Skip Navigation

Mr. Chairman and Members of the Committee: I am pleased to meet with you today to discuss the decennial census. I believe it would be most useful if, after making a few general remarks, I devote the rest of my statement to some topics that I hope your subcommittee will consider as it evaluates the Census Bureau's proposal for the 1980 decennial census.
 

Introductory Comments

The United States Constitution requires that an enumeration of the population be taken every ten years to meet the reapportionment and redistricting needs of the House of Representatives. However, since the first decennial enumeration in 1790, each succeeding census has been expanded to secure more and more information. As the scope of the decennial census has expanded, the Congress has sought answers to a series of important questions, such as:

• Why does the decennial census keep growing?
  
  
• Is the additional information worth the cost?
  
  
• Is there a better or more efficient way to collect needed additional information so that the decennial census can be simplified?
  
  
• Who benefits from the additional information that has been obtained through the growth of the decennial census?
  
  
• Is the Census Bureau providing the Congress with the information it needs to fulfill its policy making and oversight roles?

While the Congressional Budget Office (CBO) has not addressed these questions specifically, our experience as an analytic support agency of the Congress has provided certain perspectives that might be helpful to the subcommittee.
 

Governmental Need Primary

The primary rationale for the collection of additional census data, beyond the Constitutional requirement of a decennial enumeration of the population, is to provide the federal government with the information it needs to make and implement public policy. In determining whether and how given data should be collected, the test of governmental need should be the main consideration.

In choosing the appropriate mechanism to collect a given type of data, a continuing tension will always exist between the desire to have more and more data at lower levels of aggregation and the cost of such a collection--both in terms of dollars and in terms of the potential threat to citizen privacy. The modern sample survey, being less expensive and affecting fewer citizens, can provide a partial way out of this dilemma. But, even the use of surveys can not solve the entire problem, for there will always be a need for some data to be collected at the census-tract level. Analysts, and scholars--whether they are located in government, private industry, or academia--will always be able to present good arguments on why given data should be collected through the decennial enumeration rather than from a sample survey.

In evaluating these claims, the Congress should apply the following test--what information must the government have in order to make decisions and implement policies and programs? In almost all cases, the information which is needed for the various forms of policy planning, such as policy analysis and evaluation, can be obtained through sample surveys. However, the first task of the Congress is to decide which information it will need at the census-tract level and which information can most suitably be obtained through sample surveys.
 

The Need to Look Ahead

That the Congress should apply the test of the government's needs does not necessarily mean that the Census Bureau should collect less information, either through the decennial census or through its various surveys. Rather, it means that the Congress needs to look ahead to make sure that the Census Bureau collects the information that will be needed for the determination of public policy in the 1980s and 1990s.

In order to ensure the quality of the data collected, the questions and procedures to be used in a given decennial census must be pretested years in advance of the actual enumeration. Moreover, in order to ensure the comparability of census data over time, many questions on past enumeration questionnaires must continue to be included. Thus, the ability of the Congress to change the content of the decennial census once planning has begun is fairly limited.

Even the excellent new procedures instituted two years ago, by which this Committee reviews the Census Bureau's proposals for the decennial census three years before the actual enumeration, can have only a limited impact. Proposed questions can be eliminated fairly easily, but additional questions can be added only at great cost.

This process of the Congress reviewing proposals rather than setting its own agenda has meant that, more often than not, needed data has become available only after the policy determination has been made. For example, in the 1960s, the Congress and the Kennedy and Johnson Administrations became concerned with the degree and extent of poverty and discrimination in the United States. Once that concern was identified, data began to be collected. This information was used in analyses designed to support the policy-making process. But, because of the long lead time required, quality data were not available when the major policy decisions of the "War On Poverty" were being made.

Many of the major issues that the country will face over the next two decades can already be identified. To the extent possible, the Congress should make sure that, when it turns to these issues, the necessary data will have been collected so that the Congressional decision^making process can be supported by the highest quality policy analysis. Thus, along with your review of the Census Bureau's proposals for the 1980 census, it would be useful if the Committee would begin to think of those policy issues and areas that will be in the decisions-making arena in the late 1980s and 1990s.
 

Data Requirements for Policy Planning

As previously stated, the CBO believes that many of the data requirements of the policy planning process can be met through sample surveys. However, the nature of the survey data that are currently available prevents policy analysts from providing answers to many of the questions posed by decision makers. These limitations are frequently due to one or more of the following reasons:

• The data have not been collected at a sufficient level of detail.
  
  
• The data are drawn from a survey sample that does not allow analysts to make estimates for a given geographic unit.
  
  
• The data, taken from surveys that are several years old, are out of date.

These limitations can be illustrated by examining CBO's efforts to provide policy analyses related to two issues before the Congress--welfare reform and energy policy.

