June E. O'Neill

July 26, 1996

Honorable Byron L. Dorgan
United States Senate
Washington, D.C. 20510

Dear Senator:

In your letter of June 4, 1996, you asked for more information about the potential costs of operating and supporting the defenses that the Congressional Budget Office (CBO) included in its estimates of the Defend America Act of 1996 (S. 1635). You also asked for information about potential cost overruns for systems that might be part of a national missile defense. The enclosure responds to that request.

If you wish further details of our analysis, we will be pleased to provide them. The CBO staff contacts are Raymond Hall, who can be reached at 226-2840, and David Mosher, who can be reached at 226-2900.


cc: Honorable Pete V. Domenici
    Committee on the Budget

    Honorable Strom Thurmond
    Committee on Armed Services

    Honorable Sam Nunn
    Ranking Minority Member
    Committee on Armed Services

Identical letter sent to Honorable J. James Exon


In a letter dated June 4, 1996, Senators Exon and Dorgan asked the Congressional Budget Office to analyze the potential costs of operating and supporting the defenses included in its estimate of the Defend America Act of 1996 (S. 1635 and H.R. 3144). They also asked about cost growth in comparable acquisition programs.


The Defend America Act of 1996 would establish policies for deploying a national missile defense system but it does not define many of the details or characteristics of such a system. Thus, the bill is subject to interpretation over what systems would be capable of rendering a "highly effective defense" and when it would be necessary to have a "layered defense against larger and more sophisticated ballistic missile threats."

CBO's estimate of the Defend America Act of 1996 presents a range of possible costs based on its interpretation of the bill. The low end of the range is $31 billion and the high end is $60 billion, including the effects of inflation. However, these estimates do not include costs that must be paid each year to keep the defense functioning after deployment. Those recurring costs include three components:

CBO estimates that total recurring costs would be about $2 billion annually for the low-end system and about $4 billion annually for the high-end system. (Unless otherwise noted, all costs in this analysis are expressed in constant 1997 dollars.) Annual costs of this magnitude would not occur until about 2010, and assume that appropriations would be provided indefinitely. These estimates also assume that the system's basic structure would remain stable after it is deployed. However, at some point new technology or a reassessment of the defense situation could lead to changes in any of the defenses assumed in CBO's estimate, which could have a large impact on costs.

Of course, if the bill was interpreted differently and smaller or larger systems were actually deployed, costs would be different. For example, the recurring costs of certain proposals from the Air Force and the Ballistic Missile Defense Organization (BMDO) for

page 1

ground-based defenses deployed at a single site would total about half a billion dollars to $1 billion annually depending on whether space-based sensors were included. (Acquisition costs, including the costs of inflation, for these alternatives would range from about $4 billion for an Air Force system witli 20 interceptors to $13 billion for a system based on a BMDO proposal with 100 interceptors and space-based sensors.) However, we understand the bill to call for greater capability than possible with those smaller systems.

Recurring Costs of Certain Ground-Based Defenses

The system estimated by CBO to cost from $31 billion to $60 billion to acquire includes space-based components in addition to ground-based defenses. Several proposals have been made for missile defenses that use only ground-based components, including plans by BMDO and the Air Force that would deploy ground-based defenses at a single site in Grand Forks, North Dakota. Those systems fit the following descriptions:

Both of those defenses could also include a constellation of space-based sensors called the Space and Missile Tracking System (SMTS).

CBO estimates that the recurring costs for the ground-based defenses would be about half a billion dollars to $1 billion annually. These costs would provide resources for operating and supporting the interceptors and satellites, for replacing satellites, and for conducting follow-on tests.

Recurring Costs of Operating the Low-End System

The system estimated by CBO to cost $31 billion has three basic components: an initial defense consisting of 100 ground-based interceptors located at a single site with supporting ground-based radars; a constellation of 24 SMTS sensors; and a constellation of 500 space-based, kinetic energy interceptors (known as Brilliant Pebbles).

page 2

Recurring costs for that system would be about $150 million in 2003 and reach $500 million in 2010. Beyond 2010, average annual expenditures would total about $2 billion, including about half a billion dollars for operation and ground support, $1 billion for replacing satellites, and $250 million for conducting follow-on tests.

Costs of Operation and Ground Support. Once it was deployed, the low-end system would require resources to maintain equipment on the ground, including the ground-based interceptors and radars, and to support the space-based systems to be sure that they, too, were functioning properly. CBO's estimates of these costs are based on comparable costs for systems that DoD uses now and that are reasonably similar in scope and function to national missile defenses.

