Wikipedia:Reference desk/Archives/Humanities/2020 April 13

= April 13 =

Single black women's median net worth in Fed Survey of Consumer Finances, 2007-present
In 2010, the Pittsburgh Post-Gazette made national headlines exposing the fact that the Federal Reserve found that single black women had a median net worth of only five dollars in 2007. Since then, I have been unable to find anything more on that specific statistic. For example points out that blacks of all genders and marital status had a net worth of $17,600 in 2016, but there's absolutely nothing about the median net worth of single black women specifically. Can anyone determine from the 2016 and 2010-2013 data what the figures were for those years? The 2019 data isn't expected until October, for whatever reason. EllenCT (talk) 02:10, 13 April 2020 (UTC)


 * This (pdf) is the original study the article in the Pittsburgh Post-Gazette was based on. --Lambiam 08:45, 13 April 2020 (UTC)


 * I spent 20 minutes (all I can afford at this time) looking at the 2016 Survey of Consumer Finances, and the only thing that jumps out at me is that the data are all about families, not single people. DOR (HK) (talk) 15:48, 13 April 2020 (UTC)


 * I found which suggests it's fallen since, but it's not an apples-to-apples comparison. I will dig into the data today. EllenCT (talk) 19:21, 13 April 2020 (UTC)

Well, I can't repeat the $5 finding. I downloaded the pertinent data from, processed it as per and came up with :

Median net worth of single adults in 2016 dollars RACE Black        Hispanic         Other           White SEX Female   Male   Female   Male Female    Male  Female    Male YEAR 1989  1828   5345     1212   3545  20002   66051   95270  127951 1992   7186  20728     1710   1509  46957  128144   82033  102066 1995   4361  10179     4854  11337  11290   48232   83115  109620 1998   5016   8026     7672   9885  21873   33640   90061  125192 2001   4755  14496     5012  11042  10987   53381   84813  147881 2004   4959  16762     3433   9856  11573  272415   69184  158209 2007   5680   9032     3995  23027  26715   85403  122112  194500 2010   5532   8897     5305  12711  32386   11335   61710   81794 2013   5681   4949     5454   8578  11342   62454   59814   76574 2016   5230   9000     6430  11970  30000   48600   91820  103500

I wonder if the Post-Gazette can remember how they got $5 instead of $5,680. EllenCT (talk) 02:55, 14 April 2020 (UTC)
 * A link to the original study the article in the Pittsburgh Post-Gazette was based on is given above. --Lambiam 03:44, 14 April 2020 (UTC)
 * Thank you! I missed that .pdf. It looks like the figure is for ages 36 to 49. I'll give that a try. EllenCT (talk) 05:01, 14 April 2020 (UTC)

Median net worth of singles age 36 to 49 in 2016 dollars RACE Black        Hispanic          Other           White SEX Female   Male   Female   Male  Female     Male Female    Male YEAR 1989   690  31029        0  22688   20002   196288  61646  180176 1992   7680  41490    43041      0  114761    46874  50395  111517 1995   7970  20107        0  63203    7360    43691  61371  113143 1998  10608  11803    22869  31235    9324   547320  53013  123864 2001   8129    880     5283  11678   11726    62865  72647  147881 2004   8222   8139     3433  36245   98505  1586548  37072  112807 2007   1042   5963     5639  17561   13346   214753  62880  124370 2010   6854  10644    12490  16414    9284     7847  14921   57919 2013   4804   4784     4846   9073   10311    83246  16201   73208 2016   5260  13010     6100  14260   75350    59650  34580   85800

Median net worth of singles age 20 to 39 in 2016 dollars RACE Black        Hispanic          Other         White SEX Female   Male   Female   Male  Female   Male Female   Male YEAR 1989    18   2612        0   1865   11913  16503  11288  38492 1992    167  14909        0   1006    6540  78771   9056  25826 1995    876   6232      626  14955    6577  21610  16756  25056 1998   1785   3482     1445      0   13721  43821   6506  16230 2001   1693   1625        0   7722       0  62865   6367  15851 2004   2734  13353     4324   9856     190  22001   5424  20221 2007   1495   3010     2200  26634  124081  10561   6398  19686 2010    464   5305     3537   7074   19044   7571   5311  14159 2013    206   3196     4521   6681   11032  26190   4490  15012 2016    275   2785     3540   3800   17100  21000   5170  13340


