Reply to Miles Kimball
Miles Kimball has a new post that replies to my argument for a shadow fed funds target:
. . . in his post “What Ben Bernanke Can Learn from Humpty Dumpty.” Scott talks about a “shadow fed funds rate target.” I considered that phrase for my proposal, but set it aside because I thought it sounded too much like the fed funds rate target of the Shadow Open Market Committee . . . The word “virtual” not only evokes the “virtual reality” of computers that allows the impossible, but also the “virtual particles” of quantum mechanics which can arise even in empty space, stealing the energy for a ghostly existence using the opening provided by the Heisenberg uncertainty principle.
In addition to using “shadow” instead of “virtual,” there is a crucial difference between Scott Sumner’s proposal and mine. I am envisioning the Fed as using the virtual fed funds rate target to communicate what it intends to do in the near future with balance sheet monetary policy. Scott wants the Fed to use a shadow fed funds rate to communicate what it will do in the more distant future. Aside from “doing whatever it takes” to reach a nominal GDP goal, I am not clear about exactly what the Fed is communicating it will do in the future.
Kimball is new to blogging so he doesn’t yet know my fanatical opposition to interest rate targeting. In my post on the shadow fed funds target I was basically saying; “Keynesians are making a tragic mistake with interest rates targeting, but by all means if you insist on setting interest rate targets don’t stop when the economy most needs the Fed to steer it, i.e. when we are in a deep slump.” So my views are closer to Kimball’s than he suspects, I much prefer adjustments in the base as a policy instrument, and would certainly not want to stop doing those adjustments if we used a virtual fed funds target.
In my view even during normal times changes in the fed funds target are actually signals about future changes in the monetary base, so to answer his question “about what exactly the Fed is communicating,” a cut in the virtual fed funds target would be a signal to the market that the future monetary base would be higher than they now forecast. But of course no one really pays much attention to the base; they care about the interaction of changes in the supply of base money and base velocity, i.e. NGDP. So it would actually be better described as a signal about the Fed’s future NGDP intentions. Wouldn’t it be better for the Fed to just come out and say where it would like NGDP to be 2 or 3 years out? Of course, but you go into battle with the Fed you have, not the Fed you wish you had. Kimball continues:
One possible meaning of a shadow fed funds rate target that may or may not be what Scott has in mind is as follows. I hope that Scott will clarify his position, either by saying that the following interpretation is consistent with what he intends, even if he thinks the emphasis is off, or by distancing himself from the following view. Believers in Wallace neutrality as applied to the real world think that the only way the Fed can do more stimulus once the fed funds rate is at zero is to promise to overstimulate the economy in the future once the fed funds rate has lifted off from zero. For example, even with Wallace neutrality, the Fed can stimulate the economy now by promising to keep the fed funds rate at zero so long that the economy will get overheated in the future. This would be very different from what I recommend. The great virtue of balance sheet monetary policy (which works by taking advantage of departures from Wallace neutrality)””and therefore with the virtual fed funds rate target that communicates the stance of balance sheet monetary policy””is that it avoids making promises to do the wrong thing in the future in order to have the right effect now.
In theory Wallace is right and Kimball is wrong, but in practice Kimball is right, because the conditions for Wallace neutrality to hold would probably never occur in the real world. This will take some explanation, so bear with me.
I frequently argue that most people look at liquidity traps through the wrong end of the telescope. They agonize over whether this or that Fed action is enough to move the economy to a certain position. The base is viewed as an exogenous policy instrument. That’s looking at things backward. The right question to ask is how much base money does the public want to hold at each and every expected NGDP growth rate. In general, lower NGDP growth rates (and also lower NGDP levels relative to trend) are associated with lower nominal interest rates and higher demand for base money. During normal times people in most countries want to hold about 5% of GDP as base money. Right now it’s 4% in Australia, and it’s usually about 6% in America, as our currency is more popular with foreign hoarders. But when NGDP growth is unusually slow, the demand for base money can rise sharply. It’s up to about 18% of GDP in America (albeit partly due to interest on reserves) and up to 23% of GDP in Japan, which has an even lower NGDP growth rate.
