The Socioeconomic Effects of the New Markets Tax Credit and Low Income Housing Tax Credit in Low- income Metropolitan Census Tracts Michael Henderson Dissertation Proposal Defense May 5, 2015
The Socioeconomic Effects of the New Markets Tax Credit and Low
Income Housing Tax Credit in Low-income Metropolitan Census Tracts
Michael HendersonDissertation Proposal Defense
May 5, 2015
Introduction: The socioeconomic effects of place-based investment
NMTC: economic development LIHTC: affordable housing
First study to examine NMTC and LIHTC in a single framework Structurally & administratively similar Outputs are very different Target an overlapping set of census tracts; may push socioeconomic
outcomes in opposite direction What makes a census tract an attractive target for place-based
investment?
Contributions & significance Methodological improvements
Address two potential sources of bias in previous studies By considering the complex policy dynamics at play in struggling communities By better controlling for factors that affect treatment selection
Applied relevance Place-based interventions are an increasingly important part of the federal
approach to poverty Do NMTC & LIHTC target the “right” poor places? How do we define success in the low-income, high-poverty context?
Problem: Rise in Concentrated Poverty Consequences of living in a poor neighborhood (Coutts & Kawachi,
2006) School quality Crime & safety Local services & amenities Connections to labor market & employment opportunities
US is increasingly economically segregated since 2000 (Jargowsky, 2013)
Percentage of poor population living in poor neighborhoods increased in 61 of 100 largest MSAs
High poverty census tracts in 2000: 108
High poverty census tracts in 2000: 108
High poverty census tracts in 2010: 207
Federal policy perspectives People-based: individual/family level
Treat neighborhood effects by moving poor people out of poor places Place-based: community level
Target distressed, economically underperforming areas No transfer of benefits to specific individuals Treat neighborhood effects by improving conditions in the places where poor
people live
Is targeting poor places a viable means of improving the lives of poor people?
Place-based Policies Criticism: waste and inefficiency
Poor places are unable to effectively harness investments and generate positive externalities (Glaeser & Gottlieb, 2008)
Does little to help current residents; benefits are siphoned off by investors, landowners, higher-skilled commuters, and new residents (Crane & Manville, 2008; Glaeser & Gottlieb, 2008; Gurley-Calvez, Gilbert, Harper, Marples, & Daly, 2009)
Response: evaluation of Empowerment Zone (EZ) program (Busso, Gregory, & Kline, 2010) found:
Benefits generated by EZ investment in low-income census tracts significantly exceeded costs through increased productivity
Increased total employment Significant wage increases for local residents
Place-based Policies Criticism:
Tying benefits to specific places distorts the residential location decisions of the poor (Kraybill & Kilkenny, 2003)
Response: Assumes high level of mobility among populations living in places typically
targeted (Ladd, 1994) Spatial mismatch suggests a legitimate use for strategies conducive to local
employment growth (Arnott, 1998; Gobillon, Selod, & Zenou, 2007)
So where does that leave us? No clear consensus on the viability of targeting places Place-based policies are an increasingly important part of the federal
approach to poverty $82B in 2012
Important to attack structural disadvantage at multiple levels of influence
New Markets Tax Credit (NMTC) Operational since: 2003 Administered by: IRS and CDFI Fund (Dept. of Treasury) Goal of NMTC: To trigger new investment in businesses & non-
residential real estate projects in low-income communities (CDFI Fund, 2012; Marples, 2008)
NMTC eligibility criteria: Median family income (MFI) in census tract < 80% of area median income, or Poverty rate >20%
$40 billion in tax credit authority awarded
How NMTC works1. Community development entity (CDE) applies for tax credit
authority2. CDE uses tax credits to attract private investors3. Investor receives tax credit equal to 39 percent of amount invested
Received over 7 years
4. CDE uses investment capital raised to offer financing to qualified projects located in qualified low-income census tracts Typically at more generous rates than are available in the market
Research on NMTC Does NMTC trigger new investment?
Harger & Ross (2014) find that NMTC triggers additional private investment; Others find no evidence of NMTC as an investment catalyst (Forbes, 2006; Swack, Hangen, &
Northrup, 2015) Does NMTC subsidize investment that would have occurred anyway?
