In 2010, economist Mark Zandi identified weekly chain store sales in a short list of indicators he described being "often misleading". Or rather, data for which he said "the signal to noise ratio is very high and thus hard to interpret."
That was an unfortunate way to phrase that rationale, because he really meant to say the opposite. It is information that comes with a low signal-to-noise ratio that's problematic, because it's having too much noise and not enough signal that makes it tough to find out what you want to know in that situation.
Knowing that, what exactly is it about weekly chain store sales that makes it less than useful for an analyst who might be attempting to divine the direction of the economy?
Just for fun, we asked ChatGPT the question "what makes weekly chain store sales data less than useful for analysis?" Here's how the AI system responded:
Weekly chain store sales data can be less useful for analysis because it only provides a snapshot of sales activity for a given week. It doesn’t provide any information about the underlying trends or patterns that may be driving those sales. Additionally, weekly chain store sales data may not be representative of overall retail sales trends, as it only captures sales activity at a subset of retail stores.
The point about the data only coming from a small subset of retail stores is a good one. Trading Economics describes what the data specifically covers:
In the US, chain store sales refer to retail sales of goods and services in department stores excluding discount department stores. Values are based on data from the Monthly Retail Trade Survey. The survey covers companies with one or more establishments that sell merchandise and related services to final consumers, such as furniture & home furnishings; electronics & appliances; clothing & accessories; sporting goods, hobby, musical instrument, and book; general merchandise; office supply, stationery, and gift stores. Chain store sales do not include data from online-only stores.
How good is that data? We found a 1996 research paper by New York Fed economists Ethan S. Harris and Clara Vega that describes the potential usefulness of the weekly chain store sales data for analysts:
Our empirical tests provide mixed support for the use of chain store data. On the one hand, we find that weekly indexes and monthly reports from individual companies are too erratic to be useful for forecasting. On the other hand, we find that monthly chain store indexes, if given the appropriate weights in forecast models, add significantly to the accuracy of in-sample and out-of-sample predictions for several measures of consumer spending.
In other words, provided you weight it and combine it with other economic variables, you could get useful results from monthly chain store sales data. But the weekly chain store sales data is pretty close to garbage for getting any kind of accurate picture of the current state of retail sales, much less finding any trends that might be useful for forecasting their future.
Harris and Vega also cited the small fraction of consumer spending in the chain store sales data as a weakness in their 1996 paper. Since then, that fraction of total consumer spending has shrunk, where the chain store sales data omits sales at online retailers like Amazon or the discount dollar store retailers that have come to dominate consumer retail transactions in many regions of the U.S. The chain store sales data has a growing hole in what it data covers.
Let's summarize what's wrong with weekly chain store sales data. It's erratic. It doesn't cover a large enough sample of retail sales to be able to tell much about them or where they might be going. Its problems have been getting worse over time.
Now for the good news. Aside from Trading Economics' continuing tracking of the data series, weekly chain store sales data has nearly disappeared from regular media reporting. It appears to have lasted into 2020 at one outlet, but the top search results on Google include articles from 2002, 2007, and 2016. We did find weekly chain store sales mentioned in a 2023 article, but it is in reference to a different, more comprehensive weekly data series, which is not the same animal.
It's quite possible the weekly chain store sales data Trading Economics tracks to this day is the zombie remnant of a data series that's barely hanging on, but whose time has finally passed!
Previously on Political Calculations
- Less Than Useful Data in the World of Big Data
- Less Than Useful Data: Consumer Confidence
- Less Than Useful Data: FHFA House Price Index
- Less Than Useful Data: Index of Leading Economic Indicators
- Less Than Useful Data: Weekly Chain Store Sales
Image credit: Photo by Daniel Jensen on Unsplash.
Labels: ideas
Welcome to the blogosphere's toolchest! Here, unlike other blogs dedicated to analyzing current events, we create easy-to-use, simple tools to do the math related to them so you can get in on the action too! If you would like to learn more about these tools, or if you would like to contribute ideas to develop for this blog, please e-mail us at:
ironman at politicalcalculations
Thanks in advance!
Closing values for previous trading day.
This site is primarily powered by:
CSS Validation
RSS Site Feed
JavaScript
The tools on this site are built using JavaScript. If you would like to learn more, one of the best free resources on the web is available at W3Schools.com.
Other Cool Resources
Blog Roll
