New evidence on corruption and government debt from a global country panel

Purpose - The purpose of this paper is to explore the link between corruption and government debt through a regime-based approach. Design/methodology/approach - The empirical analysis makes use of a panel of 120 countries, spanning the period 1999–2015. The study makes use of the Panel Smooth Transition Regression (PSTR) methodological approach, as well as two alternative measures of corruption. Findings - The empirical results document that the relationship between corruption and debt is non-linear, while a strong threshold effect was present as well. Public debt appears to respond faster to a high corruption regime compared to a low corruption regime, while an increase in the size of the shadow economy, government expenses, the inflation rate, interest payments on debt and military expenditure all increased the debt to GDP ratio. By contrast, an increase in GDP per capita, the secondary school enrollment ratio and the ratio of tax revenues to GDP led to a fall in the debt to GDP ratio. The findings survive certain robust checks when the role of the 2008 financial crisis is explicitly considered, as well as when two separate country samples were considered, i.e. developed vs developing countries. Practical implications - Governments should aim to control both corruption and the size of the shadow economy if they really wish to reduce any high levels of their public debt. As debt levels respond faster to high corruption regimes, it is necessary that measures to reduce corruption are complemented by higher GDP per capita growth rates, enrolment rates and higher tax revenues. Originality/value - The novelty of the paper is that it investigates for the first time, to the best of the authors’ knowledge, the presence of non-linearity between corruption and government debt. It proposes non-linear panel cointegration and causality tests, as well as a non-linear panel error correction model that allows for smooth changes between regimes, hence, examining causal relationships in each regime separately.