Why do racially-segregated neighborhoods exist and persist? The standard explanation is that whites don't like to live near, say, African-Americans; when African-Americans start moving in, whites start moving out. Real estate agents steer whites away from African-American neighborhoods and vice versa. And let's not forget the legacy of de jure segregation, which herded African Americans into a few neighborhoods; residential patterns typically change slowly. The standard explanation relies on more or less overt, conscious racism.
There's clearly some of that going on. Most cities -- Austin included -- had either de facto or de jure rules that relegated African Americans to just a few, exclusively African-American neighborhoods. Although I consider Austin a fairly racially tolerant place today, it is still strikingly segregated:
(From the City of Austin's demographics map library.)
For all I know, the standard theory explains all segregation in the United States. The "tipping point" theory, however, predicts that neighborhoods will segregate racially even when everyone is fairly racially tolerant (or, if you prefer, only mildly racially intolerant). To put it another way, racially mixed neighborhoods are "unstable"; even when a city's residents have only mild racial preferences, its neighborhoods will quickly segregate once they reach the tipping point.
For example, suppose everyone is comfortable living in a racially-mixed neighborhood as long as at least 30% of the neighborhood's residents share their race. This is a pretty weak racial preference. But when a neighborhood is at a boundary -- 30% of one race and 70% of another race -- then the addition of just a few new households of the majority race will cause the minority to begin looking for other neighborhoods. Even if the minority have a wide range of preferences, small changes can cascade down until a neighborhood is racially segregated.
In this new paper (pdf), Card, Mas and Rothsetein find strong evidence that neighborhoods across the country have these tipping points. (Rorty has a good discussion.) Card, et al. looked at the racial composition of neighborhoods in 1970 or 1980 and then examined the percentage change in white population ten years later. They found that once a neighborhood's percentage of minorities exceeded a certain threshold, the white population dropped dramatically. The tipping points varied among cities. In Chicago, the tipping point was only 7% (which suggests "strong racial preferences," to put it politely).
Here is the plot for Chicago:
Each dot represents a neighborhood. The x-axis shows minorities' share of neighborhood population in 1970. The y-axis shows the percentage change in white population between 1970 and 1980. There is an obvious boundary at 7%. As long as a neighborhood was less than 7% minority in 1970, its white percentage grew between 1970 and 1980. But if the minority population exceeded 7% in 1970, its white population dropped 20% or more.
Some cities are more tolerant than others. Between 1980 and 1990, San Antonio's tipping point was about 24%. (Card, et al. didn't break out the data for Austin.) What I would like to see is a comparison of tipping points to city growth rates. My guess is that cities with high growth rates have high tipping points.