Long and variable leads (a reply to Tony Yates)
Tony Yates expresses shock that someone calling himself a market monetarist could reject Milton Friedman’s famous “long and variable lags” claim about monetary policy.
I have great respect for Friedman, but when I did my research on monetary policy in the interwar years (which is the period where it is easiest to clearly identify monetary shocks) I wasn’t able to find significant lags. The monthly WPI and industrial production indices seemed to move almost immediately and sharply after monetary shocks. If NGDP data had been available, it would have also responded quickly.
Even worse, I found that Friedman and Schwartz had misidentified monetary shocks by focusing on the monetary aggregates, which often moved ahead of or behind the actual shock. Thus the devaluation that occurred in April 1933 led to expectations of future money growth, and these expectations led to an immediate surge in prices and output. That’s what led me to coin the “long and variable leads” phrase. I thought it would be obvious to people that I know that classical theories of causation don’t really allow for cause to follow effect. Rather that it is actually expectations of future money growth that caused the near term NGDP growth surge. I used a bit of poetic license to drive the point home.
Later I learned that Woodford and Eggertsson were doing similar research from a New Keynesian perspective. Here’s Yates:
Sumner cites Woodford and Krugman as commenting on the potency of expectations, and uses this in support of his thesis that changing expectations changing things refutes the long and variable lags thesis. But I am quite sure neither of them believe any such thing. Estimated versions of Woodford’s model (for example, the original Rotemberg-Woodford model) behave just like my account above. And Krugman is a firm believer in sticky prices, talking interchangeably between IS/LM and New Keynesian models. Which behave just as I’ve explained above.
The only model I know where monetary policy has its entire effect instantaneously is the flexible price rational expectations monetary model. And in this case there is no point in monetary stabilisation policy at all. Money has no short-run effects on output. Optimal policy in this model is to set rates at zero permanently, obeying the Friedman Rule. If there are real frictions in this model, like financial frictions, there will still be a role for fiscal stabilisation, however.
I’m sure these mix-ups would get ironed out if MaMos stopped blogging and chucking words about, and got down to building and simulating quantitative models. Talking of which….
Lots of problems here:
1. Obviously I’m a believer in sticky wage/price models, and hence do not believe that monetary policy has an instantaneous effect on all prices (although it does instantly impact the prices of commodities traded in auction-style markets.) So Yates has mischaracterized my views.
2. When Woodford’s coauthor Gauti Eggertsson tried to apply their approach to the Great Depression, he read a great deal of my empirical work, and seemed to like it. Eggertsson’s 2008 AER paper on monetary policy expectations in 1933 cites three of my empirical studies. That doesn’t mean he agrees with all my views, but he didn’t seem to find them ridiculous. And in 1993 I published a paper arguing that temporary currency injections would not be inflationary, 5 years before Krugman published “It’s Baaack . . .”
3. I can’t speak for all market monetarists, but I’m extremely skeptical of the VAR modeling approach discussed by Yates. The early attempts at VAR models produced a “price puzzle,” which meant that tight money seemed to “cause” higher inflation. I remember thinking “Yeah, what do you expect when you use interest rates as an indicator of the stance of monetary policy.” Yes, some of these problems have been “fixed”, but I’m not impressed by the fact that 99% of the economics profession seemed to think monetary policy was “easy” during 2008-09.
In my view economists should forget about “building and simulating quantitative models” of the macroeconomy, which are then used for policy determination. Instead we need to encourage the government to create and subsidize trading in NGDP futures markets (more precisely prediction markets) and then use 12-month forward NGDP futures prices as the indicator of the stance of policy, and even better the intermediate target of policy. It’s a scandal that these markets have not been created and subsidized, and it’s a scandal that the famous macroeconomists out there have not loudly insisted that it needs to be done.
If and when we get out of the Stone Age and have highly liquid NGDP and RGDP futures markets, then it would be much easier to explain my views on leads and lags. In that world a change in NGDP futures prices, not a change in the fed funds rate, represents a change in monetary policy. To be more specific:
I predict that whenever the 12-month forward NGDP futures prices starts falling significantly, near term NGDP would fall at about the same time, or soon after. For instance, if we had had a NGDP futures market in 2008, then during the second half of the year you would have seen a sharp fall in 12 month forward NGDP futures. At roughly the same time or soon afterwards current NGDP would have been falling. In contrast, if the Fed had moved aggressively enough to prevent 12-month forward NGDP prices from falling, then near term NGDP in late 2008 would have been far more stable. I think that’s roughly consistent with Woodford’s view, although we may differ slightly on the lag between a change in 12-month forward NGDP expectations and a change in actual NGDP. (Nor would he necessarily accept my views on the potency of monetary policy in 2008.)
PS. The post also speculates on my views on fiscal policy. Just to be clear, I oppose attempts to force a balanced budget.
PPS. Yates’s blog is entitled “Longandvariable.” Not the specific post, the entire blog. He just needs to add “leads.”
PPPS. Yates refers to me as a “MaMoist.” I guess that’s better than being a Maoist, like that faction in the Greek party Syriza. You know, the party so many on the left now seem enamored with. The one that has a bunch of MPs that idolize history’s greatest mass murderer.
HT: Marcus Nunes
Tags:
27. February 2015 at 18:35
Sumner: “I thought it would be obvious to people that I know that classical theories of causation don’t really allow for cause to follow effect. Rather that it is actually expectations of future money growth that caused the near term NGDP growth surge.” – you’re a grown man and you believe in expectations fairies? Is this not metaphysics? If a central bank lowers rates 0.1%, and it causes a surge in NGDP, then your expecations theory is vindicated. If nothing happens, then you will argue rates should have been lowered more? At some point you must concede this is metaphysics? If not, you are only fooling yourself. Far better to conclude–as greater minds have, like F. Black–that money is neutral. No need for expecations fairies, simply agree, like Keynes did, in ‘animal spirits’. Spirits replace fairies (acknowledging that economics is nonlinear).
As for your NGDP predictions market, you still have not explained how ‘pegging’ avoids volatility in that market. One of your own followers has admitted the NGDP futures would be volatile on settlement date, even with pegging. Do you disagree? Please explain your views on how this futures market will work. I’m not the only one wondering.
