The Bureau of Labor Statistics (BLS) of the U.S. Department of Labor has tracked expenditures of U.S. consumers for more than a century. This chapter provides background for the Consumer Expenditure Surveys (CE), an overview of BLS’ recent efforts to improve the quality of the data collected in that survey, and the context within which this study was framed.
The CE is the “only Federal survey[s] to provide information on the complete range of consumers’ expenditures and incomes, as well as the characteristics of those consumers” (Bureau of Labor Statistics, 2011a). BLS has fielded surveys of consumer expenditures for more than 100 years. While initially conducted on a periodic basis, these data have been collected from households continually since the early 1980s. Since their inception, the impetus of these surveys has been to obtain information on changes in the cost of living (Carlson, 1974). In fact, the first two such surveys of the 20th century led to the development of the Cost of Living Index, which was the predecessor to the Consumer Price Index (CPI). Providing budget shares (index weights) for the CPI remains a primary reason for conducting the CE. During the Depression of the 1930s, the use of the survey expanded to include more general economic analysis. During the 1950s, 1960s, and 1970s, data on consumer expenditures were collected approximately once a decade (Bureau of Labor Statistics, 2008).
A modern version of the survey was first fielded in 1972–1973, with
the Census Bureau selecting the sample and conducting fieldwork under contract to BLS. This CE design was the first to highlight its current configuration of two separate surveys (a recall Interview survey and a Diary survey) working in tandem (Bureau of Labor Statistics, 2008).
BLS recognized the need to conduct a survey more frequently, noting that “rapidly changing economic conditions highlighted by the oil crisis in the 1970s illustrated the need for more frequent monitoring of the spending patterns of American consumers. . . . Rapid inflation—in excess of 13 percent from 1979 to 1980—further demonstrated the need for more frequent updates to the CPI budget shares than every decade” (Bureau of Labor Statistics, 2010a, p. 1). This led to changing the CE into an annual survey based on the 1972–1973 design.
The availability of microdata from these surveys opened the door to the investigation of a broad range of important questions, in the public as well as the private domains. As noted by the Bureau of Labor Statistics (1978, p. 1), the 1960–1961 survey “was also valuable in satisfying the growing interest of market researchers, government officials, and private users of data on income, expenditures, and assets and liabilities of American families.” Carlson (1974, p. 1) points out that by the time of the next periodic expenditure survey in 1972–1973, “[non-CPI] uses of the data [had] become increasingly important,” including the evaluation of economic policies, provision of supplemental information for the calculation of the National Accounts data, and market research.
The broad use of the CE for multiple research needs, in addition to the calculation of the budget shares for the CPI, persists to this day. As explained by BLS (Bureau of Labor Statistics, 2011a), “[The CE] is used by economic policy makers examining the impact of policy changes on economic groups, by businesses and academic researchers studying consumers’ spending habits and trends, by other Federal agencies, and, perhaps most importantly, to regularly revise the Consumer Price Index market basket of goods and services and their relative importance.” Jay Ryan (2010), director of the Consumer Expenditure Survey Division in BLS, said in a presentation to the June 2010 CE Data Users’ Needs Forum that the purpose of the CE is “to collect, produce, and disseminate information that presents a statistical picture of consumer spending for the Consumer Price Index, government agencies, and private data users.”
Since the 1980 makeover, BLS has improved the basic survey design of the CE (Bureau of Labor Statistics, 1983, 1997). The most important of these improvements were the conversion of the “Interview questionnaire” to computer-assisted personal interviewing (CAPI) in 2003 and a more “user-friendly” redesign of the Diary form in 2005. Other smaller changes also have been made, often associated with a regular biennial review that can initiate over 100 changes to the questionnaire and survey procedures.
Even so, modifications in the CE design have not kept pace with the changes in how consumers make (and remember) purchases, the reluctance of the public to respond to surveys, and the availability of newer survey methodology and technology.
In 2009, perceived decline in the quality of data collected spurred BLS to embark upon a concentrated effort to study and understand the potential sources of error in the CE. Known as the Gemini Project, this multiyear project seeks both to understand the potential causes of the decline in CE data quality and to design improvements that would reduce measurement error. As part of the Gemini Project’s efforts, BLS asked the National Research Council, through its Committee on National Statistics, to convene an expert panel. Box 1-1 provides the Statement of Task for the Panel on Redesigning the BLS Consumer Expenditure Surveys (referred to as “the panel” in this report).
Michael Horrigan, BLS associate commissioner for prices and living conditions, addressed the panel at its first meeting on February 3, 2011, providing more specific expectations. He called for flexible recommendations, saying that the “design recommendations should include a menu of comprehensive design options with the highest potential, not one specific all-or-nothing design.” He also stated that the “design recommendations should be flexible to allow for variation in program budget, staffing resources and skills, ability of the data collection contractors to implement,
BOX 1-1 Statement of Task
The National Research Council will convene an expert panel to contribute to the planned redesign of the Consumer Expenditure Surveys (CE) by the U.S. Bureau of Labor Statistics (BLS). The panel will review the output of a data user needs forum and a methods workshop, both convened by BLS. It will also conduct a household survey data producer workshop to ascertain the experience of leading survey organizations in dealing with the types of challenges faced by the CE and a workshop on redesign options for the CE based on papers on design options commissioned from one or more organizations. Based on the workshops and its deliberations, the panel will produce a consensus report at the conclusion of a 24-month study with findings and recommendations for BLS to consider in determining the characteristics of the redesigned CE (National Research Council, 2011b).
