Mises Wire

The Problem with the Census

Mises Wire Ryan McMaken

The Census Bureau has long known more about my family history than my family does. For instance, it was through old census forms that I discovered my grandmother changed her name from "Paula" to "Pauline" at some time after 1930. This was news even to her children. The 1930 census form also reported her place of birth — Mexico — and her native language — Spanish. Her occupation is listed as "cashier."

But the nosiness doesn't stop there. Completed forms from both 1920 and 1930 show the census taker apparently prodded the family for information on all household members' names, their citizenship status, year of immigration, and ability to read and write. In 1930, the census taker would have been instructed to ask if the householder rented the family home, the householder's salary, and "marital condition."

Looking at these forms, those who bring a skeptical mind to government forms and government programs might wonder why the government needs to know all this information. Many have said so publicly. This is why some politicians, pressured by voters, suggested people refuse certain census questions. At the time of the 2000 census, both Senate Majority leader Trent Lott and presidential candidate George W. Bush advised Americans not to answer questions "they believed invaded their privacy." That may have been good advice, especially since the Census bureau recently admitted it failed to protect the personal data collected on 100 million Americans.

Trump's Citizenship Question

On most days, though, the consensus among politicians, lobbyists, and activists is that it's very important to know this information. But what exactly must be known depends on one's political agenda.

For example, the US Supreme Court today heard oral arguments as to whether or not the 2020 census will include a question about each resident's citizenship status.

NPR reports that the court is "split along ideological lines on whether a citizenship question can be included on forms for the upcoming 2020 census."

The basic narrative as to the ideological split is this: the Trump administration has requested a new census question to help identify how many non-citizens there are in the United States. And where they are. (The questions on citizenship were abandoned after 1950.)

In contrast, the ideological left vehemently opposes the inclusion of a citizenship question for two main reasons:

First, it is claimed that a citizenship question would cause many immigrants to not fill out their census forms at all. Thus, the census would become more inaccurate, and be less reliable as a source of statistical information.

Second, a more inaccurate count would impact public policy because the census data is used to distribute welfare-state funds. As the ACLU puts it:

The federal government will use 2020 Census data to decide how to allocate $900 billion in funding for social service, health, and education programs. This money goes to everything from Medicaid to school-lunch programs to veterans’ assistance.

If the citizenship question results in a sizable undercount, states with large immigrant populations could lose funding for programs they need.

One follows from the other. They want a census count they can call "accurate" and then use it to push certain government programs.

A Rising Tide of Census Questions

The US census — which is one of only a handful of federal programs written into the US constitution — has always relied on the justification that it was needed for Congressional apportionment and redistricting. But to do this, the census need only collect information on where people live, and how many of them there are. The very earliest census forms don't collect much information beyond the total number of people, whether they are male or female, how old they are, and whether or not they are slaves.

Yet, by the 1870 census, the government was asking questions about birthplace and citizenship. Questions about occupation, literacy, and disability began even before then. In 1860, it was apparently essential for the federal government to know if a person was "deaf and dumb, blind insane, idiotic, pauper, or convict."

The fact that citizenship questions began is 1870 is significant. Prior to the 1870s, it was widely believed that the regulation of immigration was not a responsibility of the federal governments. Some state governments — especially New York and Massachusetts — enacted immigration restrictions in the mid nineteenth century. But both Congress and the Supreme court balked at the idea of imposing federal limits on migrants.

[RELATED: "American Immigration Policy 160 Years Ago" by Ryan McMaken]

This changed with sweeping new federal immigration laws passed in 1882.

At roughly the same time, federal policymakers started instructing the census takers to keep track of matters related to birthplace, immigration, and citizenship.

As the role of government expanded even further, more questions were added. These included questions on employment, housing, ethnic group, and more.

In the 1920s, Herbert Hoover, who support expanding the statistical-date role of the census bureau, became head of the U.S. Commerce Department. According to a history of the census by Robert Jenkins,

[Hoover] encouraged the systematization of business and economic statistics and their orientation toward use by business. Among other activities, Hoover directed the Census Bureau to compile the various series of business data and publish them as the monthly Survey of Current Business.

Not all of this information relies on the decennial census. But the regular census remained the cornerstone of federal data collection.

The role of census data expanded even more with the New Deal. Jenkins continues:

The [New Deal] legislation established new programs in many fields, including industry, agriculture, welfare, securities and exchange, banking and home mortgages. Corresponding to this expansion of government there developed a need fo rstatistical information to aid in the administraiton of the recovery program.

In other words policymakers needed more and more statistical information to justify new federal programs, and to claim that resources were being distributed equitably and rationally.

Murray Rothbard said as much in the Quarterly Journal of Economics in 1960. He noted the price system provided market information to firms in a marketplace. But government organizations don't use market prices, and thus:

Government intervention ... whether piecemeal or fully socialist, could do literally nothing without extensive ingathering of masses of statistics. Statistics are the bureaucrat's only form of economic knowledge, replacing the intuitive, "qualitative" knowledge of the entrepreneur, guided only by the quantitative profit-and-loss test. Accordingly, the drive for government intervention, and the drive for more statistics, have gone hand-in-hand. 

It is thus not surprising that opponents of the "citizenship question" oppose only the question on citizenship while also insisting that the census continue to collect myriads of other information.

This is unfortunate. After all, if the left is concerned about both the integrity of the total resident count — and the possibility of using citizenship information for anti-immigrant purposes — the solution is simple: abolish most of the census questions in favor of getting as accurate count as possible of the total number of residents. Given that the constitutional mandate makes no mention of counting only citizens, this would make perfect sense.

The left, of course, finds this solution unacceptable because the census exists for so much more than getting an accurate count of residents. Nowadays, it's there to help plan and justify the welfare state. It's there to argue for more free-lunch dollars in City X or County Y. It's there so ideologues can claim ethnic group A is "underrepresented" in Congressional district B. (But, if Trump-style ideologues get their way, they can use census data to argue welfare dollars ought to go to cities where the poor are citizens rather than fresh immigrants.)

Is the Data Any Good?

All of this, however, assumes the data is reliable. That's not necessarily a great assumption.

Once a census taker gets beyond the simple questions of how many people live in a household, things get iffy. As we go down the road of asking people about their ethnicity, income, and living standards, we have good reason to believe people fudge their responses. Or they get confused. Many people, to this day, are unaware that "Hispanic" — as far as the census is concerned — is not a racial designation.

Even the relatively simple census forms of old were prone to errors. For example, on the 1920 census form, my grandmother is listed as a "son" named "Paul." No male family member named Paul ever existed in that family. Who knows what other errors were recorded in other households?

Census takers themselves might also be biased as well as incompetent. For example, in both the 1920 and 1930 census, for example, my grandmother is listed as a resident alien. Yet, her children tell me she always claimed to have been born in the United States. Was she actually born in Mexico? Did she change her story after 1930? Or did the census takers in those days just mark down every Mexican-looking person with an accent as "born in Mexico." We may never know.

And yet, aggregates of this sort of data are used to justify and plan a nearly endless array of government programs, plans, and schemes. It's all done at our expense, and exists to favor certain interest groups. Yes, the citizenship question should be eliminated — along with nearly every other question as well.

image/svg+xml
Note: The views expressed on Mises.org are not necessarily those of the Mises Institute.
Support Liberty

The Mises Institute exists solely on voluntary contributions from readers like you. Support our students and faculty in their work for Austrian economics, freedom, and peace.

Donate today
Group photo of Mises staff and fellows