As for sticky prices, see my earlier post by Robert Gordon on Okum’s model (Robert J. Gordon, Working Paper No. 847, NATIONAL BUREAU OF ECONOMIC RESEARCH, January 1982). What part of Gordon’s paper, that found very little sticky prices in 20th century Japan, UK and the USA (the US had the greatest sticky prices of the three) do you disagree with?
Thank you for reading this. On occasion–as do others–I post provocatively for affect, as do you (“poetic license”). But that should not detract that there are serious points in my posts. Seriously.
27. February 2015 at 19:02
Off-topic: Sumner: “PPPS. Yates refers to me as a “MaMoist.” I guess that’s better than being a Maoist, like that faction in the Greek party Syriza. You know, the party so many on the left now seem enamored with. The one that has a bunch of MPs that idolize history’s greatest mass murderer.”
This is way off-topic, but probably H i t l er and Stalin were greater murders than Mao. Further, any Maoist party in Greece is democratically elected, and insofar as I know is not advocating violent overthrow of the state (which is illegal under Greek law, it is said that’s why the leaders of the fascist Golden Dawn were arrested). Finally, as a Greek living in the Philippines (PH), it’s interesting how history is full of paradoxes, and sometimes an enemy ‘does good’. Take the PH Maoist party “New People’s Army” (NPA, still active in PH). As outlined in James Mann’s book “Rise of the Vulcans”, influential US DoD politician Richard Lee Armitage was influential in the US government on PH policy in the 80s, and, based on Armitage’s experience in Vietnam, he concluded the NPA would take over the PH unless dictator Ferdinand Marcos was forced to leave. So, according to this theory, the US pressured Marcos to abdicate not due to non-violent resistance by Filipino “People Power” (as taught here in the PH), though I’m sure that may have been a factor, but fear of a Maoist takeover by the NPA! So ‘thank’ the NPA for free elections in PH today. Funny how history works. BTW Armitage now lobbies on behalf of the Turkish government.
28. February 2015 at 02:28
What would have been Woodford’s argument for the relative weakness of MP in 2008?
28. February 2015 at 05:33
Either you or Tony Yates is badly misinterpreting the work of Woodford, this seems like a resolvable difference.
28. February 2015 at 05:39
Ray, I hate to tell you this, but money being “neutral” has absolutely no bearing on expectations fairies or the government’s ability to change NGDP. You are in so far over your head that you don’t know what’s up and what’s down.
Gordon believes in sticky prices. I have no views on his empirical estimates. My model does not depend on sticky prices.
You said:
“Thank you for reading this. On occasion-as do others-I post provocatively for affect”
Actually it’s quite obvious you do this out of ignorance, as when you confused inability to target NGDP with money neutrality.
Saturos, I don’t know. What was he saying about monetary policy in 2008? Someone should find out.
28. February 2015 at 06:02
Sumner: “Ray, I hate to tell you this, but money being “neutral” has absolutely no bearing on expectations fairies or the government’s ability to change NGDP. You are in so far over your head that you don’t know what’s up and what’s down.”
Wow, another Houdini trick up Sumner’s sleeve! The wonders never cease. I am genuinely intrigued now by NGDPLT. So, question to the professor or his more knowledgeable followers: suppose money is neutral, meaning the Fed has no ability to affect real variables like real GDP (from the quantity theory of money). How then does NGDPLT work? Reading further on this topic, it appears Sumner may endorse the caveat* below. Or is Sumner defining “money neutrality” different from everybody else in economics? Money neutrality is defined here: http://en.wikipedia.org/wiki/Neutrality_of_money
PS– I also endorse “superneutrality”.
*Caveat from Wikipedia on Neutrality of Money: “Even if money is neutral, so that the level of the money supply at any time has no influence on real magnitudes, money could still be non-superneutral: the growth rate of the money supply could affect real variables” – is this Sumner’s trick in the sleeve?
28. February 2015 at 06:09
[…] Scott Sumner replies on The Money Illusion here, including some priceless phrases about how his research found that there were in fact, no lags at […]
28. February 2015 at 06:49
“you still have not explained how ‘pegging’ avoids volatility in that market. One of your own followers has admitted the NGDP futures would be volatile on settlement date, even with pegging.”
1- I think calling me a follower is something both Scott and I would agree is funny.
2- Volatility is a measurement of deviation from a mean. The pegged price would be the mean and the fact that it is pegged would signify that there is no deviation. Therefore no volatility.
The instantaneous volatility would occur on settlement day as the price would move from the peg price to the actual NGDP figure. Scott would hope that if all went well, the last moment volatility would be 0 as NGDP outcome would = the pegged price. Nonetheless, any difference would be the volatility.
28. February 2015 at 07:40
Saturos, You said:
“Either you or Tony Yates is badly misinterpreting the work of Woodford, this seems like a resolvable difference.”
Here’s the problem. Yates claims that I believe prices are perfectly flexible. He also thinks I’m claiming that Woodford shares my views on that. As you know that’s nonsense, I believe wages and prices are sticky (as does Woodford), and respond to monetary shocks with a lag (as does Woodford). I think that resolves the confusion.
Ray, You said:
“So, question to the professor or his more knowledgeable followers: suppose money is neutral, meaning the Fed has no ability to affect real variables like real GDP (from the quantity theory of money). How then does NGDPLT work?”
I think you are the first commenter I’ve ever had that doesn’t understand the distinction between RGDP and NGDP.
Derivs, I’d add that the Fed would stop pegging each contract long before the settlement date. After that the price could become “volatile,” but of course that wouldn’t matter, as the price would no longer be used to control monetary policy.
28. February 2015 at 08:01
Sumner:
“I have great respect for Friedman, but when I did my research on monetary policy in the interwar years (which is the period where it is easiest to clearly identify monetary shocks) I wasn’t able to find significant lags.”
That is because you cannot observe them. Neither can you observe short and constant lags. Long and variable lags is a theoretically grounded concept that enables one to understand what one is observing. You cannot “find” a long and variable lag hidden behind the objective data of exchanges.
