Thursday, August 2, 2012

Musings in central banking 1: 'printing press', fact vs fiction

The central bank 'printing press': hyperinflation, 'government is not revenue constrained', 'central banks don't need capital'? Let's review some basic concepts about banking and dispel some myths.


Pure banking (PB)

A customer uses his credit card with bank A to pay for $X worth of groceries. Grocer is with bank B. Here's what happens: A loans the customer $X and immediately instructs bank B to credit the grocer's account by $X.

To understand what goes on in the background, be mindful bank A lending the money to the customer is recording $X as an asset (the loan) and $X as a liability (the deposit). The transfer from the customer's deposit to that of the grocer therefore results in a decrease by $X of A's liability. How is it balanced?

In the old days when CB did not exist, if A held sufficient deposit at B, the latter would simply transfer $X from it to the grocer. A would then record an asset decrease by $X  balancing the above operation.

If A held no deposit at B, the latter would create a loan/deposit of $X in A's name and immediately transfer $X from the deposit to that of the grocer. The size of B's balance sheet would increase by X. In the end, A borrowed $X from B and lent as much to the customer, which is also balanced. If deposit rate is zero, for A and B to both be in business, B must lend to A at a rate inferior than that at which A lends to the customer.

CB/discount window (DW)

The central bank is created. Instead of keeping deposits at each others', they now hold deposits at the central bank, called 'reserve balances' (hence Federal Reserve). If A held sufficient deposit at the CB, the latter would debit A's deposit by $X and credit B's by that amount in compensation for the customer paying $X to the grocer.

For the case A held zero deposit at the CB, assume the only allowed way to meet its treasury shortfall is by borrowing $X from the CB (discount window). As this juncture, the role played by B in 'pure banking' is now played by the CB. In a competitive market, A lends to its customers at the CB's discount rate + a margin to cover operating cost and credit risk. In effect the CB controls the lending rate in the economy (ignore deposits by assuming their rate is zero in each class: CB—banks, banks—customers).

CB/Open market (OM)

In practice, banks can lend reserves to each other. While assuming A holds zero reserve balance, let's bring in a third bank, C, who just happens to have $X in reserve.  Instead of activating the discount window, A can borrow C's reserves and instructs CB to transfer $X to B. Typically, this is done overnight. Note that no reserve is created; only transferred.

This approach, which is the norm, amounts to C substituting a loan for reserves. A's borrowing from C is balanced by the decrease resulting from removing $X from the customer's deposit. B has an increase in reserve $X (asset) resulting from B's transfer, which is matched by an increase of $X in the grocer's deposit (liability).

The discount window is now the exception: banks in need of funds use it when they cannot find a matching party. That is no longer the primary tool of monetary policy; it is open market operations. If the interbank rate overshoots the target, CB will offer to buy assets, usually government bonds.

In our example, A sells $X worth of treasuries and receives as much in reserve that are transferred to B. On the asset side, the CB records +$X in securities held and, on the liability side, +$X in reserve. If C wants to compete with the CB, it's going to have to lower its asking rate, and that's how the interbank rate is brought back to target.

CB/Quantitative easing (QE)

Central bank buying treasury bonds to flush banking system with reserves characterizes QE. In principle, this puts downward pressure on the interbank rate based on our explanations above. But, in practice, this is often done as IB rate is already rock bottom.  It's not even clear why, therefore, QE is even pursued. By buying long dated treasuries specifically, the CB may be trying to flatten the yield curve, whether to help the government raise debt or with the hope it has some (elusive) stimulus property. For this reason, QE often comes with the criticism that the CB is usurping the government's role of 'fiscal agent'.

The CB can offset the creation of reserve resulting from buying treasuries—sterilization—by selling another asset class, say, for example, corporate bonds. Or it may do the reverse swap, say if corporate bonds are in distress.



Bank's supply of credit is not restricted by their reserves, at least not in the mechanical way of the money multiplier as taught in your average economics class. As a consequence, the thesis that an increase in the Fed's reserve held by banks poses an inflation threat is spurious.