Suppose the demand for base money exceeded the entire national debt, assuming that NGDP growth was expected to be on target. In that case Wallace would be technically right, at least if you define monetary policy narrowly, as the exchange of base money for government bonds. Consider all the “foolproof” escapes from the liquidity trap. My NGDP targeting idea, Svensson’s currency depreciation, Friedman/Kimball’s massive QE, etc. They are all supposed work by avoiding the zero bound. There is no zero bound for increasing in the base, the price of NGDP futures, or the price of foreign exchange. But here’s what might happen; you might have to buy up more than the entire national debt to hit your target. I actually don’t worry about that for several reasons. One is that you could buy other assets; gold or corporate bonds. But others would consider that fiscal policy, or at least quasi-fiscal policy. But the other reason I don’t worry about that happening is that I don’t think the demand for base money would be very high if NGDP growth was set at an appropriate target.
I am pretty sure that the Wallace claim that you’d need excessive inflation in the future to escape a zero rate trap is based on some combination of assuming inflation rate targeting (rather than price level targeting) and/or assuming that the central bank currently has an appropriate target. I.e. rates are not near-zero because of excessively tight money policy, but rather because the equilibrium interest rate is low even with “appropriate” monetary policy. Hence you need “excessive” monetary stimulus to escape the trap.
I favor level targeting, and I notice that every real world example of “liquidity traps” are cases where the central bank has allowed NGDP growth to fall far below any reasonable trajectory, with no intention to catch up. The one exception was a brief period in 1933 where FDR tired to reflate to the pre-Depression price level, and that policy was instantly effective, producing 20% WPI inflation in an economy mired in 25% unemployment and near zero interest rates. It’s the exception that proves the rule (or actually that proves the limitations of the rule.)
So here’s the right way for people to think about 1930s America, post-1995 Japan, and post-2008 America. They are all cases where the central bank had an excessively low implicit NGDP target, and hence the nominal rate fell to zero and the demand for base money was large. But if you believe level targeting, then setting a higher NGDP target would not be “promising to be irresponsible” (in Krugman’s terminology), but rather promising to be responsible. All three of those cases the central banks wasn’t trying to boost NGDP growth to an appropriate rate. Recall that in Japan, and the US during the 1930s, they even tightened policy on occasion, which shows they were “steering the economy” but along the wrong NGDP trajectory.
Society faces a very simple choice:
1. Set the NGDP target path at such a slow rate that the public wants to hold an amount of base money that exceeds the national debt. In that case rates fall to zero, and bonds become indistinguishable from money debt. You might as well monetize the entire debt. That’s the world Wallace contemplated. Obviously monetary policy doesn’t work, because there is no interest-bearing government debt for the central bank to buy. And by assumption, purchases of non-government assets are called “fiscal policy.” In that world you’d need some other tool for steering the economy. It might be interest-bearing reserves, where the rate can go negative (although interest on currency presents technical problems.) Or it might be changes in the size of the public debt. I strongly recommend against going that direction, at least until we replace currency with all-electronic money—probably around the year 2050.
2. Or you can set a NGDP target trajectory, level targeting, where the public does not want to hold an amount of base money in excess of the national debt. That’s the world we actually live in, which is why I agree with Kimball. Even QE, clumsy as it is, can “work” if done in sufficient quantities. There is some amount of QE that will speed up NGDP growth, without causing an inflation/unemployment combination that violates the Fed’s mandate. But there really are much better ways of doing policy at the zero bound. It would be much easier to set an explicit and robust NGDP growth target, level targeting, than to rely on QE to hit some vague and amorphous fed target that Bernanke won’t disclose.
A footnote. Last night I was revising Chapter 9 of my depression manuscript, and suddenly realized that FDR was using a virtual price of gold to steer monetary policy in November 1933. He kept adjusting upward the official buying price of gold, even though in a sense it was meaningless (it wasn’t the free market price.) And yet it seemed to “work”, in the sense that markets responded to the changes. When I did this research 20 years ago I concluded that it worked by sending signals to the market about future monetary policy—it showed where the government was likely to re-fix gold in 1934 (the actual figure turned out to be $35 /oz. ) Thus every time FDR raised the official price of gold (more than expected) he also raised expectations of the future money supply. Every monetary economist should spend a year studying the interwar period. It’s a gold mine of natural policy experiments.
Update: I did not mean to suggest that zero interest rates imply the public wants to hold a level of base money larger than the national debt. Rather I meant that you could envision in theory (but probably not in practice) and NGDP target path so low that the demand for base money exceeded the national debt. That scenario (unlikely to occur in reality) would also feature near-zero rates.