Long and variable lags is what explains the data we do see. If you reject long and variable lags, then you have not done so because of what the data says. You have done so theoretically, and then after being convinced of it theoretically, you then understand the data you observe as shaped by that theory.
It is also how you understand observations of money and output. You and I can agree completely on the facts of history, the correlations, the quantities, the timings, etc, but I could have, indeed I do have, a very different understanding of history because I utilize a different theory than you.
The same is true for long and variable lags. You look at today’s and yesterday’s data, and whereas I understand today’s and yesterday’s data to be in part the result of monetary actions by the central banks that stretch more than a year ago, indeed much more than that, you instead understand today’s data and yesterday’s data to be in part the result of monetary actions by central banks that stretch no more than say a day or week or whatever you define to NOT be long and variable.
I view your career as a monetarist to be one example of a long and variable lag of monetary policy. You went to the Chicago school, which was initially financed by politically connected banking families many years ago whose influence and power were directly related to “monetary policy”. The school acquired the reputation it has over many many years. It is not as if your career as a monetarist was only influenced by inflation that took place yesterday or last week.
Nobody can deny the above rationally, and yet there is no way for you to understand the above by searching through and delving into objective historthey data trying to find a long term consistent correlation. The above can only be understand ultimately by theory.
Your a priori theory has been chosen and now those are the tools you are neccessarily using to understand history. Of COURSE you “do not find any evidence of long and variable lags”. You aren’t using the theory to understand history! It would be impossible to do.
Long and variable lags are a fact of life that cannot be found IN the data via scientific correlations that presume constancy in relations (meaning, a repeating of history a million times with the only constancy being the same “monetary policy”, somehow necessarily results in you doing exactly what you are doing now all million times). Long and variable lags can only be a tool used to understand the data that treats relations and history in general as unique and non-constant.
It would be absolutely absurd to believe that the facts of today’s real economy, and the facts of today’s government, has nothing at all to do with “monetary policy” that has taken place between 1913 and whatever date juuust predates what constitutes a “short” period of time, say last week.
We are all, right now as we speak, living in a world that is in part the result of the entire history of monetary policy, indeed the entire history of the human civilization.
History is far too complex for anyone to discover a single correlative constancy in relations. Indeed, we humans who learn and act presuppose an absence of such constancy in ourselves. In order for us to even learn of any such constancy, our knowledge must be non-constant. And to the extent our knowledge influences our actions, our actions would also be non-constant. There can be no constancy assuming “evidence” that proves either short and constant lags or long and variable lags! It is a fool’s quest that enables the theorist to reaffirm what he already believes by pretending the data is speaking to him.
28. February 2015 at 09:06
‘The one that has a bunch of MPs that idolize history’s greatest mass murderer.’
It’s always been thus. I just happen to have been reading ‘Mao: The Real Story’, co-authored by a Russian, Alexander V. Pantsov, with access to Soviet archives.
After reviewing some of the nonsense that the American China Hands (John Stewart Service, John Carter Vincent, John Paton Davies) had reported about how the Chinese Communists at Yenan weren’t Communists at all, but merely democratic reformers who wanted to have good relations with the USA after the war, Pantsov writes;
‘It is amazing how easily Mao, Zhou, and the other CCP leaders were able to deceive these experienced American intelligence officers. There was nothing they didn’t promise them. In order to neutralize Washington, in the summer of 1944 Mao was even prepared to change the name of the Communist Party to New Democratic. …Ultimately…played the Americans for fools.’
28. February 2015 at 09:21
@Sumner – I understand full well RGDP and NGDP. The latter is merely the former plus some inflation; there’s nothing magical about NGDP and RGDP (money illusion and sticky prices notwithstanding). It is you who must convince the world of two things:
(1) why or if NGDPLT would work in a world where money is neutral (what model do you use?), and,
(2) why or if NGDPLT would work in a world where money is superneutral (id).
Ball is back in your court.
@Derivs – thanks, that was helpful. Perhaps you could decode for the rest of us Sumner’s statement to you? Jacques Derrida is not around to decode for us. What does Sumner mean: “Derivs, I’d add that the Fed would stop pegging each contract long before the settlement date. After that the price could become “volatile,” but of course that wouldn’t matter, as the price would no longer be used to control monetary policy.”
Why would it not matter? Aside from the trivial fact of course that the price used is months old, and presumably the price has drifted since then. Example: NGDP is predicted to be 100 six months from now from the futures market, and it is 80 today. Fed targets this, ‘pegs’ this price, and attempts to steer, by printing money, Sumner-style, to hit this target. As the months go by, the GNDP is measured to be, from first to sixth month, as 80, 85, 90, 95, 100, 105. Now on the sixth month the Fed has overshot its target of 100 by 5%. Now what? The futures will, on settlement day, adjust to account for the fact NGDP is 105 not 100 (if it’s a UK style option, which settles on the last day, these contracts will expire in the money if calls and out of the money if puts). There will be volatility on settlement date, as you say, but how does the Fed adjust the money supply to hit the target? By Open Market operations presumably, but, if money is superneutral, then the Fed is pushing on a string. So how does the Fed adjust? And what is the Fed adjusting to? The actual measured price (80, 85, 90…etc) or the futures price at what point? After pegging but before settlement/ expiration of the NGDP futures? And what does the Fed use for NGDP price after pegging? Backward looking (historic) survey data or again the forward looking prediction market? I have a dozen more questions but let’s start here please. I’m sure Sumner thinks his paper describes everything (in his mind’s eye) but I don’t think so. And for every poster bold enough (or foolish enough) to put his doubts in print, there’s a dozen more lurking in the background thinking the same thing. The blogsphere Silent Majority if you will.
28. February 2015 at 10:33
NGDP is not RDGP plus inflation.
RGDP is NGDP divided by the price level.
IF money were perfectly neutral always then changes in the quantity of money could target NGDP quite easily. While RGDP would be independent of the quantity of money, changes in the quantity of money would change the price level have an exactly proportional impact on NGDP.
The neutrality of money does not mean that changes in the quantity of money of have effect. It is rather that it only has no effect or real variables. The implication is that nominal variables all change in equal proportion to the change in the quantity of money.
