CoChat
Sign Up
Cited by 264Housing Market and EconomicsUrban, Neighborhood, and Segregation StudiesUrban Planning and Governance

The effects of low income housing tax credit developments on neighborhoods

Open access
Baum-Snow, Nathaniel, Marion, Justin·2009·Journal of Public Economics
DOI 10.1016/j.jpubeco.2009.01.001PMC PMC3824966
Open this paper in CoChat — free account

Abstract

This paper evaluates the impacts of new housing developments funded with the Low Income Housing Tax Credit (LIHTC), the largest federal project based housing program in the U.S., on the neighborhoods in which they are built. A discontinuity in the formula determining the magnitude of tax credits as a function of neighborhood characteristics generates pseudo-random assignment in the number of low income housing units built in similar sets of census tracts. Tracts where projects are awarded 30 percent higher tax credits receive approximately six more low income housing units on a base of seven units per tract. These additional new low income developments cause homeowner turnover to rise, raise property values in declining areas and reduce incomes in gentrifying areas in neighborhoods near the 30th percentile of the income distribution. LIHTC units significantly crowd out nearby new rental construction in gentrifying areas but do not displace new construction in stable or declining areas.

AI snapshot

Written from the paper’s full text.

TL;DR

This paper uses a regression discontinuity at the Qualified Census Tract threshold to estimate effects of Low Income Housing Tax Credit developments on neighborhoods. It finds higher owner turnover, lower median incomes especially in gentrifying areas, higher home values in declining/stable areas, and modest crowd-out of nearby rental construction.

Key findings

  • Tracts just above the QCT threshold receive about 6 additional LIHTC units on a base of 7 units per tract between 1994 and 1999, driven by the 30 percent higher tax credits.
  • Each additional 100 LIHTC units causes a 5.9 percentage point increase in the fraction of owners moving into a block group between 1995 and 2000, and reduces median household income by 9.3 percent overall and by 21.5 percent in gentrifying areas.
  • Each additional 100 LIHTC units raises median home values by 14.9 percent overall, with effects of 10.6 percent in stable areas and 5.6 percent in gentrifying areas.
  • Each LIHTC unit increases newly built rental units within 1 km by 0.80, with no effect on owner-occupied units; in gentrifying areas each LIHTC unit increases new rental construction by only 0.37 units, implying substantial crowd-out.

Methods

  • Regression discontinuity design exploiting the 50 percent eligible-household cutoff for Qualified Census Tract status, which gives projects 30 percent higher tax credits and generates pseudo-random variation in LIHTC units.
  • Data include HUD LIHTC projects placed in service 1987–2005, focusing on 1994–1999 metropolitan projects, merged with 1980, 1990, and 2000 census tract and block group data; applications data cover California, Texas, and New Jersey in 2004–2005.
  • Analysis uses tract-level and 1 km ring-level first-stage and reduced-form/IV estimates, instrumenting LIHTC units with QCT status or the number of blocks in qualified tracts, controlling for cubic control functions, county fixed effects, and demographic/housing characteristics.

Limitations

  • The paper states that the RD design supports causal statements only for areas that are fairly poor but not extremely poor near the QCT threshold, and that treatment effects may differ by neighborhood composition and developer incentives.
  • The paper states missing data concerns: project location is missing for 9.5 percent of projects and unit counts for 4.9 percent; only placed-in-service projects are observed, and applications data are limited to California, Texas, and New Jersey in 2004–2005.
  • The paper states crowd-out estimates may overstate effects because about 70 percent of exogenous LIHTC units are new construction while the rest are rehabilitations, and may understate crowd-out because units built 1994–1999 are compared with outcomes for 1995–1999; separate declining-area estimates are not reported due to a marginally significant first stage.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1016/j.jpubeco.2009.01.001

Cites

17 of 17 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

0 of 5 cited sources checked · 0 supported · 0 flagged

  • Findingown data

    In gentrifying areas, there is significant crowd-out of private construction; each additional LIHTC unit is only associated with 0.37 newly constructed rental units overall.

    “However, in gentrifying areas there appears to be significant crowd-out of private construction, as each additional LIHTC unit is only associated with 0.37 newly constructed rental units overall.”
    Quote in the paperSubstantial claim
  • Findingown data

    The fraction of renters entering the area is 0.8 percentage points higher for each 100 additional low-income units.

