The effect of micro-firm definitions on statistical inferences

Abstract

Regulations govern the choices that businesses make, though regulations may vary between firms of different sizes. Such variation should benefit entrepreneurs/owner-managers, who often operate (at least in the beginning) smaller businesses with lighter regulatory burdens. However, there is no universal definition of firms’ sizes (e.g. micro-firm, SME, etc.), so authors of policy and academic research are free to specify whatever definition they choose, or none at all. Intuitively, it makes sense that such inconsistency in the definition of firms will lead to different data being included in statistical models and thus, different results and policy recommendations. However, this has not been explicitly explored. In this paper we explore this notion using the case of micro-firm definitions in academic research, and then use the example of the UK (and hence EU) to illustrate the apparent difficulty in applying the “full” multi-faceted legal definition. We utilise data from English micro-firms in the tourist accommodation sector to develop a simple econometric model to apply subsets of real-world research survey data in line with different micro-firm definitions, and then compare the outcomes. We find significant differences in results, the likely inferences of those results, and thus, the policy recommendations that would be made, thereby highlighting the importance of firm size definitions. We conclude with several recommendations for how best to define micro-firms and the use of definitions more broadly – in particular that authors are advised to clearly specify what definition they have employed, so that future researchers/readers can understand accordingly.