A few notes on the GDP revisions

The new GDP figures include some pretty significant revisions of the past data.  Most people have focused on the fact that the new figures show a significantly bigger drop in 2009 GDP than originally estimated.  Even the original 2009 NGDP figures showed the biggest drop since 1938, but these have been revised further downward.

It appears to me that most of the revision is based on changes in the third quarter of 2008.  Before discussing those revisions, I’d like to talk about why that quarter is so important.  The standard view of this recession is that the housing slump of 2007 triggered a moderate banking crisis and a very mild recession at the beginning of 2008.  Then after Lehman failed, the banking crisis got much worse, and therefore the recession became much worse.

My view has always been different.  I agree with the standard interpretation of the original, and relatively mild, recession of early 2008.  But I argued that the recession worsened dramatically before Lehman failed, and that this led to a decline in NGDP growth expectations that severely reduced asset prices and made the banking crisis much worse.

One problem with my view is that it seemed like the severe fall in GDP (real and nominal) began in the 4th quarter of 2008, which was after Lehman had failed in mid-September.  Initial reports showed only a 0.3% fall in 2008:Q3 RGDP.  As a result, all I could do was rather pathetically argue that July was a strong month, and point to the fact that industrial production (which is measured monthly) suddenly began falling sharply in August 2008.  And that this showed the economic weakness had spread out of housing even before Lehman failed.

The newly revised GDP accounts paint a very different picture of the recession.  Instead of two quarters of steeply falling RGDP (2008:Q4, and 2009:Q1) there are now three really bad quarters.  The third quarter of 2008 is now estimated to have seen a 4.0% plunge in RGDP.  Nominal growth slowed abruptly from the roughly 5% norm of preceding years, to only 0.4%.  And even that is probably overstated, as the nominal GDP numbers rely on rent imputation values for housing that almost comically overstate inflation during a housing crash (as I’ve discussed in previous posts.)

So now we know that the severe recession of 2008-09 began in the third quarter.  Since Lehman didn’t fail until the quarter was almost over, there is simply no way it could explain why the recession got much worse during those summer months.  What can explain the worsening recession?  How about a Fed that refused to cut rates for nearly 6 months after April 2008, despite a steadily falling Wicksellian equilibrium interest rate.  A Fed focusing on headline inflation numbers driven up by imported oil prices, not the expenditures on American-made goods and services.

BTW, how can the initial GDP numbers be so far off in an economy that is swimming with data and high tech computers?  Four weeks after the quarter ended the RGDP growth was forecast at negative 0.3%, and only now, two years later, we find out it was minus 4.0%?  That’s a pretty major error at a particularly important moment in the business cycle.  Too bad our policy makers were blindfolded as they were making crucial policy decisions.  Time to target the forecast?

One thing that makes me think I am on the right track is that every time I learn something new, it seems to support my view of events.  In many cases this was data that was already out there, but that I had not bothered to look up.  This includes the big drop in TIPS spreads, which began well before Lehman failed.  The huge rise in real interest rates in late 2008.  The huge rise in the dollar in late 2008, the fact that the housing crash spread from the sub-prime markets to the heartland  (and commercial RE) at precisely the moment when the recession spread from housing to industrial production in August 2008.  All of this I discovered in mid-2009, after I had already begun blogging.  But now we have information that neither I nor anyone else had access to until few days ago.  And again, it is strongly supportive of my GDP —-> post-Lehman crisis view of causality, and strongly in conflict with the conventional Lehman crisis —-> falling GDP interpretation.

I eagerly await further data revisions.

PS.  It’s worth looking at the NGDP numbers in the link above.  (Bottom line)  During the 4 quarters from mid-2008 to mid-2009, NGDP actually fell 3%; a bit more than 8% below the 5.2% long run trend, not less as I had assumed.  And during the so-called “recovery” of the past 4 quarters it has fallen another 1% below the 5% trend.  Where’s the effect of that $787 billion in fiscal stimulus?

And for those who believe wage and price flexibility solves all problems, consider that with 4% NGDP growth, we’d need 3.7% deflation to get the 7.7% RGDP growth we saw during the first 6 quarters of the 1983-84 recovery.  That recovery had 11% NGDP growth. When was the last time you saw an economy growing at 8% in a period of 4% deflation?  Never?  There’s a reason for that, wages and prices aren’t nearly flexible enough to overcome that sort of nominal sluggishness.

Tags: ,


31 Responses to “A few notes on the GDP revisions”

  1. Gravatar of Doc Merlin Doc Merlin
    2. August 2010 at 07:26

    Btw if you said “income expectations” instead of NGDP expectations, I think a lot more right winged economists would support your view.

