When Military Interventions Decrease Military Power. Evidence from the French Case

ABSTRACT

Military interventions create dilemmas for military organizations which must balance security-related tasks (such as peacekeeping or crisis management operations) and defense-related tasks involving deterrence and the preparation for high-intensity operations. This article specifically examines the trade-off between security and defense tasks through an original analysis of the impact of the intensification of the French armed forces’ operational tempo on the overall availability of military equipment. We argue that the intensification of military interventions generates gaps in a country’s military capabilities. More precisely, an excessive operational tempo, understood as an unsustainable level of deployment given regeneration capabilities, can reduce the ability of military organizations to generate operationally effective forces. Through an original methodological approach, using a multilevel econometric model that estimates a specific dimension of military power applied to the French case, this article contributes to the literature on the strategic utility of military interventions by examining their structural impact on the armed forces.

KEYWORDS:

• Military interventions
• operational tempo
• availability
• multilevel models

JEL CODE:

• H59
• D29
• D74

Disclosure statement

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

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Notes

1. Notably, this analysis does not consider the costs associated with internal operations such as Sentinelle (since 2015) or Vigipirate. We argue that the cost of such operation is relatively smaller than that of military interventions : in 2015, the Sentinelle operation cost approximately EUR 170 million (and approximatively the same amount in 2016). In addition, such operations mainly imply land forces and their specific training. Before that, these operations cost less than EUR 10 million.

2. In such a curve, divided into three parts, the first part is a decreasing failure rate (infant mortality with low availability rate of equipment); the second part is a constant failure rate (availability rates of equipment reaches a plateau); and the third part is an increasing failure rate (wear-out failures with again, low availability rates).

3. Notably, this includes all the military personnel deployed outside domestic soil, including the military deployed in permanent foreign bases (e.g. Pacific bases in the US case or foreign bases for France, including bases in foreign former colonies). Contrary to other studies that focused only on combat operations, we consider that prepositioned forces can be considered a form of deployment, notably because in the French case, they very often serve as support, supply and logistic nodes for combat operations. Moreover, they also serve as first responders in case of unforeseen geopolitical crises, as illustrated by the role of the French base in Djibouti to conduct emergency evacuation operations in Chad in 2023 (Opération Sagittaire), and maintain a high level of readiness, with substantial means allocated. These bases can thus be conceptualized as rear bases in the broader context of an intervention (even though they are usually located in a different country than the one in which combat operations take place, they are still part of the intervention itself).

4. The World Bank publishes data of the surface area of countries. It would obviously be interesting to know the exact share of the territory where France deploys troops each year, or better yet, the territory covered by each type of system. However, such information is not publicly available. Our AREA variable is therefore a compromise variable that we assume is sufficient for statistical analysis.

5. Refer to appendix 2 for descriptive statistics of the variables used in the estimations. See appendix 3 for the correlation matrix of the variables.

6. The test has been done with the sigmamore option with Stata, which specifies that both covariance matrices are based on the same estimated disturbance variance from the efficient estimator.

7. For descriptive statistics on the number of observations in groups at the different levels, see appendix 4.

8. In other words, ICC is the expected correlation between any two randomly chosen measurements belonging to the same group.

9. 0.009/ ;(0.009 + 0.006 + 0.011) = 0.34.