The influence of key risk drivers on the performance of SMMEs in the manufacturing sector in KwaZulu-Natal

Abstract

Small Medium and Micro Enterprises (SMMEs) have been shown to be key contributors to sustainable socio-economic development, constituting more than 90% of private sector enterprises around the world. Inevitably, many developing countries continue to explore means aimed at enhancing the performance of small enterprises. However, despite the implementation of various interventions the failure rate of SMMEs in South Africa particularly KwaZulu-Natal (KZN) is disturbing, reaching up to 80% in the first year of operation. As such, to contribute to addressing this challenge, the study adopted a novel approach to establishing and modelling manufacturing SMMEs performance drivers. Utilising a unique three-year panel dataset, key risk drivers were established and modelled via R software version 3.6.3. To achieve the study objectives, a series of independent but related papers were carried out and these make up the main chapters of this thesis. The first chapter provided the background to the study. The second chapter explored the characteristics of manufacturing SMMEs based in KZN province. The findings showed the complexity of firm performance, indicating the heterogeneity between rural and urban based SMMEs. The next chapter, harnessing Stochastic theory aimed to establish whether SMMEs’ growth performance followed a random walk. The theoretical model was rejected, thus providing a basis for the claim that firm performance is a function of certain risk drivers. Armed with findings from the previous papers, the investigation of key drivers impacting the sales and growth performance of manufacturing SMMEs ensued. The fourth chapter, harnessing the Penrosian and strategic management theories established key drivers of SMMEs’ performance. The fifth chapter concerningly, revealed that SMME owners in the manufacturing sector are largely not aware of the impact of established drivers on their enterprises’ performance. In the next chapter, a total of five machine learning algorithms were evaluated of which Artificial Neural Network and Support Vector Machines were identified as the best algorithms for SMME sales and growth predictive modelling, respectively. The two algorithms informed the development of a dedicated machine learning application for SMMEs that’s being commercialised through the DUT Technology Transfer and Innovation Directorate.