On "Shadow Money"

moneyness, repos, reverse repos, shadow banking, shadow money
"Modern Money Hierarchies," Figure 4 on Page 14 of Daniela Gabor and Jakob Vestergaard's "Towards a Theory of Shadow Money." https://ineteconomics.org/uploads/papers/Towards_Theory_Shadow_Money_GV_INET.pdf

TMoneyPyramidhe shadow banking literature has vastly and rapidly expanded since the financial crisis, and has produced some interesting pieces, as well as some exaggerated claims, in my view.  While I am not writing today to address those claims, I still wish to question a closely linked concept that has simultaneously sprung up in the literature and in particular in the post-Keynesian one: shadow money.

One of the most elaborated and comprehensive academic research papers on this particular topic is the recently published Gabor’s and Vestergaard’s "Towards a theory of shadow money."  It’s an interesting and recommended piece.  But while I agree with some of their writings, I have to find myself in disagreement with a number of their points and examples* and in particular their central claim: that repurchase agreements ("repos" thereafter) are shadow money; that is, a type of monetary instrument used within the shadow banking system.

For some readers that might not know how a repo works, below is a concise definition provided by the IMF:

Repo agreements are contracts in which one party agrees to sell securities to another party and buy them back at a specified date and repurchase price.  The transaction is effectively a collateralized loan with the difference between the repurchase and sale price representing interest. The borrower typically posts excess collateral (the “haircut”).  Dealers use repos to borrow from MMFs and other cash lenders to finance their own securities holdings and to make loans to hedge funds and other clients seeking to leverage their investments.  Lenders typically rehypothecate repo collateral, that is, they reuse it in other repo transactions with cash borrowers.

Given repos’ (and their asset counterpart: reverse repos) properties, my view is that repos aren’t shadow money but a shadow funding instrument.  While it might not sound such a big issue, I believe the distinction is important from an analytical perspective as well as to avoid confusion.  Let me elaborate.

Gabor and Vestergaard define shadow money as “repo liabilities, promises backed by tradable collateral.”  According to them, shadow money has four key characteristics:

a) In modern money hierarchies, repo claims are nearest to settlement money, stronger in their "moneyness" than ABCPs or MMF shares.

b) Banks issue shadow money.  The incentives to issue repos are incentives to economize on bank deposits and bank reserves.

c) Shadow money, like bank money, relies on sovereign structures of authority and creditworthiness.  The state offers a tradable claim that constitutes the base asset supporting the issuance of shadow claims.

d) Repos create (and destroy) liquidity at lower levels in the hierarchy of credit claims.

They offer this chart of "modern" money hierarchy:

Shadow Money

I have to object to repos being classified as "money."

Money, as typically defined by economists, has three characteristics: it is a medium of exchange, a unit of account and a store of value.  High-powered money (the "outside money" of the financial system) currently fits this definition, as a final settlement medium.

The "moneyness" concept, a term now popularised by JP Koning’s excellent blog, asserts that various types of assets have various degrees of money-like properties.  In this quite old but classic post, JP argues that anything from beers and cattle to deposits benefits from some degree of moneyness.  In another old post, Cullen Roche provided the following good "money spectrum" chart (although I’d disagree with his outside money/deposit ranking):

scale of moneyness

Therefore, most goods and assets have some monetary properties: some can be used as media of exchange or store of value.  All represent a claim of some sort on money proper.  As a general (but inaccurate) rule, the more their price in terms of outside money fluctuate, and hence their conversion risk raises, the further away they are on the moneyness scale.  But conversion (almost) on demand also implies that, in order to have some money characteristics, a good or asset needs to be tradable.

Now let’s get back to repos as shadow money.

Repos are a liability issued by the debtor in exchange for high-powered money, of which reverse repos are the asset counterpart held by the creditor (and hence the claim on the high-powered money originally transferred, plus interest).  The debtor also transfers an extra asset (i.e. the collateral) to the creditor for security purposes at a pre-agreed haircut depending on its credit quality and market risk sensitivity.

We get here to the main point of this post: repos have little money-like property due to their non-tradability and lack of on-demand convertibility.

Indeed, a repo liability is of course non-tradable, in the same way that any debt that one owes cannot be traded for another type of liability.  It can only be refinanced and/or extinguished.  A reverse repo (or repo claim) however, could potentially have tradable properties, allowing a creditor to exchange his claim almost instantaneously on the market.  Problem is: this does not happen.  Unlike bonds or other assets, and due to the very specific features of such private agreements, there is no secondary market for repo claims.  Once a repo has been agreed upon, the contract is fixed between the two parties until maturity (or default).  Consequently, repo claims can effectively be assumed to have no liquidity.

Seen this way, it is hard to classify repos as "money," and they certainly do not deserve their third place in the moneyness hierarchy above.  So what are repos?  As I previously said, they are a funding instrument. Given that the shadow banking system makes use of repos on a large scale, we can potentially call them a "short-term secured shadow funding instrument."  And please note that repo issuance isn’t limited to banks and broker-dealers; other institutions also use them.

