I am having trouble replicating the IRA Keogh 401k income estimate for persons income aged 65+ in 2015 ASEC.

I am comparing estimation results with Table PINC-09 (link below) for persons aged 65 and older and generally find “close enough” estimates with small differences that I assume are due to topcoding. But my estimate for 401k/IRA/Keogh income (derived from INCRETI1 and INCRETI2) is off by more than 20%. This variable was affected by the 2014 redesign, but the codebook value 7 labels for SRCRETI1 and SRCRETI1 do not seem to reflect the changes in the questions, which are now supposed to account for lump sum distributions and not just “regular payments.” However, I don’t think this is the source of the problem since as of the 2015 ASEC everyone was asked the same questions. I also assume 403b and SEP plans are included even though they aren’t mentioned anywhere except in the survey questions.

Any idea what’s going on? Any help would be much appreciated.


I am unable to replicate your issue. I created a variable and named it INC_IRA, which equals INCRETI1 if SRCRETI1=“7:Regular payments from IRA, KEOGH, or 401K accounts” or equals INCRETI2 if SRCRETI2=“7:Regular payments from IRA, KEOGH, or 401K accounts”. Weighting this variable by WTSUPP produces the following results: 3,634,808 persons with a positive value of INC_IRA and a mean INC_IRA value of $12,224. The number of persons exactly matches Table PINC-09 (Both Sexes) and the mean income is off by only $6. I recommend creating a variable using the same method I have described.

Hope this helps.

Hi Tim,

I am very sorry for wasting your time, but it turns out I just made a STATA coding error. I had checked the relevant section of my do-file several times but somehow missed it. I don’t know if there’s a way you can just delete my question, but if so, please do. At least I became familiar with this very useful resource–I learned things while browsing other questions to see if anyone else had had the same problem.