NGDP now has nowhere to hide

Back in late 2008 I was horrified to see NGDP growth expectations plunging sharply.  This is exactly what a modern central bank is supposed to prevent from happening.  And yet almost no one (with a few exceptions like David Beckworth) was discussing this issue.  That got me going on this crusade, even before I started blogging.

Commenter 123 just sent me a FT article that quotes 90 experts on British monetary policy (I didn’t even know there were so many experts.)  Much of the discussion is on the merits of NGDP targeting; whether it will happen and whether it is a good idea.  Respondents are split on both questions.  Before getting into specifics, let me point out that market monetarists have already won one battle.  Never again will global NGDP plunge at the sharpest rate since 1938 with almost no one paying attention, almost no one noting that monetary policy is way too tight.  That doesn’t mean that we will have explicit NGDP targeting—but at a minimum it will be part of the discussion, part of the policy mix.  NGDP now has nowhere to hide.

Here’s one interesting critique of NGDPL:

David Blanchflower, Dartmouth College and former MPC member:

The Treasury should – and I suspect will – abandon inflation targeting which failed to deliver stability. It should broaden the remit to include growth although I am unconvinced of the merits of a numerical NGDP target not least because of the problem of data revisions. About 2/3rds of real GDP revisions down to revisions to NGDP. It makes little sense that the MPC might have to call an emergency meeting to react to data revisions. I would favour an explicit targeting of the unemployment rate as the Fed has done. The evidence is that a 1% increase in unemployment hurts four times more than a 1% increase in inflation. Unclear though that we need a committee of eight clones when Carney could do it all perfectly well on his own …! So maybe time to think about scrapping the MPC entirely.

I’m genuinely open to ideas here, but let me suggest a pragmatic solution.  Rather than take an either/or attitude toward data revisions, let’s think about the policy regime that would lead to the greatest macroeconomic stability.  I’d like to distinguish between two types of data revisions (at least for the US, I know less about the UK):

1.  Revisions that reflect new data on stuff like international trade, which comes in with a one or two month lag.  These lagging data series largely explain why the main two NGDP revisions occur within the first two months.  Most importantly, these revisions tend to change our understanding of where the economy is right now.

2.  Revisions that occur infrequently, and reflect a different way of thinking about data that was measured in real time much earlier.  For instance, a few decades ago the Italian government added an estimate of the size of the underground economy to its GDP.  If the estimated GDP suddenly rose by 15%, that obviously would not indicate that the economy was dramatically overheating, and in need of ultra-tight money.  It doesn’t change our understanding of where we think inflation and/or the output gap are right now.

I will concede that not all revisions fall neatly into one category or the other.  I recall some big revisions a couple years ago, which changed our understanding of the 2008-09 slump.  But I also believe that a panel of experts could come up with a reasonable estimate of the right GDP series to use in order to maximize macro stability.  For instance, instead of targeting NGDP 12 months out, you could target 14 months forward NGDP, where you use the second revision.  That picks up much of the data lag problem.  With that series, you don’t allow any base drift in NGDPLT.  On the other hand revisions of the sort that Italy undertook to include the underground economy obviously don’t change our current view of where we are in the business cycle, and should be accommodated with “base drift,” which means letting bygones be bygones.

Does that seem like a pragmatic solution?  Other options include targeting Nominal Gross Domestic Income, which in theory is identical to NGDP, but is less subject to revisions.  Or you could develop a model to predict revisions, and target the outcome of that model.  Any other suggestions?

PS.  I’ve always preferred targeting NGDP per capita, or per working age population, but rarely mention that fact.  It has the side benefit of keeping inflation slightly more stable in the long run, for those unenlightened folks who still believe that price inflation causes problems above and beyond the damage caused by excessive and unstable NGDP growth.

PPS.  Totally off topic, but I’ve been asked about David Henderson’s recent critique of my claim that liberal Massachusetts and Washington (state) reject progressive income taxes, whereas many southern states ruled by ultra-conservative Republicans have surprisingly progressive income taxes.  He pointed to the fact the Massachusetts constitution doesn’t allow a progressive income tax.  Yes, but in 1994 this question was put to the Massachusetts voters in a referendum, and the progressive income tax was rejected.  You can’t get much more democratic than a referendum.  Others claimed I cherry picked some states.  True, but the claim is often made (by people like Matt Yglesias) that low MTRs for the rich is the defining issue of the modern Republican Party.  It’s the only issue they really care about strongly.  So I still think these exceptions are interesting to think about.  That’s not to say there aren’t good explanations.  Some pointed to the fact that the GOP has only recently gained power in the South.  And (in comments) David noted that my referendum argument applied to voters, not Democratic politicians in Massachusetts.  Let’s watch to see if the South gradually moves to lower and flatter income taxes.