Historically, the primary use of the CE data has been for the BLS’ CPI, a Principal Federal Economic Indicator of the United States. The CPI program uses the CE data to produce the budget shares for each of 211 expenditure items, and the CE collects detailed expenditures for more than 800 items used in the construction of the budget shares. In order to produce these budget shares, the CE collects a highly detailed disaggregation of a household’s annual spending. As an additional product from the CE, BLS publishes annual expenditure tables, collapsing the 800+ items into 96 different expenditure categories. BLS also produces microlevel data files for use in basic economic research and policy analysis. The users of the microdata generally are satisfied with these more aggregated categories of spending, but they have other requirements. For example, they want a complete picture of spending, income, and assets for each household in the survey. These users also need data collected on these households at multiple time periods to facilitate investigations of how spending and income change in different conditions.
From a survey design perspective, the uses of the CE have competing requirements. Setting expectations in his original communication with the panel, Michael Horrigan stated that the “CE needs to support CPI needs” and the “CE needs to support other data users as much as possible as long as the design to meet those needs meets the needs of the core CE mission” (see Appendix B). BLS also laid out the CPI Requirements of CE in Casey (2010). In May 2011, BLS issued a separate paper entitled Consumer Expenditure Survey (CE) Data Requirements (Henderson et al., 2011), which laid out the comprehensive CE needs beyond those of the CPI. The paper states, “for purposes of this document, the CPI constraints are assumed to be suspended. This is a theoretical exercise, and in no way indicates a lack of support for the CPI program after the survey redesign. This is simply to delineate CPI versus non-CPI requirements for the CE” (Henderson et al., 2011, p. 2).
At a Redesign Options Workshop convened by the panel in late October 2011, the breadth of requirements for the CE stimulated considerable discussion. On November 11, 2011, BLS modified its expectations for the panel’s work:
Therefore, contrary to previous direction to the panel that both the CPI Requirements of the CE (William Casey, June 17, 2010) and the CE Data Requirements (Henderson, Passero, Rogers, Ryan, Safir, May 24, 2011) collectively form the requirements for the survey, the program managers ask that the panel members treat the CE Data Requirements as the mandatory requirements for the survey. The CPI data requirements document
is still helpful in terms of providing larger context for data usage, but are not requirements that the panel’s recommendations need to meet. We hope that this relaxation of constraints provides the Panel with greater flexibility in considering their recommended design changes. (See also Appendix B.)
The panel has interpreted this modification to its charge as providing it with greater flexibility in design options to consider. In particular, the panel has considered redesign options that, while supporting the CPI, do not provide the full breadth of detailed expenditures currently supplied by the CE to the CPI. Remaining is the need to provide a complete picture of spending, income, and assets for each household, and to capture data for a constant period of time and at a minimum of twice while the household is in the survey sample. There is also flexibility in these requirements: “The CE regards a complete picture of spending at the CU [consumer unit] level to be a requirement, although by using global questions, imputation, or other methods, it is not required that all expenditures be collected at the same level of detail from each CU” (Henderson et al., 2011).
It is important to note that the panel did not interpret this modification in expectations as a statement from BLS management that they have decided that the CE will not continue in the future to support the greater level of detail needed by the CPI. That is still an open question.
This report summarizes the work of the Panel on Redesigning the BLS Consumer Expenditure Surveys.
After this brief background in Chapter 1, Chapter 2 describes the many uses of the CE. The chapter notes that the CE has three critical but diverse uses, all of which have great importance for U.S. society: input into the CPI, administration of a diverse array of government programs, and research that provides insight into policy decisions such as the effects of taxes or other economic stimuli.
The current design, implementation, and costs of the CE’s two components—the Interview survey and the Diary survey—are explained in Chapter 3.
Chapter 4 summarizes the panel’s investigations into issues with the current CE. Through a workshop in June 2011 and other feedback, the panel gathered and carefully considered insights about the CE and explored how other large surveys are conducted. The panel also commissioned the development of two proposals on potential CE redesigns as a starting point to consider new directions. These workshop sessions and the two proposals, which contributed to the panel’s conclusions and recommendations, are summarized in this chapter.
Chapter 5 lays out the panel’s case about why the CE should be substantially redesigned, noting potential sources of error and respondent burden in both the Diary and Interview surveys. The chapter makes note of underreporting in both versions as compared to other sources of consumer information, most notably Personal Consumption Expenditures (PCE).
Chapter 6 presents three potential redesign options developed by the panel, as well as the panel’s recommendations in moving forward. As expressed in these recommendations, the panel urges BLS to prioritize the uses of CE data, to create a roadmap for a redesign, and to conduct targeted research to ensure that any new effort is both workable and effective.
Appendix A contains a dissent statement from three panel members followed by a response by the majority of the panel. Appendixes B through F provide additional background information on various panel activities; and Appendix G provides biographical sketches of panel members and staff.