EconoMonitor

Tales from the GDP Revisions

The 2011Q2 advance release and revised estimates [0] contained many unpleasant surprises (see Jim’s assessment; also [CR1] and [CR2] [John Taylor] [Izzo/WSJ RTE]). The below consensus growth rate, and downward revision in Q1 growth, have been discussed elsewhere. I want to focus on the implications of the revisions to the data going back to 2003 (with particular emphasis on data back to 2007).

talesrev1.gif
Figure 1: GDP from 2011Q2 advance release (blue) and from 2011Q1 3rd release (red), all in billions of 2005$, SAAR. NBER defined recession dates shaded gray. Source: BEA.
Here are the points I take away from this new information:

 

  • The output decline starting from 2008Q2 was much worse than originally estimated.
  • The increase in GDP relative to 2009Q1 (when the ARRA was passed) is the same as before.
  • The output gap is more negative than originally estimated (based on January 2011 CBO estimate of potential), and is now getting more negative.
  • Consumption and fixed nonresidential investment account for the largest components of the downward revisions in final sales of domestic product.
  • GDP per capita fell even more precipitously from 2008Q2 through 2009Q1 than originally estimated.
  • The consumption decline is also more pronounced with new data.
  • The recent employment-GDP relationship seems a little less mysterious.
  • The exports story remains unrevised.

The output decline starting from 2008Q2 was much worse than originally estimated

talesrev2.gif
Figure 2: Log GDP from 2011Q2 advance release (blue) and from 2011Q1 3rd release (red), all in billions of 2005$, SAAR, rescaled to 2008Q2 normalized to zero. NBER defined recession dates shaded gray. Source: BEA, NBER and author’s calculations.

Originally, GDP fell at a 7% SAAR rate in 2008Q4 (the quarter that Don Luskin conjectured in September 2008 might be the start of unprecedented prosperity); in the revised figures, the decline was 9.3% (rates calculated in log terms; 6.7% and 8.9% in levels terms). As an interesting aside, the advance estimate (released at the end of January 2009) for 2008Q4 growth was only -4.1% (-4.0% level terms). The 2008Q4 q/q ar change is now only rivaled by the 1958Q1 q/q change (-11.1%).

The increase in GDP relative to 2009Q1 (when the ARRA was passed) is the same as before

talesrev3.gif
Figure 3: Log GDP from 2011Q2 advance release (blue) and from 2011Q1 3rd release (red), all in billions of 2005$, SAAR, rescaled to 2009Q1 normalized to zero. NBER defined recession dates shaded gray. Source: BEA, NBER and author’s calculations.

In making this assessment, it doesn’t matter if one normalizes on 2009Q1 or 2009Q2, as most of the negative growth took place before 2009Q1. In other words, the revisions do not alter the estimated growth trajectory of the economy. If one considers the effect of the stimulus as relative to where the economy started (recalling the ARRA only really started kicking in in 2009Q2), then one’s views should not be altered on the efficacy count. On the other hand, it is even more clear now that the size of the stimulus was wholly inadequate, as highlighted in the next graph.

The output gap is more negative than originally estimated (based on January 2011 CBO estimate of potential), and is now getting more negative

talesrev4.gif
Figure 4: Log output gap from 2011Q2 advance release (blue) and from 2011Q1 3rd release (red), using CBO January 2011 version of potential GDP. NBER defined recession dates shaded gray. Source: BEA, CBO, NBER and author’s calculations.

Figure 4 is based on the CBO’s January 2011 estimates of potential GDP. Presumably, with additional information released by the BEA, the estimate of potential GDP will be revised downward over the last four years (the revisions are not substantial prior to 2007). Nonetheless, the implied output gap, at -6.9% in 2011Q1 versus the previously implied -5.3%, is daunting. The graph also highlights the fact that the output gap is widening, as fiscal stimulus is withdrawn, and the economy continues to encounter headwinds from high energy costs, faltering foreign demand for US goods, and uncertainty regarding future demand and fiscal policy.

It strikes me that this seems an odd time to embark upon fiscal consolidation that is front-loaded — i.e., has cutbacks in current or near-horizon spending. Certainly, this widening gap argues against tightening monetary policy, given the large amount of slack in the economy precludes a rapid acceleration in inflation.

Consumption and fixed nonresidential investment account for the largest components of the downward revisions in final sales of domestic product

talesrev5.png
Figure 5: Difference between 2011Q2 advance and revision and 2011Q1 3rd release for real final sales. CDIF is consumption difference, IFIXDIF is fixed investment difference, GDIF is government spending difference, and NXDIF is net exports difference, all in billions of Ch.2005$. Source: BEA and author’s calculations.

The blue shaded areas highlight the large downward revisions in consumption, particularly in 2009, but also going through 2011Q1. The fixed investment revisions are almost entirely accounted for by fixed nonresidential – not a encouraging sign. While consumption, and to a lesser extent fixed investment, are the most important, in the preceding three quarters, government spending on goods and services accounted for large shares as well.

