RJS, MarketWatch 666 Here I (run75441) am being lectured by the author, RJS. “last week you asked me to write something explaining how & why the GDP deflators were revised. So I did, expanding the paragraph I was going to write on it to six, and included it as an addendum to my usual reporting on the GDP revision at Marketwatch 666, however, three days after I sent it to you, it’s still not up on AB. I think it’s important, and that it will continue to impact seasonally adjusted data going forward, since I see no easy way that pandemic related impacts could be isolated from other impacts. Even if the agencies attempt to adjust the seasonally adjustments for pandemic impacts, it will be at best an educated fudge. That educated fudge will remain
Topics:
Angry Bear considers the following as important: GDP, Hot Topics, MarketWatch 666, politics, RJS, US/Global Economics
This could be interesting, too:
Peter Radford writes Election: Take Four
Bill Haskell writes Review of the Tax Code and Who Benefited the Most from the Breaks in It
Bill Haskell writes Lawler: Early Read on Existing Home Sales in October
Bill Haskell writes Healthcare Insurance in the United States
RJS, MarketWatch 666
Here I (run75441) am being lectured by the author, RJS. “last week you asked me to write something explaining how & why the GDP deflators were revised. So I did, expanding the paragraph I was going to write on it to six, and included it as an addendum to my usual reporting on the GDP revision at Marketwatch 666, however, three days after I sent it to you, it’s still not up on AB.
I think it’s important, and that it will continue to impact seasonally adjusted data going forward, since I see no easy way that pandemic related impacts could be isolated from other impacts. Even if the agencies attempt to adjust the seasonally adjustments for pandemic impacts, it will be at best an educated fudge. That educated fudge will remain embedded in the data thereafter. I see little mention of this in other reporting, certainly not in the media that just rewrites press releases.
Wolf Richter has mentioned errant seasonal adjustments a couple times but i don’t know that he’s explained it in any detail. Zero Hedge called out the +3.8% January retail sales report for having the “largest upward seasonal adjustment in history” but didn’t say why . . . “
4th Quarter GDP Revised to Show Growth at a 7.0% Rate After Unprecedented Revisions to Component Deflators
The Second Estimate of our 4th Quarter GDP from the Bureau of Economic Analysis indicated that our real output of goods and services grew at a 7.0% rate in the quarter, revised from the 6.9% growth rate reported in the advance estimate last month, as slower growth of personal consumption of services and of exports than was previously estimated was more than offset by greater growth of personal consumption of goods, fixed investment, and a smaller contraction of government. In current dollars, our fourth quarter GDP grew at a 14.64% annual rate, increasing from what would work out to be a $23,202.3 billion a year output rate in the 3rd quarter to a $24,008.5 annual rate in the 4th quarter, with the headline 7.0% annualized rate of increase in real output arrived at after an annualized aggregate inflation adjustment averaging 7.1%, aka the GDP deflator, was computed from the price changes of the components and applied to their current dollar change..The GDP deflator was revised from 6.9% in the advance estimate to 7.1% with this estimate…
Remember that the second estimate GDP news release reports all quarter over quarter percentage changes at an annual rate, which means that they’re expressed as a change typically a bit over 4 times of that what actually occurred over the 3 month period, and that the prefix “real” is used to indicate that each change has been adjusted for inflation using price changes chained from 2012, and then that all percentage changes in this report are calculated from those 2012 dollar figures, which would be better thought of as a quantity indexes than as any reality based dollar amounts. For our purposes, the data that we’ll use in reporting the changes here comes directly from the pdf for the 2nd estimate of 4th quarter GDP, which can be accessed directly on the BEA’s GDP landing page, which also provides links to the source data, the tables on Excel and other technical notes. Specifically, we reference table 1, which shows the real percentage change in each of the GDP components annually and quarterly since the 1st quarter of 2018 table 2, which shows the contribution of each of the components to the GDP figures for those quarters and years; table 3, which shows both the current dollar value and inflation adjusted value of each of the GDP components over the last 5 quarters; and table 4, which shows the change in the price indexes for each of the components. The pdf for the 4th quarter advance estimate, which this estimate revises, is here…
