Clarifying a Problem with NGDP Targeting
[UPDATE below.]
I fear that even you, my loyal blog readers, may not fully appreciate just how insightful my earlier post on Scott Sumner was. So let me take the humor out of it and write plainly.
In this post on the Irish “miracle” (of apparently huge economic growth), Scott wrote:
I’ve argued that NGDP targeting is not always appropriate for small open economies, citing examples such as Australia and Kuwait. Actually, it’s probably much more appropriate for Australia than Ireland. The key is whether NGDP tracks total labor compensation fairly closely. Where it does, as in the US, then NGDP targeting is appropriate. Where it doesn’t, as in Kuwait, then you want to target total labor compensation, perhaps per capita.
Look at that part I put in bold. Scott added it as almost a throwaway line, but it underscores that there are all sorts of variants of his framework, and they could (potentially) lead to huge differences in outcomes.
For example, suppose Trump becomes president and then tells Scott he wants him to design the new Fed regime. After quickly deleting all of his anti-Trump blog posts, Scott comes up with a mechanism by which the Fed has zero discretion, and simply buys/sells futures contracts (in a subsidized market) to ensure that the market always expects total labor compensation to grow at 5% per annum, with level targeting. Originally Scott had thought about doing NGDP targeting, but since he was given free rein, he decided to play it safe and go directly for total labor compensation.
Unfortunately, in yet another completely unexpected move, Trump reads the work of Bryan Caplan and decides that the only way he can pay off the national debt in 8 years is to let in 30 millions new workers per year.
Contrary to the warnings of the nativists, it turns out Bryan Caplan is right: Although certain sectors get crushed, on average real wage rates are poised to go up for the average American worker, even if we restrict our attention to the original group (before Trump came in and opened up the floodgates).
However, nobody realizes–until it’s too late–that Bryan Caplan’s cool plan (when considered by itself), plus Scott Sumner’s cool plan (when considered by itself), lead to disaster when they’re combined–using the very same framework that Bryan and Scott employ.
In the first year, with the influx of 30 million new workers, total real labor compensation is poised to jump by (say) 15%. But the new monetary regime limits the total growth in nominal labor compensation to 5%. So that means on average, nominal wage rates need to fall by 10%. [UPDATE: These numbers could work, but it is probably misleading you by me picking 15%, 5%, and 10%. I think this tripped up David Beckworth in the comments below, and he understandably thought I was confused at Step 1. In a follow-up post I’ll pick a clearer numerical example.]
If prices and wage rates were perfectly flexible, this wouldn’t be a big deal. But Bryan Caplan thinks nominal wages are not flexible, and Scott Sumner’s entire case for NGDP targeting rests on the assumption of sticky prices (in particular, sticky nominal debts). For example, someone who just bought a house with a 30-year mortgage is going to be screwed when the new Fed regime, coupled with much more open borders, causes his nominal paycheck to drop 10%. His mortgage payments don’t drop, even though food prices fall 15%.
In conclusion, my point isn’t to warn, “We better hope we don’t get Open Borders plus NGDP targeting in the same year.” Rather, my point is that Scott Sumner’s case is actually much more nuanced than I think some of his biggest fans appreciate. Yes, Sumner himself can be quite nuanced, but I’m curious: For those of you who’ve been reading him for years, did you realize the policy would blow up with rapid population growth? Or that if the Irish economy had implemented it last year, its people would be in a depression right now (according to Scott’s world view)?
One more parting shot: If you’ve ever described Market Monetarism as a policy of “encouraging steady growth in total spending,” does it matter that NGDP is actually a subset of total spending? (GDP is spending on final goods and services, not on intermediate goods.)
If total labor compensation per capita was the target your concern would not be relevant , right ?
In any case I think what Sumner is looking for is a good proxy to measure AD, which he thinks needs to be held constant to keep the economy close to its optimal output levels. This clearly is not a one-size fits all countries deal, which is what his post on Ireland was really about. In an economy the size of the US, with fairly predictable productivity and population growth NGDP seems a reasonable proxy. Obviously for Ireland it would have been a disaster, and (I assume) this would have been spotted by Market Monetarists had they been asked to identify a suitable target for Ireland and they would have chosen something other than raw NGDP.
I am sure I have seen Sumner say that he favors an NGDP target for the US for pragmatic rather than purist reasons. I think purists would favor something like the wage futures targeting standard proposed by Earl Thompson back in the 80’s. While somewhat more complicated than NGDP targeting it would (I think) address the criticisms implicit in your post.
Bob,
The point of a NGDP target (or some variant of it) is to stabilize the nominal income (or nominal wage) growth and allow output prices to fluctuate inversely to supply shocks. In your scenario, there would be benign deflation of roughly 10% along side the stable stable nominal wage growth of 5%. Real wages growth of 15% would still emerge.
