THREE ESSAYS ON HOUSING POLICY AND INEQUALITY. A dissertation presented by. Thomas PlaHovinsak. to The Department of Economics

Transcription

1 THREE ESSAYS ON HOUSING POLICY AND INEQUALITY A dissertation presented by Thomas PlaHovinsak to The Department of Economics In partial fulfillment of the requirements for the degree of Doctor of Philosophy in the field of Applied Economics Northeastern University Boston, Massachusetts July,

2 THREE ESSAYS ON HOUSING POLICY AND INEQUALITY by Thomas PlaHovinsak ABSTRACT OF DISSERTATION Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Applied Economics in the College of Social Sciences and Humanities of Northeastern University July,

3 The first chapter examines issues of vertical inequity in property assessment across Massachusetts. Several previous studies claim to identify bias in the assessment of housing value for the purposes of property taxes. This chapter highlights some problems with the methodologies used in many of those studies and provides an alternative framework for finding minimum values for assessor error and bias. Using data from the Massachusetts Office of Geographic Information, we build off previous studies in three ways. First, we briefly explain why the errors-in-variables problem can cause biased estimates of vertical inequity in property assessment and how previous solutions to this problem are based on misleading assumptions. Second, we show that a method based on hedonic price estimates using property-level observables can provide a lower bound for the extent of assessor error and bias for Massachusetts towns. Third, we explore if there are differences in vertical inequity across towns in Massachusetts. Our results show that more than 23.3 percent of the variance in the difference between assessment and sale price across Massachusetts is due to assessor error, and that several property-level and town-level features can explain assessor bias. Furthermore, high-value properties across the state are the ones most likely to be under-assessed regardless of whether or not they are located in a high-income town. The second chapter studies the aftermath of the Mount Laurel decisions and the role of the Council on Affordable Housing. In the wake of the Mount Laurel decision in the mid-1980s, the Council on Affordable Housing (COAH) was created by the NJ state legislature to implement new affordable housing requirements across the state. If NJ municipalities volunteered to fall under COAH's jurisdiction, they would agree to build their affordable housing obligation. In return, the municipality would receive legal protection from exclusionary zoning lawsuits and have the ability to engage in a Regional Contribution Agreement (RCA), a process in which a municipality could pay another municipality to build up to 50 percent of the paying municipality s affordable housing 3

4 obligation. Using data from the New Jersey Department of Community Affairs, reports generated by COAH, as well as decennial census data, I investigate three questions of interest: 1) What types of municipalities volunteered to come under COAH s jurisdiction? 2) Is there a pattern to compliance rates when examining the construction of new affordable housing? 3) What types of municipalities engaged in an RCA, and is there a pattern of rich municipalities sending their affordable housing allotment to poor municipalities? I find that those municipalities with the greatest perceived threat of potential litigation were the ones most likely to join COAH as well as fulfill at least some of their affordable housing obligation, although compliance rates were generally low throughout the time period examined. I also find a clear pattern of higher-income municipalities sending affordable housing units to lower-income municipalities, preventing economic integration within high-income municipalities and undermining the original intent of the Mount Laurel decision. The third chapter examines the results of a pilot study with Habitat for Humanity. Homeownership has remained an important aspect of U.S. policy and popular culture for decades. While some studies have attempted to prove the existence of benefits associated with homeownership, often called the homeownership effect, few studies have been able to separate benefits stemming from housing itself as opposed to neighborhood effects, and few studies have devoted their attention to homeownership-focused programs. In a pilot study using survey data collected from applicants to the U.S.-based housing charity Habitat for Humanity, I employ a quasi-experimental design to compare perceived changes in the lives of those who were selected into the Habitat for Humanity housing program to those who applied for housing but were denied. In addition, since the recipients of Habitat houses remain in the same general area as their original residence, I am able to focus on the three joint benefits of the Habitat for Humanity program: 4

5 becoming a homeowner, upgrading the quality of one's residence, and receiving a positive wealth transfer. I find that, in comparison to denied applicants, Habitat homeowners report more positive changes in their overall life, including their economic situation, their children s education, and their level of community engagement. I also find that the participating Habitat for Humanity affiliates were more likely to select those who were married and those with a high school level into the housing program. Based on these findings, I believe that more research into homeownershipfocused programs is warranted. 5

6 Dedication This dissertation is dedicated to the following people for their constant support and encouragement while I completed this research. First, to my wife, Ashley, whose constant love and reassurance made it possible to complete this work, and who helped with a substantial amount of the data collection for the third chapter. Second, to my mother, Patricia, who was constantly supportive of my efforts in this program. And third, to my father, the man who paved the way, and the original Dr. Thomas PlaHovinsak. 6

7 Acknowledgements I would like to thank several people and organizations for their assistance in the various stages of research this dissertation required. I would first like to thank my committee for their time and feedback, and also for challenging me to do the best work possible. A special thanks to my chair and advisor Bill Dickens, whose influence on both my work and enthusiasm for research has been profound. I would also like to thank my classmates that I had the great pleasure of getting to know better during my time in this program, and who were a constant source of ideas, feedback, and support. I would like to thank several members of the Council on Affordable Housing for their assistance in collecting and understanding the data used in my second article. In particular, I would like to thank Keith Henderson, Director of Planning and Policy, for his assistance in collecting the RCA data I needed, as well as John Lego, who maintains the New Jersey Guide to Affordable Housing and was immensely helpful in explaining for the guide and aspects of COAH s data collection process. I would also like to thank the New Jersey State Library for helping me locate all of the COAH reports in existence, which were a vital part of the data used in my second chapter. Lastly, I would like to thank Habitat for Humanity for their permission to study their past applicants, which was necessary to collect the data used in my third chapter. In particular, I would like to thank Pat Decker, Director of U.S. Operations, for expressing interest in the study and permitting us to conduct this research. I would also like to thank all of the employees and the directors of each of the participating Habitat affiliates in the study. Though I unfortunately cannot list their names due to the confidentiality of the study participants, their cooperation, enthusiasm, and assistance in collecting the data required made this chapter possible. I would also like to thank Thomas Spina for a generous donation to help cover the cost of the study. 7

8 Table of Contents Abstract of Dissertation... 2 Dedication.. 6 Acknowledgements 7 Table of Contents 8 Chapter 1 1. Introduction Measuring Assessor Error and Regressivity in Assessment a. The Errors-in-Variables Problem b. Previous Vertical Inequity Models 14 c. Our Methodology Data and Descriptive Statistics a. Parcel-level Data 20 b. Other Data Econometric Results a. Regressivity in Massachusetts towns. 27 b. Regressivity within-town Conclusion References Appendix 39 8

9 Chapter 2 1. A Brief History of the Mount Laurel Decision Previous Studies Data Joining the Council on Affordable Housing (COAH) COAH s Compliance Issues Regional Contribution Agreements Conclusion References Appendix 84 Chapter 3 1. Introduction Previous Studies Habitat for Humanity Data Collection and Methodology Examining the Habitat for Humanity Selection Process Examining the Perceived Benefits of the Habitat for Humanity Program a. House and Neighborhood Perceptions b. Perceived Quality of Life Changes Conclusion References Appendix 121 9

10 Chapter 1 Measuring Vertical Inequity in Property Assessment: A New Approach Using Data from Massachusetts Co-authored with Gustavo Vicentini 1. Introduction The distribution of the burden of property taxes has been researched extensively over the past several decades, with many studies exploring issues related to vertical inequity in property assessment. Particular attention has been given to the presence of what the literature calls regressivity, which is a particular type of vertical inequity. Regressivity is defined by both the literature and the International Association of Assessing Officers (IAAO) as a pattern of overassessment of low-value properties in contrast to high-value properties, typically measured by comparing the assessed value of a property to its most recent sale price. 1 Because the assessed value is the basis for calculating the tax levy of a property, regressivity shifts the burden of the tax across property owners. The focus of previous studies has concentrated on two main areas: methodology for measuring regressivity; and identifying the determinants of regressivity. The studies in the first area have proposed new methodological frameworks for measuring regressivity within an assessing jurisdiction as well as testing the validity of the competing methodologies (e.g., Paglin and Fogarty, 1972, Kochin and Parks, 1984, Clapp, 1990, Goolsby, 1997, Fairbanks et al., 2013, Birch and Sunderman, 2014). The second area of research has focused on investigating the causes of regressivity by comparing inter-jurisdictional differences (e.g., Smith et al., 2003, Ross, 2012). Particularly in the first area, a key concern has been how to solve the errors-in-variables problem 1 Progressivity, on the other hand, is a pattern of over-assessment of high-value properties in contrast to low-value properties. 10

11 when estimating the amount of regressivity via regression techniques. Since both assessed value and sale price are likely noisy measures of the market value of a property, authors have had to decide how to best address this issue so as to avoid biased estimates of regressivity. The goal of this paper is to briefly illustrate why previous methods used to measure vertical inequity are flawed, as well as to use a new approach to find minimum values for both assessor error and assessor bias. Specifically, we examine the potential indicators of vertical inequity by regressing the log difference of assessment and sale price on property and neighborhood characteristics. The features of a property (e.g., size, age, proximity to public goods) and of its neighborhood (e.g., proportion of vacant housing) should be determinants of both the assessed value and sale price of a property, but in principle they should be uncorrelated with the difference between these two if assessor error and market error are random. By using this framework, we are able to estimate a lower-bound of both assessor error from the zero-centered R 2 of this regression, and of assessor bias using a hedonic pricing model. We apply this approach to property-level data for single-family residences from 2008 to 2014 for the state of Massachusetts. We find that a number of property features explain assessor bias when controlling for town differences, and that assessor error explains approximately 23.3 percent of the variance in the difference between assessment and sale price. When examining individual towns, 2 we find that properties in highincome towns are much more likely to be under-assessed than those in low-income towns, with the exception of lowest-value properties which are very close to accurately assessed. We also see that the highest-value properties are the most likely to be under-assessed across all towns regardless of the owner s income level. 2 We note that all references to a town or towns in this paper refer to all types of municipalities within the state of Massachusetts, including both cities and townships. 11

12 The rest of this paper proceeds as follows. Section 2 describes our proposed methodology and compares it to previous studies. Section 3 describes the data and provides descriptive statistics. Section 4 contains the results, and Section 5 concludes. 2. Measuring Assessor Error and Regressivity in Assessment 2.a. The Errors-in-Variables Problem In many jurisdictions, the assessed value of a property is supposed to reflect 100 percent of its market value. The IAAO defines market value as [t]he most probable price which a property should bring in a competitive and open market under all conditions requisite to a fair sale, the buyer and seller each acting prudently and knowledgeably, and assuming the price is not affected by undue stimulus, 3 and the Massachusetts Bureau of Local Assessment ensures that each municipality s properties are assessed at fair market value at least once every three years. 4 As described in Sirmans et al. (2008), many studies measure under or over-assessment by comparing the difference between the assessed value and sale price of a property. For example, if a property sold for $500,000 and its most recent (prior to the sale) assessment was $475,000, then that property is assumed to have been under-assessed by $25,000. In addition, one area of discussion has been whether or not assessed value or sale price is a better reflection of a property s true market value, and hence whether assessed value or sale price should be on the right hand side of a regression. It is reasonable to believe, however, that there will be some degree of error in assessment. This could result from either willful bias on the part of the assessor, the assessor not correctly taking 3 Source: International Association of Assessing Officers (IAAO), Standard on Mass Appraisal of Real Property, January 2012, p Source: Graziano, Joanne. Bureau of Local Assessment. Mass.gov. (accessed May 16, 2016). 12

13 certain observable features of a property into account, or random error in assessment. In spite of these potential sources of error, the assessor s job is to approximate the true, underlying value of a property, analogous to the idea of the fair market price. However, sale price is also likely a noisy measure of fair market value. Sale price may deviate from fair market value due to transient market conditions at the time of sale, differences in bargaining power and discount rate between parties, how eager the seller is to sell the property, and other factors. Hence, there is a potential measurement error issue in both assessment (A) and sale price (P), as outlined in equations (1) and (2). ln(a i ) = ln(v i ) + ε i (1) ln(p i ) = ln(v i ) + μ i (2) Here, V represents the expected value, or the fair market value of property i that the assessor is theoretically trying to determine. We define V as the average price one would obtain if an average seller were able to repeatedly sell the same house on the open market. ε and µ are both assumed to be mean-zero error terms and independent of V. If there were random measurement error only in assessment or only in price, then the variable measured with error could be regressed on the variable that is measured without error to obtain an unbiased estimate of the direction and extent of inequity. Any values other than zero for the constant and zero for the coefficient of the variable measured without error would indicate inequity. But, since both A and P measure V with error, putting either on the right hand side of the regression will lead to biased estimates of inequity. Such a regression could potentially show regressivity where there is none. 13

14 2.b. Previous Vertical Inequity Models Sirmans et al (2008) provides an overview of the several methods that past studies have used to test for the presence of vertical inequity in an area as well as contend with the errors-in-variables problem. Building off this overview, Fairbanks et al. (2013) use a Monte Carlo simulation to test the accuracy of these models and makes several recommendations following their results. First Fairbanks et al. (2013) suggest that authors attempting to measure vertical inequity try to select areas that appear to have no predisposition to regressivity or progressivity. They also recommend using a graph of the assessment-to-price ratio and sale price as a first pass in determining the existence of vertical inequity. We believe, however, that this approach could lead to incorrect a priori decisions about the presence of vertical inequity. In other words, even if assessment is not actually regressive, because of measurement error in P the figure would display a pattern of regressivity. With the errors-in-variables problem present it is difficult to determine an area s vertical inequity without some form of proper estimation. More specifically, OLS will likely underestimate the slope coefficient, therefore potentially indicating regressivity when there may be none. This is the classical errors-in-variables problem in econometrics, and this downward bias of the slope coefficient is typically known as attenuation bias in the literature. 5 In terms of econometric models, Fairbanks et al. (2013) recommend two models in particular as a result of their simulations. First, they recommend using the model from Clapp (1990), which 5 See, e.g., Greene (2012, chapter 8). We note that measurement error in the dependent variable in a regression does not cause attenuation bias in the OLS estimator, although it does reduce the efficiency of the estimator. It is only measurement error in the independent variable that causes the bias. 14

15 uses instrumental variables to estimate vertical inequity. Clapp specifies the log of sale price as a function of the log of assessment. 6 ln(p i ) = α 0 + α 1 ln(a i ) + ε i (3) If regressivity is present, then the slope coefficient in this model is greater than one. Due to assessor error, ln(ai) is an imperfect measure of the log of market value, and therefore the OLS estimator would be biased downwards towards progressivity. Clapp proposes using a grouping instrumental variable for ln(ai) that can take three possible values, depending on property i s ranking within the distribution of assessments and sale prices: +1 if property i s assessment falls in the top one-third of assessments and its sale price falls in the top one-third of sale prices; 1 if property i s assessment falls in the bottom one-third of assessments and its sale price falls in the bottom one-third of sale prices; 0 otherwise. Clapp also recommends including some sort of time control in the model. This instrument is likely relevant for market value, because the higher the market value of a property the higher its ranking and therefore the larger the value of the instrument. However, the exogeneity of this instrument relies on the assumption that the ranking observed in assessment and sale price is the same as the ranking that one would observe in market value. This is plausible if measurement error is small enough such that the rankings across assessment, sale price, and market value are all the same, as conceded by Clapp (1990) in his paper. Clapp justifies the use of this method in his paper by claiming that this would still be an improvement over the other leading 6 Clapp (1990) justifies specifying his model with assessed value on the right-hand side by citing that in many typical sales-ratio studies, assessed value is determined before sale price, and hence a part of assessed value is negatively capitalized into sale price. See Edelstein (1979) for a full explanation of this justification. 15

16 models at the time. 7 However, if measurement errors are large enough, a property that is (for example) in the middle one-third of market value might be incorrectly allocated in the top onethird of assessment and sale price if its measurement errors are positive. This grouping misallocation would create a correlation between the instrument and the measurement errors, therefore jeopardizing the exogeneity of the instrument. In essence, measurement error would still be present on the right-hand side of Equation (3), and the attenuation bias stemming from the errors-in-variables problem would still be present. The second model that Fairbanks et al. (2013) recommend is a dummy variables regression model from Sunderman et al. (1990) for non-linear estimation of vertical inequity. 8 Their model reverts to having assessed value as the dependent variable, and searches for break-points in the data and measures slope coefficients for each segment of the data with the following equation: Ai = α00 + α10 Pi + α01 LOWi + α02 HIGHi + α11 LOW*Pi + α12 HIGH*Pi (4) where LOW and HIGH are dummy variables indicating if the sale price of property i is less than or greater than the first or second break-point in the data, respectively, and LOW*P and HIGH*P are interaction terms between the dummy variables and the sale price. Other authors have allowed for more variation by using the percentile of sale price on the right hand side. However, although the spline model does provide for a more localized and flexible measure of vertical inequity, it is silent about the potential error in measurement of P. It is not surprising, therefore, that Fairbanks 7 Specifically, Clapp writes that he believes that his model will be less biased than the Paglin and Fogarty (1972) and the Kochin and Parks (1984) models, which were the leading models for measuring vertical inequity prior to Clapp s model. See Sirmans at al. (2008) for a detailed explanation of these models. 8 We note that the Sunderman et al. (1990) model as specified in Equation (4) is labeled as a spline model by both Sunderman et al. and Fairbanks at al., though it is not strictly a spline model as defined by the econometric literature. See Greene (2012, chapter 7) for a detailed explanation of spline models. 16

17 shows that the spline model has a bias towards regressivity as this model does nothing to address the errors-in-variables problem. It is clear, then, that measurement error is still a worrying problem in vertical inequity analysis, as no method to-date has fully addressed how to purge vertical inequity regression analysis of the errors-in-variables problem. 2.c. Our Methodology To examine potential indicators of assessor bias, we will utilize a model with assessor error ε and market error μ confined to the left-hand side of our regression. We construct a new dependent variable that is the log of the ratio of assessed value to sale price, as shown in Equation (5). We choose to examine the log difference in order to examine the existence of vertical inequity in percentage terms. We estimate ln(a i /P i ) = α 0 + β 1 (PROP i ) + β 2 (X i) + γ i (5) where PROP is a vector of property observables, X is a vector of controls including the month and year in which property i was sold, as well as the town in which property i was located, β 1 and β 2 are conforming vectors of coefficients, α is the constant term, and γ is a stochastic error term. By confining the measurement error in assessment and price to the left-hand side of our regression we eliminate the attenuation bias in our parameters caused by the errors-in-variables problem. 9 Note that what is on the right hand side of Equation (5) will be zero if the assessor exactly predicts the sale price. One can think of ln(a/p) as the percentage difference between the 9 For example, see Wooldridge for a detailed discussion on running OLS regressions with measurement error and the difference between measurement error being on the left vs. right-hand side of a regression. 17

18 assessment and the sale price. If we can predict any of this difference then the assessor must be making an error as assessors have access to all the same information we do and could have used it to construct their estimates. Hence, since the variance of ln(a/p) is the variance in the difference between assessment and sale price, the zero-centered R 2 of this regression may also be interpreted as the fraction of the variance in our dependent variable ln(a/p) that is due to assessor error. However, since there are many more characteristics of a property than what we observe, adding these variables to our regression would only increase the zero centered R-squared and thus our estimate is, asymptotically, a lower bound. Now let us consider what our parameter estimates tell us about the nature of assessor errors. If assessor error and market error are mean zero, then features of a property should not be able to predict differences in ln(a/p), and no coefficients in Equation (5) should be statistically significant. Statistically significant β 1 coefficients would indicate that there are observable features of properties that are not being appropriately accounted for by assessors. For instance, larger properties might be under-assessed if the assessor under-compensates for the size of a house. It could also be, for instance, that assessors intentionally under-assess larger houses to give them a tax break compared to what they would have to pay as a result of accurate assessment. This would imply a systemic error, which is our definition of bias. We would say here that assessors are biased in favor of more expensive homes. 10 Similarly, a non-zero constant term would indicate a constant percentage bias across all properties in a town. Therefore, regressing ln(a/p) on features of a property should not yield any statistically significant coefficients if assessments are unbiased. As such, the zero-centered F-test for our entire regression is a test for the presence of bias in assessment based on our property observables. 10 We note that another possible reason for positive slope coefficients would be if housing prices are rising in the time between assessment and sale, a possibility we explore later in Section 4. 18