Growth of real personal consumption expenditures (PCE), the largest component of GDP, was revised from 3.3% down to an overall 3.1% growth rate in this 2nd estimate…that growth rate figure was arrived at by deflating the 9.61% growth rate in the dollar amount of consumer spending with the PCE price index, which indicated consumer inflation grew at a 6.3% annual rate in the 4th quarter, and which was revised from the 6.5% PCE inflation rate reported a month ago, which should have accounted for an upward revision to real PCE, were it not for spending figures that were revised that much lower.. Real consumption of durable goods grew at a 2.7% annual rate, which was revised from the 1.6% growth rate shown in the advance report, and added 0.24 percentage points to GDP, as real growth at a 13.8% rate in consumption of recreational goods and vehicles was partly offset by lower consumption of motor vehicles, furniture and durable household equipment. In addition, real consumption of nondurable goods by individuals grew at a 0.8% annual rate, revised from the 0.1% contraction rate reported in the 1st estimate, and added 0.13 percentage points to the 4th quarter’s economic growth rate, as lower growth in real consumption of clothing and footwear was more than offset by greater real consumption of gasoline and other non-durable goods. At the same time, consumption of services grew at a 3.9% annual rate, revised from the 4.7% growth rate reported last month, and added 1.76 percentage points to the final GDP tally, as a 5.5% growth rate in real health care services accounted for almost 40% of 4th quarter services growth…
Meanwhile, seasonally adjusted real gross private domestic investment grew at a 33.5% annual rate in the 4th quarter, revised from the original 32.0% growth estimate reported last month, as real private fixed investment grew at a 2.6% rate, revised from the 1.3% growth rate reported in the advance estimate, while inventory growth was little changed from the previous estimate Investment in non-residential structures was revised to indicate contraction at a 9.4% rate, up from the 11.4% contraction rate previously reported, while real investment in equipment grew at 2.4% rate, revised up from the 0.8% growth rate shown a month ago…meanwhile the quarter’s investment in intellectual property products remained unrevised at a 10.6% growth rate, while at the same time real residential investment was shown to be growing at a 1.0% annual rate, revised from the 0.8% contraction rate shown in the previous report. After those revisions, the increase in investment in non-residential structures subtracted 0.25 percentage points from the 4th quarter’s growth rate, while the increase in investment in equipment added 0.14 percentage points to the quarter’s growth rate, growth in investment in intellectual property added 0.53 percentage points to the growth rate of 4th quarter GDP, and growth in residential investment added 0.05 percentage points to the growth of GDP…..for an easy to read table as to what’s included in each of those GDP investment categories, see the NIPA Handbook, Chapter 6, page 3…..
At the same time, growth in real private inventories was revised from the originally reported $173.5 billion in inflation adjusted growth to show that inventory grew at an inflation adjusted $171.2 billion rate. That came after inventories had contracted at an inflation adjusted $66.8 billion rate in the 3rd quarter, and hence the $238.0 billion positive change in real inventory growth from the 3rd to the 4th quarter added 4.90 percentage points to the 4th quarter’s growth rate, unrevised from the 4.90 percentage point addition to GDP from inventory growth reported in the advance estimate (NB: the lack of revision to that contribution was due to the BEA’s forcing the component totals to add up the independently calculated total, analogous to a rounding error). However, since greater growth of inventories indicates that more of the goods produced during the quarter were left in a warehouse or sitting on the shelf, their increase at a $238.0 billion rate meant that real final sales of GDP were actually smaller by that amount, and hence real final sales of GDP only grew at a 2.0% rate in the 4th quarter, revised from the real final sales 1.9% growth rate shown in the advance estimate, and up from the real final sales growth rate of 0.1% in the 3rd quarter, when inventory growth had accounted for almost all of the quarter’s growth in GDP…
The previously reported increase in real exports was revised somewhat lower with this estimate, while the previously reported increase in real imports was revised just a bit lower, so on net the change in our net trade was a small subtraction from GDP, rather than negligible as was previously reported. Our real exports grew at a 23.6% rate, revised from the 24.5% rate reported in the first estimate, and since exports are added to GDP because they are part of our production that was not consumed or added to investment in our country, that growth added 2.35 percentage points to the 4th quarter’s growth rate, revised from the 2.43 percentage point addition to GDP shown in the previous report. Meanwhile, the previously reported 17.7% growth rate in our real imports was revised to a 17.6% growth rate, and since imports are subtracted from GDP because they represent either consumption or investment that was added to an other GDP component that was not produced here, their increase subtracted 2.42 percentage points from 4th quarter GDP, rather than the 2.43 percentage point subtraction shown last month. Thus, our deteriorating trade balance subtracted a net of 0.07 percentage points from 4th quarter GDP, revised from the rounded no change that had been indicated by the advance estimate..
