The Link Between Local Comprehensive Plans and Housing Affordability: A Comparative Study of the Atlanta and Detroit Metropolitan Areas

Abstract

Problem, research strategy, and findings: I question whether the strength of affordable housing policies in local comprehensive plans is associated with better affordable housing outcomes, which I measure as a decrease in the share of low-income households who spend more than 30% of their income for housing, otherwise known as cost-burdened households. I first assess the strength of affordable housing policies in 58 local comprehensive plans, counting the number of—and degree of coercion in—those affordable housing policies. I then analyze the relationship between the strength of affordable housing policies and changes in the share of low-income households with cost burden. I find that the strength of affordable housing policies is higher in the Atlanta (GA) metropolitan area than in the Detroit (MI) metropolitan area. I also find that the strength of affordable housing policies is positively associated with a decrease in the share of low-income households paying more than 30% of their income for housing in the Atlanta metropolitan area. I do not find a comparable relationship between plan strength and housing outcomes in the Detroit metropolitan area. I also find that the state role matters: Georgia provides more support and guidance for local comprehensive planning, and for affordable housing policies in those plans, than does Michigan.

Takeaway for practice: Planners should continually promote local comprehensive plans that include more and stronger affordable housing policies and advocate for greater state support for comprehensive planning and affordable housing policies because these appear to lead to a greater likelihood of implementing stronger plans.

Keywords:

• comprehensive planning
• affordable housing
• plan strength
• plan outcomes

The decline of housing affordability has been a serious issue in the United States. More than 30% of U.S. households currently spend 30% or more of their income on housing, which is considered cost burdened (Harvard Joint Center for Housing Studies, 2015). It is critical to develop affordable housing policies because low-income households that cannot find affordable housing must drastically reduce their spending on other basic needs.

Local comprehensive plans are critical in guiding and regulating urban development, including ensuring a sufficient housing supply for all residents (Berke & Conroy, 2000). Many studies assess the quality of comprehensive plans, analyzing planning documents to define, for example, whether a plan provides balanced support for sustainability principles (e.g., Berke & Conroy, 2000; Jun & Conroy, 2013). Only a few studies, however, examine how local comprehensive plans are associated with such community outcomes as improved housing affordability. It is important to understand more about the association between local comprehensive plans and housing affordability.

In this study I examine the link between local comprehensive plans and housing affordability in the Atlanta (GA) and the Detroit (MI) metropolitan areas. Georgia and Michigan differ in many ways, some of which provide an opportunity to assess the relationship between the strength of affordable housing policies in local comprehensive plans and housing affordability for low-income households in circumstances that differ by state and from those studied by other scholars (e.g., Aurand, 2014).

I examine whether housing affordability for low-income households is more likely to be improved in cities adopting a local comprehensive plan with stronger affordable housing policies. First, I assess and compare the strength of the affordable housing policies in 58 local comprehensive plans in local jurisdictions in the Atlanta and Detroit metropolitan areas. Second, I examine the association between the strength of affordable housing policies and improved housing affordability for low-income households, measured by changes in shares of low-income renters and homeowners with cost burden.

I find that the strength of affordable housing policies is greater in the Atlanta metropolitan area than in the Detroit metropolitan area and that the strength of the affordable housing policies is negatively related to a growth in the share of low-income households with cost burden. In other words, there is a positive association between stronger housing policies and improved housing affordability for low-income households. My findings also suggest that the state planning environment affects the association between local comprehensive plans and housing affordability.

I do not argue that the findings from two metropolitan areas can be directly applied to other metropolitan areas or states. I do suggest that my findings can provide insights for other parts of the country and can provide an important baseline for further exploration of the association between local comprehensive plans and community outcomes.

In the sections below I first review the literature on comprehensive planning and affordable housing and what is known about plan outcomes; the review covers the very few studies that focus specifically on the outcomes of affordable housing components in plans. I also assess research on the impact of state comprehensive planning policies and programs on plan quality. I then describe the state planning regimes in Georgia and Michigan. I explain my research strategy and methods and then evaluate the differences in plan outcomes in the Atlanta and Detroit metropolitan areas. Finally, I conclude by linking local comprehensive plans and housing affordability.

Comprehensive Planning and Affordable Housing

The 1922 Standard State Zoning Enabling Act was the foundation of land use planning in the United States. Never formally adopted by Congress, it was prepared by the U.S. Department of Commerce and adopted by 19 states before 1926, most without alteration. States that adopted the act gave local governments the power to control land use through zoning but said little about the planning needed to support zoning efforts. The Department of Commerce revised the act in 1926 to define comprehensive planning and to explain what comprehensive plans should include and how they should be used to support local land use planning. By 1930 most states had adopted one or both acts or copied modifications created by other states, although not as many states adopted the 1926 act as had adopted the original zoning provisions (Meck, 2002). The 1926 act defined a comprehensive plan as the most basic guidance for shaping growth and physical development in local municipalities. Today these plans typically contain a vision for community development–specific goals, objectives and strategies designed to support those goals, and implementation plans. A comprehensive plan, as the name suggests, covers multiple substantive policy areas, although this varies by city and state; housing has traditionally been one of the major elements along with land use, environment, transportation, community facilities, and economic development (Cullingworth & Caves, 2013).