    “We also find greater turnover among renters, as the fraction of renters entering the area is 0.8 percentage points higher for each 100 additional low-income units.”
    Quote in the paperNot a substantial claim
  • Findingown data

    Each new LIHTC unit increases the number of recently built rental units by 0.8 units within one kilometer of the project site.

    “Overall, each new LIHTC unit increases the number of recently built rental units by 0.8 units within one kilometer of the project site”
    Quote in the paperSubstantial claim
  • Findingown data

    In stable block groups, there is very little crowd-out of new rental unit construction, with an IV coefficient of 0.99.

    “In stable block groups, we find very little crowd-out of new rental unit construction, as we obtain an IV coefficient LIHTC units in the full specification of 0.99.”
    Quote in the paperSubstantial claim
  • Findingown data

    The adverse effects of LIHTC units on median block group income seem concentrated in gentrifying block groups.

    “First, the adverse effects of LIHTC units on median block group income seem concentrated in gentrifying block groups.”
    Quote in the paperSubstantial claim
  • May combine two claims. 41
  • Classified as probably not a substantial claim. 30

  • May combine two claims.

    Findingown data

    Poorer tracts receive more LIHTC developments, with a sizable portion of the income-development gradient explained by QCT status.

    “It is evident that poorer tracts receive more LIHTC developments, with a sizable portion of the income-development gradient explained by QCT status.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The IV estimate for the effect on newly built owner occupied units is −0.036, statistically insignificant and smaller than the OLS estimate of 0.09.

    “We find that the IV estimate for the effect on newly built owner occupied units to be −0.036, which is both statistically insignificant and smaller than the OLS estimate of 0.09.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The effect of LIHTC on the fraction of renters that entered within the past 5 years is estimated to be 0.011 in the stable sample, though only 0.005 in the gentrifying sample.

    “First, the effect of LIHTC on the fraction of renters that entered within the past 5 years is estimated to be 0.011 in the stable sample, though only 0.005 in the gentrifying sample.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    There are positive but statistically insignificant effects of QCT status on the number of normalized accepted units and number of accepted projects.

    “We also estimate positive but statistically insignificant effects of QCT status on the number of normalized accepted units and number of accepted projects.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Findingown data

    Splitting the sample instead on terciles of the 1990 housing value distributions within MSAs produces qualitatively similar results, though coefficients are measured with less precision.

    “Splitting the sample instead on terciles of the 1990 housing value distributions within MSAs produces qualitatively similar results, though coefficients are measured with less precision.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    In tracts below the 50 percent eligibility threshold, 67 percent of occupied rental apartments rented below LIHTC regulated rents in 2000, up from 52 percent in 1990.

    “We show that in the tracts below the 50 percent eligibility threshold, 67 percent of occupied rental apartments rented below LIHTC regulated rents in 2000. This increased substantially from 52 percent in 1990”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    In the full specification, 4 additional new units are built in QCTs just above the threshold, representing the majority of the 6 additional units of all types received by these tracts.

    “In the full specification, we estimate that 4 additional new units are built in QCTs just above the threshold, representing the majority of the 6 additional units of all types received by these tracts.”
    Quote in the paperNot a substantial claim
    • The authors state this as an assumption: Unobserved tract characteristics that influence outcomes of interest do not vary discontinuously across the QCT eligibility threshold.

  • May combine two claims.

    Findingown data

    In gentrifying areas, there is evidence of substantial crowd-out; each LIHTC unit only increases newly constructed housing by 0.37 units.

    “In the gentrifying areas, we see evidence of substantial crowd-out. Each LIHTC unit in these areas only increases newly constructed housing by 0.37 units.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    New LIHTC units cause neighborhood income to decline, and this effect is concentrated in gentrifying areas.

    “reduce incomes in gentrifying areas in neighborhoods near the 30th percentile of the income distribution.”
    • Also stated in: § 1 Introduction · plain
      “We also find that neighborhood income declines as a result of new LIHTC units nearby, and this effect is concentrated in gentrifying areas.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    New LIHTC units have a positive effect on neighborhood home values; every 100 additional LIHTC units leads to a 14.9 percent increase in median home value, though this effect is close to zero in gentrifying areas.

