Arash Vassei on market monetarism
Over at Kantoos there is a very interesting guest post by Arash Molavi Vassei on market monetarism:
Yet, none of defining characteristics of MM depend on disequilibrium analysis. They rather fit the “equilibrium always” views of money (that rely on money as a unit of account rather than a medium of exchange). In fact, some of the characteristics fundamentally depend on equilibrium reasoning.
1. New Keynesian models predict Rational Expectation Equilibria (REE) with AD-constrained output. Given the expectation channel, expected below-trend nominal spending (aka NGDP) relates to low current spending (if not countered by monetary policy). Given sluggish nominal values, current output falls below its optimal level. In contrast to Christensen’s description of NK models (fn. 8), all markets clear (not just the bond market), though at inefficient levels.
It follows that NK models are consistent with any view that traces the Great Recession to an exceptionally large AD slack. More precisely, since NK models assume that credibly committed monetary policy has full control over expected NGDP, they can accommodate the MM hypothesis that the crisis is due to a mismanagement of NGDP-expectations. In fact, whereas the much more common finance-based interpretation of the crisis asks for some kind of specifications by means of additional frictions, the MM interpretation can rely on pre-crisis variants of the NK model (Eggertson and Bernanke, Svensson, …). Scott Sumner heavily relies on such reasoning.
2. The major advantage of level targeting is that it implements a memory-based adjustment regime. If a central bank is credibly committed, then market expectations quasi-automatically correct deviations from target levels. I cannot imagine a more suitable framework for this kind of analysis than the class of REE models. This also applies to Svensson’s suggestion to target forecasts (and Sumner’s suggestion to target market forecasts). And I have no idea how disequilibrium analysis (or upon Nick’s comparative statics by means of an advanced IS-LM model) could ever improve upon such “equilibrium always” models.
There’s an old joke about putting 7 economists in a room and getting 8 opinions. So I won’t speak for the other market monetarists, who seem to find more value in the disequilibrium approach to monetary policy than I do.
I think Vassei is right, at least about my views. It is true that much of my analysis is consistent with mainstream new Keynesian thinking. But there is one area where it seems to differ sharply, the way we analyzed the policy situation in early October 2008. I attribute that to the greater emphasis that I place on market forecasts of NGDP.
New Keynesians use computer models to try to identify monetary policy shocks. The models use interest rates, and various macro aggregates. I use NGDP futures prices. An expansionary shock is a rise in NGDP future prices, and vice versa. If the central bank is too stupid to have created and subsidized trading in an NGDP futures market, then you infer NGDP expectations by looking at all sorts of other market data and consensus private economic forecasts.
In early October 2008 I went ballistic. It was at that point I realized that my views were actually far outside the NK mainstream. It was obvious to me than NGDP expectations for 2009 were collapsing, and yet the Fed wasn’t really doing anything about it. They weren’t setting policy at a level expected to produce on target NGDP growth. BTW, this failure was not due to the zero rate bound–nominal rates were above zero for the entire June to December collapse in NGDP (I’m using monthly NGDP estimates from Macroeconomics Advisers.)
And no one seemed to care! I would have expected economists to be out picketing the Fed building in DC. Yet if they said anything about Fed policy, it was generally that the Fed was doing a good job. Most said nothing about monetary policy, and merely pontificated on banking policy.
Of course there are other important differences. Most NKs favor inflation targeting, whereas I favor NGDP targeting. Most favor growth rate targeting, whereas I favor level targeting. But these differences are not decisive. You can find NKs who favor NGDP targeting, or some analogous policy for addressing the Fed’s dual mandate–such as the Taylor Rule (although this no longer includes John Taylor!) And Michael Woodford has pointed to the advantages of level targeting at the zero bound. No, it’s the targeting the (market) forecast that is decisive, at least for my views on monetary policy.
For 25 years I had almost no opinion on monetary policy. I thought the Fed was setting interest rates at a level expected to produce roughly 4% to 6% NGDP growth, which was fine by me. Suddenly in October 2008 they weren’t doing that, and almost no one seemed to care. Or I should say almost no one except us market monetarists, plus a few academic economists like Robert Hetzel, who pointed out in an early 2009 paper that Fed policy had been too tight in 2008.
NKs favor targeting interest rates, at which point the monetary base becomes endogenous. I favor targeting NGDP futures prices, at which point the monetary base becomes endogenous. Many people wrongly assume that I am a sort of “QE2-type economist,” who wants to try money supply injections to see what happens. Actually it’s just the opposite, I want an endogenous money supply, and a policy of pegging NGDP expectations (either futures contracts, or if they don’t exist, then a weighted average of various market indicators like TIPS spreads.)
PS. Don’t forget that new Keynesianism is itself a sort of quasi-monetarist model.
Tags:
19. September 2011 at 06:29
The good thing is that the MM idea is being “spread”. It was a good idea by Lars to do a “primer”.
On the “Equilibrium” thing, I say: Dynamic, yes; Sochastic, yes; GE, no!
Taylor just put a coefficient of zero on the output gap part of his “rule”. I “commented” on it:
http://www.capitalspectator.com/archives/2011/09/strategic_brief_42.html
19. September 2011 at 06:47
“And Michael Woodford has pointed to the advantages of level targeting at the zero bound.”