Superneutrality doesn’t mean that changes in the growth rate of the quantity of money have no effect. it is rather that they don’t effect any real variables, but the growth rate of all the nominal variables change in proportion.
28. February 2015 at 10:55
@Ray Lopez: “I understand full well RGDP and NGDP.” Your comments show that this is not true.
Your logic seems to be something like this:
(1) NGDP = RGDP + inflation [fair enough]
(2) Monetary policy has no effect on RGDP [wrong, but we’ll let it go]
(3) Therefore monetary policy has no effect on NGDP.
Your logic is impressively stupid. Obviously, inflation is not constant, so a lack of control over RGDP (even if true), implies nothing about whether central banks have control of NGDP.
“Fed targets this … by printing money … [NGDP] is measured to be … 80, 85, 90, 95, 100, 105 … the Fed is pushing on a string” In your strange “pushing on a string” world, how did NGDP rise from 80 to 105 via printing money in the first place? It seems your very own scenario is contradictory, where sometimes you think monetary policy has some power, and then later you claim it has no effect. In the very same example!
“Fed has overshot its target of 100 by 5%. Now what?”
Simple question, easy answer. This is the beauty of level targeting. Presumably, there is some NGDP target for the next six months, say 120. That target remains unchanged. It had originally been expected that NGDP would be 100 at this point, and a growth of 20 more would be required over the next half year. Instead, there was a slight overshoot to 105. Which means that the growth target for the next six months is now only 15 more, in order to reach the same originally-planned level of 120 at the same originally planned time.
28. February 2015 at 11:58
“Jacques Derrida is not around to decode for us. What does Sumner mean: “Derivs, I’d add that the Fed would stop pegging each contract long before the settlement date. After that the price could become “volatile,” but of course that wouldn’t matter, as the price would no longer be used to control monetary policy.””
Ray,
1- Not familiar with Mr. Derrida, but I do find it funny that his last name is essentially the latin word for “to make fun of”, and his first name is Jacques, which simply by being French also means to “make fun of”.
2- Scott is just clarifying when the contract would become unpegged, when it is freely floating is when it would have volatility. The prediction model he has in the top right of the blog is freely floating and has volatility (it moves), if it were pegged it would just be the same number every day, therefore no volatility. Essentially the future would remain pegged until X days prior to expiration, after which it would act like a normal future.
Most of the other questions you asked, I can not answer as I have no idea as to the intended structure of the contract.
Also, what you refer to is a Euro option (there is no such thing as a UK option), and all normal options settle on the final day (actually most settle the day before Futures settle – in order to avoid physical delivery from unbalanced positions as option expiration can be an unpredictable mess that needs a day to sort out), the difference being that an American option you CAN exercise prior to the final day, and a Euro option you CAN NOT. Therefore an American option is always preferable to a euro option. Regardless, it is infrequent that either type of option is exercised early, particularly in Future markets.
hope that helps..
28. February 2015 at 12:19
@Derivs: “his first name is Jacques, which simply by being French also means to “make fun of””
+1
28. February 2015 at 12:33
Thanks for this post, it really clarifies the “long and variable lags” vs. “long and variable leads” concepts, which I believe is the biggest stumbling block for many people.
I think most people intuitively believe monetary policy has “long and variable lags”. This leads them to fear monetary policy more than they should. They may acknowledge that growth and inflation are low now, but they think by the time new money has its effect growth and inflation may have already recovered and we could overshoot the target.
Your plan gets around this by targeting forward looking prices like NGDP futures, taking advantage of “long and variable leads”. And by targeting levels, if you overshoot the target then the next year’s target will be the same level, so the growth rate will be lower in that year.
As far as I’m concerned, the more posts you have explaining “long and variable lags” vs. “long and variable leads” the better.
28. February 2015 at 12:46
Some evidence for the greater stickiness of wages to prices and the procyclicality of real wages: http://www.businesscycle.com/ecri-reports-indexes/report-summary-details/economic-cycle-research-ecri-its-high-time-to-raise-rates-or-not
28. February 2015 at 15:18
What is a VAR model?
28. February 2015 at 15:48
Excellent blogging. I think today a monetary policy is more effective than ever, because curiously prices are stickier than ever — there is competition in the American market place, resulting in sticky prices as opposed to the 1970s and prices easily rising….offsetting monetary stimulation and resulting in inflation.
28. February 2015 at 16:54
Mr. Sumner, I read a lot of your papers and posts on the NGDP futures market. I think the idea is great. Really, NGDP targeting would free us from thinking about good ideas that can’t be used in practice, like NAIRU and potential RGDP. Also, allowing unemplouyment, inflation and RGDP do move around would give policy makers and the entire world clues about what is really going on in the economy (productivity shocks, monetary shocks, structural problems, etc, etc). Now, 2 practical questions: 1. what exactly is your preferred NGDP futures market proposal? 2. How many quarters in advance would you “roll over” the “current contract” (meaning, when, in number of quarters, would the central bank stop pegging a specific contract and start a new one, one quarter further away in time ?). Thank you.
28. February 2015 at 18:43
@Bill Woolsey – “NGDP is not RDGP plus inflation” – thanks for the reply, but Don Geddis disagrees (“(1) NGDP = RGDP + inflation [fair enough]”). Both of you seem equally versed in economics; one of you is wrong, who is it? Clearly you are making a fetish of NGDP since you believe it affects RGDP, due to sticky prices / wages. This is a Keynesian / MM fallacy. As for the rest of your post, indeed it is correct, but what’s the point of changing nominal prices if there’s no real effect? If superneutrality is true, then a helicopter drop of money that doubles prices overnight will have no real effect, absent some minor menu costs which these bar code – scanner days is easily fixed by the push of a button? Then what is my problem with NGDPLT? The potential for hyperinflation when the Fed prints so much money that the population panics–animal spirits–and velocity changes abruptly. Think stock market black swan jumps, not steady state continuous math equilibrium equations.