Many scholars have shown that the land use regulations that supported U.S. comprehensive planning efforts were often discriminatory, locating unattractive public facilities such as bus garages and sewer treatment plants in minority neighborhoods (Fricker & Hengartner, 2001) while allowing industrial and commercial activities to locate in poorer neighborhoods but zoning them out of more affluent communities (Nelson, Pendall, Dawkins, & Knaap, 2002; Pendall, 2000; Shertzer, Twinam, & Walsh, 2016; Whittemore, 2017a, 2017b/this issue). A wide variety of federal and state housing and other programs, moreover, favored or even required racially homogeneous neighborhoods while adopting policies that severely disadvantaged racially mixed neighborhoods (Gotham, 2000; Jackson, 1987). Cities often used zoning laws to exclude poor and minority groups from more affluent neighborhoods by prohibiting the construction of multifamily rental housing; other zoning practices raised the price of housing beyond the means of low-income and minority groups by imposing large-lot and minimum housing size requirements in wealthier or White areas (Nelson et al., 2002). Many local municipalities have been able to use land use controls to limit the supply of affordable housing and to segregate certain populations into specific neighborhoods.

Such zoning and related practices meant that some cities had no affordable housing at all; bigger cities limited affordable housing to specific neighborhoods or sprawling suburbs. More recently some states have adopted mandatory policies that govern affordable housing to address these historical inequities. States such as Florida, Oregon, and Washington have growth management strategies that require local municipalities to plan for affordable housing. Florida requires all municipalities to include a housing element in their local comprehensive plans and to address the provision of affordable housing by, for example, providing housing for all current and future residents, eliminating substandard dwelling units, and providing sites for affordable housing (Aurand, 2014).

Aurand (2014) argues, however, that the supply of affordable housing has received little attention in local comprehensive planning. Hoch (2007), in one of the few studies on the housing elements in local plans, examines affordable housing plans in Illinois that were developed in response to a statewide affordable housing planning mandate as separate plans from local comprehensive plans and prepared when municipalities and counties included less than 10% of their housing stock as affordable housing. Hoch finds that the affordable housing plans only met the minimum requirements (e.g., identifying affordable housing sites) and did not really include specific policies for affordable housing; these plans failed to mention the ways in which land use controls have excluded low-income households and did not offer solutions. Goetz, Chapple, and Lukermann (2003) also evaluate local comprehensive plans in 25 cities in the Minneapolis–St. Paul (MN) metropolitan area after the adoption of a state affordable housing mandate in Minnesota; they find that most cities did not have a sincere commitment to supplying affordable housing. Research suggests that local comprehensive plans that focus on sustainable development generally pay little attention to affordable housing because these plans emphasize only a few sustainability principles, such as a livable built environment, while giving limited attention to social equity (Berke & Conroy, 2000; Jun & Conroy, 2013, 2014), which is intimately related to the provision of affordable housing.

What Is Known About Plan Outcomes

Analyzing plan outcomes is a more complex task than analyzing plan quality, which involves analyzing planning documents based on certain criteria, because it is difficult to assess actual community outcomes years after a plan was adopted. Multiple circumstances may change, and policymakers may evaluate outcomes differently than when the plan was prepared (Aurand, 2014). It is even more difficult to identify the impact of myriad external factors on actual outcomes (Guyadeen & Seasons, 2016). Urban policies do not exist in a vacuum; as Talen (1996) notes, scholars cannot control for all factors that might create desired changes in targeted areas, such as housing affordability. This makes it difficult to assess the impact of the plan itself.

There are other uncertainties in evaluating plan outcomes. Timing is an important issue; there is no clear consensus among scholars on exactly when plan outcomes should be evaluated (Brody & Highfield, 2005). Previous studies examine plan outcomes over time periods ranging from single-digit years to more than a decade since the plan was adopted (e.g., Anthony, 2003; Aurand, 2014; Brody & Highfield, 2005). How plan outcomes should be measured is also a contentious issue. Brody and Highfield (2005), for example, examine plan outcomes by comparing environmental protection policies in local comprehensive plans with the number of wetland development permits—actual development—whereas Aurand (2014) examines plan outcomes by looking at the association between the strength of affordable housing policies in local comprehensive plans and change in the share of cost-burdened households and affordable housing units, which is a broader measure. Alexander and Faludi (1989) and other scholars suggest, however, that if a plan provides general community guidance it can be viewed as achieving its goals given the complexities in the planning process.

There are, not surprisingly, relatively few studies of plan outcomes. Alfasi, Almagor, and Benenson (2012) examine the impact of watershed protection policies in local comprehensive plans in Israel; they find that there were fundamental gaps between what plans sought and what actually occurred. Berke, Spurlock, Hess, and Band (2013) conclude that local comprehensive plans were not effective at watershed protection in North Carolina. Laurian et al. (2004) examine how well the stormwater and urban amenity management objectives of local environmental plans in New Zealand were implemented; they find that implementation of plan elements was generally low.

Guyadeen and Seasons (2016) argue that the lack of plan outcome studies is due in part to the limited number of generally accepted methodologies for evaluating community outcomes of any given plans. Oliveira and Pinho (2011) and Laurian et al. (2010) have developed methodologies to assess how well a plan achieves its goals, yet studies on plan outcomes have a relatively short history and mostly focus on whether a plan results in desired land use patterns or protects environmentally fragile areas from development rather than examining housing affordability.

Most scholars have concluded that because researchers have limited ability to study plan outcomes, they should instead focus on evaluating plan quality. Laurian et al. (2004, 2010) suggest that in the absence of agreed-upon methods to evaluate plan outcomes, researchers can assume that high-quality, strong plans bring better results. Scholars as a result have paid substantial attention to questioning what is a good and strong plan. Scholars who evaluate plan quality assess the content of adopted plans against specific criteria rather than attempting to analyze actual outcomes. Much of the recent plan quality literature focuses on how well plans incorporate sustainability as the vision for community building (e.g., Berke & Conroy, 2000; Jun & Conroy, 2013, 2014). Dalton and Burby (1994) argue that high-quality plans provide complete and detailed information about community status, address comprehensive goals, and promote policies that are intimately linked to the goals. Berke and Conroy (2000) argue that plans that provide balanced support for sustainability principles are higher-quality plans.

