The Difference Between Good Data and Bad
On Monday, I cast doubt on the many stories about how Black Friday retail sales were off to a disappointing start. This is an important story because retail is such a critical part of the U.S. economy, and because such a large share of the industry's sales occur during the roughly five weeks between Thanksgiving and Christmas. But the more important point -- at least for my purposes -- is that the initial reports, thanks to the National Retail Federation, are a case study in how to obtain meaningless data and then put it to bad use.
The NRF reported a 3.5 percent drop in spending. "Average spending per person over Thanksgiving weekend totaled $289.19, down slightly from $299.60 last year," the organization said in a statement. This information was based on asking consumers how much they figured they would spend this year versus a year ago.
It looks like this is turning out to be wrong.
I make a big deal about the retail trade group’s record of inaccuracy every year for a few reasons: it is important for investors -- and indeed, citizens -- to be grounded in reality. Most human progress is the result of the work of scientists, technologists and logicians who rely on facts and testable theories. More to the point for investors, it is painfully expensive to allocate capital based on a premise that later proves to be fundamentally false. That's the source of my ire when I spot someone pushing a narrative based on flawed data that potentially loses investors money.
This is crucial because retail sales are such a big deal. Almost 16 million people work in retail, or about 10.9 percent of the U.S. labor force. It accounts for a huge percentage of the overall economy. Retail sales provide a window into consumer sentiment, as well as corporate revenue, profits and investment decisions. By some measures, consumer spending counts for almost two-thirds of gross domestic product.
Back to holiday shopping: it is of course way too early to have the final retail sales data, but we do have some early numbers based on actual sales. First Data Corp., a point-of-sales transaction processor, says that it examined data from almost 1 million merchants and concluded that sales so far this holiday shopping season are up 9 percent from a year earlier. Furthermore, perhaps in a sign of the state of the industry's health, sales of electronics and appliances rose 26.5 percent, compared with a lackluster 2.3 percent gain last year. First Data also found that the average transaction grew by more than $41 year over year.
First Data noted that its analytical methodology "is based on actual consumer transactions rather than surveys or speculation." The company has access to this information because it processes actual credit-card and debit-card transactions.
But let's also be clear: This doesn't necessarily mean that all holiday retail sales are up 9 percent. This is because First Data doesn't track cash purchases. These are still a significant part of retail sales. This is especially true for lower-income consumers, who tend to have less available credit and/or are deeper in debt relative to income. Not capturing the spending patterns of those consumers who are less well off risks missing a potentially significant indicator about the overall health of the retail sector.
There are other forces at work. For the past 20 years, online commerce has commanded a bigger share of retail sales, and almost all of those are purchases made without cash. First Data notes that e-commerce represented 24.7 percent of all retail sales, excluding gasoline stations and fuel dispensers. That's a huge increase from 13.6 percent in 2013. With online commerce growing at double digits for years, a big chunk of that 9 percent gain reflects the shift away from bricks-and-mortar retailers (though to be sure, some of them are one and the same).
One other point: The First Data information is drawn from 940,000 merchants who participated in the program. They may, or may not be, representative of the retail industry or the sales environment as a whole. If a disproportionate number are in stronger economic regions (mostly western states), or hotter retail sectors or are more successful retailers, that could skew the results.
In other words, the First Data report is preliminary analysis, nothing more. But at least it relies on real data and not some squishy and anecdotal information like consumer recollections of spending a year ago or foot traffic in shopping malls.
All of which should remind readers of these reports from the National Retail Federation that there is nothing in them that's actionable or a basis for an investment decision.
This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.
To contact the author of this story:
Barry Ritholtz at email@example.com
To contact the editor responsible for this story:
James Greiff at firstname.lastname@example.org