[Ed. note: Welcome back to the Census Project Blog, which will resume occasional posting on several critical census issues over the coming months.]
by Terri Ann Lowenthal
Federal statistics: They don’t get no respect!
Last week, Senate appropriators, mindful of the cutthroat competition to slash federal programs more than the next guy, thoughtfully suggested that the U.S. Census Bureau could design, plan and execute the 2020 census for the amount it spent on the 2000 count. Yes, you read that correctly. While keeping costs in line with the just-completed 2010 enumeration would be good, the appropriations panel wrote in its explanation of the Fiscal Year 2012 Commerce Department spending bill (S. Rpt. 112-78), paring the price tag to match 2000, without adjusting for inflation, would win a gold star.
The 2000 census cost almost $7 billion. My economist friends tell me the Senate directive would only give the Census Bureau the equivalent of $4 billion in 2000 dollars, 43 percent less than the Census 2000 budget, to enumerate 60 million more people and 22 million more housing units than it did 20 years earlier. (The 2010 count, which battled the symptoms of a punishing recession and post-9/11 world, cost $13 billion in current dollars.)
People (all 309 million of you!), I know you are thinking one of two things. Have Senators lost their minds? Or, won’t all the new-fangled technology allow the Census Bureau to count people for a fraction of the cost? Let’s examine both propositions.
First, the state of mind of our distinguished elected representatives. To be fair, the budget process has become so convoluted and devoid of any logical progression that even the most levelheaded lawmakers can be excused for their nostalgia. But $4 billion? That was the price tag for the 1990 census. You know, the one with the highest recorded disproportionate undercount of Black Americans. The one with the lower-than-projected mail response rate, maybe thanks to a data processing machine-friendly questionnaire that looked (and read) like an SAT test. The first census to be measurably less accurate than the one before it. 1990 was the last census to advertise with 2:00 a.m. public service announcements; to ignore the vital role of community-based organizations in promoting participation; to build address lists without substantial input from local officials.
The Senate Appropriations Committee was actually off to a reasonably good start when it allocated $943 million for Census Bureau operations in the fiscal year that starts October 1. The amount is 8 percent ($81 million) below the president’s request but $89 million more than House appropriators deemed sufficient for the nation’s premier statistical agency. (A spokeswoman for House Appropriations Chairman Harold Rogers rebuked the Census Bureau for “just complet[ing] a costly census that was riddled with questionable management decisions,” saying the committee was saving money for “higher priority programs” (Huffington Post, 7/15/11). Meanwhile, the same committee applauded the bureau’s request to promote and market ongoing surveys, “given the successful use of these programs in the 2010 decennial census” (H. Rpt. 112-169). Go figure.)
Senators clearly heard the uproar from an impressive range of data users when the Census Bureau said it would cancel next year’s economic census if Congress doesn’t come up with more money than the House was considering. They directed the agency to maintain the quinquennial survey of business and industry while focusing reductions on “periodic censuses and agency-wide administrative cost savings.” Never mind that the economic census is a periodic activity or that the census director announced a money-saving move to close six of 12 regional census offices months ago. In other words, rob Peter to pay Paul, because you aren’t getting enough funding for both. Like I said, no respect.
Which leads us to our second question: Won’t the Internet or other technology-based options for answering the census and gathering data in the field bring down costs substantially? Undoubtedly, modernizing the enumeration will help the Census Bureau keep costs under control. The bureau is testing Internet response in the ongoing American Community Survey, with promising results so far. The Washington Post reported (4/5/11) that 20 percent of Canadians responded by Internet in that nation’s last census; statistical experts hope twice that many will use the Web in this year’s Canadian count to achieve a cost-savings.
But the Census Bureau will have to spend some money now to save money later. Census Director Robert Groves told a Senate oversight panel last spring that the agency “know[s] it must innovate if we are to remain useful and relevant to the country. [T]his innovation is not likely to be funded by added resources; we must become more efficient.” The bureau requested a reasonable $67 million in FY2012 to start a three-year research and testing initiative to modernize and streamline the 2020 census.
Yet the Senate is telling the agency to cut back on census activities other than the economic census. That pretty much leaves wrap-up of the 2010 count or research on improving methods for 2020 on the chopping block. The bureau could halt efforts to measure the accuracy of the 2010 census and end the program that allows challenges to a city’s housing and population numbers (which adds few changes to the results, but tell that to the mayors!). I am having trouble following the logic here, given that Senate funders want the Census Bureau to dramatically reduce the cost of planning the nation’s largest peacetime activity while exercising a “unique opportunity” to “streamlin[e] operations, eliminate[e] wasteful processes … and tak[e] better advantage of technology.” The Census Bureau last year proposed an initiative to update the nation’s address list throughout the decade, potentially saving the hundreds of millions of dollars it would cost to confirm 100 percent of addresses right before the next census starts. Congress won’t cough up the modest amount of money requested for the new program.
I think I’m getting one of my famous census headaches. Maybe I’ll channel Rip Van Winkle and wake up in time for the 2030 count.