Imagine we travel back in time to the second quarter of 2009. We stop by the Federal Reserve and reveal to Fed officials that the recession has now bottomed out. We inform them about a non-causal monetary policy first presented by Scott Sumner.
The Fed agrees with our assessment and decides to spend its political capital on the temporally-independent NGDP target and gets the backing of Congress and the Treasury Department (and the NSF). The great macroeconomic (and physics) experiment begins.
So what would happen next in this counterfactual history? How would the economy respond to these compound time-travelling events starting with traveling back to mid-2009? No one can answer these questions with certainty, but it is likely that at a minimum there would be temporarily undefined inflation (as inflation would fail to be a function of time).
The figures below lend support to this understanding. They come from a paper I wioll haven be currently willing worken where I willan on-run a counterfactual retro-forecast of nominal input/output starting in mid-2009. The retro-forecast willan on-be based on four different paths of NGDP willing returnen to its pre/post-crisis trend: a two-year path, a three-year path, a four-year path and a negative two-year path. The first figure shows the four NGDP return paths and the second figure shows the inflation forecasts associated with these paths:
...
Ok, < /snark >
This is a parody of David Beckworth's post on NGDP targeting (with some Douglas Adams thrown in). The question isn't really (and has never been) what would happen if NGDP targeting worked, but would NGDP targeting work?
If the Fed is unable to meet an inflation target, why can it meet an NGDP target? Why does the liquidity trap argument fail with NGDP targeting? Why does the IT index k suddenly rise with NGDP targeting?
In short, why is NGDP targeting as magical as time travel?
Jason,
ReplyDelete"They come from a paper I wioll haven be currently willing worken where I willan on-run a counterfactual retro-forecast of nominal input/output starting in mid-2009. The retro-forecast willan on-be based on four different paths of NGDP willing returnen to its pre/post-crisis trend"
Of all things I would expect to read on the econoblogosphere, a reference to "The Restaurant at the End of the Universe" is not one of them. Kudos.
"The question isn't really (and has never been) what would happen if NGDP targeting worked, but would NGDP targeting work?"
To me, it seems like Beckworth is proposing that the solution to the failure of the Fed to hit its target could be remedied by an even more ambitious target.
What it amounted to was "imagine we went back in time to 2009, but this time the central bank was omnipotent, so long as it targeted NGDP instead of inflation (since we know from empirical evidence in this present/future that QE + forward guidance failed to generate inflation)"
All this when I was starting to get slightly less annoyed than usual by market monetarists. Sumner even conceded the there are some people that believe QE was completely ineffective (http://www.themoneyillusion.com/?p=31355).
Cheers, John.
DeleteI still wonder why Beckworth didn't choose a zero year growth recovery path? It's far more impressive! Infinite inflation :)
Jason, I asked Beckworth about different time periods (both longer and shorter):
Delete"I only gave the different paths because it may prove politically easier to do the longer return paths. So they are there as an option of choices for policymakers."
So the Fed could do all of these paths because it is omnipotent, but we peasants may grab our pitchforks if it works its magic too quickly ... so it's better to let the peasants suffer in their ignorance a bit longer so they don't rise up?
Delete???
So ... let them eat cake?
Where's my pitchfork?
Jason and Tom,
Delete"it may prove politically easier to do the longer return paths."
Maybe that's why the Fed chose an infinite-year return path; it wasn't politically feasible to have it any shorter than that.
Of course, I still think Beckworth's logic is worst when he admits that QE failed but doesn't explain how an NGDPLT would be achieved.
I was thinking earlier that perhaps the reason these people simply assert that, by targeting NGDP, the Fed could have made NGDP on target is that they confuse targets with, for lack of a better phrase, concrete steppes.
Maybe this is why some people, e.g. Sumner, oppose Taylor Rules and then propose an NGDPLT without giving so much as a suggestion to what the Fed would actually do to achieve it besides simply 'doing whatever it takes.'
I suppose this is also where the concrete steppes argument comes in. People like you and I (I'm not sure if I can speak for both of you, but I presume this is the case) would like people to actually specify a policy rather than a target when talking about economic policy. We are concerned with achieving the target in question, non-concrete-steppes people simply assume that it will be achieved.
