The lowest 20% could be referring to the distribution, instead of specific dollar values. This way, 20% of the people are always in that category regardless how much they make. In statistics, this is called the twentieth percentile, and it’s a handy way of dealing with messy distributions like this.
In a society with great wealth inequality, the 20th percentile will be very low. If there’s more equality, it will be higher.
That's what I'm asking. 20% of $100 is $20. They're both bad incomes to have.
The distribution will also give you a specific dollar value corresponding to that percentile, and then you can use the cost of living statistics to interpret it. If you say that 20% of households earn less than 35 k$ is that 35 big or small? The income distribution alone won't tell you that. In relation to the rest of the population, it's small, but what can you actually buy with that? Interpretation requires digging into the cost of living statistics. Once you combine the two, you'll be able to tell that the lowest 20% are struggling.
Percentile based classification is good because it is simple and it won't do the interpreting for you. If you can afford to eat, but you're not eating meat every week, is that really poverty or not? If you can travel, but you're not doing that driving a Ferrari or sitting on a private jet, are you really rich at that point? These are the kinds of questions the percentile approach avoids.
BTW the 2/3 median approach is really weird. The number of people it involves changes every year, but at least it's tied to a median which actually tells you something. Using the highest income as a benchmark makes even less sense, because that number changes radically every year based on what a few billionaires are doing. It won't tell you anything about the rest of the population. Using a wild number like that as a basis of any kind is pretty much useless. Besides, the highest income of the population is really far from the rest, because the distribution has an extremely long tail.
When they say "the lowest 20% of incomes", that's referring to, if you were to plot the incomes of every American on a graph and then drop some dividers down every 20%, the people in the bottom one of those bins. It's not relative to the highest income dollar amount. The wealth inequality in this country is so bad that probably 75-85% of the population is making under 20% of the highest income by pure dollar amount.
You can track things relating to those groups, though. Like if we consider the 20th percentile and below to be the very poorest in society, no matter how rich or poor society happens to be on average or what the wealth distribution is like - we can see how the very poorest fare in terms of education, health outcomes, etc over time.
Or like, looking at the median value - sure the amount of money the median-income worker has will change, but the median-income worker could afford a standard of living much closer to someone at the 90th percentile in 1960 than today's median-income worker can.
Looking at things like this allows you to draw different comparisons.
It seems to me like you are coming to grips with the fact that statistics are nice but one mustn't trust those one hasn't forged personally.
I imagine there may be political or even practical reasons why they draw the line there. A lot of times lines remain fixed because they made more sense when they were drawn ages ago and now it's just a pain in the butt to change them. Or to keep numbers comparable with those from previous years.
Or maybe there are laws or regulations that would be triggered if you move the goal post. If you start reporting the number of poor people or their income in a different way, that could affect government spending. If that’s the case, there could be a lot of friction in trying to update the calculations.
For one thing, it maximizes the amount of information the census contains: if you’re splitting a group into bins, the splitting with the lowest entropy is the one with equal-sized bins. If the bin sizes fluctuate with time (as a proportion of the overall population), the amount of information the census captures will vary as well. If your definition of lower-income includes 100% (or 0%) of the population, then assigning someone to a category tells you nothing about them you didn't already know.
A loosely moderated place to ask open-ended questions
If your post meets the following criteria, it's welcome here!
Looking for support?
Looking for a community?
~Icon~ ~by~ ~@Double_A@discuss.tchncs.de~