Introduction

This project investigates the relationship between federal investment programs and social vulnerability in the South Atlantic Division of the United States. Specifically, it evaluates the impact of the New Markets Tax Credit (NMTC) and Low-Income Housing Tax Credit (LIHTC) programs on vulnerable census tracts between 2010 and 2020. Social vulnerability is assessed using the CDC’s Social Vulnerability Index (SVI), which includes four dimensions: socioeconomic status, household characteristics, racial and ethnic minority status, and housing and transportation access. Data were sourced from the U.S. Census Bureau, the Federal Housing Finance Agency, and were normalized to 2010 census tracts using the NHGIS crosswalk.

The project employs a mix of spatial and statistical methods, including choropleth and bivariate mapping to visualize geographic trends and relationships, correlation analyses to explore variable associations, K-means clustering to identify patterns in funding and vulnerability, and difference-in-differences (diff-in-diff) regression models to measure the effects of NMTC and LIHTC programs over time. Overall, we found no statistically significant impact of either program on social vulnerability outcomes. There was also no consistent correlation between the amount of funding received and the vulnerability of the census tracts, raising concerns about the allocation strategy. These findings suggest a need for more targeted and transparent investment practices. Future studies should investigate how and where funds were spent, and whether other non-financial factors—such as local governance or administrative capacity—played a role in determining outcomes. Stakeholders would benefit from more granular, program-level data to assess effectiveness and improve equity in resource distribution.

Data

Data from this report is collected from the U.S. Census Bureau and the Federal Housing Finance Agency. However, since census tracts have changed between 2010 and 2020, the data from 2020 was crosswalked to 2010 tracts with the NHGIS crosswalk. Throughout the South Atlantic Division section of this report, we will track indicators of social vulnerability using data from the U.S. Census Bureau and the Federal Housing Finance Agency. This data will be used in line withe Centers for Diseas Control (CDC)’s Social Vulnerability Index, which divides Social Vulnerability into four categories: socioeconomic status, household characterisitcs, racial and ethnic minority status, and housing type and transportation.

The SVI data sets contains 73057 tracts nationally. Divisionally, the South Atlantic Division contains 13706 tracts. In the South Atlantic Division, the most vulnerable state was Florida both in 2010 and in

  1. The most vulnerable by themes 1-4 in 2020 were: Broward County, Miami-Dade County, Alachua County, and Hillsborough County. For contrast, the most vulnerable by themes 1-4 in 2010 were: Broward County, Hillsborough County, Alachua County, and Miami-Dade County.

Tracts were identified as being eligible based on a number of factors. For the NMTC, census tract poverty rate would need to be equal or greater than 20%, Median Family Income would need to be less than or equal to 80% of the average, and tract unemployment to national unemployment ratio would need to be greater than 1.5. Tracts qualifying for NMTC awards would create a “flag”. Counting up these flags allow us to see which areas were the most socially vulnerable. For the NMTC, the largest ammount of flags counted was 9936. For the LIHTC, the maximum was 2457.

Data was grouped and observed in a number of ways, including clustering and diff-in-diff models. These models are used to show how different tracts are affected, as well as how the flag count relates to the amount of money invested into different tracts.

Analysis

Throughout this project, we have used various different methodologies in order to best understand our data.

Spatial analysis was used to map our division into states, counties, and tracts, in order to understand how geography may have impacted the data. Choropleth maps are shown to explain population, as well as our social vulnrability flags. Bivariate mapping is used to show relationships between SVI variables.

K-Means Clustering was used in order to discover relationships between flag counts and the amount of money invested into at-risk areas. Since there was a large variance between our maximum and minimum flag counts as well as our maximum and minimum money invested, outliers were affecting our data. Clustering allowed us to view the data in smaller groups that were similar in flag counts, and money invested. Diff-In-Diff regression was used in various models to view how the NMTC and LIHTC programs affected our tracts between 2010 and 2020.

