Why the Media Gets the Retail Sales Report Wrong

The Misconception

Retail sales this morning were strong as sales increased 0.7% from the prior month.

My View

The number above isn't necessarily wrong, but there are better numbers to use, depending on your purpose. Retail sales grew 3.4% y/y in November, which was a deceleration from October's 4.9%. Not bad but not great either.

The Details

There are two numbers that census retail sales will report -- a seasonally adjusted number and an unadjusted number. The seasonally adjusted number is the number that the media reports and that a lot of people focus on. As I'll argue, the unadjusted number is the one that really matters when trying to draw stock conclusions.

What is the seasonal adjustment number? In general, seasonal adjustments are the adjustments that try to make each period comparable. This is what the Census Bureau states:

(Estimates adjusted for seasonal variation and holiday and trading-day differences, but not for price changes)

The adjustments try to account for differences in seasons, holidays, and trading day differences. Sounds great, but it's difficult to do and there may be hazy adjustments involved. They'll make these adjustments so that you can compare November to October, even though the months might have a different number of selling days, or different holidays and purchasing patterns. The adjustments to the number attempt to adjust for these differences so that one number in October is directly comparable to one number in November.

What is the unadjusted number? This is the raw number. It tells you what the actual sales were in that period, and it won't adjust for anything. So when December numbers come out, you'll see a much higher number than in November, since December will include the all-important Christmas sales, whereas November won't include all of them. You'll also have the benefit of having one more day in December. So the December number is not directly comparable to November. 

So how can we figure out if the results were good or not? This is where the year over year growth rate comes in.  Year over year growth rates compare one period to that same period in the prior year. This controls for events that are specific to that month, such as holidays (in most cases), or days in the month. You can then compare y/y growth rates from month to month to figure out if things are accelerating or decelerating.

We should focus on the unadjusted y/y number because this is how companies report their financial results. When you hear about a company's sales, they don't make seasonal adjustments. Instead, what you'll see is the y/y change in their  raw sales. So if we're trying to extrapolate company sales implications from the retail sales data, we need to look at both in the same consistent way.

Retail sales unadjusted wasn't horrible, but it wasn't great either. Using the y/y calculation off of unadjusted numbers, retail sales were actually not as great as the 0.7% seasonally adjusted m/m growth might suggest. Here's the unadjusted y/y growth numbers for the past year. November growth decelerated to 3.4% from the 4.9% in October.

However, as I'll detail in another post later, we can dig even further. Census Bureau data has subcategories that detail performance within different areas of retail. There can be wide variances in performance, so if you're looking for specific takeaways for SPLS, or HD, or BKS, you really need to look into the relevant subcategory.