Wednesday, 4 September 2013

Banks, Non-Banks and the Medium of Exchange

Paul Krugman's comments on Tobin and banking sparked an extensive discussion on the blogosphere. Much of this discussion has related to the different status of banks and other financial institutions (NBFIs).  Throughout this discussion, the following argument keeps cropping up:

Because loans create deposits and bank deposits can be used as money, bank lending increases the total purchasing power in the economy and is therefore expansionary.  Non-bank lending merely redistributes existing purchasing power and is therefore not expansionary.

It seems to me that there are one or two ideas here that are getting muddled.  In particular, I think there is some confusion about the implications of the following two semi-facts:

- The amount of bank deposits outstanding is determined primarily by the amount of bank loans outstanding.

- Bank deposits can be used as the medium of exchange, whereas claims on NBFIs cannot.

I call them semi-facts, because I think there are grounds for questioning both, but I do think they are reasonable assertions and I'm going to assume for the rest of this post that they hold.

To examine this, I want to imagine the simple economy illustrated by the following balance sheet:


Time deposits


Savings accounts



Here the financial sector is represented by two types of lender: banks and NBFIs.  Both entities make loans to firms (households are assumed not to borrow).  What exactly constitutes the medium of account is, I think, a difficult question in real economies.  Here, I'm going to make it simple by assuming that payments may only be made using the balance of checking accounts at banks - what I have called money.  In addition, households also hold time deposits with banks.  NBFIs offer savings accounts to households and these represent their only liability.  NBFIs have to attract funds into their savings accounts before they can make loans.  As the timing of new loans does not necessarily coincide with the timing of new savings, they also hold some money balances as a float.

To give a special status to the medium of account, we are going to assume that household demand for money balances is in a fixed ratio to nominal income (Y). 

Mh = k.Y

This means that time deposits and savings accounts are not in any way substitutes for money.  However, we will assume that time deposits and savings accounts are substitutes for each other.  How much of each households wish to hold depends on the rates offered, but we will assume that a small change in rates is enough to cause significant switches from one to the other.  We are also going to assume that firms are indifferent between borrowing from banks and NBFIs.

The first thing to notice is that the immediate impact on the non-financial sector is the same for an increase in NBFI lending as it is for an increase in bank lending.  Loans (L) increase and money holdings (Mh) increase (we are assuming firms do not retain any money, so all money holdings end up with households).  An increase in bank lending would be offset by an increase in overall money (M goes up and Lb goes up).  An increase in NBFI lending involves a transfer of money (Mf goes down and Lf goes up).  But in both cases, household money (and therefore overall household wealth goes up).

If NBFIs wish to expand their business, they will need to do two things.  They will need to start lending more.  They will also need to get in more deposits.  They can do the latter by increasing their savings rates slightly, to attract money from banks.  However, it turns out that households do not need to reduce their holdings of bank debt at all.  The extra deposits required by NBFIs is exactly matched by the extra money holdings realised by households from the very loans made by the NBFIs.  All they need to do is ensure that all of the new household wealth is channelled into new savings accounts rather than new bank deposits.  But the actions of the banks are likely to help this process.  Unless the banks also wish to increase lending, they will not want an increase in interest-bearing time deposits.  If households are trying to place too much into time deposits, they will cut their deposit rates.

So loans create deposits for NBFIs as well as banks, albeit in a rather different way.  Debt levels and overall household wealth levels can rise purely off the back of a decision by NBFIs to expand their business.

But what about money, the medium of exchange?  NBFI lending does not create any new money, so unless the banks are lending, how can actual purchasing power increase?

Now we should be able to see where the confusion lies.  The quantity that is determined by bank lending is not money, but total bank debt.  Money, our purchasing power, is just a subset of that and is not fixed by the amount of lending.  If more money is needed, households can be achieve this without any new bank lending by simply reducing their holding of time deposits (obviously, there may be a slight lag here as time deposits are allowed to mature, but typically this will be small as maturity dates will be staggered).  So, we might see a growth in NBFI loans and NBFI savings accounts, a growth in the stock of money and a reduction in bank time deposits.  There is no reason why this should not be as expansionary as bank lending.

It is worth noting there is still an asymmetry here. This process cannot go on forever, because ultimately the amount of money is limited by the amount of bank debt.  Bank debt, on the other hand, can expand indefinitely (at least within the framework of this model).  However, the point here is not that bank lending and NBFI lending are identical.  The point is that it is quite wrong to say that NBFI lending doesn't add to the medium of exchange and therefore cannot be expansionary.


  1. Is it any more complicated than understanding that bringing finance to a productive process by creating deposits through bank-lending is similar to bringing existing balances in deposits and terms deposits into circulation by applying these to finance a productive process through NBFI lending can be similar in their effect on aggregate demand?

    1. Possibly not. Yet the idea that non-bank lending merely transfers existing purchasing power around comes up time and time again.