Results

NMTC Diff-In-Diff Models

Socioeconomic SVI

We fitted a linear model (estimated using OLS) to predict SVI_FLAG_COUNT_SES with treat, post and cbsa (formula: SVI_FLAG_COUNT_SES ~ treat + post + treat * post + cbsa) where treat represents NMTC program participation, post is the year of 2020 after starting period of 2010, and cbsa controls for metro-level effects.

The model explains a statistically significant and moderate proportion of variance (R2 = 0.17, F(131, 10358) = 16.00, p < .001, adj. R2 = 0.16)

The effect of treat × post is statistically non-significant and negative (beta = -0.15, 95% CI [-0.38, 0.08], t(10358) = -1.28, p = 0.200; Std. beta = -0.01, 95% CI [-0.03, 6.08e-03])

Since the effect of treat x post is not statistically significant, we cannot conclude that the NMTC program had a measurable impact on socioeconomic status-related social vulnerability and economic outcomes.

Household Characteristics SVI

We fitted a linear model (estimated using OLS) to predict SVI_FLAG_COUNT_HHCHAR with treat, post and cbsa (formula: SVI_FLAG_COUNT_HHCHAR ~ treat + post + treat * post + cbsa) where treat represents NMTC program participation, post is the year of 2020 after starting period of 2010, and cbsa controls for metro-level effects.

The model explains a statistically significant and weak proportion of variance (R2 = 0.10, F(131, 10358) = 8.60, p < .001, adj. R2 = 0.09)

The effect of treat × post is statistically non-significant and negative (beta = -0.13, 95% CI [-0.29, 0.03], t(10358) = -1.59, p = 0.111; Std. beta = -0.01, 95% CI [-0.03, 3.44e-03])

Since the effect of treat x post is not statistically significant, we cannot conclude that the NMTC program had a measurable impact on household characteristics-related social vulnerability and economic outcomes.

Racial and Ethnic Minority SVI

We fitted a linear model (estimated using OLS) to predict SVI_FLAG_COUNT_REM with treat, post and cbsa (formula: SVI_FLAG_COUNT_REM ~ treat + post + treat * post + cbsa) where treat represents NMTC program participation, post is the year of 2020 after starting period of 2010, and cbsa controls for metro-level effects.

The model explains a statistically significant and moderate proportion of variance (R2 = 0.25, F(131, 10358) = 25.92, p < .001, adj. R2 = 0.24)

The effect of treat × post is statistically non-significant and negative (beta = -0.02, 95% CI [-0.09, 0.06], t(10358) = -0.43, p = 0.667; Std. beta = -3.67e-03, 95% CI [-0.02, 0.01])

Since the effect of treat x post is not statistically significant, we cannot conclude that the NMTC program had a measurable impact on racial and ethnic minority status-related social vulnerability and economic outcomes.

Housing and Transportation SVI

We fitted a linear model (estimated using OLS) to predict SVI_FLAG_COUNT_HOUSETRANSPT with treat, post and cbsa (formula: SVI_FLAG_COUNT_HOUSETRANSPT ~ treat + post + treat * post + cbsa) where treat represents NMTC program participation, post is the year of 2020 after starting period of 2010, and cbsa controls for metro-level effects.

The model explains a statistically significant and weak proportion of variance (R2 = 0.08, F(131, 10358) = 6.93, p < .001, adj. R2 = 0.07)

The effect of treat × post is statistically non-significant and positive (beta = 4.97e-03, 95% CI [-0.16, 0.17], t(10358) = 0.06, p = 0.953; Std. beta = 5.51e-04, 95% CI [-0.02, 0.02])

Since the effect of treat x post is not statistically significant, we cannot conclude that the NMTC program had a measurable impact on housing and transportation access-related social vulnerability and economic outcomes.

Overall SVI

We fitted a linear model (estimated using OLS) to predict SVI_FLAG_COUNT_OVERALL with treat, post and cbsa (formula: SVI_FLAG_COUNT_OVERALL ~ treat + post + treat * post + cbsa) where treat represents NMTC program participation, post is the year of 2020 after starting period of 2010, and cbsa controls for metro-level effects.