GDP per capita fell even more precipitously from 2008Q2 through 2009Q1 than originally estimated

talesrev6.gif
Figure 6: Per capita GDP from 2011Q2 advance release (blue) and from 2011Q1 3rd release (red), all in billions of 2005$, SAAR, divided by population at midmonth (FREDII series POPTHM; June 2011 estimated using log first differences over 2009M01-11M05). NBER defined recession dates shaded gray. Vertical dashed lines at 2001Q1, 2009Q1. Source: BEA, FREDII, NBER, and author’s calculations.

Not only has real GDP, using revised data, not reattained pre-recession peak levels, per capita real GDP also remains below pre-recession levels.

The consumption decline is also more pronounced with new data

talesrev7.gif
Figure 7: Per capita consumption from 2011Q2 advance release (blue) and from 2011Q1 3rd release (red), all in billions of 2005$, SAAR, divided by population at midmonth (FREDII series POPTHM; June 2011 estimated using log first differences over 2009M01-11M05). NBER defined recession dates shaded gray. Source: BEA, FREDII, NBER, and author’s calculations.

So here we have vividly illustrated the “consumption disaster” Professor Mulligan dismissed back in 2009. Not only is consumption flat in aggregate terms, in per capita terms it has turned downward in the latest quarter (advance) estimate.

Okun’s Law Re-estimated

The revisions have some implications for Okun’s law. Since there are a variety of Okun’s laws — some relate to levels (or more specifically gaps) and some relate to growth rates — one has to be specific. Below, I present a scatterplot of GDP growth rates from the latest and pre-revision series, against the growth rate of private employment.

talesrev8.gif
Figure 8: Four quarter growth rates in GDP from 2011Q2 advance release and revision (blue circle), and from 2011Q1 3rd release (red +) against four quarter private employment growth, 1986Q1-2011Q2. Source: BEA, BLS via FREDII (June release), and author’s calculations.

talesrev9.gif
Figure 9: Quarter on quarter annualized growth rates in GDP from 2011Q2 advance release and revision (blue circle), and from 2011Q1 3rd release (red +) against four quarter private employment growth, 1986Q1-2011Q2. Source: BEA, BLS via FREDII (June release), and author’s calculations.

The slope coefficient of a regression of employment growth on GDP growth is actually lower with the revised series over the 2009Q2-11Q1 period, which seems at odds with the perception that employment growth has been unnaturally low; this can be reconciled by the fact that the constant in such a regression is more negative.

Formal statistical tests indicate that the relationship between GDP growth and private employment growth has experienced breaks at each of the past two recessions, and strongly rejected stability (using 1-step-ahead recursive residual tests) in the last recession. This remains true regardless of whether one uses the pre-revision or post-revision data, although the rejection of stability is slightly less marked using the revised data.

To the extent that the Okun’s Law in growth rates is of interest, the mystery is not why current (post-recession) employment growth is so slow (given GDP growth), but rather why it was so low at the end of the recession (09Q1, 09Q2). Of course, this translates into a lower level of employment for given level of GDP at the moment, given that there was no underprediction of employment growth subsequent to the recession (as one might have expected on the basis of past experience, particularly after the 1990-91 recession).

Good News: Exports Unrevised

Figure 5 highlights that in recent quarters, the external sector has in a mechanical sense not been a big contributor to GDP growth. However, the diminished positive impact associated with the external sector over the 2010Q4 and 2011Q1 quarters has been due to upwardly revised import growth. Real exports were essentially unrevised.

talesrev10.gif
Figure 10: Log exports of goods and services from 2011Q2 advance release (blue) and from 2011Q1 3rd release (red), all in billions of 2005$, SAAR. NBER defined recession dates shaded gray. Vertical dashed lines at 2001Q1, 2009Q1. Source: BEA, FREDII, NBER, and author’s calculations.

Real exports are 22.4% above 2009Q1 levels, and nominal exports are 37.7% (20.2% and 32.0% in log terms). Whether export growth can continue at this pace given uncertainty in Europe and a slowing Chinese economy remains to be seen.

This post originally appeared at Econbrowser and is reproduced here with permission.

9 Responses to “Tales from the GDP Revisions”

Intelligent CommentaryAugust 8th, 2011 at 6:07 pm

[...] “output gap.” Menzie Chinn, an economics professor at the University of Wisconsin, has estimated that the economy was about 7 percent smaller than its potential at the beginning of this year. [...]

Most Read | Featured | Popular

Blogger Spotlight

Dan Steinbock

Dr Dan Steinbock is a recognized expert of the multipolar world. He focuses on international business, international relations, investment and risk among the major advanced economies (G7) and large emerging economies (BRICS and beyond).

Economics Blog Aggregator

Our favorite economics blogs aggregated.