  2. Gravatar of Benjamin Cole Benjamin Cole
    2. August 2010 at 08:31

    A very compelling commentary by Scott Sumner. I am wrestling hard to get solid handle (or even a loose grip) on monetary policies, but the more I read Sumner and the more I learn, the more I think the Sumnerian viewpoint is the one needed now.
    BTW, I wrote a letter to Dallas Fed Prezzy Richard Fisher, and asked him it he truly believed the “ultimate goal” is price stability as opposed to economic prosperity.
    Yes, one e-mailed letter is likely to have the weight of a spitwad against a battleship. Still, when a Krugman is publicly responding to Sumner, that means traction is going on. The more Sumnerian viewpoints are brought up in public or Fed forums, the better.

  3. Gravatar of jsalvati jsalvati
    2. August 2010 at 09:29

    @Doc Merlin

    That’s not a bad idea. Maybe “dollar income expectations” or “nominal income expectations” or “spending expectations” to emphasize that it’s a nominal quantity.

  4. Gravatar of D. F. Linton D. F. Linton
    2. August 2010 at 09:58

    Why did you choose to focus on NGDP? Do you want additional Fed monetary expansion if imports rise or conversely Fed tightening if exports rise?

  5. Gravatar of Doc Merlin Doc Merlin
    2. August 2010 at 10:46


    Agreed. People see NGDP and think its just some sort of rough aggregate that doesn’t correspond to something actual. When we call it “nominal income expectations” it really brings home the microeconomics involved.

  6. Gravatar of Doc Merlin Doc Merlin
    2. August 2010 at 11:51

    Most Chicago school economists (Scott is one) say that imports and exports are red herring wrt macroeconomic policy.

  7. Gravatar of MW MW
    2. August 2010 at 12:11

    “Time to target the forecast?”

    Meaning the forecast from your proposed NGDP futures market?

  8. Gravatar of John Hall John Hall
    2. August 2010 at 12:12

    Good post Scott. If you wrote a book on this period, I would buy it.

    The only thing I would add is that the whole TIPS market got massively messed up when people started anticipating deflation (even ignoring the liquidity issues from margin calls). The principal can’t be adjusted below the initial payment. A long-dated TIP that is nearing maturity can see its principal fall to the initial payment, but a newly-issued one will not make a similar decline. Hence, the break-even inflation spreads over that period were a bit crazy. Some people look at inflation breakevens from forwards/swaps instead.

  9. Gravatar of Benjamin Cole Benjamin Cole
    2. August 2010 at 12:24

    BTW, the muchuoted Bill Gross of PIMCO says deflation is possible, and he’s investing accordingly see page C1 WSJ.
    Gross is one of those “bellwether” observers, and his observations are given weight.

  10. Gravatar of Liberal Roman Liberal Roman
    2. August 2010 at 12:59

    Here is the reason why we are in such big trouble:


    We are in the most deflationary period in 80 years and people are STILL worried more about inflation than deflation. In fact, they don’t even understand why deflation is bad. The public just doesn’t get it and some of them don’t even want to get it. THAT, above all else, is the biggest problem that we face.

  11. Gravatar of Morgan Warstler Morgan Warstler
    2. August 2010 at 13:05


    It’s pretty clear, we can use some “deflation” right now:

    1. Energy is the big drop eeping things under 1% – as noted, when will we see that at the pump? More to the point, China’s going to be buying more soon enough.

    2. Housing NEEDS to drop. 4M homes 50% underwater – please jesus will these people just walk the hell away? Stop serving the bankers and get yourselves a nice rental.

    3. Veggies – I’m going to be grateful for rain.

    4. Computers – cheaper every day please!

    5. Meanwhile, prices on everything else are headed up.


    Scott, it occurs to me… maybe all it really takes for the Fed to finally freak and target NGDP – is rapidly falling homes prices.

    You and I are on the same side!

  12. Gravatar of azmyth azmyth
    2. August 2010 at 13:32

    “One thing that makes me think I am on the right track is that every time I learn something new, it seems to support my view of events.”

    This happens to virtually everyone – beware confirmation bias! In addition to looking for confirmation, think of possible facts that would weaken or refute your hypothesis. I don’t know of any countries that managed to stabilize NGDP, but if there were, they would make an interesting test.

  13. Gravatar of Manny C Manny C
    2. August 2010 at 15:06

    Do these GDP revisions weaken the idea of an NGDP futures market given that GDP data is subject to such massive revisions?

  14. Gravatar of Benjamin Cole Benjamin Cole
    2. August 2010 at 15:07

    BTW OT but good: Krugman today in his blog referred to this paper…and it calls for higher inflation targeting in times of below-target inflation rates…
    sounds Sumnerian to me….


