homeowners in mature suburbs, to analyze how property values of single family homes in ..... Mature Suburbs in Cuyahoga, Franklin, and Hamilton Counties.
ANALYZING MATURE SUBURBS THROUGH PROPERTY VALUES
DISSERTATION
Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University
By Katrin B. Anacker, MCRP *****
The Ohio State University 2006
Dissertation Committee: Approved by Professor Hazel A. Morrow-Jones, Adviser Professor Donald R. Haurin Professor Anand Desai Professor Elena Irwin
___________________________________ Adviser City and Regional Planning Graduate Program
ABSTRACT
The United States can be characterized as a nation of suburban homeowners. More than 50 percent of Americans live in suburbs, and almost 19 percent of them live in so-called first suburbs. Nationwide the homeownership rate is almost 70 percent. For the majority of homeowners, a home represents the biggest investment in their lives and many expect their property values to increase. If mature suburbs have problems or are in decline, then property values may decrease and investment value will be lost. Recently, many mature suburbs have become concerned that history will repeat itself and they will see the same decline that central cities have witnessed. This study uses mixed methods (i.e., expert interviews and regressions) to create a definition of mature suburbs, to provide an overview of public policies that benefit homeowners in mature suburbs, to analyze how property values of single family homes in mature suburbs have behaved compared with those in central cities and developing suburbs, and to analyze what specific factors have influenced property values of single family homes in mature suburbs compared with those in central cities and developing suburbs. The geographic scope of this study is three counties in Ohio: Cuyahoga (Cleveland area), Franklin (Columbus area), and Hamilton (Cincinnati area). Expert interviews reveal that despite the common perception of a policy blindspot some policies do benefit homeowners in mature suburbs. However, current policies should be modified to benefit both mature suburbs in need and homeowners that want to ii
break even on their investment. Overall, mature suburbs are unique. Despite many stereotypes, there is variety in terms of residents and housing stock that should receive attention in future research efforts. Quantitative analyses of the mature suburbs in the study show that there is no overall suburban decline in terms of property values, although some communities should be concerned with their appreciation rates. The analyses also show that housing space and adjacency to work places and transportation networks are important variables that should be factored in by policy makers. The situation that mature suburbs are in does not seem to be hopeless.
iii
To the memories of Lydia Hoherz, who taught me tenacity; Rudolf Anacker, who taught me industriousness; Frieda Anacker, who taught me endurance; and Wilhelm Hoherz, who was not allowed to teach me but nevertheless left his mark.
iv
ACKNOWLEDGMENTS
To acknowledge everyone who had a hand in this dissertation would be daunting, as it is undoubtedly the product of a massive group effort. If I must start somewhere, it should be with my doctoral adviser, Dr. Hazel Morrow-Jones, who took time out of her extremely busy schedule to give me constructive and patient advice and encouragement as well as excellent comments. Dr. Donald Haurin and Dr. Elena Irwin offered outstanding criticism, especially of the quantitative analysis. Dr. Anand Desai helped in shaping the dissertation, especially the public policy part. Ivan Maric answered many SAS-related questions tirelessly—often while adding his special kind of humor. Wenqin Chen assisted with the preparation of several variables for the database. Dr. Thomas Bier and Dr. Steven Howe were very kind to read parts of the study in an earlier draft form. Numerous academic colleagues, namely Dr. Ned Hill, Dr. Brian Mikelbank, and Dr. Ziona Austrian, offered excellent insights during the formative stage. Supporting databases were kindly provided by Barry Bennett, Thomas Bier, Andrew Dobson, James Luebbers, Cheri Mansperger, Ronald Miller, Ivan Maric, Nancy Reger, Mark Salling, and Robert Swisher. Furthermore, several experts helped to mold my mind, including Kimberly Gibson, William Miller, Janet Keller, Howard Maier, Robert Layton, and Terry Schwarz. Kent Montlack, Lou Tisler, and Thomas Moeller also expanded my thinking on several issues. Furthermore, I would like to thank all the interviewees who took the time to meet with me during my three field trips. v
Although support from people is invaluable, research funding from institutions is also quite critical. Funding for this study was provided by the Center for Urban and Regional Analysis at The Ohio State University, the U.S. Department of Housing and Urban Development through an EDRSG Grant, the Lincoln Institute of Land Policy, the Urban Land Institute, and Lambda Alpha International through Dissertation Fellowships. Funding of my graduate studies at OSU was provided by the Center for Urban and Regional Analysis at The Ohio State University for seven quarters, by the Austin E. Knowlton School of Architecture for seven quarters, and by City and Regional Planning for one quarter. Lastly, I would like to thank my parents, Erika and Jürgen Anacker, for financial and mental support on a monthly (i.e., paying the premium of my German health insurance) and weekly (i.e., placing a phone call each weekend at 8 a.m. with German punctuality) basis. In retrospect, I have to marvel at this interesting intellectual journey that started on January 18th 2002 at 11:30 am in Derby Hall and that will be marked by its first milestone on June 11th, 2006 at 1:00 pm in the Ohio Stadium. May there be many milestones to follow the first!
vi
VITA
May 13, 1968…………………….Born — Oelde, Germany
1999………………………………Master of Arts (Public Policy and Management), The Ohio State University
1999………………………………Master of City and Regional Planning, The Ohio State University
FIELDS OF STUDY
Major Field: City and Regional Planning Specializations: Urban Studies Housing Real Estate Agricultural Economics Public Policy and Management
vii
TABLE OF CONTENTS Page Dedication………………………………………………………………………….......iv Acknowledgments………………………………………………………………….......v Vita…………………………………………………………………………………….vii List of Tables……………………………………………………………………………x List of Figures………………………………………………………………………….xiv Chapters: 1.
Introduction………………………………………………………………………1
2.
Literature Review: Suburban Change……………………………………………7 2.1 2.2
3.
Models of Urban Change………………………………………………..7 Suburban Change………………………………………………………11 2.2.1 Behavior of Property Values…………………………………………...11 2.2.2 Factors of Suburban Property Values: Selected Housing Factors……………………………………………....12 2.2.2.1 Accessibility…………………………………………………………....15 2.2.2.2 Effective Property Tax Rate……………………………………………16 2.2.2.3 Public Safety……………………………………………………………19 2.2.2.4 Public School Quality…………………………………………………..20 2.2.2.5 Other Housing Characteristics………………………………………….24 2.2.3 Factors of Suburban Property Values: Selected Socioeconomic Factors………………………………………..24 2.2.3.1 Income…………………………………………………………………..25 2.2.3.2 Race……………………………………………………………………..26 Definition, Data, and Methods…………………………………………………..29
3.1 3.2 3.3
Definition of Mature Suburbs……………………………………………30 Quantitative Data………………………………………………………...40 Methods…………………………………………………………………..51
viii
4.
Public Policies…………………………………………………………………..56 4.1 4.1.1 4.1.2 4.1.3 4.2 4.2.1 4.2.2 4.2.3
5.
Results……………………………………………………………………………68 5.1 5.2 5.2.1 5.2.2 5.2.3 5.2.4 5.3 5.3.1 5.3.2 5.3.3 5.3.4 5.4 5.5 5.6 5.7 5.8
6.