23 Responses to “NGDP now has nowhere to hide”

  1. Gravatar of StatsGuy StatsGuy
    3. January 2013 at 09:07

    Yes, Mass. is not as liberal as people think – nor is New Jersey. “Red” states are also far more pro-government than they would like to believe, as demonstrated by the net transfer of wealth.

    I was tickled with delight when Chris Christie brought up this issue of Red states being on the dole from blue states.

    BTW – I’m not sure why measurement of NGDP is so much worse than measurement of other factors – like unemployment (measured by BLS survey), jobs (subject to _notable_ monthly revisions and seasonal adjustments), etc. From an information aggregation perspective, one could argue that the Fed as a committee has access to a lot of information that helps correct for likely errors (beige books, various jobs reports, taxes, manufacturing, and more), but the onus is then to demonstrate that subjective decisionmaking with more information is better than objective decisionmaking with limited information. The latter also has the benefit of predictability.

    NGDP/working age person (or even, non-disabled working age person) would dramatically change policy btw – this is not a small detail in implementation. You might want to elaborate on why it’s better.

  2. Gravatar of W. Peden W. Peden
    3. January 2013 at 09:13

    Isn’t this one of the reasons to favour (in an ideal world) a money wage target over an NGDP target?

  3. Gravatar of Philo Philo
    3. January 2013 at 12:38

    I thought your proposal was to target *the forecast*. You shouldn’t care about revisions of data (concerning the past); you just want the forecast for NGDP one year (or whatever) in the future to be growing at a steady 5% (or whatever). It seems to me that you should dismiss Blanchflower’s worry as based on a (slight) misunderstanding of your proposal. (Or is it I who have misunderstood?)

  4. Gravatar of Geoff Geoff
    3. January 2013 at 13:57

    Dr. Sumner:

    “Does that seem like a pragmatic solution? Other options include targeting Nominal Gross Domestic Income, which in theory is identical to NGDP, but is less subject to revisions. Or you could develop a model to predict revisions, and target the outcome of that model. Any other suggestions?”

    It would be pragmatic if the Fed targeted the money supply based on a productivity rule. That would be even closer to an optimal situation, because in a laissez-faire market, the production of money would tend to correlate with the productive capacity of the economy in general. With more production, more free market money would be produced. The Fed could track a proxy for that.

    It has the benefit over NGDP/NGDI targeting by minimizing discoordination between countries. If the US is a large importer, then in a free market money would flow out, which would reduce the domestic money supply, and reduce domestic spending and prices. The US and those countries with which it has a deficit with, would be more in balance with respect to the valuations of goods imported and exported between those two countries.

    I am thinking that the same principle of deficits and surpluses is what balances individuals, firms, industries, states, etc with each other. If every individual received 5% more revenues every single year, coordination between people would be all but lost. But with fluctuating deficits and surpluses, coordination can take place.

    Same thing with a world market and country economies. Each country’s CB could produce only as much money as the country otherwise would produce in a free market. The more productive a country, the more money the country’s CB can create. The less productive, the less money. Thus, a highly productive country with high imports woould not face significant deflation. It would only experience significant deflation if it were a low productive country with high imports. Similarly, a low productive country with low imports would not experience significant deflation, and may even experience inflation.

    This, I think, is a very optimal solution, because it mimics the coordination that stable spending within countries has when it comes to surpluses and deficits, but at the world level.

    Personally, I don’t think that the US printing money, and Americans buying goods from China from a central bank that keeps the Yuan to dollar exchange rate artificially low, which has made us essentially dependent on continuous dollar inflation exporting to support our standard of living, is a sustainable, healthy situation. With all those dollars going to China, we should be experiencing declining money supply and spending here, so that our goods become less expensive and more attractive so that we can compete with China in terms of productivity. I think the US dollar system as it currently stands is keeping us down in a state of low productivity and dependency.

