In the wake of every United States Census but one, Congress has undergone a process known as “reapportionment”: the reallotment of congressional seats available to each state proportionate to its share of the national population as recorded by the (Constitution mandated) decennial census. The census of 1920 was the first-ever exception, and now, a century later, 2020 threatens to be the second.

The 1920 story: a largely rural and conservative Congress, with immigration restriction its legislative priority, learned to its dismay that the 1920 census confirmed America’s cities, especially in the northeast, to be centers of substantial population growth. If, as the Constitution required, the census were used to reapportion Congress, eleven seats would shift from rural to urban states, increasing the electoral base of left-leaning, unionized immigrant voters. Obviously, claimed alarmed congressmen, the census must be flawed. It was best to postpone the reapportionment until the nation could have a “better” census, a decade hence. And so it was—reapportioning canceled, constitutional language ignored in the process.

The 2020 story: if the Census Bureau remains seriously underfunded into FY’18, as now expected by professionals, the 2020 census will fall below the high quality standard set in 2000 and repeated in 2010. For example, the 2020 census design is based on two never-used procedures—reliance on the internet as the primary mode of response and reliance on administrative records for an expected third or so of the population who do not initially respond. Although these design innovations look good on paper, the Census Bureau planned a thorough field test, knowing that adjustments would be required. Adequate funding for the test was not provided. It’s like building a new fighter plane but skipping the flight test phase before sending it into action.

Starting in 2000, the Census Bureau made major use of paid advertising and active partnerships with trusted voices. These innovations successfully promoted census cooperation and reversed a three-decade decline in self-response rates. Although similar initiatives are in the current census plans and contracts have been signed, funding is well short of what is needed to prepare and implement the effort, which now almost a year behind schedule. Absent effective advertising and active partnerships, the 2020 census—already facing high levels of general public mistrust toward the government—will be handicapped. Small setbacks, unavoidable in any census, will be highlighted and probably exaggerated by intense media coverage, risking further erosion in public confidence. An underfunded Census Bureau will be blamed, however unfairly, and labeled as one more incompetent government agency.

Might we hear again that a flawed census cannot be relied upon for reapportionment? When Supreme Court nominees cannot get a hearing; when there are no nominees for hundreds of high-level government positions; when data are systematically withdrawn from government sites, laying blame on the Census Bureau and refusing to apportion would fit the political times.


An underfunded Census Bureau will be blamed, however unfairly, and labeled as one more incompetent government agency.


There will be resistance. The stakes in 2020 are vastly higher than they were a century ago. In 1920, congressional seats but not federal dollars were at stake. Today, the federal government allocates almost $600 billion via formulas directly tied to census counts.1

A century ago, there was no far-flung survey and polling industry that standardized sampling methods to the basic census counts; there were not commercial players in the thousands that depended on the economic census and other surveys of economic activity; there were not governors, county commissioners, and mayors who started their work day by “checking the current numbers.” There were no social security checks indexed to federal statistics. There were no social justice claims based on disparate impact or underrepresentation, concepts based in census measures and argued with statistical trend lines.

In 1920, even reapportionment was largely an inside-the-beltway argument. Not today. One political party—through gerrymandering based on census counts—enjoys a roughly two to one advantage in state legislatures and governorships, and has not done badly in securing a congressional majority, even though its voting strength is at parity with the opposition party. This has not escaped the attention of the rival party, which is now increasingly contesting this partisan imbalance in the courts and a Supreme Court ruling about gerrymandering within reach.

Because census counts, and federal statistics more generally, are widely used and the stakes are high, an underfunded 2020 census will face charges that its “flawed counts” put at risk the statistical tools needed for intelligently governing society and managing the economy and have lessened the nation’s ability to detect and correct social injustices. A likely consequence is dissatisfaction and frustration across the political and economic landscape.

If so, more than reapportionment will be at risk. Proposals to privatize the census, even federal statistics more generally, will surface. After all, the industry titans of the Internet know everything about us already. Perhaps, but the Census Bureau offers what the Internet does not. The Bureau does protect privacy and confidentiality; it does approach universal coverage, reaching even to millions of Americans beyond the borders of the Internet; it does have quality standards unmatched by the “good enough” inclinations of much of the Internet. But when these arguments are stacked up against the promise of a fit-for-purpose census at a fraction of the cost and at a speed not possible by a clunky government agency, what was unprecedented suddenly becomes plausible.

Of course privatizing and profiting from the census goes deeper than controlling costs. It reaches to basic democratic principles. If the 1920 Congress could ignore the Constitution, might the current Congress equally overlook constitutional language that assigns to the census the fundamental task of ensuring political representation proportionate to population size? At a political moment when destabilization is urged, when familiar and honored practices are pushed aside, when demands for doing things radically different are heeded, putting the Census Bureau into the mix risks eroding the core belief that the census is a public good, woven into the fabric of how power is allocated in America’s democracy.