@Don Geddis – “(1) NGDP = RGDP + inflation [fair enough]” – wrong, see Bill Woolsey’s reply upstream. Back at you Don. “(2) Monetary policy has no effect on RGDP [wrong, but we’ll let it go]” – no, don’t let it go, let it go, like Frozen’s hit song. Instead, admit it Don: you really think money illusion works? That’s really the magician’s trick behind Sumner’s illusion, isn’t it? Sucker born every minute, all those sophisticated products like TIPs designed to be money neutral are bogus, and people really will feel richer and wealth effect spend money (preserving the status quo, which any Austrian will tell you is wrong for structural reasons, but I digress) if a helicopter drop of money doubles prices overnight? Duda probably believes in this nonsense too, that’s why he’s wasting his money with Sumner. Admit it Don, you believe in fairies, expectations fairies that lead to status quo prosperity? “It seems your very own scenario is contradictory, where sometimes you think monetary policy has some power, and then later you claim it has no effect. In the very same example!” – I explained my reasoning in my reply to Woolsey above. I’ve also used in other posts the example of a eraser on a book: static coefficient of friction keeps the eraser on the book while you tilt the book…until it does not, when the dynamic coefficient of friction takes over. The Fed, by adopting NGDPLT, would be playing with the inflation dragon’s tail…a sad tale that ends badly. Inflation genie out of the bottle is another analogy you can use…hard to put back in (as the 1970s showed) once the population starts panicking and queuing up to buy gas (only a few percent difference between shortage and surplus exists in nearly any commodity, and indeed in any system, that’s why computer networks crash, as Duda can tell you, if the entire network does the same thing all at once).
@Derivs – thanks, that was a very clear explanation. UK option is what I call the Euro option. Jacques Derrida was a modern French philosopher known for his incomprehensible logic, which to his followers was brilliant, and to his detractors was rubbish. I guess it’s fitting his name means “joker joker”.
@Jose Romeu Robazzi – you are new here, if you are asking for details from Sumner on his pet project. The magic of Sumner’s NGDPLT is that it can be anything you wish it to be: it works whether the economy is money neutral, or not, whether prices / wages are strongly sticky, or not, whether there’s fiscal policy, or not, whether velocity changes or not, and I’m waiting to hear whether it works if money is superneutral or not. It’s like that magic show on Discovery channel where the magician walks across Lake Mead. I finally figured out how he does it–it’s his audience that is gullible, and believes in anything, that influences the TV viewer to also believe. Mass hysteria influencing minds (that should be a byline to this blog).
28. February 2015 at 18:52
Bill, Don, and Derivs, Thanks for trying to enlighten Ray. I fear it’s an impossible task.
Thanks Negation.
Thanks Rajat.
Doug, Vector Auto Regression. Trying to model the effect of X with past values of X and past values of other key variables.
Thanks Ben.
Jose, The first step might be to simply have the Fed peg the price of 12 month forward NGDP futures at the target. Otherwise, give the Fed discretion.
Eventually I’d like to develop point estimates of NGDP at the daily level, and use the pegging of 12 month forward NGDP futures to directly control the monetary base, via open market purchases of Treasury securities. Then I’d roll over once a day, always keeping the target being pegged 12 months ahead.
28. February 2015 at 19:29
I agree with negation, a very usefully clarifying post, thanks.
28. February 2015 at 21:18
Sumner:
“Bill, Don, and Derivs, Thanks for trying to enlighten Ray. I fear it’s an impossible task.”
Yeah but you guys can’t even agree amongst yourselves. Bill says NGDP is not RGDP plus inflation, Don says NGDP is RGDP plus inflation. How can two people who contradict each other “help” anyone on what they are contradicting themselves over?
I notice you support anyone who antagonizes you opponents, to the point of encouraging contradictions. Bad habit…
1. March 2015 at 02:54
Completely off-topic, but this is Tyler Cowen getting it completely wrong (how unusual for such a bright intellect)
http://marginalrevolution.com/marginalrevolution/2015/03/what-if-everyone-lived-like-mr-money-moustache.html
1. March 2015 at 04:57
I notice the hypermind widget shows predicted NGDP has fallen from 4.4% to 3.9%. Any reason we should care about this or just noise?
1. March 2015 at 05:21
@Michael Byrnes–Hypermind punters have little or no skin in the game, so the figure is largely noise.
Off-topic but deserves to be read; rebuts Sumner and Eichengreen who make no distinction between the Gold Standard pre-Fed Reserve and post-Fed.
The experience of free banking – ed. By Kevin Dowd (1992)- p. 21 – “Free banking’s success at maintaining peacetime convertibility (at least, in countries where governments enforced the redemption rights of noteholders and depositors) suggests that free banking was what enabled the gold standard to persist before the First World War. George Selgin (1988:40, 96) has argued that where commercial bank liabilities are convertible into a ‘base money’ (such as gold) whose supply is limited, free banks must quickly respond to changes in their reserves. Central banks, as the holders of base money for the whole banking system, do not lose reserves as quickly when they overissue, and so have more leeway in responding when losses come. The discipline that regular clearing imposed, enforcement of the laws of contract between banks and their noteholders and depositors, and free banks’ lack of power to make their notes legal tender (hence, the absence of a ‘time consistency’ problem), explain why free banking systems rarely abandoned specie convertibility during peacetime. … It may be that because much of the world had free banking during the time of the classical gold standard, the gold standard was indeed automatic then, contrary to the belief of many writers who have examined its workings (for a listing, see Bordo 1984). During and after the First World War, many countries switched to central banking or introduced measures to control the supply of commercial bank reserves, and the gold standard may have lost its automatic character as a result. (It may also have been important that, before the First World War, central banks acted more like the privately owned entities most of them were, rather than, the government appendages they all later became.)”
1. March 2015 at 05:42
Don Geddis can agree with your mistaken statements all he wants, but real GDP is found by dividing nominal GDP by the price level.
If you find the growth rates, this implies that the growth rate of real GDP is the growth rate of nominal GDP minus the inflation rate.
If you do a simple algebraic transformation, then the growth rate of nominal GDP equals the growth rate of real GDP plus the inflation rate.
You stated that nominal GDP is real GDP plus inflation. Your error was to combine levels and growth rates. Geddis either read RGDP and NGDP to mean growth rates rather than levels, or else made the same error.