Welfare Reform. This is an issue that the Congress will, in all probability, be dealing with in the near future. President Carter has indicated that he will send his proposals in this area to the Hill by the first of August. In anticipation of this issue, the CBO has been studying and evaluating--both in terms of dollar cost and distributional impact--most of the major welfare reform alternatives. Thus, we have some direct knowledge of the limitations of the present data in answering Congress's needs. It is also important to remember that the data limitations exist even though data on topics related to this policy area have been collected for almost a decade.

Given the continuing interest in the area/ one would expect that adequate information and analyses would be available to support the Congress when it takes up this issue. Many questions can be answered with the information that is available, but others cannot, due to limitations in the available data.

Frequently, the data that we have are not detailed enough to answer a given question. Thus, while we have data on total transfer income, we lack information on the amount of transfer income by specific source such as food stamps, housing assistance, and medicaid. Similarly, while we have some knowledge of the number of families on welfare at any given time, we lack data over time that would tell us the number of times that a given family is on welfare in a given year. Because of this, we cannot estimate the welfare turnover rate and the monthly case load from annual data. We lack data on the receipt of nonwelfare governmental services such as Basic Educational Opportunity Grants, educational training grants, manpower training programs, and child nutrition programs.

To some extent, the preceding represents a wish list of all the data one would need to conduct policy analysis on welfare reform. However, because welfare reform involves the possible modification of one or more interacting programs, such detailed data are needed to estimate the impact of potential program changes.

The current data are also limited by the size of the survey samples from which they are drawn. In voting on a major policy issue, Congressmen are interested not only in how the alternatives would affect the nation at large, but also how they would affect their own districts.

The data that CBO has used in its welfare reform analysis are drawn from surveys whose sample size is such that the impacts of potential changes cannot be estimated at the Congressional district or county level. If the Congress feels that it would need such analyses, it must be willing to appropriate funds for much larger samples.

Energy Policy. This is an issue that will face the Congress in the near and long term. In the near term, the Congress will be dealing with President Carter's proposed energy plan. But, as the Carter plan itself points out, energy issues will be before the Congress with greater and greater frequency over the next several decades.

Because the nation has only recently come to think of energy as a national problem, the size and quality of the energy data base is less developed than is the case for the data base associated with welfare reform. The CBO, having just completed an analysis of President Carter's energy proposals, is all too well aware of the limitations of the present data base. The gaps are larger, but the problems of the present data base are similar in type to those of the data base associated with welfare reform.

Once again, there is a lack of detail in the data that limits the scope of the analysis. This is particularly true for data on energy consumption levels. For example, we know very little about the consumption of different energy products by different income groups. Information is also lacking on the number of apartment dwellers who pay their own energy bills.

Also, as with welfare reform, the energy data base has been collected from surveys the sample size of which prevents analysis at the state much less at the Congressional district or county level. As part of its analysis of the President's proposals, CBO conducted an analysis of the proposed program's distributional effects. While that effort was able to show that the poor would not bear an undue burden under the Carter plans, it could not estimate the proposal's burden on different categories of Americans in different regions, states, or counties. Therefore, if the Congress is interested in such analyses, it will have to be willing either to appropriate funds to conduct surveys with much larger sample sizes, or be willing to expand the questions included on the decennial or mid-decade census.

Finally, much of the energy data base is out of date, having been collected before the oil embargo of 1973, While the aging of data is a continual problem in all policy areas, it is particularly so in a field like energy, that has been dramatically transformed in the recent past. Thus, for certain types of data, more frequent surveys would be useful. By looking ahead to anticipate its needs, the Congress can ensure that such data will be available when needed.
 

How to Look Ahead

I have indicated that the prime test of what questions should be included on the census is the need of the federal government. I have also stated my belief that, in order for the Congress to be able to control what questions are included on the census, it must take a long-range view. That is, it must begin thinking now about the 1985 mid-decade census, the 1990 decennial census, and what data should be collected from future sample surveys.

Given the day-to-day pressures of the legislative schedule, it will be very difficult for Members to worry about questions that need to be included on a census or survey 15 or 20 years into the future. Therefore, it might be advisable for the Congress to set up a continuing national commission that would survey the policy data needs of the federal government, states, and localities. Such a commission could then make recommendations to this Committee as to the proper priorities that should be set among the competing data needs.

By coordinating its activities with appropriate Congressional support agencies, such as the Congressional Research Service and the Office of Technology Assessment, the commission could also advise this Committee, and through it the Congress, as to which issues will require data analysis a decade or two from now. Finally, such a commission might be helpful to this Committee in overseeing the more technical details of the Census Bureau's activities.

Mr. Chairman and Members of the Committee, I appreciate this opportunity to present my views on this most important topic. I will be happy to answer any questions.