An official website of the United States government

The .gov means it's official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you're on a federal government site.

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Bureau of Labor Statistics > Consumer Expenditure Survey > Methods

Panel Survey of Income Dynamics

The Panel Survey of Income Dynamics (PSID) is the longest running longitudinal household survey on both a domestic and international scale.¹ It was instituted in 1968 to analyze income and poverty dynamics in response to former President Johnson's Great Society but has since then expanded to focus on income and expenditure data at the family and individual level as well as other specialized studies. Such studies include but are not limited to: family structure, disability and time use, child development, as well as wellbeing and daily life. The PSID collects expenditure data in a near homogenous structure to the Consumer Expenditure Surveys (CE) with a few noteworthy exceptions regarding specific categories and subcategories. The PSID is published biannually with 2021 being the most recent data available.

Website	https://psidonline.isr.umich.edu/
Release Schedule	Annual release schedule from 1968-1999, Bi-Annual release schedule since 1999.
Data Source	Data are sourced from an annual survey to now 9,000 family units, the PSID rendition of CE's consumer unit (CU). Family consumption patterns are analyzed over multiple generations, factoring in split offs when individuals leave a given family unit.
Data Type	Estimates and Packaged Data.
Collection Unit	Family units. Data have been collected on the same group of families and their offspring/descendants dating back to the start of the survey in 1968.
Sample Characteristics	The PSID began following a cluster of 4,802 family units in 1968, drawn from two independent samples. Low-income families were oversampled where they represented 1,872 family units from the Survey of Economic Opportunity (SEO). The remaining 2,930 family units are composed of a nationally representative sample from the University of Michigan's Survey Research Center. The active panel of family units adjusts annually based on factors such as individuals moving in and out of home, immigrant refreshers to the sample, demographic inflows such as births, adoptions, and marriages, as well as attritors. For a visual flow chart, reference figure 1 in the 2015 User Guide PDF linked in the methodology section of this table.
Notable Sample Exclusions	The PSID includes a category under food expenditures for "food delivered," separate from "food away from home," which is combined with food away from home to create a category comparable to CE. PSID does not break down its mortgage data under housing into interest and principal, unlike the CE, thus the two were combined in the CE for comparison purposes. For healthcare, the PSID did not have a medical services category, thus expenditures on doctors along with nursing home and hospital expenditures were combined for comparison purposes. PSID has a separate trips category that is not present in CE, thus one was generated utilizing out of town spending on transportation, lodging, and entertainment.
Methodology	https://psidonline.isr.umich.edu/data/Documentation/UserGuide2021.pdf https://psidonline.isr.umich.edu/data/Documentation/UserGuide2019.pdf#page=35 https://psidonline.isr.umich.edu/data/Documentation/UserGuide2015.pdf

Data Comparison

CE estimates from the Bureau of Labor Statistics (BLS) and PSID estimates from the University of Michigan are two frequently used sources for domestic household expenditure data. As stated above, although the purpose is the same, there are differences in how data is categorized, parsed out, and represented in the two surveys. Such heterogeneity leads to differences in total expenditures and expenditures by category. A ratio of CE expenditures divided by PSID expenditures illustrates how much consumers are spending as captured by CE relative to PSID.

Chart 1 below provides estimates of total expenditures for both surveys over a 20-year time series. Through 2013, CE estimates remained below its PSID counterpart, albeit with a narrowing expenditure gap from 2005 to 2013, the difference ranging from 1.0 to 6.1 percent. Beginning in 2015, CE expenditure data exceeded its PSID counterpart, although this trend began to reverse in 2021. The initial increase in total expenditures for CE compared to PSID resulted from growth in transportation and housing expenditures from 2011 to 2013. CE housing expenditures increased again from 2013 to 2015, but higher healthcare spending accounted for additional growth in total expenditures instead of transportation in 2017. The growth of total expenditures for CE compared to PSID in 2019 data was fueled mainly by transportation, trips, and entertainment expenditures instead of healthcare and housing expenditures. The 2021 data was the first set of data affected by the COVID-19 pandemic. As stated, 2021 was also the first year since 2007 where convergence between the two data series was observed. This can be traced back to higher food spending, specifically away from home, clothing, and trips. These trends can also be related to a change in the structure of the PSID survey. Beginning in 2015, the PSID adjusted their survey questionnaire. The survey was extended to enhance content on everything from internet access and food security to health insurance and wealth and pension recordings. For additional information on changes made to the PSID, please see The PSID Main Interview User Manual: Release 2021. For supplemental information on the differences between the two surveys, please see A Comparison of CE and PSID Expenditure Data: 1999-2011.

For more information on the figures outlined in Chart 1 below, please see the PSID tab in CE data comparisons linked below.

Analysis Methodology and Concordance:

CE estimates provided in this survey comparison were from annual integrated means tables as well as the CE's LABSTAT Database which provides time series expenditure data back to 1984.

PSID data was obtained from The University of Michigan's PSID Variable Database. Specific categories and subcategories were adjusted based on the parameters established in the "Notable Sample Deviations" section of the main data comparison table.

Supporting Documentation

CE data comparisons (XLSX)

Last Modified Date: November 8, 2023