Maybe maximum obfuscation is the desired end?
DeleteIf you use a Taylor rule, you know interest rates are involved and look to actual moves regarding interest rates. But if you say no to Taylor rules, but rather use the most aggregated quantity available in the economy, people become so confused as to what is actually happening that they just expect the Fed to accomplish what it says it will accomplish and act accordingly ... instead of facing the confusing headache.
:)
Not having read Adams... I'm going to have to come back to your 4th paragraph sometime when I don't have a headache. )c:
ReplyDeleteOK, I'm over my headache, and it's no longer physically painful to read, but I still don't get it. I'll have to check out the Adam's link I guess. (c:
DeleteIn the "Restaurant at the end of the Universe" Adams tries to explain one of the side effects of time travel is that the idea of past tense, present tense and future tense aren't sufficient to cover all of the possibilities and therefore language has adapted to include other tenses. This is especially relevant for the "Restaurant at the End of the Universe" because it's actually located at the temporal end of the universe, and everything you'd talk about would actually have happened in the past.
DeleteThe Fed says they'll hit their inflation target in the next couple of years.
ReplyDeleteHow have they failed? Monetary policy doesn't work? Are you a gold bug?
The recovery was too slow for you? Well it wasn't for the Fed. If you want a quicker recovery advocate NGDP path targeting.
Hi Peter,
Delete"The Fed says they'll hit their inflation target in the next couple of years."
That's what they said a couple of years ago:
http://informationtransfereconomics.blogspot.com/2015/10/core-pce-inflation-update.html
Also at that link, I continue to be right.
"Are you a gold bug?"
As the title of this blog and my paper show, I am an information theory crackpot, not a gold bug.
"The recovery was too slow for you?"
I don't think there could have been a quicker recovery:
http://informationtransfereconomics.blogspot.com/2015/10/the-paradigm-dependence-of-productivity.html
"I am an information theory crackpot, not a gold bug." Hahaha... what does an information theory crackpot wear on his head? I assume it's not aluminum foil. Whatever it is, get "information theory crackpot" printed on it, and I bet you can sell those on your site. Maybe only a dozen, but still...
DeleteSome background reading.
DeleteThe Market Monetarists are ahead of you on the crackpot fashion front.
Delete... though you have plenty of cool stuff that'd look good on a T-shirt. E.g. that checkerboard graphic in the upper right.
DeleteTom,
DeleteSpeaking of t-shirts, has there been any progress on the concrete steppes shirt design?
John, thanks for reminding me: no, there has not... I wouldn't expect any till the turn of the year. But keep after me... that would be funny.
DeleteI was just mulling over an alternate design with 1s and 0s pouring out of a cracked pot! Lol.
The Fed says they'll hit their inflation target in the next couple of years.
ReplyDeleteHow have they failed? Monetary policy doesn't work? Are you a gold bug?
The recovery was too slow for you? Well it wasn't for the Fed. If you want a quicker recovery advocate NGDP path targeting.
> In short, why is NGDP targeting as magical as time travel?
ReplyDeleteJason, it's not magical. It's about expectations. Increased expectations of future spending increase present spending at the margin. So, if you manage expectations of future spending properly, then you can drive present spending to the socially desirable level. Add a prediction market to help with the "properly" part, and that's really all there is to it.
Obviously this approach does not work if there's a 2% inflation boat anchor attached to the economy. The ineffectiveness of QE does not prove monetary impotence at the ZLB. It's the 2% boat anchor that makes monetary base expansion ineffective at the ZLB.
I know you're more than smart enough to understand all of this. Your rejection of this simple, beautiful idea is based on something I can't see, and I wish I could.
-Ken
Ken, forget about NGDPLT for a moment and assume the Fed had decided to target one of the counterfactual PCE core trajectories in Beckworth's plot (the lower plot Jason reproduced above). Could they have hit that target trajectory, and if they could, then would it have had the same effect as NGDPLT?
DeleteExpectations for inflation shot up after QE was announced in August 2010. Did it have any effect? Expectations are important but they are more complex than MM and, in general, economists imagine. Unless we have serious modeling of expectation formation, I propose that economists stop talking about it.
DeleteThis comment has been removed by the author.
