Separate, but how unequal? Racial neighborhood inequality in America since 1980 Presentation at segregation conference University of Chicago February 2, 2013 Glenn Firebaugh Penn State University firstname.lastname@example.org Joint work with Chad Farrell, University of Alaska
Origins of project Much interest in racial disparities in local neighborhood contexts – i.e. racial neighborhood inequality – but there is a lot we don’t know (for reasons I explain subsequently) What do we know underscores importance of what we don’t know
Things we know (1) We know that U.S. neighborhoods are very uneven with respect to poverty, median income, etc. (neighborhood income segregation) – and disparities are increasing (Reardon & Bischoff, 2011) (2) Neighborhoods are also very uneven with respect to racial composition (residential segregation) (3) We know that (1) and (2) are linked – groups unevenly spread across space, and minorities tend to reside in poorer neighborhoods (racial neighborhood inequality) • There is major literature in Sociology on concentrated poverty, neighborhood disadvantage, etc. Although object of this literatureis inequality, studies don’t use conventional inequality measures.
Things we don’t know What isn’t apparent from this literature – just how large is racial neighborhood inequality (RNI) compared to other types of inequality? Is RNI growing or declining? For which groups? Presumably, because blacks and Hispanics at every income level live in poorer neighborhoods than do whites with comparable incomes (Logan 2011), racial inequality at neighborhood level exceeds racial inequality at household level – how much larger is it? How large is RNI relative to racial neighborhood segregation? There is evidence of declining black-white segregation – is this true for black-white neighborhood inequality as well?
Origin of the project Answers aren’t apparent because studies of racial nbhd inequality do not use standard indexes of inequality. • Found only two studies that use inequality indexes - Timberlake (City & Community, 2002); Osypuk et al. (Urban Affairs Review, 2009) – but not standard ones We should be able to use standard inequality indexes for comparing racial neighborhood inequality to, e.g., racial neighborhood segregation, since both are types of inequality • RNI refers to unequal (disproportionate) distribution of groups across neighborhoods of varying economic conditions • Racial neighborhood segregation refers to unequal distribution of groups across neighborhoods
Neighborhood economic conditions where whites, blacks, Hispanics and Asians live Racial neighborhood inequality (RNI) refers to racial disparities in residential economic contexts This talk gives new results on racial neighborhood inequality in the 366 census-defined metro areas (@ 80% of US population) from 1980 to 2010, and speculate on what underlies the trends we see. • For every census tract (neighborhood) – have census data on racial composition, poverty rate, and median income • Analysis still underway – feedback welcome
Neighborhood economic conditions where whites, blacks, Hispanics and Asians live Keep in mind: The poor neighbors of (say) blacks are not necessarily black – blacks could live disproportionately (given their own incomes) among poor whites, Hispanics, and Asians. This point is key to our findings. • It also helps explain why racial inequality is greater at neighborhood level than at household level
Begin with Lorenz curves based on neighborhood poverty rates Order the 59,973 census tracts in the 366 metro areas from high to low on poverty rate • Constant 2010 tract boundaries, from Brown University’s LTDB (longitudinal tract data base). Will compare with results using constant 2000 tracts (GeoLytics). Add two columns beside the column for poverty rate • Cumulative % group 1 • Cumulative % group 2 Graph cum %s, with poorer group as X-axis
Sensitivity of RNI Ginis To 2010 vs. 2000 census boundaries: Ginis are the same using boundaries for 2010 (59,973 tracts) or boundaries for 2000 (53,138 tracts) To US as whole vs. metro-specific Ginis: Lorenz curves above are based on tracts in Detroit compared to tracts in San Diego, Atlanta, etc. But minorities are unevenly spread across regions – so how much do results differ for average RNI for the 366 metro areas that house the 59,973 tracts?
Similar results for average Ginis for 366 metro areas Metro-ignoring Ginis in 1980 vs 2010: .64 vs .37, .57vs .36, and .42 vs .34 for blacks vs. nonblacks, whites vs. nonwhites, and Hispanics vs. non-Hispanics, respectively Average metro Ginis in 1980 vs 2010: .64 vs .41, .53 vs .38, and .40 vs .33 for blacks vs. nonblacks, whites vs. nonwhites, and Hispanics vs. non-Hispanics, respectively • Sum of Ginis for each of metro area, weighted by proportion of focal group residing in that metro – i.e., a population-weighted average Gini
Still separate, but less unequal Decline in racial neighborhood inequality has gone largely unnoticed. • More attention to declines in racial neighborhood segregation – e.g., Glaeser & VigdorThe End of the Segregated Century (Manhattan Inst report, 2012) But if our results are correct, racial neighborhood inequality has declined much faster than racial neighborhood segregation …. Why?
Why is RNI declining? Overarching hypothesis: Neighborhood “sorting” of households in America has become increasingly class-based and decreasingly race-based. That is: Rising neighborhood income segregation (especially) and declining neighborhood racial segregation is behind the decline in RNI. Rest of talk explains.
Why is RNI declining? Disparities in the neighborhood economic environments of whites, blacks, Hispanics and Asians in America derive from three types of residential segregation (Quillian 2012): • racial segregation itself • within-race economic segregation • cross-race economic segregation.