    “raise property values in declining areas”
    • Also stated in: § 1 Introduction · plain
      “In addition, we find a positive effect of LIHTC units on neighborhood home values. Every 100 additional LIHTC units leads to a 14.9 percent increase in the median home value, though this effect is close to zero in gentrifying areas.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    A block group with 100 additional nearby LIHTC units has 1.2 percent slower growth in income in stable areas and 0.7 percent slower growth in gentrifying areas.

    “A block group with 100 additional nearby LIHTC units has 1.2 percent slower growth in income in stable areas and 0.7 percent slower growth in gentrifying areas.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The IV results for the full sample are in general of similar sign though larger in magnitude than the analogous OLS estimates.

    “The IV results for the full sample are in general of similar sign though larger in magnitude than the analogous OLS estimates.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Neighborhoods near the QCT eligibility threshold have an average poverty rate of approximately 25 percent, are about 30 percent black and have about 60 percent high school graduates.

    “These neighborhoods have an average poverty rate of approximately 25 percent, are about 30 percent black and have about 60 percent high school graduates.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingnone given

    Each estimate in Table 4 except that for units per project is precise.

    “Each estimate except that for units per project is precise.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    LIHTC developments depress local median household income and increase turnover in owner occupied housing units within 1 km of these projects.

    “We find that LIHTC developments depress local median household income and increase turnover in owner occupied housing units within 1 km of these projects.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    An additional 100 LIHTC units raises the growth in median house values by 14.9 percent overall, 10.6 percent in stable areas, and 5.6 percent in gentrifying areas.

    “Overall, we find that an additional 100 LIHTC units raises the growth in median house values by 14.9 percent, though the estimated effect is 10.6 percent in stable areas and 5.6 percent in gentrifying areas.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Low income units in the neighborhood raise housing values, though this effect is concentrated in stable and declining neighborhoods.

    “In addition, we find that low income units in the neighborhood raise housing values, though this effect is concentrated in stable and declining neighborhoods.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Stable areas receive 0.31 additional newly constructed renter occupied units for each LIHTC unit built, and the estimate is 0.47 for gentrifying areas.

    “Stable areas receive 0.31 additional newly constructed renter occupied units for each LIHTC unit built, and we obtain an estimate of 0.47 for gentrifying areas.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Units per proposed project seem to be higher in tracts qualifying for higher tax credits, however this is imprecisely estimated.

    “Units per proposed project also seem to be higher in the tracts qualifying for higher tax credits, however this is imprecisely estimated.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data

    There is no discernible trend in the size of LIHTC projects.

    “However, there is no discernible trend in the size of projects.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Local crowd out is stronger in gentrifying areas, as each LIHTC unit only increases new rental construction by an estimated 0.37 units in these areas.

    “Local crowd out is stronger in gentrifying areas, as each LIHTC unit only increases new rental construction by an estimated 0.37 units in these areas.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Including full controls, gentrifying areas show a first stage coefficient of 0.27 while declining neighborhoods show a first stage coefficient of 0.22.

    “Including the full set of controls, gentrifying areas show a first stage coefficient of 0.27 while declining neighborhoods show a first stage coefficient of 0.22.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The discontinuity in units per project is slightly larger for new construction, though it remains statistically insignificant.

    “The discontinuity in units per project is slightly larger for new construction, though it remains statistically insignificant.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    The average tract with fewer than 40 percent eligible households has 84.9 units per project, while the average tract in the rest of the income distribution has between 60 and 70 units per project.

    “The average tract with fewer than 40 percent eligible households has 84.9 units per project, while the average tract in the rest of the income distribution has between 60 and 70 units per project.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    The response of LIHTC units to QCT status is driven by developers' location choices rather than government preferences, as the discontinuity in units at the threshold is driven by the number of applications by developers rather than state housing authorities' acceptance rate.

    “The response of LIHTC units to QCT status seems driven by developers’ location choices rather than government preferences, as we find that the discontinuity in units at the threshold is driven by the number of applications by developers rather than state housing authorities’ acceptance rate of proposed projects.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Findingown data

    Graphs in Figure 2 exhibit noticeable discontinuities at the QCT eligibility threshold.

    “Graphs in Figure 2 exhibit noticeable discontinuities at the QCT eligibility threshold.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    New LIHTC units cause homeowner turnover to rise; 100 additional LIHTC units causes a 5.9 percentage point increase in the fraction of owners moving to the neighborhood between 1995 and 2000.