 * Well, that's not it. I'll try contacting Mariko Chang tomorrow. Thank you for your help. EllenCT (talk) 06:03, 14 April 2020 (UTC)
 * Are these figures based on very small sample sizes especially for the 'other'? Some of them are very weird, e.g. the 2004 other male comparing to other years and female and different age ranges. Does 0 mean it's actually zero or there is no data? It's median so having a few outliers shouldn't affect things much unless the sample size is tiny so the median happens to fall within what is probably an outlier. Nil Einne (talk) 13:28, 14 April 2020 (UTC)
 * I thought so too, but here are the counts for single respondents between the ages of 20 and 35:

Count of singles age 20 to 35 in the Fed SCF RACE Black      Hispanic       Other       White SEX Female Male   Female Male Female Male Female Male YEAR 1989   215   50       50   35     40   40    385  360 1992    225  105      130   60     64   45    516  630 1995    355   90       80   55     30   45    610  670 1998    325  125      105   65     55   30    615  625 2001    356   80      138   53     35   42    601  580 2004    290  155       90   92     25   35    588  528 2007    315  130      105   50     30   50    410  380 2010    471  155      301  190     56   91    707  844 2013    496  164      190  185     70  100    744  731 2016    425  179      190  120     90  176    640  760


 * I think 0 is a very common median value for net worth, because a lot of people especially in the lower age have no assets or liabilities. One thing that really jumps out at me though, is why do they have so many more black women respondents than black men? Granted, 25 samples is a pretty low value for the other-race females in 2004, and that category's outlier had 30 samples in the next 2007 survey. But that honestly seems like it shouldn't produce the anomalies observed. Well, now I know why people are so fascinated by the confidence interval of the mean. I will take another look at https://repl.it/@EllenCT/FedSCFnetworths when I get off of my absurdly scheduled teleconferences, and with luck sleep some. EllenCT (talk) 14:26, 14 April 2020 (UTC)
 * Even if 0 is the most likely outcome, which requires that there be just as many individuals with a negative net worth as with a positive net worth in the population segment, the probability of finding three median values that are exactly zero after rounding to a whole number is really small even for such low sample sizes, definitely less than 1%. Perhaps some respondents just answer "0" because they have no method for estimating the value of their assets. --Lambiam 20:04, 14 April 2020 (UTC)
 * Astonishingly there is no way for respondents who hold only cash to include it in their net worth. Therefore these counts as a subset of the previous table must rent without owning a vehicle or a bank account:

Count of singles with zero net worth age 20 to 35 in the Fed SCF RACE Black      Hispanic       Other       White SEX Female Male   Female Male Female Male Female Male YEAR 1989    50   20       15   10      0    0      5   10 1992     65    5       35    5      4    0     21   15 1995     55   10       15    5      0    0     25   13 1998     55   10       15   16      5    0     30   21 2001     46    5       55    5     10    0     45    5 2004     35   10       10    0      1    0     28    0 2007     42    9       20    0      0    0      9    0 2010     66    5       25   20      5    0     20   10 2013     45    5        1   10      0    0      5   15 2016      0    0        5   15      0   10      2    0


 * I figured out how to find plenty of negative net worth singles; send them to college:

Median net worth of singles with college degree age 20 to 39 in 2016 dollars RACE Black        Hispanic        Other          White SEX Female   Male Female    Male  Female    Male Female   Male YEAR 1989 -6717    N/A    N/A    1866    8023   27960  20524  79038 1992 11153  55243  -4562  153116    6541  228668  17827  38841 1995  42517   2349  10743  103756   11291  295897  51365  58443 1998  34349  15787 -10992  -22279   39247   34858  24566  41460 2001  88336  10209  27666   36310   37800   45997  23032  21501 2004   3179  20984  -6804  106448  175379  243127  23274  54687 2007   5941   5385  23050  -23044  184124   26866  12333  67049 2010  -5416   6135   -884    5902   40455   16580   5527  40378 2013   6960 -17833  17034   24128   30933   39451  14734  21024 2016 -10545 -12430   6791   46395   50510   42030   3750  30000


 * Well if I can get in touch with Mariko Chang and figure out how to get her numbers I'll follow up. EllenCT (talk) 02:10, 15 April 2020 (UTC)

The reason so many of the counts are divisible by five is that the SCF makes extremely heavy use of imputed data, which invalidates the use of medians. Most of these tables have so few observations as to be completely useless. For such a simple survey with so few actual respondents, they sure do take a long time to produce it. EllenCT (talk) 17:44, 15 April 2020 (UTC)