Gurley-Calvez, et. al (2009) find that NMTC changes investment patterns of individual private investors,
Crowd out a concern with corporate investors (banks) Community relevance of NMTC
Structural factors & demand for tax credits favor projects that benefit low-income populations (NMTC Coalition, 2012; Abravanel et al., 2013)
Minority CDEs rated lower & less likely to win tax credit authority than comparable white-owned CDEs (Brostek, 2009)
Research on NMTC Freedman (2012) offers only evidence to date on NMTC
socioeconomic effects: Decreases in poverty and unemployment rates No effect on home values or housing turnover Some effects may be due to changing composition of residents
Low Income Housing Tax Credit (LIHTC) Authorized in: 1986 Administered by: IRS & each state’s housing authority Goal of LIHTC: creation of new affordable housing units (Baum-Snow & Marion,
2009) LIHTC is the federal government’s primary tool for increasing supply of affordable housing
LIHTC eligibility criteria: No location restriction; tax credits dependent on agreements to restrict rent & income levels Enhanced tax credit for units placed in low-income tracts:
MFI in census tract < 60% of area income, or Poverty rate >25%
2.5 million new affordable housing units
How LIHTC works1. Housing developer applies to state housing authority for tax credit
authority2. Developer uses tax credits to attract private investors3. Investment capital used to finance construction
Does LIHTC investment Reinforce patterns of concentrated poverty, or improve amenities in low-income communities? Baum-Snow and Marion (2009) find that…it depends
Placed in gentrifying neighborhoods, LIHTC units led to declines in income and increased housing turnover
In stable or declining neighborhoods, LIHTC generated positive spillovers through increased property values
No clear evidence that LIHTC increases racial segregation (Horn & O'Regan, 2011), an issue tied to economic segregation
But, LIHTC units disproportionately used by HCV recipients (Williamson, Smith, & Strambi-Kramer, 2009)
Deng (2011) finds that LIHTC increases property values Others find evidence that LIHTC leads to property value decline (Rosenthal, 2008)
Knowledge gaps1. Unmeasured factors may drive NMTC & LIHTC site selection
Theory & evidence suggest that actors in NMTC and LIHTC have locational preferences (Baum-Snow & Marion, 2009; Freedman, 2012)
Perceptions of a census tract’s current & future viability
2. NMTC & LIHTC may push socioeconomic outcomes in opposite directions LIHTC criticism: by offering enhanced tax credit for units built in low-income
communities, LIHTC reinforces patterns of concentrated poverty(Oakley, 2008)
Study overview Study area
Low-income census tracts in 100 largest MSAs Low-income: MFI ratio below 80 percent or poverty rate above 20 percent (NMTC
eligibility criteria)
Socioeconomic change since 2000 MFI; poverty rate; unemployment rate; home values; inequality
Data sources Census (1990; 2000; 2009-2013 ACS) Publicly available NMTC & LIHTC program data Longitudinal Tract Database
1. What drives the locational patterns of NMTC and LIHTC placement? Hypothesis: NMTC and LIHTC activity is clustered in similar
neighborhood types In the census tracts most likely to experience socioeconomic ascent since
2000 Goal: to uncover and specify the latent construct of “developer and
investor preferences” To identify other meaningful factors shaping NMTC and LIHTC locational
patterns
Approach1. Principal component analysis (PCA):
On the observable year 2000 baseline predictors of socioeconomic ascent Variables to be included in study will be drawn from:
Studies using a similar approach to evaluate socioeconomic effects of policy intervention in low-income census tracts
Literature on neighborhood change and gentrification
2. Cluster analysis: Use the factors identified in PCA to develop a typology of low-income census tracts Clusters are census tracts that are similar across all factors
3. Examine patterns of NMTC & LIHTC activity across neighborhood types Are NMTC & LIHTC activity concentrated in same neighborhood types? Were these the neighborhood types most likely to experience socioeconomic ascent?