The model explains a statistically significant and moderate proportion of variance (R2 = 0.18, F(131, 10358) = 16.97, p < .001, adj. R2 = 0.17)

The effect of treat × post is statistically non-significant and negative (beta = -0.29, 95% CI [-0.75, 0.16], t(10358) = -1.26, p = 0.207; Std. beta = -0.01, 95% CI [-0.03, 6.23e-03])

Since the effect of treat x post is not statistically significant, we cannot conclude that the NMTC program had a measurable impact on socioeconomic, household characteristics, racial and ethnic minority status, and housing and transportation access-related social vulnerability and economic outcomes.

Median Income Economic Outcomes

We fitted a linear model (estimated using OLS) to predict MEDIAN_INCOME with treat, post and cbsa (formula: MEDIAN_INCOME ~ treat + post + treat * post + cbsa) where treat represents NMTC program participation, post is the year of 2020 after starting period of 2010, and cbsa controls for metro-level effects.

The model explains a statistically significant and moderate proportion of variance (R2 = 0.20, F(131, 10350) = 19.93, p < .001, adj. R2 = 0.19)

The effect of treat × post is statistically significant and positive (beta = 0.06, 95% CI [0.01, 0.11], t(10350) = 2.54, p = 0.011; Std. beta = 0.02, 95% CI [5.08e-03, 0.04])

Since the effect of treat x post is statistically significant, we can conclude that the NMTC program had a measurable impact on Median Income-related social vulnerability and economic outcomes.

Median Home Value Economic Outcomes

We fitted a linear model (estimated using OLS) to predict MEDIAN_HOME_VALUE with treat, post and cbsa (formula: MEDIAN_HOME_VALUE ~ treat + post + treat * post + cbsa) where treat represents NMTC program participation, post is the year of 2020 after starting period of 2010, and cbsa controls for metro-level effects.

The model explains a statistically significant and substantial proportion of variance (R2 = 0.42, F(131, 10116) = 56.78, p < .001, adj. R2 = 0.42)

The effect of treat × post is statistically non-significant and positive (beta = 0.02, 95% CI [-0.04, 0.09], t(10116) = 0.67, p = 0.503; Std. beta = 5.06e-03, 95% CI [-9.74e-03, 0.02])

Since the effect of treat x post is not statistically significant, we cannot conclude that the NMTC program had a measurable impact on Median Home Value-related social vulnerability and economic outcomes.

House Price Index Economic Outcomes

We fitted a linear model (estimated using OLS) to predict HOUSE_PRICE_INDEX with treat, post and cbsa (formula: HOUSE_PRICE_INDEX ~ treat + post + treat * post + cbsa) where treat represents NMTC program participation, post is the year of 2020 after starting period of 2010, and cbsa controls for metro-level effects.

The model explains a statistically significant and substantial proportion of variance (R2 = 0.37, F(123, 4962) = 23.82, p < .001, adj. R2 = 0.36)

The effect of treat × post is statistically non-significant and positive (beta = 0.05, 95% CI [-0.04, 0.13], t(4962) = 1.05, p = 0.292; Std. beta = 0.01, 95% CI [-0.01, 0.03])

Since the effect of treat x post is not statistically significant, we cannot conclude that the NMTC program had a measurable impact on House Price Index-related social vulnerability and economic outcomes.

LIHTC Diff-In-Diff Models

Socioeconomic SVI

We fitted a linear model (estimated using OLS) to predict SVI_FLAG_COUNT_SES with treat, post and cbsa (formula: SVI_FLAG_COUNT_SES ~ treat + post + treat * post + cbsa) where treat represents LIHTC program participation, post is the year of 2020 after starting period of 2010, and cbsa controls for metro-level effects.

The model explains a statistically significant and substantial proportion of variance (R2 = 0.26, F(77, 928) = 4.33, p < .001, adj. R2 = 0.20)

The effect of treat × post is statistically non-significant and positive (beta = 0.03, 95% CI [-0.37, 0.42], t(928) = 0.14, p = 0.888; Std. beta = 3.98e-03, 95% CI [-0.05, 0.06])

Since the effect of treat x post is not statistically significant, we cannot conclude that the LIHTC program had a measurable impact on socioeconomic status-related social vulnerability and economic outcomes.