    “These additional new low income developments cause homeowner turnover to rise”
    • Also stated in: § 1 Introduction · plain
      “We show that owner turnover rates are significantly higher near new LIHTC projects. On average, 100 additional LIHTC units causes a 5.9 percentage point increase in the fraction of owners moving to the neighborhood between 1995 and 2000.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Median household income is lower as a result of LIHTC units, and the effect is significantly stronger in the IV results.

    “Median household income is again found to be lower as a result of LIHTC units, and the effect is significantly stronger in the IV results.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

    Premisenone given

    An essential element required to make convincing empirical progress on understanding how exogenous changes in neighborhood attributes shift equilibrium neighborhood composition and impacts of neighborhood attributes on individual residents is some sort of random variation of people across neighborhoods.

    “An essential element required to make convincing empirical progress on both topics is some sort of random variation of people across neighborhoods.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisereasoning

    Incorporating information from surrounding tracts may improve efficiency by making use of more potential identifying information.

    “Incorporating information from surrounding tracts may also improve efficiency by making use of more potential identifying information.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Premisenone given

    An observed project is one that was both proposed and selected by a state housing authority.

    “An observed project is one that was both proposed and selected by a state housing authority.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Study design and conductown data

    There are no discontinuities in housing values, pre-existing trends of housing values, or median household incomes across the QCT threshold.

    “Importantly, there are not discontinuities in housing values, pre-existing trends of housing values, or median household incomes.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning

    There is at least 20 percent crowd out of LIHTC units within 1 km in radius, and this crowd out effect is overstated using OLS estimates.

    “Therefore, we find at least 20 percent crowd out of LIHTC units within 1 km in radius, and this crowd out effect is overstated using OLS estimates.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisenone given

    Annual rents on LIHTC units cannot exceed 30 percent of the relevant income limit.

    “Annual rents on these units cannot exceed 30 percent of the relevant income limit.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    The marked increased churn rate due to low income units is consistent with LIHTC projects inducing a downward shift in neighborhood desirability that is not fully reflected from the OLS results because developers select into gentrifying neighborhoods.

    “This marked increased churn rate due to low income units is consistent with LIHTC projects inducing a downward shift in neighborhood desirability that is not fully reflected from the OLS results because developers select into gentrifying neighborhoods.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Premisenone given

    A census tract counts as qualified if 50% of its households have incomes below 60% of AMGI, with the restriction that no more than 20% of the population of any metropolitan area may live in a qualified tract.

    “A census tract counts as qualified if 50% of its households have incomes below 60% of AMGI, with the restriction that no more than 20% of the population of any metropolitan area may live in a qualified tract.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Premisenone given

    Projects located in census tracts where at least 50 percent of households are eligible to rent an LIHTC unit are designated as Qualified Census Tracts (QCT) and receive thirty percent higher tax credits.

    “Projects located in census tracts where at least 50 percent of the households are eligible to rent an LIHTC unit are designated as Qualified Census Tracts (QCT) and receive thirty percent higher tax credits.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    As of 2001, 29 states gave extra points to projects proposed for tracts with qualified status as one of the location criteria.

    “As of 2001, 29 states gave extra points to projects proposed for tracts with qualified status as one of the location criteria.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductcitation

    For outcomes with statistically significant estimates, spatially corrected standard errors are reported using the GMM estimator described in Conley (2008).

    “For outcomes with statistically significant estimates, we also report spatially corrected standard errors using the GMM estimator described in Conley (2008).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Conley (2008)

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • May combine two claims.

    Premisereasoning

    Tracts' qualified status is assigned using decennial census data, and is thus only revised every 10 years.

    “Tracts’ qualified status is assigned using decennial census data, and is thus only revised every 10 years.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S4 (based).

    Study design and conductreasoning

    If sorting on unobservables happens using exactly the same process on each side of the threshold, the omitted variables on which the sorting occurs get differenced out by the regression discontinuity estimator.

    “If this is true, the omitted variables on which the sorting occurs get differenced out by the regression discontinuity estimator.”
    Quote in the paperNot a substantial claim
    • The authors state this as an assumption: Sorting on unobservables happens using exactly the same process on each side of the QCT threshold.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    The LIHTC program was established as part of the Tax Reform Act of 1986 to encourage the development of affordable rental housing for low-income households.

