Case Study 18, 29: From 2020 Census, GAO-18-635, August 17, 2018 and GAO-16-628, June 30, 2016
To help control costs while maintaining accuracy, the Census Bureau was introducing significant change to how it conducted the decennial census in 2020. Its planned innovations included reengineering how it built its address list, improving self-response by encouraging the use of the Internet and telephone, using administrative records to reduce field work, and reengineering field operations using technology to reduce manual effort and improve productivity. The Bureau estimated that if it succeeded with these innovations it could conduct the 2020 Census for $12.5 billion in constant 2020 dollars. By contrast, the 2020 Census would cost $17.8 billion in constant 2020 dollars if the Bureau had repeated the 2010 Census design and methods, according to the Bureau’s estimates.
As early as 2011, the Bureau began developing preliminary cost estimates of the 2020 Census in order to approximate potential savings from its plans to reengineer the census, and, according to the Bureau, to begin developing the methodology for producing the decennial life cycle cost estimates.
The Bureau’s October 2015 release of the latest cost estimate marked the transition from the “research” to “implementation” phases of the 2020 Census. According to the Bureau, this was the Bureau’s first attempt to model the life cycle cost of its planned 2020 Census, in contrast to its earlier 2011 estimate which the Bureau said was intended to produce an approximation of potential savings and to begin developing the methodology for producing decennial life cycle cost estimates covering all phases of the decennial life cycle.
However, GAO found that the October 2015 cost estimate for the 2020 Census did not fully reflect characteristics of a high-quality estimate and could not be considered reliable. Overall, GAO found the cost estimate partially met the characteristics of two best practices (comprehensive and accurate) and minimally met the other two (well documented and credible). One reason why GAO’s overall assessment was low was because the estimate was not well documented. Improving cost estimation practices would increase the reliability of the Bureau’s cost estimate, which would in turn help improve decision making, budget formulation, progress measurement, course correction when warranted, and accountability for results.
Best practices state a risk and uncertainty analysis should be performed to determine the level of risk associated with the cost estimate. The Bureau carried out such an analysis only for a portion of estimated costs for fiscal years 2018 to 2020. According to Bureau officials, they scoped the analysis narrowly to those 3 years when most of the census costs occur. GAO found that, as a result, the Bureau’s risk and uncertainty analysis (modeled costs) covered $4.6 billion, only about 37 percent of the $12.5 billion total estimated life cycle cost, and less than one-half of the total estimated cost of the census during future fiscal years.
In October 2017, the Department of Commerce announced that the projected life cycle cost of the 2020 Census had climbed to $15.6 billion, a more than $3 billion (27 percent) increase over its 2015 estimate. A high-quality, reliable cost estimate is a key tool for budgeting, planning, and managing the 2020 Census. Without this capability, the Bureau was at risk of experiencing program cost overruns, missed deadlines, and performance shortfalls.
Since GAO’s June 2016 report, in which we reviewed the Bureau’s 2015 version of the cost estimate, the Bureau had made significant progress. For example, the Bureau had put into place a work breakdown structure (WBS) that defined the work, products, activities, and resources necessary to accomplish the 2020 Census and is standardized for use in budget planning, operational planning, and cost estimation, The Bureau also improved their risk and uncertainty analysis methodology for the 2017 life cycle cost estimate, using a combination of modeling based on Monte Carlo simulation and other methods to develop the contingency estimates.
GAO reported these findings on June 30, 2016 in 2020 Census: Census Bureau Needs to Improve Its Life cycle Cost Estimating Process, GAO-16-628 and on August 17, 2018 in 2020 Census: Census Bureau Improved the Quality of Its Cost Estimation but Additional Steps Are Needed to Ensue Reliability, GAO-18-635.