Why NGDI targeting is superior to NGDP targeting

The title of this post is a joke, which I’ll explain later.

Tyler Cowen linked to an interesting article in the FT, explaining why Gross Domestic Income is much more accurate than Gross Domestic Product:

Since the start of the recession, GDI has proved the more accurate depiction of US economic performance, according to [Jeremy] Nalewaik’s work. As better data have become available and the Bureau of Economic Analysis (which calculates both) has accordingly revised its earlier estimates, it is GDP that has been adjusted in the direction of GDI rather than the other way round.

Have a look at this chart from Nalewaik’s recent paper:

Neither measure was perfect, but early GDI estimates were much closer than GDP to later revisions of both measures. Perhaps more tellingly, GDI started signaling an economic slowdown in the middle of 2007 even as GDP kept climbing. Early GDI estimates also turned out to better reflect the severity of the recession.

To give the most glaring example, the initial GDP estimate for the fourth quarter of 2008 showed that the economy contracted by 3.8 per cent. It was released on January 30, 2009 “” about three weeks before Obama’s first stimulus bill passed. That number was continually adjusted down in later revisions, and in July of this year the BEA revised it all the way down to a contraction of 8.9 per cent.

The FT is discussing real GDP and real GDI, but I believe the same applies to the nominal versions.

Some commenters have pointed to a flaw in NGDP targeting, level targeting.  If there is a sudden and massive revision in the current level of NGDP, it would force the Fed to shoot for much more or less than 5% NGDP growth over the following year.  That could be destabilizing.  Generally the revisions to current NGDP aren’t that large, but on occasion they can be significant.  Thus I now think NGDI is the better target.

The article also sheds light on the ultra low RGDP growth in early 2011, which many took as evidence that QE2 failed.  But at the time it seemed like QE2 was working, as the monthly jobs figures increased substantially, and other data such as ISM numbers showed an extremely strong economy.  So how could the RGDP numbers have been so weak?

Speaking broadly, two types of evidence suggest that the initial estimates of GDI are typically better than the initial estimates of GDP.

First, a variety of business cycle indicators that should be highly correlated with output growth-including the Institute for Supply Management surveys, the change in unemployment, some financial market variables, and even GDP growth forecasts themselves-have actually been more highly correlated with GDI growth than with GDP growth in recent decades.

That last item on the list is worth emphasizing: economists forecasting GDP growth have produced median forecasts that have tended to be more highly correlated with GDI growth than the variable they are trying to forecast.  It is also notable that GDI growth tends to predict GDP growth next quarter better than GDP growth itself.  All this suggests initial GDI growth is picking up some real fluctuations in the economy that are being missed by the initial GDP growth estimates.

Second, initial GDI growth estimates have tended to predict revisions (typically years later) to initial GDP growth estimates, especially since the mid-1990s.  So, if initial GDI growth is above initial GDP growth, GDP growth tends to revise up, and if initial GDI growth is below initial GDP growth, GDP growth tends to revise down.

For an example of the latter, just look to the recent recession.  In March 2009, the Bureau of Economic Analysis announced that real GDP declined 0.8 percent from the fourth quarter of 2007 to the fourth quarter of 2008, while their GDI calculations showed a much more substantial decline of 2.1 percent.  The latest Bureau of Economic Analysis estimates show declines over that time period of about 3 percent, using either measure.

So, while neither measure initially captured the full severity of the downturn in 2008, the picture painted by the initial GDI estimates was quite a bit closer to the revised figures (which incorporate more complete data and are generally assumed to be more accurate).

What is GDI telling us about the economy now vs what GDP is telling us?

GDI paints a less-bleak picture of the economy recently.  In the first quarter this year, the latest real GDP growth estimates show annualized growth of 0.4 percent, while the latest real GDI growth estimates show 2.4 percent growth.  Other economic indicators like the Institute for Supply Management surveys and the change in the unemployment rate were looking quite healthy over the first few months of this year, suggesting GDI was more accurate in the first quarter.  In the second quarter this year, GDP growth is currently estimated at 1.0 percent while GDI growth is 1.5 percent, so GDI again looks better, but the difference is less sharp.