    “The LIHTC program was established as part of the Tax Reform Act of 1986 to encourage the development of affordable rental housing for low-income households.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Tax Reform Act of 1986

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • May combine two claims.

    Study design and conductown data

    Only population, total housing units, and the fraction of owner occupied housing units that were detached differ significantly across the qualified threshold, and the magnitudes of these differences are small.

    “Of the many variables examined, only population, total housing units, and the fraction of owner occupied housing units that were detached differ significantly across the qualified threshold and magnitudes of these differences are small.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Premisecitation

    As argued in Eriksen and Rosenthal (2007), the magnitude and even the sign of the external effects of LIHTC construction may differ as a function of initial neighborhood composition.

    “Furthermore, as argued in Eriksen and Rosenthal (2007), the magnitude and even the sign of the external effects of LIHTC construction may differ as a function of initial neighborhood composition.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Eriksen and Rosenthal (2007)

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    All reported standard errors in the ring-level first stage are clustered at the county level.

    “All reported standard errors are clustered at the county level.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S4 (based).

    Study design and conductnone given

    The estimated coefficient β1 is consistent provided the error term εi does not change discontinuously across the threshold ei = 0.5.

    “The estimated coefficient β^1 is consistent provided the error term εi does not change discontinuously across the threshold ei = 0.5.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The error term εi does not change discontinuously across the threshold ei = 0.5.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisenone given

    Despite the population restriction, almost all (96 percent) of metropolitan census tracts above the cutoff qualify as QCTs.

    “Despite this population restriction, almost all (96 percent) of metropolitan census tracts above the cutoff qualify.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Premisenone given

    Since the program's inception, over 95 percent of units in projects supported by the LIHTC program qualified as low income.

    “Since the program’s inception, over 95 percent of units in projects supported by the program qualified as low income.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S4 (based). May combine two claims.

    Study design and conductnone given

    Blocks are chosen because they provide approximately the same number of potential LIHTC development sites regardless of spatial location and are small enough to be precisely aggregated into rings.

    “We choose blocks because they provide approximately the same number of potential LIHTC development sites regardless of spatial location and they are small enough to be precisely aggregated into rings.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    While LIHTC units increase total new rental construction at a little less than one for one in low demand areas, the private market would have created more than 60 percent of the LIHTC new construction in gentrifying areas if the program did not exist.

    “These results indicate that while LIHTC units increase total new rental construction at a little less than one for one in low demand areas, the private market would have created more than 60 percent of the LIHTC new construction in gentrifying areas if the program did not exist.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Interpretationreasoning

    Declining median area incomes points to negative peer effects or stigma of living near LIHTC projects as potential explanations for these demand shifts.

    “Declining median area incomes points to negative peer effects or stigma of living near LIHTC projects as potential explanations for these demand shifts.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Study design and conductown data

    Information on project location is missing for 9.5 percent of projects and the number of units is missing for 4.9 percent of observations.

    “Most important for our analysis, information on project location is missing for 9.5 percent of projects and the number of units is missing for 4.9 percent of observations.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    Potential LIHTC projects must meet one of two criteria: either at least 20 percent of units must be occupied by tenants earning below 50 percent of AMGI or at least 40 percent of units must be occupied by tenants earning below 60 percent of AMGI.

    “Either at least 20 percent of the units must be occupied by tenants earning below 50 percent of the Area Median Gross Income (AMGI) or at least 40 percent of units must be occupied by tenants earning below 60 percent of the AMGI.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Study design and conductnone given

    The jurisdictions used for applications data host a large number of LIHTC projects, make applications data readily available, and include states both with and without preferences to place LIHTC projects in QCTs.

    “These jurisdictions host a large number of LIHTC projects, make applications data readily available, and include states both with and without preferences to place LIHTC projects in QCTs.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning

    The effect of the tax credit on LIHTC units is stronger in areas with the fastest growth in housing values during the eighties, suggesting that developers differentially select into gentrifying areas.

    “We find that this effect is stronger in areas with the fastest growth in housing values during the eighties, suggesting that developers differentially select into gentrifying areas.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Study design and conductown data

    With only two exceptions, the estimated discontinuities at the placebo thresholds are statistically insignificant, and in most cases they are closer to zero than the estimates using the true eligibility threshold.

    “With only two exceptions, the estimated discontinuities at the placebo thresholds are statistically insignificant, and in most cases they are closer to zero than the estimates using the true eligibility threshold.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S2 (checkable).

