Defense Budget

Pursuant to a congressional request, GAO identified the reasons for the Army National Guard's unfinanced Operation and Maintenance (O&M) requirements, focusing on the: (1) way the Guard prepares its O&M budget; (2) five Army models the Guard uses to generate most of its O&M requirements; and (3) improvements the Army and the Guard are making to correct problems in their use of the models.

GAO noted that: (1) beginning with the fiscal year 1998 requirements determination process, the Guard has used five Army models to determine about 80 percent of its O&M requirements; (2) these requirements are then reviewed by senior Army leaders, who consider Army priorities as they allocate funds within the budget levels established for the Army by the Office of the Secretary of Defense; (3) since the established budget levels are usually below the Guard's total O&M requirements, the Guard has reported unfinanced requirements in fiscal years 1998-2000; (4) since it began using the Army's models, the Guard has experienced some problems with three of the models that estimate O&M requirements for funding civilian personnel, training, and base operations/real property maintenance; (5) as a result, the models did not provide reliable estimates of some O&M requirements for fiscal years 1998-2000, according to Guard officials; (6) inaccurate results were produced by the models because the Guard did not provide accurate and timely data on civilian personnel costs and did not have a complete database on its real property inventory; (7) one model did not consider certain costs that are unique to the Guard or used historical data that did not provide a good basis for determining future requirements; (8) the Guard and the Army are working to correct problems with their use of the models so that the models will provide more reliable estimates of O&M requirements in the future; and (9) the Guard is also working with the Cost and Economic Analysis Center to identify some of the unique training costs to the Guard for inclusion in the model.