June retail sales revisions up 0.2%, greater than entire July increase

According to the Advance Retail and Food Services Sales Report for July (pdf) from the Census Bureau, the seasonally adjusted advance estimate of retail and food services sales for the month was $424.5 billion, an increase of 0.2 percent (±0.5%)* over June's revised sales of $432.6 billion and 5.4% (±0.7%) above last July's $402.7 billion.  The asterisk tells us that the Census Bureau does not have sufficient statistical evidence to conclude that the actual change in July sales is different than zero, since this widely followed report is just an advance estimate of sales collected by phone, fax or mail from a sample of approximately 4,900 firms and will be revised when more complete data is available next month, and the month thereafter.  More specifically, their opening statement means they're 90% confident that seasonally adjusted retail sales for July were between $431.3 billion and $435.6 billion, or between 0.3% less and 0.7% greater than June’s.  The unadjusted estimate of July's retails sales was at $427,547 million, up from $421,636 million in June but down from the spring sales high of $444,370 million in May.  Year to date unadjusted sales, including this preliminary data, is estimated to be $2,896,307 million, which is 4.2% higher than the year to date for the first 7 months of last year.  Neither the seasonally adjusted figures nor this data is adjusted for price changes..

  The real story of this July retail sales report was not July's weaker than expected 0.2% month over month growth, but the revision of June's sales from a 0.4% increase over May to a 0.6% increase, which was treated as a footnote elsewhere; the fact is that the change to June's earlier estimated sales figures was greater than the entire increase in sales reported for July..  As originally reported, the advance estimate of June retail sales was at $422,794 million; that has now been revised to a preliminary $423,649 million, an $855 million correction.  July's sales gain of just $832 million to $424,481 million was calculated from that new, revised June baseline...so on net, this report is indicating a larger jump in retail sales than was indicated by last month's headline 0.4% increase with a less than 0.1% downward revision for May..

  Moreover, the revisions to June data brought significant revisions to sales in most retail trade sectors vis a vis what we reported last monthOn the left below, we have the image of the table showing the sales percentage for the major retail sectors from this month's report; lined up approximately next to that on the right we have the original June percentage table from last months report (archived pdf here).   Just to clarify the headings, the first estimate of each census report is labeled "advance", the second estimates, a month later, are the "preliminary" report, and the third estimate is called the "revised" report.  Here below, we see the "advance" estimate of June retail sales on the right showing a 0.4% increase in the top line total retail sales from May (as published last month), and in the middle columns we have the "preliminary" correction to that change from May at 0.6% and to the change from a year earlier at 5.9%.  But that's not all; note the change for each of the major retail business types; sales at car dealers, for instance, were originally reported up 2.1% for June; now they're indicated as up 3.3%; on the other hand, for instance, clothing stores had been showing a 0.7% gain over May, but now they show flat sales at 0.0%.

July retail table plus

June advance retail for July

Since the revision of the June report actually accounted for more of the increase in July retail sales over what was last reported than the July change, we'll take a brief look at some of the major changes that resulted in that swing.  First, as we've already alluded to, June's seasonally adjusted sales at vehicle & parts dealers were originally reported to have risen 1.8% from a seasonally adjusted $79,028 million in May to $80,461 million in June. That's now been revised to an increase of 2.9% from $79,300 million to $81,584 million in June, so combined with the May revision, the increase in June auto sales was rewritten 1.4% higher than it was reported last month.  But even with that big jump in revised auto sales, retail sales ex auto still managed to eke out a 0.1% increase, from May's $341,720 million to $342,065 million in June, versus the flat sales increase without autos & parts reported last month, because of additional upward revisions for other retail sales.

Two major retail businesses which had been thought to have been off substantially in June weren't as bad as originally reported.  At building and garden supply stores, where sales were first reported to have fallen 2.2% to $25,772 million in June from $26,346 million in May, sales actually only fell 1.6%, to $25,909 million from a revised May sales total of $26,330 million; then sales at bars and restaurants, where June's sales were first reported at $45,251 million, 1.2% below May's level of $45,821 million, Junes sales have now been revised to a much smaller decline of 0.5%, from a seasonally adjusted $45,771 million in May to a preliminary sales figure of $45,559 million in June.  Meanwhile, non-store retailers, who had been thought to have seen sales rise 2.1% from $37,154 million in May to $37,936 million in June, were revised this month to show a 1.5% June sales gain from a much lower revised May figure of $36,739 million to a preliminary sales total of $37,276 million in June.

  So, now that we understand that the widely followed "advance" retail sales estimated this week for July are less than dependable and subject to large revisions (median standard errors and coefficients of variation for various businesses are explained on the last page of the release), we can look at those estimates with an appropriate grain of salt.  Seasonally adjusted sales at motor vehicle & parts dealers, which account for a bit less than 20% of aggregate retail sales, were down 1.0% in July, from $81,584 million in June to $80,799 million in July. However, unadjusted motor vehicle & parts sales were at $86,010 million, up from June's $82,652 million but down from May's unadjusted $86,431 million, highly suggesting that seasonal adjustments, which anticipated a greater fall in June sales, skewed the reported changes in sales for these businesses over both June and July.  Seasonally adjusted sales at motor vehicle & parts dealers were still 11.8% higher than last years at $72,284 million, while unadjusted sales for those auto related businesses were 15.0% higher than last July's, which is odd because we had thought July was in the same season this year as it was last year.  With the apparent large hit to auto sales, the widely watched retail sales ex-autos, aka "core retail sales" were up 0.5% for the month, from a seasonally adjusted $342,065 million in June to $343,682 million in July...

