Updates! Scroll down....

Yesterday, the U.S. Bureau of Economic Analysis released its estimates of state level Gross Domestic Product through 2015-Q4. As it did, the BEA revised its previous quarterly GDP estimates for each state going back to at least 2005-Q1.

In doing that, the BEA's data jocks may have provided an unexpected sneak peek of how its estimates of national GDP for the U.S. will be officially changed when the agency releases is annual revisions for that data during the last week of July 2016.

In the chart below, we show the potential magnitude of the pending revision to real GDP in the U.S. The green line is how the nation's real GDP level is currently being reported by the BEA in the period from 2005-Q1 through 2015-Q4, while the black line shows how the BEA's regional data for the real GDP of the entire United States, spanning all industries, has just been reported.

There appears to be very little difference from previously reported real GDP figures from 2005-Q1 through 2006-Q3, but after that quarter, a widening gap opens up all the way through the end of 2015-Q4, at which point, the U.S.' inflation-adjusted GDP is revised downward by $324 billion (in terms of inflation-adjusted constant 2009 U.S. dollars).

Stated differently, real GDP in the U.S. is 2.0% smaller than what the BEA previously reported it to be. In terms of the percentage change of the magnitude of the revision in U.S. real GDP, the BEA's downward revision in 2016 looks to be more than double the size of 2014's and 2015's downward annual revisions.

About half of the total downward revision is concentrated during the period of 2007 through 2008, during the period leading up to and going into the first year of the Great Recession, with a second downward surge mostly taking place during 2012 and 2013, when the U.S. economy skirted the edge of an official recession in what we've previously described as a microrecession.

It will be interesting to find out if the BEA did indeed provide an unintended advance sneak peek of its upcoming annual GDP revision exercise for 2016.

Update 8:56 AM EDT: A potential, but partial explanation:

Annual Revision of Gross Domestic Product by State

In conjunction with today's release of fourth quarter 2015 gross domestic product (GDP) by state statistics, BEA is making available on its Web site revised statistics for 2008–2014 and annual statistics for 2015. The annual revision of GDP by state incorporates newly available and revised state source data, most notably the 2012 Economic Census for industries (except for manufacturing, which was previously incorporated).

Additional information on this revision will be available in an article in the July 2016 issue of the Survey of Current Business.

So yes, there is a substantial revision coming down the pike for national GDP, where the regional accounts data is indeed giving us an early sneak peek of the revisions that will be made.

Meanwhile, there is a difference between the reported national level GDP (the green line in the chart above) and the just reported regional GDP rollup for the United States (the black line in the chart above):

GDP by state for the U.S. differs from the GDP in the national income and product accounts (NIPAs) and thus from the Industry Economic Accounts' GDP by industry, because GDP by state for the U.S. excludes federal military and civilian activity located overseas, which cannot be attributed to a particular state.

That doesn't fully explain what we see in the data, because the contribution of federal military and civilian economic activity overseas to U.S. GDP is not something that just started happening beginning with 2006-Q4 - otherwise, we would see much larger differences between the reported national level real GDP and the regional rollup of state level real GDP in the period from 2005-Q1 up to that quarter.

A potential explanation for why that might be more of a factor after 2006 is the Iraq War surge, which became U.S. policy in January 2007. To the extent that U.S. military contractors added to the national level GDP, but not state level GDP during that heightened period of activity would potentially explain a portion of the difference in the period preceding 2008.

But again, military contractors were reaping the benefits of their contracts for work overseas before the surge as well. We should be seeing a much greater difference between national and regional rollup data for GDP in the period from 2005-Q1 to 2006-Q3 than what we are for that factor to fully explain what we see in the data.

On the other hand, when the national level data goes through its annual revision in late July 2016, we'll have a way to quantify how much that kind of economic activity is boosting U.S. GDP.

Update 16 June 2016: We were contacted by one of our readers who pays close attention to the BEA's revisions of GDP. Apparently, the BEA's plans for the revision of the national level GDP data will only cover the period from 2013-Q1 through 2016-Q1.

That's really weird. It's a lot harder for the BEA to disentangle the state level contributions by industry from their national level source data, but once it's done, it means that they've also updated and revised the national level data per all the most recently available information. It just doesn't make sense to sit on that much revised data and not update the national level figures to reflect the entire scope of all the revision work that has been done.

What that means is that there will very likely be an ongoing discrepancy between the state-level GDP and its rollup to the national level, and the BEA's reported national level data. The two datasets should largely match, except for that contribution by U.S. contractors who support U.S. military operations overseas, which falls outside the economic activity that occurs within the 50 states and the District of Columbia as noted by the BEA, and which should only represent a very small fractional contribution to the national GDP figures above the aggregate rollup of state level GDP for the 50 states and Washington DC.

Update 18 June 2016, 6:20 PM EDT: We've been working behind the scenes with other competent econobloggers to sort out just how big the BEA's upcoming national level GDP revision will be when it is finally posted in the last week of July 2016. In doing that, we've estimated just how big the contribution to national GDP is for federal military and civilian activity located overseas, based on the available pre-revision data we have to estimate that figure through 2015-Q3. The chart below shows that contribution and updates our findings:

We project that the maximum "likely" size of the upcoming revision for national GDP is -1.4%. We expect it will actually be less than that figure because the upcoming national GDP revision will only cover the period from 2013-Q1 through 2016-Q1, which will therefore not fully address the discrepancy that begins appearing after 2012-Q2.

