Land Use Regulations,Housing Supply,and County Eviction Filings

Abstract

Spanish

Chinese

The eviction crisis has attracted significant scholarly attention, but less is known about the links between evictions and policies that promote or impede housing supply. This research uses data from a large time-series, cross-sectional sample of U.S. counties to address this research gap. The results from several linear regression models suggest that restrictive land use regulations increase eviction filing rates, whereas elastic housing supply and inclusionary zoning reduce eviction filing rates. Controls for housing supply constraints also reduce the effect of median rents on eviction filings. Pro-renter state landlord–tenant laws have no effect on eviction filing rates.

Keywords

evictions land use regulation landlord–tenant law housing supply housing affordability

Introduction

Matthew Desmond’s Evicted: Poverty and Profit in the American City (2016) and subsequent work by Princeton University’s Eviction Lab have shined a light on a previously hidden national housing crisis affecting nearly one million renters each year: evictions (Eviction Lab 2018). For a rising number of low-income families, eviction is the first step in a downward spiral that often ends with job loss, familial disruption, mental stress, and social alienation (Desmond 2016; Desmond and Kimbro 2015). Although the proliferation of eviction moratoria during the early stages of the COVID-19 pandemic temporarily halted eviction filings, the Supreme Court’s 2021 decision to strike down the federal eviction moratorium and the expiration of many state and local eviction moratoria have elevated eviction rates to pre-pandemic levels (Haas et al. 2021).

With the expiration of pandemic-era eviction moratoria, scholars and policymakers have begun the search for more durable policy solutions to help low-income families remain in their homes amid rising rental housing prices and stagnant wages. Less than two months after the U.S. Supreme Court ended the Center for Disease Control and Prevention’s federal eviction moratorium, the U.S. Department of Housing and Urban Development (HUD 2021) issued a rule requiring a thirty-day notice prior to the eviction of HUD-assisted households for nonpayment of rent. Several states also attempted to prevent a post-pandemic eviction crisis by passing laws that encouraged or required landlords to first apply for federal emergency rental assistance before initiating an eviction process (Vasquez et al. 2022).

Despite the flurry of policy activity around the problem of evictions, little is known about the effectiveness of various policy interventions. The few empirical evaluations of anti-eviction policies have emphasized eviction moratoria, state landlord–tenant laws, and other policies directly targeted to those at risk of facing an eviction. Fewer studies have examined the impact of policies designed to improve housing affordability, despite the obvious theoretical link between rising rental housing prices and the threat of eviction. No studies have examined the role of local land use regulations and other constraints on housing supply, but theory and evidence point to potential linkages. If land use regulations restrict new housing construction in the face of rising housing demand, economic theory suggests that the price of existing rental housing will rise (Glaeser, Gyourko, and Saks 2005). If rents rise, and affordable rental housing is scarce, existing renters may have less bargaining power in lease negotiations, particularly if a large number of in-migrating renters would be willing to pay higher rents for apartments that are currently occupied. A scarcity of land for housing development may also increase the likelihood that owners of affordably priced rental apartment buildings will choose to upgrade apartment buildings to command higher rents, convert rental apartments to condominiums, or sell apartments to developers seeking land for higher-valued uses.

This paper examines the geographic determinants of county eviction filing rates, emphasizing the role of land use regulations that restrict housing supply and/or require a portion of newly constructed housing to be affordably priced. The results from ordinary least squares (OLS) and random effects regression models suggest that restrictive land use regulations are positively associated with county eviction filing rates, while elastic housing supply and inclusionary zoning (IZ) ordinances reduce eviction filing rates. These effects are strongest in models that control for unobserved county-level variability in eviction filing rates. Land use regulations also dampen the positive impact of median rents on eviction filing rates. The prevalence of subsidized housing reduces county eviction filing rates, but only in OLS models that ignore inter-county heterogeneity. I find no evidence that eviction filing rates are lower in states with renter-friendly landlord–tenant laws. These findings suggest that the scope of anti-eviction advocacy efforts should expand to include an emphasis on tenure-responsive land use planning. The findings also suggest that local land use planners can play an important role in shaping the geography of evictions.

The remainder of the paper is structured as follows. I begin with a review of existing evidence on the causes of evictions and eviction filings, followed by a discussion of the data and empirical approach. I then discuss the evidence from several regression models explaining county eviction filing rates and conclude with a discussion of the planning and policy implications of the findings.

Evidence on the Causes of Evictions

Several empirical studies of the causes of evictions rely on data collected from individuals or families to explain the household-level correlates of evictions and the loss of housing. Hepburn, Louis, and Desmond’s (2020) analysis of data from eviction court cases provides evidence of significant racial and ethnic differences in eviction rates. Furthermore, among black and Hispanic renters, females are more likely than males to face an eviction. Phinney et al. (2007) find that, among single mothers receiving welfare assistance in one Michigan county, low educational attainment is positively associated with the incidence of both homelessness and eviction. Curtis et al. (2013), in a study that relies on data from the Fragile Families and Child Wellbeing study, find that educational attainment is not associated with the probability of being homelessness but is negatively associated with the probability of being evicted or making multiple moves. Curtis et al. (2013) also find that having a child with a severe infant health condition is positively associated with homelessness but is not associated with the probability of being evicted or making multiple moves.