Your scenario is basically a large positive supply shock. Whether such a shock comes from a surge in labor or a surge in technology, the Fed would keep nominal wage growth stable–which NGDP is a rough approximation–but would be indifferent to the price level. This has ALWAYS been the implication of NGDP targeting. It allows benign deflation to emerge when there are rapid gains to the supply side of the economy.
So no, Caplan’s plan and Scott’s plan do not conflict with each other. They actually complement each other.
‘In your scenario, there would be benign deflation of roughly 10% along side the stable stable nominal wage growth of 5%’
How could this happen without some workers taking nominal wage cuts ? If the workforce has gone up by 10% and total labor compensation has gone up by the target of 5% then simple maths indicates that average nominal wages have to fall.
I agree that nominal prices would fall by more in this example, and other things equal real wages would still rise – but wouldn’t a fall in nominal wages still be an issue here ?
The average going down doesn’t necessarily mean any individual’s wage must go down, especially if there’s an influx of new workers. Especially if those new workers tend to work for below average wages.
Nominal Wage Rate = Real Wage Rate + Price Inflation
5%=15%-10%
Most people make 5% more a year and consumer prices drop 10%
That’s close to heaven.
The only problem with MM is that +5% is too high. The Hayekian 0%to +1% NGDP rule would be better and asset bubbles would be almost unheard of.
Although under +5% ngdp financial bubbles would still be common, they would not be the monstrocity like the housing bubble of the 2k nots under +6% to +8% ngdp.
Murphy, if the plan is for nominal wages per capita to grow at a particular rate, then wouldn’t an influx of 30 million new workers require a massive increase in total wage payments?
You are correct MF. Fed would have to increase OMP until per capita ninalienable wage increase’s by 5%
Nominal GDP = Nominal [profits + rents + wages]
Ngdp per capita would have to increase more than 5% to make sure that wages, a portion of ngdp, keeps up with the increase of labor.
*per capita nominal wages
Bob’s hypothetical has Scott designing a plan targeting total labor compensation, not total compensation per capita.
I think what Bob is trying to demonstrate is that if you are hemming and hawing about whether to base your target on a simple aggregate or on a per-capita value… it isn’t a trivial decision, in the case where unexpected shocks such as mass immigration can happen.
Mind you, why stop there? Maybe target median wages. Maybe only look at median wages after tax and offset by typical mortgage payments.
Maybe only target female wages in the service sector but not if she is over 50. I mean, it’s as good as anything else, right?
Andrew and Tel,
OK that makes sense.
The difference is trivial if population changes are trivial. It is not trivial if population changes are not trivial.
I think I used a misspelled email.
Nominal Wages Rate = Real Wage Rate + Price Inflation Rate
5% = 15%-10%
Most workers would be making 5% more in money wages and prices would be 10% cheaper. Very close to heaven.
By the way, make it per capita if you wish.
“Most workers” does not follow here. You can’t infer the wage distribution from the average wage.
In this case. why not?
Cantillon effect.
The Cantillon effect finto on different under NGDPLT, because the transmission mechanism is expectations.
It is a good puzzle.
The only problem I see with MM is that +5% ngdp is too high. The Hayekian 0% to +1% rule is much better. Under the Hayekian rule, asset bubbles would almost impossible.
But the MM”s +5% ngdp per capita or n-wage per capita asset bunless would be minor compared to the monstrous housing bubble of the 2k not’s under +6 to +8% ngdp.
If I understand Bob’s point correctly, here’s a more realistic scenario: Clinton becomes President, opens the borders. Her Fed chair is a Sumner reader, but a filthy casual: not understanding Sumner’s nuance her Fed chair puts in place a 5% NGDP level target. These two policies lead to disaster, under Sumner-Caplan framework, which is precisely why they would recommend a more nuanced policy.
So Bob’s point is that advocating for interventionism is irresponsible, because politicians are highly unlikely to implement the precise set of policies which happen to work together.
In Dr. Murphy’s scenario, Sumner is in charge. So Sumner would adapt to per capita, because he is Sumner, not a “Sumner casual reader.”
“suppose Trump becomes president and then tells Scott he wants him to design the new Fed regime. After quickly deleting all of his anti-Trump blog posts, Scott comes up with a mechanism by which the Fed has zero discretion, and simply buys/sells futures contracts (in a subsidized market) to ensure that the market always expects total labor compensation to grow at 5% per annum, with level targeting. Originally Scott had thought about doing NGDP targeting, but since he was given free rein, he decided to play it safe and go directly for total labor compensation.”
No, Trump asks him to design a policy, not to be fed chair. Designing the policy is a one time action, he doesn’t get to adapt it.
Sumner appears to believe he is not advocating intervention, he is proposing a general rule that “just works” like clockwork.