19 We note that we do not interpret significant coefficients as indicating systematic market error because market error, by definition, cannot be systematic. Such noise in price may stem from market conditions at the time of sale, differences in bargaining power and discount rates between parties, how eager the seller is to sell the property, and other factors. However, there is little reason to believe that larger properties would receive offers that are consistently above (or consistently below) their market value compared to smaller properties. Consider the second-order hedonic pricing model below: ln(p i ) = α 0 + β 1 (PROP i ) + β 2 (PROP i 2 ) + β 3 (X i) + γ i (6) where our variables are defined the same as in Equation (5). Since price is the expected value of a property given its characteristics, if prices systematically vary with some characteristic in our sample then this cannot be market error. It is exactly this systematic variation that assessors should be taking into account, hence any variation in ln(a/p) must be due to assessor error. 11 In addition, since all of the measurement error in price is confined to the right-hand side of Equation (6), this allows us to generate predicted prices without measurement error introduced by μ. As a result, we may then regress the ln(a/p) that we obtain from Equation (5) on the predicted log price from Equation (6) for each property to determine if there is any bias in assessment: ln(a i /P i ) = α 0 + β 1 (lnp i) + γ i (7) 11 There is still the possibility of a small part of the explained variance stemming from changes in sales prices within each year, though in a later footnote we demonstrate that this is not a concern. 19

20 where P is now exogenous. Hence, any explained variance of ln(a/p) in both Equation (5) and Equation (7) must be due to assessor error, and these equations can provide us with an estimate of assessor bias stemming from observables. The results of Equation (5) are shown Section 4.a. The regression results of equations (6) and (7) may be found in the appendix, while a graph of the results of Equation (7) is shown in Section 4.a. 3. Data and Descriptive Statistics 3.a. Parcel-level Data To estimate equations (5), (6), and (7), we use assessors parcel-level data obtained from the state of Massachusetts Office of Geographic Information ( MassGIS ). 12 For each property in Massachusetts, the data contains information on the address, geographic coordinates, assessed value, most recent sale price and date, and basic property features such as residential area, number of bedrooms, building age, and building style. The data ranges from calendar year 2008 to 2014, with different towns appearing in the data in different years. The data is available for 350 of the 351 towns in Massachusetts, with Boston being excluded. 13 From this original MassGIS dataset, we applied the following cleaning steps to obtain our final working dataset. First, we dropped towns for which data on property features were not available. Next, we kept only the properties that are single-family residences, which is the focus of this paper. Next, we kept only the properties for which a sale took place soon after the assessment. For 12 This MassGIS parcel-level dataset was downloaded from the following source and website: MassGIS Data Level 3 Assessors Parcel Mapping, Massachusetts Office of Geographic Information, available at < 13 We again note that towns in this paper refers to municipalties of any type within the state of Massachusetts. 20

21 example, among the properties appearing in the data with a 2012 fiscal year assessment, we restrict attention to only those that were sold during calendar year 2011, because a 2012 fiscal year assessment in Massachusetts means that the assessment was conducted in January of Therefore, we focus attention on properties that were sold within a short window of time (maximum 12 months) after their date of assessment. Figure 1 illustrates an example of this timeline. Figure 1 Next, in order to isolate only the valid (i.e., arms-length) sales, we matched the MassGIS parcel-level data with a dataset from the Massachusetts Department of Revenue (DOR) that contains a list of all the valid sales that took place in Massachusetts between 2008 and In Massachusetts, by state law a property s assessment should reflect the market value of the property as of the January 1 date that immediately precedes the fiscal year of the assessment. Fiscal years in Massachusetts run from July 1 to June 30. For example, fiscal year 2012 started in July 1, 2011, and ended in June 30, Therefore, a property s assessment for fiscal year 2012 would have been done in early calendar year 2011 and would have attempted to reflect the market value of the property as of January 1, (Source: Assessment Administration: Law, Procedures and Valuation Chapter 2 Mass Appraisal, Massachusetts Division of Local Services, accessed January 18, 2016, available at: < 15 This DOR valid sales dataset was downloaded from the following source and website: LA3 Parcel Search, Massachusetts Division of Local Services Getaway, available at < 21

22 This enabled us to eliminate invalid sales such as within-family transfers, sales resulting from a divorce settlement, and foreclosures. We also dropped any properties with a sale price of less than $10,000 in a further effort to prevent issues of non-valid sales affecting our results. 16 Next, we kept only properties that are at least two years old, in order to avoid cases of property tear-downs and reconstruction, and cases of new construction taking place on parcels previously used for a different purpose. Finally, we kept only towns with at least 20 valid sales within a year. After the application of all these filters, the final dataset contains valid sales for 25,971 single-family residences across 177 towns, ranging from calendar years 2008 through Figure 2 shows the geographic distribution of the Massachusetts towns that are included in our final cleaned sample, illustrating that we are able to capture data across much of the state, though towns are more concentrated in the eastern half of the state. This is mainly due to dropping a number of towns in western Massachusetts from our sample for having less than twenty valid sales. 16 We also tried alternative specifications of dropping properties of less than $50,000 and $75,000, but the statistical and substantive results remained unaltered. 22

23 Figure 2 Table 1 (found on page 27) lists descriptive statistics for our cleaned sample of valid sales. 17 Given that our original (un-cleaned) dataset provides the assessment value for the stock of all single-family properties in the state, whether they experienced a valid sale or not, we are able to check how our cleaned sample of valid sales compares to the overall stock of single-family properties. The average price in our sample is greater than the average assessed value. A t-test for difference in means shows a clear disparity with a test p-value < While properties in our cleaned sample have a mean assessed value of $411,674, the mean in the overall stock is $314, Since it appeared odd that some single-family residences would be comprised of a single room, we also ran alternative specifications omitting properties with one reported room. However, this again did not change the statistical or substantive results of our analysis. 23

24 This finding is not surprising if the owners of properties that have an assessed value greater than its perceived value are the most likely to sell. However, this just strengthens our claim to be providing a lower-bound estimate of the importance of assessor error. Since our sample is truncated compared to the population of houses in Massachusetts, and we are seeing an over-representation of properties that are accurately or over-assessed (since they have higher-than-average assessed values), this would likely bias β 1 in Equation (7) towards zero, leading us to underestimate the systemic assessor bias. This is not a strict arithmetic lower bound in the same manner as is the zero-centered R 2 from Equation (5); however, this indicates that we may safely interpret our findings in Equation (7) as conservative estimates of overall assessor bias. 3.b. Other Data We also use census block and block-group level data from the 2010 Census in order to obtain neighborhood characteristics for each single-family residence in our sample. Specifically, each single-family residence from our cleaned sample was matched to its census block and/or blockgroup to obtain socio-economic data for its surrounding area. We first attempted to match each residence to demographic data at the block level. When data were not available at the block level, we matched the residence to socio-economic data from its respective block-group. At the block level we have data on population count, percentage of the population that is black, median age of population, and the percent of housing units that are vacant. At the block-group level we have data on median income. In addition to the parcel-level data described above, MassGIS has additional datasets containing the current location (as of 2014) of various public goods across the state of 24

25 Massachusetts. The data contains the location of police stations, fire stations, hospitals, libraries, schools, town halls, and train stations. We then calculated the distance of each single-family residence in our cleaned dataset to its closest police station, its closest train station, and so on. However, most of these variables were not statistically significant in Equation (5), with only two being significant at the α =.1 level and our R2 remaining unchanged when they were added to the model. Hence, they were omitted from the final analysis. 25

26 Table 1 Descriptive Statistics Single-Family Residences Mean Std. Dev. Skewness Min 25th Percentile 50th Percentile 75th Percentile Max Assessed Value (A) $411,674 $309, $43,900 $242,400 $330,200 $474,500 $7,429,500 Sale Price(P) $435,352 $335, $45,000 $249,000 $347,900 $509,500 $6,800,000 Log (A/P) House Age (in years) Lot Size (in sq. ft.) 29, , ,712 15, , ,090,880 Residential Area (in sq. ft.) 1, ,867 Number of Rooms Observations Source: MassGIS Data - Level 3 Assessors' Parcel Mapping 1

27 4. Econometric Results 4.a. Regressivity in Massachusettes towns Table 2 presents the results a first and second-order version of Equation (5), regressing ln(a/p) on property observables. As seen in the table, several factors explain the variance in the difference between assessment and sale price, indicating that there is bias in assessment. In Model (1), both lot size and the number of rooms can predict changes in ln(a/p), as well as the black share of the population of a census block. However, a higher-order model might be more appropriate for capturing differences in mis-assessment. In Model (2), we can see that several variables have nonlinear impacts on the log difference in assessment. Interestingly, both the overall size of the property (lot size) and the size of the livable area (residential area) are both statistically significant, indicating that these are appropriately evaluated separately by assessors, though still in a biased manner. Also, house age and residential area both have a U-shaped influence on ln(a/p). An F-test confirms that the second-order terms introduced in Model (2) are jointly statistically significant. 18 Nonetheless, we are cautious in our interpretation of these variables without consulting our hedonic price estimates (see Appendix, Section A for a table of those results). 18 The cluster-robust F-stat generated using the centered R 2 is 7.23 with a p-value of <

28 Depedent Variable: Log(A/P) (1) (2) Coefficient Std. Err. Coefficient St. Err. House Age ( ) *** ( ) Lot Size (in 1000s, sq. ft.) * ( ) *** ( ) Residential Area (in 1000s, sq. feet) ( ) *** ( ) Number of Rooms *** ( ) ( ) Med hhd inc in Block Group (in 1000s) ( ) ** ( ) Population of Block (in 1000s) ( ) ( ) Median age of Block Population ( ) ( ) Black Share of Block Population *** ( ) *** ( ) Share of Houses Vacant in Block ( ) ** ( ) [House Age] *** ( ) [Lot Size (in 1000s, sq. ft.)] ** (4.60e-08) [Residential Area (in 1000s, sq. feet)] *** ( ) [Number of Rooms] ( ) [Med hhd inc in Block Group (in 1000s)] *** ( ) [Population of Block (in 1000s)] ( ) [Median age of Block Population] ( ) [Black Share of Block Population] ( ) [Share of Houses Vacant in Block] *** ( ) Constant (0.0154) (0.0236) Observations Zero-centered R-squared Zero-centered F-stat ⱡ Source: MassGIS Data Level 3 Assessors Parcel Mapping Census Data from 2010 Census Standard errors clustered by town Month, year, and town dummies included * p<0.10 **p<0.05 ***p<0.01, two-tailed test Table 2 ⱡ Calculated using robust standard errors due to cluster restrictions 28

29 The results of our hedonic pricing model show that house age has a non-linear, U-shaped influence on the price. This means that Model (2) could be capturing the diminishing influence of increased house age on the price of the property, and so focusing on second-order turning points in this case could be potentially mis-leading. However, residential area has a positive first derivative and positive second derivative based on our hedonic pricing model, so in this case we are more certain that a second-order term is appropriate to capture the impact of residential area on mis-assessment. Specifically, a property with a residential area of greater than 3,355 square feet (much larger than sample average of 1,983 square feet) is predicted to have a positive relationship with ln(a/p). So, while it is clear from an F-test that mis-assessment and property observables have a non-linear relationship, it is difficult to definitively say that a second-order model best captures the variance of mis-assessment (as opposed to a higher-order model). As for neighborhood characteristics, the median income of the block, as well as the share of the black population and share of houses that are vacant are all statistically significant. The signs of the first and second derivative of median income seem to indicate that higher-income residents could be enjoying more of a discount on their property tax bill compared to lower-income residents, although those living in the highest income areas might be getting less of a discount. Again, however, we are cautious about interpreting a specific turn around point based on these findings below considering that median income also has a negative first derivate and a positive second derivative in the results of our hedonic pricing model. We further explore the effect of area income levels later in the paper. We also see from our results that there could be some level of racial discrimination if areas with more black residents or houses owned by black residents are being over-assessed more often than those of white residents, controlling for median income. Harris (2004) had similar findings when he examined home sales in twenty-one neighborhoods in 29

30 New Haven, Connecticut, from 2000 to 2001 and found that minority-majority neighborhoods were over-assessed on average compared to other neighborhoods. Lastly, as mentioned previously, the zero-centered R 2 of our regression is an asymptotic lower bound on the fraction of the variance in the difference between assessment and sale price, or ln(a/p), that is due to assessor error. As seen in Table 2, the zero-centered R 2 for Model (2) is.233, so approximately 23.3 percent of the variance in ln(a/p) can be explained by our model, and hence that is the asymptotic lower-bound on the variance in this difference due to assessor error An overall F-test (also zero-centered) confirms with a p-value of <0.001 that our model is statistically significant, indicating that we can indeed predict error in assessment based on our observables, and that assessors exhibit some degree of bias in assessment. We also find that the town dummies are statistically significant based on an F-test with a p-value of < Hence, at least some of the variation in mis-assessment can be explained by town-specific fixed effects. Interestingly, the constant term in both models is not statistically different from zero, indicating that there is no evidence of systematic, constant assessor error across towns. It seems clear, then, that basic features of a property such as size and age play an important role in mis-assessment. Older and larger houses are much more likely to be under-assessed than 19 For reference, the centered R 2 for Model (1) is.108, and for Model (2) is We are also interested in the possibility that changes in housing prices could affect our measure of regressivity. Since our sample consists of houses that were assessed in January of a given year, and then sold at point during that corresponding calendar year, it is possible that house prices could change during that period. According to data at the Federal Housing Finance Agency ( housing prices in Massachusetts rose on average about one percent within a given calendar year. To examine if this had an impact on our estimation, we ran Model (2) from Table 2 while using the month variable as a linear time trend, and found that it was not statistically significant even at the α = 0.1 level. We also ran Model (2) using quarter dummies instead of town dummies using quarter 1 as the base. If housing prices were rising during the year, then we would expect the coefficient on later quarters to be statistically significant and negative, since houses sold later in the year would appear in the data to have been mis-assessed in a more regressive manner as prices increase. However, the coefficient on quarter 4 was not statistically significant at the α = 0.05 level, which to us is evidence that changing housing prices do not have a significant impact on our regressivity estimates. Together, we believe these robustness checks demonstrate that the results of our slope coefficients are driven by bias in assessment, and not housing price changes occurring within the calendar year. 30

31 newer, smaller houses, and we have evidence that these effects are non-linear. Again, if variation in the difference between assessment and sale price were random, then our model would not be able to predict any of the variation in the gap. Since it does, however, it indicates that there is indeed a systematic bias in the gap stemming mainly from basic features of the property. From an assessment perspective, these may be easier biases to correct for when assessing a property compared to neighborhood characters since property features are more easily ascertained by a quick visual scan of the property. While the above estimation would highlight potential indicators of assessor bias, we will also measure vertical inequity more directly by regressing ln(a/p) on the predicted price of each property, as outlined in Equation (7). As explained earlier, graphing ln(a/p) against the sale price of a property could yield a false impression as to the existence of regressivity given that price is a noisy measure of value, since it is essentially regressing assessed value on price. A better alternative is to approximate the true value of the property, V, using the property features from Equation (5) in a hedonic pricing model shown in Equation (6). In doing so, we confine market error μ from Equation (2) to the left-hand side of the regression, allowing us to generate predicted prices without measurement error introduced by μ. Figure 4 displays the results from Equation (7) using a local polynomial smoothing function. The actual regression results may be found in Section B of the appendix. As seen in Figure 4, there is clear evidence of increasing regressivity across most of the price spectrum. There is also evidence that regressivity begins to decrease for the highest-value properties, creating the non-linear shape of the regressivity, and corroborating the statistically significant second-order terms from the results in Table 2. In addition, Figure 4 again shows us 31

32 that there is no evidence of constant assessor error, indicating that assessor error is due more to property observables than some systematic amount of bias. Figure 1 Single Family Residences in Massachusetts Calendar Years Predicted Log of Sale Price Predicted Log(Price) Local Poly. Smoothing Epanechnikov kernel function bandwidth = 0.1 To reiterate, this estimate of the bias in assessment can be interpreted as a conservative estimate of regressivity. Because of the greater mean assessed value in our sample versus the population of single-family residences in Massachusetts, our truncated sampled would be biased towards zero, meaning that regressivity in assessment across the state could potentially be larger. We also want to stress that a graph such as this using predicted mis-assessment and predicted price is a much better, unbiased way to visually assess the existence of vertical inequity in an area due the 32

33 exogeneity of P. A graph that simply uses the difference in assessment and sale price or the log of (A/P) will be predisposed to showing a regressive picture because of the errors-in-variables problem. 4.b. Regressivity within-town Each town in Massachusetts has its own board of assessors, which is responsible for assessing all properties within the town. Assessing boards are typically comprised of three assessors plus assisting staff. Each board member is required to take a course on adequate assessing practices and then pass a certification exam within two years of taking office. 21 Both the course and the exam are administered by the state s DOR. Moreover, the DOR also monitors towns assessments on a regular basis in order to ensure assessment equity. 22 Despite all this state supervision and guidance, and given that town dummies were statically significant in Model (2) of Table 2, one might still expect to observe some heterogeneity in the amount of regressivity across towns. For example, it could be the case that some towns experience regressivity in assessment and others experience progressivity, with a net result of regressivity when pooling the data. To investigate this possibility, we regress ln(a/p) on property features for each property i in town j, as outlined in Equation (8) below. ln(a ij /P ij ) = α j + β 1j (PROP ij ) + β 2j (PROP ij ) 2 + ε ij (8) 21 Source: Assessment Administration: Law, Procedures and Valuation Introduction, Massachusetts Division of Local Services, accessed January 18, 2016, available at: < 22 Source: Assessment Administration: Law, Procedures and Valuation Chapter 1 Assessing Administration, Massachusetts Division of Local Services, accessed January 18, 2016, available at: < 33

34 We then generate a predicted ln(a/p) for each property in our sample based on the estimates obtained above. Next, we break up our data into 16 cells based on predicted price from Equation (6), and town median income. We match each predicted ln(a/p) from Equation (8) with one of these cells, and compute the mean value for each cell. To improve readability, we convert these 16 cells into percentages and change the sign, so that a positive number indicates underassessment/regressivity. The results are shown in Figure 5 below. Figure 5 The results above tell an interesting story about vertical inequity when comparing towns across Massachusetts. According to the income-axis, single-family properties in higher-income towns in 34

35 Massachusetts are more likely to be under-assessed, and that most residents enjoy a discount on their property taxes just by virtue of living in that town. However, there seems to be an exception for the lowest-value houses in located in the highest-income towns, which are very close to being accurately assessed. On the other hand, the price-axis of Figure 5 indicates that higher-value properties in general are much more likely to be under-assessed no matter where they are located, and that the difference in under-assessment seems to be greater when comparing the highest and lowest-value houses than when comparing the highest and lowest-income towns. Hence, it appears while the income level of a town does affect mis-assessment, the value of the property still plays a significant role, and the owners of the highest-value properties enjoy the largest discount on their property taxes regardless of location. Conversely, those with the lowestvalue houses always get the smallest break, with owners of low-value houses in high-income towns actually receiving no discount on their property taxes. These results, coupled with the evidence from our regressions in the previous section, are a clear indication that: 1) assessor error ε from Equation (1) is not random and there is indeed bias in assessment across towns; and 2) systematic assessor bias stems from town-level observables, as well as property-level observables. 5. Conclusion Past studies concerning property taxes and mis-assessment have so far been unable to adequately deal with the errors-in-variables problem present in housing data stemming from measurement error in both the assessed value and market price of a house. In particular, we argue that the Clapp (1990) method that many studies have relied on is a flawed measure of vertical inequity since his instrument keeps measurement error on the right-hand side of the regression and is most likely not an exogenous measure of assessor error. We propose several new methods that 35