Household Characteristics SVI

We fitted a linear model (estimated using OLS) to predict SVI_FLAG_COUNT_HHCHAR with treat, post and cbsa (formula: SVI_FLAG_COUNT_HHCHAR ~ treat + post + treat * post + cbsa) where treat represents LIHTC program participation, post is the year of 2020 after starting period of 2010, and cbsa controls for metro-level effects.

The model explains a statistically significant and moderate proportion of variance (R2 = 0.24, F(77, 928) = 3.86, p < .001, adj. R2 = 0.18)

The effect of treat × post is statistically non-significant and negative (beta = -0.12, 95% CI [-0.42, 0.18], t(928) = -0.77, p = 0.442; Std. beta = -0.02, 95% CI [-0.08, 0.03])

Since the effect of treat x post is not statistically significant, we cannot conclude that the LIHTC program had a measurable impact on household characteristics-related social vulnerability and economic outcomes.

Racial and Ethnic Minority SVI

We fitted a linear model (estimated using OLS) to predict SVI_FLAG_COUNT_REM with treat, post and cbsa (formula: SVI_FLAG_COUNT_REM ~ treat + post + treat * post + cbsa) where treat represents LIHTC program participation, post is the year of 2020 after starting period of 2010, and cbsa controls for metro-level effects.

The model explains a statistically significant and substantial proportion of variance (R2 = 0.37, F(77, 928) = 6.99, p < .001, adj. R2 = 0.31)

The effect of treat × post is statistically non-significant and positive (beta = 2.18e-03, 95% CI [-0.12, 0.12], t(928) = 0.04, p = 0.972; Std. beta = 9.29e-04, 95% CI [-0.05, 0.05])

Since the effect of treat x post is not statistically significant, we cannot conclude that the LIHTC program had a measurable impact on racial and ethnic minority status-related social vulnerability and economic outcomes.

Housing and Transportation SVI

We fitted a linear model (estimated using OLS) to predict SVI_FLAG_COUNT_HOUSETRANSPT with treat, post and cbsa (formula: SVI_FLAG_COUNT_HOUSETRANSPT ~ treat + post + treat * post + cbsa) where treat represents LIHTC program participation, post is the year of 2020 after starting period of 2010, and cbsa controls for metro-level effects.

The model explains a statistically significant and moderate proportion of variance (R2 = 0.17, F(77, 928) = 2.44, p < .001, adj. R2 = 0.10)

The effect of treat × post is statistically non-significant and negative (beta = -0.02, 95% CI [-0.30, 0.27], t(928) = -0.12, p = 0.903; Std. beta = -3.65e-03, 95% CI [-0.06, 0.06])

Since the effect of treat x post is not statistically significant, we cannot conclude that the LIHTC program had a measurable impact on housing and transportation access-related social vulnerability and economic outcomes.

Overall SVI

We fitted a linear model (estimated using OLS) to predict SVI_FLAG_COUNT_OVERALL with treat, post and cbsa (formula: SVI_FLAG_COUNT_OVERALL ~ treat + post + treat * post + cbsa) where treat represents LIHTC program participation, post is the year of 2020 after starting period of 2010, and cbsa controls for metro-level effects.

The model explains a statistically significant and substantial proportion of variance (R2 = 0.28, F(77, 928) = 4.65, p < .001, adj. R2 = 0.22)

The effect of treat × post is statistically non-significant and negative (beta = -0.10, 95% CI [-0.83, 0.62], t(928) = -0.28, p = 0.777; Std. beta = -7.92e-03, 95% CI [-0.06, 0.05])

Since the effect of treat x post is not statistically significant, we cannot conclude that the LIHTC program had a measurable impact on socioeconomic, household characteristics, racial and ethnic minority status, and housing and transportation access-related social vulnerability and economic outcomes.

