NOTE: Someone kindly pointed out an error in my reading of the demographic data described below. The misreading led me to make a slight overstatement in the original version of this post. I have corrected the content below accordingly.

In 2016, the Florida Housing Finance Corporation (Florida Housing) began to implement its Area of Opportunity (AofO) strategy. The developments awarded funding under the 2016 geographic RFAs are currently in underwriting or under construction. Now we can begin to see whether the AofO strategy has made a difference. This is the first of three posts on the topic.

Below, I describe the process used to collect and compare information related to tax credit financed developments. If you do not share my enthusiasm for the technical minutia of housing policy, you may want to skip past that section. There’s also a brief discussion of how I chose to measure segregation. The results follow a description of the baseline data used for comparison.

In general, the AofO policy has not yet made much of a difference. The 2016 allocation yielded a somewhat more limited range of housing options for low-income families. Looking at Florida Housing designated Areas of Opportunity specifically, the early results of the policy change are encouraging. The properties funded in these tracts create affordable housing options in diverse neighborhoods with opportunities for upward mobility.


Florida Housing has made a great leap forward in its use of GIS over the past year. Mercifully, the Corporation now collects Development Location Points (DLPs) in decimal form on the application for tax credits. This allows Florida Housing to list the latitude and longitude of proposed sites in Excel format on the applications submitted report. This is a fantastic improvement both operationally, and in terms of public transparency. Unfortunately, the DLPs for the 2016 applications are only available in Degree-Minute-Second (DMS) format on PDFs of the surveyor certification form.

We need to know where the new developments will be located in order to evaluate the neighborhood level indicators of community well being for the winning applications. Accordingly, I copied the DMS coordinates for each DLP from the PDFs into an Excel workbook. Then I converted the values to decimal form.  Adding these X,Y data to an ArcGIS map document, I intersected the points with 2010 Census Tracts using a 2016 TIGER shapefile. I then exported the records back to Excel. In this way, I compiled location information for the preliminary awards made under the three geographic RFAs. Florida Housing limited its AofO strategy to metropolitan counties. For this reason, I excluded the small county award made under RFA 2016-110, as it is located in a non-metropolitan county.

We are really after neighborhood-level data. “Neighborhood” is of course a tricky term. Census tracts are our best approximation. I downloaded data tables from American FactFinder for the 2016 5-year average of the American Community Survey (ACS). The data I used include poverty status (table S1701), median income (table S1903), employment (S2301), Hispanic identity by race (B03002), tenure (B25003), and SNAP utilization (S2201). I added each of these tables to the workbook. Additionally, I added the county level data for Hispanic identity by race, as well as poverty, for the counties which produced winning tax credit applications.

Finally, I associated the ACS data with each DLP. This was accomplished with a simple index/match function in Excel.[1] In this way, the relevant data from the ASC tables are associated with each DLP.

Measuring Segregation & Housing Choice

In my opinion, HUD defined and Florida Housing designated RECAPs are too generic to be of much use. Both designations are based on a minority concentration of 50.0% or more, and an incidence of poverty greater than or equal to 40.0%. This standard is based on academic research and federal policy that is national in scope. The consistent measure is a helpful starting point, but it is not sufficiently context-sensitive to tell us whether a state housing finance agency’s allocation criteria effectively further fair housing.

Florida is characterized by heterogeneous social geography. In Charlotte County, a tract with a minority population of 50.0% would indicate extreme racial/ethnic concentration. Such concentration could only be explained by a history of housing discrimination. However, in Miami-Dade County, and the other minority-majority counties in Florida, a minority population of only 50.0% would indicate a high concentration of non-Hispanic Whites.

Similarly, although a poverty rate of 40.0% is high anywhere in the state, the incidence of poverty varies considerably from county to county. For example, a census tract with a 20.0% incidence of poverty would be fairly typical in Miami-Dade County. However, the same value equates to 165.0% of the county-wide rate of poverty in Seminole County.

A high minority concentration is not in itself problematic, of course. It is the nexus of unusually high rates of poverty, and racial/ethnic isolation from the majority population, which perpetuates the pattern of residential segregation established during de jure segregation. I also do not mean to say that the existence of high incidences of poverty in non-Hispanic White majority areas is not a public concern. The point is that if the AofO strategy is at least in part intended to address the persistence of residential segregation, we must have some context-sensitive means of measuring it. Consequently, I identified tracts which are characterized by BOTH an incidence of poverty greater than 1.5x the county-wide incidence of poverty, AND a non-Hispanic White population less than half that of the county as a whole.

Setting a Baseline for Comparison

To establish a baseline, I identified properties awarded tax credits over several years prior to 2016. The Assisted Housing Inventory (maintained by the Shimberg Center for Housing Studies) lists 116 properties with affordability start dates between 2010 and 2015 located in the same counties as the 2016 awards. Fortunately, Shimberg reports the decimal form latitude and longitude coordinates for each site. I filtered out sites which were not listed as “ready for occupancy.”