36 effectively deal with this problem as well as demonstrate how they may be used to estimate minimum values for assessor error and bias. By regressing the log difference between assessed value and sale price on property features, the slope coefficients provide estimates of assessor bias based on these observables. We find that there is bias in assessment based on several observables such as house age, residential area, and median income of the census block-group, and that there is a nonlinear relationship between these observables and mis-assessment. We also argue that the zero-centered R 2 is an asymptotic lowerbound on how much of the variance in mis-assessment is due to assessor error (and not withinyear variance in price), which we estimate to be approximately 23.3 percent. We also use a hedonic price estimates to confirm that there is bias in assessment across most of the housing price spectrum in Massachusetts, and show that this bias is indeed regressive. Lastly we show that there is bias in assessment based on town-observables as well, and highlight that properties in higher-income towns being more likely to be under-assessed than those in lower-income towns. However, across the income spectrum, higher-value properties are still more likely to under-assessed. So, no matter where they live, higher-income residents are much more likely to get a property tax break compared to the rest of the population. Given the availability of property data on state websites and commercial websites such as Zillow, it should be reasonable for assessment boards and organizations to obtain the property features data necessary for the estimation we propose. States themselves can also use these methods to not only determine if vertical inequity is present in an area, but also determine which aspects of the house and neighborhood are the most likely causes of assessor bias as well as the degree to which assessor error is responsible for any mis-assessment. We believe that the framework described in this paper will allow future researchers to better measure the impact of 36

37 assessor error and bias on mis-assessment, which will help assessors be more accurate and ensure that residents are not over-paying or under-paying on their annual property tax bill. References Birch, J., and M. Sunderman, Regression Modeling for Vertical and Horizontal Property Tax Inequity, Journal of Housing Research, 23(1). Clapp, J., A New Test for Equitable Real Estate Tax Assessment, Journal of Real Estate Finance and Economics, 3, Edelstein, R., An Appraisal of Residential Property Tax Regressivity, Journal of Financial and Quantitative Analysis, 14(3). Fairbanks, J., Goebel, P., Morris, M., and W. Dare, A Monte Carlo Exploration of the Vertical Property Tax Inequity Models, Journal of Real Estate Literature. Greene, W., Econometric Analysis, Seventh Edition, Pearson. Goolsby, W., Assessment Error in the Valuation of Owner-Occupied Housing, Journal of Real Estate Research. Harris, L., Assessing Discriminaiton: The Influence of Race in Residential Property Tax Assessments, Journal of Land Use, 20(1). International Association of Assessing Officers (IAAO), Standard on Mass Appraisal of Real Property, January 2012, p. 17. Johnson, M., Assessor Behavior in the Presence of Regulatory Constraints, Southern Economic Journal, 55(4): Kochin, L., and R. Parks, Testing for Assessment Uniformity: A Reappraisal, Property Tax Journal, 3: Paglin, M., and M. Fogarty, Equity and the Property Tax: A New Conceptual Performance Focus, National Tax Journal, 25(4): Ross, J. M., Assessor Incentives and Property Assessment, Southern Economic Journal, 77(3):

38 Ross, J. M., Interjurisdictional Determinants of Property Assessment Regressivity, Land Economics, 88(1), Sirmans, S., Diskin, B., and H. Swint Friday, Vertical Inequity in the Taxation of Real Property, National Tax Journal, 48(1): Sirmans, S., Gatzlaff, D., and D. Macpherson, Horizontal and Vertical Inequity in Real Property Taxation, Journal of Real Estate Literature, 16(2). Smith, B., Applying Models for Verticla Inequity in the Property Tax to a Non-Market Value State, Journal of Real Estate Research, 19(3). Smith, B., Sunderman, M., and J. Birch, Sources of Variation in County Property Tax Inequities, Journal of Public Budgeting, Accounting & Financial Management. 15(4): 571. Sunderman, M.., Birch, J., Cannady, R., and T. Hamilton, Testing for Vertical Inequity in Property Tax Systems, Journal of Real Estate Research 5(3): Wooldridge, J., Introductory Econometrics: A Modern Approach, Fifth Edition, South- Western College Pub. 38

39 Appendix Section A (Hedonic Price Estimates) Depedent Variable: Log(Sale Price) House Age *** Lot Size (in 1000s, sq. ft.) *** Residential Area (in 1000s, sq. feet) 0.367*** Number of Rooms *** Med hhd inc in Block Group (in 1000s) *** Population of Block (in 1000s) * Median age of Block Population Black Share of Block Population *** Share of Houses Vacant in Block [House Age]^ *** [Lot Size (in 1000s, sq. ft.)]^ *** [Residential Area (in 1000s, sq. feet)]^ *** [Number of Rooms]^ *** [Med hhd inc in Block Group (in 1000s)]^ [Population of Block (in 1000s)]^ [Median age of Block Population]^ [Black Share of Block Population]^ ** [Share of Houses Vacant in Block]^ *** Constant 11.77*** Observations Adjusted R-squared Month and year dummies YES Town dummies YES Source: MassGIS Data Level 3 Assessors Parcel Mapping Census Data from 2010 Decennial Census Standard errors clustered by town, omitted for conciseness * p<0.10 **p<0.05 ***p<0.01 Table 3 39

40 Appendix Section B Table 4 shows the results of our regression of ln(a/p) on ln (P), as shown in Equation (7). We note that the positive constant term in this model most likely stems from the use of log price, and as shown in Table 2 and Figure 4, there is little evidence of constant, systematic assessor error. Table 4 Dependent Variable: Log(A/P) Predicted Log(Sale Price) *** ( ) Constant 0.364*** (0.0729) Observations Adjusted R-squared Source: MassGIS Data Level 3 Assessors Parcel Mapping Standard errors clustered by town in parentheses * p<0.10 **p<0.05 ***p<

41 Chapter 2 Poverty by Design: An Analysis of the Mount Laurel Decision and the Role of the Council on Affordable Housing 1. A Brief History of the Mount Laurel Decision During the 1960s, policy-makers in Camden, New Jersey, pushed for urban renewal development projects to help revitalize the city with the aim of attracting new residents. Instead, it led to most of the city s white middle class residents fleeing the city for the suburbs, leaving a mostly poor minority population behind. In nearby Mount Laurel Township, a development project was already in motion to transform the town from farmland into an affluent suburb, zoning the town in such a way that only wealthy residents would be able to afford to live there. 23 As the plan moved forward, several members of the township s historically black community proposed a plan to build affordable garden apartments to keep sections of Mount Laurel accessible to the black community. In the fall of 1970, the Mayor of Mount Laurel Township told these residents that their application would be denied, saying If you people can t afford to live in our town, then you ll just have to leave. 24 These people responded by filing a lawsuit against Mount Laurel Township 25, and in March of 1975 the New Jersey Supreme Court upheld the decision of the lower trial court in a decision that would come to be known as Mount Laurel I. This decision stipulated that New Jersey municipalities could not engage in exclusionary zoning, and had a constitutional obligation to make realistically possible an appropriate variety and choice of housing. 26 The implementation of this law, however, would not come so easily. Many municipalities openly refused to implement the decision of Mount Laurel I, and clear standards of an appropriate 23 Mount Laurel I, 67 N.J. 151 (1975) 24 Krip and Rosenthal, Southern Burlington County N.A.A.C.P. v. Mount Laurel Township 26 Mount Laurel I, 67 N.J. 151 (1975) 41

42 variety of housing were never established. As a result, lawsuits continued to be filed against a number of New Jersey municipalities, leading to a follow-up key decision by the Supreme Court of New Jersey in This would come to be known as Mount Laurel II, the main goal of which was to give the state Supreme Court s original decision more authority. 27 To help bolster this follow-up decision, the New Jersey State Legislature passed the Fair Housing Act in The Mount Laurel II decision outlined parameters for what constituted an appropriate variety of housing in New Jersey municipalities, while the Fair Housing Act created the Council of Affordable Housing (COAH), whose goal was to set yearly affordable housing requirements for municipalities that volunteered to follow the Mount Laurel II decision based on six created COAH regions. Based on the Fair Housing Act, low-income was defined as less than 50 percent of a COAH region s median income, and moderate-income was defined as between 50 percent and 80 percent of a region s median income. 28 Because of the voluntary nature of the program, however, a number of municipalities would refuse to join COAH. In addition, municipalities had to apply for membership to COAH by providing a plan that outlined how the municipality would meet its affordable housing requirement. These affordable housing requirements were then separated into three different phases or rounds of participation, each with its own affordable housing requirement. The First Round that ran from 1987 to 1993, the Second Round from 1993 to 1999, and the Third Round, which was originally slated to begin in 1999 but was continually delayed and failed to be fully implemented Mount Laurel II, 92 N.J. 158 (1983) 28 COAH Income requirements, NJ Department of Community Affairs 29 Based on COAH s calculations, the First Round rules specified that a combination of 10,849 low and moderate income housing units were to be built statewide per year, while the Second round specified that 6,465 units of similar type were to be built. For more information, see: New Jersey Department of Community Affairs COAH Status and Information. 42

43 By volunteering to come under COAH s jurisdiction, municipalities would be given several benefits. First, they would be protected from builder s remedy lawsuits, in which developers could sue municipalities for not zoning an appropriate amount of land for affordable housing. Second, these municipalities would be able to freely choose where their affordable houses would be built, as long as they were within the municipality s borders. 30 Affordable housing required as the result of a lawsuit, on the other hand, would be zoned in an area determined the courts. Finally, these municipalities would be able to engage in Regional Contribution Agreements (RCAs), in which a municipality could pay another municipality within its COAH region to build up to 50 percent of its affordable housing allotment for them, as long as it did not infringe upon a municipality s indigenous need for affordable housing as outlined by Mount Laurel II. 31 These benefits created an interesting incentive for New Jersey municipalities to fall under COAH s jurisdiction, yet they also raise several pertinent questions about the design of this policy. In theory, the Mount Laurel decision could improve the lives of low income families by ensuring the existence of affordable housing within any given New Jersey municipality; however, did the design of this program have unintended consequences on the spatial distribution of poverty across the state, exacerbating a problem that the Mount Laurel decisions desperately tried to avoid? Using data on COAH housing requirements obtained from COAH s generated reports as well as the New Jersey s Department of Community Affairs, I will attempt to answer three questions of interest for this paper. 1) What types of municipalities volunteered to come under COAH s jurisdiction? 2) Is there a pattern to compliance rates when examining the construction of new affordable housing, and how successful was COAH in getting municipalities to meet their affordable housing 30 N.J.S.A 52:27D Mount Laurel II, 92 N.J. 158 (1983) 43

44 requirements? And 3) what types of municipalities engaged in an RCA, and is there a pattern of rich municipalities sending their affordable housing allotments to poor municipalities? I find that those municipalities which perceived the greatest possibility of a builder s remedy lawsuit were the most likely to join COAH, and were also more likely to fulfill some of their new construction obligations. However, I also find that compliance rates were generally low throughout the first two rounds of COAH requirements, though there does not appear to be a selection problem at work based on the results of a Heckman two-step estimation procedure. I also find a clear pattern of higher-income municipalities sending affordable housing units to lowerincome municipalities, allowing rich municipalities to actively work against economic integration within their communities, as well as undermining the original intent of the Mount Laurel decision. The rest of this paper proceeds as follows. Section 2 describes previous attempts to study the impact of the Mount Laurel Decisions. Section 3 describes the data used in this paper. Section 4 explains in more detail the process of joining COAH and examines what types of municipalities would volunteer to join COAH. Section 5 examines the compliance rates of municipalities that joined COAH and if there is a pattern to which municipalities met their obligations. Section 6 explores what types of municipalities engaged in a Regional Contribution Agreement, and Section 7 concludes. 2. Previous Studies The literature on analyzing and evaluating the impacts of the Mount Laurel agreement is sparse and lacking a core analysis of the socio-economic impact of this policy, the aim of which was to aid those in need of affordable housing and prevent segregation within communities. The most 44

45 well-known studies explore how new affordable housing impacts individuals and families. Massey et al. (2013) examines the lives of residents in the Ethel Lawrence Homes building complex, which was created as a result of the Mount Laurel decision. Since there was not enough space to accommodate everyone who applied to live in the building complex, they compare changes in the lives of accepted and rejected applicants and find that accepted applicants had a higher employment rate and lower levels of stress. Only a few studies look at the design of RCAs as a state policy and discuss the adverse effects that RCAs could possibly have on the distribution of poverty across New Jersey [Fox (1988), Bailin and Eisdorfer (1996), and Gianetti (2001)]. These studies, however, focus on the legality of RCAs rather than evaluate their potential economic and social impacts. However, there is a growing public interest in general issues of housing policy and its impact on spatial inequality. An article published in the New York Times in July of 2016 cited a growing body of economic literature examining how towns and cities use zoning laws to reflect an anti-growth sentiment, and the potential adverse impact that can have on area development. 32 I am aware of no study to date that has explored the effect of municipality choice and the selfselection issue regarding which municipalities decided to follow COAH s recommendations, meet their affordable housing obligations, or engage in a Regional Contribution Agreement. I am also not aware of any study to date that has used the affordable housing requirements data utilized in this paper. Hence, I believe that this paper will contribute to the literature by addressing issues of zoning and inequality raised by the Mount Laurel decision, and by employing a new set of data 32 Conor Dougherty, How Anti-Growth Sentiment, Reflecting in Zoning Laws, Thwarts Equality. New York Times, July 3 rd, < 45

46 3. Data COAH s Annual Reports list municipality building requirements for municipalities that joined COAH, which I currently have for the years 1987, 1995, 1996, 1998, 2000, and However, only the 2002 report lists the required building obligation for all New Jersey municipalities, and not just those who joined COAH. 33 The 2002 Annual Report lists the 1987 to 1999 building obligation for each of New Jersey s 566 municipalities, including both the new construction of affordable housing units and rehabilitation projects. 34 New construction could be either 100 percent new affordable housing, or it could be inclusionary development in which affordable housing was mixed in with newly constructed market-value units. 35 Rehabilitation, on the other hand, involved no new construction and instead involved updating housing units that were considered sub-standard in terms of local building code requirements. 36 The 2002 report also lists which municipalities volunteered to join COAH during either the First Round or Second Round, as well as which municipalities were forced to join via a court-ordered participation. Hence, my dataset on will be treated as a cross-section of building requirements throughout this time period. I supplement this information with New Jersey municipality data from the 1990 Census accessed through the National Historical Geographic Information System (NHGIS), maintained by the Minnesota Population Center at the University of Minnesota. This data includes median household 33 I attempted to find more Annual Report, as the name Annual would imply that there should be more than six reports in total covering the timespan from 1987 onwards. However, I was unable to find evidence of any other reports other than for the six years listed. I spoke with several members of COAH, who could not locate additional reports and did not know if they had ever existed, and only these six were available from the New Jersey state library, which keeps a record off all official documents produced by state organizations. It therefore seems likely that, despite the title of some of these reports, they were produced sporadically. 34 The word municipality is chosen deliberately and used throughout this paper. New Jersey is comprised of cities, towns, townships, and several villages. Since the Mount Laurel decisions, as well as COAH building requirements, affected the entire state, a municipality refers to all 566 cities, towns, townships, and villages across the state. 35 See the New Jersey Guide to Affordable Housing for a list of which developments met which specification. 36 COAH 1987 Annual Report 46

47 income, total number of households, population demographics, and unemployment rates. 37 Descriptive Statistics on the 2002 COAH Annual Report data as well as the included 1990 decennial census data may be found in Table 1. Table 1 Descriptive Statistics New Jersey Municipalities Mean Std. Dev. Skewness Min 50th Percentile Max New Construction Obligation ,459 Rehabilitation Obligation ,431 Median Household Income $46, $15, $16,775 $42,949 $150,001 Total No. of Households 4, , ,630 91,552 Unemployment Rate 4.67% 2.33% % 4.16% 18.56% Black Share of Population 6.63% 12.13% % 1.84% 96.59% Number of Municipalities 566 Source: 2002 COAH Annual Report, 1990 Census Information on Regional Contribution Agreements comes from COAH and the New Jersey Department of Community Affairs. The data includes all RCAs that occurred from the program's inception in 1987 until the law permitting their use was ultimately repealed in For each RCA, I am able to observe which municipality was the sender and which was the receiver of affordable housing units, as well as how many housing units were to be transferred and the price 37 NHGIS provided the easiest method of matching the municipality information form the 2002 COAH report with census data. Since information for this time period is only available during the decennial census, 1990 was chosen as the representative year. As a robustness check, I also conducted the each of the analyses in this paper using the 2000 Census data, but the results remained largely the same. It is worth noting, however, that not every household was asked about employment information in 1990, and the numbers provided come from NHGIS s own estimations. Hence, while I believe they are fairly accurate, it is reasonable to believe that it is measured with some degree of error. 38 Assembly Committee Meeting Substitute for Assembly, No State of New Jersey. 213 th Legislature. <ftp:// 47

48 per housing unit. Since some RCAs were broken up into multiple parts, or in some cases the participating municipalities decided to engage in an additional RCA, each RCA transaction is treated as a separate observation in the data. Lastly, data on property sale prices for each municipality was obtained from the New Jersey Department of the Treasury website, which includes the average sale price of residential properties for each municipality in a given year from 1994 (roughly around the start of COAH s second round requirements) through 2008 (after which time RCAs were no longer allowed by law). 39 This information will be used as a proxy for the wealth of a given municipality in the year in which an RCA occurred. 4. Joining the Council on Affordable Housing (COAH) COAH divided New Jersey into six different regions, each of which was comprised of four to five counties. A map of the six COAH regions can be seen below in Figure 1. COAH then calculated affordable housing requirements at a regional level based on the guidelines outlined by Mount Laurel II. These calculations were conducted several times, but this paper will focus on the first round requirements from 1987 to 1993, and the second round requirements from 1994 to Despite the importance of these numbers in both servicing the communities in need of affordable housing across the state, as well as the potential construction burdens they could place on municipalities, specifics as to how these affordable housing requirements were calculated is absent from COAH s annual reports. However, a memo to New Jersey municipalities in the early 2000s, authored by then chairman of the State League of Municipalities Affordable Housing 39 The State of New Jersey Department of the Treasury, Division of Taxation. Accessed July 8 th, < 48

49 Committee Edward J. Buzak, described the process of shifting from the old requirements as calculated between 1987 through 1999 to the new growth share plan that COAH would be adopting going forward. In this memo, he described that the obligations calculated for the first two rounds were based on a complicated formula of speculative projected statewide and regional population, economic and employment growth which was then allocated to individual municipalities, according to land availability, employment growth, and income of residents. 40 Hence, there is at least some base evidence that these numbers were indeed based on indigenous need as outlined in Mount Laurel II, though there is some doubt as to how the specific numbers were calculated. 40 Buzak, Edward J., Build as You Grow: Understanding Your Affordable Housing Obligations, Archived January 17, Accessed June 28, < 49

50 Figure 2 Source: 2002 COAH Annual Report A simple regression of each municipality s new construction obligation on its total number of households in 1990 reveals that population explains only about 10 percent of the variation in a municipality s new construction obligation according to the adjusted R 2. However, the amount of explained variation improves to about 29.7 percent with the addition of the total number of persons 50

51 below the poverty line in 1990, 41 though the additional of median household income adds very little explained variation. This is still a low amount of explained variation considering that these are variables that should logically explain a decent amount of a municipality s new construction obligation, not only by intuition, but also by COAH s own language. Interestingly, a regression of each municipality s total rehabilitation obligation on number of households shows that population explains about 72.5 percent of the variation in a municipality s rehabilitation obligation, with the addition of the number of persons below the poverty line improving the explained variation to 83.5 percent. 42 So, it seems that rehabilitation requirements were based mostly on population, while it is unclear what most of the basis was for new construction requirements. The regression output and additional information about specification may be found in the Appendix. 43 Any municipality that was interested in volunteering to come under COAH s jurisdiction was eligible, but a joining municipality was required to present some sort of plan to COAH demonstrating that it had a feasible plan in place to meet its affordable housing obligation. If a municipality did present such a plan and had it approved, then it would be granted substantive certification, granting it protection from exclusionary zoning litigation as well as providing access to state funds from the Department of Community Affairs and the New Jersey Housing and Mortgage Finance Agency to help implement its plan. 44 If a municipality instead did not submit a plan to meet its affordable housing obligation, then it left itself open to a builder s remedy lawsuit. In this case, if a municipality was then sued by a developer and lost this lawsuit, it would be court- 41 An F-test confirms that this is a statistically significant change with test p-value of < An F-test confirms that this is a statistically significant change with test p-value of < I also tried adding in more right-hand side variables to both the new construction and rehabilitation obligation regression using the census data outlined in Section 3, but found that the additional variables explained a very low fraction of the variance in a municipalities obligations. Again, see Appendix for the output of these regressions. 44 COAH 1987 Annual Report 51

