Related BLS programs | Related articles
The impact of income imputation in the Consumer Expenditure Survey
Bill Passero
Bill Passero is a senior economist in the Branch of Information and Analysis, Office of Prices and Living Conditions, Bureau of Labor Statistics.
E-mail: passero.bill@bls.gov
With the release of 2004 data from the Consumer Expenditure Survey, the Bureau of Labor Statistics began implementing imputation for missing responses to questions about income; imputation has brought CE estimates closer to CPS estimates, but significant disparities remain between the estimates for many of the smaller components.
From 1980, the year the Consumer Expenditure Survey (CE) became a continuous survey, until 2004, no procedures were employed to produce estimates for sources of income that respondents acknowledged receiving, but for which they did not provide values. However, the release of 2004 data marked the introduction of imputation for missing income responses. With a number of years of imputed income data now available, it is possible to evaluate how well BLS imputation routines are working. The purpose of this article is to assess the impact and efficacy of imputation by comparing pre- and postimputation estimates of CE-reported income with estimates from the Current Population Survey (CPS), a large-scale household survey that has employed imputation for many years in the course of producing its income estimates.
In the next section, after a brief discussion of the background and history of income imputation in the CE, the methodology for comparing CE and CPS income estimates is presented. Then the timing of income data collection in the two surveys is examined. Timing is important because it affects the construction of matching periods for comparison. The discussion then proceeds to detail the structure and content of the income questions asked in each survey’s respective collection instrument.
Following the latter discussion, the next section of the article is dedicated to a comparative analysis of aggregate income estimates from the CE and CPS. The common income categories that can be created from the two surveys are detailed, and three alternative estimates of CE income are described. These estimates are then measured against CPS estimates. The analytical portion of this section is devoted to examining both levels and ratios of CE and CPS aggregates, for total income and by income category. The final section of the article briefly summarizes the results of the analysis and notes the direction that future refinements in the collection and imputation of income in the CE are likely to take.
This excerpt is from an article published in the August 2009 issue of the Monthly Labor Review. The full text of the article is available in Adobe Acrobat's Portable Document Format (PDF). See How to view a PDF file for more information.
Income imputation and the analysis of consumer expenditure data.—Nov. 2006.
Within Monthly Labor Review Online:
Welcome | Current Issue | Index | Subscribe | Archives
Exit Monthly Labor Review Online:
BLS Home | Publications & Research Papers
