Summary

In 2007, the U.S. Census Bureau (Bureau) estimated the 2010 Census would cost $11.5 billion, including $3 billion on automation and technology. At a March hearing, the Department of Commerce (Commerce) stated that the Field Data Collection Automation (FDCA) program was likely to incur significant cost overruns and announced a redesign effort. At that time, GAO designated the 2010 Decennial Census as high risk, citing long-standing concerns in managing information technology (IT) investments and uncertain costs and operations. This testimony is based on past work and work nearing completion, including GAO's observation of the address canvassing dress rehearsal. For IT acquisitions, GAO analyzed system documentation, including deliverables, cost estimates, other acquisitions-related documents, and interviewed Bureau officials and contractors. This testimony describes the implications of redesign for (1) dress rehearsal and decennial operations, (2) IT acquisitions management, and (3) Decennial Census costs.

The Decennial Census is at a critical stage in the 2008 Dress Rehearsal, in which the Bureau has its last opportunity to test its plans for 2010 under census-like conditions. On April 3, 2008, Commerce announced significant changes to the FDCA program. It also announced that it expected the cost of the decennial to be up to $3 billion greater than previously estimated. The redesign will have fundamental impacts on the dress rehearsal as well as 2010 Census operations. Changes this late in the decade introduce additional risks, making more important the steps the Bureau can take to manage those risks. The content and timing of dress rehearsal operations must be altered to accommodate the Bureau's design. For example, Commerce has selected an option that calls for the Bureau to drop the use of handheld computers (HHCs) during the nonresponse follow-up operation, and the Bureau may now be unable to fully rehearse a paper-based operation. Additionally, reverting to a paper-based nonresponse follow-up operation presents the Bureau with a wide range of additional challenges, such as arranging for the printing of enumerator forms and testing the systems that will read the data from these forms once completed by enumerators. Given the redesign effort, implementing GAO's recommendations associated with managing the IT acquisitions is as critical as ever. Specifically, the Bureau needs to strengthen its acquisition management capabilities, including finalizing FDCA requirements. Further, it also needs to strengthen its risk management activities, including developing risk mitigation plans for significant risks and improving its executive-level governance of these acquisitions. The Bureau also needs to plan and conduct key tests, including end-to-end testing, to help ensure that decennial systems perform as expected. According to the Bureau, the redesign and related revision of the FDCA program is expected to result in significant increases to the life cycle cost estimate for the 2010 Census. Even without considering the recent expected cost increases announced by the Bureau to accompany the redesign of the FDCA program, the Bureau's cost projections for the 2010 Census revealed an escalating trend from previous censuses. Previously, GAO recommended that the Bureau develop an integrated and comprehensive plan to manage operations. Specifically, to understand and manage the assumptions that drive the cost of the decennial census, GAO recommended, among other actions, that the Bureau annually update the cost of the 2010 Census and conduct sensitivity analysis on the $11.5 billion estimate. However, while the Bureau understands the utility of sensitivity analysis, it has not conducted such an analysis.