Miles Kimball on Market Monetarism
Marcus Nunes sent me a piece by Miles Kimball on market monetarism, clearly aimed at readers who are non-specialists. I like the article, but I’ll focus on a few areas where I at least slightly disagree:
One constant in all of these textbooks is an equation as famous for economics as E=MC2 is for physics””an equation suitable for an economist’s vanity license plate: MV=PY.
As E=MC2 is the key to understanding nuclear weapons and nuclear power, the “equation of exchange” MV=PY is the key to understanding monetary policy.
This caught my attention. While certainly a defensible argument, I believe it does more to confuse than enlighten. E=MC2 is a theory positing a specific relationship between three well-defined variables. MV=PY is essentially a definition of one variable (V), in terms of three other well-defined variables (M, P, Y). Actually not all that well defined, but at least capable of being measured independently of the other variables in the equation. I fear the average reader might assume MV=PY is a theory, just like E=MC2.
In my view people make one of two mistakes with the equation of exchange. One mistake is to assume it provides some sort of mysterious support for the Quantity Theory of Money. It doesn’t. Those that don’t like the QTM often tend to have a visceral distaste toward the equation, calling it a “mere tautology.” They often prefer C+I+G+NX=Y, which they think has profound implications for how the world works. It doesn’t either.
If one is to explain monetarism to the average reader, I think the best approach is to walk them through the “hot potato effect.” It’s not easy for people to comprehend this idea, but it’s the only real hope one has of giving actual insight into the motivating idea behind monetarism.
But what if new technology makes the natural level of output go up faster, as the digital revolution did from at least 1995 to 2003? Then real GDP should be going up faster to keep inflation steady. And that means that nominal GDP should also be going up faster.
. . .
Even if I can bring my market monetarist friends around to adjusting the nominal GDP target for what is happening with technological progress, I differ from them in thinking that the tools currently at the Fed’s disposal plus clearly communicating a nominal GDP target are not enough to get the desired result.
I have two objections to these two paragraphs. Let’s start with “Even if I can bring . . .” If Miles Kimball wants to bring us over to his side (and I’m perfectly willing to be convinced, as I don’t view NGDPLT as the absolute optimal policy, but only a close approximation), he first needs to do two things:
1. Discuss the many, many arguments that market monetarist have made for why steady NGDP growth is superior to steady inflation. We know that many new Keynesian models suggest that a flexible inflation target is preferable to an NGDP growth rate target. We reject those models and have provided many reasons for doing so. A recent example is this Nick Rowe post, although there are numerous other arguments as well. I’m working on a paper entitled “Seven unconventional arguments for NGDP targeting” (in addition to the conventional arguments.)
2. Then explain to us why all these arguments are wrong. He hasn’t done that.
As far as the adequacy of the Fed’s tools, here’s how I look at it. What is the demand for base money when NGDP rises at 5% per year, and there is no IOR? It’s about 6% of GDP. What is the national debt held by the public? About 75% of GDP? That’s a lot of ammunition. But it’s even worse for Kimball, as the current monetary base is about 20% of GDP, despite NGDP growth of less than 3% per year since 2008, and even with IOR at a rate higher than T-bill yields! Recall that slower NGDP growth leads to lower nominal rates, and hence higher demand for base money. And of course IOR increases demand for excess reserves. So it’s very likely that with a higher NGDP growth rate, and level targeting, and no IOR, nominal interest rates would be considerably higher today, and base demand would be smaller. Australia is a good example.
This may be confusing, so let me state it another way. We have a base that’s about 20% of GDP. Miles is worried that the Fed might run out of ammo trying to hit a 5% NGDP target, whereas I’m claiming that if they do it the right way (no IOR, level targeting, etc) the base is more likely to fall to 6% of GDP, than rise above 75% of GDP. (And I didn’t even included Treasury-backed agency debt, which is Treasury debt in all but name only. The Fed can and does buy that debt as well.)
