George Selgin sent me a 2000 JMCB paper by David Reifschneider and John C. Williams, who were both at the Fed:

To reduce the effects of this phenomenon, in our simulations we incorporate a notional upward adjustment to the inflation target of the policy rule to offset the aver- age effect of the positive deviations to the rule that occur when interest rates fall to zero. In this way, policy attains its inflation goal on average. As shown in the next section, this bias adjustment is a nonlinear function of the target rate of inflation, among other factors. We use this form of adjustment because of its simplicity and transparency, not because it is optimal. Intuitively, a better strategy would be to employ a conditional adjustment to the policy rule that adjusts the funds rate down immediately before or after episodes of zero interest rates; in this way offsetting movements in the funds rate would be more likely to occur when economic activity is still weak and inflation low. We consider just such a strategy in section 5 of the paper. (Emphasis added)

The zero bound need not be a big problem for central banks if they utilize level targeting. Unfortunately, most central banks use growth rate targeting, where policy becomes less effective at the zero bound. In that case, the optimal policy is to make policy more expansionary than otherwise when you are close to hitting the zero bound. This reduces the risk of a zero bound problem. Unfortunately, the Fed did exactly the opposite during late 2008:

Screen Shot 2017-10-28 at 3.59.19 PM.png

Notice that the 3-month T-bill yield had fallen to 0.62% by September 19, 2008. At that time, the Fed funds target was still way up at 2.0%. Markets were signaling (correctly) that much lower rates were on the way. The Fed should have reacted to this warning by cutting their fed funds target in September, but they did not do so.

In the past I’ve criticized Fed policy for being inappropriately tight in September 2008. Given that both employment and inflation (market) forecasts were well below the Fed’s target, they should have eased policy. That argument did not even take account of the optimal strategy discussed by Reifschneider and Williams. In other words, money was already far too tight even ignoring the zero lower bound issue. But now we see that things were actually even worse–monetary policy should have been more expansionary than would normally have been appropriate given the level of inflation and employment forecasts. That makes the mistakes of late 2008 even more unforgivable.

The Fed was acting as if there was no threat at all of hitting the zero bound, when in fact the markets were signaling that policy was about to smash into the zero bound. What was the Fed thinking? Where was the back up plan for the zero bound? We now know the Fed had no back up plan, but in that case why didn’t they do everything possible to avoid the zero bound?

If there is no guardrail on the highway, and your steering is not working very well, then don’t drive right on the edge of the cliff!

PS: Speaking of George Selgin, I strongly recommend his recent piece on interest on reserves (as well as his earlier posts). George made me realize that there are actually two problems with IOR, one demand side and one supply side. I knew that the imposition of IOR in October 2008 reduced aggregate demand, and at the worst possible time. Selgin points out that the policy has also reduced aggregate supply. It causes bank reserves to make up a much bigger share of bank balance sheets than otherwise, and this crowds out lending to the private sector. Selgin notes that this is an example of financial repression. (This explanation oversimplifies his post; read the whole thing.)

This David Beckworth post on IOR and bank reserves is also quite informative. I agree with his suggestion that IOR be brought down to the level of market interest rates. (Indeed I’d prefer IOR be set well under market rates.)