Cause and Effect

This graph from small business surveys (btyb the National Federation of Independent Businesses) has been going around the blogosphere. Matthew Yglesias and Paul Krugman comment.

Take a close look. See what I see?

The graph is nonsense.

First, the obvious: what exactly is the difference between "poor sales" and "competition from large business?" Isn't the latter just a subset of the former? I have a hard time believing there's a business owner out there saying "Well, I've encountered a lot of competition lately, which is a problem, but my sales haven't gone down." That's why competition's a problem for small businesses: they reduce their sales.

Second, really all problems are sales problems. Any expense that seems too large can be fixed with more sales. No one cares about a high cost of insurance if you're making money left and right nor more than I would care about the savings from generic brands if I made six figures a year.

What the survey really measures is what disappointing economic news is popular. In the late 1990s, during the boom, good labor was hard to come by so labor quality was a concern. During the 80s to the mid 90s, anti-government rhetoric was at its peak so the culprit was regulations. Health care costs dominated the discussions during the 2000s. Now, all the talk is about how people aren't buying things so they focus on sales.

If a business is struggling, the owner will focus first on what's being talked about. There's nothing wrong with that as long as they check it against their own business (reasonable assumption). But the "sales" option pollutes the data. If taxes were lower, wouldn't the sales problem go away? If there were fewer requirements or cheaper insurance, wouldn't that increase your bottom line just like sales would? Essentially this survey asks:

How do you increase X?

X=A-B

1. Increase A
2. Decrease B

It's both!

Update: Russ Roberts and Arnold Kling weigh in.