Air Campaign Duration and the Interaction of Air and Ground Forces

Abstract

This article explores the role that ground troops play in aerial campaigns. Specifically, it intends to explain how ground troops affect the duration of aerial bombing campaigns. It distinguishes between the effect of ground troops on the duration of aerial campaigns that are selective and on those that are more general and less discriminate (using newly coded data on aerial strategies). The effect of ground troops may vary depending on how air power is being used. Ground troops may increase the duration of aerial campaigns in cases in which selective strategies are used during the aerial bombing campaign. In these cases, states lose the possibility of quick, “clean” victories once ground troops are introduced. In the cases of the less-discriminate bombing campaigns, the effect of ground troops is the opposite: They shorten the duration of the bombing campaign, possibly through the increased effectiveness of a combined arms approach.

KEYWORDS:

• conflict
• duration analysis
• duration model
• security

ACKNOWLEDGMENTS

I would like to thank Ric Stoll, Cliff Morgan, Ashley Leeds, Cooper Drury, Sam Bell, Tobias Heinrich, Emizet Kisangani, Michael Plummer, two anonymous reviewers, and the editor of International Interactions for their help in preparing this paper. All remaining errors are my own. Data and replication files, including the appendices mentioned in this article, are available at the International Interactions Dataverse page, http://dvn.iq.harvard.edu/dvn/internationalinteractions. Please direct all questions regarding replication to the author.

Notes

¹ The Etrich Taube is a small monoplane that was used as a bomber, fighter, trainer, and surveillance aircraft. Italy and Turkey were fighting over control of present-day Libya, in what came to be known as the Italo-Turkish War, or Turco-Italian War (Johnston 2011).

² While there was no formal NATO ground presence, the alliance did employ forward air control teams on the ground to help guide aircraft to its targets (Schmitt 2012).

³ It should be noted that Obama did face some criticism from his political opposition, but much of it focused on the length of the campaign and whether he should attempt to repeal the War Powers Act. The opposition was nowhere near as strong as it was for the Afghanistan or Iraq wars (Savage and Landler 2011).

⁴ Biddle does not use the term “modern” to refer to campaigns that happened after a particular period in time but rather a type of fighting that can be observed at different periods in time. At the same time, it is important to also keep in mind that the modern system is not completely unrelated to time. As technology increases, engaging in precision bombing becomes much easier. In older wars it may have been the case that states had the intention of engaging in precision bombing, but their lack of technology made the campaigns virtually indistinguishable from general ones. The same idea applies to states with fewer resources available to them. It may be the case that they are not able to engage in selective bombing to the extent they would desire because they have old, outdated equipment that does not allow them to work with the precision required for an effective modern campaign. This means that in conducting the analysis it will be important to consider temporal effects, as well as the capabilities of states involved.

⁵ Note that the analysis does not differentiate between the different uses of air power, as given this small sample, any result that spoke to differences between the uses of air power would be too weak and unstable.

⁶ As noted by an anonymous reviewer, a target who recognizes an attacker’s intent to engage in a short campaign may choose to “bunker down” and attempt to outlast the imposition of costs.

⁷ For example, in the case of 1986’s Operation El Dorado Canyon, in which the United States retaliated against Libya for the bombing of a discotheque in West Berlin that was frequented by American service members, the bombing campaign lasted only a few hours. Despite the aircraft hitting most of its intended targets, the campaign is coded as a defeat for the United States (Allen 2007). This is because through the bombing raid the United States was unable to achieve its goals of either killing Qaddafi (Qaddafi was not in the expected location that US intelligence had placed him at) or reducing Libyan terrorist activity. Thus, the United States quickly realized that this bombing strategy was unlikely to help it achieve its aims and did not extend the duration of its bombing campaign over Libya (GlobalSecurity.org n.d.).

⁸ In this case, the dispersed tanks would be more vulnerable to ground forces than if they were clustered together defensively.

⁹ Like Allen (2007), I use only cases of actual uses of air power, not threats.

¹⁰ It is important to note that the sample size of this article is smaller than that of Allen (2007) because I used only cases in which both the attacker and target were state actors (Correlates of War Project 2008). The reason for doing this was the lack of availability of reliable capability data for nonstate actors.

¹¹ Any differences from the Allen (2007) coding are explained in Appendix C, available online. Mostly, differences in coding arise because the Allen (2007) variable included both ground and naval forces, where this one includes only ground forces.

¹² As a robustness check, I also included a post-1970 variable in the analysis, to reflect the development of more modern and precise weapons systems. The results of this analysis are included in Appendix A Tables A2 and A3, available online. There was no significant change in the results when including this variable.

¹³ I would like to thank an anonymous reviewer for the suggestion to include this variable in the analysis.

¹⁴ One consideration is whether to include any previous aerial campaigns between the attacker and the target or only those that were successful (since it is expected that states may be able to draw more useful conclusions from a successful campaign than an unsuccessful one). As a robustness check, I repeat the analysis including a variable that measures whether there was a successful aerial campaign in the past. The results, which are included in Appendix A, Table A4 (available online), are not altered in a significant way.

¹⁵ The model fits the data well, as indicated by a Wald test in which the chi-square is equal to 29.81, and the probability of obtaining the chi-squared statistic if there were no effect of the variables is equal to 0.0001. Because the sample size is so small, multicollinearity is a concern. The correlation matrix included in Appendix A shows that there is not particularly high correlation among the predictors. Not considering the interaction term, the highest correlation is 0.69 between Post-1950 and Selective Air Power.

¹⁶ As a robustness check, Appendix A, Table A6 (available online), includes the results of a Cox model. The two main variables, ground forces and selective air power, as well as the interaction term, retain their significance as well as the direction of their effect.

¹⁷ As a robustness check, I also used a competing risks model that takes into account the difference between cases in which the campaign ended because the outcome was successful and cases in which the campaign ended because the outcome was unsuccessful. The results remained very similar. These results are included in Appendix A, Table A7 (available online).

¹⁸ Note that the Maoz (2005) Dyadic MID data set ends with conflicts that began in 2001. This means that the Iraq War will not be included in it (Bushway, Johnson, and Slocum 2007).

¹⁹ Having an exclusion restriction present in a selection model avoids the problem of inflated standard errors. Models with an exclusion restriction can better address the problem of selection bias. Following (Sartori 2003), I interpret the z value of 6.95 on the capabilities variable as an indicator of it being a strong exclusion restriction. As an added robustness check, I estimated a Weibull duration Model in which I included the CINC scores of the attacker as an independent variable (Table A9 in Appendix A, available online). It did not have a significant effect on duration.

²⁰ Though in several of the WWII campaigns that included ground forces this does not appear to be the case.

²¹ As a preliminary exploration of this dynamic, in Table A1 of Appendix A, I include the results of a Weibull duration model that includes two variables that measure the alliance systems of both the attacker and target. To operationalize this, I use data from the ATOP project, which distinguishes between different types of alliances (Leeds, Ritter, Mitchell, and Long 2002). Each variable measures the number of alliances each state is involved in in a given year, with the condition that at least one of them must be a defensive or offensive alliance (as these would be the ones that would actually give a state support during an aerial campaign). As shown in Appendix A, the results remain similar, with the coefficients on the selective aerial strategies and ground troops (as well as their interaction term) remaining significant and in the expected direction.