Using Quillian’s example of neighborhood poverty rates for blacks vs. non-blacks: • within-race economic segregation refers to the extent to which poor and non-poor blacks are residentially segregated • cross-race economic segregation refers to the extent to which poor blacks are residentially segregated from non-poor individuals ofother racial groups.
Cross-race economic segregation Cross-race economic segregation contributes to racial neighborhood inequality when minorities reside “disproportionately” in neighborhoods populated by poorer members of other groups (e.g., when the nonblack neighbors of a black family tend to be poorer than the nonblack neighbors of a nonblack family with the same income as the black family). • Recall John Logan’s (2011) finding that blacks and Hispanics at every income level live in poorer neighborhoods than do whites with comparable incomes.
Why is RNI declining? The claim: Income is becoming more important, and race less important, as a residential sorter So the decline in racial neighborhood segregation has two sources: • decliningneighborhood racial segregation and • rising neighborhood income segregation (which results in declining cross-race economic segregation) Let’s unpack this further
Source 1: declining racial residential segregation The decline in black-white neighborhood segregation is well-documented, and has been observed in our results as well. But we don’t think the decline in RNI is due only to declining racial segregation: Consider results for the RNI and segregation Ginis.
Similar pattern for blacks vs. whites (instead of non-blacks) and Hispanics vs. whites
Why is RNI declining? Source 1: Declining racial segregation Source 2: Rising neighborhood income segregation (likely the more important source of declining RNI) • Reardon and Bischoff (2011) find that the percentage of families living in neighborhoods they classified as either “poor” or “affluent” increased from 15% in 1970 to 31% in 2007. • Likewise, Fry and Taylor (2012) found that residential segregation by income increased in 27 of the country’s 30 largest metropolitan areas from 1980 to 2010.
How does rising income segregation reduce RNI? Answer: by constricting cross-race income segregation Think about perfect income segregation (income- homogeneous neighborhoods). In that case: • Racial inequality at neighborhood level = racial inequality at household level • Importantly, no cross-race income segregation (minorities would not reside in poorer neighborhoods than whites with same incomes)
How does rising income segregation reduce RNI? In other words: rising income segregation declining cross-race income segregation declining RNI rising income segregation declining cross-race income segregation because neighborhood income homogeneity constrains cross-race income segregation (the more income-homogenous the neighborhood, the less that white, black, Hispanic, and Asian neighbors can deviate from the neighborhood standard)
How does rising income segregation reduce RNI? declining cross-race income segregation declining RNI Quillian (2012): Cross-race income segregation is important source of concentrated poverty in America –contributing to concentrated poverty net of the effects of racial segregation. Thus we expect reductions in cross-race segregation to reduce RNI independent of the effect of declining racial segregation.
The surprise here: much of change is 2000-2010 Is 2010 data problematic, due to switch to ACS? But note similar (though less dramatic) pattern seen even for 1980-2000 Mismatch of income & poverty data vs. tract data on race/ethnicity for 2000-2010 comparison • Tract data on race are from short-form census data for 1980-2010 • Tract data on income and poverty are based on long-form census data until 2000, then on ACS data
Next steps – testing our explanation - Is income replacing race as the “great residential separator” in American society? • Test rising income segregation declining cross-race income segregation declining RNI using data for 366 metro areas, 1980-2010
There is still lots to be done Back to my starting point – Given (a) the huge literature on residential segregation, (b) the magnitude of racial disparities in neighborhood conditions, (c) concerns about racial inequality in America, and (d) some evidence about the importance of neighborhoods, it’s surprising how little we know about racial neighborhood inequality in America – Is it growing or declining? For which groups? How is it related to racial inequality at household level? To residential racial segregation? To residential income segregation? There’s still plenty of work to be done.
Orienting observations on trends in RNI Order all 59,973 US metro census tracts from highest to lowest poverty rate in 1980 and 2010 • Constant 2010 tract boundaries, from Brown University’s LTDB (longitudinal tract data base). Compared 2000 tracts (GeoLytics). Add parallel columns for cumulative % white, %nonwhite, %black, etc., in the tracts. Here’s what we found at the 50th percentile for whites, nonwhites, blacks, Hispanics, and Asians, 1980 vs. 2010: • Poverty rate in census tract where median white lived, 1980 vs. circa 2010: 7.0% 7.8%; poverty rate where median nonwhite lived fell from 18.0%14.7%, so nonwhite/white ratio to declined from 2.6 to 1.9
Poverty rate where median black lived: 21.9% in 1980 17.4% in 2010; ratio to whites declined from 3.1 to 2.2 (from 3.3 to 2.2 using constant 2000 tract boundaries) • Poverty rate where median Hispanic lived: 15.8% in 1980 16.0% in 2010; ratio to whites declined from 2.3 to 2.1 (from 2.4 to 2.0 using 2000 tract boundaries) • Poverty rate where median Asian lived: 8.3% in 1980 8.0% in 2010; ratio to whites declined from 1.2 to 1.0 (same results using 2000 tract boundaries)
In short, poverty rate today is about the same in neighborhood where average white and Asian live, and is much more similar (than before) where average Black and Hispanic live. So: Convergence for whites and Asians at median; blacks are narrowing the gap with Hispanics; some narrowing of blacks &Hispanics with whites & Asians. Median is only one point on the Lorenz curve – now consider entire Lorenz curve