Large randomized control trial studies of homelessness such as the Family Options Study or the various Housing First experiments provide useful data sources for the study of evictions, although these studies rarely differentiate those who lose their homes to eviction from those who lose their homes for other reasons. One exception is the Australian Journeys Home study, which includes data collected from those who were housed at the beginning of the study (Ribar 2017). A few researchers relying on this source examine the influence of local housing market conditions on the incidence of homelessness and eviction. Johnson et al. (2015) find that higher housing rents and higher local unemployment rates are positively associated with the transition to homelessness. Cobb-Clark et al. (2016) find that higher rents are also associated with longer homelessness spells.

Among those studying the determinants of evictions in the United States, some exploit household- or property-level data from a single housing market to assess the influence of familial, property-level, and neighborhood-level factors on evictions. Preston and Reina (2021), relying on a panel data set constructed from eviction filings in Philadelphia, find that eviction filings are concentrated in depressed real estate markets and in neighborhoods with a high concentration of disadvantaged households and cost-burdened renters. Desmond and Gershenson (2017) rely on household-level data from the Milwaukee Area Renters Study matched to census block group data to assess the influence of a variety of neighborhood-level conditions on the likelihood of eviction. The authors find that only two neighborhood-level conditions are associated with the probability of a household being evicted: neighborhood crime rates and neighborhood eviction rates. The authors also find that social ties with other disadvantaged persons are positively associated with eviction.

Other studies rely on data from a large number of geographic areas to assess the market-level determinants of aggregate eviction rates and eviction filing rates. These studies primarily rely on state-level evictions data or the national evictions database assembled by the Eviction Lab (Desmond et al. 2018b), a data repository and research consortium led by a Matthew Desmond and a team of researchers from Princeton University. A few researchers relying on aggregate data have also included policy variables in models explaining local eviction rates or eviction filing rates. Hepburn et al. (2021) find that during the initial years of the COVID pandemic, when various local, state, and federal eviction moratoria were in place, 1.55 million fewer evictions were filed than would be expected during a normal year. Furthermore, eviction filings rose above historical averages as soon as eviction moratoria expired. While effective as a short-term public health measure, widespread bans on evictions are unlikely to be a feasible long-term policy going forward, particularly given the Supreme Court’s decision to invalidate the federal eviction moratorium. Evidence also suggests that many pandemic-era eviction moratoria were largely ineffective due to the lax penalties for landlords who violated eviction moratoria (Alexander and Lee 2021).

The most common target of anti-eviction advocacy efforts is state-level landlord–tenant law, which outlines the duties of landlords and tenants in residential lease contracts. In a large number of states, landlord–tenant laws are based on the Uniform Residential Landlord and Tenant Act, promulgated in 1972 by the National Conference of Commissioners on Uniform State Laws, or the American Law Institute’s Model Landlord Tenant Act (Glendon 1982). Despite the uniform origins of state landlord–tenant laws, most states’ laws have been substantially modified over time to become more or less friendly to tenants vis-à-vis landlords. Three states (Washington, Maryland, and Connecticut) have recently enacted tenant “right-to-counsel” laws, which provide taxpayer-funded legal representation to tenants facing eviction (National Coalition for a Civil Right to Counsel 2022). Evidence from randomized control trials suggests that while legal representation prolongs the eviction process and lowers tenants’ debt obligations, it does not always reduce the likelihood that cases will end in an eviction (Abramson 2021).

In addition to tenants’ rights to counsel, tenant–landlord laws address the maximum size of renter security deposits, landlords’ obligation to notify tenants prior to eviction, procedures for evicting tenants for reasons other than nonpayment of rent, and a variety of other obligations of landlords vis-à-vis renters. Evidence points to substantial variety in the number and strength of these provisions across states. Coulson, Le, and Shen’s (2020) comprehensive survey of state landlord–tenant laws finds that minimum notices of eviction for nonpayment of rent range from zero day to thirty days, and minimum notices for rent increases vary from seven to sixty days. Some states have no penalties for landlords who attempt to lock-out tenants or shut off utilities without a formal eviction notice, whereas other states invoke harsh penalties. Hatch (2017) relies on cluster and discriminant analysis of state landlord–tenant laws to classify states into those that are protectionist (pro-renter), those that are pro-business, and those that have contradictory policies that are both pro-renter and pro-business. Employing this approach, Hatch (2017) classifies seventeen states as pro-business, thirteen states as protectionist, and twenty states as having contradictory policies that sometimes favor renters and sometimes favor landlords.

Two studies suggest that renter-friendly landlord–tenant laws have meaningful impacts on local eviction rates. Coulson, Le, and Shen (2020) rely on data from their survey of state landlord–tenant laws to construct a tenant-right index that is included in a regression model explaining municipal eviction rates. The authors find that a unit increase in the tenant-right index reduces municipal eviction rates by 0.32 percentage points. Merritt and Farnworth (2021) include indicators of the state landlord–tenant law types identified by Hatch (2017) in a regression model explaining eviction rates and eviction filing rates at the census block group level. The authors find that while a state’s landlord protection policy regime does not have a statistically significant impact on eviction rates for the average census block group, state policy regimes are statistically significant when neighborhood racial composition is taken into account. Specifically, the gap in neighborhood eviction rates between protectionist and pro-business states is largest within majority-black neighborhoods.

Another possible long-term anti-eviction strategy is the expansion of existing social safety net programs such as Medicaid and various federal housing assistance programs. Allen et al. (2019) examine the influence of Medicaid expansion on county-level eviction rates within the state of California. The authors find that Medicaid expansion is associated with a reduction in the number and rate of evictions, with the largest effects concentrated in counties with low insurance rates prior to Medicaid expansion. Preston and Reina (2021) find that those living in public housing and in units receiving project-based rental subsidies are less likely to face an eviction filing than those living in similar unsubsidized properties.

