What the bill would do
The bill states that the Census Bureau "may only conduct the decennial census of population." That means no CPS, but also no Economic Census, which tracks the state of every economic industry every five years; no Census of Governments, another quinquennial survey tracking state and local governments; and no economic indicators on everything from home ownership rates to international trade figures.
It also means no more American Community Survey, the largest between-Census survey the bureau conducts, which provides more accurate poverty, income, education, health coverage and other statistics than the Currency Population Survey can.
What's more, it means that all the surveys the Census conducts for other government agencies will be done with. That includes the National Crime Victimization Survey, one of the best data sources we have on crime rates; the American Housing Survey that the Census conducts for HUD; the National Survey of Fishing, Hunting, & Wildlife conducted for the Department of the Interior; the Annual Survey of Jails; and various other data it provides to the Departments of Education, Transportation, Justice and more.
What that means for policy
It's hard to overstate the loss of knowledge that this bill would bring about. We wouldn't know the unemployment rate or how many people are working. We wouldn't know how many people are in the workforce, or enrolled in school, or retired. We wouldn't know how much people are earning, or how many are in poverty. We wouldn't know how many people are robbed or assaulted each year.
This has a concrete impact on government spending. Andrew Reamer, a Census expert, estimated in a paper for the Brookings Institution that $416 billion in federal spending rides on the American Community Survey alone. Almost two-thirds of that is Medicaid spending, which is distributed to states based on per capita income figures computed from the ACS. But tens of billions of dollars in highway money, Section 8 housing grants and special education funding rides on the ACS too. It's unclear how that $416 billion could be spent absent the data the ACS provides, and which Duncan seeks to ban.
So why is Duncan doing this? His office declined to comment on the record, but much of the irritation seems to derive from the fact that the ACS is mandatory, like the regular Census, and non-participants are subject to a fine of $100 to $5,000. Last Congress, Rep. Ted Poe (R-Tex.), who co-sponsors Duncan's bill, sponsored his own bill to make the ACS voluntary. That passed the House, as did another amendment from Rep. Daniel Webster (R-Fla.) to eliminate the survey altogether.
Poe's spokesperson, Shaylyn Hynes, told the Wall Street Journal, "He thinks that it is invasive and unnecessary for the government to ask about mental disabilities in your home, personal financial information, if you have a toilet and what time you leave for and return from work each day."
The trouble is that mandatory surveys produce far more reliable data than voluntary surveys. They prevent what statisticians call "selection bias," or the problems that arise when a non-random group of people opts out of a survey, potentially making the results less accurate.
Andrew Coulson at the Cato Institute opposes the survey on privacy grounds. "The Fourth Amendment to the Constitution protects [Americans] from government searches unless there is some compelling reason for them. But the reasons for making the ACS mandatory are not at all compelling," he writes. "Medical and social science researchers manage to deal with their inability to forcibly assign subjects to treatment and control groups."
It's true, social scientists "manage," but they'd be the first to tell you that voluntary surveys produce significantly worse data. When $416 billion in funding is on the line, it's hard to argue that the state doesn't have a compelling interest in having extremely accurate data.
Another argument against the survey came from Webster upon the passage of his amendment last year. He was quoted by the New York Times' Catherine Rampell as saying, "In the end this is not a scientific survey. It’s a random survey." Of course, randomness is a prerequisite for scientific validity in surveys, not a barrier to it.
Why it won't pass
Despite Webster's amendment passing, you shouldn't worry too much about the Duncan bill becoming law, and not just because it'd have a very hard time passing the Senate or getting President Obama's signature. That's because it has garnered the strong opposition of businesses. It turns out that many corporations, in particular in the retail sector, use the survey to make business decisions. For that reason, Rampell notes, business groups such as the National Retail Federation and the U.S. Chamber of Commerce issued statements supporting the ACS and, in NRF's case, explicitly supporting keeping it mandatory last time it was threatened.
Target is a great example of how businesses use the data. It uses the ACS to figure out where to locate stores, and what to stock in them. For example, in areas with higher usage of public transit, it stocks smaller units that are easier to carry on buses and subways. But they can know about stuff like that because the Census asks:
It's no secret that the business lobby holds considerable sway in D.C. In this case, data fans can be heartened by the fact that their interests and those of big business are pretty well aligned.
Update: This post originally said the Census Bureau conducts the Current Expenditures Survey and implied it conducted the Current Employment Survey; those are conducted by the Bureau of Labor Statistics. It also stated that it conducts the National Longitudinal Surveys of Youth; it does not. Earlier studies in the National Longitudinal Surveys program were conducted by Census, but those earlier studies are no longer being conducted. We regret the errors.