Retail: A tale of two charts

My friend Vitaliy just published a piece in which he notes that Amazon is hardly the only problem with retail today. For the most part, I agree. Those who follow my occasional comments on Facebook know that my opinions about retail very much match his, and also dovetail into my thesis that less dense areas are more likely to be devastated by this change than denser ones. I have tended not to focus on those matters on this blog, as I’d like to keep it mostly about data science and tech.

However, this is one of those cases where the economics of retail have something to teach us about how to best look at data, and how important it can be to choose the right data to look at.

Vitaliy’s core thesis is well-founded, in my opinion: “If you are only solving “who can fight back the best against Amazon?” you are only solving for one variable in a multivariable problem: – Consumers’ habits have changed; the U.S. is over-retailed; and consumer spending is being diverted to different parts of the economy.”

But at the same time, a more careful understanding of the data might suggest that Amazon is a lot more important than it may seem at first glance — at least as relates to much of the retail landscape and related employment — and that the lack of benefit from “savings” in other areas can easily be understood when the data is explored more thoroughly. As a data scientist, that’s the kind of thing I need to be able to do in order to tell a story that is not only compelling, but accurate.

Define your terms!

OK, so let’s start with the basic one. What is “retail?”

The US Government has one definition. It includes things like automobile sales, gasoline, and restaurants that many people don’t normally think of when the word “retail” comes up.

Wall Street has another, it mostly captures things that you buy at a mall or a Wal-Mart or a supermarket.

So when Vitaliy notes that “Ten years ago only 2.5 percent of retail sales took place online, and today that number is 8.5 percent – about a $300 billion change” he’s talking about the government numbers published by the St. Louis Fed. There’s nothing wrong with these numbers, but they need to be understood in the context of a government definition of “retail” that includes a whole bunch of things that would not be captured in a “retail stocks” index or portfolio.

This is how the government agencies that report such things define and divide the “retail landscape”:

Note that the biggest sector, cars and car parts aren’t really Amazon competitors (while Amazon does a decent business in retail car parts, the vast bulk of these are still sold by dealers or repair shops). Neither is the second, food and beverage stores (at least not yet). Restaurants are also in there. So are gas stations, building and garden materials, and furniture. Most of these are somewhat resilient in the face of online competition. Most of these would also not be thought of as “retail” to a Wall Street analyst, index-creator or fund manager.

Find the relevant comparison

For an investor in “retail,” or related sectors like mall operators, I don’t think it’s particularly relevant to consider the government definition of retail that is overly broad. Such a definition is even less relevant when you’re trying to assess the impact of an specific organization like Amazon or a larger trend like the move to online shopping. So I took a deep dive into the rest. I isolated the sectors I consider “Amazon competitors” and charted only those. (I had originally used the categories “Amazon resistant” and “Amazon vulnerable,” but discarded them. These days, everybody needs to consider themselves vulnerable.)

Look at it this way and online retail (being about 80% of the “nonstore retailers” segment) is a huge deal. It is almost as big as the “general merchandise stores” category and if you look at the time-series data, you’ll note that it’s growing. This chart also approximates much more closely the way “retail” is typically understood by Wall Street and most normal people (as opposed to government statisticians), as it excludes autos, gasoline and restaurants.

If you think about what this chart really tells you, the story is different from the one Vitaliy tells. In this chart there’s a large general merchandise category dominated by Wal-Mart, another large “non store” category dominated by Amazon, and a whole bunch of other smaller categories. It’s a story of Amazon and Wal-Mart going after each other, both directly and by trying to squeeze out as many of those other smaller players as possible. Who are all those smaller players? Pretty much everybody at your local mall, where this battle of titans has had a disproportionate impact.

Which is why “government retail” can be doing just fine, even as most retailers — particularly the niche and specialty retailers at our local malls — get killed.

Understanding the details of the data matters

It’s easy to come to the wrong conclusion when you don’t understand at a detail level what the data is saying. For example, Vitaliy notes:

“All the aforementioned factors combined explain why, when gasoline prices declined by almost 50 percent (gifting consumers hundreds of dollars of discretionary spending a month), retailers’ profitability and consumer spending did not flinch – those savings were more than absorbed by other expenses.”

Yes and no. Many of the other (non-retail) expenses that consumers face are real and have had an impact on the availability of dollars for retailers. But as far as the official “retail” number goes, the far bigger impact is that the declining gasoline sales are retail sales! All things being precisely equal, a reduction of the cost of gasoline should bring the official retail number down, not up! The fact that retail held steady even as gasoline sales (measured in dollars) declined, suggests that the money saved moved into other areas of retail, not into services or savings. Of course the truth is — as Vitaliy notes — far far more complex with many second and third order effects: reduce spending in one area, it moves to another, which may cut into a third, or generate some other ongoing expense, etc.

Not to come down too hard…

The point of this blog is to discuss uses and misuses of data, not to provide detailed economic critique. Likewise, Vitaliy’s article for a general-consumption website is hardly meant to be a detailed treatise on the economics of retail. He’s more than capable of generating that analysis for the right audience. So my apologies for using it as a bit of a punching bag. I know the author and he can take it. He’ll probably tell me why I’m wrong.

The point is to be “less wrong” and there’s more right here than wrong. Smartphones and related virtual purchases are displacing traditional retail. We are wildly over-malled and many malls are not prepared for the transition away from “shopping for stuff” towards “shopping for experiences.” The local mall here in Santa Monica could do it: replace one department store with a high-end movie theater and much of the former retail space with restaurants. That will work in some places but not in many others. The old Sears building across the street will most likely become hotel, restaurant and maybe some smaller retail space. Most places in the country, it might just sit empty as the life gets sucked out of the surrounding community. I would add to Vitaliy’s observations that the impact is geographically disproportionate. And of course, spending on services of all sorts continues to grow even as stores shutter.