Today, the U.S. Census Bureau released findings on health insurance coverage from the Current Population Survey Annual Social and Economic Supplement (CPS ASEC). For the first time, health insurance estimates will reflect important changes to the CPS ASEC. The CPS ASEC Redesign and Processing Changes blog and series of working papers  describe the changes made and how estimates may look different from previous releases. This blog focuses on one of the new measures — current health insurance coverage — and how it differs from the primary measure of coverage in the previous calendar year released in the CPS ASEC.

These new estimates of current coverage offer a snapshot of health insurance coverage in early 2019, when the CPS ASEC data were collected. It is important to note that these estimates cannot be used to predict whether coverage in 2019 will necessarily be higher or lower than 2018. As demographic, economic and policy changes may unfold over the next several months, we need to wait until the September 2020 CPS ASEC release to know what the full-year, uninsured rate will be for 2019. This blog describes how the new estimates are created and how to interpret them.

In 2014, the Census Bureau introduced a redesigned health insurance questionnaire in the CPS ASEC, based on over a decade of research, in order to capture timely and relevant information on the nation’s health insurance coverage during the previous calendar year, while also minimizing respondent burden. In addition to other changes, this redesigned questionnaire added a question about the type of coverage held at the time of interview. The primary purpose of adding this current-coverage question was to improve respondent recall regarding health insurance coverage in the previous calendar year. Recall bias was a documented issue with research from cognitive testing and large-scale surveys showing that the 14- to 16-month reference period was a challenge for respondents.

The health insurance section of the questionnaire begins with questions about current coverage. The computer-based data collection questionnaire is highly dynamic, tailoring follow-up questions based on previous answers. This adaptive design results in a variety of ways in which a person can report coverage. Despite that, in all cases, the first questions about health insurance ask about coverage at the time of interview for the most common sources of coverage that a person might have. Specifically, the starting questions are as follows:

• [For certain respondents, including respondents age 65 and older]: Medicare is health insurance for people 65 years and older and people under 65 with disabilities. Are you NOW covered by Medicare?
• Do you NOW have any type of health plan or health coverage?
• Are you covered by Medicaid, Medical Assistance, or (CHIP/or Medicare)?
• Are you NOW covered by a state or government assistance program that helps pay for healthcare, such as: State Medicaid, CHIP, Exchange/Portal, or other State Health program?
• Are you NOW covered by Veteran's Administration (VA) care?

If the person says “no” to all of the above, they get the following question:

• You are not currently covered by a health plan. Is that correct?

If a person said “yes” to one (or more) of the above questions, they get a set of questions about the source of the coverage they have now. For instance, if they have private coverage, they get a follow-up question asking whether they got the coverage through a family member or whether it was purchased directly. This is followed by a question about whether the respondent held this coverage all of the previous year:

• Did your coverage from (plan type) start before January 1, 2018?

If the answer is “no,” the questionnaire asks when coverage began and initiates a series of questions that captures information to “fill in the blanks” and provide information about the coverage they had during the calendar year. In other words, the respondent will answer questions about other coverage they had during the previous calendar year. While the full set of interview paths is more complicated than this implies, this set of questions allows the Census Bureau to produce two types of health insurance estimates from the same survey.

Health insurance coverage during the previous calendar year (annual coverage).

The main measure of health insurance coverage in the CPS ASEC examines estimates from the previous calendar year. In the CPS ASEC, people are considered to have coverage if they were covered by health insurance for part or all of the previous calendar year. People were considered uninsured if, for the entire year, they were not covered by any type of health insurance. That is, in the 2019 CPS ASEC, people are considered uninsured if they did not have coverage at any point during 2018.

Health insurance coverage at the time of the interview (current coverage).

The new processing system introduced this year gives us the ability to present a current-coverage measure for the CPS ASEC. Answers to the “now” coverage questions are used to create estimates that represent health insurance coverage held, as well as the uninsured rate, at the time of interview. The CPS ASEC interviews occur mainly in March with additional interviews in February and April, and estimates are weighted to the population as of March.

The current-coverage measure and the calendar-year measure described earlier give different and complementary information about peoples’ experience with health insurance coverage. Coverages can change during a year, and people can (and do occasionally) experience short periods during which they have no insurance.

In general, the percentage of the population who are uninsured at a specific point in time will be higher than the percentage of people who had no health insurance during that entire calendar year. Data from the 2017 CPS ASEC Research File and the 2018 CPS ASEC Bridge File offer support for that expectation.

Figure 1 shows the two uninsured rates available in the CPS ASEC — the calendar-year measure and the current-coverage measure — across three years. In 2016, 7.9% of people did not have health insurance in any month. Using the same source (2017 CPS ASEC Research File), the current-coverage measure centered around March of 2017 results in an uninsured rate of 8.6%. These rates are statistically different but do not imply anything specific about what the 2017 calendar-year uninsured rate will be. In fact, the 2017 uninsured rate was 7.9% (not statistically different from 2016), using the consistently edited 2018 CPS ASEC Bridge File.

Using the 2018 CPS ASEC Bridge File, we find a similar result. The uninsured rate for calendar year 2017 was statistically different from the uninsured rate at the time of interview (February to April 2018) at 8.7%. As expected, the percentage uninsured at the time of interview is higher than the percentage uninsured for the entire previous calendar year.

Finally, comparing the measures between years shows consistent results. The uninsured rates from the 2018 CPS ASEC Bridge File and the 2017 CPS ASEC Research File had similar results. Specifically, the uninsured rate for the previous calendar year did not statistically change between 2016 and 2017; and the uninsured rate for the time of interview did not statistically change from 2017 to 2018.

The Census Bureau released these new estimates to provide additional information that could be useful in understanding the overall health insurance coverage climate. As reported in Health Insurance Coverage for the United States: 2018, the American Community Survey (ACS) also provides information on current coverage. The ACS current-coverage measure and the CPS ASEC current-coverage measure have different reference periods. Where the CPS ASEC is collected during three months and centered around March, the ACS collects data throughout the year and the resulting measure of health coverage reflects an annual average of current health insurance coverage status.

Source: U.S. Census Bureau, Current Population Survey, 2017 Annual Social and Economic Supplement Research File, 2018 Annual Social and Economic Supplement Bridge File, 2019 Annual Social and Economic Supplement.