Exogenous variations in public debt and the private output: addressing country heterogeneity and cross-sectional dependence in a large panel

ABSTRACT

Using projections from the IMF’s World Economic Outlook (WEO) database, we explore the links between public debt and the private sector by estimating the response of private output to debt shocks – defined as debt forecast errors. We address several econometric issues through a flexible dynamic panel framework that accommodates slope heterogeneity of the coefficients and cross-sectional dependence. The overall results suggest that positive shocks in public debt are detrimental to the dynamics of private output. Nevertheless, the effect becomes neutral if countries have followed a pattern of fiscal consolidation over the previous five years. The estimates at individual-level are consistent with the general findings, except for a small group of countries. For instance, a positive and significant effect of debt emerges only for one advanced country and seven emerging economies. Although high levels of indebtedness are associated with a rapid decline in the debt coefficients, a nonlinear effect in the form of a Laffer-type curve appears to be unlikely.

KEYWORDS:

• Common correlated factors
• economic growth
• heterogeneous dynamic panel analysis
• private output
• public debt

JEL CLASSIFICATION:

• O47
• H63
• E62

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

¹ The openly accessible WEO editions published over 2006–2009 contain debt projections for the G7 economies only. Table A1, Table A2 and Table A3 report the list of countries for advanced economies, the list of countries for emerging economies and the descriptive statistics, respectively.

² Regarding the period 2010–2021, the only exception occurred in 2011 when WEO forecasts were published in September in place of October.

³ However, the WEO edition published in April 2022 exceptionally did not include predictions on public debt perhaps due to the highly uncertain developments of the pandemic and the magnitude of the fiscal responses of governments.

⁴ As suggested by Chudik et al. (2017), we focus on the variations of debt rather than its levels.

⁵ For instance, even if the discretionary fiscal policy were unchanged over the years, the stock of gross public debt would depend on various macroeconomic factors like interest rate and tax revenue, that are strictly related to cyclical fluctuations in GDP.

⁶ The DCCE models are estimated using the xtdcce2 package developed by Ditzen (2018).

⁷ We consider the p-value equal to 0.05 as a threshold for statistical significance.

⁸ Since the forecasts published in April 2020 do not contain predictions on public debt, the year 2020 is already excluded by construction in baseline estimates where shocks are built on predictions published in April of the current year.