Rating: 7.7/10.
The Promise and Peril of Entrepreneurship: Job Creation and Survival among US Startups by Robert W. Fairlie, Zachary Kroff, Javier Miranda, Nikolas Zolas
This book presents a primary economic research on startups and entrepreneurship in the United States, not limited to tech startups but encompassing all kinds of new businesses. Many surveys have defined startups as entities that begin once they hire their first employee. However, this definition is biased toward larger and more successful startups because it excludes those without any employees. In this study, startups are defined as businesses that either have at least one employee or are generating revenue. Utilizing this metric, the statistics for startups are not as promising as they initially appear.
Data on very early-stage startups is scarce, and any attempt to survey them is likely to encounter response biases. Nevertheless, this study finds ways to compile information from administrative data, which helps to avoid survey bias. For instance, businesses that initially operate without employees but later on start hiring are represented differently in the data. The majority of startups (about 3/4) do not hire any employees. However, the percentage that does employ staff increases as the startups mature. On average, a startup creates about 0.7 jobs (excluding the founder’s position), and this number gradually decreases as some startups fail. Despite this attrition, the sheer volume of new businesses means that startups collectively make a significant contribution to the economy. New startups tend to pay less than established companies, but those that survive longer tend to pay more, as they become more successful.
A large percentage of startups don’t last very long; over half of them shut down within two years. However, after that, the rates of exits start slowing down. In this study, the way “exits” are defined includes acquisitions, but typically for small startups, this means going out of business. Startups tend to have an up-or-out trajectory, where they either fail quickly (especially the ones without employees), or the surviving startups experience rapid growth. Hiring the first employee is a major milestone for a business, although this is largely uncorrelated with revenue and more of a reflection of the type of business they are in.
As a whole, most startups do not have employees: 90% of startups do not have any employees during their first year, and 75% of all businesses do not have employees. However, some businesses are started without the intention of having employees, so they define a group of “selective startups” that exclude sole proprietorships, which are typically started with the primary intention of acting as a contractor or consultant rather than a business that will hire employees. When analyzing this group of selective startups, the survival rate is slightly higher than the norm, but still relatively low, and if we only count these selective startups, approximately two jobs are created per startup.
The demographic with the highest rate of starting startups are white men aged between 45 and 54 with college degrees. Startups owned by blacks and Latinos are less likely to survive than average, whereas startups initiated by Asians are more likely to have employees and more capital. The most common source of initial funding is the entrepreneur’s own personal savings, while any form of grant funding or loans are relatively rare. The book concludes that when the data from startups are properly analyzed – not just the ones that have already hired employees – the statistics are quite dismal, so the authors argue against policies by governments that try to attract startups, as the actual amount of job creation is fairly bleak.
Overall, this book provides a fairly comprehensive analysis of the startup landscape and its outcomes. As the data comes from the entirety of the US census, there is no chance of selection bias skewing reality. One aspect I found surprising was the uniformity of the statistics across various industries: regardless of the statistic being analyzed, there was rarely any deviation from agriculture to construction to tech services. This book focuses on all industries, many of which are small businesses, so the array of businesses analyzed leans more towards small businesses, as opposed tech startups (which we commonly think of when discussing startups). For the tech industry especially, I’d be interested in statistics about funding (particularly venture capital), but this study does not include any data on these aspects and thus cannot analyze them.
