Whenever you see poverty reports like this, especially ones that seem to have a surprising conclusion, you should ask yourself whether something strange is going on with the poverty metric being used and how it compares to other ways of measuring the poverty rate. I hate debating poverty measurements because the whole thing is so arbitrary and the underlying data driving the analysis in the US is so patchy, but when something like this comes out, I have no choice.
In the below graph, I use the same dataset the study authors use, and data from the
Luxembourg Income Studyand the
OECD to plot four different child poverty rates over the last few decades.
The solid orange line is the one the study authors decided to use, with the year 1993 the one they really hone in as the baseline for most of their statistics. To produce this orange line, you first determine what the poverty thresholds were under the Supplemental Poverty Measure (SPM) in 2012. Then, after adjusting for inflation, you use those 2012 poverty thresholds to add up the number of children in poverty for 2012 and every other year. According to this measure, child poverty fell from 27.9 percent in 1993 to 11.4 percent in 2019.
Globally speaking, this is an unusual way to measure poverty. The more typical way of measuring poverty is to determine the median income of the country for a given year (adjusting for family size) and then add up the number of kids with a family income that is below half of the median. In the graph above, the red (LIS Relative), black (OECD relative), and blue (SPM relative) dashed lines use this method. The three lines differ somewhat due to slightly different ways of defining and imputing disposable income and adjusting income for family size, but they largely tell the same story: the child poverty rate declined from around 25 percent in 1993 to around 21 percent in 2019.