Even if I’m wrong, there’s several steps I’d take before throwing in the towel and endorsing Kimball’s radical currency proposal. I’d try a negative 2% yield on all excess reserves, including vault cash. But not cash held by the public. Then I’d raise the trend rate of NGDP growth to 6%. Are you worried that would create too much inflation? Then recall that far and away the biggest welfare cost of inflation is higher nominal returns on capital, and hence a higher real tax rate on capital. But the 6% GDP option assumes that nominal yields are stuck at zero! Why do you never see me mention 6%? Because I find it extremely unlikely that it would ever be necessary.
Overall I like Kimball’s article a lot. But I think he misjudged two key issues. Stable NGDP growth is preferable to stable inflation, and the Fed doesn’t need to resort to radical currency proposals—it’s not even close to being out of ammo.
PS. Marcus Nunes has a good critique of Kimball’s technology argument.
Tags:
24. April 2013 at 05:34
Miles Kimball had a really interesting way of putting his proposal on Twitter: i.e. the legalization of paper currency counterfeiting. Which is effectively the same.
https://twitter.com/AshokRao95/status/327035105723961344
24. April 2013 at 05:41
You might want to consider a title for your paper that doesn’t sound like a buzzfeed article. Or else someone might accuse your blog of being “One simple trick to fix the economy that most economists don’t want you to know” or “College professor finds cure for depression in 3 simple steps”
24. April 2013 at 06:07
I like that point you made about MV=PY really being a definition for V. The only problem is that I can never recall in any of my economics textbooks the equation V=(PY)/M, where the = in this case is actually a triple bar(http://en.wikipedia.org/wiki/Triple_bar) denoting a definition .
24. April 2013 at 06:12
I tried to get miles to not think “technology” and think “sector” because that’s really what this Internet thing does it eats sectors, almost one at a time:
https://twitter.com/morganwarstler/status/326388151209848832
Marcus Nunes response is great, it descries the Internet I remember.
The thing is under a 4.5% NGDPLT starting at say 1996, Mile’s doesn’t like that Greenspan doesn’t get to let things overheat in the name of a ‘productivity surge”
So let me explain what that felt like as an entrepreneur.
I used to go to tech conferences and everyone looked like New Jersey in shorts.
And suddenly by 1999, while I had already raised my money (early smart faster), tech conferences were packed with hottie marketing people who could even use a computer.
Meanwhile the cost of a strong C programmer in Los Angeles was heading toward $200K a year.
There were “me toos!” popping up everywhere, flooding the market with noise and when NO ONE had broadband, $1B got dumped into creating content in Hollywood for the internet.
It was a frigging NIGHTMARE.
And I blame Greenspan PRECISELY because he didn’t stick to a 4.5% (or 5%) LT, and cut all those bastards chasing me off at the knees.
I’m furious that my industry then had to sit sidelines for 3 years, until we TOOK BACK our natural place at top of food chain, while Greenspan hand fed housing.
Entrepreneurs don’t want to ride thru booms and busts like 1997-2003, and it offends me that Miles thinks the problem is that the Fed didn’t print even more money.
We’re about to see tech do the EXACT same thing to government – there will be DRASTIC public sector reduction through attrition over next ten years, then more than following ten.
Public sector labor participation / output is going to mirror the graphs you see with agriculture and manufacturing.
And none of that is an excuse to chew into savings with more inflation.
More than just proving he understands MM, Miles needs to explain his real concerns…
He sees public employees losing jobs, he doesn’t see happier citizens, 24/7 govt. services, a smart phone as a civil right, sold and demolished govt. buildings, secure entitlements, and American ingenuity shocking the globe into final submission to the free market.
Meanwhile, Miles just blocked me on twitter.
Because he doesn’t need to hear from the guys who think like me.
Epistemic Closure.
24. April 2013 at 06:20
Ashok,
Yes this was my suggestion to miles which is far more likely:
https://twitter.com/morganwarstler/status/326443533852614656
But he wanted to go for shock value.
Why oh why can’t we get better econ bloggers?
24. April 2013 at 06:25
Morgan, I am left wondering if the idea is so “radical” after all. Will people in the future think it our gold standard?