52 ordered to construct new market-value housing units as well as affordable housing units in a 4-to- 1 ratio according to their new construction obligation as calculated by COAH. For example, if a municipality s new construction obligation was 100 affordable housing units and it lost a builder s remedy lawsuit filed by a developer, then not only would the local courts take over the implementation of COAH s building requirements, but the municipality would need to build the 100 affordable housing units as well as 400 market-value housing units. COAH s own 1987 Annual Report contained language describing how it believed that the possibility of having to zone for an incredibly large amount of new housing units as the result of a builder s remedy lawsuit would encourage more municipalities to join COAH. 45 Nevertheless, given that not all municipalities volunteered to join COAH, it is clear that some municipalities valued joining COAH more than others. But what types of municipalities would be most willing to join COAH in exchange for the benefits and protections it promised? To answer this question, I use the building requirements information from COAH s 2002 Annual Report, as well as municipality-level demographic information taken from the 1990 Census. Summary information about municipality participation may be found in Table 2. Table 2 First Round COAH Participation Second Round NO YES NO YES Source: COAH 2002 Annual Report 45 COAH 1987 Annual Report 52

53 In modeling each municipality s decision of whether or not to join COAH, I assume that a municipality is beholden to its residents, who will vote to decide whether or not to join COAH. Each resident is a utility maximizer whose voting decision can be estimated from a random utility model (RUM) in which a resident s utility is a function of his or her own income level, as well as the difference between his or her income level and the income level of potential new affordable housing residents. However, since it is the median voter who will determine the outcome of the vote, I estimate the utility of each municipality s median voter using the utility function outlined below in Equation (1): U i = β 0 + β 1 ln(inc i ) + β 2 ln(inc i /LIMIT i ) + β 3 ln(nc i ) + β 4 ln(hshlds i ) + X i + μ i (1) where INC is the median household income of municipality i in 1990, 46 LIMIT is the regional income limit on families moving into new affordable housing units in that region as published by 46 When comparing data on median household income for NJ municipalities between the 1990 Census and 2000 Census, there is a change in the mean of the median household income but it is not substantial, after adjusting the 2000 median household income into 1990 dollars. A two-tailed t-test of the means fails to reject the null of a difference with a p-value of 0.061, while a one-tailed t-test with HA > 0 has a p-value of The calculated means were $48, (2000, adjusted to 1990 dollars) and $46, (1990). Additionally, any reference to median income throughout this paper refers to median household income. 53

54 COAH, 47 NC is the new construction obligation of municipality i from 1987 to 1999, 48 HSHLDS is the total number of households in municipality i in 1990, X is a vector of COAH region control dummies, and μ is a stochastic error term. The goal of parameter β 2 is to capture potential disutility from lower-income residents moving into a community, engendered by the not in my backyard mentality concerning new affordable housing. The median voter in municipality i will vote in favor of joining COAH if Ui > 0 and will vote against joining COAH if Ui < 0. I estimate the choice probabilities from Equation (1) using a binary probit model that takes the form: P(y i = 1 x i ) = Φ(β 0 + β 1 ln(inc i ) + β 2 ln(inc i /LIMIT i ) + β 3 ln (NC i ) + β 4 ln(hshlds i ) + X i) (2) where Φ is the standard normal cumulative distribution function, and yi is a binary variable that takes a value of 1 if a municipality decided to join COAH in the first or second round, and 0 if it did not. Only municipalities that volunteered to join either the first or second round are included 47 The COAH regional income limits were posted by COAH defining what constituted a median, moderate, and low income family of a given size in a particular year. The earliest available regional income limits were published in Hence, in order ensure that these limits are in line with the 1990 decennial census data, I adjust these values to 1990 dollars using the CPI. I use the (also available) 2011 income limits as a simple test of this approach, and adjusting the 2011 limits for 2000 dollars produces close estimations of the 2000 income limits. To calculate the variable LIMIT, I assume a family of four persons as the representative incoming family. Given that COAH building requirements were meant to service low and moderate income families, I calculated the average of the low and moderate income limit for a family of four (adjusted to 1990 dollars), and use that as the variable. Since there are only six COAH regions, there are only six variations of the variable LIMIT in the data. 48 Because the new construction obligation variable contained some observations equal to 0, the logarithmic transformation of NC was calculated by ln(nc +1). 54

55 in the regression; hence, court-ordered municipalities are excluded from the analysis. The results of Equation (2) are shown in Table 3. Despite using only four independent variables, Equation (2) correctly predicts the choice made by the median voter within each municipality a majority of the time, with 66 percent of those who did not join COAH predicted correctly, and 75 percent of those who did join predicted correctly. 49 Interpreting the impact of the included variables as causal, however, may prove more difficult and requires some additional thought. 49 I had also tried other specifications that included log and level versions of the other variables available in both the 2002 COAH report and the census data. However, the four variables included in Equation (2) were the only ones to remain statistically significant in every specification, and their impacts on the choice probabilities has the clearest interpretation. Moreover, the predictive ability of Equation (2) improved marginally at best using specifications that included other variables, so for prediction purposes Equation (2) as specified is sufficient. 55

56 Table 3 Dependent Variable: COAH Join Decision Coefficients Avg Mfx ⱡ ln(median Household Income) 2.481*** 0.761*** (0.769) ln(median Household Income/Income Limit) * * (0.878) ln(new Construction Obligation) 0.152*** *** (0.0454) ln(total No. of Households) 0.122** ** (0.0541) Constant *** (7.926) Observations 485 Pseudo R-squared Regional Dummies YES Source: 2002 COAH Annual Report, 1990 Decennial Census Court-ordered municipalities excluded ⱡ Average marginal effects for each covariate Robust standard errors in parentheses * p<0.10 ** p<0.05 *** p<0.01, two-tailed test Actual Correct Predictions Predicted Did not join Joined Did not join 92 (66%) 87 Joined (75%) As expected, the negative sign on the coefficient of ln(inc/limit) indicates that there is indeed a disutility incurred by the median voter the larger the income gap between current and potential incoming residents. Presumably this is because residents of higher-income municipalities would prefer to keep lower-income residents out and hence would be less likely to 56

57 agree to build more affordable housing. Conversely, the positive sign on a municipality s new construction obligation reflects an increasing willingness of the median voter to join COAH the larger the affordable housing requirement COAH has calculated. At first glance it might seem contradictory that the coefficients on the two aforementioned variables take opposite sings, since one would assume that more affordable housing would create additional disutility. However, this result is not surprising if municipalities with the largest new construction requirement were those with the greatest need, and also those at the greatest risk of being sued. In that case, there is utility gained from joining COAH because of the protection it provides form lawsuits. This point is corroborated by examining more closely municipalities whose participation was court-ordered vs. those who joined voluntarily, as seen in the comparison outlined in Table 4. Table 4 Voluntary vs. Court Municipalities Voluntary Court T-test P-value Avg. Median Household Income $49, $50, New Construction Obligation Rehabilitation Obligation Number of Municipalities Source: 2002 COAH Annual Report, 1990 Census H 0 : Statistically equal means, H A : statistically different means, two-tailed test The average median income level for court-ordered vs. voluntary participation municipalities is statistically the same at around $49,000, as confirmed by a t-test of means. On the other hand, court-ordered municipalities have a much higher new construction obligation on average, which 57

58 makes sense in light on the results from Equation (2), although their rehabilitation obligation is roughly equal (due in part to its large standard deviation). Municipalities who joined COAH under a court order were those municipalities that were tied up in litigation over their lack of availability of affordable housing at some point during this time period. If municipalities with the greatest affordable housing deficit were the most likely to be sued or tied up in litigation, these municipalities would also have among the largest affordable housing requirements in the state. Hence, as the courts began to send authority over these municipalities building requirements to COAH, these would then, by definition, be the municipalities with the greatest affordable housing need, and thus the greatest new construction obligation. Thus, the positive coefficient on new construction could also reflect that municipalities understood that those municipalities with the highest new construction requirement were the most likely to be sued, and as a result had the largest incentive to obtain litigation protection by joining COAH. 50 Recall that if a municipality lost a builder s remedy lawsuit, then it would have to zone for new units at a 4-to-1 (market value to affordable housing) ratio, which would potentially not only be a large financial burden, but also would not be looked upon favorably by residents who wanted to prevent integration within their communities. We can also see from the results that there is a positive relationship between a municipality s median household income and its likelihood of joining COAH. This is a sensible result if higher-income municipalities were the ones that most wanted to receive protection from litigation and had the most to lose in terms of leaving themselves open to the possibility of a builder s remedy lawsuit, especially in light of the positive sign on new construction. Lower- 50 Another possibility is that a municipality s new construction obligation is highly correlated with its size. However, as explained in Section 4 and in the Appendix, only about 10 percent of the variation in a municipality s new construction obligation is explained by its population. Hence, given that result and the inclusion of total household sin Equation (2), NC in Equation (2) is not merely capturing the effect of municipality size. 58

59 income municipalities, on the other hand, might have been less worried about the possibility of a builder s remedy lawsuit (though may still have wanted the benefit of being able to engage in an RCA). In fact, some of the state s lower-income municipalities seemed to believe that COAH s building requirements should not apply to them. In a 2000 interview with the then Mayor of Edgewater Park (which had a below-average median household income), Darren Atzert, he explained that he believed his town should not have to comply with COAH s building requirements, stating We have our fair share of affordable housing. 51 Ironically, Edgewater Park had been dealing with a builder s remedy lawsuit for several years at that point, and was ultimately court-ordered to join COAH. As previously mentioned, another incentive municipalities might have had to join COAH was if they knew they would engage in an RCA within their region. Although this cannot be directly tested in this model (since a municipality had to first join COAH in order to engage in an RCA), a likelihood-ratio test confirms that the COAH region dummies are statistically significant with a test p-value of <0.001, highlighting that certain conditions that vary across regions played a role in the decision to join COAH. This could give credence to the story that municipalities evaluated their potential RCA partners within their region, and that it was a strong incentive for municipalities to join. I will more closely examine the RCA decision in Section Murakami, Tomoeh, Going to Court over Housing Builder s Remedy Suits Force the Issue with Towns that won t Plan for Affordable Housing, philly.com, May 14, Accessed July 2, < 59

60 5. COAH s Compliance Issues Based on information gathered from COAH s later annual reports, it is clear that COAH was self-aware that it had an image problem by the time the First Round had completed. By the time the 1995 report was released after the completion of the First Round requirements, one of the organizations actionable goals was to improve COAH s image, which was to be accomplished by informing the public about the need for and benefits of affordable housing. 52 But how successful was COAH in rallying municipality support and participation in the name of more affordable housing? Regardless of whether or not a municipality volunteered to come under COAH s jurisdiction, each municipality was assigned an affordable housing requirement to be filled with low-income and moderate-income tenants. COAH s 2002 Annual Report also lists how much of that obligation for each category had actually been built within that time period. As it turns out, many municipalities that joined COAH did not complete any new construction or rehabilitation projects. Of the municipalities that were under COAH s jurisdiction, including both voluntary participation and court-ordered involvement, percent of these municipalities did not complete any new construction. Similarly, percent of municipalities under COAH s jurisdiction did not complete any rehabilitation projects. But what about those remaining municipalities that fulfilled at least some of their COAH obligations? Figures 2 shows the distribution of participating municipalities met-obligation for new affordable housing construction from , while Figure 3 shows a similar distribution for rehabilitation projects. As seen in both figures, although the majority of municipalities fell short of meeting 100 percent of either their new construction or rehabilitation obligation over this time period, many municipalities met their obligation. In fact, some municipalities even exceeded 52 COAH 1995 Annual Report 60

61 their obligation, though these observations have been represented as 100 percent completion in the graphs below. Moreover, it appears that among complying municipalities that there was a better project-completion rate for rehabilitation projects, with many more municipalities completing 100 percent of their rehabilitation requirement compared to new construction requirements. Figure 2 Frequency COAH Housing Compliancy New Construction Obligation Percent of New Construction Obligation Completed 61

62 Figure 3 Frequency COAH Housing Compliancy Rehabilitation Obligation Percent of Rehabilitation Obligation Completed This difference in compliance could stem from two possible sources. First, it is likely to be much more costly and time consuming to construct new affordable housing units, and it would therefore be more difficult for a municipality to meet this requirement as opposed to rehabilitation projects. However, considering that municipalities had to submit plans to meet their obligations prior to joining COAH, one would expect municipalities to reach their goals in a timely manner if those plans had been approved. Second, this result may be expected if municipalities were actively trying to avoid new affordable housing units being integrated into existing communities. For a municipality to construct new units, it had to find an appropriate space within its borders to zone for those units, which could prove difficult if those municipalities are also trying to weigh the concerns from townspeople who did not want affordable housing built near them. If, however, existing affordable housing within each municipality was located in segregated portions of the 62

63 community, then municipalities could rehabilitate those units while still preserving the segregation within their communities and appeasing the pro-segregation portion of their population. It is clear then that compliance was a huge issue for COAH. Nonetheless, this may not be surprising given COAH s lack of enforcement mechanisms. Despite the quasi-judicial powers that COAH was granted from the Fair Housing Act of 1985, it was not granted any sort of power of enforcement or repercussion for municipalities that failed to meet their obligations. This begs the question: why would any municipality have chosen to have built anything at all? Given the unorganized nature of this process, it is possible that at least some municipalities did not realize that COAH lacked substantial power to punish municipalities for not meeting their obligations. At the very least, municipalities might have built only as much of their obligations that they believed would convince COAH that they were making a good-faith effort to meet their required numbers; or, COAH might have taken the fact that the municipality had gone through the necessary steps of creating a housing plan that would grant them substantive certification as a sign of good-faith intention on its own. Yet, a non-trivial number of municipalities did in fact complete 100 percent (or greater) of their obligations. To explore if there is a selection issue in regards to a municipality choosing the fulfill all, part, or none of its new construction obligation or rehabilitation obligation, I employ a Heckman twostep estimator using both the COAH building requirements as well as municipality demographics from the 1990 decennial census as the selection variables. 53 The first stage will utilize a binary probit regression to estimate what types of municipalities fulfilled any of their new construction or rehabilitation obligation, as outlined in Equation (3): 53 See Heckman (1979) for a full explanation of the Heckman consistent two-step estimator. 63

64 P(y i = 1 x i ) = Φ(β 0 + β 1 ln(inc i ) + β 2 ln(nc i ) + β 3 ln(hshlds i ) + β 4 UNEMP i + β 5 BLACK i + β 6 COURT i + X i) (3) where INC, NC, and HSHLDS are defined as in Equation (1), UNEMP is the unemployment rate of municipality i in 1990, BLACK is the black share of the total population in municipality i in 1990, COURT is a dummy variable that is equal to 1 if a municipality was court-ordered to join COAH and 0 if it was not, and X is again a vector of COAH regional controls. The second stage will estimate how well the variables outlined above can explain the percentage of its building requirement a municipality decided to fulfill, attempting to correct for any sample selection estimated in the first stage by including the inverse Mills ratio as calculated from the predicted values from Equation (3). 54 Y i = β 0 + β 1 δ i + β 2 λ i + μ i (4) where Y is the percent share of municipality i s 1987 to 1999 new construction obligation that the municipality completed, δ represents the independent variables from Equation (3), λ is the inverse Mills ratio, and μ is a stochastic error term. Some municipalities actually completed greater than 100 percent of their obligations, and these observations were allowed to remain above 100 and were not censored at 100, although several municipalities with a completion rate of over The inverse Mills ratio is the ratio of the probability distribution function (pdf) to the cumulative distribution function (cdf) of a distribution. See Wooldridge, Chapter 17 for a complete discussion of the inverse mills ratio and its use in sample selection models. 64

65 percent were omitted from the analysis. 55 Ideally, Equation (4) would omit at least one variable from Equation (3) to act as an exclusion restriction; however, since it is most likely the case that at least some of the variables included are endogenous and that they can explain both building selection and completion, none of the variables included in the selection probit are good candidates for exclusion in the second stage. Despite this, the second stage is still identified, however this identification now relies solely on the normality assumption of the unobserved errors affecting equations (3) and (4). While this assumption is most certainly violated, the signs of the coefficients might at least be able to provide some insight as to which municipalities decided to build anything. The results of this procedure for municipalities new construction obligations are shown in Table These observations were dropped to prevent extreme outliers from having a large amount of influence on the results. Specifications which included these five municipalities had substantially different coefficient estimates than every other specification, and so they were dropped from the final analysis to prevent the conclusions from focusing on the effects of these outliers. 65

66 Table 5 OLS Heckman Probit MLE's Avg Mfx ⱡ ln(median Household Income) *** (47.95) (0.450) ln(new Construction Obligation) * 0.468*** 0.132*** (27.44) (0.0564) ln(total No. of Households) *** (10.90) (0.0739) % Unemployment Rate * * * (3.495) % Black Share of Total Population (1.108) (0.0102) Court-ordered Participation ** *** *** (50.20) (0.193) Lamba (116.8) Constant 861.3*** (544.0) (4.999) Observations R-squared 0.28 Wald Chi 2 p-value < Pseudo R-squared 0.26 Regional Dummeis YES YES YES Source: COAH 2002 Annual Report, 1990 Census Municipalities with > 1000 percent completion rate excluded ⱡ Average marginal effects for each covariate Robust standard errors in parentheses * p<0.10 ** p<0.05 *** p<0.01, two-tailed test Municipality New Construction Compliance Dependent Variable: Dependent Variable: Percentage Completed New Construction Decision 66

67 One clear contrast between the Heckman and OLS estimates is the sign on a municipality s new construction obligation. In this case, the Heckman results are slightly more intuitive, since one might expect municipalities with lower total new construction requirements to be more likely to complete a greater share of their requirements. Nevertheless, this coefficient in the Heckman model is only statistically significant at the 10 percent level, so it is difficult to tell how well this relationship holds. In addition, the coefficients on both new construction and the unemployment rate in the Heckman and OLS models were very sensitive to alternative specifications, such as changing inclusion cut-off points for a municipality completion rates greater than 100 percent, or excluding small municipalities. Hence, while the other coefficients are fairly robust to alternative specifications, new construction and the unemployment rate are not, so it is difficult to specify their precise impact on a municipality s new construction completion rate. I also investigate a similar relationship using the percentage of rehabilitation projects completed, and re-specify the twostep estimator as outlined in equations (3) and (4) appropriately. The results may be found in Table 6. Since no municipality had over a 1000 percent completion rate for rehabilitation projects, all municipalities who joined COAH are included in the estimates presented in Table 6. The probit model again shows that municipalities with higher obligations are more likely to undertake any rehabilitation project, and once again municipalities under courtordered participation seem to be putting up resistance to complying at all with COAH s requirements. In addition, λ is again not statistically significant, indicating that there is no evidence of a selection problem. As far as the Heckman and OLS results, all the signs of the coefficients are the same except for the court dummy. Interestingly, municipalities with greater median household incomes are predicted to complete more of their rehabilitation requirements. It could be, as mentioned earlier, 67

68 that rehabilitation as opposed to new construction is a preferred way to keep communities segregated, and serves as a compromise between meeting some of COAH s requirements as well as the wills of the existing residents. This could also serve the dual-purpose of diverting attention away from the fact these higher-income municipalities completed the smallest amount of new construction according to the results from Table 5. 68

