PurposeThis study aims to investigate the effect of regulatory pressure on discretionary capital management measured with the discretionary loan loss provisions (DLLP) in public (PuBs) and Private (PrBs) banks in Tunisia. Three variables are used to proxy the regulatory capital constraints: (1) the change in capital requirements, (2) the beginning of the year capital ratio (3) and the end of year adjusted capital ratio.Design/methodology/approachTo address our objective, we provide in a first step the DLLP estimation as done by Shantaram and Steven (2021). Then, in a second step based on hand-collected panel data on the 12 commercial Tunisian banks, linear dynamic model with interaction variables is conducted to discriminate between PuBs and PrBs behavior. The generelized method of moment (GMM) estimation is applied to show if the PuBs and PrBs behave differently to regulatory capital pressures. For robustness check, the discriminant analysis and the nonlinear probit and logit models are considered in a third step.FindingsThe three capital constraints affect differently the discretionary behavior of Banks. First, an increase in capital requirements makes PrBs under pressure to reduce their DLLP, which is not the case for PuBs. Second, a low capital ratio at the beginning of the year makes strong pressure on PuBs to reduce their DLLP. Third, neither PrBs nor PuBs decrease their DLLP to improve the end of year-adjusted capital ratio. The discretionary behavior of PrBs is influenced by pressures to appear well-capitalized while the behavior of PuBs is influenced by pressure to enhance their capital positions. These results are well strengthened by the discriminant analysis and the nonlinear probit and logit model investigations.?Originality/valueA few studies examined incentives based on the regulatory theory in Tunisian banks and were carried out within static linear models. Contrary to Elleuch and Taktak (2015) who tested the regulatory incentives following the publication of the (IMF, 2002), this paper tests, within linear dynamic model and nonlinear model, the effect of national prudential rules on capital management between 2006 and 2016.