Report Date: 05/10/2025

Executive Summary

This project explores the impact of tax credit programs in three of the U.S. Census Divisions: the Pacific Division (Alaska, California, Hawaii, Oregon, and Washington States; total population, 48,985,292); the West North Central Division (Iowa, Kansas, Minnesota, Missouri, Nebraska, North Dakota, and South Dakota states; total population, 21 625 690); and the Mountain Division (Arizona, Colorado, Idaho, New Mexico, Montana, Utah, Nevada, and Wyoming states; total population, 24,534,951).

It specifically looks at the impact of the New Markets Tax Credit (NMTC) and Low Income Housing Tax Credit (LIHTC) programs on geographies’ social vulnerability, structural disadvantage and financial precarity at the census tract level.

Both of these programs are national policies that utilize tax credits to encourage investors to initiate building projects for community facilities and housing within at-risk census tracts. The New Markets Tax Credit (NMTC) program seeks to stimulate private investment in low-income communities; by providing tax credits to investors who fund qualified Community Development Entities (CDEs). CDEs provide financing in favorable terms to projects in NMTC-qualified tracts. The Low-Income Housing Tax Credit (LIHTC) program is jointly administered by the IRS and HUD, and supports the development of affordable rental housing in qualified census tracts, by offering tax credits to developers.

This report utilizes data from the U.S. Census Bureau following methodologies from the U.S. Centers for Disease Control and Prevention (CDC) to asses the Social Vulnerability Index across U.S census tracts. It also analyzes measures of economic inequalities related to Median Income, Median Home Values, and House Price Index utilizing data from the US Census Bureau and Federal Housing Finance Agency.

We will examine patterns of participation in the New Markets Tax Credit (NMTC) and Low Income Housing Tax Credit (NMTC) programs, and will evaluate whether these federal polices have had a measurable impact on reducing social vulnerability and improving economic outcomes from 2010 to 2020. It is our hypothesis that census tracts that receive investment from the NMTC and LIHTC programs will experience a decrease in social vulnerability and increase in median incomes and home values that is greater than the general trends in the division.

Methods

To examine the relationship between vulnerability and federal investment, we created bivariate choropleth maps that visualized both the overall vulnerability burden and the dollar amounts awarded through NMTC and LIHTC programs. These maps allowed us to highlight areas where high need aligned with high investment, as well as places where funding may be lacking despite significant vulnerability. The spatial tools were essential for identifying geographic disparities and suggesting potential gaps in resource allocation.

To better understand relationships among variables, we conducted correlation analyses. These tests helped quantify the associations between different measures of vulnerability, such as the relationship between population size, total flag count, and median income. We applied k-means clustering to categorize counties based on their total vulnerability burden (flag count) and the dollar amount of NMTC and LIHTC awards received. This method allowed us to detect patterns across the region by grouping counties with similar combinations of need and investment, and draw attention to mismatches between social need and resource allocation.

Finally, to examine how investment or eligibility status may be associated with changes over time, we implemented a difference-in-differences (diff-in-diff) regression analysis. This quasi-experimental method compared changes in outcomes—such as economic or housing indicators—between counties that qualified for federal investment (e.g., LIHTC or NMTC) and those that did not, before and after a given policy window. By controlling for baseline differences and time trends, this approach provided a more rigorous estimate of program impacts than simple before-and-after comparisons.

Results

In all three divisions, the social vulnerability indicators follow in general the same trends as those in the United States. Compared to 2010, 2020 saw improvements in most areas. There was a reduction of the percentages in poverty, unemployed, and housing cost- burdened. The percentage of those with a high school diploma and with health insurance increased. The population in general appear to be aging, with a larger share of the 65+ population. Together with an increase in the percentage of minorities, there is also an increase in the percentage with English proficiency. There was no notable change in housing type indicators, such as the distribution between multi-unit, mobile, and other housing, as well as the percentages living in crowded and group living quarters.

The association between SVI and NMTC dollar awards is moderate to strong in the Pacific and Mountain Divisions (0.52 to 0.62), while in the West North Central Division, these associations are very strong (above 0.8). The correlation between SVI and LIHTC dollar awards is also strong in all three divisions.\

Most of the models in all the divisions didn’t return statistically significant results. In the Pacific Division the only diff-in-diff model statistically significant fining was that NMTC projects improved median income. Neither program was associated with measurable improvements at the census tract level between 2010 and 2020, suggesting limited detectable impact on social vulnerability or economic development during this period.

Recommendations

This study examined the effects of NMTC and LIHTC programs at the census tract level. While tract-level analysis offers fine-grained geographic detail, it can introduce noise; particularly in sparsely populated areas where a single tract may represent an entire county. In the divisions in this study outliers disproportionately influence the results and obscure broader patterns. Utilizing methods complementary to K-means clustering, that are more robust to the influence of outliers may offer benefits. Aggregating data to the county level could smooth these effects and better reflect how tax credit programs are actually implemented, often across regions rather than isolated neighborhoods. Though some neighborhood-level variation would be lost, county-level analysis may yield more stable and policy-relevant insights.

Second, our models treated program participation as a binary variable, whether a tract received NMTC or LIHTC funding or not. This approach does not reflect the wide range of investment levels across communities. Some areas received modest support while others benefited from major allocations. Incorporating actual funding amounts would allow us to test whether larger investments are associated with stronger outcomes. This would offer more meaningful guidance for policymakers, though more sophisticated methods would be needed to account for pre-existing advantages in better-funded areas.

Third, in the case of NMTC, funding supports a wide variety of project types, from health centers to charter schools, grocery stores, and industrial redevelopment. Many of these impacts may not register through the limited set of indicators used in this study, such as median income or housing values. Future research could explore alternate outcome measures tailored to the kinds of projects funded to better capture the full scope of NMTC’s impact.

Forth, to be able to meaningfully lift tracts out of decline, both LIHTC and NMTC efforts probably need to complement, align, and scale to proper levels. The LIHTC program goal of supporting affordable housing in high-vulnerability neighborhoods to promote revitalization competes with enabling low-income families to access affordable housing in high-opportunity areas with better schools, jobs, and safety. At the same time the New Markets Tax Credit (NMTC), which is designed to spur economic development in distressed areas, tend to fund projects across a wide range of sectors (e.g., health clinics, retail centers, charter schools) without a coordinated strategy and investment intensity. Without intentional coordination across housing, economic development, education, and infrastructure, these investments risk spot-treating symptoms without addressing the structural conditions that keep neighborhoods disadvantaged.

References

R Version

Analyses were conducted using the R Statistical language (version 4.4.2; R Core Team, 2024) on Windows 10 x64 (build 19045). For the Mountain Division, analyses were conducted using the R Statistical language (version 4.4.3; R Core Team, 2025) on macOS Sequoia 15.3.2

R Packages

Data

Readings