69 Table 6 OLS Heckman Probit MLE's Avg Mfx ⱡ ln(median Household Income) * (82.68) (101.5) (0.379) ln(rehabilitation Obligation) 67.35*** *** *** (12.06) (84.89) (0.0824) ln(total No. of Households) *** *** (16.43) (25.74) (0.0841) % Unemployment Rate % Black Share of Total Population (1.511) (2.242) ( ) Court-ordered Participation *** *** (26.25) (172.1) (0.184) Lamba (627.0) Constant (902.9) (1279.6) (4.165) Observations R-squared 0.08 Wald Chi 2 p-value Pseudo R-squared 0.11 Regional Dummeis YES YES YES Source: COAH 2002 Annual Report, 1990 Census ⱡ Average marginal effects for each covariate Robust standard errors in parentheses * p<0.10 ** p<0.05 *** p<0.01, two-tailed test Municipality Rehabilitation Compliance Dependent Variable: Percentage Completed Dependent Variable: Rehabilitation Decision 69

70 Moreover, several of the coefficients for the rehabilitation results take the opposite sign as the coefficients for the new construction results. Municipalities with higher rehabilitation obligations are predicted to have completed more of their obligations, and larger municipalities are predicted to have completed less. Again, this could stem from larger, more lawsuit-vulnerable municipalities focusing their efforts on new construction given the results from Table 5, but it is difficult to determine the exact mechanism at work without more data. Overall, since participation in COAH was largely voluntary (excluding the select municipalities that had been court-ordered to participate), one would assume that these municipalities were more motivated to meet their housing requirements than those who did not join. Yet, whether their desire to join COAH stemmed from a genuine desire to service the poor in their communities, to protect themselves from the possibility of a builder s remedy lawsuit, or simply to give the appearance of compliance, it is clear from the data that the majority of municipalities failed to meet their entire obligations from the first two rounds. There is also some indication of the types of municipalities that would meet at least some of their new construction and rehabilitation obligations, but there does not appear to be a selection problem at work based on the Heckman estimates. However, municipalities may have had other incentives for joining COAH besides the opportunity to construct new affordable housing. 6. Regional Contribution Agreements As previously mentioned, one of the key benefits that municipalities would receive if they volunteered to come under COAH s jurisdiction was the ability to engage in a Regional Contribution Agreement (RCA). A municipality that engaged in an RCA would pay another municipality within its COAH region to build up to 50 percent of its affordable housing allotment 70

71 for it within the receiving municipality s borders, as long as the indigenous need for affordable housing in the sending municipality was not threatened. 56 Considering how many RCAs occurred between 1987 and 2008, it seems that COAH was not overly concerned about the potential consequences of such an allowance, as long as the housing requirements were met. COAH s 1987 Annual Report describes the intention of allowing RCAs to occur. 57 Municipalities that sent RCA units to another municipality would have to pay a minimum of $10,000 per housing unit sent. COAH believed that this would help offset the cost of building affordable housing in municipalities where the need was greatest (though it is unclear from the language contained in the COAH reports how often a municipality had to foot the bill itself for developer costs). However, it is also possible that this gave wealthier municipalities an escape route from meeting their affordable housing requirements and also provided wealthier municipalities with a means to prevent integration from occurring within their existing communities. Between 1987 and 1999, 98 municipalities sent affordable housing units to another municipality through an RCA, while 42 municipalities received units. 58 Among the six COAH regions, Region 6 was the only region that had no municipality engage in an RCA, which as seen from the map in Figure 1 was comprised of the southern-most municipalities in the state. Interestingly, Region 6 was not only the poorest region in terms of average median household income, but also had the smallest income range, as illustrated in Figure 4. Hence, since it was comprised of relatively poor municipalities compared to the rest of the state, there might have been little incentive for any of those municipalities to engage in an RCA, especially if other regions 56 COAH 1987 Annual Report 57 COAH 1987 Annual Report 58 COAH 2002 Annual Report 71

72 exhibited a pattern of higher-income municipalities sending affordable housing units to lowerincome municipalities via an RCA. Figure 4 Median Household Income by COAH Region Decennial Census 1990 Median Income ,000 80,000 70,000 60,000 50,000 40,000 30,000 20,000 10, Given that nearly twice as many municipalities sent affordable housing units as received units, it is clear that some municipalities had an eagerness to reduce their affordable housing obligation. Considering the not in my backyard disutility captured by ln(inc/limit) in Equation (2), as well as the potential forgone property tax revenue from these properties (although at least some of these properties would have assumedly qualified for a tax credit), the perceived burden of new affordable housing must have been great in these sending municipalities. Since the results from Table 3 indicated that higher-income municipalities had the strongest probability of joining COAH, one concern would be that rich municipalities within a region 72

73 would send their affordable housing obligations to the poor municipalities. Again, COAH justified this possibility in its 1987 Annual Report by stating that this would cause building funds to flow to municipalities with potentially the greatest need for affordable housing. However, this also meant a potential worsening of the geographic separation between the rich and poor across the state, undermining the original intent of the Mount Laurel decisions. To gain an initial sense of which types of municipalities would engage in a RCA, I use the data on average sale price by municipality to determine the direction in which housing units are being sent. In this way, I can examine if higher-income municipalities are more commonly sending part of their affordable housing requirement to lower-income municipalities. To this end, I first calculate the percentile of the average sale price (compared to the rest of the state in that year) for each sending and receiving municipality in the RCA data, and then find the difference in this percentile between the sending municipality and the receiving municipality. 59 Figure 5 shows the distribution of these calculations using data from RCAs that occurred between 1994 and 2008, since sale price information is not available in digital format prior to I note that while this data does not line up precisely with the other data on the first and second round requirements, the general sending/receiving trend of RCAs was unlikely to be different between their inception in 1987 and their repeal in These calculations were all done within-year. For example, if Municipality A sent units to Municipality B in 1998, I first calculated the percentile of Municipality A and B s average sale price according to the state-wide information available from the NJ Department of the Treasury for the year I then found the difference between these two percentiles as sender receiver, or in this case as the percentile for Municipality A the percentile for Municipality B. 73

74 Figure 5 Clearly the majority of RCAs occurring between 1994 and 2008 involved a municipality with a higher average sale price sending part of its affordable housing requirement to a municipality with a lower average sale price. Among those municipalities that sent housing units, most of them fell above the 60th percentile in terms of average sale price. For municipalities that received housing units, there is a more even spread to the distribution, but still a bias towards the lower end of the price spectrum. To investigate this trend further, I again use the fair share obligation data found in COAH s 2002 Annual Report, and I utilize a multinomial random utility model in the spirit of McFadden (1974) where a municipality will choose to send housing units, receive housing units, or neither depending on which action generates the greatest predicted utility. Specifically, 74

75 municipality i will choose option j if Uj > Un where n is all other options that are not j. I estimate the predicted utility of these options using a multinomial probit model that takes the following form: P(y i = 1 x i ) = Φ(β 0 + β 1 ln(inc i ) + β 2 ln(hshlds i ) + β 3 ln(nc i ) + β 4 ln(rehab i ) + β 5 UNEMP i + β 6 BLACK i + β 7 COURT i + X i) (5) where yi is a variable that takes a value of 0 if a municipality did not engage in an RCA in the first or second round, 1 if it was an RCA receiver, and 2 if it was an RCA sender. Constructing the dependent variable in this way allows for municipalities that engaged in an RCA to be compared to those that did not as a reference group. This will help examine, for example, if the richest and poorest municipalities were the most likely to engage in an RCA of those who were under COAH s jurisdiction. All other variables are as defined in previous equations, with the addition of REHAB being the total rehabilitation obligation of municipality i between 1997 and Only municipalities that joined COAH in either the first or second round are included in the regression. Table 7 below shows the results of this model. Based on the results, we can see that the richest and poorest municipalities were most likely to engage in an RCA. A Wald test of joint significance between the coefficients from Sending Units and the additive inverse of the coefficients from Receiving Units, excluding the court dummy, confirms that their difference is not statistically different from 0 with a p-value of On the other hand, a similar Wald test between the coefficients on the court dummy for sending and receiving municipalities rejects the null hypothesis that they are additive inverses with a p- 75

76 value of This appears to indicate that the decision of whether or not to send or receive RCA units was made along two dimensions, which are being captured by the included variables (one by the court dummy and one by the other included variables). Given that municipalities under a courtorder had their own legal and political strife with which to contend, both with the courts and residents, it is not surprising that there is a separate latent variable at work affecting the decisions of these municipalities. We can also see in Table 7 that there is a clear trend for the highest-income municipalities being the senders in an RCA, and the lowest-income municipalities being the receivers, as indicated by the opposing signs on median household income. The COAH region dummy variables are also statistically significant when looking at the results of a likelihood-ratio test, again illustrating that municipalities considered the demographics of other municipalities within their region when choosing to engage in an RCA. 60 Hence, there is a strong possibility that allowing RCAs to occur undermined any potential for integration of the rich and poor within municipalities. 60 The calculated value of this likelihood-ratio test was 54.92, with a test p-value of <

77 Table 7 Coefficients Avg Mfx ⱡ Coefficients Avg Mfx ⱡ Median Household Income in 1000s *** *** *** *** (0.0120) (0.0273) Total No. of Households in 1000s * * (0.0467) (0.0471) New Construction Obligation *** *** ** *** ( ) ( ) Rehabilitation Obligation ( ) ( ) % Unemployment Rate ** ** (0.110) (0.0162) (0.134) % Black Share of Total Population (0.0143) (0.0155) Court-ordered Participation * * (0.307) (0.770) Constant *** (0.943) (1.708) Observations 406 Wald Chi 2 p-value < Regional Dummeis YES Source: COAH 2002 Annual Report, 1990 Census ⱡ Average marginal effects for each covariate Robust standard errors in parentheses * p<0.10 ** p<0.05 *** p<0.01, two-tailed test Reference Group: No RCA Participation Sending Units Receiving Units 77

78 Actual Correct Predictions Predicted Nothing Send Receive Nothing 236 (76%) 23 7 Send (66%) 0 Receive (76%) Consider that a municipality was able to send up to 50 percent of its affordable housing share to another municipality, in addition to being able to allocate up to 25 percent its affordable housing share to senior housing, 61 a substantial amount of a municipality s new construction requirement that could be allocated in such a way as to prevent serving the low and moderate income population, the target demographic of the Mount Laurel decisions. Couple that with the low compliance rate for the construction of new affordable housing across this period, and it is clear that higher-income municipalities had many avenues to avoid having to build affordable housing within their borders. So, while funds might have been flowing to poorer municipalities, it might have been a better idea to require all municipalities to build their required number of affordable housing units. Other work (such as Rossi and Weber, 1996), has shown that low-income families have better long-term economic and social outcomes if they are integrated into higher-income neighborhoods. If COAH had abandoned the use of RCAs and strictly enforced the building requirements in each municipality, there might have been more opportunities for economic success for the lower-income families of New Jersey. In considering the other variables from Table 7, again a municipality s new construction obligation is statistically significant. Municipalities that had large new construction obligations were much more likely to send housing units though an RCA, and were much less likely to receive units. This is consistent with the idea that municipalities which had to construct many new COAH Annual Report. 78

79 affordable housing units based on COAH s calculations had the largest incentives to send part of this obligation to another municipality, and that municipalities with already large new construction obligations did not want the additional burden of having to build more housing units. Although COAH, as previously mentioned, seemed to believe that the funds transacted through an RCA would at least partially offset the cost of building these units, if that were true then municipalities with large new construction obligations might not have deterred from receiving more units if the funds would have proven useful. However, it could also be an issue of space and zoning, as municipalities might not have wanted to make an already large job of zoning for all of these new affordable housing units ever larger. Interestingly, the unemployment rate in a municipality is statistically significant for municipalities that received housing units, but not for those that sent units. It could be that municipalities receiving units through an RCA believed that they could create jobs through construction projects, but it is difficult to say without more data. Although BLACK is not statistically significant, a likelihood-ratio test confirms that the unemployment rate and black share of the population within a municipality are jointly significant. Though there is evidence of some covariance between these two variables, there is not enough to worry about affecting variance estimates. 62 Lastly, the court-ordered involvement dummy variable is statistically significant for municipalities receiving units but not those sending units, yet in both cases the point estimate is negative. A likelihood-ratio test confirms that this dummy variable is statistically important to the model. 63 This result and the negative sign on the coefficient is consistent with the notion that 62 A simple univariate regression reveals that about 28.1 percent of the variation in the unemployment rate by municipality is explained by the black share of the municipality s population, according to the adjusted R-squared. Also, the variance inflation factor (VIF) between the unemployment rate and the black share of the population is small (1.39), indicating that there is little evidence of a serious multicollinearity issue. 63 The calculated value of this likelihood-ratio test was 5.89, with a p-value of

80 municipalities who were forced to join COAH did not have a large interest in engaging in an RCA, and at the very least did not want to receive even more affordable housing units from another municipality. According to the data, about 71 percent of municipalities under a court-order did not engage in an RCA, 26 percent sent housing units, while only three percent received housing units. Recall also that municipalities under a court-order were also less likely to complete new construction or rehabilitation projects, further evidence that these municipalities employed a noninvolvement mentality whenever possible. 7. Conclusion The original goal of the Mount Laurel decision was to prevent municipalities from engaging in exclusionary zoning, a direct way municipalities could segregate their communities based on income, and indirectly by race. However, although the goal was noble, the implementation of that goal as carried out by COAH appears to have been flawed from its inception. By design, municipalities were given an escape route from building the affordable housing that was required of them by COAH. It is clear that higher-income municipalities with large new affordable housing obligations had the greatest incentive to join COAH, most likely because they believed that they were the most vulnerable targets of a builder s remedy lawsuit. If they were to lose such a lawsuit, having to build all of their new construction obligation of affordable units on top of building four times as many market-value units could prove too much for a municipality in terms of time and cost. In addition, a municipality would then not be able to choose where it built those units, which would not go over well with a municipality s residents if they wanted to keep their community segregated from lower-income families. Furthermore, based on the Heckman estimator results from Section 80

81 5, municipalities that perceived the greatest threat of litigation were also the most likely to complete new affordable housing construction. On the other hand, the highest-income municipalities in the state were most likely to send affordable housing units to another municipality within their region, and the poorest municipalities were the most likely to receive those units. This illustrates a clear pattern of affordable housing flowing to the poorest municipalities in each COAH region which, while providing some amount of funds to those receiving municipalities, undermines any attempt at integrating low-income families into higher-income communities. This was potentially exacerbated by the fact that municipalities with larger new construction obligations were more likely to send some of those units to another municipality. Moreover, municipalities that volunteered to join COAH were much more likely to be the ones engaging in an RCA, as opposed to municipalities that were courtordered to join COAH. Coupled with the low compliance rates in terms of each municipality s new construction and rehabilitation obligation, it is clear that many municipalities actively avoided meeting their end of their agreements with COAH. As such, the Mount Laurel decision s goal seemed doomed to fail from the start. The richest municipalities sent their affordable housing units to poor municipalities, actively working against integration within those communities. Although municipalities were not allowed to engage in exclusionary zoning anymore by law, rich municipalities found new ways to segregate their communities through RCAs and non-compliance. If COAH had never implemented RCAs and instead strictly enforced their building requirements in all municipalities, then perhaps there could have been more opportunities for the integration of low-income families into higher-income neighborhoods across New Jersey. As these results show, however, rich municipalities will always 81

82 act upon an incentive to restrict the building of affordable housing within their borders when they can. Future work will focus in more detail on where these affordable housing units were built using GIS analysis, including those units that were and were not part of an RCA. Using this data I plan to expand upon this issue and further examine if affordable housing built as a result of the Mount Laurel decision did anything to remove the economic segregation of the rich and poor within New Jersey municipalities. References Alright, L., E. Derickson, and D. Massey, Do Affordable Housing Projects Harm Suburban Communities? Crime, Property Values, and Taxes in Mount Laurel, New Jersey. City and Community. Bailin, N, and E. Eisdorfer, The Impact of the Mt. Laurel Initiatives: An Analysis of the Characteristics of Applicants and Occupants. West Orange, NJ: Seton Hall University Center for Public Service. Fair Share Housing Center, Mount Laurel Doctrine, Fair Share Housing Center. < Fox, R., The Selling Out of Mount Laurel: Regional Contribution Agreements In New Jersey s Fair Housing Act, Fordham Urban Law Journal 16: 535. Gianetti, C., Third Time s the Charm?: The Mount Laurel Solution to Exclusionary Zoning, Quinnipiac Law Review. Heckman, J., Sample selection bias as a specification error, Econometrica 47: Krip, D., J. Dwyer, and L. Rosenthal, Our Town: Race, Housing, and the Soul of Suburbia. New Brunswick, NJ: Rutgers University Press. Massey, D., L. Albright, R. Casciano, E. Derickson, and D. Kinswy, Climbing Mount Laurel: The Struggle for Affordable Housing and Social Mobility in an American Suburb. Princeton, NJ: Princeton University Press. 82

83 McFadden, D., 1974, Conditional Logit Analysis of Qualitiative Choice Behavior, In P. Zarembka, ed., Frontiers in Econometrics, New York: Academic Press, Minnesota Population Center. National Historical Geographic Information System: Version 2.0. Minneapolis, MN: University of Minnesota < New Jersey Council on Affordable Housing, Annual Report, Trenton, NJ: New Jersey State Library Archives New Jersey Council on Affordable Housing, COAH: Opening Doors to Opportunity, Trenton, NJ: New Jersey State Library Archives New Jersey Council on Affordable Housing, People Places & Progress: The COAH Story , Trenton, NJ: New Jersey State Library Archives New Jersey Council on Affordable Housing, NJ Council on Affordable Housing Annual Report : Designing Affordable Housing, Trenton, NJ: New Jersey State Library Archives New Jersey Council on Affordable Housing, COAH 2002 Annual Report: Making NJ a Better Place to Live, Trenton, NJ: New Jersey State Library Archives New Jersey Council on Affordable Housing, Annual Report , Trenton, NJ: New Jersey State Library Archives Rossi, P., and E. Weber, What Can We Learn about Neighborhood Effects from the Moving to Opportunity Experiment? Housing Policy Debate 7(1). Rutgers, the State University of New Jersey, History, The New Jersey Digital Legal Library. < Segal, L., and D. Sullivan, Trends in Homeownership: Race, Demographics, and Income, Economic Perspectives 2: State of New Jersey. Department of Community Affairs. Trenton, NJ. State of New Jersey < Wooldridge, J Introductory Econometrics: A Modern Approach, Fifth Edition, South- Western College Pub. 83

84 Appendix As mentioned in Section 4, a regression of a municipality s new construction and rehabilitation obligations on several explanatory variables reveals that much of the basis of the new construction requirement remains unclear, while the rehabilitation obligations seem to be based mostly on population. The total number of persons below the poverty line is used as a rough proxy for the available stock of affordable housing in a municipality, since assumedly many of these people would be living in, or qualify, for some type of affordable housing. I also ran each model using the poverty rate (using the data on total population by municipality), but found that it explained much less variation in both a municipality s new construction and rehabilitation obligation. The results of these regressions may be found in Table 8 and Table 9. 84

85 Table 8 Dependent Variable: New Construction Obligation (1) (2) (3) (4) (5) (6) Total No. of Households in 1000s 10.76** 32.79*** 32.34*** 32.35*** 32.33*** 34.15*** (4.397) (4.574) (4.522) (4.516) (4.540) (4.303) Total No. of Persons below Poverty line in 100s *** *** *** *** *** (0.894) (0.850) (0.843) (0.830) (0.760) Median Household Income in 1000s 2.473*** 2.576*** 2.576*** 4.558*** (0.564) (0.579) (0.583) (0.778) % Unemployment Rate (4.281) (3.783) (4.046) % Black Share of Total Population (1.483) (1.416) Constant 99.02*** 40.17*** ** ** ** *** (17.77) (12.86) (29.80) (35.37) (36.23) (54.19) Observations Adjusted R-squared Regional Dummies NO NO NO NO NO YES Source: 2002 COAH Report, 1990 Decennial Census Robust standard errors in parentheses * p<0.10 ** p<0.05 *** p<0.01, two-tailed test 85