So the real GDP estimate for the first quarter of 2011 was probably wrong.  To be sure the RGDI number is also disappointing, but at least it isn’t horrible.  Unfortunately it will be many years before we know what actually happened, but for now I’m sticking with ISM numbers and jobs numbers as the best near-term indicators.

Even by those criteria, QE2 was far less than needed, but I’d add that ever since the Fed signaled (in the spring) that it wasn’t going to extend QE2 after June, the economy has done even more poorly than during late 2010 and early 2011.  QE2 was far too little to make a major dent in the economy, but it was probably still better than nothing, which is all we have today.

Now for the joke.  Some readers might have assumed that I am abandoning my advocacy of NGDP targeting.  No so, because NGDP and NGDI are exactly the same thing.  Thus I still favor NGDP targeting.  However government estimates of NGDP and government estimates of NGDI do differ.  And it seems like reported NGDI is the more accurate estimate of actual NGDP/NGDI.  Thus the actual implementation of NGDP targeting should involve the targeting of futures contracts with a value at maturity linked to future announcements of NGDI.



22 Responses to “Why NGDI targeting is superior to NGDP targeting”

  1. Gravatar of johnleemk johnleemk
    29. September 2011 at 06:39

    I was somewhat confused up until the last paragraph, because as anyone who’s taken Econ 101 learned, GDI and GDP are the same thing…GDP when calculated via the income approach is GDI.

  2. Gravatar of Scott Sumner Scott Sumner
    29. September 2011 at 08:52

    Johnleemk, Yes, but MEASURED NGDP and NGDI can differ.

    Sorry for the confusion, but the title warned about the joke. In any case, properly measuring NGDP is a serious issue.

  3. Gravatar of John hall John hall
    29. September 2011 at 09:05

    For the point about level targeting and revisions, my problem was more with the way to structure the futures contracts than the impact of the revisions on monetary policy. Basically, you propose one-day futures contracts that are like a weighted average of two future quarters’ NGDP. If this is the scheme, then it doesn’t suffer a problem with revisions since no contract lasts more than a day. That being said, there’s no way in hell this is anything like what will actually occur. No futures market behaves like this. What would actually occur is like a contract for the annualized growth (I’m actually partial to 4 times LN(changes to GDP)) of the current and every quarter out like five years and you could fix settlement based on like the 3rd GDP report. When one contract settles, you offer a new one. In this case, you can’t use the levels because of annual and quint-annual (is that a word?, once every 5 years) revisions. For instance, if I purchase a 1 year NGDP futures contract on levels the day before a revision, then when it is revised up/down I would profit/loss, even if the growth rate the period I care about is unchanged. If you did do level contracts, the exchange would create all these funny rules so that wouldn’t happen anyway. Hence, it is much easier to promote the idea of futures contracts based on the individual quarters growth rates. The profit/loss of the contract depends only on the change over the relevant quarter so your pnl would only be impacted by a revision to the specific quarter or to changing expectations. It is still possible to implement level targeting under such a system. If you use the log method I mentioned, then the central bank purchases equal amounts of contracts until the average of the four relevant contracts equals ln(1+target)

    That being said, if you believe in level targeting and what you’re targeting is revised, then you SHOULD change your monetary policy. I still haven’t been sold completely on level targeting, but my qualms have nothing to do with the possibility of revisions to historical data (mainly my problem is the error with which you would estimate the appropriate level in the future).

  4. Gravatar of Scott Sumner Scott Sumner
    29. September 2011 at 09:20

    John, I’m afraid I don’t follow your argument. Why are daily NGDP contracts impossible? You say futures markets don’t work this way, which is true. But this isn’t a normal futures market, it’s a Fed policy tool.

    I also don’t understand why price jumps after a revision would be a problem.

  5. Gravatar of Gabe Newell Gabe Newell
    29. September 2011 at 13:09

    Is there any reason a private agent couldn’t create a NGDP/NGDI futures market?

    Separately, if monetary policy for one currency was targeting NGDP/NGDI futures, how would a deep pocketed trader or central bank controlling monetary policy for a different currency game the system?

