[This piece was originally posted on 27 June 2016.]
Previous posts have detailed casualty estimates by Trevor Dupuy or The Dupuy Institute (TDI) for the 1990-91 Gulf War and the 1995 intervention in Bosnia. Today I will detail TDI’s 2004 forecast for U.S. casualties in the Iraqi insurgency that began in 2003.
In April 2004, as simultaneous Sunni and Shi’a uprisings dramatically expanded the nascent insurgency in Iraq, the U.S. Army Center for Army Analysis (CAA) accepted an unsolicited proposal from TDI President and Executive Director Christopher Lawrence to estimate likely American casualties in the conflict. A four-month contract was finalized in August.
The methodology TDI adopted for the estimate was a comparative case study analysis based on a major data collection effort on insurgencies. 28 cases were selected for analysis based on five criteria:
- The conflict had to be post-World War II to facilitate data collection;
- It had to have lasted more than a year (as was already the case in Iraq);
- It had to be a developed nation intervening in a developing nation;
- The intervening nation had to have provided military forces to support or establish an indigenous government; and
- There had to be an indigenous guerilla movement (although it could have received outside help).
Extensive data was collected from these 28 cases, including the following ten factors used in the estimate:
- Country Area
- Orderliness
- Population
- Intervening force size
- Border Length
- Insurgency force size
- Outside support
- Casualty rate
- Political concept
- Force ratios
Initial analysis compared this data to insurgency outcomes, which revealed some startlingly clear patterns suggesting cause and effect relationships. From this analysis, TDI drew the following conclusions:
- It is difficult to control large countries.
- It is difficult to control large populations.
- It is difficult to control an extended land border.
- Limited outside support does not doom an insurgency.
- “Disorderly” insurgencies are very intractable and often successful insurgencies.
- Insurgencies with large intervening third-party counterinsurgent forces (above 95,000) often succeed.
- Higher combat intensities do not doom an insurgency.
In all, TDI assessed that the Iraqi insurgency fell into the worst category in nine of the ten factors analyzed. The outcome would hinge on one fundamental question: was the U.S. facing a regional, factional insurgency in Iraq or a widespread anti-intervention insurgency? Based on the data, if the insurgency was factional or regional, it would fail. If it became a nationalist revolt against a foreign power, it would succeed.
Based on the data and its analytical conclusions, TDI provided CAA with an initial estimate in December 2004, and a final version in January 2005:
- Insurgent force strength is probably between 20,000–60,000.
- This is a major insurgency.
- It is of medium intensity.
- It is a regional or factionalized insurgency and must remain that way.
- U.S. commitment can be expected to be relatively steady throughout this insurgency and will not be quickly replaced by indigenous forces.
- It will last around 10 or so years.
- It may cost the U.S. 5,000 to 10,000 killed.
- It may be higher.
- This assumes no major new problems in the Shiite majority areas.
When TDI made its estimate in December 2004, the conflict had already lasted 21 months, and U.S. casualties were 1,335 killed, 1,038 of them in combat.
When U.S. forces withdrew from Iraq in December 2011, the war had gone on for 105 months (8.7 years), and U.S. casualties had risen to 4,485 fatalities—3,436 in combat. The United Kingdom lost 180 troops killed and Coalition allies lost 139. There were at least 468 contractor deaths from a mix of nationalities. The Iraqi Army and police suffered at least 10,125 deaths. Total counterinsurgent fatalities numbered at least 15,397.
As of this date, the conflict in Iraq that began in 2003 remains ongoing.
NOTES
Christopher A. Lawrence, America’s Modern Wars: Understanding Iraq, Afghanistan and Vietnam (Philadelphia, PA: Casemate, 2015) pp. 11-31; Appendix I.
