I have a question on the topcodes of the variable INCRETIR.
It’s available between 1976 and 2024 but I can only find it in the topcode tables from 1962 to 1987: https://cps.ipums.org/cps/topcodes_tables.shtml
Is INCRETIR just missing? Or is INCRETIR constructed from INCRETI1 and INCRETI2 in later years and so it inherits their topcodes? (Those are listed from 1988.)
I am asking because, if I understand the info below correctly, the sum of INCRETI1 and INCRETI2 is not necessarily the same as INCRETIR so it’s not obvious how the topcodes could be related: https://cps.ipums.org/cps-action/variables/INCRETI1#comparability_section
Because INCRETIR reports the value of retirement income from all sources, the amount reported in that variable may be greater than the sum of INCRETI1 and INCRETI2.
Many thanks for helping with this!
It sounds like you are interested in top codes to help understand cases where the sum of INCRETI1 and INCRETI2 are not equal to INCRETIR. I will share some information on the construction and composition of INCRETIR over time and direct you to the top codes where relevant.
INCRETIR is available back to 1976.
- 1976-1988: INCRETIR reports all retirement income from all sources
- 1988 - 2013: the questionnaire changed to prompt respondents to identify sources of pension and retirement income and allow respondents to detail three sources of income and their amounts (available in IPUMS CPS as INCRETIR), but only two sources and their income amounts are released in the public use data (available in IPUMS CPS as INCRETI1 and INCRETI2, SRCRETI1 and SRCRETI2 .
- 2014-2018: The ASEC introduced a redesigned questionnaire which included updates to income questions. The updated questionnaire asked separately about pensions, then retirement accounts, and lastly annuities. While the questionnaire changed, the variable availability and construction is the same as 1988-2013 and the public use data continued to report total retirement income in INCRETIR and release variables on only two sources of retirement income, including pensions and retirement accounts.
- 2019-forward: The ASEC introduced new variables that reflected the changes made to the questionnaire text in 2014, offering separate variables for retirement accounts, pensions, and annuities. INCRETIR reports only income from retirement accounts and respondents are asked to provide only two sources and amounts of retirement income (INCRET1 and INCRET2, SRCRET1 and SRCRET2). Pensions and annuities are reported in separate variables and are not included in the IPUMS CPS variable INCRETIR. Note also that beginning in 2019 there are separate variables in the original data for those 58 and over and those 57 and younger; IPUMS combines the age-specific variables.
In years where the survey collects data on more income sources than are included in the data file (i.e., prior to 2019), the sum of INCRETI1 and INCRETI2 may not always sum to the same total as INCRETIR. From 1988 through 2013, I found that there are 270 instances where INCRETIR has a greater value than the sum of INCRETI1 and INCRETI2. From 2014 through 2018, I found that there are 2,912 instances where INCRETIR has a greater value than the sum of INCRETI1 and INCRETI2. There are no such cases in 2019-forward.
As to your inquiry on top codes, beginning in 1988, INCRETIR does not have its own allocation or topcode flags in the CPS data, and instead, data quality flags are associated with the component variables.
- 1988-2018: The allocation flag variables QINCRETI1 and QINCRETI2 report the allocation status for their associated variables. TINCRETI1 and TINCRETI2 report the top code flag for their associated variables.
- 2019-forward: IPUMS CPS does not currently offer allocation or top code flags for the INCRETIR component variables, but they are available in the original Census Bureau data and I have made a note that we should consider adding them in a future release. The variable names from the original documentation are listed below for your reference.
- Allocation of first source for those 58 and over: dst_val1
- Allocation of first source for those 57 and under: dst_val1_yng
- Top code of first source for those 58 and over: tdst_val1
- Top code of first source for those 57 and under: tdst_val1_yng
- Allocation of second source for those 58 and over: dst_val2
- Allocation of second source for those 57 and under: dst_val2_yng
- Top code of second source for those 58 and over: tdst_val2
- Top code of second source for those 57 and under: tdst_val2_yng
To obtain these top code and allocation flags, you might consider downloading the data directly from the Census website, though the variable names and coding structure will differ from the IPUMS samples. You can merge observations in the IPUMS CPS BMS sample with those in the supplement using the linking keys HRHHID, HRHHID2, and LINENO (referred to as PULINENO in the CB data).
1 Like
Many thanks for this comprehensive and informative answer @Dan_Backman ! It is very helpful.
Just a small follow up question:
In the topcode tables, the Swap Value Threshold for INCRETI2 in the years 2011-2014 is not listed but given as --.
Is this a typo? (In all other years, it’s the same as for INCRETI1.)
Thanks for bringing this to my attention. I looked at the ASEC codebooks from 2011-2014, and I found some discrepancies with our documentation on the 2011-2014 Swap Value Thresholds. I have made a note with the CPS team to update our documentation. Regarding the swap threshold value of “–” in certain years, the codebooks do not specify exactly what this means, but when I look at these values in 2011, the maximum value of INCRETI2 is $58,930 and there are zero cases with a top code flag (TINCRETI2). Here are the Swap Value Threshold values across 2011-2014 for the CPS variables RET-VAL1 (IPUMS: INCRETI1) and RET-VAL2 (INCRETI2):
2011
Swap Value*
- RET-VAL1: 64,000
- RET-VAL2: –
2012
Swap Value
- RET-VAL1: 67,000
- RET-VAL2: 67,000
2013
Swap Value
- RET-VAL1: 63,000
- RET-VAL2: 63,000
2014
Swap Value
- RET-VAL1: 67,200
- RET-VAL2: 67,200