  6. Gravatar of OneEyedMan OneEyedMan
    29. September 2011 at 15:04

    If we have reached the point where we are deciding between which contract to setup, I think scott’s won. Given the costs of setting up marginal markets, it might be best to just set both up, along with CPI and PPI contracts too. Maybe even medical cost contracts too.

  7. Gravatar of Scott Sumner Scott Sumner
    29. September 2011 at 16:50

    Gabe, No, but there seems to be little interest in such a market (fortunately.)

    Research suggests that market manipulation in prediction markets is not a big problem. However if it was I’ve suggested that the Fed should be free to trade in order to offset any suspicious large trades by a single individual or institution. As long as the market is open to everyone, I have no objection to the Fed trading.

    OneEyedman, Even if I’m 100% wrong about everything, it’s a no brainer that the Fed should run an NGDP market. It would cost little, and provide incredibly useful information in real time.

  8. Gravatar of Declan Trott Declan Trott
    29. September 2011 at 17:25

    Nice one. “Income targeting” also sounds much nicer & easier to sell. Particularly at the moment – no one wants more inflation but everyone wants more income.

  9. Gravatar of Cassander Cassander
    29. September 2011 at 18:06

    The relative lack of complicated statistics required to NGDP targeting has always been one of the strongest arguments in its favor. I remember how shocked I was to learn how bogus so many macro-econ statistics were in my Empircal Macro class.

  10. Gravatar of bill woolsey bill woolsey
    29. September 2011 at 18:57

    I don’t at all agree that revisions in the data are destablizing. It appeared NGDP was on target, then we revise the data so that it turns out that it was below target. And now, the target is for NGDP growth to be extra fast.

    Stop thinking about targeting growth rates. The target is for a series of levels of GDP that are increasing. That is what expectations are about. Not any growth rate of NGDP. NGDP one year from now, two years from now, etc., will be at 17000 billion. Then 17850b. There is no, it will be growing at 5% from where it is now. Oh, it is growing 7% when I was expecting it to be only 5%. What? We expect it to be 17850b. Regardless of where it is now (and really, where we find out it was a bit later) we expect it will be 17850 in one year.

    Consider a price level target of 100. This year it will be expected to be 100. Next year 100. The year after 100. Gee, if the price level turns out to be 98, then people will expect 2% inflation. How destabilizing. How about they will expect the price level next year to be 100.

  11. Gravatar of OGT OGT
    29. September 2011 at 19:36

    I’d seen a couple of articles on the relatively better performance of the GDI measures over the GDP measures in the last year and wondered if you had picked up on those. It seems to picking up a steam finally.

    For market participants and economists trying to ascribe causation to any turns in the economy the difference is actually pretty big looking at the chart above. RGDI puts the recession starting in 2006 with a flat lined economy for two years before the financial crisis. It makes the whole episode look like even more of a Wiley E Coyote moment.

  12. Gravatar of John hall John hall
    30. September 2011 at 06:34

    I think you understate the importance of this not being as similar as other contracts. I’m going from your post “Spot the flaw in nominal…”. Basically you were saying that you use a daily futures contract that expires in like 14 months. That means that any investor has more than 365 possible contracts that they could purchase. And since they are all weighted averages of some underlying nominal GDP, there will need to be an arbitrage on all of them. Basically if I only have a view on 1 quarter ahead nominal GDP, then I have to buy the daily contract with 90 days remaining (and constantly roll that over). Then, through arbitrage, my view will roll over the term structure of futures contracts back to today’s 1-year ahead forecast. That seems reasonable. However, there is an issue with the preferred habitat of these contracts. Ie. not all of them will have anywhere near the same amount of liquidity. For instance, you seem to think the market will prefer to only trade the 90/180/270/360 day contracts. However, if there are 10 days left to the end of the quarter and I have a view on this quarter and the next. I’m going to trade the 10-day and the 100-day contracts to express my views. The expected profit on the 90 day contract is really just an average of the 10-day and 100-day contracts. So why not just simplify the structure and only have those two (or rather one each quarter)? The Fed may care about the fixed time horizon, but traders care about the simplicity with which they can express a view.

    Again, I’m basing my assumption on the idea that the Fed only cares about buying the 360 day contract, but traders will trade all kinds of contracts in order to impart their views about the structure of GDP growth. Through arbitrage, this trading will impact the 360 day contract that the Fed cares about.