Among retail trade sectors showing better than average adjusted sales increases in July, the broad category of stores specializing in sporting goods, hobbies, books or music saw sales rise 1.0% from $7,421 million in June to $7,498 million, while sales at clothing & accessories stores were up 0.9% from $20,860 million in June to 21,046 million July, and sales at food and beverage stores were up 0.8%, from $53,935 million in June to.$54,355 million in July.  In addition, sales at gas stations rose 0.9% from $45,390 million in June to $45,784 million in July as gasoline prices were up 1.0%..  Year over year sales gains for all of these better than average July retailers, however, was below the 5.4% YoY gain posted for all sales,  Sales at sporting goods, hobbies, books & music stores was were up just 2.3% from last July, clothing stores saw sales rise 4.2% from a year ago, food and beverage stores posted a one year sales increase of 3.9%, and July sales at gas stations were 4.9% greater than last years..

Other than automotive, retail sectors that saw seasonally adjusted sales decrease in July included furniture and home furnishing stores, where sales fell 1.4% to $8,398 million in July, after rising 2.5% to $8,516 million in June.  Since seasonally adjusted furniture sales wouldn't seem to be subject to much volatility, this also seems like it may be another seasonal adjustment aberration.  July sales at building materials and garden supply stores were down another 0.4% to $25,802 million in July after being down 1.6% to $25,909 million in June, suggesting more of a slowdown in home building and remodeling than is evident from other data.  Nonetheless, sales for this retail group were still 7.9% above those of a year earlier.  Meanwhile, sales at electronic and appliance stores were off 0.1%, from $8,411 million in June to $8,402 million in July, and down 0.4% year over year...other than electronic and appliance retailing, only department stores, where sales fell 4.8%, saw worse year over year sales results.

Our FRED pie chart below was constructed to give us visual picture of the relative size of each of the component retail business groups included in this report (click on any piece for a large pie).  Note that since FRED limits us to a maximum of 12 data sets per graph, we have omitted the “miscellaneous store retailers”, which accounted for $10,595 million of sales in July, or less than 2.5% of the total.  These retailer groups are arranged around the pie from those with the most sales to those with the least, so in the upper right corner, we have motor vehicle & parts dealers in blue at 19.5% of the remaining sales, and then, counterclockwise from there, general merchandise stores in red at 13.3% of sales ex “other”; food and beverage stores in green, accounting for 13.1% of sales; gas stations in orange and restaurants and bars in grey, both accounting for 11.1% of sales; non-store retailers, or online & mail order sales, in medium blue at 9.0%; building material & garden supply retailers in light green at 6.1%; drug stores in mustard coloring at 5.7%; clothing and accessory stores at 5.1% in pink; electronic and appliance stores in purple and furniture and furnishings stores in yellow, both at 2.0% each; and lastly, stores specializing in sporting goods, hobbies, books or music in light blue accounting for 1.8% of July retail sales, all ex "other"..

,FRED Graph

(crossposted from Marketwatch 666)

meta: 

Comments

The MSM pumped Retail Sales as all is well and good, ha ha

Glad you overviewed it, I skipped it trying to focus on metrics that are really unusual, which were many this week. (check out the GDP estimates by calculation, 2.3% is the new guesstimate!).

Still, the MSM was all "isn't this awesome, all is humming along, retail sales grew by 0.2%"!

Really, do these people even understand what 0.2% is? That 0.2% is chicken feed? Next month might do a historical charge on percentage changes.

Also, I am thinking economic reports are just "number feed", traders go off of the headline buzz number, have no idea what it means and HFT algos auto ftp the number in nanoseconds of it's release and input that into their trading algos, running a millisecond after.

Don't you find it strange that regardless of the details, or most of the press getting it wrong, these economic reports have a news shelf life of 30 minutes? Seriously, after max 8 hours no one reads the details, it's all just to hay to feed the trading herd.

You must have Javascript enabled to use this form.

few know where these numbers come from

you're right, 90% of the market moves on the headline number, without a clue where it comes from or how small the sample is that it comes from...it's remarkable that these reports can even get close, when surveying just 4900 out of 1.8 million firms for the advance estimates..

You must have Javascript enabled to use this form.

rjs

polls, sample size

Well, they can get reasonably close with political polls where the same size to actual voter base is much lower, so it's possible, depending on the methodology and the representative spread.

That said, pretty clear to me most economic data is not understood at all, they don't care, they are trading on it, with no comprehension and it's just like the VIX or VAR.

I'd say the people in the U.S. who care about this from an overall real economy economic health have to be less than 100,000, even though there are at least a million with a BA in Econ in the U.S.

You must have Javascript enabled to use this form.