Some studies pay attention to the factors affecting plan quality and the implementation of plan policies. Studies find that public participation makes plans stronger but, more important, positively affects the implementation of plan policies (Burby, 2003; Conroy & Berke, 2004; Conroy & Jun, 2016). Burby (2003) finds that broad stakeholder involvement leads to achieving consensus among affected parties and thus reduces the chance of opposition to proposed plan policies, thereby making plans stronger and facilitating plan implementation.

Another group of plan quality studies focuses on the effect of state-level plan mandates on local planning efforts. Many of these studies find that state plan mandates positively affect plan quality because states often provide both some financial support for mandated local planning and consistent guidelines on what the plans should contain or address. Local governments with state-level mandates may also find it easier to overcome local political opposition to mandated planning efforts (Berke, 1996; Berke & French, 1994; Conroy & Berke, 2004; Dalton & Burby, 1994; Lens & Monkkonen, 2016). Jun and Conroy (2013) analyze the quality of exurban plans in Georgia based on how well the plans provide balanced support of sustainability principles compared with comparable plans in Ohio. They report a number of reasons why state-based comprehensive planning is critical for the higher quality of exurban plans in Georgia (e.g., requiring the inclusion of the community assessment section; see Jun & Conroy, 2013). Bunnell and Jepson (2011) argue, in contrast, that state-level plan mandates do not result in better local comprehensive plans. They compare mandated comprehensive plans with nonmandated comprehensive plans and find that nonmandated comprehensive plans are more creative and less rigid because they are not required to follow mandatory guidelines. Bunnell and Jepson also conclude that nonmandated comprehensive plans are more communicative and persuasive because the plans were developed with greater motivation for successful implementation. There are studies that offer views in opposition to Bunnell and Jepson’s study; however, many studies suggest that local comprehensive plans developed under a state-level mandate are high-quality and strong plans. Thus, affordable housing policies in mandated comprehensive plans might also be stronger.

Some studies find that high-quality plans are better implemented. Burby’s (2003) study is one of the few linking plan strength—which he defines as the number of hazard mitigation measures proposed in the plan—and community outcomes. He finds that public participation positively affects plan strength and, accordingly, results in positive community outcomes. Jun and Conroy (2014) hypothesize that communities with a high-quality plan, measured in terms of their sustainability scores, are more resilient in that they rapidly recover from decline or have a less drastic change in housing value. They find evidence to support that assumption.

There are far fewer studies of the relationship between local comprehensive plans and housing affordability. Nelson, Dawkins, and Sanchez (2004) examine 242 metropolitan areas and find that mandates requiring cities to develop local housing plans were associated with modest improvements in racial integration. Both Feiock (1994) and Anthony (2003) find that comprehensive planning mandates in Florida led to less affordable housing because the state’s Growth Management Act concurrently limited housing supply. Lewis (2005) compares municipalities that developed a comprehensive plan in accordance with California’s local housing requirements with their counterparts and finds that compliance with the state mandate was not associated with subsequent local production of multifamily housing units.

Aurand’s (2014) study, in contrast, finds a positive association between the strength of affordable housing policies in local comprehensive plans (measured by the number of policies for affordable housing) and improved housing affordability for low-income households (measured by a reduction in the share of cost-burdened low-income households and a growth in the share of affordable housing units in Florida). Florida, a growth management state, mandates that local jurisdictions plan to supply adequate affordable housing for all residents, including low-income households. Aurand finds that the state plan mandate was a critical factor in the achievement of local affordable housing goals. His interviews with local planners and housing administrators indicate that Florida’s plan mandate leads local municipalities to adopt more effective policies in their plans and to more successfully implement their affordable housing policies. Aurand feels that the state mandate achieved local implementation by publicizing the housing affordability issue and educating local officials and the public about housing affordability; this in turn increased commitment to affordable housing goals (Aurand, 2014).

In summary, there have been only a few studies on plan outcomes largely because of the lack of agreed-upon methods, concerns over when to evaluate a plan’s impact, and the complexities of planning in the urban environment. There are even fewer studies examining the link between local comprehensive plans and housing affordability. The findings of the few plan outcome studies are inconclusive. Many scholars instead address plan quality, which they measure in different ways; they generally, but not always, conclude that high-quality plans are more likely to lead cities to achieve their plan objectives.

Some plan quality studies focus on the impact of state-level planning mandates on the implementation of local plan objectives. Here, too, the results are inconclusive, and most is known about Florida. Other states with different local planning requirements may have different dynamics. I conclude that there is great value in examining the link between the housing affordability elements of local comprehensive plans and the implementation of those housing objectives in states with different planning regimes to add to understanding of plan outcomes. I do so in Georgia and Michigan.

Local Comprehensive Planning in Georgia and Michigan

Georgia and Michigan provide interesting case studies of the outcomes of comprehensive plans in achieving affordable housing mandates because each state has a very different approach to how local communities address comprehensive planning and housing affordability from each other and from Florida, the state about which the most is known. Neither state mandates local comprehensive planning. Georgia, however, has a statewide comprehensive planning program that encourages local municipalities to adopt a comprehensive plan with a housing affordability component; the state uses financial incentives to encourage local comprehensive planning. Michigan does not have a statewide planning program and does not provide guidance to local governments on how to engage in comprehensive planning or include affordable housing goals in their comprehensive plans. I describe the planning environments and local comprehensive planning processes in both states below.

