The headline figure showed a mere 0.1% increase, less than the 0.3% expected by the consensus. But living up to consensus expectations isn’t the consumer’s job. Retail sales track nominal dollars, and when goods prices are falling, real, inflation-adjusted retail sales—the true bottom line—are invariably running higher.
 
For example, the 0.9% drop in gasoline sales was accompanied by a decline of a few percentage points in prices at the pump. So October gasoline sales in real dollars must have increased.

Motor vehicles and parts stores saw a sales decrease of 0.5%, but we already know unit sales of vehicles hit a 10-year high in October. So there, too, the decline in nominal dollars must have reflected price discounting, with a likely increase in real sales.
 
In any case, as part of the calculation of consumer spending in gross domestic product, the more reliable data on unit sales of autos are used, instead of the data on retail sales. Consumer spending also includes outlays for services, which aren’t tracked in retail sales, and which account for more than half of total dollars spent.
 
Real retail sales are rising, probably at an annual rate of 3%. But what also seems to be happening is that much of the cash saved on retail consumption via lower prices is being transferred to spending on services. So far this year, nominal dollars spent on services have run at 67.5% of total consumption, up from 66.6% in the comparable 2014 period—the highest share since the expansion began.
 
LATE LAST MONTH, THE Bureau of Economic Analysis reported disappointing growth in GDP for the third quarter, at an annualized rate of 1.5%. The inventory component of GDP, which clipped 1.44 percentage points from overall growth, was the main reason for the headline figure’s poor performance.
 
It now looks as though the inventory haircut was overdone, however, based on September data released last week. When the revised third quarter estimate is released on Nov. 24, it will probably show that GDP expanded at an annualized 2%. The half-percentage-point increase reflects the upward revision to inventories.
 
Such technical details leave readers the impression that GDP is shrouded in mystery. Much of the mystery can be cleared up by revealing an embarrassing fact about how the numbers are gathered.
 
GDP is a measure of what is produced domestically, and if the BEA had access to comprehensive data on that, it would report the headline figures differently. It might be interesting to know, for example, what share of domestic output went to consumption, to exports, and to any buildup of inventories. But such information would be of secondary importance, and barely get noticed.
 
The embarrassing fact, however, is that the Bureau of Economic Analysis lacks access to the economic body-scan that can fully reveal domestic production, and so must rely on the statistical equivalent of detective work to infer what it must have been.
 
Two important clues are the change in consumer spending and in exports. To the degree that consumption and exports rise, one can infer that production rose in tandem. Another clue is the change in inventories. To the degree stocks are built up or drawn down, domestic production naturally rises or falls. These numbers are prominently reported—way out of proportion to their conceptual importance—only because data on domestic production are not available.
 
The 1.44 percentage-point inventory-related haircut in the third quarter created the impression that stocks had been drawn down. In fact, there was some increase in stocks, but a much smaller one than in the previous quarter, which weighed heavily on the estimate for production.

The September data on inventories upwardly revised this increase. Hence, the upward revision to 2% growth.
 
One small wild card: imports. They are another clue in estimating production. Imports would also merit barely a mention in the GDP accounts if data on domestic production were directly available, since imports are, of course, purchased from abroad rather than domestically made.
 
But the BEA knows some goods and services that figure in consumer spending are imported, just as it knows some inventories are imported. So, again lacking direct data on what was domestically produced, it must subtract imports from overall GDP. Additional data on September imports will be released early next month.
 
The accounting device of always subtracting imports from GDP has been cited by crackpot economists as virtually scientific evidence that imports are death to GDP growth. Please don’t share this with Donald Trump.