ABSTRACT
The banking system in Tunisia suffers from non-performing loans, fuelling the aversion of banks to credit risk and hindering the financing of SMEs. The research question tackles the ‘transactional lending’ versus ‘relational lending’ relationship between a sample of 10 Tunisian banks observed over 2005–2017 and its clients, and its impact on non-performing loans. The econometric analysis first tests a linear model between non-performing loans and the determinants (Size, Ownership and Proximity) of the lending relationship, which suggests that banks are poorly committed in collecting soft information from their clients. Second, the estimation of a (conditional) threshold effect in a non-linear model between these main determinants, distinguishes two regimes, one above and one below the threshold. Three robust outcomes emerge, the (large) size of most banks in the sample constitutes a barrier to meeting loan applications from SMEs, and enhancing tighter customer relationship with a higher number of branches as well as foreign participation in the share capital of these banks would enable to meet these demands.
Acknowledgements
We gratefully acknowledge the help of Karim El-Acheche (Financial Market Board), Mehrez Ben Slama (University of Monastir, Tunisia) and Nesrine Ben Salah (University of Sfax, Tunisia) in collecting the database. We are thankful to the editor and two anonymous reviewers of the MEDJ for their comments, which helped improving our paper. The usual disclaimer applies.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1 In Tunisia, there are several definitions of SMEs. The national statistical office (INS) takes into account the number of employees (‘Micro’ enterprises with 0–5 employees; ‘Small’ 6–49 employees; ‘Medium size’ 50–199 employees, and ‘Large’ over 200 employees). Furthermore, in the context of the new Investment Law (2016), a new definition for SMEs is based on financial criteria, namely firms with total assets below TND 15 million (roughly EUR 5 million).
2 There is a strong correlation between NPLs and LLPs (0.86).