NGDP Targeting: Changing Policy Changes Relationships (Part 2)

Scott Sumner at TheMoneyIllusion suggests that Evan Soltas provides the best argument for NGDP targeting. After offering praise he provides a brief criticism regarding Evan’s interpretation of some graphs in a recent blog post:

Evan states:

The first graph shows that even the most massive amounts of monetary expansion are ineffective medicine for NGDP expansion. What matters is expectations; growth in the medium run is conditional on expectations — not on the monetary base or price level.

The second graph shows the extent to which monetary policy seems to have forgotten this. Nominal income growth expectations have been right at zero since the recession, when before that they had been stable at the 5 percent level for decades. The findings come from this study by the Chicago FRB, which I found through this Chicago Magazine article. Paging Scott Sumner…

Sumner replies:

I mostly agree, and very much like the second graph, but I’d like to slightly quibble with the first graph.  I redrew it setting both the base and NGDP equal to 100 in 1933, not 1929:

Notice that the base rose in the early 1930s while NGDP fell sharply.  That’s because base demand was engorged by two factors, banking distress and ultra-low interest rates.  The ultra-low interest rates continued between 1933 and 1944, but banking distress fell sharply after dollar devaluation and the creation of FDIC.  Thus after 1933 the base and NGDP rose at roughly similar rates, both nearly quadrupling over those 11 years.  As you know, I don’t regard the base as a reliable indicator of the stance of monetary policy, because base demand can be highly volatile under certain conditions.  But the supply of base money is still very important; indeed it’s the major factor driving NGDP over the long term.

Judging this debate I’d actually like to argue that Evan may have the upper hand. Until the Great Recession, excess reserves in the financial system were maintained at a minimal level. As has been previously shown, demand for credit (loans) creates deposits. The Federal Reserve has primarily focused on targeting the price of reserves (interest rates), letting the quantity float by always providing a sufficient amount. The amount of base money should correlate reasonably well with the increasing demand for credit under these conditions. Steve Keen has shown in his models of the economy that changes in debt directly impact aggregate demand and thereby NGDP. One should therefore expect to see a positive correlation between NGDP and base money while those circumstances persist.

However, since the Great Recession the Fed has, to a degree, altered its approach to monetary policy by paying Interest-on-Reserves (IOR). Back in November of 2010 I blogged the Fed Stands in Own Way on Monetary Policy. In that post I highlighted a report from the New York Fed asking Why Are Banks Holding So Many Excess Reserves? As the authors’ note:

if the central bank pays interest on reserves at its target interest rate,..., the money multiplier completely disappears. In this case, banks never face an opportunity cost of holding reserves and, therefore, the multiplier process described above does not even start.

Paying IOR above the risk-free-rate has resulted in a surge of excess reserves and base money uncorrelated with any increase in the demand for credit. Changing Fed policy has altered the conditions under which the previous relationship existed. The supply of base money (which was correlated but didn’t drive NGDP) may therefore no longer even show any material correlation with NGDP (as long as current condition prevail). Evan appears correct:

What matters is expectations; growth in the medium run is conditional on expectations -- not on the monetary base or price level.