Can We Count on an Accurate Count?

The 2010 Census was the most expensive in U.S. history—it cost about $13 billion from start to finish (in 2010 dollars). Using the same design, the 2020 Census would cost over $17 billion (in 2020 dollars). But the Census Bureau’s new plan for the national headcount may put the cost closer to $12.5 billion (in 2020 dollars). Is that estimate really reliable? Today’s WatchBlog explores. 

 
The Census Bureau hopes to control costs through such innovations as automating how it collects data.

Exploring a redesign

The Census Bureau has been using the same basic procedure to take the U.S. census since 1970—mail out a questionnaire and follow up in person with people who don’t respond. However, the new plans for 2020 include reengineering the address list, using the Internet and telephones to increase response, using administrative records to reduce field work, and increasing automation in field operations.

But seamlessly implementing all these new processes (many of which have not been used before in the large-scale census effort) could get tricky. For example, the Bureau still needs to decide which administrative records it will use in 2020, and where exactly they will come from. Also, it remains to be seen if the Bureau can develop and deploy the critical information systems it will need to support the massive operation.

Estimating costs

The Census Bureau believes that these new processes and other improvements will make the census more cost effective compared to prior headcounts. The problem is that the Bureau’s October 2015 cost estimate for the 2020 Census does not fully reflect the characteristics of a high-quality estimate and cannot be considered reliable. The Bureau’s cost estimate has not substantially met any of the four criteria for effective cost estimation—i.e., that estimates should be comprehensive, well documented, accurate, and credible. Without a reliable cost estimate, it is difficult for Congress to provide oversight and make effective funding decisions.

Looking forward to 2020

We have a number of recommendations that could help the Bureau identify ways to save money in 2020. For example, the Bureau should focus its resources on refining and executing the ideas with the most potential, rather than continuing to research less promising proposals. Follow us on this decade-long audit trail by checking out all of our open recommendations to the Census Bureau.

• Comments on GAO’s WatchBlog? Contact blog@gao.gov.