Reply to Arnold Kling

Arnold Kling left this comment (responding to a quote of me):

MMs predict that NGDPLT will reduce the severity of the business cycle. Not easy to test, but certainly testable over a period of time.

In fact, this is what I believe is the core irrefutable proposition of market monetarism. Suppose that the Fed announces that it is changing to a level target for nominal GDP. And suppose that the business cycle continues just the same. There are two possible responses from market monetarists.

1. Say that this was a valid test.

2. Say, “aha! They said they were doing NGDPLT, but look. NGDP was all over the map. Obviously, they were not really doing NGDPLT. They decided to do something else.”

My snark about market monetarism being irrefutable reflects my uncharitable (and untestable) belief that we would see (2).

There are a number of ways to respond to this, which vary in terms of persuasiveness.  The last item on my list might seem the least persuasive, but in practice it is by far the most important.

1.  When the Fed decided to target inflation the behavior of inflation became radically different. They didn’t produce exactly 2% inflation, but we no longer got negative 10% or positive 13%.  They can target nominal aggregates pretty well, the problem now is that they are targeting the wrong nominal aggregate (inflation.)  They should be targeting NGDP.

2.  Actions speak louder than words.  Right now the BOJ is trying to target inflation at 2%.  They are falling a bit short and I have consistently predicted that they will continue to fall a bit short (excluding sales taxes).  But at the same time something has clearly changed.  Inflation is rising.  Under the old regime they used to tighten monetary policy when there was no inflation at all.  YOU JUST DON’T DO SOMETHING LIKE THAT WHEN YOU HAVE A 2% INFLATION TARGET.  I don’t think I’d have any trouble convincing anyone that the Fed was cheating if NGDP was far below target and the Fed suddenly tightened policy sharply using conventional techniques.

3.  The most important point missed by Arnold is that I mentioned NGDPLT, not NGDP growth rate targeting.  I would still mildly disagree with him if the policy being considered was NGDP growth rate targeting, but NGDPLT makes it orders of magnitude more difficult to cheat.  Japan provides an interesting illustration.  Since 1993, the Japanese GDP deflator has fallen about 1% per year.  If they were targeting inflation at 0%, then the miss each year would be pretty small.  In contrast the level of the deflator is now almost 20% lower, and hence would be nearly 20% below a 0% target path.  It would be obvious they weren’t doing what they claim, especially during years like 2000 and 2006, when they tightened policy despite the price level being far below earlier levels.

It’s really, really hard to cheat on level targeting.  That’s why central banks don’t do it, it gives them very little discretion and they know it.  And that’s exactly why I like it.  It’s really effective, for better or worse.  The gold standard was sort of like that, but with a target (gold prices) that was much less conducive to macro stability than NGDP.

And then off course there is always NGDP futures targeting, which completely answers Arnold’s objections.



34 Responses to “Reply to Arnold Kling”

  1. Gravatar of J Mann J Mann
    10. January 2014 at 09:54

    Scott, I’ve been thinking about Jim Manzi’s book Uncontrolled, which argues that public policy should include experiments much more often. If you’re interested, I’d love to read your thoughts on whether Manzi-ism could potentially inform this debate.

    If you were in charge of US monetary policy (or world policy), are there experiments you would like to do that would potentially falsify your beliefs in any of the areas of disagreement?

    In other words, where you see yourself as disagreeing with Arnold or DeLong or whomever, is there a policy experiment that would be helpful, and what would it show?

  2. Gravatar of BJ Terry BJ Terry
    10. January 2014 at 10:18

    There is a third possibility which he doesn’t include, which is that the Fed accurately targets NGDPLT, but we still see a strong business cycle in unemployment or RGDP. This would actually be the nightmare scenario for market monetarism.

  3. Gravatar of ssumner ssumner
    10. January 2014 at 10:29

    J Mann, Yes, NGDP futures markets. Do that and all sorts of experiments become possible.

    BJ, I agree.

  4. Gravatar of Arnold Kling Arnold Kling
    10. January 2014 at 10:35

    As I read it, each of your points tries to reinforce the belief that market monetarism is true. What I want you to do is indicate how market monetarism is falsifiable.