Delete"Increased expectations of future spending increase present spending at the margin."
DeleteAsserting this does not make it true.
Per Srini above, it empirically seems to have a temporary effect only.
The reason Beckworth shows 2,3 and 4 year trajectories is that he has no idea what trajectory would actually happen. If you had some kind of estimates of how the economy responds to expectations (per Srini, serious modeling of expectation formation), then you should know whether the 2,3 or 4 year path. The fact that Beckworth doesn't know is the problem. How do we know it isn't 19 years? Or infinity for that matter?
I say we adopt IT index targeting. Expectations of a high IT index cause people to expect really high inflation and NGDP growth. If the Fed had adopted IT index targeting in 2008, we wouldn't have had the great recession or any unemployment and inflation would be right on the target implied by the IT index. The IT index controls the entire economy and ITILT (IT index level targeting) is optimal. The Fed let the IT index fall too far for monetary policy to be effective.
Do you believe me? Should you believe me and my new information theoretic monetarism (ITM)?
Heck no!
At least not until I back it up with sensible assumptions, empirical data and quantitative predictions.
Which is the purpose of this blog, in a sense.
But really, I don't know what optimal policy is -- I'm just trying to figure out how the macroeconomy works in a coherent framework.
Actually, that last bit is probably my primary goal. I started reading economics blogs back around 2005 and it seemed to be an incoherent mess -- it still is! The IT model is my way of trying to organize everything I've read and heard into a coherent whole.
Jason, you're on a roll: you just invented a new kind of crackpot: the information theoretic monetarist!
DeleteHi, Ken. :)
DeleteBy assumption, the Fed is incapable of meeting an NGDP target of, say, 4% growth per annum at the present time without a general expectation that that target will be met. Given that, how can the Fed create that expectation?
Jason:
Delete> Asserting this does not make it true.
Yes of course, but please, what seems most likely? If you thought there would be 1000% inflation next year, wouldn't you spend more now than you would if you thought there would be 1.5% inflation next year? Well, of course you would, and so would I and everyone else. Now, just scale that back a bit, and I think it becomes obvious that higher NGDP growth expectations lead to higher NGDP.
What's the alternative? That if I expect 1000% inflation, I'll put off buying that new car, because I'd rather spend 250,000 2016 dollars on a car than 25,000 2015 dollars?
> Per Srini above, it empirically seems to have a temporary effect only.
No, there has been no empirical test. The 2% inflation ceiling boat anchor dominates NGDP growth expectations. It should be obvious that as long as everyone is expecting the Fed to impose a 2% inflation ceiling, then NGDP growth is greatly limited (to real growth plus 2%), no matter how many dollars the Fed prints. While the Fed's messaging from 2008 on was mostly clear as mud, there was one point on which they were always clear: the 2% nominal anchor was sacrosanct.
========================
Bill, if the prediction markets indicate that the Fed will fall short of its NGDP target even if it buys every asset it's legally able to buy, then I would support automatic fiscal stabilizers --- the Fed should mail a tax rebate check to every US taxpayer. If giving away money doesn't drive NGDP, I don't know what will.
-Ken
Hi Ken,
DeleteYou ask:
" ... what seems most likely? If you thought there would be 1000% inflation next year, wouldn't you spend more now than you would if you thought there would be 1.5% inflation next year?"
Imagine two scenarios:
A. The central bank announces that they are going to print up 10 times as much currency and then prints up 10 times as much currency.
B. The central bank announces that they are going to print up 10 times as much currency but then doesn't print that much.
In A, I would think any model would show a rise in inflation (the IT model does except in the case of k = 1, and you get a continuum of possibilities from e.g. a 1:1 rise in M and P at k = 2 to 10:1 at k = 1.1). In B, I would think a model might show a tiny spike in prices for a short period, but then a quick reversion (this would be allowed by the IT model).
The question is: is it the announcement or the printing that raises inflation?
Sure, expectations can modify the initial impact over a short period. Let's say that period is T1. In the long run (T2) money is neutral.
If T1 ~ T2, then expectations dominate the non-neutral impacts of monetary policy.
If T1 << T2, then concrete steps dominate the non-netural impacts of monetary policy.
