Economic recovery and the perverse math of GDP reporting

Now that everyone is talking about green shoots and the potential for economic recovery, I thought I would run through the statistics of U.S. GDP with you.  The reason I am bringing this up is that there is a lot of confusion about what recovery means and positive GDP growth mean.  So, I am going to spell it out in a bit more detail here.  I will start with recession, which will lead into recovery – the thing we are all interested in right now.  Afterward, I will briefly sketch out what GDP is and throw some numbers up to illustrate a few points before wrapping up with a conclusion.

What is a Recession

In the United States, the official dates for a recession are determined by the National Bureau of Economic Research (NBER).  Wikipedia has a useful definition of the organization:

The NBER was founded in 1920. Its first staff economist and Director of Research was Wesley Mitchell. Simon Kuznets was working at the NBER when the U.S. government asked him to help organize a system of national accounts in 1930, which was the beginning of the official measurement of GDP and other related indices of economic activity. Due to its work on national accounts and business cycles, the NBER is well-known for providing start and end dates for recessions in the United States.

The NBER is the largest economics research organization in the United States[2]. Sixteen of the thirty-one American winners of the Nobel Prize in Economics have been NBER associates, as well as three of the past Chairmen of the Council of Economic Advisers, including the former NBER president,Martin Feldstein. NBER research is published by the University of Chicago Press.

Now, the Business Cycle Dating Committee at the NBER is the group that decides when a recession has occurred, usually with a significant lag (the recession starting in December 2007 wasn’t called until December 2008). As far as I know, the Dating Committee is really a virtual group of economists which don’t actually meet, but rather discuss things via e-mail, telephone, what have you in an ad hoc and informal fashion before arriving at the critical moment when they must come together and date the business cycle.  On their website there is a page called, “Business Cycle Expansions and Contractions” which has a good note in the footer about what recessions actually are. It reads:

The NBER does not define a recession in terms of two consecutive quarters of decline in real GDP. Rather, a recession is a significant decline in economic activity spread across the economy, lasting more than a few months, normally visible in real GDP, real income, employment, industrial production, and wholesale-retail sales. For more information, see the latest announcement from the NBER’s Business Cycle Dating Committee, dated 12/01/08.

I hope that clears up the fallacy of the two quarters’ decline in real GDP, a shorthand that has developed but is inaccurate.  So, to sum it up, in the U.S., the NBER has the final say regarding determining business cycles.  They are looking at weakness in five key areas – output, income, employment, production and sales in order to date a recession.


When it comes to recovery, the NBER are using the exact same methodology in reverse.  Their statement regarding the last recovery in 2001 is telling regarding how this process works.

The committee’s goal is to determine the dates of peaks and troughs as definitively as possible. The committee is careful to avoid premature judgments. The initial announcements of many of the major indicators of economic activity are preliminary and subject to substantial revisions, so it is not possible to identify the month of a peak or trough rapidly. The main reason that the committee’s decision in this episode was particularly difficult was the divergent behavior of employment. The committee felt that it was important to wait until real GDP was substantially above its pre-recession peak before determining that a trough had occurred, and to study the NBER’s past practices carefully to ensure that its decision in this episode was consistent with the dating of earlier turning points.

What is interesting to note is that at least one member of the Dating Committee, Robert Gordon, a Professor of Economics at Northwestern University, is showing his cards already.  Last week, I posted an article he wrote suggesting that the recession is about to end (See my post “Jobless claims may signal the end is near”).  Apparently, he thinks recovery may be coming.

You should notice that the NBER uses the term ‘peak’ to describe the onset of a recession and ‘trough’ to describe the onset of recovery.  So, really recession is the phase after the business cycle has peaked.  It ends and we get a recovery when the business cycle has troughed.  That’s the terminology.


This terminology makes a good entree into GDP and what it is.  Going back to Wikipedia, GDP is described as follows:

The gross domestic product (GDP) or gross domestic income (GDI), a basic measure of an economy’s economic performance, is the market value of all final goods and services produced within the borders of a nation in a year. [1] GDP can be defined in three ways, all of which are conceptually identical. First, it is equal to the total expenditures for all final goods and services produced within the country in a stipulated period of time (usually a 365-day year). Second, it is equal to the sum of the value added at every stage of production (the intermediate stages) by all the industries within a country, plus taxes less subsidies on products, in the period. Third, it is equal to the sum of the income generated by production in the country in the period—that is,compensation of employees, taxes on production and imports less subsidies, and gross operating surplus (or profits).[2] [3]

What you should notice in the definition is that GDP is a measure of “goods and services produced” not sold. And this is something I will get back to later.  What the statisticians are trying to do is value everything that is produced in the United States domestically and sum it up as an aggregate number.  That is all GDP really is.

