Thorsten Beck, one of the world’s major experts in SME finance, argues that the “size” of the SME sector in an economy does not really matter for economic development. What counts is its dynamism:

Policy efforts targeted at SMEs have often been justified with arguments that (1) SMEs are an engine of innovation and growth and (2) they help reduce poverty because they are labor-intensive and thus stimulate job growth, but (3) they are constrained by institutional and market failures. Cross-country, country-level, and microeconomic studies, however, do not support these claims. One study shows that, although faster-growing economies have a higher share of SME employment in their manufacturing sectors, it is not the size of this segment that drives growth.

… The empirical evidence thus points to the entry of new firms—which are mostly small at entry—and the possibilities for successful SMEs to grow as decisive. It is not the size of the SME segment but their dynamism that helps economic development. Recognizing dynamism rather than size of the SME sector as the source of economic development identifies another important distinction that analysts should make among the firm population: informal microenterprises, the establishment of which is often the result of a lack of alternative economic opportunities, and small formal enterprises, some of which might have high growth potential.

I do not entirely agree on the last point. Although it is true that many businesses operate for survival, my experience in Kenya is that the distinction between “informal microenterprise” and “small formal enterprise” is just not clear in the real economy.

I see this blurry boundary as a sign that dynamism can occur at an extremely small scale, namely the “micro-to-small” (MSE) segment. If we go to an even smaller scale, the “graduation” of some businesses from “self-employment” to “micro-enterprise” is also very interesting. In other words, whenever a business is growth-oriented, no matter how “micro” it is, should get more attention

More (from Thorsten Beck’s paper) here