Public capital in the 21st century: as productive as ever?

ABSTRACT

The global financial crisis and the euro area sovereign debt crisis that followed induced a rapid deterioration in the fiscal positions of countries across the globe. In the ensuing fiscal adjustment process, public investments were severely reduced in many countries. How harmful is this for growth perspectives? Our main objective is to find out whether the importance of public capital for long run output growth has changed in recent years. To this end, we expand time series on public capital stocks for 20 OECD countries and estimate country-specific recursive vector autoregressive (VAR) models. Results show that the effect of public capital shocks on economic growth has not increased in general, although results differ widely between countries. This suggests that the current level of public investments generally does not pose an immediate threat to potential output. Of course, this could change if low investment levels are sustained for a long time.

KEYWORDS:

• Public capital stock
• economic growth
• global financial crisis
• macroeconomics

JEL CLASSIFICATION:

• E22
• E62
• H54

Supplemental Material

Supplemental data for this article can be accessed here.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ The resulting series for public and private capital stocks, as well as data on real GDP, will be available from the on-line appendix to this article. This document also provides an overview of the exact data source for each series.

² This way we also avoid including spending on military equipment in the investment series, which are assumed not to be important for the production process.

³ The Conference Board Total Economy Database, January 2014 version, http://www.conference-board.org/data/economydatabase/.

⁴ Another potential explanation for our I(2) results is that our capital stock series contain structural breaks. Failure to account for a structural break leads to a reduction in the ability to reject a false unit root null hypothesis Perron (1989). Therefore, we perform Zivot-Andrews and Philips-Perron tests allowing for a break in the intercept and the deterministic trend where appropriate, with the trimming parameter set at 0.10. The evidence remains inconclusive (results not shown).

⁵ Of course, these are quite strong assumptions. We therefore performed a robustness check with different ordering of the variables but this does not affect results much. The impulse response functions for different orderings of the variables are available on request.

⁶ Charts 3 and 4 report one standard deviation (dark grey) and two standard deviation (light grey) bands around the central estimate. Confidence intervals for impulse responses from VAR-models are notoriously wide (see e.g. Runkle 1987), as the uncertainty on each model parameter translates into uncertainty around the impulse response. Therefore Kamps (2005), for example following up on Sims and Zha (1999), presents 68%-confidence intervals.

⁷ Results not shown, but available upon request.

Additional information

Funding

This research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sectors.