GDP as reported

From my view point point, the interesting bit about GDP is NOT inflation (i.e. real vs. nominal GDP), but rather the fact that the number which is reported is a first derivative.  It is the change in GDP which is reported, not the actual number.  And, this is significant because generally a business cycle troughs when the change in GDP goes from being negative for a significant period to being positive for a significant period. Equally, a business cycle peak (read recession) occurs when the change in GDP goes from being positive for a significant period to being negative for a significant period. So, it is the change in GDP that everyone cares about.  But, reporting the change brings in a few statistical anomalies that are important.

Negative numbers are bigger than positive numbers

The first problem with measuring a first derivative is that negative percentage changes have statistically greater effect than positive percentage changes.  Let me give you an example.  Say GDP starts at 100 and it drops 20% to 80, if it rises 20%, you get back to 96, not 100.  The 20% down move is the equivalent of a 25% up move.  So, in a country like Latvia, where GDP is down 20% annualized, you need an even large push to the upside of 25% to get back to even.  Net, net, this means recovery has to be either stronger or longer than the recession to get back to the pre-recession level of production.

Subtracting less negative means positive

Thinking about production as opposed to sales again, you have to look at inventories.  The NBER is not fooled by inventory builds because they look at both industrial production and retail sales.  But, since GDP is a pure production statistic, inventory builds distort the picture.  For example, say your economy produces $980 worth of stuff one quarter that gets sold. But it also sells a lot of stuff, $20 worth, out of inventory.  If next quarter, you need to sell just as much stuff ($1000), guess what, GDP growth goes up automatically (Remember, we are not talking about GDP, but GDP growth). The inventory purge means you are producing less to meet demand than you would otherwise need to. So, when comparing one quarter to the next, unless you purge just as much stuff or unless demand goes down, you need to produce more. Therefore, you get an automatic uptick in GDP growth.

in my post “GDP: 4th quarter 2008 was worse,” I said this about Q1 GDP:

clearly the consensus was much too bullish as I predicted. And the main culprit was inventories, which were purged at a $136.8 billion annual rate. That is enormous. In fact, Q1 2009 saw the largest inventory purge ever. As a percent of GDP, you have to go back to Q4 1982 to get a more liquidationist reading.

Now, if you think about this in the context of what I just presented, it makes it pretty clear that you don’t need to build inventories to get an uptick in GDP growth.  All you need to do is purge fewer inventories.  Subtracting a less negative number is the same as adding a positive number.  And I doubt we will be purging an annualized $136.8 billion in inventories going forward.

Recovery does not mean recovery

My final thought on the statistics here has to do with starting from a lower base.  Before the Great Depression in 1929, the U.S. had nominal GDP of $103.6 billion.  By 1933, this had dropped to $56.4 billion due to deflation and a decrease in production.  Over the next four year, GDP growth soared. It was 17.0% in 1934,11.1% in 1935, 14.3% in 1936 and 9.7% in 1937.  That’s some serious growth, right?  Well, GDP was only $91.9 billion in 1937, a full 11.3% lower than it had been 8 years earlier. In fact, it wasn’t until 1941 that we attained the nominal production output of 1929.

What this should illustrate is that working from a lower base makes an increase in GDP comparatively easier than when working from a higher base.  Yes, it was a powerful recovery, but it did not get the United States back to the same productive level for many years.  In that sense, recovery does not mean recovery immediately.


Having run through this, I have a few thoughts.

  1. The massive inventory purge we just witnessed is more significant than people realize.  The numbers I ran through should help demonstrate this. Would it be wild to think Q3 could see a positive GDP growth number?  Perhaps, especially with an idled auto sector. Nevertheless, we should think outside the box here.
  2. The same logic for inventories (i.e. subtracting a less negative number is the same as adding a positive) is at play with imports – but in the opposite direction.  As recession took hold, U.S. trade shrank, reducing both imports and exports. But, since the U.S. is a debtor nation, the shrinkage in imports has been much larger than the shrinkage in exports (why else do you think we saw horrific declines in exports from Asia in January and February?).  That has been a plus for statistical GDP growth. If we do get a recovery, unless people save much more, this same dynamic in reverse will be a drag on GDP growth.
  3. It should be pretty clear now that just because we get a recovery does not mean the United States is going back to where it was.  After all, the financial services sector is still fairly sick and it comprised an outsized percentage of GDP. So, unless Geithner, Summers and Bernanke can reflate the sector as they seem to be trying to do, the U.S. is going to need to make up the slack from somewhere else.  This suggests that the recovery will be weak.