Current Policies………………………………………………………….56 Federal Policies…………………………………………………………..57 State, Regional, and County Policies…………………………………….58 Local Policies…………………………………………………………….61 Current Suburban Tools………………………………………………….63 “Sticks”…………………………………………………………………..63 “Carrots”…………………………………………………………………65 Other Tools………………………………………………………………67
Spatial Patterns…………………………………………………………...68 Themes from Expert Interviews………………………………………….73 Population………………………………………………………………..73 Resource/Expenditure Imbalance………………………………………..75 Housing…………………………………………………………………..77 School Quality…………………………………………………………...79 Descriptive Statistics……………………………………………………..80 Metropolitan Area Characteristics……………………………………….80 Comparisons among Central Cities……………………………………...80 Mature Suburbs Compared to Central Cities and Developing Suburbs…84 Mature Suburbs in the Three Central Counties…………………………..93 House Price Behavior/Appreciation……………………………………..97 Regression Models……………………………………………………..108 Discussion of Results…………………………………………………..122 Recommended Policies…………………………………………………127 Suburban Advocacy…………………………………………………….131
Summary, Conclusion, and Future Research…………………………………..135
References………………………………………………………………………………146 Appendix………………………………………………………………………………..170
ix
LIST OF TABLES
Table
Page
1
Variables Used in Quantitative Analysis………………………………………43
2
Variable Means for Cuyahoga, Franklin, and Hamilton Counties for Properties Sold between 1985 and 2000 (Population of Sales)………………..81
3
Descriptive Statistics of Cuyahoga County, Cleveland, and Cuyahoga County’s Mature and Developing Suburbs…………………………………….86
4
Descriptive Statistics of Franklin County, Columbus, and Franklin County’s Mature and Developing Suburbs…………………………………….88
5
Descriptive Statistics of Hamilton County, Cincinnati, and Hamilton County’s Mature and Developing Suburbs…………………………………….91
6
Descriptive Statistics of Mature Suburbs in Cuyahoga, Franklin, and Hamilton Counties (combined and separate)…………………………………..95
7
Sale Price Appreciation 1986 – 1999: Cuyahoga, Franklin, and Hamilton Counties (combined and separate)…………………………………..99
8
Sale Price Appreciation 1986 – 1999: Cuyahoga County, Cleveland, and Cuyahoga County’s Mature and Developing Suburbs ……………………….101
9
Sale Price Appreciation 1986 – 1999: Franklin County, Columbus, and Franklin County’s Mature and Developing Suburbs …………………………103
10
Sale Price Appreciation 1986 – 1999: Hamilton County, Cincinnati, and Hamilton County’s Mature and Developing Suburbs ………………………..105
11
Sale Price Appreciation 1986 – 1999: Mature Suburbs in Cuyahoga, Franklin, and Hamilton Counties (combined and separate)………………......107
12
Log-Linear Weighted Least Squares Regression: Cuyahoga, Franklin, and Hamilton Counties (combined and separate)………………………………….109
x
13
Log-Linear Weighted Least Squares Regression: Cuyahoga County, Cleveland, and Cuyahoga County’s Mature and Developing Suburbs…………111
14
Log-Linear Weighted Least Squares Regression: Franklin County, Columbus, and Franklin County’s Mature and Developing Suburbs…………..113
15
Log-Linear Weighted Least Squares Regression: Hamilton County, Cincinnati, and Hamilton County’s Mature and Developing Suburbs…………115
16
Log-Linear Weighted Least Squares Regression: Mature Suburbs in Cuyahoga, Franklin, and Hamilton Counties (combined and separate)………..117
17
R Squareds in Central Cities and Mature and Developing Suburbs in Cuyahoga, Franklin, and Hamilton Counties…………………………………..119
A.1
Cuyahoga County: Number of Occupied Housing Units and Proportion of Homeowners in Municipalities…………………………………………………171
A.2
Franklin County: Number of Occupied Housing Units and Proportion of Homeowners in Municipalities…………………………………………………175
A.3
Hamilton County: Number of Occupied Housing Units and Proportion of Homeowners in Municipalities…………………………………………………177
A.4
Interviewees in Cuyahoga County……………………………………………..180
A.5
Interviewees in Franklin County……………………………………………….183
A.6
Interviewees in Hamilton County………………………………………………184
A.7
Cuyahoga County: Places and Designation According to Author’s Definition..187
A.8
Franklin County: Places and Designation According to Author’s Definition….190
A.9
Hamilton County: Places and Designation According to Author’s Definition…191
A.10
Multicollinearity: Cuyahoga, Franklin, and Hamilton Counties (combined)…..193
A.11
Multicollinearity: Cuyahoga County…………………………………………..194
A.12
Multicollinearity: Franklin County…………………………………………….195
A.13
Multicollinearity: Hamilton County……………………………………………196
A.14
Breusch-Pagan and White Tests for Cuyahoga, Franklin, and Hamilton Counties (combined and separate)……………………………….……………..197 xi
A.15
Population 1990, 2000, and Change from 1990 to 2000 in Cuyahoga County’s Suburbs ……………………………………………………………..198
A.16
Population 1990, 2000, and Change from 1990 to 2000 in Franklin County’s Suburbs……………………………………………………………..202
A.17
Population 1990, 2000, and Change from 1990 to 2000 in Hamilton County’s Suburbs……………………………………………………………...204
A.18
Cuyahoga County: Places and Designation According to Author’s Definition and Community’s Eligibility for Cuyahoga County’s Housing Enhancement Loan Program (HELP)………………………………………….207
A.19
Hamilton County: Places and Designation According to Author’s Definition and Community’s Eligibility for Hamilton County’s Home Improvement Program (HIP)…………………………………………………..210
A.20
Population Over 65 Years of Age 1990, 2000, and Change from 1990 to 2000 in Cuyahoga County’s Suburbs…………………………………………..212
A.21
Population Over 65 Years of Age 1990, 2000, and Change from 1990 to 2000 in Franklin County’s Suburbs…………………………………………….219
A.22
Population Over 65 Years of Age 1990, 2000, and Change from 1990 to 2000 in Hamilton County’s Suburbs…………………………………………..223
A.23
Sale Price Appreciation Based on Moving Averages 1986 – 1999: Cuyahoga, Franklin, and Hamilton Counties (combined and separate)……….228
A.24
Sale Price Appreciation Based on Moving Averages 1986 – 1999: Cuyahoga County, Cleveland, and Cuyahoga County’s Mature and Developing Suburbs……………………………………………………………229
A.25
Sale Price Appreciation Based on Moving Averages 1986 – 1999: Franklin County, Columbus, and Franklin County’s Mature and Developing Suburbs……………………………………………………………230
A.26
Sale Price Appreciation Based on Moving Averages 1986 – 1999: Hamilton County, Cincinnati, and Hamilton County’s Mature and Developing Suburbs……………………………………………………………231
A.27
Sale Price Appreciation Based on Moving Averages 1986 – 1999: Mature Suburbs in Cuyahoga, Franklin, and Hamilton Counties (combined and separate)……………………………………………………….232
xii
A.28
Elasticities for Cuyahoga, Franklin, and Hamilton Counties (combined and separate)…………………………………………………………………233
A.29
Elasticities for Cuyahoga County, Cleveland, and Cuyahoga County’s Mature and Developing Suburbs…………………………………………….235
A.30
Elasticities for Franklin County, Columbus, and Franklin County’s Mature and Developing Suburbs…………………………………………….237
A.31
Elasticities for Hamilton County, Cincinnati, and Hamilton County’s Mature and Developing Suburbs…………………………………………….239
A.32
Elasticities for Mature Suburbs in Cuyahoga, Franklin, and Hamilton Counties (combined and separate)…………………………………………..241
xiii
LIST OF FIGURES
Figure
Page
1.
Mature Suburbs in Cuyahoga County …………………………………………70
2
Mature Suburbs in Franklin County…………………………………………...71
3
Mature Suburbs in Hamilton County…………………………………………..72
4.
Sale Price Appreciation 1986 – 1999: Cuyahoga, Franklin, and Hamilton Counties (combined and separate)……………………………………………..98
5.
Sale Price Appreciation 1986 – 1999: Cuyahoga County, Cleveland, and Cuyahoga County’s Mature and Developing Suburbs………………………..100
6.
Sale Price Appreciation 1986 – 1999: Franklin County, Columbus, and Franklin County’s Mature and Developing Suburbs………………………….102
7.
Sale Price Appreciation 1986 – 1999: Hamilton County, Cincinnati, and Hamilton County’s Mature and Developing Suburbs…………………………104
8.
Sale Price Appreciation 1986 – 1999: Mature Suburbs in Cuyahoga, Franklin, and Hamilton Counties (combined and separate)…………………..106
xiv
CHAPTER 1 INTRODUCTION “[…] first suburbs are caught in a policy blindspot. Unlike central cities, they are not poor enough to qualify for many federal and state reinvestment programs and not large enough to receive federal and state funds directly. Unlike newer suburbs, they are ill suited to federal and state programs that focus on building new infrastructure and housing rather than maintaining, preserving and renovating what is already built. In general, first suburbs are undermined by large federal and state policies that, on balance, facilitate sprawl and concentrate poverty. These policies set the dominant ‘rules of the development game’ that ultimately shape metropolitan growth in ways that undermine older communities.” Robert Puentes and Myron Orfield, “Valuing America’s First Suburbs: A Policy Agenda for Older Suburbs in the Midwest” (2002)
The United States of America can be characterized as a nation of suburban homeowners (Alba & Logan, 1991; Berger, 1960; Dobriner, 1963; Douglas, 1925; Guest, 1971; Harris, 1943; Lake, 1981; Logan & Golden, 1986; Mieszkowski & Mills, 1993; Mills & Price, 1984; Muller, 1981; Pinkerton, 1969; Schnore, 1972, 1963b, 1957, 1956; Schnore & Sharp, 1963; Wood, 1972; Wood, 1958). The majority of Americans live in suburbs (Orfield, 2002), and almost 19 percent of them live in so-called first suburbs (Puentes & Warren, 2006). Nationwide the homeownership rate is almost 70 percent (2000 U.S. Census). For the majority of homeowners a home represents the biggest investment in their lives and many expect their property values to increase. If mature suburbs have problems or are in decline, then property values may decrease and
1
investment value will be lost. With a large proportion of home owners living in mature suburbs, this is a serious public policy concern. The repercussions of change in mature suburbs can be linked to arguments used by opponents of sprawl. Outward movement requires public investment in infrastructure at the urban fringe, but this infrastructure is already in place in mature suburbs (Howe et al., 1998). Land at the urban fringe, formerly used primarily for agriculture, is now used for residential development. However, residential development does not pay for itself in terms of tax revenues provided versus services required (American Farmland Trust, 2002; P. Feldmann, personal communication, November 3, 2003). The broader society pays for this inefficiency and for duplicate infrastructures. At the local level, municipalities depend on tax revenues, the most important of which are income tax, sales tax, and property tax. To oversimplify, when people who are well off reside in expensive homes in a municipality, income tax and residential property tax revenues are relatively high and therefore municipal services and amenities, including schools, are usually of high quality. When mature suburbs have problems property values are usually affected. This, in turn, affects tax revenues and services such as schools, parks and roads. If the revenue situation declines, then the municipality might have problems meeting the requirements set by the public sector (e.g., state school boards) and the expectations of residents. It will become more difficult to keep homeowners and attract potential homebuyers. Property values may thus continue to cycle downward. After the 1950s many central city neighborhoods experienced an out-migration of middle-income households and an in-migration of lower-income households (Bradford & Kelejian, 1973; Ellen, 2000; Guest, 1972; Guterbock, 1976; Palen, 1998) although lower-
2
income households moved out as well (T. Bier, personal communication, March 22, 2006). These developments, related to the phenomenon of filtering, have affected property values. It is commonly assumed that unless a central city housing unit is in a gentrified neighborhood its real value has either remained stagnant or declined since the 1950s (D. Beach, personal communication, September 9, 2003; Berry, 1985; Bier & Post, 2003; Gale, 1979; Glass, 1964; Grebler, 1952; Hoyt, 1960; Lowry, 1960; Ratcliffe, 1949; Sternlieb, 1969; Sternlieb & Burchell, 1973; Sternlieb et al., 1974; Varady, 1986; Wilson, Margulis & Ketchum, 1994; Wyly & Hammel, 1999). Until the 1990s it was also commonly assumed that most mature suburbs had steady if not increasing real property values. However, recently many mature suburbs have become concerned that history will repeat itself and that they will see the same decline that central cities have witnessed. Recent research supports this fear (e.g., Bier, 1991, 2001; Bier & Howe, 1998; Howe et al., 1998; Hudnut, 2003; Lucy & Phillips, 1995, 1997, 2000, 2001a, 2001b; Orfield, 2002; Puentes & Orfield, 2002; Puentes & Warren, 2006; see also Blumberg & Lalli, 1966; Bollens, 1986, 1988; Clay, 1979; Connolly, 1973; Farley, 1970; Frey, 1985, 2000; Guest, 1978b, 1980; Keating, 1994; Kim, 2003; Logan & Alba, 1995; Marshall & Stahura, 1979; Nelson, 1980; Pendleton, 1975; Roof & Spain, 1977; Rose, 1965; Rubinowitz, 1973; Schneider & Phelan, 1993; Schnore, Andre & Sharp, 1976; Stahura, 1979b, 1983). Mature suburbs are also often called first ring suburbs (Rokakis & Katz, 2001), inner-ring suburbs (First Suburbs Consortium Housing Initiative, 2002), inner-ring cities (Advisory Commission on Intergovernmental Relations, 1984), inner suburbs (Sutker, 1974), first suburbs (Puentes & Orfield, 2002; Puentes & Warren, 2006), first-tier
3
suburbs (Hudnut, 2003), older suburbs (Kotkin, 2001; Lucy & Phillips, 2001a, among others), older hubs (Listokin & Beaton, 1983), or mature suburbs (Listokin & Beaton, 1983), among other labels. These labels are used interchangeably in the literature (Hudnut, 2003), indicating that there is not yet a generally accepted definition. The multitude of labels also indicates that not all troubled suburbs are mature suburbs, and that not every mature suburb is a troubled suburb (Fernandez & Pincus, 1982; see also Bollens, 1988; Mikelbank, 2004). A definition of mature suburbs will be presented in Chapter 3. The primary objective of this research is to analyze property values in the mature suburbs of the three major metropolitan areas in Ohio as case study areas. The terms property value, house price, and sale price—although there are differences between them—will be used interchangeably in this dissertation for the sake of clarity and simplicity. In this study the following research questions are asked: •
How can mature suburbs be defined?