86 Table 9 Dependent Variable: Rehabilitation Obligation (1) (2) (3) (4) (5) (6) Total No. of Households in 1000s 18.79*** 7.984*** 8.000*** 7.953*** 8.024*** 8.160*** (2.772) (2.188) (2.191) (2.160) (2.152) (2.145) Total No. of Persons below Poverty line in 100s 2.426*** 2.416*** 2.517*** 2.607*** 2.582*** (0.516) (0.521) (0.528) (0.544) (0.533) Median Household Income in 1000s *** *** *** (0.142) (0.150) (0.158) (0.187) % Unemployment Rate ** (2.395) (1.894) (2.016) % Black Share of Total Population *** *** (0.461) (0.456) Constant *** *** 31.05*** 46.19*** (11.34) (7.436) (11.18) (12.03) (11.06) (14.72) Observations Adjusted R-squared Regional Dummies NO NO NO NO NO YES Source: 2002 COAH Report, 1990 Decennial Census Robust standard errors in parentheses * p<0.10 ** p<0.05 *** p<0.01, two-tailed test 86

87 Chapter 3 The Perceived Benefits of Homeownership-Focused Programs on Low-Income Families: A Pilot Study with Habitat for Humanity. 1. Introduction Homeownership has traditionally been viewed as a core tenet of the American dream, a dream that has been encouraged and subsidized by government policies for much of the country's modern existence. For instance, homeowners are able to deduct both the interest on their mortgage and any property tax payments from their federal income taxes. 64 But what has a national focus on homeownership accomplished for the average family, and is homeownership still a worthwhile goal for low-income families in particular? In the past some researchers and social organizations have noted the strong correlation between income and homeownership (for example, see Segal and Sullivan 1998), though any causality underlying this relationship is still debated. Given the attention brought to this potential link, however, as well as the recent national focus on income inequality within in United States, the past and present emphasis placed on homeownership comes as no surprise. Though to uncover the potential positive benefits associated with homeownership, more research is clearly needed. Although there are many organizations that focus on reducing poverty through homeownership, Habitat for Humanity, a U.S.-based charity that builds houses for low-income families, provides a unique opportunity to study the population in need of affordable housing as well as the potential impact a program focused on homeownership has on low-income families. 64 Homeowners are also able to receive several other benefits, such as not being taxed on the imputed rental income they receive, or on their capital gains from the sale of their house (to a limit). See a Q&A by the Tax Policy Center for more information. < 87

88 Using a new set of data collected through a pilot study in conjunction with several Habitat affiliates in New Jersey, I employ a quasi-experimental design to compare those who were selected into the Habitat housing program with those who applied for housing but were denied. I do this to study several key aspects of the Habitat for Humanity program. In addition to examining what population Habitat for Humanity s affordable housing program serves, I am able to test the perceived changes in an applicant s quality of life and living conditions in light of three joint benefits provided by the Habitat program, specifically: becoming a homeowner, upgrading the quality of one s residence, and receiving a positive wealth transfer. 65 Furthermore, since those who move into Habitat houses stay in the same general area as their original residence, it makes the program ideal for studying these perceptions apart from the influence of neighborhood characteristics, which has been a complicating factor in many previous studies. The interpretation of these results will rely on several key assumptions about the data that will be outlined in Section 4; however, as a pilot study this data provides an opportunity to evaluate the need for research into homeownership-focused programs such as Habitat for Humanity. I find that among the affiliates that participated in this study, there is a skew among those accepted into the Habitat housing program towards those who are married as well as those who have a high school level of education in comparison to the pool of those who were not selected for the program. I also find that Habitat homeowners report more positive changes in their family s life since the time of application compared to denied applicants, and that further research into homeownership-focused programs such as Habitat for Humanity is warranted. 65 The wealth transfer stems the house being a substantial asset selected applicants have acquired at a low cost (compared to market value, as explained later), as well as Habitat s goal of lowering an applicant s monthly incometo-rent ratio. 88

89 The rest of this paper proceeds as follows. Section 2 briefly discusses some previous works that study the potential benefits of homeownership for low-income families. Section 3 outlines more details about Habitat for Humanity as an organization, its housing program, and its program philosophy. Section 4 describes the survey and experimental design utilized to collect the data used for this paper, as well as the key assumptions required for inference. Section 5 compares participating Habitat homeowners to the population that Habitat serves, as represented by the denied applicant pool. Section 6 examines how Habitat homeowners perceive their lives have changed since moving into Habitat houses, as well as their general evaluation of their houses and neighborhoods, and Section 7 concludes. 2. Previous Studies One aspect of the relationship between income and homeownership is something that many have experienced first-hand. As families earn more, they are able to either buy a first house, or buy a better quality house. However, whether or not homeownership itself causes any positive changes in the lives of individuals and families is still unclear. While other studies have employed a quasi-experimental design in an attempt to measure the effect of moving a low-income family into new housing, with Moving to Opportunity (MTO) remaining one of the largest and most widely-cited examples, 66 these studies have typically focused on combined housing and neighborhood effects. Moreover, much of this work focuses on the effect of moving low-income families into more stable rental housing, not distinguishing between any potential benefits that stem specifically from renting versus owning the residence. 66 For instance, see follow-up papers on the Moving to Opportunity experiment such as Rossi and Weber (1996). 89

90 On the other hand, several studies have tried to isolate the impact of homeownership itself apart from any neighborhood effects, though doing so remains a challenge. A number of studies from the late 1990s and early 2000s found that homeowners enjoyed positive economic and social benefits compared to renters even after controlling for family differences (Green and Hendershott 2001; Dietz 2003). Other studies have found that children of homeowners showed better academic performance compared to children of renters (Green and White 1997; Haurin, Parcel, and Haurin 2003). The general estimation strategy in these papers was to examine an outcome variable of interest, include a dummy variable for whether a member of a family or an entire family rented or owned their residence, and included as many additional control variables as possible on the righthand side. However, it remained difficult in these studies to account for geographical and neighborhood effects, as well as to account for unobserved heterogeneity between homeowners and renters. As such, more recent studies have begun to question the strength of some of these previous findings, leading to a debate over the existence of any potential benefits stemming from the socalled homeownership effect (Barker and Miller 2009; Mohanty and Raut 2009; Holupka and Newman 2010). Thanks to several key aspects of Habitat for Humanity s housing program, including the fact that housing recipients are purported to remain in the same general area, using a quasi-experimental design to study a homeownership-focused program could yield additional insights as to the potential benefits of owning a house (along with other aspects of the program). 90

91 3. Habitat for Humanity As previously mentioned, Habitat for Humanity s core mission is to build houses for lowincome families in need. 67 Potential Habitat housing recipients must have an income that falls between 25 to 60 percent of the median income for their affiliate's serving area, but other than the income requirement, the final judgement of a family being in need is determined by each Habitat affiliate on a case-by-case basis. Although all Habitat for Humanity affiliates are under the same organizational umbrella, in examining their selection processes and from conversations with several affiliate directors, each affiliate s selection process and funding is a function of each individual affiliate. Since affiliates receive little financial support from the central branch of the organization, they frequently rely on donations of money, land, and materials. As a result, some affiliates have more resources at their disposal than others. 68 In speaking with local affiliate directors as well as managers at Habitat for Humanity s main office in Atlanta, Georgia, they consistency made it clear that Habitat is not giving away a handout, and is instead providing low-income families an opportunity to be homeowners through a partnership with Habitat for Humanity. However the relationship between the applicant and Habitat is labeled, those selected into the Habitat program do receive what is in essence a large positive wealth transfer. Habitat applicants must be able to make monthly mortgage payments of no more than thirty percent of their monthly income directly to Habitat at a no interest, or zero 67 Habitat for Humanity U.S. Affiliated Organization Policy Handbook. 68 The author can personally attest to this asymmetry after having viewed large differences in the aesthetics and upkeep of the different participating affiliate offices. 91

92 percent, mortgage rate. 69 Although the house is not costless for the applicant, it is still a much more favorable mortgage than most would receive on the open market. Moreover, unlike other housing assistance programs, Habitat housing recipients must also contribute sweat equity in addition to making their monthly mortgage payment. 70 This process involves logging at least 200 hours towards building either one's own house, or the house of another Habitat housing recipient. Habitat s Organizational Policy Handbook describes the rationale of sweat equity hours through the lens of both a life of service to others as outlined by the Christian faith (though Habitat accepts applicants of all faiths), as well as increased feelings of self-sufficiency. 71 Conversations with several directors at participating Habitat affiliates also revealed that some affiliates allow children to contribute to sweat equity requirements by reducing the required number of service hours if children submit a report card with good grades. Many who are accepted into the program either lived in transient housing or lived with a relative or friend prior to moving into their Habitat houses. Some who are accepted by the program already live in houses, though they may be considered inadequate for health or space reasons. However, only a single Habitat homeowner in the sample I collected reported that she owned her prior residence. Lastly, as previously mentioned Habitat housing recipients typically stay in the same general area as their original transitional housing. In speaking with the directors of the participating affiliates, they explain that this generally happens because: 1) it allows applicants to remain a part of their familiar community (though the merits of this based on the results of studies such as MTO are unclear); and 2) since many affiliates rely on donations of land, these donations 69 Several directors at participating affiliates indicated that the mortgage amount typically reflects the cost of the materials to build the house since the land is generally donated, though this is not an official rule, and in general the goal is simply to keep an applicant s rent-to-income ratio at or under the 30 percent threshold. 70 Habitat for Humanity, Affiliate Operations Manual Family Selection, 2012: US Affiliated Organization Policy Handbook 2013: 29 92

93 tend to come from only a few geographic areas within an affiliate region. Whatever the reason may be, it provides a unique opportunity to remove any neighborhood effects from the analysis, allowing me to focus on the other joint benefits those selected into the Habitat housing program receive: becoming a homeowner, improving the quality of one s residence, and receiving a positive wealth transfer. 4. Data Collection and Methodology The data collection was conducted over a two-year period with permission from Habitat for Humanity s central office in Atlanta, 72 and in conjunction with another graduate student in Northeastern University s Department of Sociology and Anthropology. Through both a mailed and online survey, we gathered information from three different groups of Habitat applicants from the participating affiliates: 1) those who had been accepted into the Habitat program and were currently living in a Habitat house (referred to as Habitat homeowners ); 2) those who had been accepted into the Habitat program but had yet to move into their new homes (referred to as future Habitat homeowners ); and 3) those who had applied for a house through Habitat but were denied (referred to as denied applicants ). This will allow me to examine how Habitat homeowners have benefited from the program by comparing those who were selected into the program to those who were not, making it possible to form a plausible counterfactual for Habitat homeowners and conduct an accurate measure of the impact of homeownership for this group. 72 Specifically, we received expressed written permission from Pat Decker, the director of all U.S.-based Habitat operations, to contact all the affiliates in New Jersey requesting their participation in this study. This permission was granted after both she, and Habitat s on-staff legal team, reviewed our study and approved its design as well as ultimately the survey itself. This was in addition to the usual permission obtained from Northeastern University s Institutional Review Board (IRB). 93

94 Though we received permission to contact all the Habitat affiliates in New Jersey, 73 ultimately, data was collected from the above mentioned groups from five New Jersey Habitat affiliates. 74 In order to ensure that the survey was in line with the confidentiality standards of both Habitat for Humanity as well as Northeastern University s Institutional Review Board we, as the researchers, were not allowed to know any identifying information about any potential survey participants until they signed and returned the consent form included with the survey itself. Hence, members of the participating Habitat affiliates were responsible for the actual mailing of the surveys, although we could provide certain materials to participating affiliates to assist them in this process. As a result, however, we are unable to know any information about those who did not consent to take part in the study. Since the information on those who did not consent to participate in this study is most certainly not missing at random, ideally even some general measure, such as the average income of those who did not respond, would be invaluable for evaluating how severe the selection problem is in the data. Nonetheless, due to confidentiality issues, this information remains unavailable. The survey was conducted between April and May of Study participants were given six weeks to respond either by mailing the consent form and completed survey in the provided envelope, or by following the enclosed web address to complete the consent form and survey online. The online version of the survey attempted to replicate the paper version in every respect possible, and was created using tools in Google documents and was hosted through Yahoo 73 New Jersey was selected to conduct this pilot for several reasons. First, since my colleague knew one of the directors of a New Jersey Habitat affiliate personally, we believed that we could rely on this director to help introduce us and our study to the other directors in the state. Second, my colleague and I are both from New Jersey and have family that still lives there, and New Jersey is a relatively small state geographically. Since this study required numerous in-person meetings with personnel at the participating affiliates, both of these factors made travel arrangements significantly easier. New Jersey also has an incredibly diverse population within its relatively small borders, providing the variation needed for our sample. 74 I had originally reached out to all 17 Habitat for Humanity affiliates in New Jersey after receiving written permission from Pat Decker. Nine affiliates had originally expressed interest in participating in the study, but only five were able to ultimately participate due to logistic issues and changes in affiliate directors. 94

95 Domains. As an additional incentive to complete the survey, all consenting participants were entered into a random drawing for one of two $100 American Express gift cards. Please see Appendix to view copies of the consent forms and surveys mailed to all three study groups. In order to preserve the confidentiality and privacy of the study participants, completed consent forms and surveys were returned directly to Northeastern University. At the completion of the data collection period, we provided a list of names to each affiliate indicating who had consented to be part of the study so that we could receive a copy of their original applications. The consent form contained clear language that this was a study being conducted by graduate researchers at Northeastern University and not affiliated with Habitat for Humanity. It further mentioned that all consent forms and surveys would never be viewed by anyone at Habitat for Humanity, only the student researchers and our advisors. Nonetheless, it is still a concern that study participants may have incorrectly concluded that Habitat for Humanity was directly involved in the study, and thus may have influenced their willingness to participate or influenced some of their answers regarding their general feelings towards Habitat for Humanity. The survey asked participants general questions about their experiences with Habitat, information about their houses and neighborhoods, perceptions of how they believe their lives had changed since applying for a Habitat house, as well as some more specific questions related to assets, expenses, employment information, community engagement, and children's grades and behaviors. When appropriate, the survey asked questions pertaining to current information and information from the time of application, in order to collect pre- and post-application information. For example, a participant was asked how many hours per week they work on average currently, as well as how many hours per week they worked on average at the time of application to Habitat. Whenever possible, pre-application information was collected from the participant s original 95

96 application to Habitat (which they consented to release). The application contains personal demographic information in addition to information about employment, income, and savings. 75 Since Habitat requires some form of validation for all information listed on the application (tax returns, pay stubs, etc.), this is a more credible source of pre-application information when available as it is less subject to measurement error. Table 1 shows survey response rate information for all three groups taking part in the study, including both the mailed and online survey. Because of the small number of study participants, affiliate names have been omitted for purposes of confidentiality and anonymity. As seen from the table, the five participating affiliates differed in both size and response rates. For both current and future Habitat homeowners, there was a more favorable response rate compared to denied applicants. The low response rate for denied applicants, though disappointing, is not surprising and likely stems from: 1) a feeling of bitterness or resentment over not being selected for a Habitat houses; and/or 2) the denied applicant incorrectly believing that the survey was being conducted by Habitat for Humanity. These potential negative feelings towards Habitat almost certainly contributed to the low response rate for this group. Moreover, surveys sent to denied applicants relied on the last address of the applicant known to Habitat, and some of those applicants had since moved (those that were returned marked return-to-sender ). 75 The directors of the participating affiliates also indicated that a credit report is run on all program applicants, though we were unable to receive the results of these reports. 96

97 Affiliate Surveys Sent Table 1 Habitat Homeowners Future Habitat Homeowners Denied Applicants Surveys Received Response Rate Surveys Sent Surveys Received Response Rate Surveys Sent Surveys Received Response Rate Without Return-to- Sender # % % % 7.94% # % % % 8.60% # % % % 50.00% # % % % 5.36% # % % % 0.00% Total: % % % 6.36% 97

98 Due to the small sample and response rate, there are several potential sources of bias which may make interpretation of results difficult. First, there is likely self-selection in both those who responded to the survey, as well as in which questions a particular respondent decided to answer. Second, there is likely heterogeneity among those in the treatment and comparison groups that is unobservable to the researcher, but is observable to the employees of each Habitat affiliate. Since the ultimate determination of who is in the most need of a house is determined by each affiliate (as long as the applicant meets the income requirement), there are likely aspects of this decision that are known to employees at that affiliate, but unknown to the researcher. Lastly, there is always the potential for measurement error when relying on self-reported data, particularly when asking respondents for information about past events (though I attempt to mitigate this by using information from the applicant s original application to Habitat whenever possible). As such, I make several strong assumptions about the data in order to preliminarily interpret the results presented in this paper. First, I assume that any missing data (both in terms of participation and question response) is missing at random, and that study participants are drawn at random from the population Habitat serves. Second, I assume that there is no unobserved heterogeneity in the applicant selection process, and that housing applicants are drawn at random as long as they meet Habitat s income requirement. Third, I assume that respondents are able to perfectly recall both current and past information, and that question responses contain no measurement error. Though these assumptions are most certainly violated, they will allow me to use the results of this paper to evaluate the potential for future research into Habitat s program. If across the board there are no statistically significant results when comparing Habitat homeowners and denied applicants, then further research into the program may not be needed. However, if there are statically significant results, although their direct interpretation is questionable, it 98

99 provides evidence that further research into the program is warranted to examine the potential benefits associated with it. 5. Examining the Habitat for Humanity Selection Process To reiterate, Habitat for Humanity considers Habitat homeowners as entering into a partnership with the organization. As such, one initial question that Habitat homeowners and future Habitat homeowners were asked pertained to their satisfaction with this partnership, the results of which may be seen in Figures 1 and 2. Overall, Habitat homeowners in the sample seemed to have a positive outlook regarding their partnership with Habitat, with no participant in the sample responding that they were Very dissatisfied. However, similar to the issue described concerning the low response rate for the denied applicant group, it is very possible that this is a skewed distribution based on this sample, as one might expect those with most positive overall feelings about Habitat to be more likely to participate in this study. Concerning future Habitat homeowners, there are very positive feelings about the partnership, though this reflects how the participants feel so far considering that they have not yet moved into their new houses. Nevertheless, this distribution could reflect the optimism felt among these soon-to-be Habitat homeowners, although again it is possible that participants had difficulty understanding that the survey was independent of Habitat, and future Habitat homeowners feared revealing negative opinions of Habitat for Humanity prior to moving into their new homes. 99

100 100.00% 90.00% 80.00% 70.00% 60.00% 50.00% 40.00% 30.00% 20.00% 10.00% 0.00% Figure 1 Opinion of Partnership with Habitat Habitat Homeowners Very dissatisfied Somewhat dissatisfied Neutral Somewhat satisfied Very satisfied % 90.00% 80.00% 70.00% 60.00% 50.00% 40.00% 30.00% 20.00% 10.00% 0.00% Figure 2 Opinion of Partnership with Habitat Future Habitat Homeowners Very dissatisfied Somewhat dissatisfied Neutral Somewhat satisfied Very satisfied Considering that Habitat affiliates select those with the greatest need for affordable housing, another relevant question to examine is how the sample of selected housing applicants compares to the population that Habitat for Humanity serves, which is assumed to be represented by the denied applicant pool. Although, as mentioned previously, Habitat does have minimum and maximum income requirements for applicants, these will aid in examining if there is an over or under-representation of a certain demographic category among the Habitat homeowners as 100

