The extreme stance of monetary policy is such a critical factor in the financial markets here that it is worth spending a bit more time on the relationship between interest rates, inflation, and the monetary base.
Let's return to the concept of "liquidity preference" - basically the "demand curve" for base money (currency and bank reserves). To make this operational, we define liquidity preference as the amount of base money that individuals choose to hold per dollar of nominal GDP, given any particular level of short-term interest rates. The chart below shows this "demand curve" for money in monthly data since the 1940's. Notice that when interest rates are high, there is a significant "opportunity cost" to holding base money, so people cut back on the balances they hold. When interest rates are low, people are willing to hold a greater amount of non-interest bearing money per dollar of GDP.
What's critical about liquidity preference is this - while there are numerous combinations of T-bill yields, monetary base, and nominal GDP that will produce equilibrium (demand for money = supply of money), the three variables are "jointly constrained." For example, suppose that there is upward pressure on interest rates which reduces the attractiveness of non-interest bearing cash. If the Federal Reserve does not reduce the monetary base sufficiently to move left to the appropriate point on the "demand curve," the burden of adjustment is instead thrown onto nominal GDP. Since variations in real GDP have a fairly limited range, the majority of that adjustment is forced to take the form of price increases (i.e. inflation) sufficient to bring the ratio of the monetary base to GDP down to the appropriate level.
Similarly, if the Fed creates a great deal of base money, and the components of nominal GDP (real GDP and prices) are fairly "sticky", short-term interest rates will decline to a level sufficient to ensure that the additional money is held (this has been essentially the story of QE2).
If you like equations (if not, skip this paragraph), by our estimates, about 96% of the historical variation in U.S. money demand is described by a fairly simple equation relating the Treasury bill yield ("i") and the amount of monetary base per dollar of nominal GDP (M/PY): i = exp(4.25 - 129.87*M/PY + 84.42*M/PY_lagged_6_mos). In some of the recent pieces I've written, I've used the "steady state" of this equation, which is i = exp(4.27 – 45.5*M/PY). See the original "Sixteen Cents" piece for further details.
Below, I've plotted this liquidity preference relationship as a 3-dimensional surface. This is essentially the "policy surface" faced by the Federal Reserve. Nearly all of the historical data is captured by periods where the amount of monetary base per dollar of GDP changed by less than 1 cent either way over any 6-month period. The blue marbles on the graph represent actual data points since the 1940's. The marbles on the floor along the right side of the graph aren't technically off the policy surface, but are clearly outliers because of the abrupt shift in monetary base per unit of GDP that occurred between 6-month periods during the recent financial crisis, which were associated with a plunge in interest rates toward zero.
As should be evident, the historical liquidity preference relationships we've been discussing are very tight. This is why I am so adamant that quantitative easing is an irresponsible policy - we know how these variables are related. Specifically, it will be nearly impossible to normalize interest rates, even slightly, without a massive contraction in the Fed's balance sheet. Likewise, as we approach 17 cents of monetary base per dollar of nominal GDP, even the slightest exogenous interest rate pressure will imply the need for massive reversals in the monetary base in order to avoid steep inflationary pressures. My hope is that my previous comment Will the Real Phillips Curve Stand Up? makes it clear that there is very little "tradeoff" between unemployment and general price inflation.
Since a picture is often worth a thousand words, I've included a few additional perspectives of the "policy surface" that the Federal Reserve faces here. The chart below is a head-on view. The point at the far right shows the present stance of monetary policy. Charles Plosser of the Philadelphia Fed is quite correct that normalizing interest rates to about 2.5% would imply a reduction of nearly 50% in the Fed's balance sheet, but as I noted two weeks ago, the required cutback in the balance sheet is extremely front-loaded, as a non-inflationary move to a Fed Funds rate of just 0.25% would require a reduction in the monetary base from about 17 cents to less than 13 cents per dollar of GDP, taking the monetary base from .6 trillion to less than trillion - effectively reversing QE2 in its entirety.
The following chart shows the policy surface, along with actual data points, from the origin. We are now way out on the flat part of the curve. Again, in order to achieve even a slight endogenous increase in interest rates, the Fed will have to reduce the monetary base sharply. Alternatively, in order to offset the inflationary pressure from a slight exogenous increase in interest rates, the Fed would be forced to respond with a sharp tightening in its balance sheet. Both normalizing policy, and failing to normalize it, now present the economy and the financial markets with hazards that, in my view, were needless in the first place.
As a final note, the Fed does have an additional policy tool - the ability to pay interest on bank reserves, in hopes of preventing base money from becoming an inflationary "hot potato." The difficulty here is that the Fed's balance sheet is now leveraged 51-to-1. Even 0.25% of annual interest on reserves works out to about 12% of the Fed's total capital. The Fed is already in a position where a 35 basis point increase in long-term interest rates would effectively wipe out that capital. Though the Fed does earn interest on the Treasury debt it holds (which is remitted back to the Treasury for public uses), it would still take an increase in long-term interest rates of less than 1% to wipe out the Fed's capital as well as its entire net interest margin. So while the Fed might have the latitude to pay another 0.25% of interest on reserves, every extension of its present policy course, be it more quantitative easing, or paying interest on existing bank reserves, substantially increases its already untenable level of balance sheet leverage, and the likelihood that the public will quietly need to subsidize that balance sheet.
