Firm survival and innovation: direct and indirect effects of knowledge for SMEs

Abstract

ABSTRACTThis paper explores the effects of innovation on firm survival by using data from a representative survey of small and medium manufacturing firms in the province of Salerno, Italy. We innovate upon the literature by (a) comparing the impact of different sources of internal and external knowledge (including universities) on the probability of firm survival; (b) assessing the mediating impact of the human capital of workers and entrepreneurs on learning from these knowledge sources. Finally, we measure the impact of different types of innovation on firm survival. Our evidence upholds the link between innovation and firm survival, particularly for product and organisational innovation. Results regarding the impact of different sources of knowledge highlight the roles of employee training, the human capital of entrepreneurs and workers and the productivity of university departments providing relevant knowledge. Other elements of external knowledge, such as proximity to the University of Salerno or being in the city of Salerno, are significant facilitators of survival only if mediated through high levels of the human capital of entrepreneurs and workers.KEYWORDS: Internal and external knowledge; absorptive capacitySMEsuniversity collaborationhuman capitalJEL CLASSIFICATIONS: L20O3D22I2 Disclosure statementNo potential conflict of interest was reported by the author(s).Notes1 There is a common presumption that firm survival is ‘good’. However the literature on firm exit emphasises the distinction between voluntary entrepreneurial closure and failure (Bates Citation2005; Coad Citation2014; DeTienne, McKelvie, and Chandler Citation2015; Headd Citation2003; Khelil Citation2016; Wennberg, Delmar, and McKelvie Citation2016).2 Hyytinen, Pajarinen, and Rouvinen (Citation2015) and Fernandes and Paunov (Citation2015) also apply probit models to the study of innovation and survival. Their analyses, however, do not allow for the joint determination of innovation and survival and hence do not yield estimates of the direct and indirect effects of innovation on survival.3 It is worth noting that in 2016, 93% of EU manufacturing firms had fewer than ten employees (Muller et al. Citation2017).4 Note however that Trushin and Ugur (Citation2021) find that firms in hazardous environments can mitigate the detrimental effects of these environments on survival through R&D expenditure.5 Interestingly, Holl, Peters, and Rammer (Citation2023) report a similar spatial range for German manufacturing firms in a related context (the impact of patents on innovation persistence).6 This finding is consistent with the theoretical model of Desmet and Parente (Citation2012), where a large firm size is essential for the introduction of effective cost-saving technologies.7 The survey was conducted by the Centre for Economic and Labour Policy Evaluation at the University of Salerno and funded by the Sichelgaita Foundation in Salerno.8 LMAs, akin to British travel to work areas, are defined by the Italian Statistical Office (ISTAT) as subregional geographical areas where a given group of people live and work.9 Becattini (Citation1990, 38) defines an industrial district as follows: ‘A socio-territorial entity which is characterised by the active presence of both a community of people and a population of firms in one naturally and historically bounded area. In the district, unlike in other environments, such as manufacturing towns, community and firms tend to merge’. The two districts in the province of Salerno compare with six districts in the Campania region, and with 156 districts in all Italy (ISTAT Citation2005), providing further evidence in favour of a large and well-articulated economy.10 According to the response rate, we obtain a total representativeness of 98.87% with a total error of 0.25 and an error of 0.5 at the economic activity sector level (firms interviewed = 462, population = 11,805). For additional technical details regarding sample design, see Amendola et al. (Citation2013).11 In SMEs, the manager and the owner (or the main shareholder) are almost always the same person (Lobonțiu and Lobonțiu Citation2014), which we indicate as the entrepreneur.12 We exclude six firms from the OPIS survey that have missing values for the variables used in this study.13 A final remark related to the use of binary variables is that various regressors for which information was available as continuous variables (e.g. firm size or entrepreneur's age) are converted to categorical dummies. We do this to reduce the multicollinearity among the regressors, a point we will come back to in the following subsection.14 We have 14 two-digit ATECO sectors that are gathered into seven broader categories. The four Pavitt sectors are built using three-digit ATECO sectors that cut across the seven ATECO categories. Hence, there is no collinearity between the Pavitt and ATECO dummies. Further information is provided in the Supplementary Material (Table A2).15 Below we provide a brief introduction to these probit models. Following a suggestion from an anonymous referee, further details about these models are provided in the Supplementary Material.16 The mean VIF was less than 2.5 in all cases and was also below 5 for most variables in all equations (for three variables it is higher than 5 but lower than 7.5).17 For discrete variables, the sum of the direct and indirect effects does not equal the total marginal effect because the procedure for calculating the direct and indirect effects is designed for continuous variables and is used with some approximations for discrete variables.18 Maximum likelihood methods are notoriously affected by problems of convergence in the presence of a reduced number of degrees of freedom (see, e.g. Altonji, Elder, and Taber Citation2005).19 We thank two anonymous referees for suggesting these robustness checks.20 The nature of our dataset prevents a deeper analysis of risk preference along the lines of Hyytinen, Pajarinen, and Rouvinen (Citation2015).21 We have attempted interacting the entrepreneur's high educational level and the share of graduate workers with all available sources of internal and external knowledge, including the dummies for employee training, MIPAAF lab, district (which controls for important contacts with rivals and customers), and the productivity of university departments. However, to ensure a more compact presentation, we only report specifications that have significant interactions, either with the entrepreneur's high educational level or with the share of graduate workers.