It is also possible that the federal military and civilian activity located overseas will turn out to be significantly larger than what the BEA's data has previously indicated, which would also narrow the gap between how national GDP is currently being reported and what the just-revised aggregate state GDP suggests it will turn out to be.

This is really pretty cool because this is the first time that this kind of predictive exercise has even been possible thanks to the BEA's more timely reporting of its regional GDP data series. As our long time readers know, we're very happy whenever we're both on the cutting edge of doing new analysis and can share our work in progress in developing a brand new kind of analysis to get to the real story behind the numbers in the economy.

Update 27 July 2016: This is so cool! We had an extended contact today with an analyst who works for the BEA, who wanted to know more about how we came up with our estimate of how real GDP would change using the BEA's recently revised regional account data.

That was fantastic timing, because we just happened to have our spreadsheet open because we were updating it with state level GDP data from 2016-Q1 that was just released today.

In going over that material, we discovered a problem that threw off our calculations, with the effect that the results we had obtained weren't matching what they were getting from 2014-Q3 onward in replicating our analysis. We were able to trace the problem back to the GDP deflator that we used to convert nominal GDP data to inflation-adjusted "real" data, where we were multiplying the nominal data by the GDP deflator data for the national level data (the data whose revision is set to be published on Friday, 29 July 2016) instead of the GDP deflator data that applies for the aggregate 50 states plus Washington DC, which just happened to be in the spreadsheet column next to it.

After we made the appropriate correction, all our results from 2014-Q3 onward snapped into place and all results matched, from the period from 2005-Q1 through 2014-Q2, where there was never any problem, and now from 2014-Q3 through 2015-Q4! The following chart shows the latest and greatest for what we expect from Friday's GDP revision based only on the state level GDP revision from June!

As for the outcome of the analysis, the error we made with using the incorrect GDP deflator overstated the amount by which real GDP is likely to be revised by 0.5% of GDP, so instead of the maximum likely change of -1.4% that we had previously calculated, the maximum amount by which national GDP might change as a result of the revisions the BEA has made to its GDP data for the 50 states plus Washington DC is -0.9%, with the maximum discrepancy now taking place in 2013-Q3. This data is indicated in the chart above by the dark-green line.

But since the BEA's national revision will cover the period from 2013-Q1 to the present, the revision of the national level GDP will not include the now-confirmed $55 billion discrepancy that opens up between the national GDP data and the national aggregate state level GDP in 2012. We are therefore now estimating that the most likely revision that will be made to the national GDP data on Friday, 29 July 2016 will be an adjustment of -0.6% in 2013-Q3. This data is indicated in the chart above by the bright red line (and the blue arrow at 2013-Q3).

We'd like to thank everyone who provided their useful assistance in getting to this point - in terms of collaborative effort, this has been one of the biggest projects we've had the pleasure of working on since launching Political Calculations. It's always exciting when we get to break brand new ground and do analysis that was never possible before, and we appreciate your shared enthusiasm!

On a final note, if you happened to have come across this post by way of Econbrowser, you might want to pass along information back to the author who pointed you in this direction that their observations have been described as "not relevant". We're pretty sure that particular author hears that a lot....

There's more that we'd like to be able to discuss on the topic of the upcoming national-level GDP revision, but that will have to wait until after that data is released. Until then, what you see above represents the most that anyone can reasonably glean about what the revision will look like based only on data that is already available to the public.

Update 29 July 2016: The verdict is in! The chart below shows the revisions in national level real GDP from 2013-Q1 through 2015-Q4....

Compared to our final pre-revision prediction from 27 July 2016, we were off by 0.1% of GDP through 2015-Q4, where we had previously projected no change.

More significantly however, national-level real GDP was increased by 0.2% of its previously reported figure in 2013-Q3, where the aggregate GDP data for the 50 states and the District of Columbia had instead indicated that a 0.6% of GDP decline for that quarter was most likely. That means that the amount of GDP that the U.S. generated in that quarter from Overseas Federal Military and Civilian Government Activities was revised to be significantly much higher than the BEA had previously indicated, in effect, adding an additional 0.8% of GDP on top of what the BEA now indicates was generated within the actual territory of the United States.

We'll dig deeper into the national-level GDP revision periodically over the next several weeks!

Update 2 August 2016: Added boldface type to selected text above. Somehow, one particular critic is so committed to a non-relevant narrative that they chose to completely ignore the now emphasized text where we were happy to describe the actual error we made, just so they could continue their pursuit of their preferred narrative.

Given their history, that is perhaps not so surprising. In fact, in their obsessive pursuit, they have even managed to check off two new categories from the "How to Detect Junk Science" checklist.

If you can't tell from our discussion above, we were quite happy to both challenge our assumptions and to share our data and calculations with others where they could be reviewed, which was a key part of what led to our making real progress in understanding what the BEA's 2016 GDP data revisions are really communicating. More on that in a new post, soon!