You’ll be tempted to reply: “what about deposits?  They have no secondary market and are not tradable either.”  This isn’t strictly accurate.  While they are both promises to pay a certain amount of money proper at a certain date, there is a very specific difference between deposit liabilities ("on demand" ones especially) and repo liabilities.  Banks themselves are deposits’ secondary market: deposits can be ‘traded’ within the bank’s own balance sheet and swapped for cash on demand.  And when dealing with a counterparty that does not hold an account with the same bank, banks take over the responsibility of transferring the underlying funds (i.e. high-powered money).

If repos aren’t "money", what else could be considered "shadow money?"  Well, assets provided as collateral do have liquidity, tradability, and therefore some "moneyness."  Those assets can sometimes be used in further transactions.  This is why I am wondering whether or not there isn’t some confusion with "shadow money" proponents’ terminology.  While their writings clearly emphasise the "shadow money" nature of repos themselves (and Poszar seems to be using the same definition here), many other academic authors have instead referred to the most commonly-used types of repo collateral (high quality and highly liquid sovereign and corporate bonds) as "shadow money" (which indeed makes more sense to me, although I do not fully adhere to this concept either).

There are plenty of things worth discussing regarding this theory of shadow money and the use of repos in general, but the money-like properties of repurchase agreements isn’t one of them.  Let’s focus on their funding properties instead.


*I also believe that their shadow money expansion theory is subject to the same critique as other endogenous outside money theories, such as MMT’s.

PS: The fact that repos are backed by marketable collateral does not confer any specific monetary property to repo claims.  Marketable collateral is used in many other types of lending transactions, in particular in private banking-type lending.  Also, repos and any other collateralised lending are expected to be repaid at par, independently of the valuation fluctuations of their underlying collateral.

PPS: Baker and Murphy build on Gabor’s and Vestergaard’s piece and just published a blog post that argues for a new "investment state," in a typical post-Keynesian interventionist fashion.

[This article originally appeared on Spontaneous Finance]

  • Mike Sproul

    Why does anyone care if something is called money or not? Some things, like Federal Reserve Notes, checking account dollars, and credit card dollars, are used as money quite often. Other things, like Walmart gift certificates, Canadian tire money, and my own paper IOU's to my local grocer, are used as money only occasionally. It's just a question of where different things fall on the moneyness spectrum,

    • Julien Noizet

      I agree Mike. My point being that all the examples you gave were tradable assets (or liabilities, depending on the perspective), whereas repo claims aren't. Hence I believe they rank quite low in the hierarchy of money.

  • Bill Woolsey

    I think overnight repurchase agreements are money–as long as your bank will make funds available when you don't roll it over for an additional day. The collateral is no doubt important for some reasons, but not for the "moneyness" issue.

    • Julien Noizet

      Bill, independently of the collateral issue, if you believe that overnight reverse repos can be considered money because they are overnight, then it should also be the case of any other type of overnight loans.

  • George Machen

    It’s not that overnight repos themselves are money (transactions media), but that they represent demand deposits that *should* be counted in M1 to the extent that they reduce demand deposit balances *reported* at business day-end.

    Like commercial banking business & retail sweep accounts and overnight Eurodollars, overnight repurchase agreements fall under the category of instruments that are not themselves directly used as purchasing media (transaction balances generally accepted as means of final payment for things offered for sale in markets), but instead are demand deposits in use during the daytime that get switched to other accounts at the end of each day, right before the depository institutions have to report their demand deposit liability amounts to the monetary authorities.

    From the individual banks’ points of view, these account “reclassifications” are motivated by avoidance of reserve requirements. Reserve requirements on checkable deposit liabilities are a “tax” on banks; consequently, banks are eager to find ways to reduce such non-earning assets. For example, before bank closing for the day, much of (large) consumer retail demand deposit balances, which as of January 21, 2016 impose an up to ten percent reserve requirement on the bank, might be transferred to a money market deposit account, which carries a zero reserve requirement. The next morning, right before bank opening, the funds are swept back into the customer’s demand deposit account for use during the day to effect payments. This reshuffling of funds occurs “behind the back” of the retail depositor (and the bank keeps the interest). As recently as 2004 (the last time I looked at the numbers), consumers and small businesses thought they had retail demand deposit totals twice as large than are reported to the Federal Reserve. Business sweep accounts work similarly, except that funds are swept into money market mutual funds, Eurodollars and repurchase agreements; the business customers share with their banks in the overnight account earnings; and the procedure explicitly is known and initiated by the corporate treasurers.

    Much of the increases in M1-type transactions balances to the current unprecedented levels have been “hidden” within financial innovations that do not get counted in the official M1 monetary aggregate, namely overnight repos, overnight sweep accounts, overnight Eurodollars and a portion of retail money market mutual funds.