastructure given its “salient risks and limitations.” Existing wholesale payments systems are already incredibly effective. Blockchain-based versions will have to show they are capable of matching throughput rates of the incumbents, while also providing new features.

As for blockchain-based retail CBDC, the only takers to date are minuscule island nations. This may indicate there is a case to be made for adopting a distributed ledger format in certain geographies. Or it may mean the product is dead on arrival.