Georgia and Local Comprehensive Planning

Georgia has encouraged but not mandated all local municipalities to adopt a comprehensive plan since 1989 when the state Planning Act was passed. Georgia’s statewide comprehensive planning program is incentive based (Jun & Conroy, 2013). Georgia certifies local municipalities that adopt a comprehensive plan as Qualified Local Governments, which makes them eligible to receive grants and various forms of assistance from the state (Growth Strategies Reassessment Task Force, 1998). The Georgia comprehensive planning program requires local comprehensive plans to be consistent with statewide planning goals and objectives, which include a housing affordability goal and related objectives. The state goal is “to ensure that all residents of the state have access to adequate and affordable housing.” The state’s housing opportunities objective is that “quality housing and a range of housing size, cost, and density should be provided in each community, to make it possible for all who work in the community to also live in the community” (Georgia Department of Community Affairs, 2017). The state comprehensive planning program guides local municipalities to achieve the state housing goal and objective through its Supplemental Planning Recommendations. These recommendations suggest, for example, that local governments develop homebuyer education programs to enable each household to make a better decision when purchasing a home. The recommendations also suggest providing mixed-income housing in new developments, explicitly integrating the poor with middle-income residents (Georgia Department of Community Affairs, 2016). Georgia requires municipalities adopting a comprehensive plan to designate responsibility for each plan project and identify funding sources for the project (Jun & Conroy, 2013), which seems likely to lead to plan policies being better implemented.

Michigan and Local Comprehensive Planning

Michigan permits, but does not require, communities to adopt a comprehensive plan; the state does not have a state-based comprehensive planning program, nor does it provide guidelines to local jurisdictions on how to conduct comprehensive planning. In 2008, Michigan adopted the Michigan Planning Enabling Act, which consolidated Michigan’s three existing planning statutes—the Municipal, Township, and County Planning Acts—into one law. Major provisions in the act require communities to coordinate the planning processes they have with adjacent municipalities. The 2008 act also requires zoning regulations to support the vision and policies in a comprehensive plan by, for example, promoting mixed-use development in a municipality that envisions a more walkable community (Michigan State University Extension, 2008). Michigan, unlike Georgia, has no statewide goals, objectives, or requirements about conformance of local comprehensive plans to the state’s goals and objectives. More important, Michigan provides no financial support or assistance to local jurisdictions adopting a comprehensive plan, nor does it provide guidelines for local jurisdictions to facilitate the implementation of plan elements and policies.

The literature suggests, overall, that the existence and strength of state mandates, programs, and policies are crucial aspects of plan quality that may well determine plan implementation, thereby resulting in positive community outcomes. Georgia provides far more support for local comprehensive planning efforts, and especially planning for affordable housing, than does Michigan. I question whether the affordable housing elements of local comprehensive plans in Georgia lead to more effective plans in improving housing affordability than those in Michigan.

Understanding Plan Outcomes Under Different State Planning Regimes

My review of the literature suggests that little is known about

• the outcomes of most plans, let alone the affordable housing component of local comprehensive plans;

• the impact of various elements of plan quality on community outcomes or housing affordability; and

• the role of different state planning regimes—mandates, program, and polices—on the outcomes of the affordable housing components in local comprehensive planning efforts.

Therefore, it is important to understand more about how local comprehensive plans are associated with housing affordability for low-income households. More specifically, I examine whether the strength of affordable housing policies in local comprehensive plans is related to improved housing affordability for low-income households, which is measured by a reduction in the share of cost-burdened low-income households. To examine the association between local comprehensive plans and housing affordability, I propose two specific hypotheses:

• The strength of affordable housing policies in local comprehensive plans is greater in the Atlanta metropolitan area than in the Detroit metropolitan area.

• Affordable housing policies are more influential in improving housing affordability for low-income households in the Atlanta metropolitan area than in the Detroit metropolitan area.

I first assess and compare the strength of the affordable housing policies in 58 local comprehensive plans in local jurisdictions in the Atlanta and Detroit metropolitan areas, focusing on low-income households, defined as those whose income is less than 80% of the area median income. I expect that when affordable housing policies recommended in a comprehensive plan are implemented, the share of cost-burdened low-income renters and homeowners—those who pay 30% or more of their income for housing—can be reduced. The two metropolitan areas provide an opportunity to assess the relationship between the strength of comprehensive plans and housing affordability in circumstances that differ from those Aurand (2014) studied in Florida. These two metropolitan areas face differing state planning regimes, economic challenges, and housing markets from each other and from Florida.

An Overview of the Two Metropolitan Areas

I study the Atlanta metropolitan area in Georgia, or more specifically the Atlanta–Sandy Springs–Marietta metropolitan area. The Atlanta metropolitan area is the largest metropolitan area in Georgia; its population was 4.1 million in 2000 with an area of 8,376 square miles. I study the Detroit metropolitan area in Michigan, or more specifically the Detroit–­Warren–Livonia metropolitan area. The Detroit metropolitan area is the largest metropolitan area in Michigan; its population was 4.5 million in 2000 with an area of 3,913 square miles. The two metropolitan areas had a similar number of municipalities, 110 in the Atlanta metropolitan area and 114 in the Detroit metropolitan area, although the average population size of each jurisdiction was lower in the Atlanta metropolitan area (11,780 people) than in the Detroit metropolitan area (27,785 people) in 2000. The two metropolitan areas had roughly comparable populations in 2000 and provide a basis for a comparative study examining the subsequent change in the share of low-income households with cost burden between 2000 and 2010.