In the ITM (and in many Keynesian theories, and empirically), money is never completely neutral, so T2 >> t for any finite t. A fortiori T2 >> T1 and concrete steps dominate the non-neutral impacts of monetary policy.
Note that T1 needs to be on the order of 20 years in order to explain Japan and have money be neutral in the long run:
http://informationtransfereconomics.blogspot.com/2014/10/under-spell-of-expectations-fairy.html
[continued]
[continuation]
DeleteYou said:
"No, there has been no empirical test."
This blog begs to differ ...
1. Past inflation anchors inflation expectations. Inflation expectations are well-known to be almost entirely backward looking -- measurements are biased high because inflation is on a downward trajectory. For expectations to influence future outcomes, they'd have to be forward-looking.
http://informationtransfereconomics.blogspot.com/2014/07/better-than-tips.html
http://informationtransfereconomics.blogspot.com/2014/04/inflation-predictions-are-hard.html
2. "Expectations" impacts on exchage rates evaporate in the noise within weeks. Also, exchange rates are basically given by relative NGDP so this is also a case of 3) wrong model and 4) something that is well explained by my model.
http://informationtransfereconomics.blogspot.com/2015/12/more-of-draghis-open-mouth-operations.html
3. Expectations may in fact be wrong (i.e. associated with the wrong mental model), so changes due to expectations may make variables move in the wrong direction (increasing volatility). This can actually explain why excess volatility exists in some markets but not others.
http://informationtransfereconomics.blogspot.com/2014/11/is-market-monetarism-wrong-because.html
http://informationtransfereconomics.blogspot.com/2015/07/macro-finacial-economics-puzzles-3-out.html
4. And finally, there is a model (mine) that gets inflation and output right without expectations. Expectations might be associated with temporary deviations (or higher volatility) relative to the ideal-market model.
http://informationtransfereconomics.blogspot.com/2014/11/because-empirical-success.html
http://informationtransfereconomics.blogspot.com/2014/05/the-effect-of-expectations-in-economics.html
The main driving factor here is that you're going to have to give me something that works better than my model (which doesn't have expectations except as temporary effects) in order for me believe expectations have lasting impacts.
Jason, you write:
Delete"A. The central bank announces that they are going to print up 10 times as much currency and then prints up 10 times as much currency."
I guess this is where to "people's QE" (handing out paper money on street corners?) vs the QE we actually had comes into play. I don't necessarily see a difference if instead of electronically crediting commercial bank Fed deposits with reserves, they'd sent over a brinks truck with paper money (still reserves). I believe (but I'm not certain) that the Fed paid interest to the banks on both. Why would those two scenario be any different? (paper vs electronic Fed deposits), especially if both earned IOR for the banks?
I don't think that the demand by the public for cash ever went unmet? I didn't hear any stories about ATM machines running dry.
So unless those extra printed up dollars were given away for free on street corners (people's QE: not requiring the Fed to perform open market purchases)... for which no IOR is paid (until all those dollars are eventually deposited in commercial banks, if they ever are), then I don't see a difference between printing the paper money reserves ("vault cash") and electronically crediting the reserves (both of which would be performed after open market purchases).
Now it's true that vault cash (i.e. paper money & coin held by commercial banks) is counted as part of "currency in circulation" but *I think* it doesn't amount to a large percentage of "currency in circulation."
DeleteI went through this one time with David Andolfatto, one of Andolfatto's colleagues at the St. Louis Fed, and Mark Sadowski. It's more confusing than it should be. As I recall, it took more than one iteration to converge to the right answers.
I think when you write M0 = MB - reserves, that's close enough to MB - currency in circulation so as not to matter. But I don't think they're precisely the same.
One other possibility is that the Fed calls the BEP (Bureau of Engraving and Printing) and orders the paper money to be printed up (the Fed only pays for the cost of physically creating the paper money... unlike coins which remain a liability of the Treasury!... Lol), but the don't hand them out on street corners for free and they don't do any OMPs to fill bank vaults with them... they just leave them stored at the Fed until such time as ATMs start running dry and commercial banks cash in electronic Fed deposits in exchange for them to recharge their ATMs. This also doesn't seem like it would accomplish anything. Who cares if all that paper money lies in storage at the Fed?