Although no studies have included measures of local land use regulations in models explaining local eviction rates or eviction filing rates, economic theory points to possible connections between land use regulations and evictions. In jurisdictions where land use regulations reduce the responsiveness of new housing supply to increased housing demand, the existing housing stock will absorb all short-term housing demand shocks. Competition among renters for scarce existing housing may increase landlords’ incentives to replace low-paying renters with renters willing to pay more. In jurisdictions facing an influx of new renters, existing renters have less bargaining power in these negotiations because in-migrating renters would be willing to pay higher rents for the rental housing units that are currently occupied. In cities where land for housing development is scarce, owners of affordably priced rental apartment buildings may also be more likely to upgrade older apartment buildings, convert rental apartments to condominiums, or sell apartments to developers seeking land for higher valued uses. Somerville and Mayer (2003) provide evidence that is consistent with these arguments. The authors demonstrate that in highly regulated housing markets, existing affordable homes are more likely to filter up and become unaffordable to those earning low incomes. No studies have extended these arguments to explore the influence of land use restrictions and other housing supply constraints on eviction rates at the county level.

Data and Method

In the empirical analysis, I examine the impact of land use restrictions, IZ requirements, and housing supply elasticity on county eviction filing rates, controlling for a variety of housing market and sociodemographic conditions that have shown to be related to the likelihood of eviction. Specifically, I estimate several different specifications of the following linear regression model:

\ln (E_{i t}) = X_{i t}^{'} β + α_{i} + u_{i t}

(1)

where E_it is the eviction filing rate for county i at time period t (t varies from 2009 to 2016), β is a vector of estimated coefficients, and X_it is a vector of independent variables. E_it the eviction filing rate for county i at time t, defined as the number of eviction filings divided by the number of renters living in county i at time t. Given that E is a percentage bounded between 0 and 100, I transform E_it into its natural log prior to estimation.

I focus on eviction filings rather than actual evictions for two reasons. First, the publicly available data provided by the Eviction Lab, the source for the dependent variable in this study, only report eviction filings. Second, unlike actual evictions, eviction filings capture serial evictions, which occur when a given rental household receives multiple eviction notices while living at the same address, possibly without ever being evicted. Evidence from Leung, Hepburn, and Desmond (2021) suggests that many landlords rely on serial evictions as part of their normal business practice as a way to collect rent and fees. If landlords rely on serial evictions to exercise leverage against renters in lease renegotiations, as suggested in the previous section, failing to account for serial evictions may underestimate the impact of housing supply constraints on eviction practices.

With the panel structure of the data, it is possible to control for time-invariant county-specific effects, which I estimate using a random effects model that decomposes the overall random error term into a county-specific error term, represented by α_i, and an overall error term that varies across counties and across time (u_it). If α_i ≠ 0, and the error term varies systematically across counties, then the random effects model provides a more efficient estimate of the model’s parameters than the traditional OLS regression model. Given that some elements of X_it do not vary across time, it is not possible to estimate fixed effects models that treat α_i as county-specific intercepts.

The vector X_it includes year-specific dummy variables along with several variables discussed in the previous section that have been hypothesized to be associated with local eviction filing rates. These include various demographic characteristics, including the percent of the population that is black, percent of the population that is Hispanic, the percent of seniors aged sixty-five or older in the population, and the percent of single-person households. I also include the local unemployment rate to capture the influence of local economic conditions and population density to control for a range of factors that affect differences in eviction rates along the urban–rural spectrum.

I include several measures of local housing market conditions, including the percent of households spending more than 50 percent of their income on housing costs, the percent of households who are renters, and the percent of housing units that are subsidized through one of the HUD subsidy programs or the Low Income Housing Tax Credit (LIHTC) program. As reported in research by Leung, Hepburn, and Desmond (2021), the effect of median rent on eviction filings is nonlinear, with lower eviction filing rates occurring in neighborhoods with low and high median rents. To capture these nonlinearities, I include median gross rent and its square.

I also include two state-level policy variables. The first is a state-level measure of the strength of state landlord–tenant laws. I rely on Hatch’s (2017) classification of state landlord–tenant laws to create a dummy variable equal to 1 if the state is a protectionist state and 0 otherwise. I also include a dummy variable equal to 1 for Maryland and 0 otherwise to capture Maryland’s unique eviction filing process which produces a high case volume compared with other states. According to Desmond et al. (2018a), the eviction process in Maryland begins with an eviction filed in court, compared with other states where evictions begin with an out-of-court notice delivered to the tenant.

The primary variables of interest are two different measures of local land use regulations and one variable that captures the responsiveness of housing supply to demand shocks. The first regulatory variable is the Wharton Residential Land Use Regulatory Index (WRLURI) developed by Gyourko, Saiz, and Summers (2008). The WRLURI is based on surveys of over 2,000 jurisdictions in major metropolitan housing markets. The WRLURI is aggregated from responses to multiple surveys within each U.S. Census–defined metropolitan area and increases in value with the average restrictiveness of local land use regulations within a given metropolitan area. The second regulatory variable is equal to the number of IZ regulations adopted by jurisdictions within the county. To calculate this variable, I rely on the census of IZ ordinances developed by Wang and Balachandran (2021) and use geographic information system (GIS) software to count the number of IZ ordinances within each county at of the end of 2008.

