What is money? Much more on that soon, it gets to the heart of what I am thinking about these days, and what I hope to write about on these pages. I think we're about to have something of a crisis with respect to the meaning of money, and that the resolution of this should not be old bromides like the gold standard, but something completely different.

In the meantime, on Brad Setser's wonderful blog of which I'm a breathless groupie, regular commenters HZ and DF touched upon the question. I wrote a long response, then decided it wasn't really appropriate to Brad's comments, as my response was long and has nothing to do with Brad's main post. So, I'll put my mini-essay here, as a teaser for the much more that is to come on the subject.

Steve Randy Waldman — Friday March 3, 2006 at 1:05pm permalink
Iasius (mail):
If I understand you correctly, debt is money basically when it's debt that is not likely to be simply rolled over when it's due?
3.3.2006 6:30pm
Steve Randy Waldman (mail) (www):
Iasaius — No, debt is money when it is likely, as an empirical matter, to be used as one side of an exchange.

Imagine that you and HZ and I live in a village, in which we and everyone else know each other and one another's business. I happen to know that HZ owes you 20 chickens. I have some corn that you want, but you have nothing that I want. Instead you "sign over" your claim to 5 of HZ's chickens. I don't even want chickens, but I know other people in the village who have things I want, and who do want chickens.

In this scenario, HZ's debt to you has become money. We are exchanging IOUs -- much easier and more efficient than exchanging goods themselves -- and we are exchanging IOUs for things we don't necessarily even want, based on a "consensus value" that others who have things we may want will place on those IOUs.

Now, say HZ owes you not 10 chickens, but a large home. This is debt, just the same as the 10 chicken IOU, but it's unlikely to be money. Why? If I want to sell you corn, it is hard for me to accept a claim to 0.01% of the house HZ owes to you. It's not impossible, given a large pool of homes, sophiticated securitization, and all that jazz, but in our village setting, claims on tiny fractions of homes are unlikely to be used as money.

Note that none of this has much to do with whether the debt rolls over. Actually, an indefinite term or automatic rollover is increases the likelihood that debt becomes money, because usually people don't want delivery of whatever backs the debt -- chickens, gold, taxation offsets. They want to continue to exchange the claims they have received for unrelated goods and services they may want in the future.

The question of whether debt is money is primarily a question of convenience. Can the debt conveniently be used as a medium of exchange? States have refined to an art the crafting of debt for use as currency. They have made packaged debt in the form of low-denomination bearer notes, and developed elaborate electronic systems for transferring finely graded claims. They have increased the relative convenience of using their debt for exchange, by outlawing and therefore threatening inconvenient consequences, for using alternative forms of debt as exchange. So, the form of government debt we call currency has attained a near monopoly on the supply of money.

But money is always whatever people use as money. There is no magic quality that makes some debt money and other debt not. There are qualities intrinsic to some debt that make it convenient as a medium of exchange (or a store of value, or a unit of account). But debt is only money when people commonly exchange it for goods and services.

The US Treasury's debt to China, sitting in a metaphorical vault and rolling over, is very much not as an emprical matter participating in any economy as money. It could, in the future, and that would be inflationary. But for now it's only debt.
3.4.2006 3:55am
HZ:
Steve,

Thanks for your generosity in hosting the discussion. I enjoyed reading your well-written comments here.

I agree with your broader definition of money on a transactional basis, which is also the frame of mind in which I raised the original questions.

In your response to Iasaius you explained well why not all debt are money (really cash) like. It is clear that currencies and non interest bearing checking account are intended for transactions (they are inferior value stores even though some do stuff cash under their matresses). It is less clear how debts in other forms are used.

Let's we go back to our fundamental equation MV=PT: on the RHS is what our goals are based, we want price (P) to be stable to reduce inefficiency, we assume that the number of transactions T will steadily go up reflecting growing economic activities. So in determining monetary policy we want to control the LHS to achieve the objective of price stability. The problem is, if we define transaction T broadly enough, money (M) and velocity (V) become largely unobservable and their definitions are inter-dependent.

An operational definition of an observable money is to strictly interpret money as borrowings from the banks: thus M0, M1, M2 reflect progressively broader definitions of money that are measureable and controllable through central bank levers. But in a developed economy like the US bank lending is about 35% of the total credit, these operational measures do not reflect the total economic activity.

Let's continue using examples: suppose we have a farmer who needs a loan to buy seeds from a seed supplier and pay it back after harvest. First suppose the farmer goes to the bank to get a loan and his supplier deposit the payment back into the bank and it can be loaned out again. This is classic money creation through the bank. However since the bank has reserve requirement the amount of money that can be created this way is well controlled and observable. By involving the bank as an intermediary in their debt/credit relationship, a portion of the bank's reserve is tied up. The aggregate of such borrowings are limited by the amount of reserve supply and reserve requirement. An economy based on such relationships are easy to regulate from the central bank's point of view. But now suppose the bank securitize the loan and sell the note to the supplier instead (in reality such securitizations will be spread over many borrowers and investors to have a statistically stable credit risk profile, but here we keep the players to three to make things simple). It is as if the farmer directly advanced the loan from the supplier (a financial barter, if you will). Now that the bank and the supplier (in his investor role) swapped note and deposit, the loan is off the bank's book (such borrowings are no longer reflected in our M1/M2) and it no longer has reserve tied up and can go merrily creating more other loans.