  15. Gravatar of Morgan Warstler Morgan Warstler
    2. August 2010 at 15:26


    Scott, you’ll need to do a better job than Krugman, because my biggest goal now is to keep getting him to FAIL at explaining why we need to worry about inflation.

    All he’s got for experience is a bunch of old savers in Japan to and now in just free discussion, he makes it sound down right sexy. Oh wait, no sex talk.

    It’s just, it really seems like if you assume correctly that housing prices are falling and will continue to fall…

    Insisting deflation is bad just becomes an excuse for OBNOXIOUS debt forgiveness for the bad apples and banks.

    Isn’t there a more honest discussion to be had about inflation / deflation MINUS the heavily weighted housing category doomed to fall?

  16. Gravatar of Morgan Warstler Morgan Warstler
    2. August 2010 at 16:02


    June 2010 “All items less shelter” Y/Y 1.9 – almost the damn Fed target.

    Sept. ’09 3.9
    Dec. ’09 3.9
    Mar. ’10 2.5
    June ’10 -2.6 And the big dump June is energy?

    I’m really nervous the whole thing we’re yelling “oh no deflation!” about is a mixed bag of inflation and a heavily weighted fall off in rents, and some cheaper motor fuel prices last three months I do not see at my pump.

    Question: How do we get reposition the inflation discussion minus rents? Otherwise it feels like a scam.

    Moreover, perhaps the Fed refuses to worry about inflation driven by falling home prices, because they know that’s the smart way to deal with something so aberrant.

  17. Gravatar of Joe Joe
    2. August 2010 at 18:08

    Professor Sumner,

    I enjoyed your commentary, it was illuminating as always. I second John Hall’s opinion. I have just one quick question.

    My understanding of how tight money causes recessions is that Say’s law breaks down when money demand unexpectedly jumps up, causing an “equal” fall in AD, which causes a fall in production, output, employment, and prices.

    Your explanation is somewhat similar, that a jump in money demand caused NGDP expectations to fall which caused asset prices to fall, ergo the crisis. For me the problem is that this step-by-step explanation is at a vast macro level, in which you have abstracted the actions of millions of market participants into a vast… abstraction. I can’t visualize whats happening on the ground, on the micro level, at the level of actual persons working in their offices in front of their computers.

    Could you possibly reformulate, if you have time of course, this same step by step explanation, from the jump in money demand to fall in asset prices, but what individuals were doing “on the ground.” What were all those finance people thinking, doing, talking, physically with each other up to the point that Lehman fell?

    The difficulty for me with macro is that everything is explained at such an abstracted distance that I often can’t actually visualize what physical people are actually doing. I suppose, as an example, how you would explain this to freshman econ students…. granted, I know a little bit more than they would. I just think this would be very useful for many readers who don’t comment.

    If this would be a possibility I would be very much obliged.



  18. Gravatar of JimP JimP
    2. August 2010 at 18:48

    Looks like the Fed is going to do just as little little little as possible.


  19. Gravatar of scott sumner scott sumner
    3. August 2010 at 04:35

    Doc Merlin, Yes, I think eventually NGDP targeting will have to be sold as nominal income targeting.

    Benjamin, Thanks.

    J Salvati, I agree.

    D.F. Linton, No, I don’t want the Fed to focus on trade, or any other component of NGDP, just the total. A drop in exports may or may not affect NGDP, it depends what happens to other components.

    AS for why do I want to target NGDP, that is a complicated issue. If you Google the two money illusion posts with Selgin in the title, there is a discussion of the pros and cons of targeting NGDP. I believe the more recent one is better.

    The basic idea is that you keep inflation fairly low, and also stabilize the business cycle.

    MW, Yes, but even if we lack a NGDP futures market, we can use markets like TIPS to construct forecasts. The Fed ignored TIPS market signals in September 2008.

    John Hall, Yes, I have discussed that issue elsewhere. But if you use long dated TIPS that are almost ready to mature, it is not a significant problem.

    Liberal Roman, That’s why nominal income may be a better target. The public understands that there is too little income. I find that most people don’t even understand the definition of the term ‘inflation.’ I often ask my class “Suppose all incomes and all prices rise by 10%, has the cost of living actually increased?” Most say no, although of course it has increased 10%. People wrongly think of inflation as a bad thing that reduces their standard of living, and not as something that also boosts incomes.

    Morgan, You said;

    “You and I are on the same side!” That has me worried! 🙂

    azmyth, Good point. But small countries are exposed to swings in international demand for goods, much more than the US. So they have a hard time stabilizing output even if they stabilize NGDP.