The two metropolitan areas were, however, similar in population size in 2000 but very different in terms of socioeconomic characteristics and population trends. The Detroit metropolitan area is located in the Rust Belt, which has struggled economically. Between 2000 and 2010, annual unemployment rates were higher than the national average (U.S. Department of Housing and Urban Development & U.S. Department of the Treasury, 2013). Both the central city and metropolitan area lost population between 2000 and 2010 in the Detroit metropolitan area. The central city population declined from 945,300 people to 711,300 people, and the metropolitan population declined from 4.5 million to 4.3 million between 2000 and 2010. The Atlanta metropolitan area is located in the Sun Belt region and has experienced population growth in recent decades through the inmigration of people looking for better job opportunities and warm weather. The city of Detroit experienced a significant population loss between 2000 and 2010, but the population of Atlanta was relatively stable between 2000 (421,300 people) and 2010 (422,800 people). Atlanta’s metropolitan population, however, grew significantly from 4.1 million to 5.3 million people between 2000 and 2010. Socioeconomic characteristics and population trends suggest that housing affordability may be more problematic in the rapidly growing Atlanta metropolitan area. These differences imply that researchers should control for metropolitan differences when analyzing the relationship between the strength of affordable housing policies in local comprehensive plans and the change in the share of low-income households with cost burden between 2000 and 2010.

Examining the Association Between Local Comprehensive Plans and Housing Affordability

There is no established protocol to examine the association between local comprehensive plans and housing affordability. I adapt Aurand’s (2014) method to test my two proposed hypotheses. Aurand considers a comprehensive plan to be strong if it has a large number of individual affordable housing policies; he then analyzes whether plan strength is positively related to improved housing affordability for low-income households, which he measures as a reduction in the share of cost-burdened households and growth in the share of affordable housing units. He finds a positive association between plan strength and improved housing affordability for low-income households, thereby confirming the link between local comprehensive plans and housing affordability in Florida.

I evaluate plans from cities in each metropolitan area that had 5,000 or more people in 2000—the initial time point to analyze subsequent change in the share of low-income households with cost burden between 2000 and 2010—because very small municipalities are less likely to adopt a comprehensive plan, and housing affordability data are not often available for small municipalities. As of 2000, there were 60 municipalities in the Atlanta metropolitan area and 68 municipalities in the Detroit metropolitan area with more than 5,000 people. Thirty (50%) of the cities in the Atlanta metropolitan area and 28 (41%) of the municipalities in the Detroit metropolitan area with a population greater than 5,000 had adopted a comprehensive plan between 1996 and 2008.¹

To identify the strength of the plan, I initially count the number of specific policies in each comprehensive plan that are closely linked to the provision of affordable housing; ultimately it is possible for a plan to have 23 affordable housing policies. I build on Aurand’s (2014) typology of 20 policies that are typically addressed in affordable housing for low-income households, and I include three additional plan policies—preferential tax treatment, homeowner assistance, and information and education for low-income households—that I found in some of the comprehensive plans that I reviewed.²

I then go beyond Aurand’s (2014) approach (of simply counting the mention of a specific policy) and score each policy present in the plan using a 2-point system. I code words like encourage, consider, intend, or should as 1; I score a required policy, one using words such as shall, will, require, or must, as 2.³ The scoring system does not take into account frequencies of plan policies for affordable housing but only the presence of each policy. I only use the highest score for policies mentioned more than once; for example, if I originally code the inclusionary zoning policy as 1, but the plan later discusses that policy using words like shall, will, require, or must, I recode the inclusionary zoning policy as 2.

The highest possible plan score is 46 given that there are 23 plan policies for affordable housing, and each policy can have a maximum score of 2. My scoring system recognizes that the degree to which a policy is implemented can be related to the intensity of the policy requirement (Berke & Conroy, 2000; Jun & Conroy, 2013, 2014).

Table 1 shows the 23 plan policies grouped into six categories: cost reductions, zoning, location or land use map, use of public money, household assistance, and other alternatives that are linked to housing affordability. A one-stop permitting system, for example, allows cost reductions for developers by expediting affordable housing projects and thus improves housing affordability. Inclusionary zoning that requires a given share of new housing to be affordable for low- and moderate-income households can also improve housing affordability. A community land trust can improve housing affordability by leasing at affordable prices housing that is built on the land held in a land trust by a nonprofit organization.

Table 1. Potential housing policies linked to affordable housing in comprehensive plans.

Cost reductions	Use of public money
Reduction of impact fees for affordable housing One-stop permitting Excess city-owned land or vacant lots given to affordable housing organizations Preferential tax treatment for building affordable housing	HOME Investment Partnership Program Community development block grant Cooperation with housing authority State program (identified) to provide affordable housing Interlocal agreement to provide affordable housing
Zoning	Household assistance
Allow higher-density development Allow accessory dwellings Allow a variety of housing types (or multifamily) Inclusionary zoning	Homebuyer assistance Homeowner rehabilitation assistance Homeowner assistance (e.g., foreclosure prevention program) Information and education for low-income households about affordable housing opportunities
Location or land use map	Alternatives
Identified criteria/locations for low-income housing Identified criteria/locations for mobile homes and manufactured housing	Community land trust Support of fair-share housing Job training or job creation as an affordable housing solution (income strategy) Information and education for developers and nonprofits on providing affordable housing

Note: All items but those italicized are from Aurand (2014).

The mean strength of affordable housing policies in the 58 plans in both the Atlanta and Detroit metropolitan areas is 3.41 out of the highest possible score of 46. The low mean score suggests the low priority of housing affordability in local comprehensive plans in the two metropolitan areas.

To identify the association between plan strength and improved housing affordability for low-income households, I run regression analyses in which the dependent variables are growth between 2000 and 2010 in two commonly used measures of housing affordability: shares of low-income renters and shares of low-income homeowners with cost burden, or those spending 30% or more of their income on housing (U.S. Department of Housing and Urban Development, 2010). The dependent variables measure changes in the relative ratios of each municipality’s shares of low-income renters and homeowners with cost burden to its metropolitan area’s shares of low-income renters and homeowners with cost burden. I use these relative ratios to reduce the effect of other differences between the Detroit and Atlanta metropolitan areas. I also confirm that housing affordability in specific sample municipalities is not statistically significantly different between the two metropolitan areas in the initial time point of the year 2000, thereby providing a basis for analyzing subsequent changes in housing affordability between 2000 and 2010.

