Monday, May 18, 2015

Interest on Reserves during the Financial Crisis and Going Forward

Here is a graph of:

1) Bank Deposits

2) Bank Deposits, net of excess reserves

3) GDP

Notice that bank deposits grew fairly steadily throughout the crisis and recession.  They leveled off shortly in 2010, after the crisis.  (This lagged decline in deposit growth happened also after the 1990-1991 recession, where deposits were relatively flat from 1992 to 1995.  Here is a previous IW post on the similarities between 1990 and 2007.)

But, if we look at deposits, net of excess returns, there is an unprecedented volatility in bank assets, beginning in the fall of 2008, coincident with the sharpest fall in NGDP since the late 1940s.  Because of the zero lower bound and the role of reserves in current monetary policy, this measure has continued to be volatile.  But, since 2008, this volatility has not been associated with extreme and predictable variations in NGDP.

I think the reason is that, since the implementation of QE, changes in reserves have come from the injection of cash into the economy.  The Fed purchased long duration treasuries, and subsequently sold short duration securities back to the banks, that were in the form of excess reserves.  This was an exchange of liquid assets, and I think it facilitated an exchange of equity based real estate ownership that replaced debt based ownership.

But, the original shock to deposits minus reserves was not associated with this exchange.  That period was chaotic.  In this next graph, the red line is bank deposits, net of excess reserves, plus all federal reserve assets.  The purple line is bank deposits, net of excess reserves, plus conventional open market operation assets (in which I include treasuries and MBS).  The difference between these two measures (between red and purple) is the "lender of last resort" activity that the Fed engaged in during the crisis.  We can see that there was none of this before 2007, a small amount stemming from the failures earlier in the recession, and a sharp rise in September and October 2008, after the Lehman failure.  Since then, these assets have been slowly declining.

If we just look at Fed assets associated with a typical injection of currency as part of open market operations, (the purple line) we can see that in late 2008, until QE1 kicked into gear, there was a flight of cash from banks back to the Fed.  We can see here that, since then, reserves and QE injections have roughly matched, so that deposits minus reserves plus Fed assets has grown as steadily as deposits.

We should remember that when the bottom fell out of the economy in the fall of 2008, the Fed was far from the zero lower bound.  The recession, as defined now, had begun in December 2007, but that was not declared by the NBER until November 28, 2008.  So, we were not officially in a recession in September 2008, the Fed Funds rate was at 2%, and the FOMC decided to hold the rate steady in defiance of market expectations of a decrease, as a prophylactic against inflation.

It now seems clear that the Fed was taking an extremely tight stance at that meeting.  This has not been the case since the implementation of QE.  Or, at least, bank liquidity has not been the constraint on nominal economic expansion since then.  I have previously looked at interest on reserves during the crisis period.  Here is a graph from that post, comparing effective Fed Funds rate, target Fed Funds rate and the IOR rate.  When the Federal Funds rate was at 1%, and IOR was pegged at 1% also, in November and early December, the effective Fed Funds rate was well below the target rate.  And, it  was during this period that bank deposits were piling into excess reserves.

The IOR rate is a contracted rate, controlled by the Fed.  As far as I know, it remains fixed.  But, the Fed Funds rate is a target, which the Fed influences by buying and selling securities in open market operations.  Normally, the Fed would raise the effective Fed Funds rate by selling securities, sucking cash out of the economy.  But, clearly what we needed in late 2008 was not a sharp tightening of monetary policy.

What I think happened was that the Fed had set the Fed Funds policy above the neutral rate.  After the disastrous September 16 FOMC decision, even before IOR was implemented, banks were collecting reserves in an unprecedented way.  Since banks were hoarding cash, the effective Fed Funds rate was already pushing well below the target rate.

Once the IOR rate was pushed above the neutral rate, the only remedy would have been to push IOR back down.  But the Fed didn't do that, so while bank deposits grew at a normal pace on banks' liability ledgers, hundreds of billions of dollars were being put back to the Fed as reserves on the asset ledgers.

It seems to me that a big difference regarding the monetary base between using IOR and Fed Funds is that when the Fed uses the Fed Funds rate, it is generally in control of the quantity of cash it is injecting or pulling out.  But, with IOR, the banks are in control of the quantity of cash being exchanged.  The Fed's institutional inertia in the face of that problem in November 2008 was disastrous.

A further problem with an IOR based policy is that IOR below the neutral rate may have minimal effect on reserves and bank balance sheets.  Currently, expansion in bank credit is not encumbered by limited reserves.  (In modern banking, I'm not sure that it ever is.)  But, if IOR is pushed above the neutral rate, as it was in late 2008, there is a tipping point where reserves come rushing out of the banks.  It's like a lender-initiated bank run.

I wonder if this is a signal to look for if the Fed begins pushing the IOR rate up.  I suspect the neutral rate is currently above the Fed Funds rate.  But interest rates aren't the constraint on faster nominal expansion.  (Regulatory pressures on mortgage availability are strong.  Banks have assets of about $7 trillion in loans to businesses and households, and half of that is real estate.  So, any growth in credit is only coming out of half the balance sheet.)  Currently, banks are expanding their loans and leases and reserves are slowly being converted to currency.  But, if the IOR rate pushes above a tipping point, and reserves start to grow while the Fed's asset base remains level, that could be a real problem.

The Fed didn't react quickly to that problem in 2008.  Would they act quickly next time?  There are a lot of folks who see "asset bubbles" all over the place, who are impatient for the Fed to raise rates.  I suspect that if this problematic scenario plays out, bubbles will be blamed.


  1. Interesting post; not sure about one item.

    When the Federal Reserve buy bonds from the 22 primary dealers, it then credits the commercial bank accounts of those dealers by an equal amount. This creates the reserves.

    So, I am not sure what this means:

    "The Fed purchased long duration treasuries, and subsequently sold short duration securities back to the banks, that were in the form of excess reserves."

    1. You're right. My wording is a bit awkward there. And, I'm really no expert on the microstructure of Fed operations. I was just trying to get at the idea that when the Fed buys securities from the primary dealers, the dealers will purchase other securities with the money, so the money doesn't travel immediately into a time deposit at the banks.