  2. Money can be saved by the owner under-the-mattress, in bank accounts, in NBFIs, and in wallets. Banks and NBFIs differ from under-the-mattress and wallets in that they can make loans based on deposits without the direct action of the deposit owner.

    Now, does a deposit resulting from a bank loan add to the money supply, or does a bank loan merely move savings into actively trading hands?

    It seems to me that BOTH options complete. This then would offer two methods of measuring money supply, both accurate.

    If correct, our challenge is to develop data and correlated models for both measures of money supply.

  3. I didn't really include cash in this model, in order to keep it simple. However, we could quite easily assume that what I call money represents a mix of checking account balances and privately-issued banknotes. The latter could obviously then be kept under the mattress or in wallets, so this would be part of Mh.

    There are of course different measures of the money supply. Most measures beyond the narrowest include some form of bank deposit, so would be increased by the deposit resulting from a loan.

  4. I don't get how NBFI lending can be - in any way that is important for economics - just a transference of purchasing power. If money is deposited with the NBFI then the intention of its owner is that it isn't going to be spent in this period. It could be of course, but their decision is that it isn't going to be, in some latent sense there is purchasing power there but it isn't being used to purchase anything.

    Instead the NBFI is acting as something of an intermediary and the deposited money is being lent to a firm, in this model, who intends to invest it productively. That then sets into action a set of monetary flows that would not have otherwise occurred.

    1. I agree, but this idea of only transferring purchasing power seems to have a real hold, even with some professional economists.

  5. There's a lot of things that went on during the various blog debates. I think the "mainstream" argument was that if the banks extend loans, and investors redistribute their portfolios, you could see that others would pay down their loans to cancel out the expansionary impact of the original loan. Yes, it is possible. Your model appears to preclude such behaviour (I believe) as the total cancellation appears unlikely (which seems to be good match with actual data).

    However, bank finance is not the only way to create demand via credit. Industrial firms routinely extend credit via receivables and vendor finance. The line between "financial" and "nonfinancial" firms is somewhat blurred.

    1. Agreed. I haven't allowed for reflux in this model, because I'm assuming that the borrowers (firms) spend everything. In the real world, I think loans get paid down a bit, but I don't think it's a hugely important effect.

      I was just picking out one part of blog discussions, being one that particularly interested me. I specifically focused on NBFIs as that had seemed to be the subject of the debates, but I agree that this point applies to credit generally.

    2. 'Reflux' for the simple flow model you have set up may not matter, but have you read Minsky (I'm sure you must) on 'hedge' 'speculative' and 'Ponzi' positions that firms can adopt - these positions can include a portfolio of many other 'monies', including what are described as assets, brought into play to support the firm's productive operation. The basic problem that economists have is that they don't understand to the fullest extent that capitalism is a money economy - they don't really understand that money is money of account. Hence a firm with a position with both a NBFI and a bank say, will have a money balance sheet that includes both. Relationships between these positions can be complex - for example a firm may borrow from an NBFI based upon an asset as collateral, to service a loan from a bank. Broadly, the 'money supply' is in part built out of these many types of financial operations, contingencies and necessities.

    3. I'd tend to agree with that, Jim. Treating bank money as special and ignoring other financial balances seems arbitrary to me. They may not all matter equally, but they all matter.

  6. Nick I think your model is saying in simple terms that Agg demand is Income +change in debt (bank credit) -change in savings. Once you understand that drawing down a savings account is disavings you can derive the argument above. This is a change in the amiount of money in circulation, and only money in circulation affects prices, but not a change in the money held in account. I derive an almost identical result from simple identities here

    1. Andrew,

      I haven't included much in the way of equations in this post, as it wasn't necessary for what I wanted to say, but it is certainly implied that, for each sector, demand (meaning expenditure) = income + change in debt - change in savings. Of course, this should be implied by any consistent national balance sheet and is integral to the way the flow of funds are drawn up in the national accounts. I would add that this identity provides the framework for my UK macro model, described on this site.

      When you say money in circulation, do you mean the balance of sight deposits ("ready" money) as opposed to time deposits. If so, I'm not sure I agree with you. My view is that all assets and liabilities impact on spending (although some may impact more than others). The amount of ready money is more to do with how much of their wealth people choose to keep in that form.

  7. I have some thoughts on money supply expansion that I put into a new post that can be found at

    In the post, I make the assumption that banks expand the money supply (deposits) by creating a new deposit while not affecting any preexisting depositors. The result should be a ratio of two deposits per loan. This is not found in the relevant data. An explanation for the difference is offered.

    Based on my post comments, I would conclude that NBFIs do not expand the money supply unless they also offer creditable bank-like guarantee for deposits.

  8. Late to the party... I liked the post and -- first time for everything -- understood the balance sheet.

    Quoted you (for my purposes). Thanks.