Analyzing changes in housing affordability between 2000 and 2010 could be problematic because the Great Recession occurred during this time period after a housing boom. Various confounding factors (e.g., foreclosures) may affect housing affordability as I define it. I recognize the challenge and take into account the effect of the recession when both testing the hypotheses and interpreting the results. In addition, I do not examine how specific affordable housing policies affect housing affordability for low-income households. A goal of future research should be to examine the impact of specific affordable housing policies on improved housing affordability as well as to use other measures of improved affordability.

In my regression analyses I evaluate the strength of affordable housing policies in the plan—the number of policies and the degree to which those policies are voluntary or not—as the key variable affecting subsequent changes in housing affordability for low-income households. I also include other local-context variables that might influence changes in housing affordability for low-income households, including population size, the number of years from plan adoption to 2010, the percentage of low-income households, and the percentage of vacant housing units.⁴ I include other explanatory variables to control for the baseline conditions in each municipality. I assume that municipal conditions in 2000 are predictive of subsequent changes in housing affordability for low-­income households between 2000 and 2010.⁵

The percentage of low-income households can increase the growth in cost-burdened households because these households have a higher demand for affordable housing. The percentage of low-income households can also act as a control for the income distribution in the municipalities I am studying. I include a variable for population size because larger municipalities may have more resources to develop affordable housing than smaller municipalities as well as greater planning staff capacity to implement plan policies (Aurand, 2014; Conroy & Jun, 2016). I include the growth in total population because population growth tends to increase housing prices and thus can lower housing affordability for low-income households. I include the percentage of vacant housing units to control for housing characteristics in each municipality. Higher vacancy rates usually reflect lower housing demand, which can lead to greater housing affordability for low-income households (Aurand, 2014). I include the number of years since plan adoption to the year 2010 because older plans give cities more time to implement plan policies.

Determining How Plan Strength Affects Affordable Housing

My analyses suggest that overall, communities in the Atlanta metropolitan area have comprehensive plans that include more affordable housing polices and those with more mandatory language, and thus are stronger plans, than those in the Detroit metropolitan area. Table 2 shows that the mean and median strength of affordable housing policies in the sample plans are greater in the Atlanta metropolitan area (4.90 and 4.00, respectively, out of 46) than in the Detroit metropolitan area (1.82 and 1.00, respectively, out of 46). The highest strength of affordable housing policies among local comprehensive plans in the Atlanta metropolitan area is 18.00 out of 46, whereas the highest strength of affordable housing policies among community comprehensive plans in the Detroit metropolitan area is 7.00 out of 46. The Atlanta metropolitan area also has fewer communities that lack any type of affordable housing policy: A fourth (7 of 28) of the sample municipalities in the Detroit metropolitan area have no affordable housing policies in their comprehensive plans, whereas less than 4% (1 of 30) of comparable communities in the Atlanta metropolitan area lack such policies.

Table 2. Strength of affordable housing policies.

Variable	Atlanta (GA) metropolitan area	Detroit (MI) metropolitan area
Mean strength of affordable housing policies	4.90	1.82
Median strength of affordable housing policies	4.00	1.00
Highest strength of affordable housing policies	18.00	7.00
% Comprehensive plans with no specific affordable housing policies	3.3%	25.0%
Note: The highest strength possible of affordable housing policies is 46 (23 × 2).

To test my first hypothesis that the strength of affordable housing policies in local comprehensive plans is greater in the Atlanta metropolitan area than in the Detroit metropolitan area, I run a t-test analysis of the mean score of the strength of affordable housing policies in the two metropolitan areas. The strength of affordable housing policies in local comprehensive plans in the Atlanta metropolitan area is statistically significantly stronger/higher (4.9) than in the Detroit metropolitan area (1.8; t = 3.8, 
p < .001), which supports my first hypothesis that the strength of affordable housing policies in local comprehensive plans is greater in the Atlanta metropolitan area than in the Detroit metropolitan area.⁶

I examine the link between local comprehensive plans and housing affordability by running regression analyses using the strength of affordable housing policies as the key independent variable, as shown in the Technical Appendix. I find that the strength of affordable housing policies in the Atlanta metropolitan area is negatively related to the growth in both low-income renters and homeowners with cost burden. In other words, there is a positive association between the strength of affordable housing polices and improved housing affordability for low-income households.

My regression analysis for the Detroit metropolitan area, however, shows that the strength of affordable housing policies is not statistically significant for the growth of either low-income renters or homeowners with cost burden. The fact that the strength of affordable housing policies is statistically significant only in the Atlanta metropolitan area supports the second hypothesis that affordable housing policies are more influential in improving housing affordability for low-income households in the Atlanta metropolitan area than in the Detroit metropolitan area. The relationships between other control variables and the dependent variable are discussed in the Technical Appendix.

In summary, I find overall that the strength of affordable housing policies in local comprehensive plans has a statistically significant positive effect on improved housing affordability for low-income households in the Atlanta metropolitan area. The strength of affordable housing policies in local comprehensive plans in the Detroit metropolitan area does not have a statistically significant effect on improved housing affordability for low-income households.

Evaluating the Differences Between the Atlanta and Detroit Metropolitan Areas in Plan Outcomes

I question in this section why the strength of affordable housing policies is related to the change in housing affordability for low-income households only in the Atlanta metropolitan area and suggest that having a state-based comprehensive planning program in Georgia makes a difference.

