Testing fiscal sustainability in OECD countries: new evidence from the past centuries

ABSTRACT

Using three complementary approaches that examine the fulfilment of the intertemporal budget constraint while avoiding explosive behaviour, this study examines the stationarity and the sustainability of public finance for six industrial countries over the period spanning from 1870 to 2017. According to Fourier DF unit root tests, time-varying fiscal reaction functions and threshold reaction functions, we find that longer-run debt sustainability is not rejected for the UK, Sweden, and for the US. The evidence is somewhat equivocal for Canada, Italy and Portugal.

KEYWORDS:

• Fourier DF unit root test
• debt sustainability
• public debt
• primary balance
• time-varying
• threshold regressions

JEL CODES:

• C22
• E62
• H62

Acknowledgment

The authors are grateful to Katharina Priedl for her precious language advice.

Disclosure statement

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

Notes

¹ U.S. Office of Management and Budget and Federal Reserve Bank of St. Louis. https://fred.stlouisfed.org/series/GFDEGDQ188S, October 12, 2022.

² Afonso (2005), Afonso and Rault (2010) and Afonso and Jalles (2014).

³ https://www.macrohistory.net/database/.

⁴ https://www.jamelsaadaoui.com/wp-content/uploads/2022/10/appendix-22-April-7.pdf.

⁵ Enders and Lee (2012) suggest using single frequency in Fourier Dickey-Fuller unit root tests.

⁶ According to Enders and Lee (2012), minimizing the sum of squared residuals is equal to maximizing the F test by imposing the restrictions $\mathit{a}=\mathit{b}=0$ on equation (4)(4) $y_t=\mathit{c}+\mathit{a}.\sin \left(\frac{2\pi k^s\mathit{t}}{\mathit{T}}\right)+\mathit{b}.\cos \left(\frac{2\pi k^c\mathit{t}}{\mathit{T}}\right)+\mathit{\beta }{y}_{\mathit{t}-1}+{\epsilon }_t$(4) . For more details, please refer to Enders and Lee (2012).

⁷ Using the Engle-Granger’s cointegration test between the primary surplus and the public debt, we reject the null of no cointegration at the 1% level in Sweden, the US, Canada and the UK. We reject the null at the 5% level in the case of Italy. We fail to reject the null at the 10% level for Portugal. The results are available upon reasonable request.

⁸ Interesting extensions of this work could be the use of the Fourier quantile unit root test as in Bahramian and Saliminezhad (2021) or the Fractional frequency flexible Fourier form (FFFFF) for panel cointegration test as in Olayeni, Tiwari, and Wohar (2021).

⁹ We use the initial/lagged debt as the threshold variable as in Fotiou (2022).