# INCCAPG B/W gap

I am puzzled as to why Black, Non-H INCCAPG > White, Non-H INCCAPG in the 2022 ASEC:

I can see that Blacks are only 6.5% (210/3275), but still, isn’t the \$4K gap kinda crazy?

Does anyone know (or can speculate) on what is going on here?

Thank you,

Humberto

With the help of a friend, I figured out what is going on here.

One thing that I neglected to mention is that I was cutting off values for INCWAGE < 20000 in the question above. I have explored different cutoffs, but this is not driving the result.

What explains things is that INCCAPG has an extremely long tail. Of the 210 Black obs, the four top values sum to 2,619,999. Averaged over 210 people, the capital gains income of these 4 alone comes to \$12,475. The total income of all 210 people is 210 x \$24,014 = \$5M, so these four people alone make more than half of all the capital gains in the entire sample of Blacks in 2022.

I got INCVAPG for 2021 and the B-W gap is positive over all INCWAGE cuts. This is because the 2021 sample does not include the high value outliers that are in the 2022 sample.

Hope I didn’t waste anyone’s time on this.

Humberto

I’m glad to hear that you were able to figure out what was happening in the data. I unfortunately have been unable to replicate your results using either the microdata or the online analysis tool and would like to help figure out what is happening. In the screenshot below, I run your analysis using the means program in the online analysis tool, setting INCCAPG as the dependent variable and filtering to in-universe values, RACE as my row variable, and restricting the sample to non-hispanics (HISPAN = 0) in the 2022 ASEC. I find that White non-hispanics who earn any capital gains earn on average \$23,307 as opposed to the \$20,606 earned by Black respondents. Restricting the analysis to those with INCWAGE < 20000 increases the gap further: White respondents earn \$20,049 while Black respondents earn \$6,266. 284 of these respondents are White and 18 are Black. Please let me know if I’ve misunderstood your analysis.

Thanks for this reply. I don’t see a way to attach a file – how can I share a file with you? You can email me directly at hbarreto at depauw dot edu.

I also noticed that there’s a supplement weight, sdawt, that I do not have in my data set. I have asecwt and asecwth. Are we looking at the same data?

Thanks again,

Humberto

You can attach a file by replying to this post and clicking the upward pointing arrow on the top toolbar. You can also email us at ipums@umn.edu. SDAWT is the name for ASECWT in the online analysis system.

It won’t let me upload an Excel workbook with macros, it’s .xlsm extension.

I’ll strip the macro and try again.
inccapgNOMacrosVersion.xlsx (3.4 MB)
Yes, without macros, it let me upload it.
Please download it and see what I did. I think there is enough documentation to see if you get the same results as I did.
Thank you,
Humberto

While reviewing code and analyses such as this is beyond the scope of our user support team, I was able to take a quick glance at the document. I believe the issue is that you’re using unweighted counts in your analysis when instead your averages should be weighted using ASECWT. The IPUMS-CPS samples are weighted, with some records representing more cases than others. This means that persons and households with some characteristics are over-represented in the samples, while others are underrepresented. This blog post explains how to calculate weighted averages.

No, I don’t the weights are driving this result at all. I checked and White and Black non-Hispanic weights are pretty similar. The answer is that there are a few extremely high inccapg values for Blacks. It’s definitely an outlier issue.

I did not mean to have you continue working on this – I thought you were the one who wanted to replicate what I had done. Please forgive me if I am confused on this and thank you for your replies.

Humberto