  5. Gravatar of ssumner ssumner
    10. January 2014 at 10:39

    Arnold, It seems to me I directly addressed your concern. I claim that if the Fed adopts NGDPLT it will either succeed or it will be patently obvious to everyone (including you) that it isn’t trying. If it succeeds (overwhelmingly likely) and RGDP is still unstable (quite possible) then we (MMs) lose.

    Where do I lose you?

  6. Gravatar of Becky Hargrove Becky Hargrove
    10. January 2014 at 10:55

    The way in which it is falsifiable is actually what you struggle with. MM is a natural vortex for the intersection of time and resources. However, finance can distort that vortex by insisting that the intersection doesn’t really matter.

  7. Gravatar of TravisV TravisV
    10. January 2014 at 11:01

    Scott Grannis:

    “Weak December jobs report likely a fluke”

  8. Gravatar of Mads Llindstrøm Mads Llindstrøm
    10. January 2014 at 12:35

    BJ writes: “which is that the Fed accurately targets NGDPLT, but we still see a strong business cycle in unemployment or RGDP.”. That seems sensible, except that according to MM inflation is a meaningless number (see “MMs say inflation is a meaningless number;”). Then inflation adjusted NGDP = RGDP should also be a meaningless number. You cannot judge a theory by a meaningless number.

    I do kind of agree with BJ though, as I don’t understand why MMs consider inflation a meaningless number.

  9. Gravatar of Mads Llindstrøm Mads Llindstrøm
    10. January 2014 at 12:43

    OK, I posted to quickly. Sorry about that. I read and I am wiser. Feel free to delete my posts.

  10. Gravatar of Negation of Ideology Negation of Ideology
    10. January 2014 at 12:45

    Arnold Kling –

    It seems to me the only way to test Scott’s proposal is to implement it. That means create an NGDP futures market and have the Fed offer to buy and sell the futures from the public in unlimited amounts at a fixed price. So if the Fed puts into place the NGDP futures proposal it would be considered a failure if:

    1) Actual NGDP routinely comes in dramatically different than expected NGDP. (Would this also disprove the EMH?) It doesn’t have to be exact of course, perhaps in actual NGDP is as stable as the Great Moderation we’d call it a success.

    2) As BJ Terry said, actual NGDP is very stable, but real GDP loses it’s association with NGDP and you have a real business cycle and wildly fluctuating inflation.

    3) Real GDP growth is unusually low for an extended period of time. This one may be tricky, because other factors are involved, but if real growth averages 1% and inflation averages 4% I think most people would consider market monetarism a failure.

    And probably many others. I don’t think any of my three failure scenarios are likely, but I think that would be the test.

  11. Gravatar of wufwugy wufwugy
    10. January 2014 at 13:54

    What I don’t understand is why NGDPLT is controversial in the first place. How does anybody think that volatile NGDP fluctuation is better than steady NGDP growth? The biggest no-brainer there ever was is that when NGDP drops below a certain level, the Fed picks it up, and when NGDP rises above a certain level, the Fed drops it. I am merely a layman, but it absolutely baffles me that economists didn’t all agree on growth level targeting decades ago. The other option is growth volatility. Who seriously thinks that’s a good idea? The road back from a recession is always said to be in NGDP growth, so why do so many economists have a hard time with preventative measures that keep NGDP from falling?

    Anti-NGDPLT is like driving your car till you run out of gas, admitting you need more gas, yet then not only refusing to get more gas but refusing to acknowledge that if you refilled your tank when it got below a certain level, you wouldn’t have run out of gas in the first place

  12. Gravatar of Craig Fratrik Craig Fratrik
    10. January 2014 at 14:23


    I totally believe in MM, but I think you are missing Arnold’s point. (Your response about inflation is persuasive but not defninitve. Here is how I see the argument.

    Scott wants (I’m guessing at the preferance order):

    1. Monetary policy implemented via NGDP futures market.
    2. Discretionary policy targeting NGDPLT along with subsidized NGDP futures markets.
    3. Discretionary policy targeting NGDPLT
    4. Discretionary policy targeting NGDP growth informed by NGDP futures markets.
    5. Discretionary policy targeting NGDP

    Arnold probably doesn’t want to discuss #1, and is more interested in discussing 2-5, but probably in reverse order.