101 compared to the denied applicants; however, these will not help in determining overall sample selection within the treatment groups or control group. To this end, I employ a binary probit model that takes the following form: P(y i = 1 x i ) = Φ(β 0 + β 1 DEMO i + β 2 INC i + X i) (1) where y is equal to 1 if applicant i is a Habitat homeowner or a future Habitat homeowner and 0 if applicant i is a denied applicant; β 1 is a confirming vector of coefficients on DEMO, a vector of demographic information for applicant i at the time of application including variables such as age and race, as well as marital and educational status. INC is the real weekly income of applicant i at the time of application, and X is a vector of Habitat affiliate control dummies. It is worth noting that in the survey measures of race and ethnicity were combined into a single question asking the participant to select the category that best describes themselves. 76 For presentation purposes, the responses to this question are presented as describing the applicants race. Since each observation comes from a single application to Habitat, the information used for Equation (1) comes from whoever is listed as the primary applicant. 77 Because of the differing 76 This approach has seen some growing support in recent years as opposed to asking separate questions about race and ethnicity when collecting survey data. The U.S. Census Bureau itself in fact is currently deciding how to best present demographic questions of race to survey participants. The 2020 Decennial Census could employ this type of strategy of asking respondents to describe themselves instead of inquiring as to their race and ethnicity separately. The National Content Test is also using this new approach. For more information, please refer to the sources outlined in an article published by the Pew Research Center. < Accessed July 20, The categories included in the survey were African American/Black, American Indian, Asian or Pacific Islander, Hispanic/Latino, and White/Caucasian. 77 Each affiliate s application form distinguishes between the primary applicant and any co-applicants. Even married couples list a primary and a co-applicant. In the large majority of cases in this sample, the person listed as the primary applicant was either the person who was employed if the co-applicant was not, or the person who had a greater weekly income or worked full-time. 101

102 group response rates as outlined in Table 1, inverse probability weights are used to weight each observation by the overall response rate of their group. This will allow me to weight up the comparison group response by allowing each denied applicant observation to represent a relatively larger group of people drawn from that population. 78 Table 2 reports the output of Equation (1). 78 Specifically, observations for Habitat homeowners are assigned a weight of 1/.1737, future Habitat homeowners of 1/.4828, and denied applicants of 1/

103 Coefficients Avg Mfx ⱡ Coefficients Avg Mfx ⱡ Age (0.0208) ( ) (0.0227) ( ) Born in U.S ** ** ** ** (0.814) (0.178) (0.970) (0.195) Children at Application (0.439) (0.106) (0.451) (0.104) Real Weekly Income (in 100s) (2014 dollars) (0.0576) (0.0141) (0.0611) (0.0144) Race (Ref: White) Black * (0.579) (0.166) (0.590) (0.169) Hispanic ** *** ** *** (0.897) (0.148) (1.059) (0.147) Marital Status (Ref: Single) Married 1.058** 0.287** 1.064* 0.280** (0.516) (0.130) (0.570) (0.138) Divorced, separated, or widowed (0.479) (0.113) (0.502) (0.112) Education (Ref: < High school) High school 1.841** 0.398*** 1.886** 0.406*** (0.859) (0.120) (0.957) (0.129) Some college (0.848) (0.121) (0.921) (0.126) College (0.856) (0.119) (0.927) (0.123) Constant (1.637) (1.822) Observations Pseudo R-squared Affiliate Dummies YES YES Model (1) includes current and future Habitat Homeowners in treatment group Model (2) excludes future Habitat homeowners from treatment group ⱡ Average marginal effects for each covariate All demographic information taken from "primary applicant" "Some graduate school" dropped from output for lack of observations Inverse probability weights based on group response rate included Robust standard errors in parentheses * p<0.10 ** p<0.05 *** p<0.01, two-tailed test Table 2 Dependent Variable: Treatment Dummy (1) (2) 103

104 Model (1) from Table 2 contains the full specification of Equation (1) which includes both current and future Habitat homeowners in the treatment group, while the Model (2) specification omits future Habitat homeowners from the treatment group as a robustness check. From the results we can see that even when controlling for the income of applicants and affiliate specific fixed effects, certain demographic groups are more or less likely to be represented in the treatment versus control group. Applicants that are married (compared to single applicants) in the sample are more likely to have been selected into the Habitat program when controlling for other demographic information. There also appears to be a positive skew towards selecting those with a high school level of education, but not those with higher levels of education. These results could reflect that married and more educated applicants are perceived by Habitat affiliates as having more financial stability, and that these families will have an easier time making the monthly mortgage payments to Habitat, though these positive coefficients persist even when controlling for income. Interestingly, those who were born in the United States are less likely to be in the sample when controlling for other demographics, with a similar negative likelihood for self-identified Hispanics. Based on these results, it appears that non native-born Hispanic applicants are more likely to have been selected to participant in the Habitat program than native-born Hispanic applicants. We can also see from the results of Model (2) that the coefficients change very little, indicating that these estimates are not being driven by the inclusion of future Habitat homeowners in the treatment group. As an additional robustness check, I also ran Models (1) and (2) from Table 2 using weekly household income, combing the primary and co-applicant s weekly income rather than only using the primary applicant s weekly income. I also included an additional dummy variable in this specification for applicant i having a working spouse or partner. However, neither coefficient on the newly included variables was statistically significant, nor did the original 104

105 coefficients change statistically nor substantially. Hence, though the sign of the point estimate makes sense if those in greatest need for housing have the lowest incomes (in addition to the maximum income limit), the effect of weekly income is not statistically significant in Equation (1). One might suspect that the population a specific Habitat affiliate serves is a byproduct of the population of its surrounding area. For instance, if an affiliate serves an area with a large black population, it might be expected that the affiliate would have a larger share of applicants that are black. A likelihood ratio test confirms that the affiliate dummies are jointly significant, 79 which is likely a reflection of the differing amount of funding and resources available to each affiliate. For example, an applicant might be more likely to be selected by a particular affiliate (after controlling for other differences) simply because that affiliate has more resources at its disposal and is able to build more houses in any given year. It is important to note, however, that even when controlling for these affiliate-specific effects, several key demographic variables remain statistically significant. It would be interesting to see in a full-scale, multiple state or nation-wide, study how these demographic patterns changed and how reflective they are of Habitat s selection process in relation to the demographics of its local population. 79 The calculated value of this likelihood-ratio test for Model (1) was 8.59, with a test p-value of However, they are not jointly statistically significant in Model (2) at 6.07 and a test p-value of This is likely a result of the future Habitat homeowners group being relatively small compared to the others and that the overwhelming majority of these participants were drawn from a single affiliate, as seen in Table 2. Nonetheless, this does not diminish the overall importance of the Habitat affiliate fixed effects, and is more a reflection of the limited sample used to represent this group. 105

106 6. Examining the Perceived Benefits of the Habitat for Humanity Program 6.a. House and Neighborhood Perceptions Although the sample used for this pilot study is a small and restricted, it is possible to preliminarily examine the perceived benefits of the Habitat housing program by Habitat homeowners by comparing their responses to those of denied applicants. One category of questions presented to survey participants pertained to house and neighborhood characteristics, including the following questions: Overall, how would you describe the condition of your home? 80 Overall, how would you describe the size of your lot in comparison to your neighbors? Overall, how satisfied are you with your neighborhood? To gain a sense of how Habitat homeowners currently view their houses and neighborhoods in comparison to denied applicants, I utilize an ordered probit model that takes the following form: P(y i = j x i ) = Φ(β 0 + β 1 HABITAT i + β 2 DEMO i + β 3 APP i + X i) (2) where y may take on any response value j, and j is an ordinal scale of response values whose responses are outlined in Table 4. HABITAT is a dummy variable that takes a value of 1 if applicant i is a Habitat homeowner and 0 if applicant i is a denied applicant, where β 1 is the primary coefficient of interest for determining fixed differences in group responses. DEMO is defined as in Equation (1), although for this model I use current demographic information as opposed to information at the time of application. APP is the number of years that have passed since applicant 80 Denied applicants were asked about the condition of their residence instead of their home. 106

107 i applied for a house through Habitat, 81 and X is again a vector of Habitat affiliate control dummies. The maximum likelihood coefficients from Equation (2) are shown in Table 3, while Table 4 shows the difference in the estimated average marginal effect for Habitat homeowners versus denied applicants for each response category of each question. In examining the coefficient on HABITAT in the models from Table 3, we can see that both the condition of the respondent s residence and the estimated size of the lot in comparison to that of their neighbors are both statistically significant and positive. This result in Model (1) is expected since one would assume that Habitat housing is an upgrade compared to an applicant s pre-habitat housing, which presumably is very similar to the type of residences in which denied applicants currently live. The positive coefficient on HABITAT in Model (2), on the other hand, was not initially anticipated. The directors of the Habitat affiliates involved in the study each indicated that one of their goals is to build houses for selected applicants that keep with the aesthetic of their neighbors, including similar color, shape, and size. The overall goal of which is to allow Habitat homeowners to blend in and not be as easily identified as living in affordable housing. One might therefore assume that the parcel lots of Habitat homeowners would be relatively similar to those of their neighbors. However, in observing the difference in the average marginal effects in Table 6, we can see the positive probabilities for responses of somewhat larger and much larger are relatively small (as is the mean difference), making Habitat applicants only a bit more likely to respond that their lot size is larger than denied applicants. These positive average marginal effects in this case could indicate that either Habitat is receiving donations of parcels of land that are 81 APP is included to control for time-period effects since application data spans from as early as 1988 to as late as 2013 (although most of the data comes from post-2000). In essence, APP acts as a type of linear time trend. 107

108 occasionally larger than those of surrounding houses, or that Habitat homeowners have a slight tendency to overestimate the size of their property, perhaps due to positive overall feelings about their experiences with Habitat or due to being first-time homeowners in many cases. 108

109 Dependent Variable: * p<0.10 ** p<0.05 *** p<0.01, two-tailed test 109 Table 3 (1) (2) (3) Home Condition Lot Size Neighborhood Coefficients Coefficients Coefficients Treatment Dummy 2.299*** 1.147*** (HABITAT) (0.441) (0.438) (0.336) Age (0.0239) (0.0213) (0.0182) Born in U.S *** * (0.485) (0.634) (0.487) Children at Application *** (0.725) (0.750) (0.486) Years since Application ** *** (0.0354) (0.0317) (0.0282) Race (Ref: White) Black (0.357) (0.494) (0.613) Hispanic *** (0.544) (0.691) (0.591) Marital Status (Ref: Single) Married 0.762* 0.923* (0.456) (0.489) (0.395) Divorced, separated, or widowed (0.435) (0.427) (0.301) Education (Ref: < High school) High school *** 0.907* (0.644) (0.852) (0.473) Some college *** (0.700) (0.850) (0.520) College 1.299* 3.122*** (0.768) (0.841) (0.533) Some graduate school (0.789) (1.077) (1.041) Observations Pseudo R-squared Affiliate Dummies YES YES YES All demographic information taken from "primary applicant" Inverse probability weights based on group response rate included Robust standard errors in parentheses

110 Home Condition Lot Size Neighborhood Table 4 Difference in Average Marginal Effects Habitat Homeowners vs. Denied Applicants Poor Fair Good Excellent ** *** *** *** (0.0625) (0.0811) (0.0718) (0.0581) - Much smaller Somewhat smaller About the same Somewhat larger Much larger *** * *** ** * (0.0695) (0.0304) (0.0603) (0.0603) (0.0161) Very Somewhat Somewhat Very Neutral dissatisfied dissatisfied satisfied satisfied (0.0571) (0.0332) (0.0127) (0.0226) (0.0571) ⱡ Mean difference calculated as each cell times its response value number summed by row. It is the expected value of the difference of the average response (number) between groups by question. Mean Difference ⱡ Of particular note is the fact that HABITAT in Model (3) is not statistically significant, indicating that Habitat homeowners evaluate their current neighborhoods similarly to how denied applicants evaluate them. This is the result one would expect given that one of the key elements of the Habitat for Humanity housing program is that applicants remain in the same geographic area, and this result may is reinforced by every estimated average marginal effect in the third row of Table 4. Even though Habitat homeowners are in a slightly new area (they may have moved to a new block or to a different part of their neighborhoods), in general these homeowners are in the same general area as they were before they moved into their new houses. Hence, we would expect that having a new house would not change these respondents opinions of their neighborhoods, and in fact this provides evidence that these respondents are able to distinguish between 110

111 neighborhood and non-neighborhood related benefits. This will aid in the interpretation of the next set of estimates. 6.b. Perceived Quality of Life Changes In addition to the previous questions, survey participants were asked to rate their experiences in several broad categories since the time of application (for consistency between the treatment and control groups): Overall, how do you feel your family's economic and career outcomes have changed since applying for a home? Overall, how do you feel your child(ren)'s educational experience has changed since applying for a home? Overall, how do you feel your family's level of community involvement has changed since applying for a home? Overall, how do you feel your family's life has changed since applying for a home? Survey participants were then asked to rate their experiences in these categories on a scale of 1 through To gain a sense of how Habitat homeowners currently view their lives in comparison to denied applicants, I again utilize an ordered probit model that takes the following form: P(y i = j x i ) = Φ(β 0 + β 1 HABITAT i + β 2 DEMO i + β 3 INC i + β 4 APP i + X i) (3) 82 The five point Likert scale was selected as the simplest and most accessible way for study participants to evaluate any life changes since they applied to Habitat. 111

112 where y may take on any response value j, and j is an ordinal scale of 1 through 5 as outlined in Table 6. DEMO, INC, and APP are defined as in Equation (1), although I again use current demographic and income information, and X is again a vector of Habitat affiliate control dummies. Table 5 shows the maximum likelihood coefficients from Equation (3), and Table 6 shows the difference in the estimated marginal effect estimates for Habitat homeowners versus denied applicants. 112

113 Dependent Variable: (Likert Scale 1-5) Table 5 (1) (2) (3) (4) Economic Children's Situation Education Life Improvement 113 Community Engagement Coefficients Coefficients Coefficients Coefficients Treatment Dummy 3.137*** 1.439*** 0.932* 4.310** (HABITAT) (0.676) (0.454) (0.545) (1.786) Age * (0.0357) (0.0209) (0.0262) (0.0533) Born in U.S ** ** (1.021) (0.637) (0.596) (2.115) Children at Application ** *** 3.826*** (1.484) (0.532) (1.240) (1.080) Weekly Income (in 100s) ** (0.0846) (0.0540) (0.0781) (0.153) Years since Application *** ** (0.0460) (0.0388) (0.0405) (0.125) Race (Ref: White) Black ** (0.678) (1.139) (0.992) (0.837) Hispanic * (0.865) (0.883) (0.842) (1.864) Marital Status (Ref: Single) Married 1.497** 0.931* *** (0.671) (0.558) (0.606) (0.601) Divorced, separated, *** or widowed (0.607) (0.571) (0.582) (1.208) Education (Ref: < High school) High school 5.168*** 4.047*** 2.618* (1.355) (0.941) (1.383) (1.369) Some college 4.917*** 5.657*** ** (1.252) (1.049) (1.393) (1.243) College 4.747*** 4.269*** (1.196) (1.093) (1.330) (1.117) Some graduate school 7.388*** 6.118*** (1.429) (1.470) (1.859) (1.210) Observations Pseudo R-squared Affiliate Dummies YES YES YES YES All demographic information taken from "primary applicant" Inverse probability weights based on group response rate included Robust standard errors in parentheses * p<0.10 ** p<0.05 *** p<0.01, two-tailed test

114 Life Improvement Economic Situation Children's Education Community Engagement Significantly worsened Table 6 Difference in Average Marginal Effects Habitat Homeowners vs. Denied Applicants Somewhat worsened No change Somewhat improved Significantly improved *** * *** (0.1035) (0.0947) (0.1134) (0.0647) (0.0678) ** ** *** *** (0.0596) (0.0606) (0.0488) (0.0421) (0.0541) * * (0.0197) (0.0729) (0.0400) (0.0284) (0.0714) Significantly Somewhat Somewhat Significantly No change decreased decreased increased increased ** ** ** * ** (0.0931) (0.1628) (0.1457) (0.0524) (0.0976) ⱡ Mean difference calculated as each cell times its response value number summed by row. It is the expected value of the difference of the average response (number) between groups by question. Mean Difference ⱡ From Table 5, we can see that for all four questions HABITAT is positive and statistically significant, though the magnitude of the coefficients and the levels of statistical significance vary. A perhaps more insightful look at comparative responses for the questions listed above comes from the difference in the average marginal effects shown in Table 6. Here we can again see a similar trend in responses among all four questions, with Habitat homeowners being less likely to report negative evaluations of each of these broad areas, and more likely to report positively (compared to denied applicants). In particular Habitat homeowners were, on average, about 28 percent more likely to report that they believed that their lives had significantly improved since moving into their Habitat houses, and on average answered about one response category greater than denied applicants. 114

115 Although this measure alone cannot indicate how much this positive outlook translates into substantive, measureable life changes, at a minimum it indicates that Habitat homeowners are relatively optimistic about the changes that have occurred in their lives, and that there are least some potential mental/personal welfare benefits associated with their Habitat experiences. Nevertheless, because there are several potential benefits of the Habitat program operating at once, it is difficult to pinpoint if this optimistic outlook stems primarily from becoming a homeowner, upgrading to a higher-quality residence, or receiving the positive wealth effect associated with Habitat s mortgage payment structure. Perhaps surprisingly, Habitat homeowners appeared to be much more pessimistic about changes in their children s educational experiences. Although on average they are about 13 percent more likely to respond that their children s educational experiences have significantly improved, they do not have a statistically different likelihood of reporting that their children s educational experiences have significantly worsened or that there was no change. In addition, the estimated expected difference in response values is relatively small at , or about half of a response value category. As mentioned in Section 2, several studies from the late 1990s and early 2000s reported finding that children of homeowners had better academic achievement than children of renters. However, from this preliminary evaluation it appears that Habitat homeowners may not share in this view, and it warrants further investigation before coming to a definitive conclusion. 115

116 7. Conclusion Homeownership is most certainly a major life goal of many Americans, and both research and federal funding have been directed towards affordable housing for low-income families. But while many studies use purely observational data to measure the impact affordable housing has on these families, Habitat for Humanity provides a unique opportunity to employ a quasi-experimental design to study the possible benefits associated with a homeownership-focused program. In addition, while other experimental design studies such as Moving to Opportunity have focused on neighborhood effects, the fact that Habitat homeowners remain in the same general area permits me to narrow my focus on the three joint benefits specific to the Habitat program: becoming a homeowner, upgrading the quality of one s residence, and receiving a positive income effect due to the mortgage payment structure setup by Habitat. The sample used in this pilot is no doubt small and restricted, and I make several assumptions about the data to permit preliminary interpretation of the results presented. First, that all missing data is missing at random and there is no self-selection issue. Second, that there is no unobserved heterogeneity among the study groups and that qualified housing applicants are drawn at random. Third, that respondents are able to accurately recall current and present information and their responses are free of measurement error. While these are strong assumptions and most certainly violated, they allow me to use this pilot study to evaluate the potential for future research of homeownership-focused programs such as Habitat for Humanity. Overall, both Habitat homeowners and future Habitat homeowners report very positive feelings about their partnerships with Habitat for Humanity. In comparison to denied applicants, those selected into the Habitat program seem to be slightly older, more likely to married, and more educated, perhaps due to their perceived financial stability in making mortgage payments to 116

117 Habitat. This result holds even when controlling for both real weekly income and affiliate-specific fixed effects, as well as when omitting future Habitat homeowners from the treatment group. Habitat homeowners also report that their houses are in better overall condition compared to denied applicants, and that their lot sizes are slightly larger, though they feel the same as denied applicants in terms of evaluating their neighborhoods. Since Habitat homeowners remain in the same general area as their original residence, this last finding provides evidence that treatment group participants are able to distinguish between benefits stemming from neighborhood effects versus the three joint benefits of the Habitat for Humanity program studied in this paper. Lastly, Habitat homeowners report much more positive changes in their lives since applying to Habitat in comparison to denied applicants, including an improvement in their economic situations, their children s educational experiences, their levels of community engagement, and their overall lives in general. To be clear, these self-evaluations alone are unable to separate the impact the three joint benefits associated with the Habitat program outlined above; however, they seem to indicate that Habitat homeowners have an improved sense of satisfaction and well-being since moving into their Habitat house, which could help improve other aspects of their lives as well. Considering the overall net benefits reported by Habitat homeowners in this sample in comparison to denied applicants, clearly more attention and research into homeownership-focused programs such as Habitat for Humanity is required. Future work will examine other parts of the data that have been collected from this pilot study. Since I also have address information for all consenting study participants, I also plan to supplement the survey data in the future with census block and block-group data to measure the influence of any neighborhood characteristics on survey participants responses. Additionally, since the survey used for this pilot study covered a broad 117

