I am keeping this article as brief as possible. For more information, a lot of this ground is covered in my recent MMT book.
Academic MMT View — No
This first thing to note is that from what I have seen of academic MMT discussions, “money” and “bonds” are distinct. One thing that I see is that “money” shows up in two ways in most MMT writing.
The most common is that “money” is a stand in for “government money” — which is the monetary base. This corresponds exactly to the way money and bonds show up as M and B in both heterodox and mainstream models that do not have a banking system. (This leads to the straw man attack that “MMT ignores that most ‘money’ is created by banks!”)
If we want to look at “broad money,” bank money is seen as “pyramided” on top of the monetary base. (Which answers the previous straw man attack.)
I cannot claim to have encyclopaedic knowledge of the MMT academic literature, but I have never seen a proposal to include government bonds within a “money supply” measure. Most of the discussions involve replacing government bond issuance with (government) money issuance. If the two governmental liabilities were equivalent, why would this matter?
Straw Man Attacks — Yes
One prolific MMT critic made up a theory a few years ago that “MMT is the quantity theory of money with bonds as money” based upon pulling it out of his nether regions. (I will not name the critic, given his obnoxious online behaviour mixed with bald-faced lying.)
The only theory that roughly matches that description is “the Fiscal Theory of the Price Level,” (FTPL) since the price level is proportional to the stock of government debt (assuming that future real fiscal balances are unchanged). Although some people want to draw parallels between MMT and the FTPL based on the simplistic observation that fiscal policy determines the price level, the actual mechanics are obviously different.
Over the past few years, I have seen some instances of non-academic MMT fans making statements on the internet along the line that “government bonds are money.” I have no idea where that came from, but the rest of this article discusses the plausibility of the idea. The answer depends upon what your definition of money is, but focusing on that question misses the more important question: why do we care? (Spoiler: the answer to that question is: we don’t, if we know what we are doing.)
What is Money? (Sigh)
In order to answer the question “Are government bonds money?” we need to define the word “money.” (The fun thing about the Twitter debate I saw was that everyone was just using random definitions of the word “money.”) There are two basic approaches.
(Non-controversial) Use the monetary aggregates (M0, M1, …). For any given country, whomever compiles the monetary aggregate data (could be a statistical agency or the central bank) has a list of instruments that fit into each aggregate. The overall definition is generally the same across countries, but each country has its own little weird fixed income instruments that might get included.
(Controversial) Define by the role “money” is supposed to play. If one reads the Austrian literature, this is something they love to do.
The advantage of the first approach — using monetary aggregates — is that it is pretty easy to answer the question. Do “government bonds” appear in the aggregates? The answer is “no” if we mean “directly appear,” although “money market funds” will typically make the cut in broad monetary aggregates. And guess what? A money market fund is just a pass-through vehicle for unit holders for its assets — and those assets will include short-dated government bonds/bills and repurchase agreements (“repo”) on government bonds.
The disadvantage of the first approach is that it does not answer the question “Why are these particular instruments showing up in the various monetary aggregates?” So one might better phrase “Are government bonds money?” as “Should government bonds be included in monetary aggregates (ignoring their pre-existing indirect inclusion)?” That is the question I am looking at here.
As an aside, this theoretical debate shows a weakness of the approach of many physical scientists who make their way into economics: they assume that they can apply whatever mathematical techniques they studied in university can be applied economic data, and that we can create theory just based on the data. The problem with that the data are defined with an implicit economic theory behind them. (As an aside, I avoided that fate by finding out that neoclassical economists already applied mathematics that I studied in university, and the results stank.)
The problem with defining “money” based on its roles is that everyone needs to agree with what properties “money” is supposed to have, and whether real world instruments exhibit that property. Guess what? Not everyone agrees (which is completely unsurprising given the dysfunctional state of economics).
I see two main approaches.
An enumerated list of properties that “money” must have — stable value, used in transactions. This usually ends up aligning with the M1 — the monetary base (“government money”) and bank deposits. (And gold, if you are an Austrian.)
An approach that also takes into account how instruments are used in portfolios, which ends up including instruments that would be considered “cash.” This is how we end up with the wider monetary aggregates.
If we take a dogmatic version of the first approach, we are stuck with M1 in practice, since bank deposits and government money have legal privileges in transactions. The problem with that dogmatism is that it says that things like M3 are not “money supply” measures, when even desperate Monetarists had to use them in a desperate attempt to rescue their failed theories.
Unit of Account
The notion of a unit of account can cause confusion. Debt instruments are denominated in a unit of account — e.g., normally the U.S. dollar in the United States. Just because the unit of account is terms of government money in a class of transactions that does not mean that government monetary instruments are used. For example, a client can do many types of transactions with a broker “counterparty” in a margin brokerage account in terms of U.S. dollars without any use of bank or government money. (Only inflows/outflows to a bank account would count.)
Problems With The Narrow Definitions of Money
Using a definition of money designed to exclude everything other than government money and bank money has a number of problems.
