2019 ASEC Updates

In 2014, Annual Social and Economic supplement (ASEC) to the Current Population Survey (CPS) was collected using an updated survey instrument. This new questionnaire was designed to allow better estimates of income and better assess income from retirement accounts, updated to include questions about marketplace health insurance coverage and health insurance coverage at the time of interview. In 2015, a further change to family relationship questions allowed respondents to specify whether their spouse or unmarried partner was of the same or opposite sex.

In 2019, the Census Bureau completed an overhaul of its editing procedures to take full advantage of the improvements to the survey in 2014 and 2015. In addition to making new variables available on retirement income and various types of health insurance coverage, the updated processing system also offers improved imputation methods and more detailed data quality flags.

An in-depth explanation of the 2014 ASEC survey redesign can be found here. Further discussion of these survey changes, subsequent Census Bureau data processing updates, and implications for IPUMS CPS users are discussed below.

Changes to Family Relationship Variables

Prior to 2015, respondents were only able to specify whether or not they had a spouse. Prior to 2010, these couples were edited to be opposite-sex couples; between 2010 and 2018, same-sex married couples were edited to become same-sex unmarried partners. In 2015, the survey was updated to better reflect the realities of family structure in the United States.

In 2015, family relationship questions were updated in the following ways:

• Respondents may now specify an opposite-sex or a same-sex spouse and an opposite-sex or same-sex unmarried partner. These designations are available as new detailed codes in the IPUMS CPS RELATE variable.
• Parents are now designated as gender-neutral parent 1 and parent 2 rather than as mother and father.

Census Bureau editing procedures that refered to “husband” and “wife” and “mother” and “father” were updated to allow for same-sex couples. As parents are no longer described as mothers and fathers by the survey, the IPUMS CPS variables PELNMOM, PELNDAD, PEMOMTYP, and PEDADTYP are not available after 2018 and are replaced with PELNPAR1, PELNPAR2, PEPAR1TYP, and PEPAR2TYP in 2019. IPUMS-constructed family pointer variables have also been updated to account for these newly report-able types of relationships.

These relationship updates have implications for family units and, as a result, income and health insurance imputations and poverty estimates. For more information on the changes to relationship in the 2019 ASEC, see Changes to the Household Relationship Data in the Current Population Survey. For detailed breakdown of the effects of these processing updates on poverty measurement, see Updating the Current Population Survey Processing System and Bridging Differences in the Measurement of Poverty.

Changes to Income Questions and Processing

Improvements to the income questions in 2014 include the following:

• Questions about income receipt and sources were separated from questions about income amounts for all income questions except wage and self-employment earnings This shortens the survey and guards against respondent fatigue by collecting all sources of income in a first pass and income amounts from each reported source in a second.
• The family income screener question was removed to allow families earning $75,000 to be asked questions about means-tested benefits such as TANF and SNAP. Some families eligible for benefits were previously being screened out by this question.
• Income questions are now ordered based on household characteristics.

  Default		Low Income		Householder or Spouse 62+
  1	Unemployment/Workers Compensation	1	Unemployment/Workers Compensation	1	Unemployment/Workers Compensation
  2	Social Security/SS for Children	7	Public Assistance/TANF	2	Social Security/SS for Children
  3	Supplemental Security Income (SSI)/SSI Children	8	Food Stamps (SNAP)	3	Supplemental Security Income (SSI)/SSI Children
  4	Disability	2	Social Security/SS for Children	4	Disability
  5	Veterans	3	Supplemental Security Income (SSI)/SSI Children	5	Veterans
  6	Survivor Benefits	4	Disability	6	Survivor Benefits
  7	Public Assistance/TANF	5	Veterans	9	Pensions
  8	Food Stamps (SNAP)	6	Survivor Benefits	10	Annuities
  9	Pensions	9	Pensions	11	Retirement Accounts (within)-Withdrawls or distributions
  10	Annuities	10	Annuities	12	Other Income Earning Assets (outside of retirement)
  11	Retirement Accounts (within)-Withdrawls or distributions	11	Retirement Accounts (within)-Withdrawls or distributions	13	Property Income
  13	Property Income	13	Property Income	8	Food Stamps (SNAP)
  14	Educational Assistance	14	Educational Assistance	14	Educational Assistance
  15	Child Support	15	Child Support	15	Child Support
  16	Financial Assistance from friends or relatives	16	Financial Assistance from friends or relatives	16	Financial Assistance from friends or relatives
  17	Other Income	17	Other Income	17	Other Income

• If respondents did not give a value for income received from a given source, the interviewer now follows up with a question about income ranges. Respondents who did not give a specific value were asked if they received over $60,000, between $45,000 and $60,000, or less than $45,000 from this income source. Those who replied that they earned less than $45,000 from this source were further asked if they earned more than $30,000, between $15,000 and $30,000, or less than $15,000 from this income source.
• Retirement income questions were redesigned to collect more accurate data. Income from retirement accounts, pension plans, and annuities are now collected via separate questions. New questions were added for interest earned from retirement accounts. The results of these updated questions are available in IPUMS CPS as SRCRET1, SRCRET2, INCRET1, INCRET2, SRCPEN1, SRCPEN2, INCPEN1, INCPEN2, INCRANN, INCRINT1, and INCRINT2.
• Questions were added to assess capital gains income (see INCCAPG).
• Questions on Social Security and Supplemental Security Income were clarified.
• Respondents who were unsure of the amount of interest or dividend income they had received last year were asked to provide information on other assets they owned which might be used to impute income.