    Manny C, Not a problem as long as the revisions are unpredictable (in terms of direction.) If they are not predictable, then the initial estimate is an efficient forecast of the final number, which is what you want.

    Benjamin, Yes, I’ve seen that argument by Blanchard. It is basically a way of avoiding liquidity traps.

    Morgan#2, I don’t see housing as doomed to fall.

    Morgan#3, Last year I pointed to the rents problem. It was obvious to me that inflation was being greatly overstated in 2008-09, which meant that later it would be understated. We may be approaching that understated point due to the lagged fall off in rents. The government does a horrible job measuring the cost of housing.

    Joe, The reason we are in this mess in the first place is that people assume there are easy micro analogies for macro shocks. But there aren’t. Most people can visualize changes in interest rates, hence that is the explanation usually offered. Unfortunately interest rates don’t play much of a causal role, they respond to macro changes.

    The closet micro analogy I can give you is Keynes concept of “confidence” which I interpret as expectations of NGDP growth one, two and three years out. Of course businessmen don’t literally think in terms of NGDP, but there vague concept of how “business” is expected to be doing next year, or how the “economy” is expected to be doing, is basically about NGDP.

    Here’s where micro analogies break down. When the Fed increases the money supply permanently, expected future NGDP will rise due to the “hot potato” concept. People don’t want to sit on a lot of cash forever. But then the fallacy of composition comes into play, what is true for the individual is not true for the group. As individuals we think we can get rid of excess cash holdings, and in a sense we can. But the group cannot, as the money is just passed from one person to another. So it the individual level the attempt to get rid of excess cash seems unimportant, we just swap if for some other good, service or asset. But as the group level if we are all trying to get rid of extra cash, it drive up AD and NGDP.

    To summarize:

    1. A permanent increase in money tends to raise future expected NGDP due to the “hot potato” process.

    2. Higher future expected NGDP means more business confidence, higher current asset prices and more current investment.

    The exact opposite occurs if there is an increase in money demand, not supply. Now people try to accumulate more money. This is what happened in late 2008. Even though the Fed did supply more, it was all demanded, and then some, partly because of fear, partly because of interest on reserves.

    Both processes seem mysterious because they are based on expectations. The public and investors is looking at a complicated picture, and trying to forecast NGDP growth when we don’t even know exactly what the Fed’s future plans are.

    JimP, Thanks, I may do a post.

  20. Gravatar of MW MW
    3. August 2010 at 05:43

    Scott, I think you mean an “unbiased” forecast, not an “efficient” forecast (assuming you’re using the term in a statistical sense)..?

  21. Gravatar of JimP JimP
    3. August 2010 at 06:49

    I must say – articles like this fill me with shame and horror – that something like this could be and is going on right now – to millions of people.


    I look at Bernanke and Obama about this.

    Especially at Obama. Bernanke is a Republican – so what would you expect?

    But Obama? He apparently could care less.

    He is a DINO.

    A Democrat in Name Only.

  22. Gravatar of B.B. B.B.
    3. August 2010 at 07:00

    Good post.

    First, let’s lobby for more money spent on measuring GDP more accurately and more quickly. Bad data hinder the Fed. For example, in October 1990 (!) Alan Greenspan said the US was not in recession because the data said it wasn’t. Ooppps. As a result, Greenspan did not ease fast enough. It happened again in 2008. Data was flawed.

    That is the problem with targeting nominal GDP. It keeps getting revised under you.

    Second, the Fed can target instead of GDP a set of variables that have few revisions. For example, at the not seasonally adjusted level, revisions tend to be small for the unemployment rate, consumer price index, producer price index, median home sale price, housing starts, electrical power production, steel production, and money/credit measures. And of course financial variables like stock and bond prices and commodity prices are never revised.

    May I suggest your next line of research. Connect optimal Fed policy to a set of indicators that are reliable: timely availability, small revisions, stable and strong correlations with ultimate goals.

    Milton Friedman thought M2 fulfilled these criteria. I think he was wrong, not historically but in this new age. Let’s find the new M2.

  23. Gravatar of JimP JimP
    3. August 2010 at 08:11

    The FT – on IOR. Scott is linked.


  24. Gravatar of Joe Joe
    3. August 2010 at 11:04

    Professor Sumner,

    Thanks for the great response.

    My interpretation is that your explanation of the “hot potato” process is effectively what Mishkin covers in his chapter of “Monetary Transmission Mechanism.” He’s just going into much greater detail of how monetary policy boosts AD, NGDP, and rGDP.

