Stephanie Myers last week had a piece in the Austin Post entitled, “Are Families with Children Being Forced Out of the City?” “Forced” is a loaded term, but families with children undeniably are a declining share of central Austin’s population. Families with children have been declining in share for decades and, as Myers notes, are now declining in absolute numbers in the central city as well.
Why this is happening is up for debate. A fruitful debate, though, requires accurate facts, and I think it’s important to correct the article’s impression that the 78704 zip code somehow is “bucking this trend” and seeing a rise in children. (To orient the out-of-town readers, 78704 is the zip code directly south of downtown Austin.)
The Census data show, clearly and unambiguously, that the population of children in 78704 declined between 2000 and 2010. Back in March 2011, I ran a table showing that nine of the eleven census tracts in ’04 lost children, and the zip code lost nearly 1,400 children overall.
A few months later, I published this chart showing that the population in the zip code declined for every single year of age between birth and seventeen.
The loss of children was broad and deep.
There’s yet another way to measure the change, and that is relative to the Austin average. Divide a census tract’s percentage of population under 18 by the citywide percentage of population under 18. This number (itself a percentage) tells you the concentration of children in the tract relative to the citywide average. This is sometimes referred to as a “location quotient.” (The Census Bureau used location quotients extensively in its recent report on U.S. metro area.)
For example, in 2010, 11.15% of the residents of census tract 13.03 (Zilker) were under 18 while 22.2% of all Austin residents were under 18. Children thus were concentrated in tract 13.03 at just 50.2% of the citywide average concentration.
I calculated the location quotients for children under 18 for all of the the 78704 tracts, using both 2000 and 2010 census figures (including the 2000 citywide averages). Between 2000 and 2010, the concentration of children declined, relative to the city average, in every single census tract.
Children were more underrepresented in 78704 in 2000 than 2010, both in absolute terms and relative to their citywide share.
Note that in 2000, 78704 had only one census tract (13.07 – Galindo) with a share of children above the city average. By 2010, though, Galindo’s share had plummeted to 80% of the city average.
Don’t make the mistake of thinking that 78704’s low share of children is due to an overabundance of VMU-style apartments, or to an overabundance of rental housing in general. Owner-occupied housing in 78704 is simply less likely to be used to raise children than owner-occupied housing in the city as a whole.
Citywide, 33% of owner-occupied homes were occupied by a household containing one or more related children under 18 in 2010. The percentages were far below that for every single ’04 census tract. The closest was Barton Hills (19.01), where approximately 25% of the owner-occupied homes contained a related child. Only approximately 18% of the owner-occupied housing in tract 13.03 (Zilker) contained a related child in 2010. Across the entire zip code, only 20.5% of owner-occupied homes contained a related child. (Comparable data is not available for 2000, unfortunately.)
I haven’t seen any persuasive evidence that the population of children in ’04 is increasing. On the contrary, we've recently heard a lot of complaints that short-term rentals are further “hollowing out” '04, making it less desirable for families with children. (I’d have more sympathy for this position if these neighborhood associations hadn’t worked so hard to make them inhospitable to larger households.)
The article cites school enrollments, but using raw school enrollment figures to track changes in population is tricky because of transfers. For example, of the approximately 380 students enrolled at Barton Hills Elementary, approximately 180 are transfers. Of course, some of these transfers are from neighboring schools. Calculating the population of school-age children requires accounting for both the source school and destination school for all these transfers, as well as homeschooled and private-schooled children. And it requires doing this for multiple years, because we're not concerned with what might be a random fluctuation from one year to the next, but an apparently long-term, secular trend. I haven't seen anyone do this analysis.
I'm also skeptical of ad hoc recounts performed by local residents. To be clear, I'm not questioning the integrity of their counts, but merely their relevance to comparisons of 2010 census counts and 2000 census counts. It's virtually certain that the census canvassers missed people in 2010 in the Zilker neighborhood, for example, but it's virtually certain that they missed people in every neighborhood in 2010, and in every neighborhood in 2000. If you want an ad hoc survey to count as compelling evidence against the decline thesis, you need a good explanation why the census takers would be more likely to miss children in 2010 than 2000. Again, I haven't see one.