This 2014 questionnaire redesign, coupled with a processing overhaul competed in 2019 resulted in the following improvements:

• The earnings imputation model was modified to use income brackets. The new model ensures that imputed values fall within the range the respondent specified in the followup questions described in point four above.
• Income imputations were updated to use more variables in the imputation model.
• American Housing Survey data is no longer used to impute mortgage or property values.
• Means-tested benefits are no longer capped based on family income and these benefits can now be imputed to higher-income households. Some respondents’ energy assistance benefits had been removed by this cap prior to the overhaul.
• Data quality flags for income variables have been updated to provide greater detail about whether or not a range was used in imputation and to better differentiate between item and supplement nonresponse imputations. Data quality flags for IPUMS CPS variables are found on the “Flags” tab.
• Topcodes were increased for some income variables, as the real value of these has declined over time. IPUMS CPS variables with updated topcodes in 2019 are INCASIST, INCDISA1, INCDISA2, INCINT, INCOTHER, INCRENT, INCSURV1, INCSURV2, INCVET, and INCRETIR.

Users should be aware that, as marital status and spouse characteristics are used in income imputation models, the updates to relationship variables described above may also impact imputed income values.

For more detail on changes to income data collection and processing, see Processing Changes to Income in the CPS ASEC. More information on the use and accuracy of bracketed income responses can be found in Are Bracket Responses Accurate?

Changes to Health INsurance Coverage Questions and Processing

Beginning in 2014, collection of health insurance coverage information in the ASEC was expanded and updated in the following ways:

• Questions were added about ACA marketplace coverage, and whether that coverage was subsidized or not. These variables begin with MRK, MRKS, and MRKU prefixes in IPUMS CPS, respectively, and are part of the ASEC Health Insurance variable group.
• Questions about various types of health insurance coverage at the time of interview were added. These variables have a NW suffix and are also part of the ASEC Health Insurance variable group in IPUMS CPS.
• Limited information about subannual coverage is now included on the public use file for several types of coverage. These variables indicate whether the respondent had coverage for part or all of the previous calendar year. Relevant IPUMS CPS variables include ANYPART, PRVTPART, PUBPART, and CAIDPART.

As with income, in order to fully leverage the redesigned survey, the health insurance section of the ASEC underwent a processing overhaul that was completed in 2019. Health insurance data processing was updated in the following ways:

• Subannual coverage information was used to improve consistency in types of concurrent coverage. Information about concurrent coverage can be found in the IPUMS CPS variable MULTCOV.
• Rather than being imputed one individual at a time, health insurance variables are now imputed as part of health insurance units and information used to fill in missing data is taken from other health insurance unit members with reported data.
• Infants who were born after the calendar year (that is between January and March of the survey year) and thus could not have had any type of insurance coverage during the previous calendar year, are now coded separately from those who were in the household during the previous calendar year and did not have a given type of insurance coverage. Affected IPUMS CPS variables include PRVTCOVLY, PUBCOVLY, GRPCOVLY, DPCOVLY, TRCCOVLY, VACOVLY, and INHCOVLY.
• What constitutes private insurance changed as part of this update. Prior to 2019, Only employment-based (GRPCOVLY) and direct-purchase (DPCOVLY) coverage were considered private. Beginning in 2019, TRICARE coverage (TRCCOVLY) is also considered private.
• Some IPUMS CPS health insurance variables were re-named to impose a variable naming convention that accommodates variables regarding previous year coverage as well as current coverage.

  Old IPUMS Name	New IPUMS Name
  CAID	CAIDLY
  CARE	CARELY
  CHAMPUS	TRCCOVLY
  CHAMPVA	CHAMPVALY
  INDIANH	INHCOVLY
  PRIVDEP	DPDEPLY
  PRIVOUT	DPOUTLY
  PRIVTYP	DPTYPLY
  PRIVWHO1	DPWHO1
  PRIVWHO2	DPWHO12
  PRIVOWN	DPOWNLY
  GROUPOWN	GRPOWNLY
  GROUPDEP	GRPDEPLY
  GROUPTYP	GRPTYPLY
  GROUPOUT	GRPOUTLY
  SCHIP	SCHIPLY
  HIGROUP	GRPCOVLY
  HICOVNOW	ANYCOVNW
  QCAID	QCAIDLY
  QCARE	QCARELY
  QGROUPOW	QGRPOWNLY
  QPRIVOWN	QDPOWNLY
  QPRIVDEP	QDPDEPLY
  QGROUPOU	QGRPOUTLY
  QPRIVOUT	QDPOUTLY

For more information on changes to health insurance coverage survey questions and data processing and the effects of these changes, see Health Insurance Coverage in the 2017 CPS ASEC Research File and Health Insurance in the United States: Evaluating the Effects of Changes in the CPS ASEC.