The models also include the metropolitan area-level estimate of housing supply elasticity developed by Saiz (2010). Unlike the WRLURI which only captures regulatory restrictiveness, the measure developed by Saiz (2010) is a direct measure of the responsiveness of housing supply to housing demand shocks. To construct this measure, Saiz relies on measures of local regulatory restrictiveness and other geographic constraints that limit developable land area to estimate the percentage increase in housing supply as a function of percentage increases in housing prices. As the Saiz (2010) measure is a function of the WRLURI, I include each of these variables separately in the estimated models to avoid multicollinearity. I also estimate models that omit the housing supply elasticity and land use regulatory variables to assess the confounding influence of these variables on other estimated coefficients. Of particular interest is the confounding influence of housing supply constraints on the size of the median rent coefficients.

The WRLURI and housing supply elasticity measures are measured at the metropolitan area level using 1999 metropolitan area definitions, and the IZ variable is measured at the county level. Unlike the other covariates included in the model, the housing supply constraint variables are only available for a single year, so the effects of housing supply constraints are assumed to be constant across time.

The panel is unbalanced, so some counties have more years of eviction filing data than others. After removing counties that were not in 1999 Census-defined metropolitan areas, the final analytical sample includes 451 counties with an average of 6.8 years of data. These counties represent 53 percent of all counties in 1999 Census-defined metropolitan areas. The final sample size (N × T) is equal to 3,082 time-series, cross-sectional observations, which is equal to 20 percent of the full Eviction Lab sample. All counties in the final sample have nonzero eviction filing rates. The data sources and descriptive statistics for each of the variables included in the analyses are listed in Table 1.

Table 1.

Descriptive Statistics.

Variable	Data source	M	SD
Ln(Eviction filing rate)	Desmond et al. (2018b)	1.92	1.00
Eviction filing rate	Desmond et al. (2018b)	11.20	15.63
% black in total population	2009–2016 5-year American Community Survey	11.94	13.34
% Hispanic in total population	2009–2016 5-year American Community Survey	11.97	14.18
% Age 65+ in total population	2009–2016 5-year American Community Survey	13.69	3.69
% One-person households	2009–2016 5-year American Community Survey	25.94	5.07
Unemployment rate	2009–2016 5-year American Community Survey	8.39	2.63
Population/sq. mi.	2009–2016 5-year American Community Survey	1,139.27	4,680.61
Median gross rent	2009–2016 5-year American Community Survey	860.58	217.55
Median gross rent squared	2009–2016 5-year American Community Survey	787,909.40	435,092.90
% Extreme rental cost burden	2009–2016 5-year American Community Survey	22.85	4.72
% Renters	2009–2016 5-year American Community Survey	31.08	10.33
% of housing units subsidized	2009–2016 HUD Picture of Sub. Households, 2019 HUD LIHTC Database	16.68	7.53
Maryland	Desmond et al. (2018b)	0.04	0.19
Protectionist state landlord–tenant law	Hatch (2017)	0.17	1.46
No. of IZ ordinances in co.	Wang and Balachandran (2021)	0.23	1.25
WRLURI	Obtained from A. Saiz, as calculated by Saiz (2010)	−0.02	0.74
Housing supply elasticity	Obtained from J. Gyourko, as calculated by Gyourko, Saiz, and Summers (2008)	2.31	1.23

Note: HUD = U.S. Department of Housing and Urban Development; LIHTC = Low Income Housing Tax Credit; IZ = inclusionary zoning; WRLURI = Wharton Residential Land Use Regulatory Index.

Findings

I begin with an examination of the results from OLS regression models that do not control for county-level random effects. To control for unspecified heteroskedasticity, all models use robust standard errors, as suggested by Huber (1967) and White (1980). Table 2 displays the results from three models. The first omits controls for land use regulations and housing supply elasticity; the second includes the WRLURI and the IZ variable; and the third includes the housing supply elasticity variable along with the IZ variable. Given that the dependent variable is log-transformed, I choose not to report the raw coefficients but instead report the percentage change in the eviction filing rate from a one standard deviation unit increase in each independent variable. This percentage change is equal to (exp(βX_s) − 1) × 100, where β is the raw coefficient, and X_s is the standard deviation of variable X, holding the other independent variables constant.

Table 2.

OLS Regression Results.

Variable	Model 1		Model 2		Model 3
Variable	(exp(βX_s) − 1) × 100	Sig.	(exp(βX_s) − 1) × 100	Sig.	(exp(βX_s) − 1) × 100	Sig.
Constant	11.18		35.31	**	111.55	***
% black in total population	43.49	***	43.57	***	47.26	***
% Hispanic in total population	−7.38	***	−7.76	***	−8.32	***
% Age 65+ in total population	−19.85	***	−19.74	***	−20.98	***
% One-person households	10.93	***	10.37	**	8.84	**
Unemployment rate	23.95	***	23.62	***	19.96	***
Population/sq. mi.	19.75	***	17.75	***	15.78	***
Median gross rent	137.91	***	99.67	***	69.25	***
Median gross rent squared	−56.06	***	−47.02	***	−41.19	***
% Extreme rental cost burden	9.34	***	8.07	***	8.18	***
% Renters	−34.64	***	−32.88	***	−31.00	***
% of housing units subsidized	−3.81	**	−4.34	**	−5.60	***
Maryland	39.67	***	37.43	***	37.98	***
Protectionist state landlord–tenant law	1.07		1.39		2.31
No. of IZ ordinances in co.			−8.33	***	−8.84	***
WRLURI			5.79	**
Housing supply elasticity					−15.59	***
N × T	3,082		3,082		3,082
R ²	.47		.47		.49
Wald Test, Year Effects	6.14	***	5.50	***	5.23	***

Note: All models include year fixed effects and robust standard errors. OLS = ordinary least squares; IZ = inclusionary zoning; WRLURI = Wharton Residential Land Use Regulatory Index. DV: Ln(Eviction Filing Rate).

p < .10. **p < .05. ***p < .01.

