What We Found
The Census Bureau implemented new technologies and other innovations for the 2020 Census, but also made a series of late design changes, such as delaying operations in response to COVID-19, that put the quality of the census at risk. Evaluations of innovations and late design changes are critical for 2030 Census planning.
Since our 2019 High-Risk Report, ratings for the leadership commitment criterion regressed, and the other four remain unchanged.
Leadership commitment: partially met. The Coronavirus Disease 2019 (COVID-19) pandemic caused the Census Bureau to pause and delay several field data collection operations. For example, the Bureau delayed nonresponse follow-up, when the Bureau follows up with households that do not initially respond to the Census, by 3 months. Given this unexpected stop in operations, the Department of Commerce (Commerce) on April 13, 2020, requested a 4-month delay in delivering the apportionment numbers to the President—statutorily due by December 31, 2020—and sought legislative relief for the delay.
However, according to Bureau officials, Commerce directed the Bureau to plan for a census without legislative relief, and on August 3, 2020, the Bureau publicly announced it would deliver the apportionment numbers by December 31, 2020. According to senior Bureau officials, they were not consulted on the decision to change the date for delivering the apportionment numbers and had approximately 96 hours to develop a revised plan of operations without legislative relief.
To meet the statutory deadline, the Bureau shortened the time to collect data by 1 month and the response processing operation by approximately 2.5 months. Compressing these time frames, increased the risk to the accuracy and completeness of the census count.
Subsequently, this decision to shorten data collection time frames and deliver the apportionment numbers by December 31, 2020, was challenged in court. The Bureau ended field data collection on October 15, 2020, after the U.S. Supreme Court determined that it could do so, and began working on a timeline to deliver apportionment numbers. However, data anomalies found during the processing of census responses have delayed the delivery of apportionment numbers, which as of February 2021 had not been delivered to the President.
Moreover, during the summer of 2020, Commerce created four new political appointee positions at the Bureau—Deputy Director for Data, Deputy Director of Policy, Senior Advisor to the Deputy Director for Policy, and Counselor to the Director. Senior Bureau officials told us that it is unprecedented to create new senior-level political appointee positions during the decennial census and that the appointees’ roles in the 2020 Census were often not clear. On January 19, 2021, all four political appointees resigned, and on January 20, 2021, the Director of the Census Bureau retired.
Capacity: partially met. Before the COVID-19 outbreak, the Bureau had an estimated $2.03 billion in total unobligated contingency funds for 2020 operations. As of December 2020, the Bureau anticipated that these contingency funds will be more than sufficient to address both the initial COVID-19 response and design changes, estimating that the Bureau would still have at least $187 million in contingency funding.
To ensure its resources are effectively targeted for the 2030 Census, it will be important for the Bureau to follow both cost estimation and scheduling best practices. For example, the Bureau needs to implement our recommendation to improve the credibility of schedules for the 2030 Census. While we found the Bureau took steps toward conducting quantitative schedule risk analyses with its master activity schedule for the 2020 Census, it effectively ran out of time to do so. To ensure the Bureau has sufficient resources and time to complete all the activities for the 2030 Census, this recommendation will remain open.
Action plan: partially met. According to Bureau officials, in 2019 they began the planning for the 2030 Census. The focus of 2030 planning is to reduce risk during peak operations through the work done earlier in the decade. The Bureau has developed 5 guiding principles for 2030 Census planning:
- follow disciplined management practices;
- simplify designs, solutions, and methods;
- distribute program work, resources, and costs more evenly across the census life-cycle;
- minimize field data collection with alternative data sources wherever possible; and
- manage stakeholder communications and expectations throughout the decade
However, in planning for 2030, the Bureau will not fully understand the quality of the data collected for 2020 until it completes all of its planned evaluations. The Bureau has a series of planned operational assessments, coverage measurement exercises, and data quality teams that are positioned to retrospectively study the effects of design changes made in response to COVID-19 on census data quality.
The Bureau is updating its plans for these efforts to examine the range of operational modifications made in response to COVID-19, including the August 2020 and later changes. We have previously noted that late design changes create increased risk for a quality census.
As part of the Bureau’s assessments, it will be important to address a number of concerns we identified about how late changes to the census design could affect data quality. These concerns include (1) how the altered time frames affected population counts during field data collection and (2) what effects, if any, compressed and streamlined post-data collection processing of census data may have on the Bureau’s ability to detect and fully address processing or other errors before releasing the apportionment and redistricting tabulations.
Over the next year, addressing these concerns and providing transparency over what is known and not yet known about census quality will help the Bureau increase public confidence in the quality and completeness of 2020 Census data products, despite all of the challenges the Bureau faced. These actions will also help inform future census planning efforts.