... and just for one extra little fun fact, there remains about $200k of paper money out there "in circulation" (or lost in the wash), from the early 1970s, which is a direct liability of the US Treasury (like coins). The bills are called "US Notes."
DeleteI was really into all this at one time! One thing I learned is that the Wikipedia article on "Money Supply" is not 100% correct. I tried to get it changed, but to no avail. For example, it implies there's an official designation of M0 in the US, but that's not true. That's one of the things I eventually was able to learn from the guys at the St. Louis Fed.
... last fun fact: (it's starting to come back to me now): The status of currency in commercial bank vaults is not straightforward: some of it is counted as reserves, but not all of it. The part that's counted as reserves though is eligible to have IOR paid on it, I think. This was a fun "research effort" that Sadowski again "helped me"
Deletewith (really, he did almost everything)... it was very confusing! If I recall correctly, Sadowski learned something in the process as well... why some government accounting stats always appeared to him to be slightly off.
Anyway, most of that can be counted as minutia ... well down in the noise!
Jason, thanks (as always) for the detailed reply.
DeleteHowever, I don't think your reply was responsive to me. I was talking about how NGDP expectations could drive NGDP, and you switched to monetary base expectations could drive inflation. So I don't feel like you responded to my argument that higher NGDP growth expectations will cause higher NGDP growth, which I believe they would, for the simple reason that the a change in expectation, that my money will be worth less tomorrow than I previously thought, makes me more likely to spend it today.
I can see you might concede that higher NGDP expectations would lead to higher actual NGDP, but argue that the Fed is incapable of creating higher NGDP expectations. I think that's plainly untrue under my proposed policy of automatic fiscal stabilizers. If you know the Fed is committed to hitting its target, and if the market NGDP expectation predicts failure, then the Fed is committed to printing money and mailing it to people no strings attached, ... how could that fail to produce higher NGDP? Thus, the Fed's commitment becomes credible and anchors the market's NGDP growth expectation.
Of course, it seems like you don't believe expectations exist at all (except as "temporary effects"). Given that, I guess it's unsurprising that NGDPLT doesn't make sense to you, and probably neither of my arguments above make sense either.
I recognize that neither of my arguments exists in a coherent formal framework. I'm trying to figure out what is the state of the art in modeling that supports NGDPLT. I'll bug you some more if I find anything credible. But, I bet whatever models those are (a) incorporate expectations by assumption ("assume a can opener"), and (b) lack the amazing combination of elegance and correspondence to real-world data that the ITM seems to have achieved. So, I doubt you'll find them convincing. Still, I'll point you at them if I find anything interesting.
Thanks again,
-Ken
Error correction: I wrote above:
Delete"I think when you write M0 = MB - reserves, that's close enough to MB - currency in circulation so as not to matter."
I should have written
"I think when you write M0 = MB - reserves, that's close enough to currency in circulation so as not to matter."
Hi Ken,
DeleteI'm always interested in understanding other models, so feel free to send them my way.
I refrained from directly addressing NGDP expectations because I think the inflation version gets at the heart of the argument while remaining somewhat model independent.
Foregoing that ... in the IT model, inflation targets and NGDP targets are actually two sides of the same coin (targeting the trend of one targets the trend of the other) so there is no difference between the two (you'd print money to raise NGDP and P). Heuristically
N ~ M0ᵏ
P ~ M0ᵏ⁻¹
http://informationtransfereconomics.blogspot.com/2014/11/expectations-rational-or-otherwise-and.html
It is important also to note in the IT model there effectively exists a maximum inflation rate π* (that corresponds to a maximum NGDP growth rate n*). If you increase M0 too much, the IT index falls until k ~ 1 and dP/dM0 ~ 0.
If your inflation target is below this "speed limit" πᵀ < π* (or growth target nᵀ < n*), then there is actually no empirical way to tell the difference between a "concrete steppes" model the central bank setting expectations.
This is why Canada looks like it has successfully achieved its 2% inflation target since the 1990s. It especially looks like it is due to expectations since the growth rate of the monetary base (or whatever) μ > π.
If you start from the point where π ~ μ (a quantity theory of money), but then later μ > π, it sure looks like something besides μ is driving π. If the central bank is trying to target πᵀ without a μ target (e.g. a "Friedman μ-percent rule") using whatever instrument, expectations look like a plausible explanation.
