What is a firm census in a developing country? An answer from Ghana

A burgeoning literature in economics uses firm census data to provide explanations for the very large differences in income per capita across countries. Much of this literature takes for granted that the coverage of firm censuses across and within countries is similar. In this paper we use data from four Ghanaian firm censuses conducted between 1962 and 2014 to show that the coverage of each census was very different. Treated as is, the four censuses show dramatic and unbelievable changes in the scale of manufacturing production in Ghana over this period. As a result, we examine and document important changes in what undertaking a “firm census” has meant over 50 years in Ghana, as well as documenting variation in the coverage of firm censuses from several other African countries. We show that it is possible to obtain a believable evolution of the firm size distribution in Ghana over the period for which we have firm microdata, but that this requires substantial work to understand how the coverage of each firm census has varied over time. Our paper shows that the coverage of firm censuses both within and across countries can differ quite dramatically, and that this can impact research that uses firm census data.