Record startup filings mask decline in businesses planning to hire

Record numbers of Americans are applying to start new businesses, fueling speculation that artificial intelligence is ushering in a new era of entrepreneurship. However, a recent Bloomberg opinion argues that while AI is making it easier than ever to launch a company, there is little evidence so far that it is increasing the number of startups with the potential to become major employers or economic drivers.

According to data released by the U.S. Census Bureau on July 9, the United States recorded 531,423 seasonally adjusted business applications in June, capping what was the strongest first half of a year in the 22-year history of the series. The surge has been widely interpreted as an early sign of an AI-powered entrepreneurial boom.

The Bloomberg opinion, however, suggests that headline application figures may not tell the full story. It points to another indicator collected during the registration process: whether applicants specify when they expect to begin paying wages to employees. According to the piece, filings that include plans to hire workers have been declining for nearly two years and continue to fall by double-digit percentages, even as total applications reach record levels.

The opinion piece cites research by Jorge Guzman of Columbia Business School and Scott Stern of the Massachusetts Institute of Technology (MIT), who examined 28 million business registrations across 32 U.S. states. Their study found that only about 0.07% of newly registered firms reached an initial public offering or a significant acquisition within six years. More than one-third of those successful companies came from the top 1% of firms identified as having high growth potential at the time they were registered.

Bloomberg Opinion also references research by economists Vincent Sterk, Petr Sedláček and Benjamin Pugsley, who concluded that much of a startup's future growth is determined before it officially begins operating, with a relatively small number of high-potential firms accounting for a disproportionate share of economic growth.

The article argues that AI is reducing the cost and effort required to perform routine entrepreneurial tasks, such as drafting legal documents, writing marketing materials, coding basic software and preparing business plans. It cites a working paper by Junhui Jeff Cai and co-authors examining 12.8 million Chinese business registrations, which found that new business formation accelerated in regions with strong AI expertise following the emergence of ChatGPT, particularly among smaller firms.

However, the Bloomberg article contends that these efficiencies primarily lower administrative barriers rather than addressing the qualities that distinguish successful startups from unsuccessful ones.

The article also highlights studies examining AI's impact on worker performance. Research by Erik Brynjolfsson of Stanford University, Danielle Li, and Lindsey Raymond of MIT found that AI assistants significantly improved productivity among less experienced customer support workers, while providing only modest benefits for experienced employees.

Meanwhile, a separate study involving 640 Kenyan entrepreneurs who received GPT-4-based business mentoring found no overall increase in business performance. According to the research cited in the Bloomberg Opinion piece, stronger entrepreneurs tended to benefit from the AI advice, while weaker businesses often performed worse, despite receiving similar recommendations.

AI is making business creation easier and boosting the number of registrations, but there is not yet clear evidence it is producing more high-quality startups. Instead, the opinion piece suggests that the long-term economic impact will depend less on the ease of forming companies and more on whether those businesses ultimately grow, hire employees and generate lasting economic value.

By Sabina Mammadli