Noah Smith and Zachary David refute the EMH
By Scott Sumner
Suppose markets are efficient at pricing assets. And suppose everyone believes they are efficient. In that case (some argue) no one would have an incentive to gather information, and trade on that information. But in that case, speculators and arbitragers would have no incentive to push prices to equilibrium in the first place. So the EMH would seem to be self-refuting.
But if the EMH were false, even a tiny bit false, then speculators would have an incentive to jump in and try to take advantage of any tiny imperfections. So that’s an interesting paradox.
In fact, we do see lots of trading. We see speculation and arbitrage, even though asset prices are usually close to a position where risk-free arbitrage is almost impossible. That tells me one of two things, probably both:
1. The EMH is not precisely true—some people do unearth information not available to the overall market, and do profit on that information.
2. The EMH is 100% true, but people don’t think it’s true. These non-believers in the EMH trade assets, trying to make a profit. If Tesla is priced at $200/share, some people think it’s worth $180 and others think it’s worth $220 (say after some new information comes out.) These people trade shares, because there is a diversity of views.
Either of those assumptions are enough to rescue the EMH. Some economists would deny that assumption #1 rescues the EMH, but they suffer from a lack of wisdom. They put too much weight on the literal meaning of words, and not enough on the actual meaning. If the EMH is almost true, but not precisely true, it is still truer than virtually any other theory in the social sciences, including basic workhorses like “supply and demand”, and “comparative advantage.” A theory that is 99% true is essentially “true” in any social science.
Sometimes when people are first confronted with NGDP futures targeting, they sort of lose their minds, and engage in bizarre circular reasoning. They make arguments that they would never make for any other market. They say, “If the theory works, then why would anyone trade NGDP futures? After all, they would not expect to make a profit.” Well, expected by whom? The market, or the person doing the trade?
These claims are equivalent to saying that arbitrage won’t cause the interest parity condition to hold, because if it caused the interest parity condition to hold, then no one could make money by doing arbitrage. In other words, these skeptics prove too much.
In fact, people would certainly trade NGDP contracts if they had diverse views as to the future expected NGDP, and they would also trade NGDP futures contracts if (unrealistically) they all had identical views. I hope the diverse views case is obvious; so let me focus on the (far less realistic) identical views case.
Let’s suppose the Fed has a 5% NGDP target, and the base is at $1 trillion. All traders have the identical view that in order to hit the 5% target, the Fed must boost the base up to $1.03 trillion. Some of my skeptics would claim that a base of $1.03 trillion is not an equilibrium as investors would not expect any profit at that money supply level. This is what they mean by the policy “not working”. But that’s a very narrow way of thinking about success. I’ve never argued that in equilibrium the market price will be precisely equal to the market expectation. Again, nothing in the social sciences works perfectly. I’ve always acknowledged that there would be a (probably small) risk premium embedded in NGDP futures prices. That’s obvious, but is it a problem?
The basic argument for NGDP futures targeting is that the risk premium, the difference between the market price and the market expectation, would be relatively small. And by “relatively small” I mean of no macroeconomic significance. You might have a 5.0% futures price and a 4.8% or 5.2% market expectation. But that’s a smashing success in macroeconomics!
In the case above, where investors thought a monetary base of $1.03 trillion was needed to hit the NGDP target, they would take a short position on NGDP futures, pushing up the monetary base, at least somewhat closer to $1.03 trillion. How close? Close enough where further trades are not worth the bother. Where the expectation of profit is just counterbalanced by the risk. Again, that’s likely to involve a NGDP growth rate that’s really, really close to 5.0%, just as arbitrage in foreign exchange markets keeps interest rate differentials pretty close to the forward premium or discount on a currency, but not exactly equal.
Some recent posts by Zachary David and Noah Smith criticized my Mercatus NGDP futures targeting proposal. Even if their criticisms were completely correct, it would not affect my views on NGDP futures targeting, as the plan I currently advocate (Woolsey’s index futures convertibility) does not require any trading to occur. The Fed uses discretion, constrained by its reluctance to lose boatloads of money.
In fact, however, their criticisms are completely wrong, for not one but two reasons. First, even if the EMH were precisely true, people would trade because there is a diversity of views regarding future levels of money needed to hit the NGDP target (i.e. future levels of velocity). Their criticism proves too much, as it implies that no one would trade in any market for information reasons, only for liquidity reasons (say needing to sell stock for retirement spending.)
And it’s even more incorrect than that, because they wrongly assume that NGDP targeting must work perfectly in order to be a sensible macroeconomic stabilization policy. An expected future NGDP that fluctuated between 4.8% and 5.2% would be a massive success. Might it fluctuate further? Sure, anything is possible, but that would trigger exactly the sort of trading that Smith and David suggest would not occur, thus limiting the fluctuation. Their criticism of NGDP futures targeting exposes a lack of awareness as to the role of arbitrage in the financial markets. It’s like saying that if one opened the floodgates between two lakes with uneven levels, the water levels would not equalize, because once equalized, water would no longer have an incentive to flow to the lower lake!
They also discuss market manipulation, which I doubt would be much of a problem. After all, manipulation would open up even bigger profit opportunities for speculators making the opposite bet on NGDP futures, and side bets elsewhere. And, AFAIK, we did not see the sort of manipulation they discussed under either the gold standard or Bretton Woods, other cases where the central bank moved the money supply as needed to peg a key asset price. And finally, the version of NGDP futures that I am currently advocating (a band of 3% to 5% on NGDP futures contracts) is not at all susceptible to manipulation.
Before anyone else tells me why NGDP futures targeting cannot work, I implore them to think about how ordinary asset markets work. What causes prices to be at the efficient levels? And if they are not at the efficient levels, how close are they?
PS. Question for manipulation fear-mongers. Which big investment bank will do the manipulation, and which investment bank will be the sucker that gets taken to the cleaners? Is manipulation really as easy as it looks? Might they compete to “manipulate”?
PPS. Last time Noah Smith criticized the plan, he made the false claim that it was subject to “Goodhart’s Law”. Before that he suggested that asset markets are too irrational. He clearly doesn’t like the idea, and no matter how many objections I shoot down, he seems to come up with another. I wonder if some of my critics are annoyed that someone from lowly Bentley College came up with a clever idea. Perhaps Zachary David would find the idea less “goofy” if endorsed by John Cochrane. Or would he say that Cochrane doesn’t understand basic finance?
PPPS. In the future, I encourage people to actually read my Mercatus paper, before jumping in.
HT. Dilip, Patrick R. Sullivan