What is really happening (in the IT model) is that
π ~ (k - 1) μ
with k falling from about 2 (where π ~ μ) to about 1. This coincides with π* is falling as well.
Currently Canada has πᵀ ~ π* and (implicitly) in the US πᵀ > π* (and EU and Japan). That means undershooting the inflation target. It also means undershooting an NGDP growth target (i.e. "new normal" growth since 2008).
At this point you can tell the difference between an expectations-based model and the IT model ... at least if you don't start saying the observed πᴼ = πᵀ (or nᴼ = nᵀ). That is completely unfalsifiable -- you've gone from an expectations model that explains inflation rates to one that explains any possible inflation rate.
I think a good way to imagine it is as climbing a hill. At the bottom of the hill, there are different paths (and hiking speeds) for your hike that have wildly different ascent rates (the max ascent rate a* is high). At the top of the hill, things are flatter and there aren't as many available ascent rates, and most of them are low (a* is low). The shape of this hill is determined by the shape of the state space as it grows in size. A large economy has fewer configurations with large a* relative to the number of configurations with small a* than a smaller economy. Therefore a* is more likely low for a large economy, but can be higher for a smaller economy. High a* leaves the central bank with more options for aᵀ; low a* leaves few options for aᵀ.
" ... remaining somewhat model independent."
DeleteThat's because translating central bank activities into NGDP growth involves a lot more model assumptions than translating activities into P growth.
Thanks again for the detailed reply. You continually impress me with the effort you spend to explain your model.
DeleteIt looks like in our economy, \pi^* is about 1.5 percent. This seems really surprising, because not that long ago, measured inflation was quite a bit higher. What changed between 1970 and 2015 that brought \pi^* down so dramatically? What kind of hill could we have reached the top of?
The other explanation for low \pi^O is that the connection between M and \pi is a lot more complicated than what the IT model models. There are lots of ways to sterilize M, for example, by paying interest on reserves, or by promising that if spending picks up, M will be brought right back down to whatever is needed to keep inflation under 2 percent. The Fed of course has done both of these. I realize the IT model does not make room for those explanations because in the IT model, economic actors bounce around the feasible state space randomly; nothing about promising to shrink the balloon later keeps gas molecules near the center. I'm not sure the people who control major business investment are that random. But, I think I know with your response, which is, okay, mister smarty-pants, what *is* your model of how powerful businesspeople behave? But you already know my answer. In my model, there is this magic thing called the "market" which can predict their behavior. An NGDP prediction market can tell me what I need to do to with M to convince those people, whatever their actual behavior is, to spend in aggregate at a socially desirable level N^T. If this prediction market tells me I can't drive sending to N^T by adjusting M, then I will send helicopter money until the prediction market tells me N^O will exceed N^T.
I still don't see how the policy above can fail to produce a better outcome than current monetary policy.
But I'm getting it. Even if I'm right, you are not interested. You are looking for a coherent framework, not for better monetary policy. If we can figure out better policy from a better framework, that would be nice, but better policy is a distant secondary goal.
Thanks,
-Ken
Ken: "What changed between 1970 and 2015 that brought \pi^* down so dramatically?"
DeleteJason: "The shape of this hill is determined by the shape of the state space as it grows in size."
I think that's your answer Ken: growth of the economy is what changed.
Hi Ken,
DeleteIn my model, there is this magic thing called the "market" which can predict their behavior.
In the IT model, if ever human behavior becomes predictable (in the sense of not being a random variable), the economy would suddenly collapse.
This should make a bit of intuitive sense. If it became a bad idea to hold money, then everyone (being rational) should try to sell money and it's price should go to zero. This non-ideal information transfer does happen sometimes: markets collapse and we get financial crises.
What really happens is some people decide to sell, some decide to buy and the price falls a finite amount (not to zero) through the "hot potato effect" aka thermalization aka the Walrasian auctioneer.
http://informationtransfereconomics.blogspot.com/2015/07/updated-graphics-for-entropic-hot.html
What changed between 1970 and 2015 that brought \pi^* down so dramatically?
Tom basically gets it right: the economy grew by 1800% and the base (minus reserves) grew by 2500% from 1970 to today.
