Digital currencies, including potential central bank digital currencies (CBDC), have generated a lot of interest over the past decade, since the emergence of Bitcoin. The interest has only grown in recent months because of a desire for contactless payment methods, stemming from the coronavirus pandemic. In this post, we discuss a common distinction made between “token-based” and “account-based” digital currencies. We show that this distinction is problematic because Bitcoin and many other digital currencies satisfy both definitions.
Tokens vs. Accounts
It is common to make a distinction between account-based and token-based digital currencies, as in this report from the Committee on Payments and Market Infrastructures (on which some of us collaborated), this working paper from the Bank of Canada, this staff discussion note from the International Monetary Fund, and this note from the Bank for International Settlements, among many other sources. In all of these papers, the distinction the authors make is that an account-based system requires verifying the identity of the payer, while a token-based system requires verifying the validity of the object used to pay. This distinction adopts concepts presented in an article by Charles Kahn and Will Roberds, which was written before the current usage of the word “token” to denote a unit of a cryptocurrency became popular.
A pair of examples is useful to understand the distinction between token-based and account-based systems. The canonical example of a token-based system is currency. If I pay for a coffee with currency, the only thing the merchant needs to worry about is that the bill I hand over is not fake. If the bill is valid, then it can be used to make a purchase. In particular, the merchant doesn’t need to know anything about me. Other examples of store-of-value systems are gold or silver coins, or stored-value cards.
A good example of an account-based system is the Fedwire® Funds Service. If a participant submits an instruction to the service to transfer funds to another participant, the instruction must be transmitted in accordance with the Reserve Banks’ security procedures, which include access control features designed for the purpose of verifying that the instruction the Reserve Banks receive is an authorized one. Because these security procedures limit the likelihood of the Reserve Banks acting on an unauthorized instruction, they also limit the likelihood that fraud is facilitated using the service. This is why in an account-based system, establishing a process for verifying the identity of the would-be payer is important. Bank deposits, which can be used to make payments, are another example of account-based systems.
Neither system is perfect. Store-of-value systems are susceptible to counterfeiting (the U.S. Secret Service was created in 1865 to combat counterfeiting). This report by the U.S. Treasury states that the value of counterfeits in circulation is most likely around $70 million, with an estimated upper bound of $200 million. Meanwhile, the 2016 Bangladesh Bank cyber heist, demonstrates that account-based systems are not foolproof either. In this case, hackers reportedly compromised the central bank of Bangladesh’s computing systems and submitted instructions to the Federal Reserve Bank of New York to pay unauthorized parties, resulting in transfers of $81 million from an account belonging to the central bank of Bangladesh without its knowledge.
Are token-based and account-based digital currencies distinct?
Use of the token-based and account-based terminologies for digital payments is problematic because it does not create mutually exclusive categories. If one sought to specify a taxonomic hierarchy of digital payment methods, these terms would not partition members of the higher taxonomic rank. We demonstrate this by considering the case of Bitcoin.
Bitcoin uses public-key cryptography. Each Bitcoin address has associated with it a pair of keys, one public and one private. Only someone who knows the private key can spend Bitcoins associated with the address to which that key is related. This protects Bitcoin users, as long as they can keep their private key secret.
Bitcoin fits the definition of an account-based system. The account is a Bitcoin address, and the private key is the proof of identity needed to transact from that account. Every time a Bitcoin user wants to spend Bitcoin, that user must verify their identity by using their private key. It is not relevant whether the system requires users to reveal their true identity. Rather, what matters is whether a user must follow a process the system has developed for verifying the identity that they established within the system, whatever that may be. Analogously, a bank that wants to move funds through the Fedwire Funds Service has to comply with the Reserve Banks’ security procedures, which includes a set of access control features.
Bitcoin also fits the definition of a token-based system. When someone wants to spend a Bitcoin, the protocol verifies its validity by tracing its history. The current transaction history is used to verify the validity of the “object” being transferred, as other token-based systems also do. With Bitcoin, the object is a UTXO, which is only valid if it has not already been spent. With cash, the object is a note, which is only valid if it is a genuinely issued from the central bank.
One important difference between Bitcoin and a physical payment object, such as a dollar bill or a gold coin, is whether the payee can ascertain the unit’s validity with reasonably high confidence. For example, dollar bills have security features, which are easy to identify and difficult to counterfeit. In the case of a digital currency, it is not possible for the payee to ascertain the authenticity of the digital payment instrument independently.
To Sum Up
Classifications can be a powerful tool to organize and communicate ideas. The main allure of distinguishing between account-based and token-based is to highlight a defining feature of certain new, emerging forms of digital currency. But if a digital currency can be both token-based and account-based, then the classification loses its power to meaningfully distinguish between new and existing methods of digital payments. Furthermore, it may slow down progress in understanding intrinsic differences between the growing set of digital payment options and technologies. Future classifications could modify the definitions of the terms account-based and token-based to more clearly distinguish them. In the meantime, perhaps these terms should be retired to avoid further confusion.