Please note, I’m not saying I agree with Sumner. In fact, I would argue that no possible clockwork rule exists, neither Sumner’s target nor any other. The marketplace is an information processing machine, in order for the Fed to operate within this machine it must have some information destroying capability. Thus the invention of “fedspeak”, or that famous dictum, “When it gets serious, you have to lie.”
But then, I do believe that the purpose of the Fed is intervention, unlike what Sumner has convinced himself.
The existence of the Fed is intervention.
Andrew, everybody knows that the existence of the fed is intervention.
Nothing new, just NGDPLT is less disruptive given the federal government already owns more then half the economy through money monopoly..
That’s unacceptable.
And in my experience, quite a few people don’t know that.
Sumner does not believe fed rules are not government intervention, just less intervention.
MMs are not anarchists. So Sumner would have to be an anarchist first to then convince himself that government should not intervene and that targeting is not intervention.
You just have different world view then Sumner has.
I’m going to blog a specific example later, after I do “day job” stuff, but very quickly:
Guys, the whole point of the last post and this one, is that Scott’s phrase “perhaps per capita” is critical. So David (Beckworth), I think you missed that. You are telling me what would happen if the number of workers were constant, and their productivity increased 15%. That’s not the same thing as the quantity of workers shooting up by (say) 13% while productivity goes up 2%.
With a huge surge in immigration (or a sudden explosion in birth rates, or a bunch of retired people / women suddenly entering the work force), it’s not enough for the Sumnerian Fed to target total labor compensation. That too will lead to disaster, using Sumner’s own model.
Instead, you need to target per capita labor compensation.
That’s what my two posts were trying to illustrate.
Actually per capita only covers your immigration or birth rates scenarios. To cover retired people, or women (wow, that’s a dated scenario, actually) scenarios, you’d need to target labor compensation/civilian labor force. Which also covers dis-incarceration or war demobilization scenarios.
Andrew_FL ha, great! You’re right, I’m getting tripped up too, this stuff gets tricky.
Dr. Murphy,
A. Sumner and Larry White have discussed the fact that ngdp per capita is the target. It’s just that right now that is not important because policy makers are far from implementing. And since we have you guys to remind us the per capita issue plus Andrews’ thoughtful view that we should use the size of the civilian labor.
Thanks for the reminder, but thus has been addressed before.
B. A large population increase, labor size increase, increases productivity because increases the division of labor. So is not the economic problem thathat you seem to picture.
“One more parting shot: If you’ve ever described Market Monetarism as a policy of “encouraging steady growth in total spending,” does it matter that NGDP is actually a subset of total spending? (GDP is spending on final goods and services, not on intermediate goods.)”
Yeah this is kind of my hobby horse, so I probably should comment on this, as well.
This seems to be another area of important nuance MM’s don’t like to talk about in front of the children. The appropriate expenditure measure to target is something they debate in their technical writings, though.
Sumner’s labor-centric economics actually takes him in the wrong direction, with an even narrower measure in total labor compensation per capita.
Personally I think about what most closely resembles Hayek’s “Effective Money Stream” from Prices and Production. Maybe something like total money payments for physical goods and services per number of people in the civilian labor force.
Both this post and the previous one were entertaining and interesting. I hope this kind of thing gets NGDPLT fans thinking.
But I wonder, Bob, how you even have the patience to get that far. For me, NGDPLT is just a giant homunculus fallacy:
1 – Targeting inflation doesn’t work.
2 – So let’s target nominal spending instead.
3 – But how do we know how much to boost the money supply to result in an X% increase, unless we have some measure of inflation?
This is just *one* formulation of the NGDPLT fallacy, though. There are plenty more. For example:
1 – Floating currencies are intended to automatically inflate and deflate as per market money demand, rather than commodity demand.
2 – Currency futures already exist.
3 – But after years of continued business cycles, it’s clear that this system hasn’t solved all business cycle problems, hence:
4 – NGDPLT fans propose a futures market for the business cycle to smooth out money demand.
5 – But wait, what were those currency futures supposed to be?
Any way I look at NGDPLT, it’s like Inception, a dream-within-a-dream, circular and homunculus-like. When I present these arguments to NGDPLT fans, the response is always the same: Step One, assume I just don’t get the basic architecture of it; Step Two, re-hash Sumner’s arguments against the Taylor Rule, as though that is what I am arguing for; Step Three, insist that it’s possible to create an NGDP futures market; Step Four, call me an idiot.
I really don’t get it anymore.
it is my understanding that NGDP targeting does not prevent “good deflation”. it does not prevent prices from falling in the presence of a positive supply shock. That is a feature, not a bug.
But the point of 5% as opposed to the Productivity Norm is to avoid good wage deflation. Good wage deflation is viewed as a bug.
I get it now, thanks.