Why This Counts: Maximizing Our Data Using the Consumer Expenditure Survey

Almost all BLS statistical programs are based on information respondents voluntarily give us. We want to squeeze as much information as we can out of the data respondents generously provide. Limiting respondent burden while producing gold-standard data is central to our mission.

Let’s take a look at how one program, the Consumer Expenditure (CE) Survey, squeezes every last drop of information from the data to provide you, our customers, with more relevant information.

What is the Consumer Expenditure Survey?

The CE survey is a nationwide household survey that shows how U.S. consumers spend their money. It collects information from America’s families on their buying habits (expenditures), income, and household characteristics (age, sex, race, education, and so forth). For example, we publish what percentage of consumers bought bacon or ice cream and how much they spent on average.

A little back story: The first nationwide expenditure survey began in 1888. BLS was founded in 1884, so the CE Survey is one of our first surveys! It wasn’t until 1980 that we began publishing CE data each year, however. A 2010 article, The Consumer Expenditure Survey—30 Years as a Continuous Survey, provides more historical information.

How is the CE program doing more with what we have?

We’ll briefly look at four different areas, starting with the most recent improvements:

• Limited state data
• Higher-income data
• Generational data
• Estimating taxes

Limited State Data – Starting with New Jersey

• Regarding geographical information, the CE survey is designed to produce national statistics. Enough sample data are available to produce estimates for census regions and for a few metropolitan areas.
• Up to now, however, we did not produce state data. The CE program recently published state weights for New Jersey, which will allow for valid survey estimates at the state level for the first time.
• State-level weights are available for states with a sample size that is large enough and meet other sampling conditions.
• Right now, the state-level weighting is experimental. We provide state-level weights to data users to gauge interest and usefulness.

 Higher-Income Table

• We evaluated the income ranges of the published tables and found that over time more and more households were earning more, and the top income range had not increased to keep pace. To provide greater detail, we divided the existing top income range of “$150,000 and over” into two new ranges: “$150,000 to $199,999” and “$200,000 and over.” We integrated these changes into the 2014 annual “Income before taxes” research table, allowing more robust analysis for our data users.
• In addition, we added four new experimental cross-tabulated tables on income without the need for additional information from our respondents.

Generational Table

Grouping respondent information by age cohort can be helpful, since a person’s age can help to predict differences in buying attitudes and behaviors. The CE program has collected age data for years, but never grouped the data into generational cohorts before. A Pew Research Center report defines five generations for people born between these dates:

• Millennial Generation: 1981 or later
• Generation X: 1965 to 1980
• Baby Boomers: 1946 to 1964
• Silent Generation: 1928 to 1945
• Greatest Generation: 1927 or earlier

The 2016 annual generational table shows our most recent age information for the “reference person” or the person identified as owning or renting the home included in the CE Survey. In 2016 we wrote a short article on Spending Habits by Generation, including a video, which used 2015 data. We’ve updated the chart using 2016 data:

Editor’s note: Data for this chart are available in the table below.

Estimating Taxes

CE respondents used to provide federal and state income tax information as part of the survey. These questions were difficult for respondents to answer.

Starting in 2013, the CE program estimated federal and state tax information using the TaxSim model from the National Bureau of Economic Research and removed the tax questions from the survey. As a result, the quality and consistency of the data increased, and we have reduced respondent burden!

If you have any questions or want more information, our staff of experts is always around to help! Please feel free to contact us.

This is just one example of how we at BLS are always looking for ways to maximize our value while being ever mindful of the costs—and one of those important costs is the burden our data collection efforts place on our respondents. Maximizing our data means providing gold-standard data to the public while reducing the burden on our respondents—a true win-win!