When he talks about "NGDP targeting", Scott Sumner actually means the following (quoting Williamson):
In its current incarnation, here's how NGDP targeting would work, according to Scott Sumner. The Fed would set a target path for future NGDP. For example, the Fed could announce that NGDP will grow along a 5% growth path forever (say 2% for inflation and 3% for long run real GDP growth). Of course, the Fed cannot just wish for a 5% growth path in NGDP and have it happen...One might imagine that Sumner would have the Fed conform to its existing operating procedure and move the fed funds rate target - Taylor rule fashion - in response to current information on where NGDP is relative to its target. Not so. Sumner's recommendation is that we create a market in claims contingent on future NGDP - a NGDP futures market - and that the Fed then conduct open market operations to achieve a target for the price of one of these claims.So basically, the Fed would do its utmost to keep NGDP futures prices on a certain path.
I have a problem with that. The problem is called "excess volatility". According to some theories, asset prices should be an optimal forecast of (discounted) future payouts - for example, the price of a stock should be an optimal forecast of discounted future dividends, etc. An optimal forecast should not respond to "noise"; in other words, if something happens that doesn't affect dividends, it shouldn't affect the forecast. This means that actual dividends should be more variable than prices - the dividends should have lots of "surprises".
But we can actually look at whether or not this is true! All we have to do is wait for actual dividends to come in, and then see whether past prices bounced around less than the dividends, or more. Robert Shiller was one of the first to do this, and here is what he found:
The jagged black line is the S&P 500, and the lighter, less jagged lines are dividends discounted by various discount rates. It's easy to see that no matter what discount rate we use, stock prices are a lot more variable than the fundamental value of stocks. This means that there is "noise" in stock prices - prices may respond to information about dividends, but they also respond to some other stuff that has nothing to do with dividends. They display "excess volatility". Lots of researchers have tried to kill the excess volatility puzzle, but none have really succeeded. In other words, markets may be "efficient" in the sense that you can't predict future returns, but those returns are probably going to depend partly on things other than fundamental value.
Now back to NGDP futures targeting. An NGDP futures price will probably experience excess volatility too. It will bounce around more than changes in actual NGDP. This means that the Fed will be trying to hit a very volatile moving target. One quarter, futures prices will soar, the next month they will crash, and the Fed will be trying to keep up, tightening dramatically in the first quarter and loosening dramatically in the second. This will happen even if a true optimal forecast of NGDP (which of course no one really knows) is relatively stable! Asset market volatility will cause policy volatility.
If you think that policy volatility has no costs, this is fine. But if you think that the Fed bouncing around wildly from quarter to quarter sounds scary, you should be scared of NGDP futures targeting. For example, volatile Fed policy, even if it adheres rigidly and credibly and permanently to an NGDP futures targeting rule, may cause expectations to become more volatile if a substantial number of economic actors fail to believe that NGDP targeting will succeed in hitting the target. That could cause volatility in things like inflation and real GDP.
Of course, this is also true of any Fed policy rule that takes market prices as a measure of expectations (for example, using TIPS spreads as a measure of inflation expectations) and then responds to those "expectations". This is one reason why, contra John Taylor, I support combining rules with judgment. But even in terms of rules, we can make guesses about excess volatility. If excess volatility is an increasing function of actual volatility, then using NGDP futures should worry us more than using inflation expectations. Real GDP bounces around a lot more than inflation, as this graph from Stephen Williamson shows:
If excess volatility is, say, 50% of actual volatility, then using NGDP futures in Fed policy is going to cause a lot more bouncing around than using inflation futures alone.
Anyway, this is not to say that NGDP targeting is a hopeless idea. But the futures-market aspect of what Scott Sumner is proposing relies on a version of market efficiency that is much stronger even than what most finance professors would accept.
Update: In an email exchange, Scott Sumner has clarified the nature of his NGDP futures market proposal. In a nutshell, he proposes that the Fed act as a market maker, buying and selling infinite quantities of NGDP futures at the target price. Demand for NGDP futures would then be used to determine Fed policy; if NGDP futures demand increased, the Fed would commence open-market operations to bring down expected NGDP. The price of NGDP futures would not move, but demand would swing from positive to negative, moving Fed policy as it swung.
It seems to me that the concerns expressed in this blog post apply to this sort of market as well. In the stock market, prices move in response to demand (not, generally, supply); we can infer from excess price volatility that demand exhibits excess volatility as well. Now, if prices could not move (which is really the same thing as banning resale), it's possible that excess volatility would diminish or even vanish. However, this is far from obvious. For example, if excess volatility is caused by overconfidence, as many behavioral finance theories predict, excess volatility would persist in Sumner's proposed setup.
If people believe that noise is actually information, that noise will directly feed through into Fed actions, causing Fed actions to bounce around too much.