From an article in Bloomberg:
In 2007, half of families had a net worth of $139,700 or more and half fell below this level. By 2016, the midpoint dropped to $97,300 — a decline of $42,600. Families ranked in the top tenth of net worth have enjoyed a sizable gain since 2007: a $132,100 rise in net worth to reach almost $1.2 million.
As is common with reporting on these sorts of numbers, the writer makes it sound as if the families in each percentile grouping remained constant. In fact, the data do not tell us that families ranked in the top tenth of net worth in 2007 gained $132,100 on average. What they tell us is that the threshold for a family to be ranked in the top tenth rose by that amount. At the 50th percentile, the data tell us that the threshold for a family to be at that percentile dropped by $42,600. We cannot use these figures to say what happened to families that were at the 50th percentile in 2007.
I am not contesting the inference that the distribution of wealth became more concentrated. I am just asking for the wording to describe the precise meaning of the data rather than imposing a misinterpretation.
For more on the wealth survey, see Timothy Taylor.
Because it is hard to describe means that I hope you point out the people doing it well rather than just nailing the errors.
Wealth and income distribution are important to do well, not just to shoot down.
I think wealth data should be taken with grain of salt, since they rely on surveys with each person valuing their own property ( I know my house is worth more than my neighbors, and I am sure they know theirs is worth more than mine) and including or excluding assets (how much is a defined benefit pension worth, or tenure).
I think it is better to focus on income, where their is more reliable data from tax returns (everyone has to file under penalty of perjury, set method to calculate income, and a disincentive to inflate income). If one looks at income, the different income groups have had about the same share of income since 1999 or so (while the income share of top groups increased in both the 80s and 90s).
“I am just asking for the wording to describe the precise meaning of the data rather than imposing a misinterpretation.”
You’re asking for rather a lot, I’m afraid. While not all misinterpretations due to imprecision are deliberate, accuracy isn’t much of a priority in highly politicized matters.
The other potential problem is that the families included in the entire set is changing. The wealth of immigrants arriving after 2007 would be expected to impact the analysis. Disaggregation of immigrants and native-born might provide a clearer picture.
This happens All. The. Time.
The way I usually state it is, you can’t use cross-sectional data to make longitudinal claims.
It’s such a subtle and pervasive error of language, it can be hard to point out even to folks who should know better.
Net worth does not include periodic income streams such as pensions or Social Security. Perhaps a future article will attempt to account for these; watch this space, if we model income streams as net worth in the future we’ll link out.