The various rules and requirements in Georgia’s comprehensive planning program might explain a large share of the differences in the statistical significance of the strength of affordable housing policies on the change in the share of low-income households with cost burden between the two metropolitan areas. I test this assumption by running an additional regression model that includes all samples from both metropolitan areas and a dummy variable for a plan from Georgia and an interaction variable between the dummy variable indicating a plan from Georgia and 
the strength of affordable housing policy (excluding the strength of affordable housing policy due to collinearity). I find that the interaction variable is negatively related to the growth in the share of low-income homeowners with cost burden; the strength of affordable housing polices in local comprehensive plans adopted in Georgia is negatively related to the growth in the share of low-income households with cost burden.

The interaction variable is not statistically significantly related to the growth in the share of low-income renters with cost burden; this may be because using only two metropolitan areas provides a low level of variability. However, the statistical significance of the interaction variable still supports the hypothesis that Georgia’s comprehensive planning program plays an important role in improving housing affordability for low-income households at the local level.⁷

How does Georgia’s state-based comprehensive planning program promote housing affordability for low-income households? Local plan policies may be better implemented in Georgia than Michigan because of the assistance that the state of Georgia provides local jurisdictions in the Atlanta metropolitan region. I do not examine whether a specific policy was implemented or not, but studies suggest that stronger or high-quality plans are better implemented (Aurand, 2014; Burby, 2003; Laurian et al., 2004). Overall plan quality, which I do not directly address, may be higher in the Atlanta metropolitan area than in the Detroit metropolitan area because of the state-based comprehensive planning program. It is likely that overall plan quality is lower in the Detroit metropolitan area in the absence of a supportive state program, leading to lower levels of implementation of affordable housing policies.

The role of public participation in the two states may also explain some of the differences between the two metropolitan areas (Burby, 2003; Conroy & Berke, 2004; Conroy & Jun, 2016). Georgia’s comprehensive planning program requires public participation in local comprehensive planning processes, which accordingly is likely to lead to better implementation of plan policies. Georgia municipalities preparing a plan are required to identify stakeholders, use techniques to encourage participation, and schedule public meetings. Georgia requires localities to hold two public meetings as part of the local planning process to allow a diverse body of stakeholders and interest groups to express their desires and opinions (Georgia Department of Community Affairs, 2017). Georgia’s comprehensive planning program identifies specific participation techniques such as strengthening public information resources by, for example, creating and maintaining a website that is dedicated to public participation and holding open houses, public hearings, and public workshops that inform the public of the activities and progress of certain projects. The state also recommends a general schedule for public participation.

Georgia’s minimum planning requirements also promote the implementation of plan policies by requiring a short-term work program that sets the implementation schedule, designates responsibility for each project, and identifies funding sources. The city of Cartersville’s (GA) plan, for example, reports that the city will adopt a density bonus for which the city council and planning department are responsible and states that the city plans to use its general fund for the program. The city’s comprehensive plan also states that the city will work cooperatively with the independent housing authority in providing affordable housing using the housing authority’s budget for the plan policy but that the housing authority is in charge of providing affordable housing.

I do not argue that having a state-based comprehensive planning program is the only reason why the strength of affordable housing policies in local comprehensive plans is positively related to a reduction in the share of low-income households with cost burden in the Atlanta metropolitan area and not in the Detroit metropolitan area. It is hard to deny the importance of the state role, however.

The Link Between Local Comprehensive Plans and Housing Affordability

Local comprehensive plans can play an important role in ensuring affordable housing for low-income households, but few studies have examined the link between local comprehensive plans and housing affordability. To fill the gap in the literature, I proposed two hypotheses:

1.	The strength of affordable housing policies in local comprehensive plans is greater in the Atlanta metropolitan area than in the Detroit metropolitan area.
2.	Affordable housing policies are more influential in improving housing affordability for low-income households in the Atlanta metropolitan area than in the Detroit metropolitan area.

I tested these hypotheses to examine whether the strength of affordable housing policies in local comprehensive plans is associated with improved housing affordability for low-income households.

I first evaluated the strength of affordable housing policies in local comprehensive plans in the Atlanta and Detroit metropolitan areas to examine the relationship between local comprehensive plans and housing affordability for low-income households. The t-test analysis showed that the strength of affordable housing policies is statistically significantly higher in the Atlanta metropolitan area than in the Detroit metropolitan area, as hypothesized. Next, I ran regression analyses by including local context variables that can influence housing affordability for low-income households in addition to the strength of affordable housing policies variable in examining changes in housing affordability for low-income households. My regression estimates suggested that the strength of affordable housing policies is positively related to improved housing affordability for low-income households—a reduction in the share of low-income households with cost burden—in the Atlanta metropolitan area but not in the Detroit metropolitan area, which supports the second hypothesis.

Not many studies have paid attention to how local comprehensive plans are associated with certain community outcomes. There has been a default view that a plan will be satisfactorily implemented or has other positive values. The finding that the strength of affordable housing policies is positively related to improved housing affordability for low-income households in the Atlanta metropolitan area suggests that stronger plans result in better community outcomes. Thus, planners should continually promote and advocate local comprehensive plans, including more and stronger affordable housing policies.

Planners should recognize at the same time that changes in housing affordability after plan adoption can vary by planning regime. Statewide comprehensive planning programs can create an environment in which local affordable housing policies are more likely to be implemented. Michigan’s current planning legislation neither explicitly promotes affordable housing nor addresses sustainable development that is linked to social equity and thus housing affordability.