Larger data sets are more stable and therefore more reliable. Comparing more than 100 properties funded over several years to a handful of properties funded in a single year may lead to incorrect conclusions. For this reason, I also evaluated the properties awarded 9% tax credits under the 2013 geographic RFAs. This allows a year-to-year comparison pre- and post- policy change.

Census Tracts – Preliminary Awards – 2016 Geographic 9% Tax Credit RFAs

Average of All Tracts Range
Labor Force Participation Rate 58.8% 43.1% – 77.1%
Tract Median Income $39,467 $9,693 – $64,444
Homeownership Rate 49.5% 2.5% – 84.1%
Incidence of Poverty 27.2% 4.4% – 75.9%
SNAP Utilization Rate 29.5% 6.8% – 85.3%
Total Minority

(Other than Non-Hispanic White)

60.6% 16.7% – 98.9%

A high percentage of the properties are located in census tracts with overlapping characteristics of poverty and racial/ethnic isolation. 5 of the 19 properties (26.3%) funded in 2016 are located in tracts characterized by BOTH an incidence of poverty greater than 1.5x the county-wide incidence of poverty, AND a non-Hispanic White population less than half that of the county as a whole.

Census Tracts – In Service – 2010-2015 9% Tax Credit Properties

Average of All Tracts Range
Labor Force Participation Rate 58.9% 26.1% – 77.1%
Tract Median Income $35,719 $13,660 – $110,521
Homeownership Rate 40.0% 3.7% – 92.2%
Incidence of Poverty 27.5% 2.5% – 56.0%
SNAP Utilization Rate 29.5% 2.8% – 64.4%
Total Minority

(Other than Non-Hispanic White)

67.4% 4.7% – 99.7%

35 of the 117 properties (29.9%) are located in tracts characterized by BOTH an incidence of poverty greater than 1.5x the county-wide incidence of poverty, AND a non-Hispanic White population less than half that of the county as a whole.

Census Tracts – 2013 Geographic 9% Tax Credit RFAs

Average of All Tracts Range
Labor Force Participation Rate 58.5% 35.5% – 79.5%
Tract Median Income $32,733 $16,219 – $60,882
Homeownership Rate 35.3% 2.9% – 82.3%
Incidence of Poverty 31.3% 15.1% – 61.2%
SNAP Utilization Rate 26.5% 5.5% – 49.2%
Total Minority

(Other than Non-Hispanic White)

62.6% 12.2% – 97.6%

5 of the 17 properties (29.4%) awarded funding under the 2013 geographic RFAs are located in tracts characterized by BOTH an incidence of poverty greater than 1.5x the county-wide incidence of poverty, AND a non-Hispanic White population less than half that of the county as a whole.

In General, there is No Meaningful Difference

In general, there are few meaningful differences between the 2016 allocation and Florida Housing’s competitive tax credit allocations over previous years. There are some positive results. The average tract median income has increased. Also, the average total minority concentration has decreased. Additionally, the 2016 sites are in tracts with a slightly lower average incidence of poverty, and higher average rates of home ownership. However, the ranges indicate a narrowing of neighborhood choice. During the 2010 – 2015 period, at least some properties were funded in tracts with home ownership rates as high as 92.2%, the highest incidence of poverty was 56.0%. In tracts containing sites awarded tax credits in 2016, the highest rate of home ownership is 84.1%; the highest incidence of poverty is 75.9%. Although the range of tract median incomes indicates that properties were funded in middle-income neighborhoods in 2016, at least a few were funded in more affluent census tracts during the earlier period.

But…Geographic AofOs Yield Measurably Better Spatial Outcomes

Of the 19 properties funded through the geographic RFAs in 2016, 4 are located in Florida Housing designated Areas of Opportunity. The proposed developments are located in 4 unique census tracts. The average rate of labor force participation is 64.4%; the average tract median income is $49,142; the average incidence of poverty is 17.4%; and, the average rate of home ownership is 65.3%. Interestingly, the average minority population is 70.8%; excluding the 2 Miami-Dade AofO tracts, it is 65.2%.  Nonetheless, non-Hispanic Whites are present in these communities. In the 2 Miami-Dade tracts, non-Hispanic Whites are present at more than double the county-wide rate in one tract, and just under the county-wide rate in the other. In the other AofO tracts, non-Hispanic Whites account for about one-third of the population.

Funded properties in geographic Areas of Opportunity offer affordable housing options in neighborhoods where both diversity, and upward mobility (in the form of work and home ownership), are normative. Moreover, although these are not the most exclusive neighborhoods, they are characterized by relatively moderate incidences of poverty.

The data are available here. Feel free to take a look and let me know in the comments section below if you see any potential improvements.

[1] =index(ACS_variable_array,match(tractnumber,acs_table_tract_array,exact_match))