•
What public policies benefit homeowners in mature suburbs?
•
How have property values of single family homes behaved in mature suburbs compared to single family homes in central cities and developing suburbs?
•
What specific factors influence the property values of single family homes in mature suburbs compared to single family homes in central cities and developing suburbs?
Different authors have used different measures of decline. Some, such as Lucy and Philips (2000), Orfield (2002), and Puentes and Warren (2006), use resident-related 4
socioeconomic variables (see also Bier & Howe, 1998). Others, such as Bier (2001) and Margulis (2002), use housing-related variables. An example of a resident-related socioeconomic variable is Average Household Income. However, using this variable to analyze suburban decline entails potential issues. American society has been characterized by a high proportion of 1-person households, and this proportion has increased over time. Whereas in 1990 24.56 percent of all households were 1-person households, by 2000 25.82 percent of them were (1990 and 2000 U.S. Census). Similarly, in 1990 the average household size was 2.71 people per household whereas in 2000 it was 2.67 people (1990 and 2000 U.S. Census)—not to mention changes in the number of workers within households (S. Howe, personal communication, March 12, 2006). It is assumed that in the future the proportion of 1-person households will continue to increase while the average household size will continue to decrease. Typically, 1-person households (i.e. those with one earner) have lower household incomes than households with several earners, and a decrease in household size means a decrease in household income. Taking household income as a variable symptomatic of suburban decline is therefore risky. A decrease in household income does not necessarily translate into suburban decline—it might only reflect a broader societal trend. Along with this trend, the number of households in some mature suburbs might decrease due to vacancies. A better alternative to household income as a measure of decline is a housingrelated variable such as sale amount of housing unit. Housing units are fixed in space and reflect the characteristics of their surroundings. They are not influenced by the characteristics of in-movers and out-movers per se. Therefore, this dissertation will use the sale amounts of housing units as a measure of potential decline.
5
This dissertation is structured as follows: Chapter 2 will be a literature review followed by data and methods in Chapter 3. Chapter 4 will provide a policy analysis and Chapter 5 will cover the empirical results. The final chapter will provide a summary, a conclusion, and suggestions for future research.
6
CHAPTER 2
LITERATURE REVIEW: SUBURBAN CHANGE
2.1. Models of Urban Change Many human ecology and urban economic models predict continual outward growth of urban areas. This outward growth is typically driven by those who have upper and middle incomes moving out, although those who have lower incomes move out as well (T. Bier, personal communication, March 22, 2006). On the other hand, people who have fewer choices with regard to location often remain in inner neighborhoods, along with some higher-income residents (S. Howe, personal communication, April 2, 2006). Many of these inner neighborhoods have had problems, sometimes severe ones, and they have often been characterized as neighborhoods in decline. As discussed above, many mature suburbs have been concerned that history will repeat itself and that they will see the same decline that central cities have witnessed. Discussed below are select theories that fall into two groups: first, those that primarily focus on land use, and second, those that stress land values, house prices, and rents. Theories that focus primarily on land use were suggested by Burgess (1925), Hoyt (1939), and Harris and Ullman (1945), among others. Burgess (1925) discussed growth and change in cities, stating that the expansion of a city can be illustrated by a series of
7
concentric circles, each containing different types of economic and cultural groups of businesses and residents. However, he did not provide much rationale for the assignment of activities to the zones (Muth, 1969). Burgess stated that the inner zones tend to extend their areas by invading the next outer zone, and that urban expansion can be described in terms of extension, succession, and concentration. The zone of deterioration, or zone in transition near the Central Business District (CBD), is expected to expand outwards over time. Hoyt (1939) criticized and modified Burgess’s concentric circle theory, suggesting a sector theory based on rents at the block level. Confirming Burgess in part, Hoyt admitted that there is an upward gradation of rents from the center to the periphery. However, he observed that this is not the case from the CBD to the edges of the city in all directions: instead of concentric rings there would be a pattern of sectors. With respect to the pattern of movement of residential rental neighborhoods, Hoyt observed that high rent residential areas pull the growth of the entire city in the same direction, aptly called the point of highest rents or the high rent pole. “Residential rents grade downward from this pole as lesser income groups seek to get as close to it as possible. This high rent pole tends to move outward from the center of the city along a certain avenue or lateral line” (Hoyt, 1939, p. 114). Harris and Ullman (1945) overlaid the internal structures suggested by Burgess (1925) and Hoyt (1939) and suggested the multiple nuclei model due to the existence of physical factors and separating factors such has rent levels and transportation facilities and lines. Their model distinguishes among different types of land uses in general and among different levels of residential uses based on rents. Harris and Ullman observed that
8
certain groups profit from cohesion, such as retail or financial and office-building districts, while other activities are detrimental to each other, such as factory development and high-class residential development. Spatial theories that stress land values, house prices, and rents were suggested by Alonso (1964), Muth (1969), and Mills (1972), among others. Alonso (1964) introduced a theory of urban land values to urban land uses. He developed the residential bid price curve for individual residents in different spatial settings. For example, Alonso analyzed rent patterns of two competing neighboring centers of the same size; rent patterns of two competing neighboring centers of unequal size; a possible rent and occupancy pattern for two complementary centers with an island of high-income population; a possible rent and occupancy pattern for two complementary centers with populations side by side and an island of low income; and the rent and occupancy patterns for a city with a center and a high-status road. Based on theories discussed above, as well as others, Muth (1969) developed a theory of the spatial pattern of the housing market in cities. He considered the influence of accessibility to the CBD upon the location of households, the effects of the age of buildings and neighborhoods, and the determinants of housing quality or condition, among other phenomena. Mills (1972) developed several models of urban structure. One of his models concerns the urban area residential sector. He shows that land values are very high near the centers of large urban areas, although they decrease rapidly with distance and the curve flattens in the suburbs. He also shows how transportation costs and the substitution of land and other inputs cause downtown land values to be much higher than those in the suburbs.
9
These and other models suggested by human ecologists and urban economists have shown both that cities typically expand outward and that residents with higher incomes move outward. Residents with lower incomes typically remain near the center of a city. Continued movement to the urban fringe raises the question of whether the urban models hold for suburbs and whether mature suburbs will face decline in the future as some inner-city neighborhoods have experienced over time. The theories discussed above were suggested several decades ago when central cities and their neighborhoods were the focus. Mature suburbs have since moved into the academic limelight, and it is important to investigate whether Burgess’s (1925) zone of deterioration or zone in transaction that used to be near the central business district has expanded over time to such a degree that it has reached the mature suburbs. Where do mature suburbs stand in terms of property values and appreciation rates in comparison with central cities and developing suburbs, as discussed for central cities and their neighborhoods by Hoyt (1939) and Harris and Ullman (1945)? Would the observations of Muth (1969) and Mills (1972) apply to mature suburbs and, if not, how would their models have to be modified?
10
2.2 Suburban Change Suburban change can be analyzed through selected housing factors and selected socioeconomic factors to answer two of the four research questions posed in Chapter 1: How have property values of single family homes behaved in mature suburbs compared to single family homes in central cities and developing suburbs? What specific factors influence the property values of single family homes in mature suburbs compared to single family homes in central cities and developing suburbs? In this dissertation, suburban property values will be analyzed through house price behavior (Section 2.2.1) and through selected aspects of property values such as accessibility (Section 2.2.2.1), property tax rate (Section 2.2.2.2), public safety (Section 2.2.2.3), and school quality (Section 2.2.2.4), among with other housing factors (Section 2.2.2.5). Aspects of suburban property values will also be analyzed through socioeconomic factors such as income (Section 2.2.3.1 ) and race (Section 2.2.3.2).