Finally, there was also an upward revision to real government consumption and investment in this 2nd estimate, since the real contraction rate for the entire government sector was revised from a 2.9% rate to a 2.6% rate.. However, real federal government consumption and investment was seen to have shrunk at a 4.5% rate in this estimate, revised from the 4.0% contraction rate shown in the advance estimate, as real federal outlays for defense shrank at a 6.1% rate, revised from the 5.7% contraction rate shown previously, and subtracted 0.24 percentage points from 4th quarter GDP, while all other federal consumption and investment was revised from a 1.6% contraction rate to contraction at a 2.2% rate, which subtracted 0.06 percentage points from 4th quarter GDP.. Meanwhile, real state and local consumption and investment was revised from shrinking at a 2.2% rate in the first estimate to shrinking at a 1.6% rate in this estimate, as state and local investment spending shrank at a 7.3% rate and subtracted 0.14 percentage points from 4th quarter GDP, while state and local consumption spending was slightly lower but had a negligible impact on GDP. Note that government outlays for social insurance are not included in this government GDP component; rather, they are included within personal consumption expenditures only when such funds are spent on goods or services, thus indicating an increase in the output of those goods or services…
An Additional Note on the Major Revisions to the GDP deflators with this Report
This month’s report also included unprecedented major revisions to the percentage changes of the deflators used to adjust the various GDP components for changes in prices…we’ve already mentioned that the overall change in the price index for personal consumption expenditures was revised from a 6.5% rate in the advance estimate to a 6.3% rate in this month’s report; that change was driven by a downward revision of the change in the price index for non-durable goods from an 11.0% rate to one of 9.8%, while the price index change for PCE services was revised from 4.3% to a 4.4% rate. At the same time, the change in the price index for fixed investment was revised from 8.0% to 8.9%; most notable among the revisions of its component indexes was the change in the deflator for residential investment, which was revised from 9.5% to a 12.1% rate. There were also revisions to the deflators for exports and imports; the export deflator was revised from 5.9% to 6.4%, while the import deflator was revised from 5.8% to 5.6%; the effect of those revisions would be to lower real exports while increasing real imports. There were also revisions to the deflators used on government consumption and investment; the overall government outlays deflator went from 7.4% to 7.7%, as the deflator for non-defense federal government outlays was revised from 5.4% to 6.1%. Incorporating the effect of all those component deflators meant that the deflator for GDP overall was revised from the the 6.9% rate indicated by the advance estimate to a 7.1% rate with this month’s estimate. In the more than half dozen years i’ve been checking this metric closely, i’ve rarely seen deflator revisions greater than ±0.1%. You can view the deflators used in the advance estimate in table 4 here, while the deflators used in this month’s estimate are in table 4 here…
Normally, we would expect an upward revision in a deflator to result in a corresponding downward revision to its associated real component, and vice-versa, simply because these deflators correspond directly to the chained dollar inflation factor used to convert the change in nominal dollar spending into the change in the real component, thus indicating the change in output of the associated GDP component. That played out for some of the revisions we have noted; for instance, the revision in the deflator for non-durable goods from 11.0% to 9.8% was coincidental with a revision in the real growth of non-durable goods from a 0.1% contraction to growth at an 0.8% rate. On the other hand, the revisions to some deflators seemed to have the opposite effect that one would expect; the revision in the deflator for residential investment from 9.5% to 12.1% coincided with an upward revision to real residential from a contraction at a 0.8% annual rate to growth at a 1.0% rate…
The BEA doesn’t give much information on the reason for the unprecedentedly large revision to those deflators; at one point, in addressing the downward revision to the PCE price index change, they mention a downward revision in the consumer price index for gasoline in the technical notes accompanying this month’s report (pdf), but gasoline only accounts around 2% of GDP and 3% of PCE, certainly not large enough to move the overall needle 0.2%. For the most part, the deflators used for GDP are derived from data provided by the Bureau of Labor Statistics in their monthly releases of the Consumer Price Index, the Producer Price Index, and the Import-Export Price Index, which did undergo annual revisions to seasonal adjustment data in January, which the BLS does every year, and i have noted inordinately large revisions to seasonal adjustments this year, not just for prices, but also for other economic metrics, including most notably employment..
Tto compute seasonal adjustments, agencies take several years of historical data on each metric and come up with average on how much the data should be expected to change each month, after also taking into account other factors such as holidays and weekends in each month, and then adjust the current month’s data to account for that typical month over month change. The result of the adjustment is supposed to give us a representation of how much the data would have changed after eliminating seasonal influences; for instance, if retail sales are normally expected to fall 20% from December to January based on the normal end of Christmas shopping, and they only fell 18%, then the post adjustment figures would show seasonally adjusted sales rose 2%…
Hence, to compute this year’s revisions to seasonal adjustments, the historical data being used now includes almost two years of data that has been impacted by the pandemic, and that adjustment is being applied to new data as if all the economic impacts were due to seasonal changes. The result is that most economic data from months when the pandemic has been severe, ie, generally November through February, has been revised upward, or reported higher, whereas economic data from the months when the pandemic has historically ebbed, ie, generally May, June, and July, is being revised lower, and thus i expect those months will be reported lower going forward. Similarly, the supply chain shortages that arose in the wake of the pandemic certainly had an inordinately large impact on prices that would have been incorporated into the seasonal adjustments that certainly weren’t seasonal in nature.. But i can’t see any easy fix to that problem; how one could separate out the impacts of the pandemic from those that were due to real seasonal factors; ie, how could one tell whether people shopped less in January because of Covid, or because of the bad weather?
It appears that despite the large revisions to the GDP deflators that accompanied this month’s report, the overall revision to GDP from the advance estimate to this second estimate was in the range that we’d consider normal. The most likely reason for that seems to be that not only did the price factors undergo a seasonal adjustment revision, but at least some of the underlying data also did as well, which would probably be in a similar direction and magnitude to the deflator revisions. That would have a mitigating effect on the deflator related revisions, and vice versa; ie, this month’s upward revision to the growth rate of seasonally adjusted current dollar GDP from 14.33% in the advance estimate to 14.64% in this estimate was partly offset by the upward revision in the GDP deflator from 6.9% to 7.1%…