    Interpretationreasoning

    To disentangle the supply decisions of firms from the preferences of state housing authorities, it is worthwhile to know about projects that were rejected.

    “To disentangle the supply decisions of firms from the preferences of state housing authorities, it is worthwhile to know about projects that were rejected.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Study design and conductreasoning

    The empirical analysis always controls for county fixed effects to maintain continuity of baseline variables across the tax credit discontinuity.

    “In order to maintain continuity of baseline variables across the tax credit discontinuity, we always control for county fixed effects in the empirical analysis to follow.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Interpretationreasoning

    The difference between IV and OLS either suggests that sorting biases OLS estimates towards zero, or that the local estimates at the QCT threshold do not generalize to the broader sample.

    “This either suggests that sorting biases OLS estimates towards zero, or that we are obtaining local estimates at the QCT threshold that do not generalize to the broader sample.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Study design and conductreasoning

    Since tracts are often too large to assess treatment effects that may decay quickly with distance, the study examines the effect of geographically close LIHTC units on outcomes at the census block group level.

    “Since tracts are often too large to assess treatment effects that may decay quickly with distance, we examine the effect of geographically close LIHTC units on outcomes at the census block group level.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisenone given

    Each state is required to have a Qualified Allocation Plan (QAP) to determine whether applications for developments merit receiving the tax credit.

    “Each state is required to have a “Qualified Allocation Plan” (QAP) to determine whether applications for developments merit receiving the tax credit.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S4 (based).

    Study design and conductnone given

    The expected value of the covariates captured in Z should be equal in QCT and non-QCT tracts near the threshold if the identification strategy is valid.

    “The expected value of the covariates captured in the vector Z should be equal in QCT and non-QCT tracts near the threshold if this identification strategy is valid.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Premisecitation

    A standard strategy for evaluating the marginal willingness to pay for a local amenity is to estimate the response of housing values to exogenous shocks in the amenity.

    “A standard strategy for evaluating the marginal willingness to pay for a local amenity is to estimate the response of housing values to exogenous shocks in the amenity (Rosen, 1974 and Ekeland et al. 2004).”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites Rosen, 1974 source
    • cites Ekeland et al. 2004 source

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Interpretationreasoning

    The slight positive impact of LIHTC units on housing values even in gentrifying neighborhoods may reflect capitalization of the expected future amenity value of the LIHTC developments once they become market rate housing.

    “The slight positive impact of LIHTC units on housing values even in gentrifying neighborhoods may reflect capitalization of the expected future amenity value of the LIHTC developments once they become market rate housing.”
    Quote in the paperNot a substantial claim
64 of 64

  • The paper relies on this without support.

    LIHTC subsidized units are almost entirely populated by households below the 30th percentile of the income distribution.

    “Since LIHTC subsidized units are almost entirely populated by households below the 30th percentile of the income distribution, new developments may generate a decline in local amenities because they lead to an influx of lower than average income residents in most neighborhoods.”

  • The paper relies on this without support.

    Developers’ incentives to build low income versus market rate housing may differ markedly by local economic conditions.

    “Developers’ incentives to build low income versus market rate housing may differ markedly by local economic conditions.”

  • The paper relies on this without support.

    The probability a project is proposed in a particular tract, and whether this proposal is ultimately accepted, is likely to be related to some unobserved tract characteristics that also influence the change in the outcome.

    “The probability a project is proposed in a particular tract, and whether this proposal is ultimately accepted, is likely to be related to some unobserved tract characteristics that also influence the change in the outcome Δyi.”

  • The paper relies on this without support.

    The error term εi does not change discontinuously across the threshold ei = 0.5.

    “The estimated coefficient β^1 is consistent provided the error term εi does not change discontinuously across the threshold ei = 0.5.”

  • The authors state this as an assumption.

    There are no unobservables that influence an outcome of interest and are correlated with QCT status, conditional on controls including the interacted polynomial f(e).

    “The basic identifying assumption is that there are no unobservables that influence an outcome of interest and are correlated with QCT status, conditional on controls including the interacted polynomial f(e).”

  • The paper relies on this without support.

    Sorting on unobservables is likely to remain within groups of tracts on either side of the QCT threshold.

    “It is important to recognize that sorting on unobservables is likely to remain within groups of tracts on either side of the QCT threshold.”