So, most of this suggests an early but weak recovery.  That’s pretty much my baseline view too. I would appreciate your thoughts on this as well.

NBER – Wikipedia
Business Cycle Expansions and Contractions – NBER website
July 2003 announcement of business cycle trough – NBER website
Determination of the December 2007 Peak in Economic Activity – NBER website

  1. Tim Jones says

    One definitive lesson we learned in the Great Depression (GD) was that when reality set in, the mainstream just started to ignore the problem even existed. I think the lesson this time will be similar in that the most likely scenario will be that we are “recovered” even as the economy will continue to give up >= 400K jobs per month….

    Focusing on financial recovery and GDP data I think obscures the real issue which is that we most likely are back to is the same question we asked in the 1960’s… Which is… How does a post industrial nation continue to grow when it is too expensive to create anything, and produces limited enhanced productivity gains?? Originally the answer was that globalized consumption / asset based pyramiding approach would work until its industrial peers caught up, but this approach appears to have run its course now. Meaning, private and public debts are already higher than they were in the GD so additional sustainable growth thru debt enhancement is unlikely.

    The only current option left right now is to do mass currency devaluation to allow further expansion by debt devaluing which is a logical short term approach. I think that even the FED would agree, however, that this is a temporary measure only meant to just buy more time… The long term risks of this approach are no doubt many…

  2. rfreud says

    Hi Edward,

    Thanks for this and for many informative and thoughtful posts. I want to take this opportunity to ask for your thoughts on a question or two that have been bothering me.

    Speaking of inventories, we have today a huge overhang in housing inventory. Also, in the past, housing construction jobs have been a key element in a recovery and a target of government stimulus and contra-cyclical monetary policy. I would be interested in your thoughts as to how these sectors nuance your guardedly optimistic view of the GDP statistics. Housing inventory is not, if I am not mistaken, included in the inventory statistics you discuss.

    Secondly, and forgive me for going off topic here although it is a question about economic statistics, I would be interested in any insight you might be able the share about how the secondary market in mortgage backed securities, synthetics, derivatives and swaps impact the money supply and how this is or is not measured. One of my pet theories is that expansion of the money supply through these instruments was not included in the statistics which are used to guide the economy.

    You will get that my theme is how well our economic statistics describe our reality.

    Thanks very much.

    1. Edward Harrison says

      I’m fairly optimistic (for me) on residential housing because we have already seen 30% down on residential fixed investment and it’s back to 1998 levels. Perhaps, we go down another two-three quarters before we trough. CRE is another story though, with investment still at 2005 levels. I reckon we could see those numbers cut in half easy.

      Some of this is going to depend on the shape of the real economy obviously.

      As for the excess credit supplied by all of that stuff, there is a general consensus that the shadow banking system operating in those realms and has been decimated. Part of why the economy is flat on its back has nothing to do with commercial banks and more to do with the deleveraging in shadow banking. The fed has stepped in to fill that void, not to fill a void left by commercial banks.

      So, I think you are dead on regarding the importance of credit supplied as a result of derivative markets.

  3. gaius marius says

    great analysis as usual, mr harrison.

    did you happen to see anything of the standard chartered note on inventories the other day? ft alphaville covered it a bit. expecting worse than q1 in q2 — and i have to say this corroborates well with the gathering decline in consumption, which david rosenberg noted has deepened from january to april, as well as accelerating reductions in loans and leases — but more upbeat about h2.

    there’s been a few anecdotes about order quality as well — that “real” cash orders aren’t really coming in, lots of highly predicated sales and therefore lots of cancellations, making the orders picture (glum as may be) look better than it is. that of course means inventory reductions could surprise negatively. can you provide color on this? is that typical of the turns?

    again — really fine blogging. thanks.

  4. Anonymous says

    We are in one of those odd moments when quarter-to-quarter annualized and year-on-year (both real) reading conflict. The second quarter’s Q-Q of -1% was a great improvement over Q-1, but the YoY -3.9% was worse than January-March.

Comments are closed.

This website uses cookies to improve your experience. We'll assume you're ok with this, but you can opt-out if you wish. Accept Read More