As suggested by the R² values, the models explain slightly less than half of the variation in eviction filing rates. Wald tests for the joint statistical significance of the year effects (coefficients not reported for brevity) confirm that models with year-specific fixed effects outperform models that omit year effects.

Consistent with previous research, I find that eviction filing rates are higher in jurisdictions with a higher percentage of black households. Somewhat unexpectedly, I find that the percent of Hispanic households is negatively associated with eviction filing rates. Eviction filing rates are also higher in counties with a younger population, a higher percentage of one-person households, higher unemployment rates, and higher population densities.

Several housing market conditions are associated with county eviction filing rates. Consistent with Leung, Hepburn, and Desmond (2021), I find that median gross rent is nonlinearly related to eviction filing rates. Eviction filing rates are also higher in counties with higher extreme rental cost burden rates and a smaller percentage of renters relative to owner-occupants. Consistent with Preston and Reina (2021), I find that county eviction filings are negatively associated with the percentage of units that are subsidized, suggesting that housing subsidies provide a safety net that insulates subsidized households from the threat of eviction. As hypothesized by Desmond et al. (2018a), I also find that eviction filing rates are higher in Maryland, due to the state’s unique eviction filing process.

Regarding the primary policy variables of interest, I find that housing supply constraints have a larger influence on county eviction filings than pro-renter protectionist state landlord–tenant policies. In all three models summarized in Table 2, the coefficient on state landlord–tenant policies is small in magnitude and statistically insignificant from zero. In contrast, all housing supply variables are statistically significant. A one standard deviation unit increase in the WRLURI increases the county eviction filing rate by 5.8 percent, whereas a one standard deviation unit increase in housing supply elasticity reduces the county eviction filing rate by 15.6 percent. I also find that a standard deviation unit increase in the number of IZ ordinances within the county is associated with a reduction in county eviction filing rates of between 8 and 9 percent.

Controls for local land use regulations and housing supply elasticity also dampen the impact of median gross rent on eviction filing rates. Comparing model 1 with models 2 and 3, controls for housing supply conditions reduce the total percentage impact of median gross rent from 81.86 percent in model 1 to 52.65 percent in model 2 and 28.06 percent in model 3. This interaction between housing supply constraints and median rent is expected, given that restrictions on housing supply have been shown to inflate the asking price for rent. The statistical significance of the coefficients on land use regulatory measures in models that also control for median rent suggests that rents alone fail to fully capture the effect of land use regulations on county eviction filings.

Table 3 reports the results from random effects models that control for unobserved county-level heterogeneity in eviction filing rates. The overall R² values are comparable to the OLS models, but the Breusch and Pagan (1980) test statistics, which test the null hypothesis that county-specific error components are jointly equal to 0, indicate that the random effects models outperform the OLS models that ignore random effects. Wald test statistics indicate that year fixed effects are also jointly significant, so all random effects models include year-specific dummy variables (coefficients not reported for brevity).

Table 3.

Random Effects Model Regression Results.

Variable	Model 1		Model 2		Model 3
Variable	(exp(βX_s) − 1) × 100	Sig.	(exp(βX_s) − 1) × 100	Sig.	(exp(βX_s) − 1) × 100	Sig.
Constant	107.90	***	127.94	***	213.60	***
% black in total population	60.99	***	61.09	***	61.73	***
% Hispanic in total population	−3.01		−3.92		−4.74
% Age 65+ in total population	−17.93	***	−18.65	***	−20.32	***
% One-person households	9.48	***	9.55	***	9.51	***
Unemployment rate	5.67	***	5.49	***	5.22	***
Population/sq. mi.	16.16	***	14.75	***	12.15	***
Median gross rent	66.69	***	58.31	***	50.19	***
Median gross rent squared	−38.30	***	−35.84	***	−33.80	***
% Extreme rental cost burden	0.48		0.21		0.23
% Renters	−31.49	***	−30.93	***	−30.34	***
% of housing units subsidized	3.49		3.16		3.03
Maryland	34.09	***	30.72	***	31.72	***
Protectionist state landlord–tenant law	−0.49		−0.17		0.61
No. of IZ ordinances in co.			−8.95	***	−10.18	***
WRLURI			12.39	**
Housing supply elasticity					−20.73	***
N × T	3,082		3,082		3,082
R ²	.43		.44		.47
Breusch–Pagan test stat.	7,659.57	***	7,726.17	***	7,783.44	***
Wald test, year effects	49.53	***	49.88	***	51.82	***

Note: All models include year fixed effects and robust standard errors. IZ = inclusionary zoning; WRLURI = Wharton Residential Land Use Regulatory Index. DV: Ln(Eviction Filing Rate).

p < .10. **p < .05. ***p < .01.

The overall pattern of results from the random effects models is comparable to the OLS models, but there are several important differences worth noting. First, the coefficients on several variables (percent Hispanic, percentage of extremely cost-burdened renters, percentage of rental housing units that are subsidized) are no longer statistically significant at the .05 level once random effects are taken into account. Second, the size of the effect of median gross rent is much smaller, whereas the effects of local land use regulations and housing supply elasticity are larger in magnitude. These findings suggest that models that do not control for random effects overestimate the effect of rent and underestimate the effect of local land use regulations on eviction filings.

