External and domestic political determinants of defence spending: a time-series cross-section analysis of EU member states

ABSTRACT

We investigate external and domestic political influences on defence burden of the 28 EU member states, based on time-series cross-section data from 1993 to 2017. Our research is one of the most comprehensive studies on EU members' defence, as it includes many potential explanations. Our findings do not support freeriding by EU members, both NATO allies and non-NATO members, on US commitment in the defence and security field. Relatedly, NATO membership does not come up in the analysis as an instantaneous guarantee for an effective special relationship with the US. However, there is evidence of freeriding by EU members not belonging to the Alliance on EU-NATO member states. Domestic political factors are also important: right-wing parties in government are associated with heavier defence burden, while upcoming elections tend to reduce it. In general, we find that to provide an accurate explanation of military spending, one needs to take into account a multitude of different factors and, moreover, pay attention to conditional effects.

KEYWORDS:

• Military spending
• EU member states
• NATO membership
• freeriding
• defence politics

Acknowledgments

The authors are indebted to the two anonymous reviewers for their valuable comments, which greatly improved the article. Its earlier version had also benefited from a presentation at the UACES 2019 Annual Conference in Lisbon. The authors extend their gratitude to the participants of the event for their relevant suggestions.

Disclosure statement

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

Notes

1 Our dataset misses the earliest years of the 1990s, due to the unavailability of data on Russia’s military spending for 1991 and first-differencing of this variable. In the case of lagged predictors, however, earlier years are also used (1991 and 1992) in order not to lose observations. The data are available from the authors upon request.

2 See in particular Caruso and Di Domizio (2016), Christie (2019), and Spangler (2018).

3 Among control variables, population also turned out non-stationary.

4 This does not mean that both types of terrorism are unrelated, which is, however, a relatively recent field of study (Enders et al. 2011) and beyond the scope of the present research.

5 We owe this insight to one of the anonymous reviewers.

6 In Model 1, however, the marginal effect of an upcoming legislative election in case there is no international involvement is on the borderline of conventionally accepted levels of statistical significance (at 10 per cent). The previously found effect, therefore, may have been determined by EU members’ experience in the 1990s, when defence spending was on average higher. This requires further research to specify our results.

7 We are indebted to one of the anonymous reviewers for this suggestion.