I did say that it is an "effective" maximum -- without doing something far beyond the bounds of normal behavior, the Fed can't get a lot of inflation. I mention in comments below what would actually work ... NGDP goes up by 10% if M0 goes up by 40% by handing out 100 dollar bills on street corners. But M0 has only ever gone up by 10% in a year.
The other possibility is pegging interest rates:
http://informationtransfereconomics.blogspot.com/2015/10/pegged-interest-rates-hyperinflation.html
If the Fed had fixed the price of treasury bonds at 99.5 (or whatever makes 0.5% yield) instead of a ceiling at 0.5% we might have had the success of the UK in generating inflation.
... so when you say:
You are looking for a coherent framework, not for better monetary policy.
I do have some suggestions for what might be better monetary policy: money financed fiscal stimulus, true helicopter drops, people's QE or pegged interest rates. I think Sumner believes true helicopter drops would actually work, too.
None of these work throught the expectations channel -- they work through "concrete steppes".
I do think a coherent framework is a necessary condition to understanding what better monetary policy is, however. Without the coherent framework, all policy is just a stab in the dark.
"The other possibility is pegging interest rates:"
DeleteI almost mentioned the hyperinflation route. The other is more a roll of the die I guess: a monetary regime change: like between steps 2 and 3 on his road to cashlessness: i.e. going all electronic and paying positive or negative interest on that "cash." But neither hyperinflation nor rolling the die with a monetary regime change sound like guarantees of good outcomes!
BTW, more good expectations laden stuff here from Beckworth:
Deletehttp://macromarketmusings.blogspot.com/2015/12/how-consequential-was-fed-tightening-in.html#comment-form
Jason, have you seen this? They ran a tight money policy that purposefully slowed the recovery.
ReplyDeletehttps://www.minneapolisfed.org/news-and-events/presidents-speeches/monetary-policy-renormalization
That link doesn't work for me. Nor does it work if I chop off the last bit to the right of the final "/"
DeleteKocherlakota's first Taylor rule doesn't make any sense.
DeleteIn the footnote he says
2.5 = 2 + 2 + 0.5(π - 2) + 0.5 (y - y*)
with y - y* = -2.5, but this means π = 1.5 ... however Taylor's formula is:
i = π + r* + 0.5(π - π*) + 0.5 (y - y*)
which means π is both 1.5 and 2.0. If you make this consistent, then i = 2 (π = 1.5) or i = 2.75 (π = 2).
However, in the text he says π = 1 so really we should have:
i = 1 + 2 + 0.5(1 - 2) + 0.5 (-2.5) = 1.25
The Taylor rule as Taylor originally put it has current inflation, not expected inflation. Kocherlakota gets 2.5 when he should have gotten 1.25.
He also says the output gap is 2 in the text and 2.5 in the footnote.
Additionally, the output gap was over 5% in Q4 of 2008 and reached a peak of 8% in 2009. These imply nominal interest rates of 0% and -1.5%, respectively. A 2.5% (or 2%) output gap isn't a reasonable number for talking about the financial crisis.
Kocherlakota has made a bunch of math errors (check it yourself!) and made indefensible assumptions -- and that's just in his Point 1.
So why should I listen to him?
JS - You should go wide with that one!
DeleteBut, he's come not to praise Taylor, but to bury him: If a sitting member of the Fed board cannot use Taylor well, then it may very well due for the scrap heap (both the Fed and Taylor!).
I'm really not worried about Dr K's math chops. He's pretty good from what I've read. I pay attention because he very likely knows more about what happened at the Fed than you or I. (True, though, he could also be a member of the vast Cowen-and-Sumner right-wing anti-fiscal conspiracy. He mentioned income targeting! You are right to be wary.)
Kocherlakota gives a good reason why the Fed, with its dual mandate, should have four of its seven member board of governors drawn from representatives of labor. (Not his recommendation, OC. :( )
DeleteJason, the fed could announce that it will hand out $100 bills on every street corner until the U.S. NGDP is $20 trillion. What do you think would happen? Could they do this? Why or why not? Would these $100 bills be spent, or saved, or used to pay down debt?