2.2.1 Behavior of Property Values Researchers are interested in trends of property values for a multitude of reasons. The values of owner-occupied homes represent more than twice the equity located in corporations—the 1990 gross value of owner-occupied homes in the United States was $4.6 trillion while corporate equity was only $2.4 trillion (Poterba, 1991; see also Skinner, 1989). Less than ten years later the gross value of owner-occupied homes in 1998 had almost doubled to $9.4 trillion (Glaeser, Gyourko & Saks, 2004; see also Abraham & Hendershott, 1996; Capozza & Seguin, 1996; Case & Shiller, 1987, 1988, 1989, 1994). Thus, while every individual homeowner has a major interest in what
11
happens to property values, the nation overall has a very large stake in the value of owner-occupied residential property. The behavior of property values affects household wealth and, potentially, consumer spending (Di, 2005). Selling an appreciated home results in resources that can be spent on a higher priced home that might be located further away from the city (see Skinner, 1989), or in the ability to spend on other goods and services. Although most home owners expect price appreciation, there is no guarantee that it will happen. Whereas some research (e.g., Case & Shiller, 1987; Case & Mayer, 1996) provides the overall rate of appreciation over time, other research (Case & Shiller, 1994) differentiates among different price tiers, with low, middle, and high tiers ranked by price and the behavior of these price tiers over time.
2.2.2 Factors of Suburban Property Values: Selected Housing Factors Burgess (1925) pointed out that variations in land values [sic] indicate “perhaps the best single measure of mobility, and so of all the changes taking place in the expansion and growth of the city” (Burgess, 1925, p. 61). In the United States, property values typically contain the price of the structure as well as the price of the land on which the structure is located. Once a structure is built the land price component is subsumed under the price of a property. Thus, land values for urbanized areas are difficult to obtain. Another measure has to be taken for analysis.
12
Hoyt (1939) hypothesized that the average block rent is representative of a series of other housing factors, such as tenant occupancy, the condition of dwellings, units lacking private baths, nonwhite occupancy, units lacking central heat, and overcrowded dwelling units. Hoyt used average monthly rents for all tenant-occupied and vacant units in addition to owner-occupied units in selected cities 1 . Hoyt’s hypothesis resembles a regression in which rents are influenced by independent variables. Because Hoyt’s analyses were undertaken in the mid-1930s when homeownership rates—especially in urban areas—were low, rents were an appropriate variable. Now, however, homeownership rates are high (see Tables A.1, A.2, and A.3 in the Appendix), especially in the suburbs, so property values are the more appropriate variable. They are also able to be collected in a more timely way. With respect to differences in property values within metropolitan areas, earlier studies were undertaken by Burgess (1925), Hoyt (1939), Harris and Ullman (1945), Alonso (1964), Muth (1969), and Mills (1972), as discussed above. More recent studies have been undertaken by Archer, Gatzlaff, and Ling (1996) and Brasington (2002b). Some studies have focused on property values of metropolitan areas as a whole, others have dealt with a comparison of property values between different metropolitan areas, and others have focused on a comparison of property values within a metropolitan area. Factors that contribute to property values can be grouped into supply-side factors (i.e., factors related to the construction or provision of homes) and demand-side factors (i.e., factors related to the homebuyer) (Colton, 2003). Most studies list several factors from each group as explanations for the variety in property values.
1
See remark in Hoyt (1939) in Figure 11, page 33.
13
Examples of supply variables that influence the variation in property values are price of farmland (Ozanne & Thibodeau, 1983; Manning, 1986, 1989), number of municipalities in the metropolitan statistical area (Ozanne & Thibodeau, 1983), zoning (Glaeser, Gyourko & Saks 2004, 2005; Hamilton, 1978; Jud, 1980: Mark & Goldberg, 1986; Pogodzinski & Sass, 1991), construction costs (Poterba, 1991; Witte, 1975), percent change in building costs (Manning, 1986, 1989; Abraham & Hendershott, 1992), and number of building departments per 100,000 population (Manning, 1986, 1989). Examples of demand variables are population level and growth rate (Manning, 1986, 1989), employment (Abraham & Hendershott, 1992), income-related variables such as total real household income (Manning, 1986, 1989) or mean income of metropolitan households (Fortura & Kushner, 1986), change in real income (Poterba, 1991; Abraham & Hendershott, 1992), percentage change among high-income households (Manning, 1986, 1989), variation of nonelderly singles (Ozanne & Thibodeau, 1983), proportion of households that are non-family households (Fortura & Kushner, 1986), and changes in the real after-tax financing costs (Abraham & Hendershott, 1992). Other variables that influence the variation in property values are anticipated rate of inflation (Fortura & Kushner, 1986), expectations of future property values based on past price movements and hearsay, popular clichés, casual observations, and images people have (Case & Shiller, 1988). Obviously, there is a wide range of possible variables one could choose to explain house prices. The selected housing factors chosen for this research are accessibility, property tax rate, public safety, public school quality, and other housing characteristics.
14
These factors will be justified and discussed in detail later in this section, and they are also part of the quantitative model discussed in Chapter 3.
2.2.2.1 Accessibility Location and accessibility influence households’ commuting costs and disposable income, therefore affecting property values. Monocentric urban models (e.g., Muth, 1969) predict declining rent gradients per unit of land [sic] and suggest that the CBD is the point of maximum accessibility. Other advantages, in the simplest version of the model, include the fact that the CBD is the only location of employment (Jackson, 1979) and retail. Therefore, households who have the least to spend on transportation tend to locate near the CBD in high density housing—to reduce the per unit cost—while those with more money can afford to choose to spend it on more land at the edge of the city, with the longer commutes that entails. The difference in costs—primarily transportation costs—associated with the two locations influences the difference in land rents. Over the last several decades urban structure has changed from a monocentric to a polycentric form (Kim & Morrow-Jones, 2005). Major places of employment and retail are now often located along transportation lines. Polycentric urban models (Heikkila et al., 1989; McDonald & McMillen, 1990) suggest that the transportation system, with its major arteries, has influenced the rent gradient. Whereas some accessibility studies find a declining rent gradient (Anas, 1981; Gin & Sonstelie, 1992; Voith, 1991), other studies come to different conclusions (Blackley & Follain, 1987). The monocentric model suggests that the CBD is the point of maximum ability, but alternative measures of accessibility are suggested by polycentric models, such as the distance to rail lines (Bajic,
15
1983; Benjamin & Sirmans, 1996; Damm et al., 1980; Gatzlaff & Smith, 1993; McDonald & Osuji, 1995) or the distance to a highway access point (Bruinsma, Rienstra & Rietveld, 1997; Kockelman, 1997a, 1997b; Ridker & Henning, 1967). In current urban areas the importance of commuting may have declined (Robert Layton, personal communication, September 10, 2003; Kim & Morrow-Jones, 2005). Both Distance to the CBD and Distance to a Highway Access Point will be part of the quantitative model discussed in Chapter 3.
2.2.2.2 Effective Property Tax Rate Housing units are embedded in neighborhoods, which are, in turn, embedded in jurisdictions. This section examines the literature on the effect of taxes and some public services on property values and their appreciation (except for school districts, which will be discussed below). Capitalization occurs when lower taxes or higher public service levels lead to higher property values and vice versa (Brasington, 2002a; see also Edelstein, 1974; Orr, 1968; Palmon & Smith, 1998; Pollakowski, 1973; Richardson & Thalheimer, 1981; Richman, 1967; Yinger, 1982). Assuming negative capitalization, taxes to be paid in the future should cause the homebuyer to discount the price he or she is willing to pay. Thus, property values should be negatively impacted by higher taxes (Muth, 1969). Coefficients of a variable related to taxes should be negative: The higher the future taxes for a home owner the lower the current property value, all other things being equal. Neighborhood and jurisdiction effects on property value appreciation are part of the capitalization debate that addresses the question of whether capitalization of taxes and
16
public services into property values occurs. The property tax rate may also capture the effect of additional services that are not explicitly accounted for in the hedonic equation (Brasington, 2002a). There seems to be general agreement on capitalization in most cases (for exceptions see Chinloy, 1978; Hyman & Pasour 1973a, 1973b; Wales & Wiens, 1974). One of the supporters of capitalization is Oates (1969 and 1973), who used per capita public expenditures in his analysis and argued that if a community increases its property taxes unaccompanied by an increase in the output of local public services, then the bulk of the increase in taxes will be capitalized in the form of reduced property values (because the landowner’s income is reduced). Whereas there is general agreement on capitalization (see Mieszkowski & Zodrow (1989) for an excellent overview), there is less agreement on whether there is undercapitalization or overcapitalization (see Church, 1974; Crane, 1990; Ihlanfeldt & Jackson, 1982; Mieszkowski, 1972; Palmon & Smith, 1998, among others), whether capitalization is negative or positive, and on the degree of capitalization. Oates (1969), for example, suggested negative capitalization to a large extent (supported by Edel & Sclar, 1974; Lewis & McNutt, 1979; Smith, 1970b; Wicks, Little & Beck, 1968, among others). Assuming everything else is constant, full capitalization of taxes occurs when differences in property values exactly equal the present value of variations in expected tax liabilities. Yinger et al. (1988) pointed out that the degree of capitalization depends on the information available, on the expectations of the home buyer, and on the sources of change in property tax rates in a particular community (see also Linneman (1978) with respect to specification issues). Results for jurisdictions in different metropolitan areas
17
range from a capitalization rate of about 7 percent (Rosen, 1982), to 35 percent (Li & Brown, 1980), to 60 percent (Goodman, 1983), to 62 percent (Palmon & Smith, 1998). Results for different communities within a jurisdiction range from 9 to 79 percent, depending on data (Yinger et al., 1988). The capitalization of property values and local spending decisions are linked (i.e., local spending is higher in places with a greater degree of capitalization, as Hilber and Mayer (2001) argued in the case of Massachusetts; see also Bradbury, Mayer & Case, 2001). In turn, fiscal variables and amenities are capitalized to a much greater extent in areas with little available land (i.e., an inelastic supply of residential land). Suburban locations with little available land have relatively high spending on local schools, even after controlling for other factors that might affect demand for education. These ranges show the complexity of both the capitalization phenomenon and the housing markets. With respect to property taxes specifically, there are several types of tax rates that could potentially be used for an analysis. First, there is the stated tax rate; second, there is the effective tax rate (i.e., the house’s property tax payment divided by its market value (Case, 1978)); and third, there is the mean assessment to price ratio. Chinloy (1978) recommended the effective rate, since the actual rate will lead to biased estimates because
18
the home owners’ tax credits are not factored in properly in the actual rate (for a different opinion see Ihlanfeldt and Jackson (1982)). Thus, analyses in Chapter 3 will use the Effective Property Tax Rate.