  • The paper relies on this without support.

    Whether locating in a qualified tract or not, developers will still find it to their advantage to locate in gentrifying areas.

    “In particular, whether locating in a qualified tract or not, developers will still find it to their advantage to locate in gentrifying areas.”

  • The authors state this as an assumption.

    Sorting on unobservables happens using exactly the same process on each side of the QCT threshold.

    “For econometric identification, we need that such sorting on unobservables happens using exactly the same process on each side of the threshold.”

  • The authors state this as an assumption.

    The empirical model in (2) captures the data generating process for x at the census block level.

    “Assuming the empirical model in (2) captures the data generating process for x at the census block level, inclusion of the aggregated polynomial control function terms ensures that the instrument is orthogonal to the error term ε~g.”

  • The authors state this as an assumption.

    Missing information on location does not differ across the 50 percent eligibility cutoff.

    “Assuming that missing information on location does not differ across the 50 percent eligibility cutoff, missing data does not adversely affect our identification strategy, though it does add noise to the error term.”

  • The authors state this as an assumption.

    Unobserved tract characteristics that influence outcomes of interest do not vary discontinuously across the QCT eligibility threshold.

    “In order for the regression discontinuity strategy to yield consistent parameter estimates, unobserved tract characteristics that influence outcomes of interest must not vary discontinuously across the policy threshold.”

  • The paper relies on this without support.

    In 2003 the QCT criterion was expanded to allow tracts with a poverty rate exceeding 25 percent to qualify as QCTs.

    “In 2003 the QCT criterion was expanded also to allow tracts with a poverty rate exceeding 25 percent to qualify as QCTs.”

  • The authors state this as an assumption.

    If all profitable projects are proposed, then the response of proposed units to QCT designation yields the supply response of LIHTC units to higher anticipated tax credits.

    “If all profitable projects are proposed, then the response of proposed units to the QCT designation yields the supply response of LIHTC units to higher anticipated tax credits.”
13 of 13

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • Table 4 column range excludes a column discussed

    Minor

    The text says the specifications are in columns 2-4, but then describes a specification in column 5.

  • Ekeland citation year differs from reference year

    Minor

    The text cites Ekeland et al. as a 2004 work, while the reference list gives 1994 for the corresponding work.

2 of 2

Cites

Flagged first, then by citations

Research works printed but not linked: 1

  1. Schwartz, 2006, The external effects of subsidized housing investment, Regional Science and Urban Economics, 36, 697

Cited by

Sorted by citations

Related

Authors
Baum-Snow, N., & Marion, J.
Affiliations
  • Brown University (US)
  • University of California, Santa Cruz (US)
Corresponding author
Baum-Snow, Nathaniel
Venue
Journal of Public Economics
Publisher
Elsevier BV
Published
2009-06
Volume
93
Issue
5-6
Pages
654-666
ISSN
0047-2727 (ISSN-L)
DOI
10.1016/j.jpubeco.2009.01.001
PMC
PMC3824966
OpenAlex
W2048157984
Cited by
264
References
17
Citations by year
2026: 20 · 2025: 25 · 2024: 14 · 2023: 11 · 2022: 19 · 2021: 13 · 2020: 30 · 2019: 16 · 2018: 16 · 2017: 10 · 2016: 9 · 2015: 28 · 2014: 13 · 2013: 10 · 2012: 11
FWCI
19.27
Type
journal article
Language
en
Open access
Licence
www.elsevier.com/tdm/userlicense/1.0/ (tdm)
Locations
  • Journal of Public Economics (publishedVersion)
  • PubMed (publishedVersion)
  • Europe PMC (PubMed Central) (acceptedVersion)
  • PubMed Central (acceptedVersion)
  • RePEc: Research Papers in Economics (submittedVersion)
Funders
Eunice Kennedy Shriver National Institute of Child Health and Human Development (R24HD041020)
Topics
  • Social Sciences › Economics, Econometrics and Finance › Economics and Econometrics › Housing Market and Economics
  • Social Sciences › Sociology and Political Science › Urban, Neighborhood, and Segregation Studies
  • Social Sciences › Urban Studies › Urban Planning and Governance
Keywords
Low-Income Housing Tax Credit, neighborhood effects, gentrification, property values, income distribution, displacement, housing policy, tax credits, census tracts

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-11-04.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-11-04.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review