“Fedwire®” is registered service mark of the Federal Reserve Banks.
Rod Garratt holds the Maxwell C. and Mary Pellish Chair in Economics at the University of California Santa Barbara.
Michael Lee is an economist in the Federal Reserve Bank of New York’s Research and Statistics Group.
Brendan Malone is a senior financial institution and policy analyst in the Federal Reserve Board of Governors’ Division of Reserve Bank Operations and Payment Systems.
Antoine Martin is a senior vice president in the Bank’s Research and Statistics Group.
How to cite this post:
Rod Garratt, Michael Lee, Brendan Malone, and Antoine Martin, “Token- or Account-Based? A Digital Currency Can Be Both,” Federal Reserve Bank of New York Liberty Street Economics, August 12, 2020, https://libertystreeteconomics.newyorkfed.org/2020/08/token-or-account-based-a-digital-currency-can-be-both.html.
Disclaimer
The views expressed in this post are those of the authors and do not necessarily reflect the position of the Federal Reserve Bank of New York or the Federal Reserve System. Any errors or omissions are the responsibility of the authors.
RSS Feed
Follow Liberty Street Economics
In reply to Svein Olnes: Thanks for your interest. We’re aware of Lewis’s post and his different take on the relative confidence a payee can have for determining the validity of a dollar bill versus a digital currency. As we tried to explain in our post, and others have observed as well, dollars have readily identifiable and well-known security features that allow holders to establish them as authentic in the moment, while a digital currency must be verified through a third-party.
Thanks for an interesting blog post where you try to clarify concepts that are a bit ambiguous. Whether you succeed or not I will not try to judge, but I will point to your paragraph on how to ascertain the validity of a dollar coin/bill or a bitcoin payment. I think your argument here is wrong, and I will point to Antony Lewis’ comments in his own blog “Bits on Blocks” for a clarification.
Thanks for this post and the examples, they really help to understand the distinction between token-based and account-based systems. Very valuable information for beginners.
Dear Sirs, The cases of Hong Kong’s Octopus and Japan’s Suica e-money system reinforces your point that the token based and account-based digital currencies are not mutually exclusive. The Octopus was the first contactless e-money system based on FeliCa Operating System which records which accounts the digital money has moved. (Wikipedia described South Korean UPass as the first contactless e-money system, but it is wrong. The UPass started as just an electronic contactless card system not e-money system.) The Suica e-money also record accounts. Both Octopus and Suica cards are regarded as token based as they are transferrable like notes and coins. In summary, i agree with your assessment that the distinction between token based and account based does not mean much if any.
Gabriele, thank you for your comment. We think of “identity” in a broad way. You are correct that it is possible to transact with Bitcoin while remaining anonymous. Nevertheless, to transact with Bitcoin any user must obtain a public-private key pair, which corresponds to that person’s identity within the system. Regarding your second point, whether a currency can be counterfeit is only one aspect of what makes it secure. There are many well documented cases of people losing their Bitcoins because they were not able to keep their private key secret.
Jessica, thank you for your comment. As you note, it is common in central bank and academic circles to use the term ‘digital currency’ for any system that allows for electronic payments, include Fedwire. This is also the way we think of digital currency in this blog post. Our main point also holds for the more narrow definition you prefer, as the example of Bitcoin should make clear.
How do you define digital currency? That seems to be the main question for where I start to see a disconnect between what I’ve read in the crypto space and what I’ve seen in the Fed/IMF/BIS world. Based on the crypto space, my current understanding for a simple definition of digital currency is “programmable money”, meaning it is tokenized and it’s a digital bearer instrument. To me, it’s not a question of whether money is token- or account-based, it’s really a question of whether or not it’s a token to determine if it’s a digital currency. Fedwire, while there is a digital aspect, is not a system for transferring a digital currency. It’s a transfer of messages back and forth between some intermediaries to update their ledger or record of accounts. A ‘general purpose account’ provided by a central bank would also not be considered a digital currency if it’s just having access to an account with no programmable money. This seems at odds with what I’ve read from the Fed or BIS, where the term ‘digital currency’ is used more loosely, so that something like inter-bank transfers and reserve balances would fall under the ‘digital currency’ categorization.
Bitcoin IS NOT an account based currency because no identity is necessary and the right to spend is NOT guaranted by any institution. The conditions to spend the funds are set by the spender and the protocol and can perfectly be met by a machine (not the same in an account based system traditionally intended). Bitcoin is the least counterfeitable currency to date because its validity is redundantly verified by thousands of computers which act as anti counterfeiting machines.