Survival determinants for Brazilian companies, 1996 to 2016

Abstract

This paper examines the survival determinants of small- and medium-sized Brazilian companies from 1996 to 2016. A nonparametric survival model (Kaplan–Meier hazard function) and a semiparametric model (Cox proportional model) were used to study a period larger than the ones in previous studies. Applying these methods to a sample of 43,865 industrial firms, we analyzed survival rates by size, region and technological intensity of companies, in order to understand which elements influenced the survival of firms. Our main findings was that small companies had the lowest survival rates compared to medium and large companies, in all groups of sectors classified by technological intensity. However, for small companies, the highest survival rate of those classified as having medium technological intensity and those located in Northeast region of Brazil stand out. Both are interesting findings because, first, they go against common sense that assumes that participation, growth, and survival of small companies are linked to low technology sectors, and second, that survival is not linked to deep Brazilian regional inequalities. In addition, the semiparametric survival analysis model showed that the variables that best explain the greater probability of survival of Brazilian small- and medium-sized enterprises (SMEs) are presence in less concentrated markets, sectoral growth, productivity levels and presence in sectors with high technological intensity.

Appendix

The Annual Industrial Survey (PIA-Enterprise) is carried out by IBGE and, in the current model, began in 1996. The PIA aims to identify the structural characteristics of industrial companies (classified in sections B and C of NACE 2.0) in Brazil. The periodicity of the survey is annual. The geographic scope of the research is national, with results published for Brazil, Large Regions and Federation Units. General information for the total industry, which does not infringe on confidentiality laws on company data, is made publicly available in consolidated reports and data from the IBGE data library. The PIA presents a variety of economic and financial information from companies, such as information on employees, costs and expenses, personnel expenses, gross and net revenues, value of industrial transformation, among other aspects of companies with one or more employees (BRASIL 2019a).

PIA includes companies actively registered in the Central Register of Enterprises (CEMPRE). CEMPRE is a database of registration and economic data of companies in Brazil. The database combines information from IBGE's Industry, Construction, Commerce and Services surveys with data from the Cadastral Maintenance System of the Central Register of Companies (Simcad) and the Annual Social Information Report (RAIS). CEMPRE contains a great deal of information on companies, such as the corporate name and the code of legal nature of the companies (BRASIL 2019b).

It is noteworthy that IBGE allows the request of individualized information by companies and special tabulations from PIA and CEMPRE. However, for access to microdata from these databases a research project must be approved by the IBGE Evaluation Committee, which allows the use of the encrypted microdata database in an assisted manner in the IBGE's Restricted Access Room (SAR), located in Rio de Janeiro. It should be noted that this was the procedure adopted in this paper, and the variables created from the microdata and the estimations of the models were carried out at SAR/IBGE, in order to achieve the main goals of the paper.

Table 10 shows the list of data used from the PIA and CEMPRE from 1996 to 2016 for constructing the variables used in the survival models.

Table 10 List of variables extracted from PIA and CEMPRE (1996–2016)