You Can Guess Workers Are Scared When New Start

There’s been a big surge in filings for new businesses. The data, business applications, is from the Census Bureau and, ironically, may have as much bad to say about the economy as good. Possibly even worse.

Before getting into analyzing the data, let’s understand it, because it’s a little confusing.

The Internal Revenue Service gets applications (Form SS-4) for business tax identifications called employer identification numbers, or EIN. It only requires a simple form that requires various information. One is the intended business structure, like a sole proprietorship, partnership, corporation, S-corporation, or limited liability company (LLC). Also needed is the Social Security number or taxpayer ID number of the responsible party in control of the organization. If a third party does the work, maybe a friend or family member, lawyer, tax professional, organization, or company, they need explicit authorization.

It’s pretty easy, happens online, and doesn’t mean anything other than there is a tax identification number for that business. When it comes to file taxes, the business uses the EIN rather than a Social Security number.

The Census Bureau takes this data for all types of businesses (using the North American Industry Classification System, or NAICS) and reports it. Below is a graph, via the Federal Reserve Bank of St. Louis, of monthly business application data.

However, there area two distinct types of applications. One category is called high-propensity, which means having a strong likelihood of turning into businesses with eventual payroll. They include applications from a corporate entity, applications that say the new business will hire employees, ones that provide a planned data for first wages paid, or applications for businesses in specific industry codes with that tend to hire people. Those industries include accommodation and food services; construction; manufacturing; retail; professional, scientific, and technical services; educational services; and healthcare.

Non-high-propensity applications are those without the same likelihood of eventual payroll. That doesn’t mean they won’t, but it does include many of the businesses that someone will start on the side.

The graph below shows the lines for all applications (blue) and high-propensity (green). The space between the two lines (the difference between the levels) is the number of non-high-propensity applications.

The single largest jump in business applications was from April 2020 to July 2020. Absolutely stunning. Also stunning, that in the second quarter of 2020, the unemployment rate jumped to 13.0%, the highest average unemployment rate jump on record. The two, especially the non-high-propensity applications, are certainly correlated. Stating an absolute causality isn’t mathematically possible, but given the firings, the remote working, the fear, it also would require an impossible degree of credulity to assume that people weren’t frightened into starting businesses. They needed alternatives to the normal expectations of what the economy could offer.

That brings us to the current uptick in non-high-propensity business formations that started in January 2025. Various factors might correlate. One is the sudden explosion of generative artificial intelligence.

Bloomberg’s reporting suggests that AI is fueling a startup boom (there is a free link until 19-July-2026). The theory is that the capabilities of AI is enabling many people to start new businesses. They wrote that “the AI-enabled startup surge is so robust it should yield many lasting companies even after the weaker ones peter out, said Aaron Terrazas, an economist who works with small business services firm Gusto.”

That might be one factor, but there’s also the chance that the fear of 2020 is here in another form. Much of the conversation regarding AI has been about how many jobs would be eliminated. The concern seems both broad and deep. When the population hears one expert after another speculate on how much of the national work force would become, as the English say, redundant, looking for an escape is a rational response. There could even be a crossover, with people afraid of the AI impact using AI to create some business that might provide an economic haven.

Again, there’s no proof of fear being a significant reason for the surge in business formations but pretending that it couldn’t be seems like a mistake.

Some will make it. Many won’t.