  5. Gravatar of Benjamin Cole Benjamin Cole
    3. January 2013 at 17:12

    Here’s a toughie:

    The FOMC minutes reveal deep divisions in the committee, and Richard Fisher is in hysterics about inflation, very publicly so.

    So how can there be Fed “credibility” that it will stick to NGDP targets?? Will not market players fear the Fed will cave in to the Fishers?

    I think we have to clarify who is in charge of Fed policy and that it has to be one person. The FOMC board is becoming a hindrance to sound monetary policy.

  6. Gravatar of Bob Murphy Bob Murphy
    3. January 2013 at 18:07

    Of course I don’t like your views, but I do congratulate you Scott for causing such a turnaround in a short time. It is really impressive, and I can tell me kids I knew you before you were famous!

  7. Gravatar of Bob Murphy Bob Murphy
    3. January 2013 at 18:08

    * my kids

  8. Gravatar of Becky Hargrove Becky Hargrove
    3. January 2013 at 18:13

    Neither of the points you raised about potential revisions seem to represent significant impediments, unless data collection is a lot more difficult than I realize. Rather, should Nominal Gross Domestic Income be implemented, it seems your second point would (eventually) matter but for a different set of reasons. The informal economy in Italy for instance has been adapted to include “hidden” money in formal measurements, while a growing knowledge based economy in the U.S. is a different thing entirely: it wants to be formal but does not yet have a way to be expressed through money. IOW changing definitions of work over time can effect long term optimization for NGDILT but should not really be a problem in the present.

  9. Gravatar of jknarr jknarr
    3. January 2013 at 18:57

    They already call emergency meetings to react to events. I’d prefer them to be called in order to calmly address incremental better information, rather than the current policy of reacting to systemic blowups.

    Change the regularity of monetary actions to fit the information flow, and/or weigh actions in accordance with information arrival. Keep a board to conduct oversight of the data and reaction function, with independent audits – not ever one guy, even Carney.

    I’d avoid building bells and whistles on the clean version of NGDPLT. The point is to keep the ultimate goal of economic activity matched with monetary accommodations, not to influence their distribution or character. Population, technology, capital tend to grow slowly, and the black-death-or-cold-fusion counter arguments are silly: they don’t happen in any practical sense.

    The real goal ought to be identifying variables that would lead to a leveraged blowup, one which would steer policymakers to abandon NGDPLT (if you want to refine the proposal).

    BD/CF/per capita/revisions questions are not nearly as important as addressing debt formation under NGDPLT, i.e. reserves, the role of the Fed, and moral hazard in banking under NGDLT. Assume 5% per annum stable NGDP for the next 30y, and debt looks very different.

    A more exciting topic, anyway.

  10. Gravatar of Rien Huizer Rien Huizer
    4. January 2013 at 07:11


    Some glimpse of a technology here. Things are moving forward.

  11. Gravatar of ssumner ssumner
    4. January 2013 at 08:12

    Statsguy, See W. Peden’s comment about wages.

    Philo. But it could be a problem for level targeting. (not growth rate targeting) Consider the Italian example I provided. After the revision, do you want to aim to cut NGDP 10% over the next year? Obviously not.

    Geoff, I view money demand as being too unstable.

    Ben, Thanks, I’ll take a look.

    Thanks Bob.

  12. Gravatar of flow5 flow5
    4. January 2013 at 09:59

    “I am unconvinced of the merits of a numerical NGDP target not least because of the problem of data revisions”

    Roc’s in N-gNp are proxies for all transactions “PT”. Roc’s in MVt = PT. Revisions to M & Vt don’t vary to the degree the 3 gdp releases do (monthly figures for advanced, second, & final). And roc’s in MVt are cummulative numbers. They are calculated over several qtrs (not just for 3 months) & reflect constant lags.

    It takes big changes in MVt to produce small changes in gdp figures. Any error at the end of the lag (where they inevitably crop up) either becomes too small to have a meaningful impact or is more than obvious.

  13. Gravatar of flow5 flow5
    4. January 2013 at 10:47

    And if you focuse on the short-run (1 qtr) you make big errors. Take “Black Monday”. The roc in MVt crashed from 16 to 4 (from Aug to Nov). This overlapped 2 gDp releases.

