Real-time uncertainty in budget planning: evidence from euro area countries¹

1 The views expressed in this article are those of the authors and do not necessarily represent the views of the Bank of Finland.

Abstract

Using rich panel data including potential output for euro area countries, we analyse budget balance forecasts and their errors. We find that budget balance forecasts are systematically biased and subject to mean reversion (tendency towards more balanced budgets). A robust result is that errors in budget balance nowcasts contribute to errors in budget balance forecasts. In addition, we find that nowcasted macroeconomic conditions can affect over-optimism in budget balance forecasts. Overall, our results emphasize the central role of nowcasting in the EU fiscal framework.

Keywords:

• Fiscal policy
• real-time data
• economic crisis

JEL Codes:

• E62
• E32

Disclosure statement

No potential conflict of interest was reported by the authors.

Acknowledgements

The authors would like to thank the Editor and two anonymous Referees for useful comments. We also thank Athanasios Orphanides for his valuable suggestions. We are grateful to participants at XXXVIII Finnish Economic Association’s Annual Meeting (2016), INFER Workshop on European Integration in the Aftermath of Debt Crisis (National Bank of Slovakia, 2016), Banco de España conference on Fiscal Sustainability, XXI Century (2016), the 33^rd CIRET Conference (2016), and the Finnish Economic Association’s XXXIX Annual Meeting (2017) for constructive comments.

Notes

1 The views expressed in this article are those of the authors and do not necessarily represent the views of the Bank of Finland.

1. See e.g. Kukk and Staehr (2015) for EU fiscal framework.

2. The level of potential output is estimated using e.g. statistical techniques or economic relationships such as the production function, or combined methods. The estimates are characterized by significant uncertainty, which is largest for the most recent years. IMF estimates potential outputs using several methods, for example a production function approach (see De Masi 1997).

3. See Orphanides and van Norden (2002) for challenges to measure output gap in real time.

4. See Cimadomo (2014) for a survey of real-time fiscal policy literature.

5. Jonung and Larch (2006) and Strauch, Hallenberg, and von Hagen (2004) find that governments tend to make over-optimistic growth forecasts. Baldi (2016) builds a two-period model to examine the effects of fiscal policy rules on budget deficits and forecasting biases in official budget outlooks.

6. Austria, Belgium, Finland, France, Germany, Greece, Ireland, Italy, the Netherlands, Portugal and Spain.

7. Our sample is restricted to 2004–2014. Particularly the output gap estimates are still likely to be subject to substantial revisions in the following vintages.

8. Beetsma and Giuliodori (2010) analyse fiscal plans and their implementation separately in OECD countries.

9. The average nowcast errors of budget balances were clearly positive for the GIIPS countries in post-crisis period. For both country groups growth nowcast errors were slightly negative in the pre-crisis years and marginally positive in the after-crisis years. Output gap nowcast errors have been systematically negative and potential output growth nowcast errors positive.

10. Period t + 1 refers to the forecasted budget year. For example, the forecasted budget balance change in 2012 (ΔBB_{2012,2011|2011} = BB_2012|2011 – BB_2011|2011) is explained by BB_2011|2011 and OG_2011|2011 (Equation 1(1) ) or by BB_2011|2011 and GDP_2011|2011 (Equation 2(2) ).

11. In terms of a simple AR(1) process s _t  = αs _t−1 + ɛ _t , mean reversion means that 0 < α < 1 (see Hillebrand 2003; Mpatswe, Tapsoba, and York 2011). For mean reversion in fiscal policy, see also Lau and Baharumshah 2009.

12. According to Bernoth, Hallet, and Lewis (2008) poolability of fiscal policy reaction functions for 14 European countries cannot be rejected.

13. In final-data estimations of Equations (1(1) )–(2(2) ), mean reversion is systematically stronger than in Table 1 (available upon request).

14. These two components cannot be separated in Frankel and Schreger (2013) data.

15. Avellan and Vuletin (2015) examine implications of output forecast errors on fiscal procyclicality. They define the forecast error of output growth as the difference between ex-post and predicted changes in output.

16. Turner et al. (2016) decompose output gap revisions into revisions in real GDP level and potential output level.

17. These two variables represent decomposition of nowcasted output gap change.

18. Due to multicollinearity problems, actual government revenues and expenditure could not be used as controls.

19. Detailed information can be downloaded from https://ec.europa.eu/info/publications/fiscal-rules-database_en.

20. Broadly, the estimation results in Tables 1–6 remain qualitatively unchanged if robust standard errors (with a correction for autocorrelation and heterogeneity) are used to the extent estimation is feasible (available upon request).