2.2.2.3 Public Safety Public safety, a local public good, is difficult to define and, thus, difficult to measure. In many cases, the inverse is taken: An increase in the level of public safety can be interpreted as a decrease in the probability that an individual will become a victim of crime and vice versa. Individuals have different levels of aversion to different types of crimes. Also, various types of crime tend to be highly collinear (Clark & Cosgrove, 1990). Studies use different variables as proxies for public safety. Clark and Cosgrove (1990) chose murder (per capita) as a representative type of violent crime for two reasons. First, measures of the number of murders tend to be accurate; and second, because murders receive relatively more media attention than other types of crime, homebuyers are likely to be aware of them. Other proxies are the number of serious crimes (i.e., murder, forcible rape, robbery, aggravated assault, motor vehicle theft, and arson) per 100,000 residents (Brasington, 2002a), the number of major crimes (details not specified in Kain and Quigley (1970)), and the proportion of people between 16 and 21 years old who are high school dropouts (Li & Brown, 1980; see also Alba, Logan & Bellair, 1994; Liska & Bellair, 1995; Liska, Logan & Bellair, 1998; Logan & Messner, 1987; Logan & Schneider, 1984; Stahura, Huff & Smith, 1980). The Number of Murders
19
per 1,000 People was chosen for the analysis in Chapter 3, based on Clark and Cosgrove’s (1990) argument.
2.2.2.4 Public School Quality Much of the capitalization debate has addressed the effect of school quality on residential property values. Public schools exert an important influence on residential location decisions and the demand for housing (Hayes & Taylor, 1996; Jud & Watts, 1981). More local expenditures are devoted to schooling than to any other public service. School quality is one of the large components of the house price premium, and it can be measured in many ways. Nevertheless, “quantifying the link between school characteristics and house prices has been exceedingly difficult” (Downes & Zabel, 2002, p. 2). The scale of the data with respect to measurement and analysis is important. The larger the scale (i.e., the smaller the area covered) the more precise the results seem to be. For example, Downes and Zabel (2002) suggested that school-level variables are significantly better in describing property values than district-level data, assuming the consistency of the assignment of a house to a particular school building. Many measures of school quality can be differentiated between measures of input and measures of output. However, other factors, such as a student’s hard-to-disentangle genetic endowment and his or her family’s socioeconomic status (e.g., parental education, family income, family size, etc.) or peer group characteristics, determine individual achievement and, thus, school output measures (Hanushek, 1986, 1996, 1997;
20
Henderson, Mieszkowski & Sauvageau, 1978; Summers & Wolfe, 1977; see also Coleman (1966) quoted in Hanushek (1986)). There are several input variables that can be used in order to test capitalization, such as the student/teacher ratio and expenditure per student. These variables are often highly correlated with each other, so it is difficult to be sure when one has used the best measure. With respect to the student/teacher ratio, Grether and Mieszkowski (1974) found that in their study the effect of the student/teacher ratio on house price is significant and negative, although the effects of this variable are varied when it is used in interaction terms. Brasington (1999) found that the student/teacher ratio is consistently capitalized into house prices. Expenditures per student are an often-used measure of school resources. Such expenditures can be used to reduce class size, purchase equipment, or fund a broader range of courses (Crone, 1998). Nevertheless, since this indicator is a measure of inputs, not of outputs, it is unsatisfactory to many authors (see Rosen & Fullerton, 1977). Gustely (1976) included noneducational and educational expenditures per capita among other variables, and concluded that the expenditure variable is insignificant though negative. Brasington (1999), on the other hand, found that expenditure per student is consistently capitalized into property values, although he pointed out that variations in spending may have nothing to do with the quality of output provided (see also Oates, 1969; Pogodzinski & Sass, 1991; Sonstelie & Portney, 1980). Many studies use test scores as a measure of output, along with school report cards and other similar items (Black, 1999; Weimer 2001). Although test scores may not reflect school quality properly because of neighborhood or family characteristics or peer
21
group or other effects, they are visible, public, and accessible to homebuyers or movers who factor this information into their home buying and selling decisions. In the majority of studies, test scores have a positive and significant impact on property values. For example, Kain and Quigley (1970) found a positive coefficient for public school achievement, although they use caution in their interpretation because they think that the influence of better schools on value may be represented by the residential-quality variables. Conversely, Jud and Watts (1981) and Jud (1985) concluded that the reading test scores of elementary public schools (i.e., district mean reading scores in third grade) have a strong, positive, and statistically significant impact on property values (see also Walden, 1990). In addition to these studies, Weimer and Wolkoff (2001) also found that the average score of the English Language Arts (ELA) exam at the elementary school level is significant. In sum, improving the quality of an elementary school, especially in terms of its test scores, has a large potential for increasing property values in its surrounding neighborhoods. Haurin and Brasington (1996) found that public school quality has a very large positive impact on real constant-quality property values, more on a per lot basis than on a per square foot of land basis. Figlio and Lucas (2000) departed from the traditional approach of proxying school quality by using test scores, instead examining the influence of school report cards (i.e., letter grades) on property values. In tune with studies that use test scores, they found a positive capitalization. Despite the advantages the measures of outputs such as test scores and report cards offer, they are also criticized because they do not provide unambiguous school measures. Hanushek and Taylor (1990), among others, argued that, for example, SAT scores are misleading because of measurement problems, since in many cases they do not
22
differentiate between students in public and private schools. Also, there is a selection bias since the proportions of students taking the test varies dramatically across states. Crone (1998) pointed out that test scores might not necessarily represent what the school has contributed to the academic development of a student. For example, students with higher innate abilities tend to have higher test scores than those with lower innate abilities. Therefore, improvement (see Reinhard (1981) who included variable improvement in reading level between first and third grade (measured in months)) or value added measures (Hanushek, 1986; see also Brasington & Haurin, 2006) might be worth analyzing. Also, neighborhood or family characteristics, peer group effects, and reputation (see Brasington, 1999; Henderson, Mieszkowski & Sauvageau, 1978; Jud, 1985; Phares, McKenna & Werner, 1974; Summers & Wolfe, 1977; among others) might be significantly related to achievement. In sum, there is consensus that school quality exerts an important influence on property values. However, quantifying this link has been difficult since both inputs and outputs are influenced by direct as well as indirect factors. In addition to this, the measurement of the direct factors and selection bias complicate matters. Nevertheless, the Proportion of Students who Passed the Statewide Mathematics Test in 6th Grade was chosen as a measure to analyze school quality in Chapter 3 below.
23
2.2.2.5 Other Housing Characteristics Property values are not only influenced by the accessibility, property tax rate, public safety, and school quality of a neighborhood, but also by the structural variables of the housing unit such as size and age (Ball, 1973; Kain & Quigley, 1970; Rosen, 1974; Sirmans & Macpherson, 2003). Previous studies came to the conclusion that square footage, or a proxy (Clark & Herrin, 2000; Grether & Mieszkowski, 1974; Ridker & Henning, 1967), and the number of full bathrooms is positively correlated with property values (Clark & Herrin, 2000; Grether & Mieszkowski, 1974; Ridker & Henning, 1967). With respect to the influence of the age of the housing unit on property values, results are complex since the relationship is probably non-linear and nonmonotonic (Appraisal Institute, 2001; Cannaday & Sunderman, 1986; Chinloy, 1977; Goodman & Thibodeau, 1997; Hotelling, 1925; Hulten & Wykoff, 1981; Margolis, 1982; Palmquist, 1979; Rubin, 1993). On the one hand, a negative coefficient of age can be interpreted as depreciation (i.e., decreased productivity and more costly maintenance), while on the other hand, a positive coefficient of age can be interpreted as a premium, or vintage, effect (Clapp & Giaccotto, 1998; Malpezzi, Ozanne & Thibodeau, 1987; Randolph, 1988).
2.2.3 Factors of Suburban Property Values: Selected Socioeconomic Factors As mentioned above, suburban change can be measured by analyzing changes in the socioeconomic status of residents—something that many studies have done (Burgess, 1924, 1925; Chlodin, Hanson & Bohrer, 1980; Collver & Semyonov, 1979; Farley, 1964; Guest, 1974, 1978a; Pinkerton, 1973; Schnore, 1963a, 1964; Schnore & Jones, 1979; Smith, 1970a; Stahura, 1979a; Winsborough, 1963; Wirt et al., 1972). Many authors who
24
have focused on suburban decline use resident-related socioeconomic variables (Lucy and Philips, 2000; Orfield, 2002 and Puentes and Warren, 2006). However, because status is difficult to measure quantitatively, alternative variables will be used in this dissertation. The following discussion will focus only on home owning residents. Since the homeownership rate in suburbs is usually high, homeowners there have a lot at stake and are sensitive to community changes. Therefore, owner occupied properties are a good surrogate for overall residential properties.