This view is acknowledge by central bankers (Monetarist experiment tried and failed) and gaining acceptance in academia (Steve Keen).

Government is not revenue constrained (MMT)

Printing money is an expression used indiscriminately and writers who use it without explaining what it is exactly are probably bluffing. That's a general criticism, which does not target Modern Monetary Theory. Quite the contrary, but, as I now argue, their technicality is deceptive.

Because CB's capacity to create reserves is infinite, the MMT school argues that 'the government is not revenue constrained' or, equivalently, is 'never insolvent'.  That's because, ultimately, the treasury has the guarantee of the CB. They claim government is, instead,  'inflation constrained', that is, it can spend (provided CB subservient) until output gap is closed. As part of their credo, they stress that government spending does not crowd out the private sector.

This definition of 'revenue constrained' is weak. To create a loan/deposit, a private bank needs a credit worthy customer that asks for a loan.  Granting that loan, it should be noted, does limit the next credit worthy customer's chances of getting a loan. There is no crowding out. In fact, as we have learned from the US bubble, private banks can lend indiscriminately so long as markets' expectation of a boom becomes a self fulfilling prophecy.

MMT claims government (via CB) 'not revenue constrained' but implies that private banks are revenue constrained. Yet I don't see the difference.

I don't know definitely MMT is wrong, it's just a hunch. They tend to emphasize OM, which is complex and therefore amenable to fudging. The DW model is probably sufficient for thinking about macro policy. They frequently resort to obvious imagery of 'electronic keyboards to create reserve balances'  (to underscore gov not revenue constrained) or 'taxes paid in dollar bills that the government shreds' (debt is not really debt). This mix of complex and visually obvious does not convey the right impression.

Their practical recommendations are shared by Keynesian economics. AFAIK they don't rely on the 'not revenue constrained' assumption.

CB losses are fictitious

Thanks to its privileged position the CB could attempt to extract rent for itself. That would be self defeating because, in principle, the CB serves the public. It's objective, instead, is to set the interbank rate at that it deems optimal for the economy. But it doesn't follow from this that the CB incurring a loss is nobody's loss: it decreases its 'net worth' $-for-$, which has a corresponding entry as an asset in the treasury's book. If the CB incurs losses on a chronic basis, taxpayers are on the hook for it. The CB should extract just enough rent to cover its operating cost, not more, not less.

As noted previously, losses don't impair the capacity of the CB to 'print money' or, more exactly, create reserves. Technically, but all it would take for private banks to do so is to decree that they can because that's what it boils down to. And, in fact, it has happened. We don't know the true extent of the banks losses from the financial crisis. They were allowed to change their accounting rules to avoid bankruptcy. In the EU banks that have passed regulators (EBA)' stress tests successfully later were found to be loaded with bad debt.

What if the CB incurred such losses it had to be recapped? Again, the CB can operate with negative equity, just as a private bank could: it's a matter of convention/decreeing it. Since the premiss is it that it's not right for private banks to go into negative equity, why would it be right for the CB? Creating reserves wouldn't replenish equity. The government that is stakeholder would either have sell assets or borrow from the banking system to buy the equity.

Arguments that CB losses don't matter are usually supported by fuzzy explanations that require big leaps of faith.


UPDATE: The text that appeared here was integrally moved to the next post as it occured in hindsight that mixing theory and practice was a bad idea.


  1. "MMT claims government (via CB) 'not revenue constrained' but implies that private banks are revenue constrained."

    MMT makes no such claim. (What would this even mean?) Banks are CAPITAL constrained.

    1. Message from ECB-Watch awaiting moderation at 2012 presidential candidate and MMT figure Warren Mosler:

      I have challenged MMT in my blog by contrasting pure banking with a system managed by the Fed. The main difference is that the Fed has monopoly over supply of reserves. No biggie as MMT would have you believe and it doesn't follow from this that government is 'not revenue constrained' or that government debt is different in nature from private debt.

  2. Exactly, what does 'not revenue constrained' even mean?

    1. 'Letsgetitdone' (as I recall) replied here but I deleted by error, sorry. Re comment, if you like.