'What should the Fed do?' is the wrong question. The right question is “Where should the Fed go?” Several commenters asked me for a critique of Miles Kimball’s new post on monetary policy. I like him a lot, but I’m afraid this post will be mostly negative. Of course that’s not surprising, as Kimball is like 99.99% of other economists; he looks at monetary policy through the wrong end of the telescope.
Most economists think in terms of actions the Fed takes with interest rates or the money supply to move the economy. In my view that has things backward. The correct approach is to first decide where you want to go. If you want NGDP to be 11% higher two years from today, you let the money supply and interest rates respond endogenously until NGDP is expected to be 11% higher. I prefer an explicit NGDP futures market, but could easily live with Lars Svensson’s approach, which is to adjust monetary policy so that the central bank’s internal NGDP (or inflation) forecast is equal to the policy goal.
The reason many are pessimistic about various tools like QE is that they don’t understand that the Fed doesn’t have a robust policy goal. The Fed seems pretty content with current NGDP. In that case, there’s no reason to expect QE to work, because the markets would not expect it to be permanent. I can probably make these points clearer by criticizing specific comments by Kimball:
So, interestingly enough, once the nominal Fed funds rate is zero, the level of the money supply doesn’t matter very much any more!
This is a common misconception. Indeed even Paul Krugman makes this argument, despite the fact that his own (expectations trap) model says otherwise. Consider the old “double the money supply” thought experiment so loved by the monetarists. They claim it leads to the quantity theory result that the price level will also double, at least in the long run. Now let’s consider two possibilities:
1. The money supply doubles, but the extra money is expected to be withdrawn 60 days later.
2. The money supply is permanently doubled.
Krugman would say in case one the doubling has no effect, and in case two the QT prediction applies. That’s because liquidity traps aren’t expected to last forever. Say it’s expected to end in 10 years. Then the price level would rapidly double in year 10. But if that was expected, then asset prices would soar in year 9, in anticipation. By backward induction we get asset prices soaring immediately, and the doubling of the money supply has a powerful stimulative effect, even at the zero bound. But only if expected to be permanent.
Krugman knows all this, but what he seems to overlook (and I suspect Kimball overlooks as well) is that it applies almost equally when nominal interest rates are 5%. A doubling of the money supply might do a little bit more at 5% interest than 0%, but not much. If expected to last only 60 days, asset prices won’t move much, as markets will understand the economy will go right back to the old situation in just 60 days. Fed funds rates will fall briefly, but that won’t be enough to trigger much long term investment. Investors care about where the economy will be in the long run.
So monetary policy is ALWAYS LARGELY ABOUT THE EXPECTED FUTURE PATH OF POLICY. I can’t emphasize this enough. And that’s what Kimball pays far too little attention to. He talks about monetary policy in a very mechanical way, as buying various assets. In fact, it’s mostly about signaling future policy intentions. Now I’m certainly not saying asset purchases don’t matter. I supported QE2. But the reason QE2 boosted asset prices was mostly because it signaled a new Fed determination to boost AD, whatever it takes. In other words, it was a indirect signal of future policy intentions.
Just to be clear, I’m not claiming the Fed should say the entire recent monetary base increase is permanent—that could lead to hyperinflation. Rather they should say enough will be permanent to boost 2 year forward NGDP by a certain figure, say 11%. But if they do that, then they don’t even need to talk about money, just set your NGDP target and let the market determine the appropriate levels of interest rates and money. Failing to see the endogenous nature of interest rates can lead to bad reasoning:
All interest rates matter for the economy, so lowering almost any interest rate will stimulate the economy.
Kimball has yet to encounter my “never reason from a price change” maxim. Interest rates plunged yesterday, and that was extremely bearish news for the economy (as we saw from the stock market crash.) Kimball would probably respond “OK, but other things equal lower interest rates are surely a form of stimulus.” But “other things equal” prices never change. Something must cause them to change. OK, how about “lower interest rates caused by monetary policy actions are stimulative.” Not really, a highly contractionary policy announcement can cause expectations of recession, which depresses long term rates. OK, how about short term rates? In December 2007 a highly contractionary policy surprise caused the stock market to crash at 2:15 PM, and caused bond yields from 3 months to 30 years to fall. I’ll grant you this, a stimulative monetary policy surprise will usually cause 3 month bond yields to fall, and will almost always cause the fed funds rate to fall. But as we saw from the previous post, even top Fed officials seem to think lower long term rates are “good news,” so I can’t emphasize enough the folly of focusing on interest rates as an indicator of monetary policy.
It’s beyond my comprehension why our monetary policymakers have not spent a few million dollars setting up and subsidizing trading in an NGDP futures market. Monetary policy decisions can wipe out trillions of dollars in asset values in a single day. Millions of jobs are at stake. And our policy makers act like they aren’t interested in having a real time indicator of the stance of monetary policy. Instead they fiddle around with the money supply and interest rates, like a ship captain steering a huge ocean liner through a thick fog, to a destination undetermined.
Monetary policy is all about shaping expectations of NGDP growth over the next two years. Nothing else really matters. Everything else should be subservient to that goal. Until we start thinking of policy in those terms will be stuck in an Alice in Wonderland world where low rates might be easy money, or tight money. Where an enlarged monetary base might be easy money, or the belated response to a tight money policy that drove NGDP so far down that nominal rates fell to zero.
PS. Kimball’s post had some good points, such as the focus on many different asset prices. But there is too much talk of the Fed directly affecting them via asset purchases, not enough talk about how they indirectly affect them via changing expectations of future NGDP, which then impacts current asset prices. He also does the reductio ad absurdum of the Fed buying up all of planet Earth. I like that example, but (unlike Kimball) I’d characterize it is a success if it failed to boost NGDP. After all, wouldn’t it be nice if America owned the whole planet, and we could all kick back and live off the work effort of others?
PPS. Lars Svensson is one of the 0.01% who looks through the telescope from the right direction.