24. April 2013 at 07:30
Scott: “While certainly a defensible argument, I believe it does more to confuse than enlighten. E=MC2 is a theory positing a specific relationship between three well-defined variables. MV=PY is essentially a definition of one variable (V), in terms of three other well-defined variables (M, P, Y).”
Actually I think this is a pretty good comparison. E=MC2 describes relationship between relativistic mass and energy. For some cases, that may have different flavor compared to newtonian definition of “rest mass”. For instance photon has no rest mass, you physically cannot “weight” it. However you can “measure” its relativistic mass as a byproduct of the aforementioned famous equation. And this is important, because this mass is very real, it has implication even for Newtonian physics. For instance if you would “confine” intense rays of photons into a sphere, it would generate gravity field equivalent to an object with the rest mass equal to relativistic mass related to the energy of photones in the sphere.
PS: Even saying all that, I think that it is actually theoretically possible to directly measure V (unlike rest mass of photons). If you would put RFID chips or electronic tag into every coin, paper note or electronic cent – and then measure it in cash registers or during money transfers I imagine that you could come up with V “independently” of other measures.
24. April 2013 at 07:49
Legalizing counterfeiting is a joke like #mintthecoin.
The natural play would be, and Miles should admit it, reducing size of currency denominations and then outlawing cash transactions over $x. I can see being able to shop at grocery or restaurant with cash, paying every bill, and buying every major good online.
Ashok,
The other issue that Miles (and Sumner) and other econs don’t grok is that once we move to digital currency…
We get tighter money.
Once you are pure digital currency, with “account” and “pin code” there’s very little friction between switching from dollars to Bitcoin to whatever.
We’d see digital dollar policy become far more worried about storehouse of value.
Otherwise, people will switch.
Again the logic assumptions is pretty simple…
SWITCHING currencies is far easier in a purely digital space.
COSTS of switching go down, less overhead in exchange, less margins.
LAWS to outlaw switching are nearly futile, its technically impossible, like in the music industry, surviving becomes about meeting the demands of the consumers… either you give them exactly what they want (storehouse of value) or they break your laws.
Think about the markets for drugs, guns, gambling, hookers, and off the books work for hire in a cashless society – that alone makes an alt.currency market liquid.
If you want to be Keynesian in this future, you can, but it becomes FAR MORE about NEVER growing state in good times, than it is about state helping in bad times…
And that’s not why peeps are Keynesian.
24. April 2013 at 07:56
“And that’s not why peeps are Keynesian.”
Truer words were never spoken. Awesome comment.
24. April 2013 at 09:32
Prof Sumner said:
“Then recall that far and away the biggest welfare cost of inflation is higher nominal returns on capital, and hence a higher real tax rate on capital.”
Question: If the govt only taxed real, inflation-adjusted capital gains (or got rid of capital gains tax entirely), would it make sense to target inflation at 4-5% per year? Or to target NGDP growth at 7-8% per year?
It seems like there would be benefits from having higher trend inflation, because it would provide more cushion against the ZLB and shortfalls in AD.
24. April 2013 at 09:46
JV Dubois —
MV = PY vs E=MC^2
Mass and energy and the speed of light are well defined and observable. The implications of E=MC^2 are that mass is equivalent to energy vs. separate properties and that there is a universal speed limit. Both were revolutionary.
V is an abstract that cannot be observed. We define it in terms of M, Y and P. M, Y and P are not so well defined or easy to observe either (what definition of money? Mb, M1, M2, M3, MZM?, what measure of inflation?)
It isn’t until you take the derivtive of the MV = PY equation that things get interesting… then we say
%Δ M + %Δ V = %Δ P + %Δ Y
Then we make the tenuous assumption that dV is effectively 0, and changes in P and Y can be controlled solely by M.
Y = C+I+G+X-M becomes interesting when you say C = C’ + cY
Y = 1/(1-c)(C’ + I + G + X-M)
But it is less usefull because you can’t talk about % changes. People will say that C is 70% of the economy but dC and dI are variability, so changes in I are more likely associated with a recession than changes in C.