2.2.3.1 Income Neighborhood incomes are related to neighborhood house prices in a positive fashion—typically, the higher the household income the higher the property value. For example, Manning (1986, 1989) suggested that higher demand for property (i.e., the number of households and the expected inflation rate, among other factors) causes property values to increase since the supply of land is upward sloping (i.e., increasing) (see Browne (1982) but also Ozanne and Thibodeau (1983) for mixed results). He concluded that higher rates of increase in after-tax household income contribute to higher rates of property value appreciation in a significant fashion. Similarly, Fortura and Kushner (1986) came to the conclusion that a one percent increase in the income of householders raises property values by 1.11 percent. Nellis and Longbottom (1981; see also Buckley & Ermisch, 1983) concluded that a 10 percent rise in real permanent income leads to a rise in property values of 18 percent. More specifically, there are several measures that could potentially be used for an analysis. Ozanne and Thibodeau (1983) started out with average income per household,
25
but adjusted this variable by calculating the average logarithm of income and factoring in a recession. Manning (1986) used the change in real household income and adjusted for interurban variation in household living costs and the change in household non-monetary income received in the form of urban amenities (e.g., low air pollution, favorable climate, etc.) (see also Roback (1982) and Rosen (1979)). Fortura and Kushner (1986) suggested taking the median income of households or the mean income of households (Fortura & Kushner, 1986). Nellis and Longbottom (1981) used the personal disposable income. Analyses in Chapter 3 use Average Household Income per Census Tract, as will be discussed below in more detail.
2.2.3.2 Race Many studies argue that the racial composition in a neighborhood has an influence on property values and their appreciation (see Hoyt, 1939, among others). McEntire (1970) pointed out that the fear of financial loss is important. Sutker, Gilman, and Plak (1974) noted that the most feared concomitant of racial transition in a neighborhood is the expected influx of lower-status residents who are presumed to change the character of the neighborhood by failing to maintain property, increasing the incidence of theft and vandalism, lowering standards in the schools, introducing variant life styles, and so on. This litany of possible negative outcomes is expected to lead to declining property values. The literature differentiates between a static racial composition and a dynamic one. With respect to a static racial composition, some studies found a negative effect of the racial composition in the neighborhood on property values (for example Bailey, 1966 (for property values in the blocks near the block of sale); Fisher, 1923; Hoyt, 1939;
26
McMichael & Bingham, 1923; Schneider, 1927). The reason for the negative effect is, as McEntire (1970) pointed out, that most non-Hispanic Whites are not willing to live in areas near other racial groups, and that the demand of other racial groups for housing is not sufficient to replace the vanished non-Hispanic White demand. Hence, property values must fall. Other studies also found a negative effect of the racial composition in a neighborhood on property values, but distinguished effects based on socioeconomic levels, physical condition of the neighborhood before entry, and value ranges of homes in the area (for example, Babcock, 1932; Babcock, Massey & Greene, 1938; Hoyt, 1933; McDonald, 1953; Neiswanger, 1937; Parker, 1943; Ross, 1955). Still other studies argue that there is either no effect or a positive effect from the racial composition in a neighborhood on property values (Abrams, 1955; Beehler, 1945; Laurenti, 1960; MacPherson & Sirmans, 2001; Myrdal, 1944; Jud, 1985; Weaver, 1948). With respect to a dynamic racial composition, there is the threshold issue, called the “tipping point,” “tip-point,” “scare point,” “preference point,” “wish-to-leave point,” “willingness-to-enter point,” “tipping in,” or “tipping out” (Grodzins, 1958; Schelling, 1978; Sutker, Gilman & Plax, 1974; Taeuber & Taeuber, 1965; Wolf, 1963; see also Holifield, Holloway & Morrow-Jones, 2000). Many non-Hispanic Whites prefer to live in primarily non-Hispanic White neighborhoods, whereas many African Americans prefer to live in neighborhoods that are more integrated. The threshold concept predicts that after a community has reached a certain proportion of minorities, its racial composition will change rapidly until it is entirely African American (Sutker, Gilman & Plax, 1974). Farley et al. (1978; see also Schelling, 1978) concluded that the neighborhood tipping point in 1978 was 70 percent non-Hispanic White and 30 percent African American,
27
implying that neighborhoods that had a greater proportion than 70 percent non-Hispanic Whites would remain stably integrated. Goering (1978), however, questioned whether there is any evidence that demonstrates the existence of a racial tipping point. He argues that neighborhoods are too diverse from a historical, demographic, and social standpoint to justify an iron law of demographic transition. Recent works argue that the change in the proportion of African American residents in a neighborhood has a negative effect on property values (Coate & Vanderhoff, 1993; Kim, 2001, 2003; MacPherson & Sirmans, 2001). Despite the numerous studies that have discussed the effect of race on property values, and despite the fact that regressions hold everything constant, it is still somewhat unclear what race stands for and whether it is a single effect (Boichel et al., 1969; Holifield, Holloway & Morrow-Jones, 2000; Kim, 2003; B. Mikelbank, personal communication, September 9, 2003). Nevertheless, the Proportion of Non-Hispanic Whites was chosen for the analysis, as discussed below. In summary, suburban change in terms of property values is influenced by housing-related factors as well as socioeconomic and geographic factors. After having discussed the theoretical framework of this dissertation, the next chapter will define mature suburbs. Chapter 4 will discuss public policies and Chapter 5 will show how property values of single family homes behaved in mature suburbs in selected study communities, along with exploring what specific factors influence property values of single family homes in mature suburbs in a more empirical fashion.
28
CHAPTER 3
DEFINITION, DATA, AND METHODS
This dissertation focuses on the central counties of the three largest metropolitan areas in Ohio: Cuyahoga (Cleveland), Franklin (Columbus), and Hamilton (Cincinnati). The cities are distinct enough to provide a range of possible situations. Cleveland, an industry-based city, was founded in 1796, 2 while Cincinnati, home to many big corporations, was founded in 1778 (as Losantiville). 3 Both are completely hemmed in by their suburbs. Columbus, founded in 1812, 4 is not landlocked because it has annexed actively (Bier & Howe, 1998). Columbus’s employment is oriented toward government, education, medicine, and financial services, as well as manufacturing. The population figures (Cuyahoga County: 1,393,978; Franklin County: 1,068,978; Hamilton County: 845,303—all based on the 2000 U.S. Census) provide some range of city sizes to consider. The mature suburbs of the three counties range from very troubled (e.g., East Cleveland in Cuyahoga County (Margulis, 2002)) to quite well off and apparently healthy (e.g., Upper Arlington in Franklin County (Morrow-Jones, 1998)).
2
The Encyclopedia of Cleveland History. (n.d.). Retrieved April 23, 2006 from http://ech.cwru.edu/timeline.html 3 City of Cincinnati, Ohio, Fire Department (n.d.). Retrieved April 16, 2006 from http://www.cincinnati-oh.gov/cityfire/pages/-6674-/ 4 City of Columbus, Ohio, Columbus Planning Division (n.d.). Retrieved April 23, 2006 from http://www.columbusinfobase.org/areas/cityof.asp
29
All counties used in this study are in the same state, so the same state-level policies and economic conditions apply. The existence of a statewide First Suburbs Consortium, which will be discussed in more detail in Chapter 5, indicates that these communities are on the leading edge of a trend that acknowledges and addresses issues in their mature suburbs that the rest of the nation is only beginning to notice. This dissertation will employ both quantitative and qualitative data. Hammersley (1996) pointed out that researchers can combine quantitative methods and qualitative methods in three ways: first, using one to verify the findings of the other (i.e., triangulation); second, using one as the groundwork for the other (i.e., facilitation); and third, to explore different aspects of the same research questions (i.e., complementarity). This study uses qualitative methods as facilitation for quantitative methods, and also as triangulation in order to explore different aspects of the same research questions. The exact methods employed depend on the data, as will be discussed below.
3.1 Definition of Mature Suburbs The first research question, as stated in Chapter 1, asks how mature suburbs can be defined. Up until now, definitions of mature suburbs have been scarce and inconsistent. Assuming Burgess’s (1925) model, mature suburbs would surround a central city in a concentric ring. Applying Hoyt’s (1939) model, mature suburbs would surround a central city in selected sectors. There is no generally accepted definition of what constitutes a mature suburb in the literature, although suburban decline has been discussed. Fernandez, Pincus, and Peterson (1982) offered the following indicators of a “suburb in trouble”: (1) deteriorating physical plant and low or deteriorating quantity and
30
quality of public services; (2) persistent inability to finance the municipal budget except by borrowing, or existence of tax rates that drive out residents and businesses; (3) rapid demographic change, which may lead to such social problems as discrimination, higher crime rates, poverty, and flight from the suburbs; and (4) relative economic decline or low-level stability, measured by level of per-capita income and change in per-capita income and by decline of local manufacturing services and trade. Unfortunately, many of these indicators are difficult to measure and difficult to obtain, especially in a consistent fashion for several communities over time. Bier (1991) suggested that declining suburbs are characterized by low-priced homes, low incomes, and their adjacency to their central city. In a later study Bier (2001) pointed out that many of these suburbs have small populations, some deteriorated structures, a lack of resources beyond their own tax base, and a lack of political power. “In comparison with their central city, they [the suburbs] may be virtually impotent in competition for fiscal resources” (Bier, 2001, p. 11). More specifically, Bier defined Cleveland’s first suburbs as follows: (1) they are fully developed—they have little or no vacant land for new construction and therefore cannot expand their tax base through construction; (2) they have no population growth—population may be declining; (3) the average age of real estate is at least 50 years; (4) they have relatively dense infrastructure (e.g., streets, sewers, water pipes, etc.) and the infrastructure is old enough to require extensive maintenance 5 ; and (5) they have lower income growth and property value growth than newer outer suburbs (T. Bier, personal communication, May 28, 2003). However, attempts to apply Bier’s definition to Franklin County (H. Morrow-Jones,
5
See also Haughwout (1997) who suggests the presence of a public sewer hookup as a dummy for older, inner-ring suburbs.