Table 4 displays the results from a regression that duplicates the model summarized in model 2 of Table 3 for a restricted sample of counties within tight housing markets. I define tight housing markets as those with housing vacancy rates less than or equal to the median vacancy rate of 10.1 percent for all counties in the sample. Table 4 suggests that in tight housing markets, increases in a county’s black population percentage and increases in a county’s unemployment rate both have much larger effects on eviction filing rates than in the full sample of counties. These findings suggest that populations at risk of being evicted face an even higher likelihood of being issued an eviction notice in tight housing markets where fewer alternative housing options exist. As hypothesized, land use restrictions also have a larger impact on eviction filing rates within tight housing markets. Specifically, within tight housing markets, the percentage increase in evictions due to a one standard deviation increase in regulatory restrictiveness is about 4 percentage points higher than in the model estimated for the full sample.

Table 4.

Random Effects Model Regression Results (Tight Housing Markets).

Variable	(exp(βX_s) − 1) × 100	Sig.
Constant	156.44	***
% black in total population	107.39	***
% Hispanic in total population	−4.65
% Age 65+ in total population	−18.97	***
% One-person households	10.41	**
Unemployment rate	12.00	***
Population/sq. mi.	14.11	**
Median gross rent	38.33	**
Median gross rent squared	−29.24	***
% Extreme rental cost burden	1.69
% Renters	−33.17	***
% of housing units subsidized	−4.59
Maryland	28.97	***
Protectionist state landlord–tenant law	0.27
No. of IZ ordinances in co.	−7.67	***
WRLURI	16.34	**
N × T	1,541
R ²	.58
Breusch–Pagan test stat.	3,791.58	***
Wald test, year effects	50.27	***

Note: All models include year fixed effects and robust standard errors. IZ = inclusionary zoning; WRLURI = Wharton Residential Land Use Regulatory Index. DV: Ln(Eviction Filing Rate).

p < .10. **p < .05. ***p < .01.

To get a better sense of how controls for land use regulations affect the estimated impact of rent on eviction filing rates, Figure 1 displays the percent change in eviction filing rates with increases in median gross rent equal to one, two, three, and four quartiles of the distribution of median gross rent, holding other control variables constant. This percent change is calculated as (exp(β_mX_m + β_m2X_m₂) − 1) × 100, where X_m is the median gross rent quartile, X_m₂ is the median gross rent quartile squared, β_m is the coefficient on median gross rent, and β_m2 is the coefficient on median gross rent squared. All simulations displayed in Figure 1 are based on the full model with random effects included (Table 3). I also simulate the effect of changes in median rent for the regulatory restrictiveness model estimated for tight housing markets only (Table 4). In Figure 1, “Baseline Model” refers to the model from Table 3 that omits controls for land use regulation and housing supply elasticity.

Figure 1.

Percent changes in eviction filing rates with quartile increases in median gross rent.

As graphically displayed in Figure 1, increases in rent from the minimum value of median gross rent to various higher quartiles are associated with substantial percentage increases in eviction filing rates, even considering the nonlinear impact of rent. Importantly, these percentage changes decline in magnitude in models that include controls for housing supply constraints, particularly with rent increases to the fourth quartile of the distribution of median gross rents. The largest reduction in the size of the rent effect is in the model estimated for tight housing markets that controls for regulatory restrictiveness and the number of IZ ordinances. These findings suggest that regulatory restrictiveness, particularly within tight housing markets, accounts for a large portion of the impact of median rent on eviction filing rates. These findings also imply that failing to control for the presence of land use regulations will tend to overestimate the impact of median rent on eviction filing rates.

Conclusion

This paper analyzed the determinants of county eviction filing rates between 2009 and 2016, emphasizing the role of land use regulations and housing supply elasticity. The results from several OLS and random effects models demonstrate that eviction filing rates increase with the restrictiveness of local land use regulations and decrease with the elasticity of housing supply and the prevalence of local IZ ordinances. These effects are larger in magnitude in models that control for county-level random effects and in models estimated for a subsample of counties within tight housing markets. Controls for land use regulation also reduce the effect of median rents on eviction filing rates. None of the estimated models provide evidence to support the claim that pro-renter state landlord–tenant laws reduce eviction filing rates.

These results provide new evidence to inform anti-eviction policy efforts. On the global stage, human rights advocates have long recognized the connection between land use planning and forced evictions. In 1997, the United Nations Committee on Economic, Social and Cultural Rights promulgated General Comment Number 7, which interpreted the human right to housing to include protection against forced evictions from urban redevelopment projects and “unbridled speculation in land” (Committee on Economic, Social and Cultural Rights 1997, 3). In 2016, the United Nations Human Settlements Programme (2016) published a how-to guide on tenure-responsive land use planning that outlined the steps that nations could take to ensure that land use planning protects, rather than threatens, the security of residential tenure for those with insecure land rights.

Within the United States, progress toward tenure-responsive land use planning is threatened by conflicts between tenant-rights advocates, who call for statutory reforms of state landlord–tenant laws, and YIMBY (Yes in My Backyard) advocates, who prioritize the removal of local regulatory barriers to housing production. Standing alongside tenants-rights advocates are Public Housing In My Backyard (PHIMBY) advocates who call for large-scale government-subsidized housing programs (Axel-Lute 2019). The evidence presented in this paper suggests that the best anti-eviction strategy likely combines several tools in the housing policy toolbox, including land use reforms that reduce constraints on housing supply and IZ policies that require a portion of new housing to be affordably priced. I also present evidence that government-subsidized housing reduces county eviction filings, although this effect disappears when inter-county differences in the unobserved determinants of eviction filings are considered. Consistent with Preston and Reina (2021), I find that evictions are concentrated in disadvantaged counties with higher unemployment rates, which suggests that government assistance to disadvantaged households may go a long way toward reducing the risk of evictions during times of economic hardship, as suggested by Allen et al. (2019).

