Determination of the Regional Impact on Innovation with an Ordinal Logit and a Multilevel Analysis

Although innovation is considered as an engine for the economic growth of the firm, it remains a risky and complex process. However, in the current context of international competition, firms can’t survive without innovating. The identification of the main determinants of an innovation effort by a firm is a subject that has been much discussed in recent studies. This research aims to identify the main determinants of the innovation decision of Tunisian companies. It evaluates the impact of R&D efforts and other factors reflecting the company's ability to cooperate on the various dimensions of innovation (product, process, organization and marketing). In addition, we will check up whether the 24 governorates of Tunisia contribute equally to innovation within companies. To do this, we used data on a sample of 620 firms observed by the Ministry of Higher Education and Scientific Research in 2008 covering 24 governorates. Methodologically, we propose in the first place an ordinal logistic regression in which the endogenous variable expresses the number of type of innovation made by each firm in the survey (one type, two types, three types or four types of innovation). The exogenous variables selected are essentially the activities of R&D of the company, cooperation with research centres’, the qualifications of the employees, the foreign capital share and the turnover. The results of estimating a multi-level model shows that innovation differs from one governorate to another. The results of the estimation of an ordinal logit and a multi-level model show up that expending on research and development is a key determinant of innovation (odds ratio equal to 5.3) and that it differs from one governorate to another (the variance of weights assigned to level 2 (governorates) equal to 27.7%)