    I’m not sure how I can explain about the reference point any better. First off, you should want to offer real and nominal GDP contracts, along with a price deflator contract, and you should want to keep the structure the same for both (most people focus more on real GDP or prices and then a simple arbitrage will join them together). If you have a futures contract representing x% of real GDP, then re-basing real GDP will result in a significant change in the value of the contract, which will impact settlement. That should be obvious. The second point is with regards to more normal revisions (yearly, quarterly). Let’s say again I’m a trader with 10-days left in the quarter and views about the quarter after this one ends. The only way for me to express my view is to go long the 100-day contract and short the 10-day contract and roll it over daily. If I do this, then I will not be impacted by any revision since I have hedged it by shorting the 10-day one. However, let’s say I have a view about the upcoming quarter (the one ending in 10 days). Again, I would need to buy the 10-day contract and sell the -80-day contract (I presume there would be a -80-day contract, since you say the settlement would not occur until 2-3 months after the quarter ends). I guess my point about all this isn’t really that it isn’t possible to hedge yourself to avoid the risk of any revisions. It’s more that it is a hassle to have to purchase two contracts and roll them over daily. It makes no damn sense and would cost a significant amount in trading. Why not just offer the contracts directly?

    I’m also not saying the Fed can’t target a weighted average of a term structure of futures contracts and then buy and sell them in proportion to their weights. They could even create a synthetic contract that mimics the the daily contract you discuss. However, I don’t understand why you would start with the synthetic contract when you can alternately start with a firmer foundation that would be much easier to work with in practice? Also, if you work with fixed quarter contracts, then you can offer options on them, which I believe is very important.

  13. Gravatar of John hall John hall
    1. October 2011 at 06:09

    I want to revise my above post. I was thinking this morning and realized that since every contract automatically loses a day. There’s no need for explicit rolling over. Its automatic. That being said, I still believe there will be a preferred habitat on only the contracts that are necessary to express views about the specific quarters’ nominal GDP growth. The trade on all the others will simply be arbitrage. Further, I believe most people will use two contracts to do this to hedge the risk from revisions. Hence, I think it makes more sense to use only some sort of log growth contracts and if you want to do level targeting create a synthetic product from the underlying growth contracts.

  14. Gravatar of Scott Sumner Scott Sumner
    1. October 2011 at 16:39

    Declan, Glad you liked it. I was thinking about your comments when I wrote it.

    Cassander. Yes, I also like the simplicity.

    Bill, Suppose for 20 years in a row we underestimate NGDP growth by 1%. So actual growth is 4% not 5%. But the labor market had adjusted to that mistake. Now you try to make up the 20%, plus 5%? You shoot for 25% NGDP growth in 12 months? That doesn’t seem right.

    John, I think you misunderstood the proposal. The only contracts that matter are those with 14 months to go. Those are the only contract that affect the money supply. All the other contracts issued earlier are irrelevant. It doesn’t matter if liquidity is zero. And the contracts that do matter are guaranteed liquidity by Fed trading subsidies.

  15. Gravatar of Scott Sumner Scott Sumner
    1. October 2011 at 16:41

    OGT, Good point about the flat line. And yet unemployment didn’t rise very much. I suppose the housing collapse reduced measured productivity.

  16. Gravatar of flow5 flow5
    1. October 2011 at 16:57

    “sudden and massive revision in the current level of NGDP”

    Aggregate monetary purchasing power = monetary flows (MVt)&
    MVt = NGDP

    You never look in the rear-view mirror. MVt is always forward-looking. Revisions to MVt don’t ever exist.

  17. Gravatar of John hall John hall
    1. October 2011 at 17:33

    I’m puzzled. If there’s something I’m blatantly misunderstanding here, please help me out. I’m focused mainly on the practical way anyone would trade in this market. I am ignoring the issue of how the monetary policy mechanism will actually work. It is completely unrealistic to think that traders will freely enter the market under the scheme you mention. I now acknowledge it could work if people actually care about the older contracts, only in the sense that they could replicate the way that I would prefer it to be set up.