    But if I understand you correctly, you’re arguing that the typical mechanisms are incomplete. You’re saying that there’s an additional mechanism thats even more important than what Mishkin describes. You’re saying that the mere “expectation of higher future NGDP” in of itself has the power to boost AD, NGDP, and rGDP.

    If what I wrote is correct, then I think I finally get it.


  25. Gravatar of Morgan Warstler Morgan Warstler
    3. August 2010 at 11:31

    Denninger helps here:


  26. Gravatar of Benjamin Cole Benjamin Cole
    3. August 2010 at 11:53

    The cover of today’s WSJ (for those of you who still read paper newspapers) is on the FED and QE.
    The battle has been joined, on favorable terms I would say. People are talking about QE!
    If I understand Sumnerian Physics correctly, all we need from here is for the Fed to strongly signal it wants NGDP up by five percent, and that the Fed will engage in QE and other related measures to reach that goal.

    Scott Sumner–It is difficult, but not impossible, to place and editorial in the WSJ. Have you ever tried?

  27. Gravatar of Silas against spending 1 « Good Morning, Economics Silas against spending 1 « Good Morning, Economics
    3. August 2010 at 14:39

    […] at explaining the most important way that monetary disequilibrium affects economic activity (link) When the Fed increases the money supply permanently, expected future NGDP will rise due to the […]

  28. Gravatar of Doc Merlin Doc Merlin
    3. August 2010 at 14:50

    “Milton Friedman thought M2 fulfilled these criteria. I think he was wrong, not historically but in this new age. Let’s find the new M2.”

    Because of Goodhart’s law, once you target something, the target stops working for what you want it to do. So, while I think NGDP targeting will help us, I also think that NGDP will become less useful after we target it, and regulatory policy more so.

  29. Gravatar of Bonnie Bonnie
    4. August 2010 at 02:45

    It might be possible that the FOMC had no clue just how far the rabbit hole went when Bernanke was out giving speeches in 3Q 2007 about fighting inflation with a triamphant grin from ear to ear. That doesn’t explain not paying attention when things started getting really bad and it looked like there was no bottom, however. Perhaps Prof Sumner is giving the Fed the benefit of the doubt when he discusses the “blindness” theory, but I don’t buy it because I saw the whole thing unfolding like a slow motion trainwreck from numbers I pulled out the Fed’s own Flow of Funds reports combined with CPI, wages and commodities data.

    It also doesn’t explain the reluctance to ease when Bernanke was answering questions Monday (broadcast on FBN, but I couldn’t find it in the CSPAN archive) he admitted that the Fed didn’t stabilize a sharp decline in NGDP in 2008 and faliure to do so made things markedly worse (Sumner is vindicated — and I’m surprised he hasn’t linked a youtube of it). There’s also no explanation for the apparent ~1% inflation rate target for here and now. Why not zero or some fraction of negative instead if there is no difference between fractions of a percent or even whole ones for that matter? Obviously there is a difference in the ability to sustain economic activity at a reasonable pace.

    I don’t understand why this whole thing isn’t some huge scandal and our political system isn’t out for Federal Reserve heads.

  30. Gravatar of Doc Merlin Doc Merlin
    4. August 2010 at 23:15

    ‘That recovery had 11% NGDP growth. When was the last time you saw an economy growing at 8% in a period of 4% deflation? Never? There’s a reason for that, wages and prices aren’t nearly flexible enough to overcome that sort of nominal sluggishness.’

    Even if wages and prices were perfectly and instantly flexible. The existence of nominal contracts over time would be enough to add significant stickiness to the system.

  31. Gravatar of scott sumner scott sumner
    6. August 2010 at 06:16

    MW, I forgot the context, but I probably meant “optimal” in a Ratex sense.

    JimP, Obama may be more ignorant than evil.

    B.B. You said:

    “That is the problem with targeting nominal GDP. It keeps getting revised under you.”

    And that’s exactly why I favor targeting NGDP forecasts, not actual NGDP.

    NGDP expectations are the new M2.

    JimP, I’m glad the FT is now paying attention.

    Yes Joe, that is what I am saying.

    Morgan, Thanks for the chart—quarterly data would look even more dramatic.

    Benjamin, Yes, I am trying.

    Doc Merlin, There’s a huge difference between M2 and NGDP. NGDP is the goal variable, M2 is not. So I don’t care how NGDP correlates with other variables.

    Bonnie, I actually agree. My point is that the data they rely on was flawed. But they should be using forward looking data like TIPS spreads, so I still think they had plenty of data to avoid this fiasco.

    Doc Merlin. Debt contracts don’t directly affect unemployment however, they affect the financial system,

Leave a Reply