Decennial census could cost billions more

The Census Bureau's calculation for how much the 2010 census will cost is flawed and the total price tag for the decennial count could be billions of dollars more, according to a report released Wednesday by the Government Accountability Office.

Comment on this article in The Forum.GAO concluded in the report that the bureau's original estimate of $11.8 billion was not reliable because Census did not document all expenses, omitted some costs and failed to include up-to-date assumptions. Furthermore, it did not hire an independent party to validate its cost estimate.

Census revised its estimate in April when it announced it was abandoning plans to use handheld computers to record data from households that did not return census forms and instead would use the more costly paper forms. The bureau said the census would cost between $13.7 billion and $14.5 billion. GAO now says that estimate was likely too low because the original forecast was flawed.

"Without improvements in its cost estimation practices, the bureau's ability to produce and maintain well-documented, comprehensive, accurate and credible life-cycle cost estimates . . . as well as its ability to make informed decisions and effectively manage costs, will continue to be hampered," GAO reported.

The final cost could eclipse $14.5 billion, even if Census doesn't encounter more problems with other technology projects or general operations. For example, the bureau failed to adequately document how it calculated the costs ofmajor decennial programs, thereby making it impossible to verify the accuracy of its estimate.

In addition, Census' estimate did not include the potential cost of fingerprinting the almost 600,000 temporary workers that it would hire to go door to door to count households. The bureau said it would fingerprint workers with the assistance of the FBI. Census explained in a written response to GAO that the cost of fingerprinting was not included in the estimate because the decision to move forward with the program was not finalized until recently.

Census also failed to update its assumptions about how productive workers would be using new technology to record new addresses, an operation that will take place next year. The bureau learned during last year's dress rehearsal that it most likely overestimated how quickly workers could canvass neighborhoods to input new addresses. GAO reported that if Census revised its productivity assumptions, the cost to conduct the address canvassing would add $270 million to its estimate.

Finally, Census did not quantify the risks of the program or ask an independent party to provide an estimate, making it impossible to validate the total cost of the count, GAO said. In addition, the bureau did not have the staff available or policies in place to calculate an accurate cost estimate.

GAO recommended that Commerce Secretary Carlos Gutierrez direct the bureau to update its original assessment to reflect the most recent data and perform analyses to validate its cost estimates.

GAO also urged the bureau to establish policies that would result in a more accurate cost estimate for the 2020 decennial. Accurate decennial estimates are crucial because bureau executives use the assumptions they're based on to formulate Census' budget and appropriations requests. Without an accurate cost estimate, Congress' ability to perform oversight would be constrained, GAO added.

Census generally agreed with GAO's findings. The bureau is working to develop a budget integration tool that will support the cost estimation process and is considering hiring additional experienced estimators.