    However, you say that earlier contracts don’t matter, but, and I can’t say this in any nicer way, this is ridiculous. All those older contracts will be freely traded. And the ones that most closely provide the ability for investors to express their views on nominal GDP are going to be the ones traded. Ie. the ones at the end of the quarters. If you think there won’t be traders who want to take the VERY simple arbitrage to ensure that all other contracts are priced off the end of quarter contracts, then you’re mistaken. If they do follow this arbitrage, then it will impact the price of the contract that the Fed actually cares about.

    My question is, why make it more complicated with 360 possible contracts, when you can have four. It doesn’t make any sense. You say that the Fed will guarantee liquidity on the ones that “DO” matter, but I wonder, why not just adopt a more simplified structure in the first place.

  18. Gravatar of Scott Sumner Scott Sumner
    3. October 2011 at 15:53

    flow5, I have no idea what you are talking about, but then I’m stupid, like Milton Friedman.

    John, I don’t think you understand the proposal. Suppose the subsidy is $10,000,000. Now suppose that no one enters the market. That’s your assumption, right? Now I decide to enter. The Fed pays me the entire $10,000,000. I’m rich. Is that your hypothesis? If not, don’t be so quick to dismiss my ideas.

    The purpose of having daily contracts is to allow daily adjustments in the money supply. You could do weekly, I’d be fine with that. But not quarterly, that’s too infrequent for monetary policy changes. People are going to wait until the last minute before buying these contracts from the Fed, because the price is fixed until they are sold, and obviously the longer you wait the more information you have.

  19. Gravatar of TheMoneyIllusion » A comment on Q3 NGDP TheMoneyIllusion » A comment on Q3 NGDP
    27. October 2011 at 09:32

    […] initial NGDP numbers show a 5% growth rate, a bit better than expected.  But as I discussed in this post, the initial estimates of NGDI are actually better predictors of actual NGDP than the initial NGDP […]

  20. Gravatar of Why GDP and CPI numbers are psuedoscience [Draft] « Intellectual Detox Why GDP and CPI numbers are psuedoscience [Draft] « Intellectual Detox
    29. December 2012 at 13:55

    […] Jeremy Nalewaik has pointed out that GDP tends to be adjusted in the direction of the GDI estimates (GDP and GDI should be indentical, GDP is calculated by adding up expenditures while GDI is […]

  21. Gravatar of Why GDP and CPI numbers are psuedoscience [Draft] | Intellectual Detox Why GDP and CPI numbers are psuedoscience [Draft] | Intellectual Detox
    29. December 2012 at 19:05

    […] Jeremy Nalewaik has pointed out that GDP tends to be adjusted in the direction of the GDI estimates (GDP and GDI should be indentical, GDP is calculated by adding up expenditures while GDI is […]

  22. Gravatar of Axel Axel
    19. March 2013 at 04:53

    @Scott Sumner,
    I understand this discussion is old and my comment can be completely irrelevant: please excuse this if it’s the case but I’m lost.
    One remark about the NGDP or NGDI future mecanism you discuss with John. If there is no player and the fed is the only counterpart making a 10trn / 10trn bid offer on the NGDI, you just receive nothing in cash from trading with it at its price. I mean a future is unfunded – or i’m missing something. if you think NGDI will be massively above 10trn, you can buy at the fed and wait for settlement to pocket in the difference. Even if someone else thinks NGDI will be lower than Fed price, noone will ever make a market which is different from fed sticky price reference as it would provide for an arbitrage opportunity. Therefore any investors would trade with the Fed at its ref price, unfunded contracts. This would not create any money until cash settlement, ie 18 months from contract inception. So this would mean no monetary stimulus/contraction up until 18 months. I guess this is why you advocate daily or weekly contracts. But don’t we have there a ramp up issue? is there any clarity on policy transmission mecanism this would introduce? (settlement would adjust M supply 1 quarter after the actual NGDP miss the targetted level right & it’s hard to assess for how many contracts – as speculation could create/destroy much more money than needed with uncertain impact on real economy).
    If you have detailed post on this ‘future’ system I would be happy to read.
    thank you for your incredible blog & out of the box ideas

Leave a Reply