One issue is that usability in transactions of M1 components is conditional.
The reality that money laundering is a major concern for organised crime tells us that banknotes use in transactions is not guaranteed. If you try to purchase a $1 million home with $100 bills from a couple that consists of a zealous RCMP officer and Canada Revenue Agency auditor is going to get you answering a lot of questions. The only financial institutions that deal with banknotes are banks, as part of a entirely separate cycle of transactions aimed at supporting retail/inter-household transactions (and the underground economy).
Bank deposits are only “money” until the bank goes bust. Deposit insurance with a limit of $100,000 is not a whole lot of help to large firms or even a household that is doing something like receiving payment for a house.
Settlement balances at the central bank (“reserves”) do not have size concerns like banknotes, but they can only be used in transactions between banks.
The next — and most important — issue is that not all transactions are intermediated by M1 components. In order to exclude instruments that are not in M1 from your definition, you need to add riders that rely on the legal privileges of M1 components. And once we see that many transactions are not undertaken with M1 components — e.g., anything involving debt/receivables — we need to ask “why do we care about this definition of ‘money’?”
Finally, we run into the issue that measured balance sheets of “monetary” instruments can bear no relationship whatsoever to transaction capacity. The case to keep in mind is the “no reserves” interbank clearing model that the Canadian banking system (roughly) followed pre-pandemic. (I discuss this in my book Understanding Government Finance. Hey, three book plugs in one article!)
In that model, banks are expected to end the day with a $0 balance with respect to the payments system — implying they start the next day with a $0 balance. The Canadian banks merrily transfers oodles of bucks to each other intraday (the oodles is typically best understood as a percentage of nominal GDP rather than a dollar amount), and they only worry about doing transactions to square up their balance at the end of the day.
We can compare this to the pre-2008 American system where the convention was that banks had to hit a (bank-dependent) target balance based on a percentage of a certain class of deposits (a.k.a. “reserves”). To an outside observer, the end result is the same — a bank starts the day with a target balance, and they square up at the end of the day at that target. (The target is adjusted periodically, based on lagged deposit data.)
If we want to be subversive, we would realise that these conventions are arbitrary results of culture. The convention could just as easily be that all banks are expected to have a specific negative balance at the end of the day (implying a loan from the central bank to private banks). This would imply a negative aggregate “reserves” balance supporting all inter-bank transactions. (The monetary base also includes banknotes, so the monetary base would only be negative if the target settlement balance was negative.) Try sticking that negative balance into your Monetarist models, suckers!
If we do not make arbitrary restrictions on the transaction uses of “money” and just look at usage in portfolio, “money” becomes somewhat synonymous with “cash” as it used in portfolio theory: short-dated fixed income instruments.
By that standard, Treasury bills and aged bonds with a short remaining maturity qualify. What about the rest of the government bonds? Well, one of the reasons many investors hold government bonds is that they can be used to raise funds in the repo market at the risk-free rate. Using repo for dealing with large short-term liquidity swings can be more attractive than trying to trade out of private commercial paper.
In my online discussion, one counter-point was that the repo market is unreliable (oh no, 2019). Banking systems are also unreliable — bank holidays happen. Neoclassical central bankers do not like thinking about the plumbing of the financial system, but the reality is that if they want monetary policy to work like their models assume, they need to keep the government bond repo market functioning. (Repo in private securities can be periodically burned down to keep everyone on their toes.)
As such, I can see why someone might attempt to lump government bonds in as a liquidity reserve, but it’s a bit of a stretch to lump a 30-year bond in the “cash” asset class.
Should We Care? (No.)
Regardless of the previous semantic arguments, the elephant in the room is the utter irrelevance of the answer. Why do we care about “money” in the first place as an analytical concept?
The Quantity Theory of Money — lol.
Keynes’ “liquidity preference” theory for interest rate determination — also lol.
Trying to correspond to other macro models featuring “money” — I am largely unimpressed. All that such models tell us that the private sector allocates between “government bonds” and “government money.” Well, guess what. If we divide government liabilities into two classes, by definition, private sector holders of government liabilities have to have some rule to define that portfolio allocation within the model (as otherwise the model is indeterminate). Although these models can be used as teaching models, their correspondence to the real world is questionable. As my previous discussions noted, real world transactions often involve credit and can be done with zero (or even negative!) holdings of instruments.
What about the study of financial crises (a suggestion given to me)? If we want to understand a financial crisis, we need to keep track of the instruments involved — their status as “money” does not matter. E.g., the creation of a lot of banknotes (somehow) was not going to help the financial system in 2008. Neoclassicals at central banks have come to believe that central bank purchases of government bonds “to create money” will solve all manners of ills — but is that really true, or just a reflection of repressed Monetarism? People used to sacrifice animals (or even people) as part of construction projects, does that mean that sacrificial magic should be part of the civil engineering curriculum?
This debate in itself is not particularly significant (other than possibly showing up in internet MMT arguments), but it does provide another vantage point on the problematic nature of “money” in economic theory.