  14. Gravatar of Britmouse Britmouse
    4. January 2013 at 15:35

    Scott, this is very interesting and it’s a good critique from Blanchflower, probably the only good one I’ve seen since Carney’s speech, and there’s been a lot of noise.

    I want to look at the data more to understand the impact of revisions we’ve had in the UK. I’m not sure if “base drift” is going to always be a sufficient solution.

    Could there not be a case where a methodology revision not only shifts up historic levels but also growth *rates*? I could see this happening with Owner-Occupier Imputed Rents for example; that portion of NGDP is mostly the invention of statisticians.

    If a historic 5% NGDP path was revised into a 5.1% path purely due to a methodology change, wouldn’t it be desirable to accommodate that by targeting a new 5.1% path – not merely to target a new 5% path with “base drift”?

    I think this kind of thing has happened in the UK. We do have that particular problem discussed previously, where some of NGDP is derived from volume data + deflators, but there have been some significant methodological changes recently too. I’ll try to blog a writeup on this and look at the data for NGDP by the income method to see how much that gets revised.

  15. Gravatar of ssumner ssumner
    4. January 2013 at 17:46

    Britmouse, That’s a good point, but in practice I think changes in growth due to methodological changes are small enough that they can be ignored. I’m pretty sure the issue is changes in levels. So I’d keep the 5% and shift the level if necessary.

  16. Gravatar of Philo Philo
    5. January 2013 at 19:39

    No matter what one proposes as a target for the monetary authorities–NGDP or anything else””the objection might be raised: “But the authorities might in midstream *redefine the target magnitude*, producing a discontinuity in policy”? But why take such an objection seriously? Obviously the proposal will have assumed an interpretation of the target magnitude that is *uniform and unvarying through time*.

    I don’t consider it incumbent upon you to discuss how your policy should be revised if it became clear that something other than NGDP (as you presently conceive it) was really the appropriate target. This would be a *meta*-topic, about making the transition from one target to another, rather than the specification of a particular target.

    (Regarding the Italian example: governmental estimates of the underground economy shouldn’t be used for making economic policy, since they are mere guesses, not at all “hard” data.)

  17. Gravatar of ssumner ssumner
    6. January 2013 at 11:23

    Philo, Good observation.

  18. Gravatar of UK NGDP Targeting: Sceptics Round-Up « uneconomical UK NGDP Targeting: Sceptics Round-Up « uneconomical
    11. January 2013 at 04:15

    […] – David Blanchflower’s response in the FT questionnaire was the most astute.  Scott and Lars have both done posts on this question, but I’ll add the […]

  19. Gravatar of Oh, You Want Flexible? We Can Be Flexible! « uneconomical Oh, You Want Flexible? We Can Be Flexible! « uneconomical
    7. February 2013 at 07:09

    […] him refer to notes, he just knows the data.  That is excellent!  Scott is surely right that NGDP will never again be forgotten – that’s progress at […]

  20. Gravatar of Opinion: The case for NGDP targeting (1) Opinion: The case for NGDP targeting (1)
    22. February 2013 at 06:00

    […] estimate the future path of nominal GDP and therefore the appropriate policy stance. There are also ways of adapting the policy regime to […]

  21. Gravatar of TheMoneyIllusion » Bean on NGDPLT TheMoneyIllusion » Bean on NGDPLT
    27. February 2013 at 08:40

    […]  In that case revisions are not a problem.  In the case of NGDPLT, I’d refer you to this earlier post.  The article […]

  22. Gravatar of A practical critique of NGDP targeting | Basil Halperin A practical critique of NGDP targeting | Basil Halperin
    1. January 2015 at 10:47

    […] then the Fed should just target NGDP per capita. Indeed, Scott Sumner has often mentioned that he actually would prefer an NGDP per capita target. To be frank, I think this is an even worse idea! This would require the Fed to have a long term […]

  23. Gravatar of Basil Halperin's Critique Of NGDP Targeting – PakiMania Basil Halperin's Critique Of NGDP Targeting - PakiMania
    23. February 2017 at 05:07

    […] then the Fed should just target NGDP per capita. Indeed, Scott Sumner has often mentioned that he actually would prefer an NGDP per capita target. To be frank, I think this is an even worse idea! This would require the Fed to have a long term […]

Leave a Reply