24. April 2013 at 09:52
Monetarists asking other monetarists about the core problems of NGDP targeting is like the Italian mafia asking the Irish mafia what is wrong with Italian mafia methods of enforcement. They’d only get superficial, tactical, strategy related quibbles and complaints, and little trenchant and penetrating analysis.
24. April 2013 at 10:05
Geoff,
I don’t know that I agree with the implication that such conversations have no value.
First, it’s not really reasonable to expect monetarists to spend all of their time debating the basic viability of a monetarist approach. I understand that you have issues with it, and a lot of your arguments ring true to me. But we all live in a subjective world, and subjectively they believe your arguments are wrong. So it makes sense to move on and develop their ideas further, amongst like-minded people.
Second, most of scientific research consists of people learning “nope, that’s not it.” Finding dead ends is productive in itself – it’s just not glamorous. Whereas you think that monetarism can be rejected based upon simple logical consideration, others might later be persuaded by what they see as supporting empirical evidence.
24. April 2013 at 10:06
But Geoff, please tell us more about how recessions are caused by large numbers of people simultaneously deciding to go on year-long vacations.
24. April 2013 at 10:06
In other words, I think criticism and debate is productive both from within and outside of any organization or belief system.
24. April 2013 at 10:29
JV, You said;
“Even saying all that, I think that it is actually theoretically possible to directly measure V (unlike rest mass of photons). If you would put RFID chips or electronic tag into every coin, paper note or electronic cent – and then measure it in cash registers or during money transfers I imagine that you could come up with V “independently” of other measures.”
This is a common misconception. V is not the average number of times that dollars are spent each year. V is defined as NGDP/M, nothing more, nothing less.
Jake, Maybe, but I doubt there’d be much benefit to NGDP growth above 5%, or 6% tops.
24. April 2013 at 14:49
IIRC, Irving Fisher defined V as the transactions velocity of money, i.e. MV = PT, so J. V. Dubois would be right as far as Fisher’s monetarism.
I’m not sure when the switch to focusing on V as the income velocity of money began, but I’m glad it did, since T is dubiously useful. (Although I do think that studies of money’s relation to NGDP + asset transactions are interesting.)
24. April 2013 at 23:58
I am pretty sure that V could be measured by tracking money. For instance imagine this – you just “know” or just assume that transactions on newly produced goods and services are X% of all money transactions. And you have perfect control over money – you can perfectly track them.
I can hear “Oh, but then it is just residuum of NGDP and M”. But then the same may be said for theory of relativity. If you know “m” and “c” you know “E”. If you know “E” and “c” then you automatically know “m”. It is just a trivial observation.
For instance if you want to measure energy of a subcategory of objects called “this rock” (the same as you want to measure subcategory of all transactions, that is transactions on newly produced goods and services). The way you do it is that you find out its mass first – you pretty much put the rock on weight (or measure how defined energy impulse changes its speed) and then calculate an Energy using the famous equation. There is no “energymeter” that will enable you measure total energy of a rock directly. Would it make sense to say “Hey, you cannot measure energy of a rock without measuring its mass. Therefore if we are talking about energy of this rock it is just some virtual variable, a residuum of rocks mass and energy”
Money velocity is not some residuum. It has real-world implication, like actual money changing ownership. There are other things like that in physics. For instance relationship between frequency and wavelength of a sinusoidal wave that propagates at constant speed. If you know one, you automatically know the other. But it would be silly to say that wavelength is just a residuum of frequency, because it is frequency we usually measure directly.
25. April 2013 at 02:46
But maybe I do not understand your point of what residuum is. Do lets simplify things:
1) Assume that you know what money is (paper currency, money on accounts). You measure money in some lowest unit, like (cent or fraction of cents). Some of these units cents are grouped into bulks, like in coins or paper notes
2) Assume that for each money unit you know how many times it changed ownership during last year as part of a transaction where some newly produced good or service was exchanged for it.