(continue on next post as there seem to be a limit on each post)
3.4.2006 8:12pm
HZ:
Now we have the question of how such ex-banking borrowings (bonds and stocks, bank loan securitization, MBS/ABS/CDO, and some of the biggest players outside of banking, the GSEs, use even higher leverage than banks for loans kept on their books) enter into our fundamental equation. If we take the view that we are concerned with total economic activity on the RHS we will have to either have a broader definition of M (M* > operational definition of M) or attribute the effect to an increase in the velocity. I would argue that the increase of velocity is insufficient: consider that immediately after the collapse of the Argentine currency, people were forced to barter, money (strictly defined) goes to zero (or infinity depending on your view) effectively, velocity couldn't plausibly be argued to have gone up, yet the economic activity, while depressed, didn't change nearly as much. Of course the equation was tautologically true if you strictly define everything: in which case T defined as transactions conducted using bank money also goes to zero, so while the equation holds it is no longer interesting. The relevance here is that entering into debt/credit relationship outside of the bank is akin to financial bartering (instead of goods and services bartering in post-collapse Argentine, and which also exists to a certain degree here in the US) and needs to be reflected in a broader definition of money, if central banks want to regulate the economy rationally. Going back to the example you used, modern fianance inventions created all kinds of fractional ownerships of debts and equities that can be easily swapped and behave as bank money substitutes (or reduce demand for bank money).

Taking the broader view of money makes it a lot less clear how money supply should and could be regulated. I find this a fascinating subject and am working to have a better understanding of it.
3.4.2006 8:15pm
Steve Randy Waldman (mail) (www):
HZ,

First, thanks for the very thoughtful comments!

Taking the broader view of money makes it a lot less clear how money supply should and could be regulated. I find this a fascinating subject and am working to have a better understanding of it.


Exactly. I think that at this moment, to not take a very broad view of money is to miss most of the action. Let's begin with the equation you use to frame things, the equation of exchange:

MV = PT


It is, as you point out quite aptly, both a tautology, and not so informative when the simplifying assumptions (or expected results) of monetarism fail to hold. For the quantities in the equation to be meaningful, there has to be a unit of money, all transactions must be occur via exchanges of this money, and this money must have a finite, quantifiable supply.

As you point out, the equation doesn't capture very much if people barter. It doesn't capture so much in a multicurrency world, where the money supply in any single unit of value varies with currency fluctuations. If one restricts oneself to a world without barter, in which all transactions do occur in a single currency whose quantity is effectively controlled by a banking system, then one has forced all valuation-dependent variation onto the velocity term. If one concludes that velocity is resistant to change, you get to the monetarist worldview that made Greenspan emperor for a day.

But the real world is moving fast. You point out that only 35% of total credit is bank credit. For simplicity, let's just sloppily call the rest "securitized". Let's presume a neutral central bank, and a banking system in which the only ways money is lent are via demand deposits (reserve controlled) and securitization (unregulated). M1 is reserves and demand deposits. We assume banks have created demand deposits up to the maximum permitted by their reserves, so M1 is a fixed constant.

I want cash, and have an appreciated house. With a bank's ephemeral intermediation, I sell you a securitized claim on my house, and you give me a lot of money, which you take from your bank, and I drop into mine, making the whole thing a net wash for the banking system. I now have cash, you have a security, which trades in liquid markets. The money supply remains unchanged, but our perceived liquidity has nearly doubled! Next we have an empirical question: Does perceived liquidity increase the right-hand side of the exchange equation, PT? Am I more likely to spend money, now that I have cash in hand rather than home equity of the same value? Is the increase in my likelihood to spend fully offset by the decrease in your likelihood, having exchanged cash for a not-quite-as-liquid security?

These aren't simple questions. They depend, among other things, on what kind of actors holds the securities and cash. If a central bank accumulates securities and sequesters them, as a matter of policy never liquidating them to purchase ordinary goods and services, only the price and transaction volume of securitizable assets might be affected. If investor-consumers underwrite securitization, the additional liquidity is likely to carry through to PT more broadly. (Note though that even government and credit card spending now represent securitizable "assets".)

But the various definitions of money supply don't much capture this dynamic. As you note, tautologies always hold, and we can account for any effect in PT by broadening M or letting V take up the slack. But I feel like we are trying to plot the orbits of the planets with Ptolemeic charts. Everything works, but the conceptual framework we are using solve our problem is cumbersome, we can think of better ways.

I don't have a better way all worked out, but I do have some ideas. (I'll try to put some of those ideas together and make that a fresh post.)
3.5.2006 9:23am
Steve Randy Waldman (mail) (www):
Some of those ideas are up here.
3.7.2006 9:31am
Note: You don't have to register to comment! Just use the pull-down menu to comment as "Guest".