QE, irrelevant or not?

Stephen Williamson has been thumping the drum on the irrelevance of quantitative easing for some time now. See here, here, here, here, here, here. I jumped into the comments of his most recent on this issue, and have done so here and here as well in the past.

I've had problems squaring Steve's irrelevance theory with the very real fact that in the day-to-day drama of financial markets, traders with large portfolios think QE is very relevant. Because they are large traders, and because they think it is relevant, quantitative easing IS relevant. One way to square this is to conclude that both Steve and the markets are right, but it depends on how you approach the problem.
Steve is approaching the question from the Miller-Modigliani framework, which carries with it all sorts of assumptions. If those assumptions hold, he's probably right about irrelevance.
But as commenter Richard Serlin consistently points out, MM only holds under conditions that don't actually characterize the real world. Back in 2010, I think Steve was still willing to acknowledge those limitations.

So, the Fed swaps interest-bearing reserves for long-maturity Treasuries. Why does this matter? In a world with complete markets, as in Backus and Kehoe, of course it can't matter. The private sector will just undo this. However, we don't live in a world with complete markets. The question then is: what are the frictions in the world that could make QE2 matter, and matter significantly? Then, we want to know whether this intervention somehow makes us better off. Unfortunately, there is not good theory out there, or sufficient empirical evidence, to tell us much about what to expect.

My guess is that because Steve hasn't been able to find any frictions that might be relevant, he has become more willing to use an unfiltered version of MM in order to come out in favor of QE irrelevance. His is a strong-form version of MM applied to QE.

On the other hand, I think (along with many market participants) that QE is very relevant. In answer to Steve's 2010 query about "what real-world friction might make QE2 matter, and matter significantly?" I noted:

When the Fed bought up Agency MBS in 2009 it picked up about 25% of the outstanding float. With QE2 ($600b purchases, about $7T longer term debt outstanding) it is picking up less than 10% of the float.

The larger the buyer & the smaller the market the more will prices be pushed as a buy order is executed. So relatively speaking the Fed can skew prices much more in agency MBS markets than the broader Treasury markets. 

A smart buyer tries to hide and slowly accumulate so as not to buy at inflated prices. Naive buyers who announce purchases ahead of time will pay inflated prices. 

Operating in small markets and pre-announcing is the best way to overpay for assets, thereby reducing your financial strength, and thus diminishing the value of your liabilities. 

So I'd say that QE2 will probably reduce the value of the dollar, a liability of the Fed, but not as much as the agency purchases reduced the dollar. They will overpay to a lesser degree.

The friction, or departure from perfect markets, is that large actors like the Fed are not price takers but can impact prices.

Last fall, Steve seemed to have acknowledged that the ability to dominate a particular market like MBS, and therefore set its price, might create the sort of frictions that make QE relevant:

First, as I argue here, the asset swaps cannot matter. Second, while the QE1 purchases of mortgage-backed securities (MBS) may have mattered (possibly in some bad ways), under current conditions MBS purchases by the Fed cannot make any difference unless the Fed purchases dominate the market, which they will not. 

See here too.

I've adopted a "front-running" theory of QE in the past. Under ideal MM conditions, markets are perfectly competitive. All traders are price takers (which is also, incidentally, an assumption of infinite liquidity). But in actual market conditions, traders in financial markets are almost always price-makers to some smaller or larger degree. Markets don't offer infinite liquidity. How does this work in real life? Securities in formal exchange environments trade within a narrow band called the bid-ask spread. Buyers will often exhaust the amount of  some security available for offer and will have to move the price up higher in order to secure the amount they desire. Large desperate buyers can move markets quite significantly before reaching their requirements.

Other traders know that their competitors are fully capable of moving markets should they get desperate to get liquid (or less liquid) and will try to front-run them. Front-running means to properly anticipate the actions of a large buyer and accumulate a position ahead of that buyer at a relatively low price, then sell to them at higher prices. Front-runners do this by looking for "tells" and other sorts of indications that might give away other traders' intentions.

This explains why real-life traders play their cards close to their chest. They are anxious to avoid giving out any information that might lead their competitors to front-run them.

In an MM world in which all traders are price-takers and information universally available, the sort of scenario in which a desperate trader is front-run by other traders who have sniffed out their intentions simply doesn't arise. Price takers in an MM world don't have to worry about getting their orders filled at the right price - they get automatically filled by definition. Liquidity is infinite. And no one bothers to sniff out intentions since prices are given and not capable of being pushed up by desperate or foolish competitors.

So what is the Fed in a non-MM world? The Fed is a massively large trader who has the ability to move even the largest of markets. It is the 10,000 pound price-maker in securities markets.

Second, the Fed is a dumb-trader. Smart traders do not pre-announce their intentions. To do so would attract front-runners, push up the prices of their targeted securities, and force the trader to pay an expensive price for them. Do this often enough an you go bankrupt. In announcing large purchases, the Fed is allowing itself to be feasted upon by front-runners. Prices of securities get bid up in the market that the Fed is targeting, and this quickly works its way through the full array of asset prices. By the time the Fed consummates its purchases, the market price has already adjusted to take those buying pressures into account.

Steve likes to point out that the private sector can undo/replicate QE. I don't doubt he is right in a MM world. But in a non-MM world, no one in the private sector is large enough of a price maker to "outmake" the Fed. In perfect markets, a bank using repo-markets to finance trillion dollar purchases might be possible, but Lehman/Bear demonstrate how unforgiving overnight markets can be. Secondly, private sector participants in securities markets simply do not preannounce their purchasing intentions. They are not willing to be a dumb-trader and sacrifice investor capital. As a result, private actors cannot undo/replicate the Fed. The Fed is a wholly unique asset buyer, both large enough and dumb enough to have an effect on the broad market.

See Brad Delong and Miles Kimball on QE and MM