4. Apply findings to second research question
2. What are the effects of NMTC and LIHTC activity? Predictions:
The presence of NMTC improves socioeconomic outcomes The presence of LIHTC leads to worse socioeconomic outcomes
Approach Propensity score matching
Approach for making causal claims in observational studies where random assignment is not possible
Propensity score: the predicted probability that a census tract would receive a treatment given pretreatment attributes
Variation that allows for calculating separate predicted probabilities of falling into each of several different treatment conditions
NMTC and LIHTC NMTC only LIHTC only Neither NMTC nor LIHTC
Treatment assignments
Mean Minimum MaximumNMTC and LIHTC 2.05 0.00 41.43NMTC only 7.40 1.22 48.18LIHTC only 10.67 0.03 75.72Neither 79.89 18.49 97.14
Treatment
Min, max, and mean predicted probabilities ofcensus tracts that did receive a particular treatment combination would have received that treatment combination, given pretreatment characteristics
Matching variables Match on baseline characteristics that:
Affect the treatment assignment, and controlling for treatment assignment, affect socioeconomic change over time
(Ho, Imai, King, & Stuart, 2007)
Matching variableDemographic% White 38.04 32.97 40.04 40.71% Black 37.14 31.88 33.65 27.06
Socioeconomic stabilityUnemployment rate 13.20 11.92 10.42 9.71Poverty rate 31.20 27.04 24.97 21.57
74.20 70.42 70.25 67.77
52.59 56.50 57.72 62.00
Socioeconomic trajectory
51.54 36.80 35.23 30.01
Housing characteristics% Vacant Housing Units 12.14 9.14 9.85 8.47
Human Capital
62.10 62.41 62.82 61.35
14.79 14.71 12.50 14.56
log population density 8.37 8.55 8.09 8.50
MFI in census tract / MFI in MSA
Home value growth rate (1990-2000)
Population % age 25+ with high school diplomaPopulation % age 25+ with bachelor's degree
Descriptive Statistics of Matching VariablesNMTC and LIHTC
(n=332)NMTC
(n=1200)LIHTC
(n=1730)Neither
(n=12995)
% households in current residence under 10 yrs
ModelTreatment Comparison1 NMTC and LIHTC NMTC only2 NMTC and LIHTC LIHTC only3 NMTC and LIHTC Neither4 NMTC only LIHTC only5 NMTC only Neither6 LIHTC only Neither
Difference-in-difference models and predicted effect of treatment
+-
Treatment effect-+No difference+
Difference-in-difference of census tract socioeconomic characteristics
2000-2010Mean
difference Std. Std. t-valueMedian Family Income
Both vs. NMTC 11364 879 -2068 1239 -1.67Both vs. LIHTC 6680 742 2571 1049 2.45 *Both vs. Neither 8096 772 1154 1088 1.06NMTC vs. LIHTC 6908 439 4324 623 6.95 ***NMTC vs. neither 11170 497 327 705 0.46LIHTC vs. neither 8776 307 -2352 434 -5.42 ***
Significant at the * 5% level, ** 1% level, *** .1% level
Comparison Tracts Treatment Tracts
Treatment effect
Difference-in-difference of census tract socioeconomic characteristics
2000-2010Mean
difference Std. Std. t-valuePoverty Rate
Both vs. NMTC -2.38 0.70 0.65 0.99 0.65Both vs. LIHTC 0.39 0.68 -2.19 0.96 -2.28 *Both vs. Neither -0.45 0.70 -1.35 0.98 -1.38NMTC vs. LIHTC 0.57 0.34 -1.97 0.48 -4.07 ***NMTC vs. neither -2.29 0.36 -0.22 0.51 -0.43LIHTC vs. neither 0.44 0.26 1.43 0.37 3.82 ***
Significant at the * 5% level, ** 1% level, *** .1% level
Comparison Tracts Treatment Tracts
Treatment effect
Difference-in-difference of census tract socioeconomic characteristics
2000-2010Mean
difference Std. Std. t-valueUnemployment Rate
Both vs. NMTC 0.035 0.005 -0.010 0.007 -1.37Both vs. LIHTC 0.038 0.005 -0.013 0.007 -1.98 *Both vs. Neither 0.038 0.005 -0.013 0.007 -1.84NMTC vs. LIHTC 0.043 0.002 -0.012 0.003 -3.65 ***NMTC vs. neither 0.030 0.002 -0.006 0.003 -1.59LIHTC vs. neither 0.048 0.002 0.003 0.003 1.34
Significant at the * 5% level, ** 1% level, *** .1% level
Comparison Tracts Treatment Tracts
Treatment effect
Conclusion Isolating the neighborhood effects of place-based policies is difficult
Need to take into account the complex policy dynamics at play in struggling neighborhoods
Need to understand the motivations of key actors in market-driven programs What does success look like in the high-poverty context?