Like most Market Monetarists, I advocate nominal GDP _level_ targeting, it think that kind of confusion is a major problem.
To claim that neutral money means that nominal GDP level targeting provides no benefit is not the same thing as saying that it is impossible. You said it was impossible if money is neutral. Absolutely false.
I think having nominal GDP too high is just as much of a problem as having it too low. The way to keep it from being too high is to be ready to reduce the quantity of money or raise the demand to hold money if and when it is rises too high.
Regardless, I think it is simply necessary to stop increasing the quantity of money to bring a halt to hyperinflation.
For money to be neutral, all prices must be perfectly and instantly flexible. That is a necessary condition, though perhaps not sufficient. My view is that money is approximately superneutral in the long run and the approximation is even close to neutral in the long run.
I don’t believe the money is neutral or superneutral in the short run mostly because of sticky prices and wages. Some prices are very flexible, and others are more sticky. Some wages are flexible, but most are very sticky. Wages are prices–the price of labor. Some prices are very flexible but many prices are very sticky.
Because of sticky prices and wages, shifts in spending on output result in changes in both real output and prices. This shows up directly in nominal GDP. The impact on prices will show up in measures of the price level. And When you divide nominal GDP by the price level, the shift in output should show up in shifts in real GDP.
Keeping nominal GDP growing on a stable growth path is desirable to avoid unnecessary and harmful disruptions in production and employment.
However, if prices and wages were all pefectly flexible, the shifts in spending on output would solely generate fluctuations in the price level. These would be undesirable as well causing unanticipated redistributions between creditors and debtors. This makes contracts for particular operations less desirable. With a decrease in the price level, it is possible that the result could be bankruptcy and reorganization, a very costly process.
Further, it is important to explore how spending on output can change. It occurs due to an imbalance between the quantity of money and the demand to hold it. Adjusting the quantity of money to the demand to hold will prevent these disruptions in the first place. However, because money is a nominal good and the nominal demand for money depends on its purchasing power, an elastic quantity of money requires a nominal anchor–the price level or nominal GDP are possible candidates.
Nominal GDP is better for two reasons. When productivity shocks shift the price level, nominal contracts are better for borrowers and lenders. Now, if you look at just one shift, a stable price level is better for either borrowers or lenders. But with productivity shocks going both ways nominal GDP rule is better over the long run. This is true even if all spot prices and wages are perfectly flexible.
With prices and wages being sticky, then a second reason why nominal GDP is better with productivity shocks is that a supply shock directly changes the price level. Reversing that change requires monetary disequilibrium to force a readjustment of all prices and wages to return the price level to target. This disrupts production and employment for no good purpose. Nominal GDP level targeting doesn’t try to fix the price level and so changes in productivity that cause changes in particular prices and so the price level.
I must admit I gave up reading your posts long ago. I rarely get beyond the first sentence. You seem to be remarkable in your ability and willingness to criticize market monetarism and Sumner in particular using a variety of incompatible perspectives. The impression this gives is you don’t understand any of them.
And that apparently includes some basics like what nominal GDP is.
By the way, many macroeconomists are just applied mathematicians. All they know is some models. In these models, there is typically one consumer good. Real output is the number uf units of those consumer goods produced. And it depends on the real interest rate. The price of that one consumer good is the price level, and it depends on the previous level of the price level plus the inflation rate. The inflation rate depends on expected inflation rate and the deviation of the number of units of the consumer good that were produced and the potential output of that consumer good.
OF course, the real interest rate depends on the central bank’s policy and the expected inflation rate. The policy is typically based upon deviation of inflation from target and the number of consumer goods produced from the potential number that could be produced.
Now, it is possible solve the model for the growth rate of the number of units of consumer goods produced. The model is set up to determine the inflation rate (the rate of change of price of the consumer good.)
The amount of money spent on the consumer good and the amount of money the producers of the consumer good earns plays no role in the model at all. The amount of money spent on the consumer good would be nominal GDP and the amount of money received for the consumer good would be nominal income. They are the same.
Now, if you are an economist and not a mathematician/physicist, then it is natural to believe that spending on output and the amount earned from selling output is important. And that these sorts of models must be constructed with that reality in the background. If you are not an economist, then not so much.
So, with inflation being the key variable to be discovered and the growth rate for the production of the consumer good readily calculable, nominal GDP might well be thought of as the growth rate of the production of the consumer good plus the inflation rate, which is the rate of change in the price of the consumer good.
Market Monetarists believe that the amount of money spent on output is important and the amount earned from selling resources is important. Nominal GDP is a measure of that in the real world.
In the real world, firms determine what they will produce and the prices they will set based upon how much money they can earn from selling their products. They determine the resources such as labor and and what they will pay according to how much money they expect to make from selling their product.
They don’t determine production according to the real interest rate and set prices based upon the deviation that about from potential output.
Nominal GDP is not actually calculated by taking an already determined inflation rate and adding it to an already determined growth rate of real output.
And while we do find spending on any one good by multiplying the price times the quantity, and so in a one good world, like the one most macroeconmists study, spending would be calculated by price times quantity, with many goods you have to add the price times the quantity of many goods. After summation, you have nominal GDP. Then you have to estimate a price level from all of those prices. It comes out as an index, a percent of prices in one period relative to that of an arbitrary base year. Then you divide and get an estimable of real output, which is in base year dollars. What the money value of output would have been if prices had been at the level that they were in the arbitrary base year used to describe the price level.
Nominal GDP is not found by multiplying real GDP times the price level. Real GDP is found by taking nominal GDP and dividing it by the price level.
1. March 2015 at 08:16
@bill woolsey – Prior to reading your prolix missive above, I thought you were a PhD in economics, maybe even a professor. Now I know you are just a crank. 😉
“Nominal GDP is not found by multiplying real GDP times the price level. Real GDP is found by taking nominal GDP and dividing it by the price level.” – uh, Bill, what Don and I are saying is: Nominal GDP (level) = Real GDP level + Inflation rate * Real GDP level = Real GDP level * (1 + inflation rate). Pretty simple really.