Future studies will need to expand the study areas, use different measures of plan outcomes, and address other confounding issues to confirm the robustness of the findings in this study. There is great value in assessing local comprehensive plans, examining the effect of plan policies for affordable housing in additional metropolitan areas and states to test the results in this study. It would be useful for future research to expand the timeframe to see whether the findings in this study are applicable in the coming decades given that the 2000s was a unique period.

Finally, future studies will need to take into account other factors in analyzing the link between local comprehensive plans and housing affordability. Other municipality-specific factors for which my measures do not effectively control may affect the change in housing affordability for low-income households. The presence of strong nonprofit organizations and/or advocacy groups, a high level of institutional commitment, and political leadership in specific municipalities may all act to promote the implementation of local comprehensive plan policies, thereby improving housing affordability for low-income households. Additional case studies for in-depth understanding of the mechanism and barriers to effective implementation of plan policies for affordable housing will contribute significantly to the literature.

This study is one of the few attempts at quantitatively examining the link between local comprehensive plans and housing affordability, and it sheds some light on the role of the state in local comprehensive plans and of those local plans in community outcomes. I do not claim that I can directly generalize the findings from two metropolitan areas to other metropolitan areas or states. I argue that my research can provide insights for other parts of the country and an important baseline for further exploration of plan outcomes.

Acknowledgments

I would like to thank Marisa Laderach and Hyun Jeong for assistance in preparing the data.

Research Support

This work was supported by the Ministry of Education of the Republic of Korea and the National Research Foundation of Korea (NRF-2016S1A3A2925463).

Supplemental Material

Supplemental data for this article can be found on the publisher’s website.

Additional information

Funding

This work was supported by the Ministry of Education of the Republic of Korea and the National Research Foundation of Korea (NRF-2016S1A3A2925463).

Notes on contributors

Hee-Jung Jun

Hee-Jung Jun ([email protected]) is an associate professor in the Department of Public Administration and Graduate School of Governance at Sungkyunkwan University, Korea.

Notes

1.  At first, I intended to examine the effect of local comprehensive plans adopted between 1996 and 2005 on subsequent changes in housing affordability between 2000 and 2010. However, in the plan selection process, I noticed that a relatively large number of plans compared with the total possible obtainable plans were adopted after 2005. To retain a large enough sample size, I also included plans adopted between 2006 and 2008. In the United States, decennial census data are no longer available, and scholars use 5-year average census data as an alternative form of 2010 census data. Thus, I used 2006–2010 census data as 2010 census data for those municipalities that adopted a comprehensive plan between 1996 and 2005 and 2008–2012 census data as 2010 census data for those municipalities that adopted a comprehensive plan between 2006 and 2008 to consider the time lag in examining the subsequent changes in housing affordability between 2000 and 2010. In making the decision, I confirmed that there was no statistically significant difference in the frequency of plans adopted between 2006 and 2008 between the two metropolitan areas by running a chi-square test. I also included a dummy variable indicating dependent variable values of different time in the regression analyses. The dummy variable was not statistically significant in either the Atlanta or Detroit metropolitan area models, and thus I omitted it from the final analyses.

2.  As this study applies Aurand’s (2014) study to other metropolitan areas, I measured the strength of each plan by counting the number of policies for affordable housing as undertaken in his study. Burby (2003) and Brody and Highfield (2005) assess plan strength by counting the number of specific hazard mitigation policies and environmental policies to protect wetlands, respectively. In future studies, researchers may want to analyze how plan quality, for example, measured by sustainability scores as in Berke and Conroy’s (2000) study, rather than plan strength, is associated with changes in housing affordability for low-income households.

3.  In evaluating plan strength, I considered each housing policy to have equal strength. Although a housing policy can have a different level of impact from another policy on the supply of affordable housing, the impact may vary by municipality depending on community conditions. Therefore, as Burby (2003) and Aurand (2014) did, I developed an equally weighting additive index for the strength of affordable housing policies. I also did not differentiate individual policies that might focus on housing affordability in either rental or housing markets because rental and housing markets are interrelated. For example, financial assistance for first-time homebuyers that can enhance housing affordability for low-income homeowners can also improve housing affordability for low-income renters. This is because the reduction in low-income renters because they become homeowners with the assistance means an increase in affordable rental housing units available to the number needed (Aurand, 2014).

4.  I considered percentage of White people, percentage of rental housing units, and other factors that might affect subsequent changes in housing affordability for low-income households. However, given that there were a small number of samples in this study, percentage of White people and percentage of rental housing units were highly correlated (more than .6) with percentage of low-income households and thus omitted from the regression analyses. None of the demographic and housing characteristics variables included in the regression analyses were highly correlated (less than .6).

5.  Other time-differenced estimators were not used because they were endogenously related to changes in housing affordability. For instance, a change in low-income households between 2000 and 2010 is endogenously associated with the change in housing affordability for low-income households.

6.  One might assume that housing affordability was not a serious issue in the Detroit metropolitan area because of its depressed housing markets. There was no statistically significant difference, however, between local municipalities in the Atlanta and Detroit metropolitan areas in the share of cost-burdened renters and homeowners in 2000: The mean was 54.2% of low-income renters with cost burden in the Atlanta metropolitan area and 51.9% in the Detroit metropolitan area 
(t = 0.75, p = .46) and 48.1% and 44.8%, respectively, of low-income homeowners (t = 1.33, p = .19). In other words, these insignificant differences provide baseline conditions for the comparison in changes from 2000 to 2010 in housing affordability between the two metropolitan areas.

7.  In addition, one might think that increasing diversity associated with population growth in the Atlanta metropolitan area is related to the change in housing affordability rather than the strength of plan policies for affordable housing. I ran additional regression analyses for each metropolitan area by including an interaction variable between population growth and percentage of White people, but the interaction variable was not statistically significant in any of the models.