Does government debt crowd out capital formation? A dynamic approach using panel VAR

We estimate a panel vector autoregression model using data for 127 countries from 1980 to 2017 in order to identify the dynamic relationship between public debt and the growth of capital formation. Our results provide evidence for the crowding-out effect of government debt and the subsequent drop in output growth. The impulse response functions for sub-samples of countries reveal two remarkable results. First, the response of capital formation to a shock in debt appears to be consistent across different income categories of countries, and does not depend on the size of debt-to-GDP ratio. Second, the magnitude and persistence of this effect is lower for the high-income countries. The results obtained are robust to various model specifications as well as for alternative proxies of debt and capital formation. • We estimate a panel VAR to evaluate the effect of public debt growth on capital formation. • Our results provide evidence for the crowding-out effect of debt and a drop in income growth. • There is some variation in the persistence of this effect across countries. • The findings do not depend on the size of public debt-to-GDP ratio. • The results are robust to various model specifications and proxies of debt and capital formation.