    Let’s go with 3, and he asks how might things unfold:

    1. The Fed does it’s job perfectly and we have stable NGDPLT growth from here until eternity.

    2. The Fed says its trying to target NGDPLT, but it doesn’t hit it.

    2.a It basically hits it (MMs probably consider this most likely), and we are still in situation 1

    2.b It consistently undershoots or overshoots.

    2.c It is pretty erratic, and consistently misses it in both directions (worse than we’ve seen with inflation targeting).

    Under 1: MMs say “See, you’re welcome for creating trillions of dollars of wealth in the economy.”

    2.a: basically 1.

    2.b: Arnold asks, what would MMs say? Might they say, “The Fed isn’t trying hard enough.” It certainly is what they are saying to the Fed now about trying to stimulate the economy. (I understand that forward guidance is what MMs want the most form the fed right now, and that size of their balance sheet is not a measure of “trying,” even though many people think it is.)

    2.c: MM response? I’m not exactly sure, the biggest push would be for futures markets and maybe we could learn something from the consistent misses to make them less consistent.

    Anyway, the way I would phrase Arnold’s question is:

    Assume you get policies 3 or 5, as in the Fed says, this is our goal. How might your system fail? BJ Terry’s answer is a very good one. Arnold’s point is, how would you respond if the Fed tries to (or says it’s trying to) target your preferred variable, and the fail at the task. “Try harder” is not a very good response. Also of note to me, what are the ways you imagine it is _possible_ for the Fed to fail at the task?

  13. Gravatar of wufwugy wufwugy
    10. January 2014 at 14:34

    I don’t understand how the Fed can “try to target” yet miss it. I thought it was consensus that the Fed can essentially induce as much inflation or deflation as it wants

  14. Gravatar of Daniel Daniel
    10. January 2014 at 14:48

    Yeah, it’s hard to wrap your head around that.

    It speaks very poorly for the economics profession that so much intellectual energy is spent developing theories around the supposed inability to debase fiat money.

  15. Gravatar of ssumner ssumner
    10. January 2014 at 15:06

    Craig, You misunderstood my point. I am saying that if the Fed adopts NGDPLT, and actually appears to try (to Arnold’s satisfaction) they will succeed. If not, MM loses.

    Daniel, You nailed the key point. When MM started out we had to convince people that monetary policy was effective at the zero bound. We’ve won that argument, and are on to the next one–is NGDPLT best.

  16. Gravatar of Paul Zrimsek Paul Zrimsek
    10. January 2014 at 19:10

    After the past five years it’s hard to conceive of a Fed policy– short perhaps of returning to the gold standard at the pre-Civil War parity– that everyone would agree constitutes “not trying”. If you share Sumner’s disdain for “concrete steps” it’s tricky to come up with any criterion other than the final result by which to judge whether or not they’re trying. “Stabilize NGDP growth, or do not. There is no try.”

  17. Gravatar of Craig Fratrik Craig Fratrik
    10. January 2014 at 20:20

    Scott, fair enough. I think the statement, “if the Fed appears to try to target NGDPLT and fails, then MM loses” is what Arnold was looking for. He was (is) worried that “appears to try,” === “is successful,” and any failure would imply inadequate effort. One key distinction between you two is his hypo is announcing a target, and yours is appears to try (to his satisfaction).

    New point: critics of MM seem to worry about what they call asset bubbles. (I’m with Scott that these might be pure hindsight labeling.) Another way NGDPLT might fail in the eyes of others is that asset prices fluctuate significantly enough, people might say that MM is fueling asset bubbles.

  18. Gravatar of dtoh dtoh
    10. January 2014 at 23:06

    I think the question for Arnold is:

    Do you believe that Fed policy (specifically the purchasing of assets) does not have an impact on output, demand and prices?


    Do you believe that it does have an impact but that the Fed can not control it well enough to hit a target?

    Until you define the question, the discussion of fasifiabilty is pointless, and…..

    IMHO, there is no question but that…

    1. Everyone (except Cochrane and Krugman) now agree that monetary policy can increase demand and output even at the ZLB.

    2. Fiscal policy can impact demand and output.

    3. Monetary policy can offset fiscal policy (or vice versa).

    4. The Fed can hit any target (price, real output, or NGDP) that it wants.

    There are only two question where there is still substantial disagreement:

    5) Should we rely solely on monetary policy to increase output, or should it be a combination of fiscal and monetary policy.