Ground-Based Interceptors. The system of 100 ground-based interceptors and the associated radars would cost $150 million a year to operate by 2003, when it would be fully operational. This estimate is based on the operating costs of an air defense system (Patriot), a strategic missile system (Minuteman III), and the Safeguard antiballistic missile system, which the Army operated in the early 1970s. The Safeguard system consisted of two types of interceptors--a long-range Spartan missile and a short-range Sprint missile--to defend Minuteman missiles. The cost to operate 100 Safeguard interceptors at a single site reached nearly $200 million in today's dollars.

Space-Based Sensors. The annual cost to operate a constellation of 24 SMTS satellites from the ground would be about $150 million by 2007, when all satellites would be deployed. This estimate is based on the cost of operating the 24-satellite Global Positioning System (GPS) and its nuclear detection system, which is similar to SMTS in constellation size, missions, and communications network. These costs include about $100 million to maintain the constellation's position in orbit, to ensure that it is working properly, and to provide maintenance for the ground systems that support the satellites. Another $50 million would be needed for conununications and data processing.

Space-Based Interceptors. It is hard to know how the space-based interceptors would be managed and controlled from the ground primarily because the constellation would be so large -- about 20 times bigger than the largest constellation that DoD operates now. CBO assumes that operation and ground support for the space-based interceptors would entail costs for three components:

CBO estimates that the first two components would cost $50 million to $125 million annually and would involve 300 to 900 people. The range reflects uncertainty about the marginal increases that would be required in Los Angeles and Colorado. CBO estimates that the third component would cost $100 million to $125 million and would involve several hundred civilian contractors and enlisted personnel.

Costs of Replacing Satellites. In addition to operating the defenses, DoD would have to replace the space-based components periodically because, like all satellites, their performance will deteriorate over time. Satellites are designed to operate for a specified length of time in orbit, often called the mean mission duration. The mean mission duration is a function of many factors, including the amount of rocket fuel available for orbital corrections, the ability of critical electronics to withstand the radiation that a satellite encounters, and the length of time that special cooling systems can operate to keep infrared sensors like those in SMTS, Brilliant Pebbles, and space-based lasers cold enough to detect cold warheads in space. In the past, some satellites have exceeded the planned mean mission duration and others have fallen short. Planners typically use the mean mission duration to estimate the rate at which satellites would have to be replaced.

The SMTS program office plans to build satellites that last 8.5 years, on average. With a 24-satellite constellation, an average of about 3 satellites would have to be launched each year to sustain the system. CBO esfimates that this would cost $400 million annually beyond 2010, including the cost of the satellites (about $100 million each) and the cost to launch three satellites each year into orbit ($60 million).

Based on an average service life of 10 years reported in earlier studies, DoD would have to launch an average of 50 Brilliant Pebbles a year. Assuming that 10 satellites would be deployed on each launcher, CBO estimates that this would cost $500 million annually beyond 2010 (5 launchers at $60 million each and 50 Pebbles at $4 million each).

Costs of Follow-On Tests. After the defenses are deployed, DoD would have to test them on a regular basis. These tests are necessary to ensure that the systems continue to meet performance standards despite the effects of time and day-to-day use, and to validate hardware and software changes.

page 4

CBO assumes that an average of 1O to 12 follow-on tests a year would occur, evenly divided between the ground-based interceptors and the space-based interceptors. The Air Force conducted at least six follow-on tests a year for the Minuteman III intercontinental ballistic missiles during the first 15 years of operations. Because DoD would probably want to have similar confidence in the reliability and performance of a missile defense, CBO used this same number for both types of interceptors. The number of tests may be lowered as the system matures in much the same way as the number of postproduction flight tests has been lowered for Minuteman and Trident D5 missiles. CBO estimates that the recurring cost of those tests would total $250 million annually. CBO assumes that retired Minuteman missiles currently in storage would serve as targets.

Recurring Costs of Operating the High-End System

The system that CBO estimated would cost $60 billion to build includes 20 space-based lasers and two sites with 100 ground-based interceptors each, in addition to the components included in the low-end system. Adding those components to the low-end system would boost the recurring costs to about $4 billion a year after 2010, including about $1 billion for operation and ground support, about two and a half billion dollars for replacing satellites, and $300 million for conducting flight tests.

