In this post we’ll take a deeper look at the relationship between children and housing, and then do a simple model of Burnet Rd demographics, assuming current housing trends, to predict where our community’s demographics end up in a generation. We’ll end with some intriguing news from SoCo.
Before I start, a quick disclaimer. I am not a statistician. I am offering data with the hope it will be further refined and validated by City staff. I’ve presented these data and no one has called out any obvious errors, but nor to my knowledge has there been a formal review.
You might recall from a previous post that in Vancouver, communities with lots of multi-bedroom housing also have lots of kids. I’m using a statistical measure called R-squared (R^2). Broadly speaking, R^2 measures how close real world data (dots on a scattergraph) “fit” to a line generated by your mathematical model. An R-squared value of 1 would mean that your equation perfectly describes the real world data. (Statisticians dream about this. It never happens.) In most cases, an R^2 result of .8 or .9 means if you know one thing, you can predict the other thing.
In northern cities, the R^2 fit between kids and multi-bedroom housing is strong. In Vancouver, it was 0.81 for all neighborhoods, and 0.9 for middle and upper income neighborhoods. In Boston, R^2 = .8. Portland R^2 = .87 (fit against a log curve). Seattle R^2 = .9.
We don’t see these correlations in southern cities like Raleigh, Dallas, or Austin. When I run this for all census tracts in Travis County, I get R^2 = .35. That doesn’t mean a lot.
What’s Going On in Austin, Texas?
I’m not entirely sure why multi-bedroom housing in a place like Vancouver so closely predicts children, but doesn’t in Austin. Here are some possibilities.
If you take out Travis County’s lower income communities and just run a correlation on middle class neighborhoods, you get a more respectable R^2 of 0.6. That’s enough to visually distinguish the arc of yellow dots (middle class communities) in the graph below.
One conclusion you can’t cleanly draw from these data is a correlation between a given housing type and poverty. Lower income communities are found both in areas with lots of apartments (like Rundberg), and also in suburban communities (like the area north of Austin Bergstrom International Airport). Desperate people will cram into any kind of housing – efficiencies, ranch houses, expropriated mansions.
What is striking about Rundberg is how quickly and sharply the area declined. The housing there is not that old – 1970s-80s. The main driver here was probably economic, but the high concentration of big, aging apartments may trap poverty. One hint – median incomes on the northwest corner of Lamar-Rundberg are $10,000 higher than on the southeast corner. The southeast corner has a much higher concentration of big apartment blocks with single occupancy units.
What’s going to happen on Burnet Rd?
The future is here! Burnet Rd already has a neighborhood that reflects the kind of demographics we could have for the entire area a generation from now. It’s “Baja Brentwood.” In the housing/children graph, southern Brentwood’s housing is close to 50% efficiencies and one-bedrooms. And its weight of children in the population has fallen to just 11%.
Gentrification is in full swing. The housing mix continues to shift towards more single occupancy units. Families are locked out of half the housing market, and are competing with everybody else for the other half. As housing rents skyrocket in Brentwood and throughout the urban core, it becomes harder for families to remain. Less safe outdoor conditions from chronic on-street parking, and the relative decline of family-oriented area services, accelerate the outflux.
This last graphic attempts to model, in broad brush strokes, what kind of housing mix we’ll have on the middle part of Burnet Rd (from 2222 to Anderson) in 25 years. Assuming at trend construction of new apartments on land already zoned for them, plus some ‘transition zone’ housing along the edges of the corridor as a result of CodeNext, the area’s housing mix winds up as majority single occupancy units – not too different from what we have in Clarksville near Downtown (54% efficiencies and one-bedrooms; children at 10% of the population).
This happens despite the transition zones. The demographic weight of the big apartments just overwhelms everything else.
Let’s end on a cautiously positive note. One of the outliers on the housing-to-children graph above is Bouldin Creek, west of South Congress. The 2014 American Community Survey (ACS) data show a surprisingly high 22% kids in the population, despite a continued decline in the weight of multi-bedroom housing (61% multi-bedroom units.)
A closer look reveals a little Bouldin baby boom.
For years, the story in Austin and other cities has been predictable – those young families will pack up and leave for the suburbs. But if Bouldin represents a millennials-driven shift in market demand, it could become easier to make the case to RECA for housing that doesn’t exclude families.
Burnet Corridor Plan >