    6) What is the right target for Fed policy (prices, employment/unemployment/output, or NGDP).

  19. Gravatar of Jeff Jeff
    11. January 2014 at 06:50


    Not so fast. There’s not much empirical evidence that fiscal policy has ever had much impact independent of monetary policy, and in the simplest and most straightforward DGSE models fiscal policy has no effect. Ricardian Equivalence is the default.

    Even if you disagree with this view, it is still the case that (as Scott has said repeatedly) fiscal policy is very costly and monetary policy is cheap. So there really isn’t any reason to even consider fiscal policy until you’ve actually tried and failed with Scott’s monetary policy prescription.

    Finally, MM says that monetary policy works by affecting expectations. What Scott wants to target is not NGDP levels or growth rates, but the expected level of NGDP. The Fed may or may not be able to hit an NGDP target precisely, but using Scott’s NGDP futures scheme or something like it would enable the Fed to precisely target expected NGDP.

  20. Gravatar of ssumner ssumner
    11. January 2014 at 08:27

    Paul, You said:

    “After the past five years it’s hard to conceive of a Fed policy- short perhaps of returning to the gold standard at the pre-Civil War parity- that everyone would agree constitutes “not trying”.”

    Not at all, When the BOJ tightened policy in 200 and 2006 almost everyone would agree they were not trying to push inflation up to 2%. Right?

    Craig, Whatever system we have there will be lots of “asset bubbles,” but of course no actual asset bubbles, just the scare quotes variety. So yes, NGDP targeting will get blamed for that problem, just as inflation targeting has been blamed.

    dtoh, Why do you think monetary policy can hit a RGDP target?

  21. Gravatar of dtoh dtoh
    11. January 2014 at 08:52

    I don’t think monetary policy can hit ANY RGDP target, but I think it can hit AN RGDP target if the target is not too high. If policy can hit an NGDP target and the price component turns out to be large, then there is a natural arbitrage to buy production capacity now to take advantage of higher future prices for goods. Good NGDP targeting will therefore generally yield RGDP growth at or near the production capacity of the economy. Hitting any RGPD target below the production capacity to me seems trivial as the Fed has amply demonstrated over the last 6 years.

  22. Gravatar of dtoh dtoh
    11. January 2014 at 09:14

    If there is an AD shortfall, then IMHO there is no question that Ricardian equivalence does not apply.

    I totally agree that monetary policy is better and more effective than fiscal policy, but to win the MM debate, Scott and others need to be open to a debate about fiscal policy rather than dismissing it out of hand. The NKs like fiscal policy for a lot of reasons that have nothing to do with output and efficacy, but it is a debate which they will lose and therefore they prefer talking about MM at the ZLB or the fasifiability of MM. In order to win, MMs need to do the following.

    1. Convince everyone that monetary policy always works…even at the ZLB. (I think this has been done. Even DeLong has capitulated on this point. The only people still seriously debating this is ideologues like Krugman and Cochrane who’ve sold their souls to the politicians).

    2. Force the NKs to have the debate about monetary policy (generally not MM specific) versus fiscal policy. (It’s not possible to have the debate if MMs deny that fiscal policy won’t work).

    3. Once 1) and 2) have been accomplished, it’s trivial to convince people that NGDPLT is better than a Taylor rule or an inflation target.

    Right now, the MMs are fools for focusing on 3) rather than having than forcing the NKs into an open debate about 2).

  23. Gravatar of dtoh dtoh
    11. January 2014 at 09:17

    Sorry. 2) should read “if MMs deny that fiscal policy WILL work.

  24. Gravatar of Jeff Jeff
    11. January 2014 at 17:07


    As Scott has pointed out dozens of times since he started this blog, the debate in the profession was over 20 years ago. Until the financial crisis, virtually everyone agreed that fiscal policy was useless as a discretionary policy tool. What the proponents of fiscal policy (Krugman and his sycophants) are arguing for is not New Keynesianism, it’s crude old-style Keynesian stuff that was discredited long ago. It survives only because it gives political cover to those who want to increase government spending for other reasons.

