PROJECTOpportunity Zones: Maximizing Return on Public Investment

Opportunity Zones: Maximizing Return on Public Investment

Brett Theodos, Carl Hedman, Brady Meixell, Eric Hangen

Zones investment score dataset (.xlsx download)

Background

The Tax Cuts and Jobs Act included a new federal incentive—Opportunity Zones—to spur investment in undercapitalized communities. Local areas (defined by census tracts) are eligible for selection as Opportunity Zones if they are Low Income Communities (LICs) under the high poverty or low median income definitions established for the New Markets Tax Credit program. Also eligible for selection are census tracts contiguous to LICs if median family income does not exceed 125 percent of the qualifying tract. Roughly 56 percent of tracts in the US are eligible for selection as Opportunity Zones.

Governors of the 50 states and 5 territories, and the mayor of the District of Columbia (“governors”) are charged with selecting 25 percent of the eligible tracts (or at least 25 tracts for states and territories with fewer than 100 eligible tracts) as Opportunity Zones. Non-LICs can represent no more than 5 percent of tracts selected. Governors have until March 21 to make selections and can take an additional 30 days if they request an extension. Once selected, Opportunity Zones keep the designation for 10 years. There is no provision in the statute to change which communities are classified as Opportunity Zones.

Apart from the exclusion of a few “sin” businesses, the activities and projects Opportunity Funds can finance are broad. Funds can finance commercial and industrial real estate, housing, infrastructure, and existing or start-up businesses. For real estate projects to qualify, the investment has to result in properties being “substantially improved.”

Given the breadth of eligible investment types, Opportunity Zones must be carefully selected to ensure the return on the public investment is maximized and will lead to gains for low- and moderate-income residents. To guide selection, we prepared a dataset, for all eligible tracts, ranking them in terms of the investment flows they are already receiving and the social and economic change they have experienced.

Investment Score

We developed a score of investment flows to tracts based on four components: commercial lending, multifamily lending, single-family lending, and small business lending.

Opportunity Zones will spur equity investments into tracts, but information about existing equity flows is not available at small areas of geography across the dimensions of interest. As such, we present debt flows as one means towards understanding local capital access, but note the important distinction. Further, we have not incorporated other types of capital flows that matter to tracts—notably local, state, federal, and philanthropic funding. While the investment score provides information about capital access for LICs and contiguous tracts, local knowledge will help contextualize, clarify, and even correct the understanding conveyed via this score.

Commercial lending. To develop this measurement, we used 2011–15 CoreLogic data of loans to commercial, industrial, agricultural, and exempt properties geocoded to 2010 census tracts, excluding single loans that totaled $100 million or more. We summed the total investment over the five-year period at the census tract level and created an annual average. We divided that average investment amount by the number of workers employed in the tract to create an investment-per-employee ratio. We obtained employment data from the Longitudinal Employer-Household Dynamics Origin-Destination Employment Statistics Workplace Area Characteristics at the census block level and then aggregated them to the census tract level. We calculated this measure only for census tracts with 200 or more employees.

Multifamily lending. We developed a measure of multifamily lending using 2011–15 CoreLogic data of loans to multifamily properties (five or more units) coded to 2010 census tracts. We excluded single loans that totaled $100 million or more. We summed the total investment over the five-year period at the census tract level and created an annual average. We divided that average investment amount by the number of multifamily units in the tract to create an investment-per-multifamily-unit ratio. We obtained the tract multifamily unit data from the 2011–15 American Community Survey. This measure was calculated only for census tracts with 200 or more multifamily units.

Single-family lending. We compiled a single-family lending measure for 2011 through 2015 using tract-level Home Mortgage Disclosure Act records. We considered only home purchase loans. We took the average amount and total number of home purchase loans per tract to arrive at estimated average total loans over the five-year period. We divided the average annual amount by the total number of single-family units in each tract to arrive at an average annual level of Home Mortgage Disclosure Act single-family lending per single-family unit. This measure was calculated only for census tracts with 200 or more units of single-family housing as obtained from the 2011–15 American Community Survey.

Small business lending. We compiled a small business measure at the tract level for the years 2011 through 2015. We obtained lender-level Community Reinvestment Act loan amounts for small businesses from the annual aggregate Community Reinvestment Act data files for 2011 through 2015, available through the Federal Financial Institutions Examination Council. We excluded likely credit card loans by dropping records if the average size of their loans made under $100,000 was less than $10,000. Collapsing five years of data by tract, we arrived at a total sum, which we used to obtain the average annual amount of Community Reinvestment Act lending. We then divided this amount by the number of small business employees in each tract. We considered any private-sector employee working at a firm with up to 19 employees as an employee of a small business. We obtained these data from the Longitudinal Employer-Household Dynamics Origin-Destination Employment Statistics Workplace Area Characteristics at the census block level and then aggregated them to the census tract.

We created z-scores for each component measure (commercial, multifamily, single-family, and small business) unless the tract did not meet the cutoff criteria. We averaged the z-scores to create a composite investment score. If a tract did not meet the eligibility threshold for a given category, it was averaged based on the other categories. Then, looking at only LICs and contiguous tracts, we ranked tract z-scores relative to other LIC and contiguous tract scores within the same state or territory. We then created a decile ranking of the composite z-scores for all eligible LIC and continuous tracts, meaning that each LICs and contiguous tract has a ranking from 1 (low) to 10 (high).

Socioeconomic Change Flag

Eligible tracts that have gentrified may need federal investment support less than other tracts that have not. Gentrification is challenging to measure at a national-level, and ideally would incorporate local data. (For example, see this compilation of studies.)

To help inform the decision of which tracts will maximize the return on public investment, we created a flag for tracts that have experienced high levels of socioeconomic change. But local knowledge will be needed to validate, verify, and modify the information presented.

Tracts received a socioeconomic change flag if they were more than 1 standard deviation above the mean of all national census tracts on the composite socioeconomic change index we developed. This index was composed of four indicators measuring the change in their respective values between 2000 and 2016. We obtained all 2012–16 data from the 2012–16 American Community Survey, and we obtained all 2000 data from the 2000 Decennial Census.

We included the following four measures in these calculations:

  • Percentage point change in the share of residents with a bachelor’s degree or higher
  • Dollar change in median family income
  • Percentage point change in the share of non-Hispanic white residents (which for example, can help to explain difference in assets, not just incomes)
  • Change in average housing burden
    • We created the housing burden measure by calculating z-scores for two housing measures and averaging them: (1) change in the tract’s median home value divided by change in the metropolitan statistical area’s median household income, and (2) change in the tract’s median gross rent divided by change in the metropolitan statistical area’s median household income. For rural areas outside metropolitan or micropolitan statistical areas, we used the county median household income instead. Any tracts with fewer than 100 units of rental-occupied housing were scored by the home value measure alone, and any tracts with fewer than 100 units of owner-occupied housing were scored by the rent measure alone.

To receive a score in the first three demographic indicators (educational attainment, median family income, and race), tracts had to have at least 100 residents. We created z-scores for each of these three factors and then averaged them with the housing burden z-score. We calculated this score for all US tracts, not just LICs and contiguous tracts. Any tract 1 standard deviation or more above the mean is flagged as a “1”, having experienced sizable socioeconomic change across these dimensions.