There are a huge number of policies devoted toward increasing the number of small businesses. The assumption, it seems, is that small businesses are generating more spillovers than large businesses, in terms of innovation, increases in the labor match rate, or indirect welfare benefits from creative destruction. Indeed, politicians like to think of these “Joe the Plumber” types as heroic job creators, although I’m not sure what that could possibly mean since the long run level of unemployment is constant and unrelated the amount of entrepreneurial churn in whatever economic model or empirical data you wish to investigate.
These policies beg the question: are new firms actually quick-growing, innovative concerns, or are they mainly small restaurants, doctor’s offices and convenience stores? The question is important since it is tough to see why the tax code should privilege, say, an independent convenience store over a new corporate-run branch – if anything, the independent is less innovative and less likely to grow in the future. Erik Hurst and Ben Pugsley do a nice job of generating stylized facts on these issues using a handful of recent surveys of firm outcomes and the stated goals of the owners of new firms.
The evidence is pretty overwhelming that most new firms are not the heroic, job-creating innovator. Among firms with less than 20 employees, most are concentrated in a very small number of industries like construction, retail, restaurants, etc, and this concentration is much more evident than among larger firms. Most small firms never hire more than a couple employees, and this is true even among firms that survive five or ten years. Among new firms, only 2.7% file for a patent within four years, and only 6-8% develop any proprietary product or technique at all.
It is not only in outcomes, but in expectations as well where it seems small businesses are not rapidly-growing innovative firms. At their origin, 75% of small business owners report no desire to grow their business, nonpecuniary reasons (such as “to be my own boss”) are the most common reason given to start a business, and only 10% plan to develop any new product or process. That is, most small businesses are like the corner doctor’s office or small plumbing shop. Starting a business for nonpecuniary reasons is also correlated with not wanting to grow, not wanting to innovate, and not actually doing so. They are small and non-innovative because they don’t want to be big, not because they fail at trying to become big. It’s also worth mentioning that hardly any small business owners in the U.S. sample report starting a business because they couldn’t find a job; the opposite is true in developing countries.
These facts make it really hard to justify a lot of policy. For instance, consider subsidies that only accrue to businesses below a certain size. This essentially raises the de facto marginal tax rate on growing firms (since the subsidy disappears once the firm grows above a certain size), even though rapidly growing small businesses are exactly the type we presumably are trying to subsidize. If liquidity constraints or other factors limiting firm entry were important, then the subsidies might still be justified, but it seems from Hurst and Pagsley’s survey that all these policies will do is increase entry among business owners who want to be their own boss and who never plan to hire or innovate in any economically important way. A lot more work here, especially on the structural/theoretical side, is needed to develop better entrepreneurial policies (I have a few thoughts myself, so watch this space).
Final Working Paper (RePEc IDEAS) which was eventually published in the Brookings series. Also see Haltiwanger et al’s paper showing that it’s not small firms but young firms which are engines of growth. I posted on a similar topic a few weeks ago, which may be of interest.
A Fine Theorem 
Scott Berkun has also made contributions to the body of research on the subject of small businesses, innovation and jobs creation. His conclusions were, as conventional thinking understands it, counter-intuitive – just as those describe above are thought to be.