As for your sticky wages assumption being necessary for nominal GDP to become real GDP (along with money illusion I might add), you are correct. Sumner would never admit to the same thing, as he prefers being opaque. But the evidence, as evidenced by R. Gordon in his paper I’ve cited repeatedly, is that instead of sticky wages/prices in 20th c. USA, UK and JP (the three countries Gordon looked at) as being 10% responsible for changes in output, and output being affected by quantity 90% of the time, it turns out it is 90%, completely opposite to expectations (and Gordon, a sticky wages/prices advocate admits he was shocked). In short, prices change radically, not quantity, in the P*Q accounting. So much for ‘sticky wages’. Sticky prices are even more radical. Though it’s true a lot of merchants have sticky prices, as they don’t want to sell below cost, many merchants find out that it’s best to dump stuff at a loss rather than hold onto it, and clog up your inventory. Hence no sticky prices either.
PS–don’t read me, you might not learn something.
1. March 2015 at 11:43
“You stated that nominal GDP is real GDP plus inflation. Your error was to combine levels and growth rates. Geddis either read RGDP and NGDP to mean growth rates rather than levels, or else made the same error.”
Correct – I also assumed Ray was talking about growth rates. The correct equations are:
RGDP = NGDP/(Price Level)
RGDP Growth = NGDP Growth – Inflation Rate (Approximately)
So Bill Woolsey is correct – the first equation describes the level. Most of us, including me (and I assume Don) gave Ray the benefit of the doubt and took Ray’s comment to mean the second equation. Strange that Ray then attacks Don for giving him the benefit of the doubt.
The more important point is that RGDP is calculated from taking NGDP and adjusting it by prices, not the other way around. This is true whether we’re talking about levels or growth rates.
1. March 2015 at 12:05
@Ray Lopez: “…what Don and I are saying is…”
Please, please, I beg you … don’t put me in the same category as you on any subject. I don’t want to be tainted by the association. (I find it amusing that my post was sharply critical of your confused misunderstanding, but all you seemed to get out of it was, “we agree!”)
You know, when you wrote “I also endorse “superneutrality”“, all I could think was, “I bet if superneutrality were sentient, it would reject that endorsement.”
1. March 2015 at 12:39
Ray:
As best I can tell, your equations imply that the price level is equal to 1 + the inflation rate.
No it isn’t.
1 + the inflation rate is about 1.02, though for the CPI the most recent figure is .94.
The CPI is 2.33
2.33 does not equal .94
You said:
NGDP = RGDP + inflation x RGDP.
NGDP = (1 + inflation)x RGDP
But NGDP = Py and RGDP = y
Py = (1 + inflation) y
P = (1 + inflation)
If we calculate the inflation rate as (Pt – Pt-1)/Pt-1 then we can see that the relationship you describe only hold true if the price level last period was 1, that is, 100.
It is not true at any other time, like now, when the price level is 2.33.
Wage stickness doesn’t make nominal GDP into real GDP. Why would anything think that?
The question isn’t whether prices or wages fluctuate more than output. It is whether real output fluctuates at all in response to shifts in money supply or demand. It shouldn’t shift at all.
However, your explanation of Gordon’s result is incoherent.
Just as you do not understand that RGDP = NGDP/P, I doubt you know much about Gordon either. I sure don’t trust your report.
1. March 2015 at 18:10
@bill woolsey – “As best I can tell, your equations imply that the price level is equal to 1 + the inflation rate.” – no, you misread me. I am saying you can find nominal GDP from real GDP by chaining real GDP, in geometric fashion, times an inflation rate. This is simple arithmetic, not rocket science. As for stickiness and Gordon’s paper, don’t take my word for it, Google his 1980s paper, read it, and judge for yourself.
1. March 2015 at 18:19
@Don Geddis: we agree on the fact NGDP = RGDP + inflation. For both rates (simple addition) or for levels (geometric compounding for every period, in a chained fashion). Not for the rest of your post.
Tell me, why do you think NGDPLT is good, if it cannot (I assume this is your view?) affect real GDP? Stable prices?
1. March 2015 at 20:29
@Ray Lopez: “why do you think NGDPLT is good?”
Because nominal shocks can have real effects (due to sticky wages and debts). NGDPLT essentially eliminates nominal shocks, thus removing an important class of unnecessary economic damage.
2. March 2015 at 04:39
Mr. Sumner and Ray, i am a practitioner, a hedge fund manager. Although I am interested in the theoretical ideas underpinning models of the real world (economy), my interest in the NGDP futures is because it is worthwhile exploring it as a practical rule. I don’t really care if it will work, or if it should work. Let’s SEE if it works! There is no excuse for any central bank not to do it in a limited scale, just to test the concept. In the end, if central bankers don’t want to relinquish discretion in making policy, at least they will have another input for their reasoning. As a practitioner I am very skeptical of any body of “smart guys in a room” trying to measure quantities like NAIRU or potential GDP, or the velocity of money. It does not matter. Pegging NGDP is a simple enought, theoretically strong enough concept to prompt regulators to explore it beyond writing and criticizing papers. That is why I am interest on how it would work in practice. And I don’t know how much traction this is getting in the central developed economies, but anyone interested in making this happen should be trying to convince regulators everywhere, in smaller counties, maybe israel, or Brazil, where the central banks could be more willing to inovate, or where there are advanced derivatives exchanges that could be willing to create these contracts with or on behalf of their central banks.
2. March 2015 at 08:31
Ray, are you forgetting about RGDP growth??? Lol what a joke
2. March 2015 at 15:34
Michael, The market is probably too small at this point to achieve a high level of efficiency. Recall it’s not a true futures market, so arbitragers can do much to correct market inefficiencies. The iPredict market will be an actual futures market.
Also, in my view those price fluctuations are “small” in a macro sense, even if “large” in a market inefficiency sense.
Ray, Eichengreen and I make no distinction between the pre and post 1913 gold standard?
And I find it amusing that when it’s shown that you didn’t know what NGDP was, you change your statement and act like you knew it all along. Very classy, as usual.