Regardless of one’s stance on the YIMBY-PHIMBY debate, the findings from this paper imply that planners can play an important role in shaping the geography of housing affordability and residential tenure security. By working with local district courts to collect real-time eviction filing data, local planners can develop early-warning indicators of impending housing instability crises. The findings also have implications for planning education. Most graduate planning programs in the United States offer introductory courses in housing policy and land use planning, but these courses do not always address the connection between land use policy and the security of residential tenure. Moreover, most planning programs offer housing policy courses as specialized electives rather than as part of the core curriculum. The findings from this research suggest that all planning students, particularly those specializing in land use planning, would benefit from coursework that trains students to evaluate the housing market impacts of plans and policies.

I conclude with a discussion of the limitations of the study along with suggestions for future research. First, with aggregate evictions data, I am unable to differentiate individual-level effects from ecological or network effects. The finding that eviction filing rates are higher in counties with high unemployment rates, for example, may mean either that those who are unemployed face a higher risk of eviction or that in counties where unemployment is concentrated, all renters are at a higher risk of eviction, whether they are employed or not. Second, I only study eviction filings, so it is possible that the findings reported in this paper do not generalize to evictions filed that end in actual evictions. Third, the policy variables included in this study are defined for large geographies and are based on aggregate indices that may fail to account for subtle differences in policy design or implementation. Furthermore, cities, not counties, have jurisdiction over land use regulation in most states. I am also unable to exploit temporal variation in the regulatory variables to control for county-level fixed effects. Future research that combines time-varying microdata on evictions with a large sample of land use regulatory regimes would help to improve the precision of the estimates presented in this research. Qualitative research that examines the behavior of landlords and renters before and after local zoning reforms would also enrich the study of the link between evictions and housing supply.

Supplemental Material

sj-xlsx-1-jpe-10.1177_0739456X221118104 – Supplemental material for Land Use Regulations, Housing Supply, and County Eviction Filings

Supplemental material, sj-xlsx-1-jpe-10.1177_0739456X221118104 for Land Use Regulations, Housing Supply, and County Eviction Filings by Casey J. Dawkins in Journal of Planning Education and Research

Footnotes

Acknowledgements

The author wishes to thank the U.S. Department of Housing and Urban Development (HUD) for providing funding that supported this research. This paper was completed as part of the HUD Scholar in Residence program.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Funding for this research was provided by the U.S. Department of Housing and Urban Development (HUD).

ORCID iD

Casey J. Dawkins

Supplemental Material

Supplemental material for this article is available online.

Author Biography

Casey J. Dawkins is a Professor of Urban Studies and Planning at the University of Maryland and a Research Associate at the University of Maryland’s National Center for Smart Growth. His research interests include housing justice; U.S. housing policy analysis; metropolitan housing market dynamics; the causes, consequences, and measurement of residential segregation by race and income; affordable housing and transit-oriented development; and the link between land use regulations and housing affordability.

References

Abramson

Boaz.

2021. “The Welfare Effects of Eviction and Homelessness Policies.” Stanford University Working Paper. https://stanford.edu/~boaza/evictions_abramson.pdf

Alexander

Anne Kat

Lee

Sarah

. 2021. “Preliminary Analysis: A Year of Eviction Moratoria.” Princeton University. https://evictionlab.org/one-year-of-eviction-moratoria/

Allen

Heidi L.

Eliason

Erica

Zewde

Naomi

Gross

Tal

. 2019. “Can Medicaid Expansion Prevent Housing Evictions?” Health Affairs 38 (9): 1451–57.

Axel-Lute

Miriam

. 2019. “YIMBYs: Friend, Foe, or Chaos Agent?” Shelterforce: The Voice of Community Development, February 19. https://shelterforce.org/2019/02/19/yimbys-friend-foe-or-chaos-agent/

Breusch

T. S.

Pagan

A. R.

1980. “The Lagrange Multiplier Test and its Applications to Model Specification in Econometrics.” Review of Economic Studies 47:239–53.

Cobb-Clark

Deborah A.

Herault

Nicolas

Scutella

Rosanna

Tseng

Yi-Ping

. 2016. “A Journey Home: What Drives How Long People Are Homeless?.” Journal of Urban Economics 91:57–72.

Committee on Economic, Social and Cultural Rights. 1997. General Comment No. 7: The Right to Adequate Housing (Art. 11 (1) of the Covenant): Forced Evictions. Geneva: United Nations Office of the High Commissioner for Human Rights.

Coulson

N. Edward

Thao

Shen

Lily

. 2020. “Tenant Rights, Eviction, and Rent Affordability.” SSRN Working Paper. doi:10.2139/ssrn.3641859.

Curtis

Marah

Corman

A. Hope

Noonan

Kelly

Reichman

Nancy E.

2013. “Life Shocks and Homelessness.” Demography 50:2227–53.

10.

Desmond

Matthew

. 2016. Evicted: Poverty and Profit in the American City. New York: Crown.

11.

Desmond

Matthew

Gershenson

Carl

. 2017. “Who Gets Evicted? Assessing Individual, Neighborhood, and Network Factors.” Social Science Research 62:362–77.

12.

Desmond

Matthew

Gromis

Ashley

Edmonds

Lavar

Hendrickson

James

Krywokulski

Katie

Leung

Lillian

Porton

Adam

. 2018a. Eviction Lab Methodology Report: Version 1.0. Princeton: Princeton University. www.evictionlab.org/methods

13.