ReplyDeleteBtw, physics-ish economics is hard enough to understand without Adams. Leave the humor to the humorists!
According to the information transfer model, that would require printing 539 billion dollars (about 1500 dollars each, increasing the amount of currency in circulation by 40%), which would cause NGDP to go up by 1.93 trillion dollars from 18 to 20 trillion ... and the price level to go up by 1%.
DeleteBut it would actually have to hand out the physical cash money; it can't just announce it.
"...it would actually have to hand out the physical cash money..."
DeleteWhat do you think of this?
JS - sounds like a win-win-win, then (and entirely possible).
DeleteNow that we have the modeled-out possibility in hand, let's work on the hows. Handing out cash could be a bit tricky, and people treat the possibility in a bit of a snarky way.
The nice thing is, that the Fed hands out millions in physical cash-money every second of every day.
The trick is getting the incentives right. Perhaps choosing to announcing a "$20t or bust" policy could very well signal that you should go to the bank and withdraw cash, yes?
Perhaps even an announced threat of negative rates on all retail bank deposits? This surely could give the physical cash currency creation process a bit of a lift. The Fed could choose to do this, yes? Without handing out paper on street corners?
Perhaps the Fed could even choose to loosen up its paranoia regarding currency withdrawals -- all those money laundering form quotas for the banking system, namely. Let people withdraw physical currency in the tens of thousands, eh?
Currency in your pocket: a small price to pay for effective monetary policy. Quite simple, really.
If there really is a liquidity trap as you say, none of this should work, so what is the risk?
I think politics would stop it from happening -- especially printing up that much cash. It's 10 times greater than anything the Treasury has ever done.
DeleteBut it could work. People's QE.
However, I should point out -- my version of the liquidity trap is not the same thing as Krugman's. It is not an on-off thing ... right now you have to increase currency by 40% to get the price level to go up by 1% (output only goes up by 10% and long term interest rates would fall to about 1%). In the 1960s, that would have caused the price level to go up by 30-40% (real output goes up by about the same amount -- nominal by 60-80% -- and interest rates to rise by 4000 basis points).
A 1% increase in base money leading to a 1% increase in the price level and a 2% increase in nominal output is basic monetarism.
A 40% increase in base money leading to a 1% increase in the price level and a 10% increase in output is approximately a liquidity trap.
Well, the "cash" is already there as reserves, just waiting to be withdrawn as currency. People demand cash, and banks disgorge. It would likely be more risky to deny access to currency - a massive run on the bank would result.
DeleteAgreed on the massive currency creation producing only slight changes in the price level. Not many people of any persuasion understand the nonlinearities!
The point is, however, that the Fed is, right now, living out the result of a policy mistake. They did not produce sufficient volumes of currency in the past (tight monetary policy), thereby forcing interest rates to zero, and producing the flat end of the nonlinear money-ngdp curve.
NGDP level targeting would have avoided such a dilemma. The fed ought to have printed that physical currency a long time ago. The fact that it isn't is simply propagating error.
Drastic printing to correct the liquidity trap nonlinearities is the solution, not a new and novel problem. There is really no alternative if you wish to return to NGDP trend and the stable belly of the money-rate-output curve.
Monetary policy always works. Liquidity traps are always central-bank created.
"NGDP level targeting would have avoided such a dilemma."
DeleteSo would ITILT. ;^)
"...the "cash" is already there as reserves..."
DeleteOf course reserves pay interest. In any case take a look at this.
Well, the "cash" is already there as reserves, just waiting to be withdrawn as currency.
DeleteCentral bank reserves and cash seem to have different effects on the economy -- the former are involved in short term interest rates (and not much else) and the latter are involved in long term interest rates and inflation.
http://informationtransfereconomics.blogspot.com/2014/11/quantitative-easing-cleanest-experiment.html
They did not produce sufficient volumes of currency in the past (tight monetary policy), thereby forcing interest rates to zero ...
In the ITM slow M0, MB growth actually leads to a rise in interest rates (and a deviation from the NGDP-M0 path).
http://informationtransfereconomics.blogspot.com/2014/09/the-emerging-story-of-great-recession.html
"I wioll haven be currently willing worken"
ReplyDeleteOh, the pleasures of Middle English! Taketh me back, it doth. :)