Serial Entrepreneurship: Learning by Doing?

GAICIL Insights

Forget the garage startup.¹ Most entrepreneurship happens at the hair salon, the repair shop and the corner restaurant. Serial Entrepreneurship: Learning by Doing?, by Francine Lafontaine and Kathryn Shaw, studies that everyday economy and asks whether business experience makes owners better at it. The data cover all establishments started between 1990 and 2011 to sell retail goods and services in Texas. An entrepreneur is the owner of a new business; a serial entrepreneur opens repeat businesses.

The answer is yes, and measurably. Serial entrepreneurs operate 25.6% of businesses. Their businesses last longer: prior experience increases the longevity of the next business opened. If the owner of a new establishment has owned one prior business, the probability of exit for that establishment falls by 7 percent. Results with owner fixed effects suggest past experience imparts valuable business skills. Entrepreneurs learn by doing.

The setting gives the result unusual weight. Small retail entrepreneurs may create fewer jobs per business than large technology entrepreneurs, yet small retail matters because there are so many establishments. Across Texas retail from 1990 to 2011, small owners founded 2.33 million establishments, a count that includes 46,000 small-owner establishments affiliated with national chains through franchising. Large owners, defined as those with 21 or more establishments in the data period, open only 120,000. Restaurants, standard retail stores, repair shops, hair salons: this is entrepreneurship as most people encounter it.

Now introduce artificial intelligence as a substitute teacher. Small business owners increasingly have AI tools for inventory management, pricing, marketing and customer service. Those tools can partly replace the experience serial entrepreneurs gain the hard way. A first-time owner with AI-driven analytics may dodge mistakes that would otherwise be learned only through failure. If AI compresses the learning curve, first businesses could survive longer and the gap between novices and veterans could narrow. The tuition of failure gets cheaper.

Experience will not lose its value, though. Owner fixed effects show skills accumulate within individuals through running businesses. AI may complement those skills, letting experienced owners manage more effectively. Human experience plus AI assistance could make serial entrepreneurs even more productive. The veteran with the new tool may be the most formidable small-business owner the economy has seen.

An unexpected pattern opens a further possibility. Serial entrepreneurship is not especially prevalent in this sector, the authors note, possibly because most retail owners are small proprietors unable to maintain more than one business at a time. AI might relax that constraint. By lightening the managerial burden of each establishment, it could let small owners run several businesses at once, widening the scope for learning across ventures. The one-shop owner could become a three-shop owner.

Policy implications point to the revolving door of business life. If experience raises the longevity of subsequent businesses, barriers to re-entry after failure impose a real cost by discouraging experienced entrepreneurs from trying again. Licensing requirements, compliance costs and bankruptcy rules all shape how easily owners can start anew. Simplifying them lets the economy harvest the skills entrepreneurs accumulate. Every unnecessary license fee is a small tax on learning.

The data infrastructure deserves recognition. The Texas records cover all retail establishments over two decades, letting the authors track owners across businesses and measure returns to experience precisely. Comparable data elsewhere would let policymakers evaluate how AI adoption affects business survival and the returns to experience, an obvious research priority as AI reaches Main Street.

Pulled together, entrepreneurship is a craft developed through practice. Serial entrepreneurs run more durable businesses because they have learned from earlier ones. AI may speed that learning, extend it to first-time owners and let experienced owners do more. Policies that keep the doors of entry open, and let owners try again after failure, ensure that accumulated entrepreneurial skill, amplified by new tools, keeps powering growth across the many small businesses that make up the economy.

A closing challenge: the biggest AI productivity story of the decade might unfold far from Silicon Valley, in thousands of small shops whose owners suddenly have a strategist in their pocket.