Desmond

Matthew

Gromis

Ashley

Edmonds

Lavar

Hendrickson

James

Krywokulski

Katie

Leung

Lillian

Porton

Adam

. 2018b. Eviction Lab National Database: Version 1.0. Princeton: Princeton University. www.evictionlab.org

14.

Desmond

Matthew

Kimbro

Rachel Tolbert

. 2015. “Eviction’s Fallout: Housing, Hardship, and Health.” Social Forces 94 (1): 295–324.

15.

Eviction Lab. 2018. “National Estimates: Eviction in America.” Princeton University. https://evictionlab.org/national-estimates/

16.

Glaeser

Edward L.

Gyourko

Joseph

Saks

Raven

. 2005. “Why Is Manhattan So Expensive? Regulation and the Rise in Housing Prices.” Journal of Law and Economics XLVIII:331–69.

17.

Glendon

Mary Ann

. 1982. “The Transformation of American Landlord-Tenant Law.” Boston College Law Review 23 (3): 503–76.

18.

Gyourko

Joseph

Saiz

Albert

Summers

Anita

. 2008. “A New Measure of the Local Regulatory Environment for Housing Markets: The Wharton Residential Land Use Regulatory Index.” Urban Studies 45 (3): 693–729.

19.

Haas

Jacob

Rangel

Jasmine

Garnham

Juan Pablo

Hepburn

Peter

. 2021. “Preliminary Analysis: Eviction Filing Trends after the CDC Moratorium Expiration.” Princeton University. https://evictionlab.org/updates/research/eviction-filing-trends-after-cdc-moratorium/

20.

Hatch

Megan E.

2017. “Statutory Protection for Renters: Classification of State Landlord-Tenant Policy Approaches.” Housing Policy Debate 27 (1): 98–119.

21.

Hepburn

Peter

Louis

Renee

Desmond

Matthew

. 2020. “Racial and Gender Disparities among Evicted Americans.” Sociological Science 7:649–62.

22.

Hepburn

Peter

Louis

Renee

Fish

Joe

Lemmerman

Emily

Alexander

Anne Kat

Thomas

Timothy A.

Koehler

Robert

Benfer

Emily

Desmond

Matthew

. 2021. “US Eviction Filing Patterns in 2020.” Socius: Sociological Research for a Dynamic World 7:1–18.

23.

Huber

Peter J.

1967. “The Behavior of Maximum Likelihood Estimates under Nonstandard Conditions.” Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability I:221–33.

24.

Johnson

Guy

Scutella

Rosanna

Tseng

Yi-Peng

Wood

Gavin

. 2015. “Entries and Exits from Homelessness: A Dynamic Analysis of the Relationship between Structural Conditions and Individual Characteristics.” Final Report No. 248, Australian Housing and Urban Research Institute, Melbourne.

25.

Leung

Lillian

Hepburn

Peter

Desmond

Matthew

. 2021. “Serial Eviction Filing: Civil Courts, Property Management, and the Threat of Displacement.” Social Forces 100 (1): 316–44.

26.

Merritt

Breanca

Farnworth

Morgan D.

2021. “State Landlord-Tenant Policy and Eviction Rates in Majority-Minority Neighborhoods.” Housing Policy Debate 31 (3–5): 562–81.

27.

National Coalition for a Civil Right to Counsel. 2022. “The Right to Counsel for Tenants Facing Eviction: Enacted Legislation.” www.civilrighttocounsel.org

28.

Phinney

Robin

Danziger

Sheldon

Pollack

Harold

Seefeldt

Kristin

. 2007. “Housing Instability among Current and Former Welfare Recipients.” American Journal of Public Health: Research and Practice 97 (5): 832–37.

29.

Preston

Gregory

Reina

Vincent J.

2021. “Sheltered from Eviction? A Framework for Understanding the Relationship between Subsidized Housing Programs and Eviction.” Housing Policy Debate 31 (3–5): 785–817.

30.

Ribar

David C.

2017. “Early Research Findings from Journeys Home: Longitudinal Study of Factors Affecting Housing Stability.” The Australian Economic Review 50 (2): 214–19.

31.

Saiz

Albert

. 2010. “The Geographic Determinants of Housing Supply.” The Quarterly Journal of Economics 125 (3): 1253–96.

32.

Somerville

C. Tsuriel

Mayer

Christopher J.

2003. “Government Regulation and Changes in the Affordable Housing Stock.” Economic Policy Review 9 (2): 45–62.

33.

United Nations Human Settlements Programme. 2016. “Tenure Responsive Land Use Planning: A Guide for County Level Implementation.” HS Number HS/075/16E. https://unhabitat.org/sites/default/files/documents/2019-05/tenure-responsive-lup-a-guide-for-country-level-implementation_.pdf

34.

U.S. Department of Housing and Urban Development. 2021. HUD to Issue Rule Protecting Tenants Facing Evictions for Non-Payment of Rent in HUD-Assisted Properties. Press Release, HUD No. 21-167. Washington, DC: U.S. Department of Housing and Urban Development.

35.

Vasquez

Jade

Calabro

Alayna

Johnson

Kim

Gallagher

Sarah

. 2022. Tenant Protections and Emergency Rental Assistance during and beyond the COVID-19 Pandemic. Washington, DC: National Low Income Housing Coalition.

36.

Wang

Ruoniu

Balachandran

Sowmya

. 2021. Inclusionary Housing in the United States: Prevalence, Practices, and Production in Local Jurisdictions as of 2019. Oakland: Grounded Solutions Network.

37.

White

Halbert

. 1980. “A Heteroskedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroskedasticity.” Econometrica 48:817–38.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.04 MB