Median Income Economic Outcomes

We fitted a linear model (estimated using OLS) to predict MEDIAN_INCOME with treat, post and cbsa (formula: MEDIAN_INCOME ~ treat + post + treat * post + cbsa) where treat represents LIHTC program participation, post is the year of 2020 after starting period of 2010, and cbsa controls for metro-level effects.

The model explains a statistically significant and substantial proportion of variance (R2 = 0.31, F(77, 926) = 5.47, p < .001, adj. R2 = 0.26)

The effect of treat × post is statistically non-significant and negative (beta = -4.13e-03, 95% CI [-0.13, 0.12], t(926) = -0.06, p = 0.949; Std. beta = -1.73e-03, 95% CI [-0.06, 0.05])

Since the effect of treat x post is not statistically significant, we cannot conclude that the LIHTC program had a measurable impact on Median Income-related social vulnerability and economic outcomes.

Median Home Value Economic Outcomes

We fitted a linear model (estimated using OLS) to predict MEDIAN_HOME_VALUE with treat, post and cbsa (formula: MEDIAN_HOME_VALUE ~ treat + post + treat * post + cbsa) where treat represents LIHTC program participation, post is the year of 2020 after starting period of 2010, and cbsa controls for metro-level effects.

The model explains a statistically significant and substantial proportion of variance (R2 = 0.41, F(77, 880) = 8.00, p < .001, adj. R2 = 0.36)

The effect of treat × post is statistically non-significant and positive (beta = 4.72e-03, 95% CI [-0.17, 0.18], t(880) = 0.05, p = 0.958; Std. beta = 1.36e-03, 95% CI [-0.05, 0.05])

Since the effect of treat x post is not statistically significant, we cannot conclude that the LIHTC program had a measurable impact on Median Home Value-related social vulnerability and economic outcomes.

House Price Index Economic Outcomes

We fitted a linear model (estimated using OLS) to predict HOUSE_PRICE_INDEX with treat, post and cbsa (formula: HOUSE_PRICE_INDEX ~ treat + post + treat * post + cbsa) where treat represents LIHTC program participation, post is the year of 2020 after starting period of 2010, and cbsa controls for metro-level effects.

The model explains a statistically significant and substantial proportion of variance (R2 = 0.40, F(32, 241) = 4.98, p < .001, adj. R2 = 0.32)

The effect of treat × post is statistically non-significant and positive (beta = 0.09, 95% CI [-0.19, 0.36], t(241) = 0.64, p = 0.523; Std. beta = 0.03, 95% CI [-0.07, 0.13])

Since the effect of treat x post is not statistically significant, we cannot conclude that the LIHTC program had a measurable impact on House Price Index-related social vulnerability and economic outcomes.

Discussion and Recommendations

From our bivariate maps, we can see that there is a high ratio of flags to population in West Virginia, eastern North Carolina, southern Georgia, and across Florida. We can also see that the tracts appear to be very similar between 2010 and 2020. Although this model does not show total populations and may not be the best way to view this data, it could suggest that the programs did not have the expected affect.

Suprisingly, when we look at our amount of flags compared to the amount of money invested, we see that there is not a strong correlation. This would suggest that the amount of money being spent may not have been decided based on social vulnerability factors soley. Due to this bias in our data, we viewed our data in groups so that we may see how different flags and spending levels correlate. However, even in this model, we cannot conclude that the amount of money spent had a significant impact on social vulnerability due to the amount of flags between 2010 and 2020 remaining similar.

Finally, when we viewed the impact of both the LIHTC and NMTC programs onto different social vulnerability outcomes, we cannot conclude that either programs had an impact on any of the outcomes.

Due to the large varience in flags and amount of money spent, as well as the lack of significance in the outcomes, it would be beneficial to study further into how the money was spent specifically. It does not appear as if funding was going where it was most needed. Although this study could not conclude that social vulnerability flags were related to the amount of money invested, it would be interesting to see what other factors or decision makers were impactful. Understanding how the money was spent could unlock further understanding into how the money could be used to see a better result.

References

Packages

Analyses were conducted using the R Statistical language (version 4.4.0; R Core Team, 2024) on Windows 10 x64 (build 19045)

Data