118 range of areas, further investigation of this pilot data will help determine the areas in which a larger study of either Habitat or a similar program should focus its attention. References Albright, L., Derickson, E., and D. Massey, Do Affordable Housing Projects Harm Suburban Communities? Crime, Property Values, and Taxes in Mount Laurel, New Jersey, City and Community. Barker, D., and E. Miller, Homeownership and Child Welfare, Real Estate Economics 37(2): Briggs, X.S., Popkin, S., and J. Goering, Moving to Opportunity: The Story of an American Experiment to Fight Ghetto Poverty. New York, NY: Oxford University Press. Campbell, D.T., and J.C. Stanley, Experimental and Quasi-experimental Designs for Research. Boston, MA: Houghton Mifflin Company. Collinson, R., Ellen, I. G., and J. Lidwig, Economics of Means-Tested Transfer Programs in the United States, Low Income Housing Policy (Forthcoming) Cummings, P.M., and J.D. Landis, Relationships between Affordable Housing Development and Neighboring Property Values, Working Paper 599, University of California at Berkeley, Institute of Urban and Regional Development. DeLuca, S., Neighborhood Matters: Do Housing Vouchers Work? Boston Review. Dietz, R., The Social Consequences of Homeownership. Unpublished paper, written for the Homeownership Alliance. Freeman, L., and H. Botein., Subsidized Housing and Neighborhood Impacts: A 42 Theoretical Discussion and Review of the Evidence, Journal of Planning Literature 16: Galster, G.C., Tatian, P., Santiago, A., and K. Pettit, Why Not in My Backyard? Neighborhood Impacts of Deconcentrating Assisted Housing. New Brunswick, NJ: Rutgers University/Center for Urban Policy Research Press. Goetz, Edward G, Clearing the way: Deconcentrating the Poor in Urban America. Washington, D.C.: The Urban Institute Press. 118

119 Green, R., and P. Hendershott, Home Ownership and the Duration of Unemployment: A Test of the Oswald Hypothesis, unpublished. Green, R., and M. White, Measuring the Effect of Homeowning: Effects on Children. Journal of Urban Economics 41: Habitat for Humanity, Affiliate Operations Manual Family Selection, Habitat for Humanity, U.S. Office. < pdf> Habitat for Humanity, U.S. Affiliated Organization Policy Handbook, Habitat for Humanity, U.S. Office. < Proselytizing.pdf> Haurin, H., Parcel, T., and J. Haurin, Does Homeownership Affect Child Outcomes? Real Estate Economics 30(4): Holupka, S., and S. Newman, The Effects of Homeownership on Children's Outcomes: Real Effects or Self-Selection? Forthcoming Paper. Likert, R., A Technique for the Measurement of Attitudes. Archives of Psychology 140: Ludwig, J., Duncan, G.J., Gennetian, L.A., Katz, L.F., Kessler, R., Kling, J.R., and L. Sanbomatsu, Neighborhood Effects on the Long-Term Well-Being of Low- Income Adults, Science [Internet] 337: Ludwig, J., Duncan, G.J., Gennetian, L.A., Katz, L.F., Kessler, R., Kling, J.R., and L. Sanbomatsu, What Can We Learn about Neighborhood Effects from the Moving to Opportunity Experiment? American Journal of Sociology 114: Madden, J.F, The Changing Spatial Concentration of Income and Poverty among Suburbs of Large US Metropolitan Areas, Urban Studies 40(3): Massey, D.S., and S.M. Kahaiaupuni, Public Housing and the Concentration of Poverty, Social Science Quarterly 74(1): Massey, D., Albright, L., Casciano, R., Derickson, E., and D. Kinsey, Climbing Mt. Laurel: The Struggle for Affordable Housing and Social Mobility in an American Suburb. Princeton, NJ: Princeton University. Mohanty, L., and L. Raut, Home Ownership and School Outcomes of Children: Evidence from the PSID Child Development Supplement, American Journal of Economics and Sociology 68(2):

120 Nguyen, M.T., Does Affordable Housing Detrimentally Impact Property Values? A Review of the Literature, Journal of Planning Literature 20(1): Rossi, P., and E. Weber, What Can We Learn about Neighborhood Effects from the Moving to Opportunity Experiment? Housing Policy Debate 7(1). Sampson, R.J., Morenoff, J.D., and T. Gannon-Rowley, Assessing 'Neighborhood Effects': Social Processes and New Directions in Research, Annual Review of Sociology 28: Sanbonmatsu, L., Marvokov, J., Porter, N., Yang, F., Adam, E., Congdon, W.J., Duncan, G.J., Gennetian, L.A., Katz, L.F., and J.R. Kling, The Long-Term Effects of Moving to Opportunity on Adult Health and Economic Self-Sufficiency, Cityscape [Internet] 14(2): Segal, L., and D. Sullivan, Trends in Homeownership: Race, Demographics, and Income, Economic Perspectives 2: Shadish, W., Cook, T., and D. Campbell, Experimental and Quasi-Experimental Designs for Generalized Causal Inference. 2 nd edition. Cengage Learning. Tighe, J.R., Public Opinion and Affordable Housing: A Review of the Literature, Journal of Planning Literature 25(1): Wooldridge, J., 2012, Introductory Econometrics: A Modern Approach, Fifth Edition, South- Western College Pub. 120

121 Appendix 1.A: Consent Form Habitat Homeowners NU HSRP Rev. 1/20/2013 ```````````````````````````````````````````````````````````````````````````````````````````````````````````` Northeastern University, Department Name of Investigator(s): Williams Dickens, Ph.D., and Thomas PlaHovinsak, Ph.D. Candidate Title of Project: The Impact of Homeownership on Low Income Families: A Case Study with Habitat for Humanity Informed Consent to Participate in a Research Study We are inviting you to take part in a research study associated with Habitat for Humanity and Northeastern University. This form will tell you about the study. When you are ready to decide if you would like to take part in the study, you may sign the form. You do not have to participate if you do not want to. If you decide to participate, you must sign this statement. If you would like a copy of this consent form, please write your return address on the prestamped return envelope. Why am I being asked to take part in this research study? We are asking you to take part in this study because you previously applied for housing assistance through Habitat for Humanity. Why is this research study being done? The purpose of this research is to discover the impacts of homeownership on individuals and families. We will explore if homeownership causes any changes in economic and career outcomes, health and physical well-being, youth changes, and community engagement. This research study can provide us with new information about how owning a home can impact personal lives. What will I be asked to do? 121

122 If you decide to take part in this study, you will be asked to complete a survey and provide us with permission to look at your original application for a home at your local Habitat affiliate. The survey can be found behind this consent form. This will allow the researchers to compare your current social and economic physical well-being to the state of each category at the time of applying. This will further allow us to measure if there was an objective and significant change due to owning a home. The survey will then ask you about your children before moving into your Habitat home and after moving into the home (if applicable) as well as your community engagement. All Habitat homeowners in the state of New Jersey will be asked to take part in this survey and consent process. Where will this take place and how much of my time will it take? You will be asked to complete the survey in the comfort of your own home. The survey is 5 pages long and should take you approximately 20 minutes. You can mail it back to us in the pre-stamped envelope in this packet. Will there be any risk or discomfort to me? The only foreseeable discomfort is related to your previous life circumstances. The survey questions may cause slight psychological distress because they may cause you to think about times of your life associated with high levels of stress and/or anxiety. However, we perceive this discomfort to be minimal. If you feel any discomfort during or after completion of the survey, please call your local affiliate office and ask to speak with the Family Relations coordinator or the executive director. Will I benefit by being in this research? There will be no direct benefits to you for taking part in the study. However, the information learned from this study may help the researchers provide policy implications for individual homeowners as well as individuals who living in inadequate housing conditions. The information will also allow us to provide local affiliates with suggestions for further improvement of the homeownership program. 122

123 Who will see the information about me? The information you provide will be confidential and in the strict confidence. Your name will not be associated with any of the information you provide. Only the researchers on this study will see the information about you. No reports or publications will use information that can identify you in any way or any individual as being of this project. The survey is coded with a unique identification number, which only the researchers can link to you. Once the survey is returned, the affiliates will leave it sealed and use the external number on the envelope to pair your survey with your original application. They will then copy your original application, and black out any proprietary information, including your name, social security number, name of employer(s), bank accounts, asset-specific information, etc. In rare instances, authorized people may request to see research information about you and other people in this study. This is done only to be sure that the research is done properly. We would only permit people who are authorized by organizations such as the Northeastern University's Institutional Review Board to see this information. No individual at Habitat for Humanity will view the returned survey or any information linked to your name. The data will be maintained in order to conduct a panel study in the future. It will be locked in an office cabinet on Northeastern's campus. If I do not want to take part in the study, what choices do I have? If you do not want to take part in this study and/or are unwilling to consent to the release of your original application, it is not necessary for you to complete the survey. However, please check the "decline to participate" box at the bottom of this consent form and return it using the pre-stamped envelope. What will happen if I suffer any harm from this research? No special arrangements or payments will be made for counseling if you experience any psychological distress as a result of your participation. Can I stop my participation in this study? 123

124 Your participation in this research is completely voluntary. You do not have to participate if you do not want to and you can refuse to answer any questions. Even if you begin the study, you may quit at any time. If you do not participate or if you decide to quit, you will not lose any rights, benefits, or services that you would otherwise have as a Habitat homeowner. Who can I contact if I have questions or problems? If you have any questions about this study, please feel free to contact Thomas PlaHovinsak via at plahovinsak.t@husky.neu.edu or via telephone at (908) You can also contact William Dickens, the Principal Investigator, via at w.dickens@neu.edu or via telephone at (617) Who can I contact about my rights as a participant? If you have any questions about your rights in this research, you may contact Nan C. Regina, Director, Human Subject Research Protection, 960 Renaissance Park, Northeastern University, Boston, MA Tel: , n.regina@neu.edu. You may call anonymously if you wish. Will I be paid for my participation? You will entered in a lottery for one of two $100 American Express gift card as soon as you complete and return the survey. Will it cost me anything to participate? You will not incur any costs from your participation in the study. Is there anything else I need to know? You must be at least 18 years old to participate in the study. 124

125 Appendix 1.B: Consent Form Future Habitat Homeowners NU HSRP Rev. 1/20/2013 ```````````````````````````````````````````````````````````````````````````````````````````````````````````` Northeastern University, Department Name of Investigator(s): Williams Dickens, Ph.D., and Thomas PlaHovinsak, Ph.D. Candidate Title of Project: The Impact of Homeownership on Low Income Families: A Case Study with Habitat for Humanity Informed Consent to Participate in a Research Study We are inviting you to take part in a research study associated with Habitat for Humanity and Northeastern University. This form will tell you about the study. When you are ready to decide if you would like to take part in the study, you may sign the form. You do not have to participate if you do not want to. If you decide to participate, you must sign this statement. If you would like a copy of this consent form, please write your return address on the prestamped return envelope. Why am I being asked to take part in this research study? We are asking you to take part in this study because you were accepted as a partner family with a Habitat for Humanity affiliate. Why is this research study being done? The purpose of this research is to discover the impacts of homeownership on individuals and families. We will explore if homeownership causes any changes in economic and career outcomes, health and physical well-being, youth changes, and community engagement. This research study can provide us with new information about how owning a home can impact personal lives. What will I be asked to do? 125

126 If you decide to take part in this study, you will be asked to complete a survey and provide us with permission to look at your original application for a home at your local Habitat affiliate. The survey can be found behind this consent form. This will allow the researchers to compare your current social and economic physical well-being to the state of each category at the time of applying. This will further allow us to measure if there was an objective and significant change due to owning a home. The survey will then ask you about your children before moving into your Habitat home and after moving into the home (if applicable) as well as your community engagement. All Habitat homeowners in the state of New Jersey will be asked to take part in this survey and consent process. Where will this take place and how much of my time will it take? You will be asked to complete the survey in the comfort of your own home. The survey is 5 pages long and should take you approximately 20 minutes. You can mail it back to us in the pre-stamped envelope in this packet. Will there be any risk or discomfort to me? The only foreseeable discomfort is related to your previous life circumstances. The survey questions may cause slight psychological distress because they may cause you to think about times of your life associated with high levels of stress and/or anxiety. However, we perceive this discomfort to be minimal. If you feel any discomfort during or after completion of the survey, please call your local affiliate office and ask to speak with the Family Relations coordinator or the executive director. Will I benefit by being in this research? There will be no direct benefits to you for taking part in the study. However, the information learned from this study may help the researchers provide policy implications for individual homeowners as well as individuals who living in inadequate housing conditions. The information will also allow us to provide local affiliates with suggestions for further improvement of the homeownership program. 126

127 Who will see the information about me? The information you provide will be confidential and in the strict confidence. Your name will not be associated with any of the information you provide. Only the researchers on this study will see the information about you. No reports or publications will use information that can identify you in any way or any individual as being of this project. The survey is coded with a unique identification number, which only the researchers can link to you. Once the survey is returned, the affiliates will leave it sealed and use the external number on the envelope to pair your survey with your original application. They will then copy your original application, and black out any proprietary information, including your name, social security number, name of employer(s), bank accounts, asset-specific information, etc. In rare instances, authorized people may request to see research information about you and other people in this study. This is done only to be sure that the research is done properly. We would only permit people who are authorized by organizations such as the Northeastern University's Institutional Review Board to see this information. No individual at Habitat for Humanity will view the returned survey or any information linked to your name. The data will be maintained in order to conduct a panel study in the future. It will be locked in an office cabinet on Northeastern's campus. If I do not want to take part in the study, what choices do I have? If you do not want to take part in this study and/or are unwilling to consent to the release of your original application, it is not necessary for you to complete the survey. However, please check the "decline to participate" box at the bottom of this consent form and return it using the pre-stamped envelope. What will happen if I suffer any harm from this research? No special arrangements or payments will be made for counseling if you experience any psychological distress as a result of your participation. Can I stop my participation in this study? 127

128 Your participation in this research is completely voluntary. You do not have to participate if you do not want to and you can refuse to answer any questions. Even if you begin the study, you may quit at any time. If you do not participate or if you decide to quit, you will not lose any rights, benefits, or services that you would otherwise have as a Habitat homeowner. Who can I contact if I have questions or problems? If you have any questions about this study, please feel free to contact Thomas PlaHovinsak via at plahovinsak.t@husky.neu.edu or via telephone at (908) You can also contact William Dickens, the Principal Investigator, via at w.dickens@neu.edu or via telephone at (617) Who can I contact about my rights as a participant? If you have any questions about your rights in this research, you may contact Nan C. Regina, Director, Human Subject Research Protection, 960 Renaissance Park, Northeastern University, Boston, MA Tel: , n.regina@neu.edu. You may call anonymously if you wish. Will I be paid for my participation? You will entered in a lottery for one of two $100 American Express gift card as soon as you complete and return the survey. Will it cost me anything to participate? You will not incur any costs from your participation in the study. Is there anything else I need to know? You must be at least 18 years old to participate in the study. 128

129 Appendix 1.C: Consent Form Denied Applicants NU HSRP Rev. 1/20/2013 ```````````````````````````````````````````````````````````````````````````````````````````````````````````` Northeastern University, Department Name of Investigator(s): Williams Dickens, Ph.D., and Thomas PlaHovinsak, Ph.D. Candidate Title of Project: The Impact of Homeownership on Low Income Families: A Case Study with Habitat for Humanity Informed Consent to Participate in a Research Study We are inviting you to take part in a research study associated with Northeastern University and Habitat for Humanity. This form will tell you about the study. When you are ready to decide if you would like to take part in the study, you may sign the form. You do not have to participate if you do not want to. If you decide to participate, you must sign this statement. If you would like a copy of this consent form, please write your return address on the pre-stamped return envelope. Why am I being asked to take part in this research study? We are asking you to take part in this study because you previously applied for housing assistance through Habitat for Humanity. Why is this research study being done? The purpose of this research is to discover the impact of living quarters on individuals and families. We will explore if the type of living quarters a family lives in impacts economic and career outcomes, health and physical well-being, youth changes, and community engagement. This research study can provide us with new information about how different types of living quarters impact personal lives. What will I be asked to do? If you decide to take part in this study, you will be asked to complete a survey and provide us with permission to view your original Habitat for Humanity application. 129

130 The survey can be found behind this consent form. Completing this survey and consenting to the study will allow us to compare your current social and economic well-being to the state of each category at the time of applying for house through Habitat for Humanity. This will further allow us to measure if there was an objective and significant change due to a change in your living quarters. The survey will then ask you about your children before applying for housing assistance and after applying for housing assistance (if applicable) as well as your community engagement. Where will this take place and how much of my time will it take? You will be asked to complete the survey in the comfort of your own home. The survey is 5 pages long and should take you approximately 20 minutes. You can mail it back to us in the pre-stamped envelope in this packet. Will there be any risk or discomfort to me? The only foreseeable discomfort is related to your previous life circumstances. The survey questions may cause slight psychological distress because they may cause you to think about times of your life associated with high levels of stress and/or anxiety. However, we perceive this discomfort to be minimal. Will I benefit by being in this research? There will be no direct benefits to you for taking part in the study. However, the information learned from this study may help the researchers provide policy implications for individual homeowners as well as individuals who are living in inadequate housing conditions. Who will see the information about me? The information you provide will be confidential and kept in the strictest confidence. Only the researchers on this study will see any information about you. No reports or publications will use information that can identify you in any way or any individual as being of this project. When your survey is received, it will be coded with a unique identification number and your name will be removed from all information. We will inform the housing agency that we received your consent form and that you are interested in taking part in the study. Then the agency will copy your original application, and black out any proprietary information, including 130

131 your name, social security number, name of employer(s), bank accounts, asset-specific information, etc. In rare instances, authorized people may request to see research information about you and other people in this study. This is done only to be sure that the research is done properly. We would only permit people who are authorized by organizations such as the Northeastern University's Institutional Review Board to see this information. No individual at Habitat will view the returned survey. The data will be maintained in order to conduct a panel study in the future. It will be locked in an office cabinet on Northeastern's campus. If I do not want to take part in the study, what choices do I have? If you do not want to take part in this study and/or are unwilling to consent to the release of your original application, it is not necessary for you to complete the survey. However, please check the decline to participate" box at the bottom of this consent form and return it using the pre-stamped envelope. What w ill happen if I suffer any harm from this research? No special arrangements or payments will be made for counseling if you experience any psychological distress as a result of your participation. Can I stop my participation in this study? Your participation in this research is completely voluntary. You do not have to participate if you do not want to and you can refuse to answer any questions. Even if you begin the study, you may quit at any time. If you do not participate or if you decide to quit, you will not lose any rights, benefits, or services that you would otherwise have in reference to your current relationship with Habitat for Humanity. Who can I contact if I have questions or problems? If you have any questions about this study, please feel free to contact Thomas PlaHovinsak via at plahovinsak.t@husky.neu.edu or via telephone at (908) You can also contact 131

132 William Dickens, the Principal Investigator, via at or via telephone at (617) Who can I contact about my rights as a participant? If you have any questions about your rights in this research, you may contact Nan C. Regina, Director, Human Subject Research Protection, 960 Renaissance Park, Northeastern University, Boston, MA Tel: , n.regina@neu.edu. You may call anonymously if you wish. Will I be paid for my participation? You will be entered into a lottery for one of two $100 American Express gift cards as soon as you complete and return the survey. Will it cost me anything to participate? You will not incur any costs from your participation in the study. Is there anything else I need to know? You must be at least 18 years old to participate in the study. 132

133 Appendix 2.A: Survey Habitat Homeowners 133

134 134

135 135

136 136

137 137

138 138

139 139

140 Appendix 2.B: Survey Future Habitat Homeowners 140

141 141

142 142

143 143

144 144

145 145

146 146

147 Appendix 2.C: Survey Denied Applicants 147

148 148

149 149

150 150

151 151

152 152

153 153