    To see this more clearly, imagine you wanted to do a really big fiscal stimulus. Like Obama discovered, there aren’t enough “shovel-ready” projects available to rapidly increase spending on infrastructure and the like. Tax cuts are the only practical way to rapidly increase the deficit. But you didn’t see Krugman calling for tax cuts, because fiscal stimulus isn’t really what he was after.

  25. Gravatar of Paul Zrimsek Paul Zrimsek
    11. January 2014 at 20:23

    I don’t recall whether everyone actually agreed that the BOJ was tightening policy in 200? and 2006, but I’m certainly willing to take your word for it. There still seems to be a lot of territory between that and the sort of measures you’d need to move NGDP back towards its target in a 2009-type situation. (Consider how many people thought money was easy in 2009.)

    Still, the agreement that we might need to import our measure of ease vs. tightness from outside market monetarism in order to test it properly is a step in the right direction.

  26. Gravatar of dtoh dtoh
    11. January 2014 at 21:37

    I don’t disagree at all. My comments still hold though. PK&Co are just using ZLB, MM or whatever they can grasp as a red herring in order to avoid the ral monetary vs. fiscal policy debate. Scott and others should not take the bait.

  27. Gravatar of ssumner ssumner
    12. January 2014 at 09:08

    Paul, With NGDPLT I don’t the think the Fed wouldn’t have needed to do much in the “concrete steppes” area back in 2009.

  28. Gravatar of ssumner ssumner
    12. January 2014 at 09:09

    Everyone, It’s not a question of fiscal vs monetary policy. Unless we move to barter we will always be using monetary policy. The question is how do we use it? And do we want to also use fiscal policy? And if so, why? Summers and Krugman are giving very different arguments for fiscal policy.

  29. Gravatar of dtoh dtoh
    12. January 2014 at 12:19

    I agree. Let me re-frame the question more precisely.

    Do we use fiscal policy to solve (or help solve) cyclical AD problems?

    I think that’s the question you need to force the debate to focus on. If there was no debate on this question, most everyone (PK included) would agree that NGDPLT is the right target. The problem is that many NKs are so desperately clinging to a fiscal approach, that they simply find it more convenient to use a debate on the efficacy of MM as a smokescreen to confuse the more general question of the relative efficacy of fiscal versus monetary policy. IMHO, you’re playing right into their hand.

    If I were you, I’d do a post boldly proclaiming FISCAL POLICY CAN SOLVE AD SHORTFALLS and then go on to explain why monetary policy is more effective. Once you smoke these guys out on the fiscal policy issue, winning the NGDPLT debate will be easy.

  30. Gravatar of Jeff Jeff
    13. January 2014 at 04:25

    Dtoh, what evidence is there that fiscal policy has ever successfully solved an aggregate demand problem?

  31. Gravatar of ssumner ssumner
    13. January 2014 at 05:04

    dtoh, I don’t think fiscal policy can solve AD problems, because I don’t think the central bank will allow it to. If Congress had done less austerity in 2013 then the Fed would have done less stimulus. One can make a better argument that fiscal policy can solve other problems like bubbles, but I’m skeptical of that.

  32. Gravatar of dtoh dtoh
    13. January 2014 at 13:19

    But that’s a different argument.

    Your argument really rests on the Fed being more nimble than Congress. What would happen if Congress passed a law dictating a specific amount and rate of growth of the base?…Or what would happen if Congress passed a law that set a different target than the CB, and the law automatically adjusted income tax rates and monthly withholding if the economy was not on target.

    Bottom line, unless the Fed acts to offset it, fiscal policy will impact AD. If it didn’t we wouldn’t even be talking about an offset.

  33. Gravatar of ssumner ssumner
    14. January 2014 at 20:55

    dtoh, Suppose fiscal stimulus was offset by crowding out of private investment. Would you still say it works? And that crowding out was a different issue?

  34. Gravatar of dtoh dtoh
    15. January 2014 at 11:17

    No, but when AD is growing slowly, it is certainly possible to conceive a scenario where the level of Fed OMO is fixed and where increased government spending or tax cuts could result in both higher AD and higher private investment.

Leave a Reply