Costs of Operation and Ground Support. In addition to the nearly $500 million required for the low-end system, the high-end system would require resources to deploy the additional ground-based interceptors and the space-based lasers. CBO's estimates of these costs are based on costs for reasonably comparable systems that DoD uses now.

Ground-Based Interceptors. Deploying 100 ground-based interceptors at two additional sites would increase operation and ground support costs by $250 million a year by 2010. This estimate is based on the estimate described above for operating a single-site system at Grand Forks. Although the high-end system would add two times more ground-based interceptors, total costs would not triple because the additional sites would not require extra early-warning radars.

Space-Based Lasers. The costs to operate a system of space-based lasers are difficult to determine. On the one hand, it is possible that much of today's infrastructure for space operations could support a constellation of 20 space-based lasers with added costs of about $100 million a year. For example, some people would be assigned to DoD's Satellite Control Network in Los Angeles and some people would be assigned to Falcon AFB in Colorado, but there could still be some economies in operations of several space systems. On the other hand, space-based lasers could be far more technically complex than currently deployed satellites. Based on the relationship between investment and operation and support costs for three current space systems, operating costs for space-based lasers could reach $500 million annually.

page 5

Costs of Replacing Space-Based Lasers. If the laser satellites lasted 10 years in orbit, an average of two satellites would have to be launched each year to sustain a 20-satellite constellation. CBO estimates that this would cost about $1.6 billion a year beyond 2OlO. Combined with the satellite replacement costs for the low-end system ($1 billion a year), DoD would have to spend an average of roughly two and a half billion dollars annually after 2010 to replace the satellites in the high-end system.

Replacement costs could be lower, according to recent plans, if BMDO can extend the average service life of the lasers to 16 years by replenishing their chemical fuel while the satellites are in orbit. Those savings would be partially offset, however, by the costs of refueling the lasers in space. CBO did not use that option in this estimate because it involves new techniques and much more uncertainty in the costs.

Costs of Follow-On Tests. DoD would have to test the lasers operafionally on a regular basis. These tests would be necessary to ensure that the system continued to meet performance standards. CBO estimates that the annual recurring cost of five or six follow-on tests a year would total $50 million and bring the total cost of follow-on testing to $300 million a year. CBO assumes that retired Minuteman missiles would serve as missile targets. Instead of expending an entire satellite during such a test, the space-based laser would use up some of its laser fuel. The price of this fuel consumption is reduced time in orbit, and the overall cost is reflected in the estimate for replacement lasers.


Senators Exon and Dorgan also asked for information about cost overruns in procurement programs with the size, scope, and complexity of a national missile defense system. One useful framework for examining this question would compare a system's currently projected costs with earlier estimates based on original plans and estimates. Between the early stages of a program and the time it concludes (perhaps 20 years or more later), many things change. The differences in estimated costs could be divided into at least six categories:

Still, it is difficult to disentangle the effects of policy changes from changes in estimated costs or capability. For purposes of a given cost estimate, including the estimate for S. 1635, CBO assumes that policies are fixed and that any policy changes would be the subject of future Congressional acfion and the corresponding cost estimate. For the estimate of S. 1635, CBO included $3 billion to hedge against technical and schedule risks in the underlying DoD estimate not caused by policy changes.

Studies by RAND and the Institute for Defense Analyses (IDA) have examined cost growth in a wide array of weapons systems. Both studies indicate that, after the effects of inflation and changes in quantities are factored out, costs of weapons programs have tended to grow. The studies focus on how costs have exceeded expectations after the start of engineering and manufacturing development (known as Milestone II), a point in a program at which the design is fairly well defined. None of the systems that would make up a national missile defense under S. 1635 have reached that point yet.

According to RAND, the systems that are most comparable to a national missile defense -- satellites, electronics, and missiles -- have experienced cost growth averaging 32 percent to 35 percent, close to the average for DoD programs after Milestone II. But the average in these categories masks a wide range in cost growth. For example, the MX missile ended up costing 2 percent less than anticipated, whereas Patriot and Pershing II missiles both experienced 50 percent cost growth. Likewise, the Global Positioning System experienced only 6 percent growth whereas costs of the DSCS III satellites grew by 57 percent.

It would be inappropriate, however, to apply an average percentage cost growth from the RAND and IDA studies to CBO's estimate for S. 1635. The studies show wide variation around the average cost growth for similar systems, some of which have experienced very little cost growth. Also, the studies measure cost growth from Milestone II, but none of the systems envisioned in CBO's estimate for S. 1635 have reached that point yet, and some are several years away.

page 7