31
personal communication, May 13, 2003) indicated that the definition did not necessarily fit other local contexts. Similar efforts were conducted for Hamilton County (R. Miller, personal communication, September 23, 2003) with similar results. Lucy and Phillips (1995, 1997, 2000, 2001) examined suburban decline in 554 suburbs in the 24 most populous urbanized areas as of 1960. They pointed out that decline has many dimensions, although they argue that the decline in suburban family or household income relative to metropolitan levels of the same measures is a good indicator. Among other potential aspects of decline, Lucy and Phillips added that race is an indicator of decline. African Americans as a group have a lower income than nonHispanic Whites. Thus, increases in the proportion of African Americans in the suburbs would be expected to be accompanied by relative income declines. Some potential indicators suggested for future research are the availability of funds for reinvestment, the number of newly constructed units (minus the number of units demolished and converted), housing affordability expressed as the median value of housing divided by median income, crime indicators that factor in both police reports and beliefs about safety, and school indicators based on the performance or the characteristics of students and/or free lunch eligibility. Orfield (2002) argued that many suburbs are beginning to experience rapid social changes, but that they lack the local resources to deal with them. He suggested several measures of municipal fiscal characteristics (e.g., revenue capacity and expenditure needs) and several measures of sociopolitical environments. He conducted a cluster analysis of 4,606 incorporated municipalities and 135 unincorporated areas in 25 metropolitan regions. The two tax-capacity measures are 1998 tax capacity per household
32
and growth in tax capacity from 1993 to 1998. The four characteristics of the sociopolitical environments—cost measures— are as follows: proportion of elementary students eligible for a free- or reduced-price lunch program, population density, population growth, and age of the housing stock. All of the approaches discussed above have merit, but none incorporate the experience of officials in suburban communities. Thus, the first task of this research was to gain access to local, non-codified knowledge through expert interviews (Ary, Jacobs & Razavieh, 2002; Judd, Smith & Kidder, 1991). First, approval from OSU’s Institutional Review Board (IRB) was sought. The following questions served as a starting point for the expert interviews, although typically discussions ranged freely from this base. •
How do you think property values of single family homes in [specific mature suburb] have behaved over the past decades?
•
What factors do you believe have influenced property values of single family homes in [specific mature suburb]?
•
Can you think of any variables specific to the housing units that would increase or decrease property values of single family homes in [specific mature suburb] significantly?
•
Can you think of any variables on the neighborhood level that would increase or decrease property values of single family homes in [specific mature suburb] significantly?
•
Can you think of any variables linked to the public sector that would increase or decrease property values of single family homes in [specific mature suburb] significantly? 33
•
What can public policy do in order to influence the behavior of property values of single family homes?
Next, recruitment letters were sent out to experts based on a database provided by Kimberly Gibson of the Mid Ohio Regional Planning Commission and the Central Ohio First Suburbs Consortium (FSC). In order to broaden the interviews beyond those involved in the First Suburbs Consortium, an Internet search was used both to update the information provided and to identify additional communities and respondents. In the case of non-academic institutions, letters were sent to the heads of each organization, as they were assumed to have a heightened awareness of issues related to mature suburbs. It was requested that these high level officials assign specialists as interviewees. In the case of research institutions in the three areas, letters were sent to selected faculty members who had special expertise in their areas. As can be seen in Tables A.4, A.5, and A.6 in the Appendix, thirty-two recruitment letters were sent out to potential interviewees in the Cleveland area (including three letters to people who worked at non-academic institutions), twelve were sent out to specialists in the Columbus area, and 42 were sent out to experts in the Cincinnati area (including one duplicate letter to the City of Cincinnati). In the Cleveland area 19 interviews were conducted (response rate: 65.52% (19/29) or 59.38% (19/32)); in the Columbus area nine interviews were conducted (response rate: 75% (9/12)); and in the Cincinnati area 22 interviews were conducted (response rate: 53.66% (22/41) or 52.38% (22/42)). The low response rates might be attributed to several factors. First, potential interviewees may have been available because of busy schedules or difficulties getting in 34
touch. In the case of elected officials, local elections were scheduled for November 2003 so many potential interviewees were busy campaigning several weeks prior to the elections. Also, the majority of interviews were conducted in September 2003, which was the end of summer when some potential interviewees were out-of-town for extended periods of time. Furthermore, one community was at the brink of bankruptcy so their financial issues were understandably prioritized. Cincinnati’s low response rate might be attributed to its spatial structure, with numerous small villages that employ part-time mayors and part-time staff (L. Haaf, personal communication, September 18, 2003; J. Hammon, personal communication, September 22, 2003; J. Looye, personal communication, September 24, 2003). A second issue might be lack of awareness of issues suburbs face. Recruitment letters contained the title of this study (mature suburbs), so potential interviewees might have been under the impression that their suburb was not mature or that their suburb did not have issues that were worth mentioning. A third factor might have been doubts about being the right person or having the right competence. Some interviewees expressed doubts as to whether they were competent enough to be interviewed, and these doubts could not be removed during follow-up recruitment phone calls. Fourth, some people were unwilling to sign paperwork that was required by the IRB before interviews. There were fewer formal interviews in the Columbus area because of the author’s personal knowledge of the area and because several informal interviews were undertaken. As Ary, Jacobs, and Razavieh (2002) point out for survey research, nonresponse is a serious problem that should not be ignored if the study is to have validity. Respondents tend to differ from nonrespondents in characteristics such as education,
35
motivation, and interest in the topic. In order to increase the number of interviewees, several follow-up phone calls were placed for the time frame of the visit. Increasing the number of interviewees beyond the time frame might have confirmed present results or possibly revealed that current suburban issues are not present in other communities. The interpretation of the qualitative data will take into account the potential sample bias and other possible issues. Of those who responded (21 interviewees total in the Cleveland area; nine interviewees in the Columbus area, and 24 interviewees in the Cincinnati area), the majority was male: 80.95 percent in the Cleveland area, 66.66 percent in the Columbus area, and 70.83 percent in the Cincinnati area. Also, the vast majority was assumed to be non-Hispanic White. Only one interviewee was perceived to be African American (Village of Woodlawn). Another interviewee was assumed to be non-Hispanic White although he worked for a village that had an African American majority (Village of Lincoln Heights). It can be assumed that interviewees know best about those issues related to their gender and race. Having a high proportion of male non-Hispanic White respondents implies a systematic bias in terms of responses which could be rectified by increasing the sample. However, since many office holders are male non-Hispanic White, increasing sample size would not necessarily correct this bias. Tables A.4, A.5, and A.6 in the Appendix also show the affiliations of the interviewees, which illustrates the geopolitical municipal structure of the central counties. Cuyahoga County is characterized by many mid-sized cities and only two townships at the fringe of the county, whereas Hamilton County has many small-sized cities, villages,
36
and townships, some of which are suburban in character, sprinkled through the entire county (Howe et al., 1998; J. Looye, personal communication, September 24, 2003). Franklin County, on the other hand, is characterized by Columbus’s annexation activities in the 1950s and 1960s, which nevertheless left numerous villages and townships untouched (Howe et al., 1998; see also Stahura & Marshall, 1982). These expert interviews identified commonalities among mature suburbs and helped to create the definition that will be discussed later in this chapter. Other interview results will be discussed in Section 5.2. The following issues were identified by multiple respondents as major aspects of mature suburbs. A mature suburb that shares a boundary with the adjacent central city of the area can experience long-term residential succession as central city residents spill across its boundary (Z. Austrian, personal communication, October 10, 2003; D. Brooks, personal communication, September 18, 2003; D. Feinstein, personal communication, October 10, 2003). Residential succession in inner cities and adjacent areas often means that the socioeconomic status of the incoming population is lower than that of the population in place (Logan & Semyonov, 1980; S. Neal, personal communication, September 17, 2003; L. Ochsendorf, personal communication, November 26, 2003). Some in-movers might have difficulties maintaining aging housing units, which translates into lowered property values and reduced property tax income for the municipality. At the same time, some inmovers might have a higher need for social services—including community policing— than residents with a higher status, which places a burden on the municipality (D. Beach, personal communication, September 9, 2003; R. Bittner, personal communication,
37
September 18, 2003; D. Brooks, personal communication, September 18, 2003; Hoyt, 1939; Mueller, 1981; Orfield, 2002). A mature suburb that does not share a boundary with unincorporated areas and has a high proportion of residential land use yet low proportion of vacant land as well as an old building stock will likely face tax income issues in the future (V. Barney, personal communication, November 3, 2003; R. Bittner, personal communication, September 18, 2003; D. Brooks, personal communication, September 18, 2003; W. Creager, personal communication, September 17, 2003; R. Corrigan, personal communication, September 9, 2003; P. Feldmann, personal communication, November 3, 2003; Howe et al., 1998; J. Kocevar, personal communication, September 11, 2003; J. Magill, personal communication, November 6, 2003; K. Montlack, personal communication, September 11, 2003; S. Neal, personal communication, September 17, 2003; D. O’Leary, personal communication, September 24, 2003; D. Savage, personal communication, September 18, 2003; B. Siegel, personal communication, September 16, 2003). There will be property tax revenue issues compared with mature suburbs that are able to annex, because the property tax is more difficult to increase. In most cases, annexed areas lie further out, so the annexed building stock is younger. This can translate into higher tax revenues These issues identified by the interviewees have led to the definition of a mature suburb used in this research. A mature suburb has at least two of the following three characteristics: •
it shares a boundary with the adjacent central city of the area;
•
it does not share a boundary with an adjacent unincorporated area (called a township in Ohio); and/or 38
•
it falls in the lower half of all municipalities in each county ranked by proportion of vacant residential parcels. In other words, it has little land available for development. 6
Municipalities that have already spilled over into adjacent counties are excluded from this definition. Municipalities that could potentially annex unincorporated areas in the adjacent counties—assuming that they have land available for development—are also excluded. Tables A.7, A.8, and A.9 in the Appendix show the categories into which each municipality in the three study counties falls. In sum, the qualitative analysis has provided an answer to the first research question: How can mature suburbs be defined? This analysis forms the background for the quantitative analysis.