So now what is exactly your claim? Is it that you cannot calculate NGDP from this information? Because clearly you can. NGDP can be calculated by sum of individual multiplications of money units with frequency it was exchanging ownership during last year.
Or is it that you cannot possibly such an information? Because you evidently can. It is just matter of tracking two variables:
1) Ownership (and changes of ownership) of each individual money unit during last year. This is surely physically possible.
2) Tracking nature of good or service that was traded while exchanging money, specifically if it was subset of transactions where money was exchanged for newly created goods or services. Also this is for sure physically possible.
Now I know that there may be alternative ways, alternative measurements that will help you calculate V. One such method is measuring (and estimating) flows (in pieces, kilograms etc.) of real newly produced goods and services as they leave factories and service providers and then multiplying it with measured (and estimated) prices for these goods and services (plus checking warehouses for inventories and chacking them as they leave/enter country). Then you divide number with total stock of money.
But it still is physically different way of measuring things.
25. April 2013 at 04:15
I define V as 1/k, where k = md/y. That makes V defined as y/md. “y” is real income and md is the demand to hold real money balacnes.
The equation of exchange is an equilibrium condition derived from Ms = Md.
I am not sure who decided that V is defined as Y/M, but I am pretty sure they are called Keynesians and this is just one of their many bits of intellectional ammunition to provide window dressing to expanded public works employment.
It is sad that they apparently teach this at Chicago as well. I suppose it fits in with Friedman’s hyper empiricism. It is just an empty tautology until we find a measure of M (and Y) that generates an estimate of V with low errors.
The “equation of exchange” as tautology is the same thing as saying that Ms = Md is a tautology because all the money that exists is held by someone, and the demand for money is how much money people are actually holding.
And that is the same as the argument that quantity demanded is always equal to quantity supplied for each and every good because the amount people buy is equal to the amount people sell.
But if this sounds familiar, it is the same as the argument that saving equals investment is a tautology. You can look at it that way, but it is just bad economics.
As is the notion that the equation of exchange is a tautology.
25. April 2013 at 14:25
Tyler Joyner:
“I don’t know that I agree with the implication that such conversations have no value.”
Well, it depends on whether you consider the superficial quibbling as sufficiently valuable. For me, it is somewhat valued, but very little, comparatively speaking.
“First, it’s not really reasonable to expect monetarists to spend all of their time debating the basic viability of a monetarist approach. I understand that you have issues with it, and a lot of your arguments ring true to me. But we all live in a subjective world, and subjectively they believe your arguments are wrong. So it makes sense to move on and develop their ideas further, amongst like-minded people.”
But I think their arguments are wrong, and so it makes sense for me to move on with engaging those arguments. If retreating to ivory towers is the “sensible” route, then consider the arguments not making much sense.
“Second, most of scientific research consists of people learning “nope, that’s not it.” Finding dead ends is productive in itself – it’s just not glamorous. Whereas you think that monetarism can be rejected based upon simple logical consideration, others might later be persuaded by what they see as supporting empirical evidence.”
The problem with that is that historical data alone is insufficient to singling out one theory over another rival theory as superior or inferior. If two variables, A and B, tend to be positively correlated over long stretches of history, there is no way to convince someone who demands to be shown empirical evidence for the theory that A is not causing B, that A is in fact not causing B. For he would continually point to the correlation and believe “the evidence suggests that the theory, A causes B, is confirmed.”
If someone else advances the theory that B causes A, then history would also show “evidence” that he is right.
And, if someone advances the theory that B increases and decreases despite A, the data would also show “evidence” that he is right.
At some point, empiricism will be revealed as insufficient. I am just skipping what are needless steps.
Daniel:
“But Geoff, please tell us more about how recessions are caused by large numbers of people simultaneously deciding to go on year-long vacations.”
I would if I ever said that in the first place.
Tyler Joyner:
“In other words, I think criticism and debate is productive both from within and outside of any organization or belief system.”
Agreed.
26. April 2013 at 05:18
Geoff,
Are you trying intentionally to fail the Turing test ?