2. March 2015 at 21:02
@Don Geddis – thanks for making that distinction about NGDP affecting RGDP explicit. I thought that was the case, but I don’t recall Sumner ever saying so, perhaps because he feels it is an obvious point. Nevertheless, as I feel sticky prices and wages are not that sticky, I feel the nexus between RGDP and NGDP is weak.
@Jose Romeu Robazzi – I agree, but the problem with adopting NGDPLT is that it’s not a central banking rule that is ‘bounded’. Meaning this: if a central bank (cb) says ‘we need 4% NGDP this quarter”, and attempts to do so via open market operations, and the economy refuses to rise, the cb will simply print more and more money buying more and more paper (government bonds, but even commercial paper) from banks. This is a dangerous experiment since at some point the economy could explode into hyperinflation. It’s an experiment not worth doing, akin to shutting down an engine during an approach to landing, and restarting it.
@Sumner–you are old, perhaps your memory is faulty, but late last year in the comments section you specifically stated there’s no difference between post-WWI / Fed 1913 Gold Standard and pre-this-period Gold Standard, since, inter alia, during both periods central banks sterilized gold inflows. I think Eichengreen also fails to make a distinction (from memory). And a careful reading of NGDP, RGDP shows I know what the differences are, I just wanted to see what Don G thinks the differences are, and how they affect each other.
Question for Sumner: do you adopt the clear explanation of the nexus between RGDP and NGDP provided by Don Geddis above? I repeat it: “@Ray Lopez: “why do you think NGDPLT is good?” Because nominal shocks can have real effects (due to sticky wages and debts). NGDPLT essentially eliminates nominal shocks, thus removing an important class of unnecessary economic damage. I bet you don’t because it will make explicit how you feel NGDPLT works, and being clear and explicit is an enemy to your idea, which, like a good magician’s trick, depends on misdirection, smoke and mirrors. You’re welcome.
3. March 2015 at 03:38
Ray, I think your reasoning is good, I haven’t thought about this long enough like you guys, but it seems to me that under NGDP targeting the volume of the stabilizing actions by the monetary authority could be huge. But by design they will be both expansionary or contractionary, therefore, any excess could be reverted faster than decision on a comittee of experts, with all their biases at work. And before it is used for actual policy making, it could be tested in a limited way, let’s say, up to 100 million contracts per period. Just to see if it’s “recommendations” are in the right direction.
Mr Summner, your idea of rolling over every day, to a 12 month period makes this contract look much more like a OTC swap, the person entering the contract will exchange the actual NGDP measured from D0 to D365 against the target of the monetary authority. This may be convenient from the point of view of contract design, but since NGDP comes out every 3 months, you would have to come up with a rule to adjust the payoff of these contracts, in the days between the NGDP announcements. I am probably repeating ideas already discussed in this forum, excuse me for that. Maybe to make things simple, we could come up with actual futures contracts, settling on the first day of each quarter, and using the third NGDP revision to calculate the payoff. Let’s say, the contract maturing April 1st, 2016, would settle with the actual NGDP growth of full year 2015, and would be traded between January 1st, 2015 and March 31st, 2015. If I had to design it, I would go along these lines …
3. March 2015 at 11:11
Ray, Where is the link to what you claim I said? But, yes, there was sterilization under both systems, so there were SOME similarities.
And yes, I agree with the italicized statement about NGDPLT.
Jose, I’ve suggested the US come out with daily NGDP estimates, by first creating monthly estimates, then interpolating.
3. March 2015 at 19:54
@ssumner – “Where is the link to what you claim I said?” — as MF can attest, you have a selective memory. Since you don’t use forum software like Simple Machines where I can easily search for comments (in fact you cannot search for comments), then it’s hard to find this comment, but it’s around the time we were ‘debating’ (if that’s the right word) the anti-gold but excellent 1982 paper by Harvard economist Richard N. Cooper. But, if you are now retracting this statement, and/or adopting what I said earlier, that there are differences (big I claim, I suppose you would say not so big) between pre-Fed Reserve and post-Fed gold standard policies in the USA, then I agree with you.
4. March 2015 at 07:15
Mr. Sumner, this is similar with inflation linked treasuries we have in Brazil, the statistics bureau partners with an association of market participants and collects expectations on the current month inflation. The treasury securities linked to it are adjusted daily based on the daily interpolation of that expectation number. It works fine, but sometimes when the expectation is a little off, there are last day adjustments that can be not so insignificant. The market is big though, participants got used to this mechanics. But as a money manager, i have to tell you, I would prefer the quarterly futures, it is simpler and easier to trade. Politically, it is less prone to criticism, at least in the earlier stages, when people are getting used to it …
4. March 2015 at 07:44
Ray, You said:
“But, if you are now retracting this statement, and/or adopting what I said earlier, that there are differences (big I claim, I suppose you would say not so big) between pre-Fed Reserve and post-Fed gold standard policies in the USA, then I agree with you.”
Very funny. “I see you are now agreeing with me. . . well not really, but I want to be able to claim you are agreeing with me, so I’ll pretend you are and then say I told you so.”
Jose, Thanks for that info.
5. March 2015 at 18:00
True enough, monetary policy responded immediately to an injection of legal reserves (before excess reserves were remunerated). But monetary lags have been mathematical constants for over 100 years. Any current injection is accumulated in the past flow.
Of course, NO economist yet understands this. So NO economist understands money and central banking period. I am the Alpha and the Omega.
6. March 2015 at 02:08
Don Geddis:
“Please, please, I beg you … don’t put me in the same category as you on any subject. I don’t want to be tainted by the association.”
You have already tainted yourself by believing you are tainted by association, rather than the content of your ideas. Anti-intellectual crank city, dead ahead.
12. June 2015 at 10:44
Ray,
http://www.themoneyillusion.com/?p=28144
19. November 2015 at 08:33
[…] It is funny how Milton Friedman is wheeled out repeatedly for this hoary old chestnut. He was really on to scare politicians and policymakers out of trying to fine-tune the economy. Friedman was probably just wrong on this point, as rational expectations theory has shown and the markets demonstrate day in day out by reacting immediately to unexpected changes in monetary policy, and to unexpected inaction in monetary policy to changes in the economic outlook. It is hard to see why it has gained status as a near religious truth. […]