DEIMS and DEPPS History and Current Models
DoD`s projections of annual DoD demand had its start in 1980 when the Office of the Secretary of Defense (Program Analysis and Evaluation) (OSD(PA&E)) created DEIMS -- Defense Economic Impact Modeling System. DEIMS was designed primarily to supply industry with information about long-term defense spending plans. This information allowed firms to size production capacity better while protecting classified, proprietary, and contract specific data.
In 1996, OSD(PA&E) and Interindustry Forecasting at the University of Maryland (INFORUM) completely restructured DEIMS. The restructuring was so substantial that the model is now called the Defense Employment and Purchases Projection System, or DEPPS, which consists of three defense-spending models:
  • , a detailed interindustry model, which forecasts defense demands by 320 industries, broken out by major appropriation and procurement category.
  • , a state-level (regional) model, which determines the effect of defense expenditures by major procurement category on each state, at an aggregate level of 97 industries (unlike the more detailed IDEPPS).
  • , a skilled labor model, which summarizes the requirements generated for various occupational classifications of employment in each industry.
Two input-output models feed DEPPS. The first model, called the Long-term Interindustry Forecasting Tool (LIFT), is a 97-sector input-output model embedded in a macroeconomic model. LIFT establishes the macroeconomic environment as well as industry controls. The second input-output model, called Iliad for Interindustry Long-run Integrated and Dynamic model, uses the macroeconomic forecasts and the industry controls from LIFT to further divide the economy into 320 industries. LIFT and Iliad were developed by INFORUM.

Because the framework for the projections was redesigned in 1996, and the definitions of industries for which projections are reported have changed, reliable comparisons between pre- and post-1996 projections are precluded. Three additional factors influence the reliability of comparisons between the current estimates and other post-1996 projections: (1) internal changes in the DEPPS model; (2) differences between defense budgets resulting from changes in the Future Year Defense Program (FYDP); and (3) economic assumptions.
The internal structure of DEPPS is revised occasionally. Such was the case for, "Projected Defense Purchases: Detail by Industry and State Calendar Years 2002 Through 2007", when LIFT was changed to a 1987 base year input-output table, and the number of sectors increased from 85 to 97. Structural changes such as this make it extremely difficult to compare projections over time. Most internal model changes, however, reflect revisions in data sets rather than changes in a model's fundamental architecture. For example, in 2002 the occupational matrix used in LDEPPS (the skilled labor model) was updated with the Bureau of Labor and Statistics (BLS) 1998 - 2008 projections. In addition, industry-level employment data used in LIFT have been updated through 1998. These types of changes do not preclude comparisons because the essential framework of the model remains unchanged.

Changes in the defense budget overall also influence cross-year comparisons. Each year`s projections are made using the defense portion of the President`s budget submission to Congress. The defense budget is developed each year through an involved process and can change significantly from year to year. Consequently, the projections also change. For example, the projections for 2003 are based on a defense budget that is more than 20 percent higher than the 2001 budget.

Finally, variability in the projections, and assumptions about the behavior of the economy as a whole and of specific sectors within the economy, also affect DEPPS outcomes. Many basic economic assumptions have been revised over the past year, in part due to the continued strong growth of the U.S. economy. The revised assumptions imply higher average productivity growth than was implicit in 2000's projections. Other economic changes include higher global oil prices and higher interest rates. These changes affect the 2003 projections. As this discussion illustrates, there are many reasons why estimates shift from year to year. Structural changes in the DEPPS model greatly complicate comparisons with other years` projections. The other possible sources of change--DEPPS data revisions, the defense budget, and broader economic assumptions--do not preclude comparisons, but isolating the effects of any single factor is a challenge. Ultimately, while comparisons between estimates are possible, they should be made taking the underlying reasons for the changes into account.


Rules and Regulations