Monitoring: partially met. In looking forward to the next decennial census, senior Bureau officials told us they will build on lessons learned from 2020. For example, the Bureau actively monitored the COVID-19 pandemic and made necessary changes to census operations. Specifically, Bureau leadership used data to make real-time decisions about area census office re-openings during COVID-19.
However, as of January 2021, the Bureau continued to face uncertainty about schedules and plans related to disclosure avoidance for 2020 Census data products expected to be released starting in 2021. Disclosure avoidance protects the confidentiality of respondent data, especially at lower levels of geography.
According to the Bureau’s Chief Scientist, plans and schedules will need to be updated if the release dates for data products, such as redistricting data, change due to the operational impacts from COVID-19. In the fall of 2020, the agency had pushed some disclosure avoidance milestones from August 2020 to November 2020, due to schedule uncertainty and operational impacts from COVID-19.
Additionally, as of February 2021, the Bureau still needed to complete the IT testing and implementation activities required to support its post enumeration survey—a survey that is independent from the 2020 Census and intended to provide estimates of census quality. The Bureau plans to deploy the final systems to support its post enumeration survey by November 2021.
We previously reported on shortcomings in the Bureau’s management of the IT systems testing activities including, for example, being at risk of not meeting near-term milestones planned for completing system integration testing.
Demonstrated progress: partially met. In August 2019, the Bureau provided an update to the 2020 Census cost estimate, which we found to be sufficiently reliable. For example, the Bureau had implemented a system to track and report variances between actual and expected cost elements. Tools to track these variances are important because they enable management to measure progress against planned outcomes and prepare for 2030.
However, as of January 2021, a recommendation we made in April 2019 to identify and implement corrective actions within prescribed time frames for cybersecurity weaknesses had not been fully addressed. The Bureau had made some progress toward addressing this recommendation by reducing the number of corrective actions that it considered “high” or “very high” risk. Nevertheless, as of November 2020, 106 of the 174 total open “high” and “very high” risk corrective actions (about 61 percent) were delayed past their scheduled completion dates.
In December 2020, the Bureau’s information security officials attributed their current delays in addressing the corrective actions to technical challenges and dependencies between systems. According to those officials, the Bureau conducts quarterly briefings with system and information security stakeholders to discuss in-depth the delayed corrective actions. However, cybersecurity will continue to be an area to watch as the Bureau processes data to be included in upcoming data products that are to be released starting in 2021.
The U.S. census is mandated by the Constitution and provides vital data for the nation. Census data are used, among other purposes, to apportion the seats of the U.S. House of Representatives; redraw congressional districts in each state; and allocate billions of dollars each year in federal financial assistance.
Through 2023, the 2020 Census is estimated to cost approximately $15.6 billion after adjusting for inflation. To achieve cost savings, the Bureau implemented several new innovations including the development of new and modified IT systems. However, these innovations were not fully tested as budget uncertainty caused the Bureau to scale back testing in 2017 and 2018.
The 2020 Decennial Census was first added in 2017 as a high-risk area. Moreover, both the 2000 and 2010 Censuses were high-risk areas. For this update, we are changing the name of the high-risk area because risks continue beyond 2020 and may threaten the 2030 Census.
In March 2020, COVID-19 caused the Bureau to delay its 2020 operations and when the Bureau resumed operations in May 2020, it faced a new set of operational and public safety challenges. These delays, the resulting compressed time frames, and continued uncertainty over COVID-19 had the potential to undermine the overall quality of the count.
As of January 2021, we have made 113 recommendations related to the 2020 Census, 20 of which have not been fully implemented. Commerce generally agreed with our recommendations and is taking steps to implement them. Moreover, in our April 2020 priority recommendation letter to Commerce we identified 10 recommendations as priorities, none of which have been fully implemented over the past year. To make continued progress for the 2030 Census it will be essential for the Bureau to
- improve the credibility of schedules, including conducting a quantitative risk assessment;
- update and implement its assessments to address data quality concerns we have identified, as well as any operational benefits;
- address cybersecurity weaknesses in a timely manner; and
- continue to address our recommendations, especially those designated priority recommendations.
Congressional Actions Needed
In 2019 and 2020, we testified in six congressional hearings focused on the preparations for and implementation of the decennial census. Going forward, continued oversight will be needed to ensure that 2020 Census evaluations are completed as scheduled and that the Bureau has the resources it needs to begin planning the 2030 Census. Moreover, given the importance of the decennial census to the nation, it will be imperative for Congress to provide oversight of early planning for the 2030 Census.