27. April 2013 at 06:18
W. Peden, T has no relationship with GDP at all. None. So it’s good we no longer use MV=PT.
JV, V cannot be calculated by tracking the average number of times money is spent, because V is not the average number of times money is spent.
Bill, I’m just using the definition in all the textbooks.
I don’t like the distinction between desired and actual money demand. For me, there is no difference after a week, so the gap is not macroeconomically important.
29. April 2013 at 03:20
Scott: “V cannot be calculated by tracking the average number of times money is spent, because V is not the average number of times money is spent.”
And I see that you keep saying that + that V is just “residuum” of NGDP. It is as if I would say “Wavelengths is not distance over which the wave shape repeats itself. It is just a “residuum” of frequency….
Except that it is. Wavelength IS the distance in some basic distance units (like meters) over which the shape of the wave repeats. And V really IS the calculated as sum of interactions where some basic money unit (like thousand of cent) is spent on newly produced goods and services during some time.
Hey, but maybe repeat it another 100 times that it isn’t so and then you actually convince yourself. But let us go through some basic logic. Logic is a thing how you construct arguments. In order to prove conclusion wrong you have to prove that some logical operation that reaches the conclusion is wrong – not how many times you say it is wrong.
So sum it up again:
1) It is physically possible to track how many times ownership of money is changed for newly produced goods or services.
2) It is physically possible to know the total stock of money
3) Therefore V REALLY IS “average”number of times some basic money unit is spent
29. April 2013 at 05:54
I know that this probably wont be read, but I just want to add one real example of what “residuum” really is. And it concerns special theory of relativity. There is this equation:
E = m/(1-v2/c2)-1
E is enrgy of a particle, m is its mass, v is its velocity and c is the speed of light. This equation basically says that there cannot be a particle traveling faster than speed of light.
Except the fact, that what if there was such a particle? That would mean that v2/c2 is greater than 1. That would mean that (1-v2/c2) is a negative number which means that sqare root of (1-v2/c2) is imaginary number. That means that if “E” is a real number, then has “m” to be an imaginary number. And this is hard for physicist to swallow.
This interpretation of equation gives birth to fantastic theoretical particle called “tachyon”. It is always moving faster then light and it gains energy as it slows down. Which (for various reasons concerning a phenomenon called Cerenkov radiation is not possible)
So long story short, “tachyons” do not exist. They are just a “residuum” of an otehrvise valid equation under some specific condition without any physical interpretation/meaning. You can define them mathematically and it is interesting to think about them, but it is all there is – just a mathematical curiosity.
But money velocity is not like this. It is not some residuum of some equation. It is real, we can even measure it. The “birth” of money velocity could have been a result of something else, but it does not mean it is an residuum. It is perfectly reasonable to imagine a world where NGDP was “discovered” after people played around with Velocity.
29. April 2013 at 06:28
JV, V would represent the average number of times a dollar is spent if and only if:
1. All dollars were spent on items in NGDP, and not at garage sales, etc.
2. All purchases involved dollars spent, and not other assets.
Neither is even close to being true. Hence V has nothing to do with the average number of times a dollar is spent.
Consider that there are different definitions of M.
30. April 2013 at 04:07
Scott: I am perfectly aware of that. That is why I always say that we need to track if money was exchanged for “newly produced goods and services” – that is NickRoweSpeak for things counted into the GDP. That excludes second hand sales – if they do not incorporate some service (such service sold by real estate agent selling old house)
The point is that measuring V is physically possible. For instance you can create incentives for for people reporting what they purchase and then ask about it every time you register money leaving their wallet/bank account. You can create incentives on the side of the seller and then pair these transactions. Or if you can have everybody wearing Google Glass and then maybe some technology will enable us sift thorugh data reports to automatically identify not only point of purchase but also if newly produced goods and services were purchased.
So it is theoretically and physically possible to measure frequency with which ownership of money is being exchanged for newly produced goods and services. Therefore “V” is real, it has “physical interpretation”. Maybe you cannot measure V given current technology or the costs of measuring it are too high. But it does not mean that V is just some irrelevant residuum.