How can you credibly only do this at the zero bound?
“Many people wrongly assume that I am a sort of “QE2-type economist,” who wants to try money supply injections to see what happens.”
MOST REAL PEOPLE (the ones who matter) need you to spend time talking about the positive effects of targeting when money is too loose how it would have avoided 2008 altogether.
The fact that you start at Oct. 2008, is WHY,m EXACTLY WHY, most people think you are some QE2 type economist.
I know this, because if you didn’t have that bit I’d call you all DeKrugner
19. September 2011 at 08:27
Marcus, Good points. It’s a zero coefficient on output, as you say.
Morgan, I agree about Woodford, I think level targeting should be done all the time, not just during liquidity traps.
I started favoring NGDP targeting in the late 1980s (in my first published paper) At that time it would have meant tighter money. I’m not like those Keynesians who just come over when it means easier money.
19. September 2011 at 10:30
Scott, my god…
Pay attention!
You want conservatives to listen to you BLOG ABOUT how under 4% NGDP we would have tightened rates in 2005, 2006 and saved America.
IF your ideas can’t tell that story, you aren’t gong to get real attention.
IF your ideas can tell the story, my god man, stop burying the lede – put the meat in the window!
19. September 2011 at 16:01
This and the Taylor post got me thinking about level targeting and measurement error. I am about to teach Orphiandes’ (2003) paper which argued that the 70s inflation was mainly the result of the Fed’s mismeasurement of potential output, so using a Taylor rule with real time data wouldn’t have prevented it. With serially correlated measurement error, targeting changes is safer than levels.
How can level targeting get around this? (I know your plan uses futures, but at some point the futures have to be settled based on actual data.) You pointed out how GDP back in ’08 has recently been revised. Is a statute of limitations & chaining enough to handle it?
19. September 2011 at 16:08
And Taylor is advocating change to the legal mandate of the Fed, not the Taylor rule. He talks about the 80s and 90s as when the Fed got it right, and that is the period the Taylor rule fits best. Maybe he is just being inconsistent but I don’t read him as advocating a zero coefficient in output (more like you need a price mandate to force them to follow the Taylor rule).
20. September 2011 at 00:53
[…] Scott Sumner, Josh Hendrickson and Nick Rowe have some interesting responses to Arash’s […]
20. September 2011 at 08:22
Declan, There is no way you could get the 1970s with level targeting. The NGDP growth rate was 11%. Let’s say the target was 4%. In that case, after 10 years NGDP would be double the target, and the Fed would be shooting for a 50% reduction in the following year. Obviously that can’t happen–level targeting prevents anything like the 1970s from happening.
I also recall that in years like 1971/72/73, the forecast NGDP growth rate was close to 10%, so it wasn’t just an ex post problem. The Fed was way too expansionary ex ante.
I don’t follow your second comment. If he’s not advocating a zero coefficient on output, he’s expressed himself very poorly.
20. September 2011 at 16:13
Scott,
I didn’t say you would have got the 70s with NGDP level targeting. (That was just an example of how measurement issues can be dangerous for level targeting in general.)
But how would it deal with the recent GDP revisions? If I understand correctly, they made a big revision to 2 year old data. How would your scheme handle this, or (if I have got this specific case wrong) any large revision made after the futures contract expires? This might seem unlikely, but so did the zero lower bound a few years ago. (Or a NAIRU of 6% + productivity slowdown in 1970!)
Taylor may have expressed himself poorly. But it’s pretty clear that:
1. He is explicitly advocating changing the Fed’s legal mandate, not the Taylor rule. (He does not mention the Taylor rule at all in the article.)
2. He still thinks that policy in the 80s and 90s was good.
3. The Taylor rule fits the 80s and 90s well.
My conclusion is that he does not advocate changing the Taylor rule ie a zero coefficient on output.
How is this compatible with not mentioning output/employment in the legal mandate? Maybe he thinks that a sole price mandate will force actual behavior more like the Taylor rule (like if you set a 50 speed limit, you know people will still do 55, but they might not do 100).
The way he describes the 80s and 90s is consistent with this interpretation (the Fed talked mostly about prices but set rates like it was following the Taylor rule).
21. September 2011 at 10:58
@Declan,
You don’t actually target NGDP. You target expected NGDP. It doesn’t require you to know what potential output is.
As for revisions, that’s a matter for whoever writes the NGDP futures contract to worry about. Personally, I don’t think that the contracts would actually be for NGDP, they’d be for some other nominal quantity that is easier to measure and verify.
21. September 2011 at 15:23
Jeff – I never said NGDP targeting required knowledge of potential output. I said:
1. Level targets IN GENERAL are more demanding of accurate measurement than growth targets, as shown by the potential output example of the 70s.
2. NGDP targeting is not immune to this problem, as shown by the GDP revisions Scott talked about a month or two (?) ago.
Maybe you have your own plan, but I thought Scott’s depends on NGDP futures contracts, which would obviously require that NGDP be measured and verified at the time the contract was settled.
26. September 2011 at 11:33
Declan, I discussed this on a more recent comment thread–but agree it is a problem. I just don’t see it as a major problem.
You could slightly modify the level targeting rule to adjust for this problem.