How Much Is Too Much Debt for South Africa? A Threshold Nonlinear Autoregressive Distributive Lag (T-NARDL) Perspective


  • Andrew Phiri Nelson Mandela University
  • Asanda Fotoyi Nelson Mandela University



Optimal debt, optimal public investment, economic growth, endogenous growth model, threshold nonlinear autoregressive distributive lag (T-NARDL) model, South Africa


Since the global financial crisis, South African fiscal authorities have acquired debt at a faster rate compared to other economies, which, according to recent growth theory, implies that the economy has a lower debt tolerance or threshold level than previously thought. Our study presents country-specific debt threshold estimates for the South African economy based on reduced form regressions derived from an endogenous growth model of public debt that incorporates public investment as a channel through which debt can influence economic growth. We estimate the reduced form regressions using the threshold nonlinear autoregressive distributive lag (T-NARDL) cointegration model which we apply to quarterly time series spanning from 1960:q1 to 2020:q4. We identify debt thresholds of 47 percent which are much lower than those predicted by previous panel-based studies. Similarly, the corresponding public investment threshold estimate of 2.8 percent is lower than that prescribed by previous literature. However, our study shows that both public debt and public investment are too high to be growth enhancing and we provide policy recommendations based on these findings.