Many investors say they are not concerned about inflation, created by the Federal Reserve asset purchases, since the reserves are not being loaned out, and simply sit on the Fed's balance sheet as excess reserves. I decided to test this hypothesis.
First, I downloaded the following annual data series from the St. Louis Fed's data site, FRED:
|Adjusted Monetary Base||AMBNS||1959-2012|
|Consumer Price Index||CPIAUCSL||1959-2012|
The data was downloaded using Fred's excel tool which made an annual time series for each year, out of the monthly average for said year. I then took the adjusted monetary base and subtracted the excess reserves to arrive at a data series of reserves, which do indeed support loans and lending.
I then imported this data into the econometrics software Gretl. I made a time series of log difference returns from the yearly reserves and CPI. I then preformed a cross-correlogram comparing reserves (ld_AMB) and CPI (ld_CPI) as shown below:
(click to enlarge)
This shows that CPI has the greatest, and statistically significant, correlation to the increase in the money base two years ago.
I then created an OLS regression analysis of CPI and the adjusted money base from 2 years ago (robust standard errors were used because of autocorrelation of the residuals). The results showed that 14.9% of the movement in CPI could be explained by the amount of change in base money two years ago, from 1962-2012. This is shown by the adjusted R^2 statistic.
The number .48176 is the number to multiply by the change in the money base from two years ago to arrive at an inflation forecast.
The model does show a likelihood of increased inflation in the coming years because of the Fed's expansion of its balance sheet; even if most of the asset purchases are sitting as excess reserves, enough reserves have been created to push inflation higher.