6
Age of the suburb was not used as a characteristic of a mature suburb because many suburbs grow in leaps and bounds. Thus, using median age might not be the best measure. Obtaining data related to the age of the suburb was also a serious difficulty.
39
3.2 Quantitative Data The primary data source for the quantitative analysis in this research is the PaceNet database of sales transactions purchased from First American Real Estate Solutions (FARES). This database contains all property transactions between 1985 and 2000 in Cuyahoga, Franklin, and Hamilton Counties. Sales transactions that had either a missing price or an entry of 0 were eliminated, as were sales transactions with no transaction date. In addition, all transactions that involved properties that sold before the end of construction (example: sale date 1990, year housing unit built 1991) were eliminated because unfinished homes might sell for a discount. A few cases had a discrepancy between sale date and year housing unit built from one year up to ten years, and these cases were also dropped. Only single family residential land use transactions were used in this analysis. Single family homes are the basis of the suburbs, and so they are the primary concern of this research. The property values of other properties may reflect different factors than property values of single family residential units. To keep the analysis clear and focused on the key suburban land use, this dissertation examines only single family unit transactions, most of which are assumed to be owner occupied. Nevertheless, not considering multi-unit structures means the results only apply to single family homes. The proportion of owner-occupied units is 0.49 in Cleveland and Columbus and 0.39 in Cincinnati, whereas the proportion of owner-occupied units is almost always higher than 0.70 in the mature and developing suburbs in the three counties (the exception is 0.68 in mature suburbs of Hamilton County). As can be seen in Tables A.1, A.2, and A.3 in the Appendix, the proportion of owner-occupied units is quite high in
40
these communities, with most of the single family homes owner-occupied. Thus, the focus is on owner occupied single family homes. All properties that are located on more than five acres were eliminated. In this database the variable sale price is not broken down into land price and house price. Whenever properties are located on large lots the land price component will be a larger part of the sale price, so the lot size threshold of five acres was selected to keep properties comparable. Since the three study communities are located in urban areas, the vast majority of properties is located on smaller lots. Properties that had a sale price of more than $2 million were eliminated in order to prevent problems with large residuals that influence results. 7 Ideally, only arms’ length transactions should be kept in the database in order to eliminate bias. This could be tested by comparing sale prices to assessed values to see if any prices were far too low (i.e., sold to a family member for a low price). Unfortunately, these arms’ length issues could not be addressed because assessed value was not provided by FARES, nor are values consistently reassessed by the county auditors in a very short time frame. The database contains only properties that sold at some point in time between 1985 and 2000. In other words, it does not contain all properties in the three study communities. This invites, as Haurin and Hendershott (1991) point out, the possibility of non-randomness. For example, properties in the lower price tier might sell more frequently than properties in middle or higher tiers (or vice versa). Thus, some price
7
Pindyck & Rubinfeld (1998). In the Cuyahoga County subset 18 outliers, in the Franklin County subset 20 outliers, and in the Hamilton County subset 61 outliers over $2,000,000 were eliminated. The high number of outliers in Hamilton County can partly be attributed to properties located in the Village of Indian Hill.
41
classes of properties might be disproportionately represented in the database. 8 Furthermore, properties in appreciating neighborhoods might sell more frequently because home sellers might use the profits for down payments on the next properties on their way up the housing ladder. Those selling in declining neighborhoods might be overrepresented as people flee. The database has information on the exact address and the physical characteristics of each unit. Some variables are nominal (e.g., building style), and thus had to be respecified as dummy variables. Dummy variables for the sale season (winter, spring, summer, fall) were also created. Most transactions occurred during the summer, so the summer dummy variable was used as the omitted one in order to prevent perfect multicollinearity (Hamilton, 1992). The third research question introduced in Chapter 1 asks how property values of single family homes have behaved in mature suburbs, and the fourth research question asks what specific factors influence the property values of single family homes in mature suburbs. Table 1 below provides a list of variables used to answer these research questions.
8
About 5 percent of a community’s homes sell in one year. Since this study encompasses 15 years of data the data set probably represents about 75 percent of the housing stock in a community (T. Bier, personal communication, April 29, 2006).
42
Raw Variable Sale Amount in $ at Year of Sale Sale Date Sale Date Census Tract (basis: 1990 boundaries) Census Tract (basis: 1990 boundaries)
Effective Property Tax Rate in Tax District at Year of Sale 9 Number of Murders per 100,000 Population in 2000 Pass Rate on the State 6th Grade Mathematics Test in School District in 2000 Proportion of Non-Hispanic White Household Heads in Census Tract 1980, 1990, 2000
Transformed Variable Sale Amount in $ at Year of Sale Dummy Sale Season Sale Year Distance from Geographic Centroid of Census Tract to Downtown (miles) Distance from Geographic Centroid of Census Tract to Nearest Interstate/State Route Access Point (miles) Effective Property Tax Rate in Tax District at Year of Sale Number of Murders per 100,000 Population in 2000 Pass Rate on the State 6th Grade Mathematics Test in School District in 2000 Proportion of Non-Hispanic White Household Heads in Census Tract at Year of Sale
Source PaceNet PaceNet PaceNet calculation based on PaceNet calculation based on PaceNet
county auditors (Cuyahoga, Franklin, Hamilton County) State of the Cities Data Systems (SOCDS) 10 Ohio Department of Education 11 2000 U.S. Census
Table 1: Variables Used in Quantitative Analysis
9
The effective property tax rate provided by auditors of Cuyahoga, Franklin, and Hamilton Counties. The auditors calculate the effective tax rate by using the taxes of the prior year and the taxable value in the current year for property taxed in the prior year. Dividing the taxes by the value (and multiplying by 100 to convert to a rate per $100 of value) provides the effective tax rate. Carole Keeton Strayhorn. Texas Comptroller of Public Accounts (n.d.). A Guide for Setting Tax Rates: Truth-In Taxation. Available from Window on State Government (Texas) website, http://www.window.state.tx.us/taxinfo/proptax/tnt00/03tnt00.html 10 U.S. Department of Housing and Urban Development, HUDUser. (2003). State of the Cities Data Systems (SOCDS), 2003 [Data file]. Available from SOCDS Web site, http://socds.huduser.org/FBI/violent_crime.odb. Several data sets were analyzed but the SOCDS data set had the least amount of missing data. 11 Ohio Department of Education. (2003). Local Report Card District Ratings, 2002 (2000-2001 School Year) [Data file]. Available from ODE Web site, http://www.ode.state.oh.us/reportcard/ratings.asp
43
Table 1 (continued)
Average Household Income in Census Tract 1979, 1989, 1999 Topography Lot Size Square Footage of Unit at Year of Sale Number of Bathrooms of Unit at Year of Sale Year of Construction of Unit
Average Household Income in Census Tract at Year of Sale Dummy Rolling Topography Lot Size Square Footage of Unit at Year of Sale Number of Full Bathrooms of Unit at Year of Sale Age of Unit in Years at Year of Sale
2000 U.S. Census
PaceNet PaceNet PaceNet PaceNet PaceNet
In addition to the information provided in the PaceNet database, four housing factors and two socioeconomic factors related to suburban change were added based on the literature reviewed in Chapter 2 (see also Andersson, 2000). With respect to housing factors, accessibility, tax rate, public safety, public school quality, and selected housing characteristics were chosen. This study uses two accessibility variables. First, it uses the Distance from the Geographic Centroid of Each Property’s Census Tract to Downtown to account for the remaining importance of the CBD as a place of employment, retail, and other amenities (see Anas, 1981; T. Bier, personal communication, September 10, 2003; Gin & Sonstelie, 1992; Muth, 1969; and Voith, 1991). Second, it uses the Distance from the Geographic Centroid of each Property’s Census Tract to the Nearest Interstate or State Route Access Point to account for accessibility to all parts of the metropolitan area (see Bruinsma, Rienstra & Rietveld, 1997 and Kockelman 1997a, 1997 b; R. Layton, personal communication, September 10, 2003; B. Mikelbank, personal communication, September 9, 2003).
44
Assuming that amenities and disamenities are capitalized into property values (Mieszkowski & Zodrow, 1989; Oates, 1969, 1973, among others), this study will use the effective property tax rate provided by the auditors of the three counties (as suggested by R. Layton, personal communication, September 10, 2003). As discussed above, the effective tax rate is supposedly unbiased (Chinloy, 1978; see also Ihlanfeldt and Jackson (1982) for an alternative opinion and Case (1978) for an excellent summary). Following Clark and Cosgrove’s argument (1990; see also T. Bier, personal communication, September 10, 2003), the Murder Rate (per 1,000 Population) was included for 2000 at the municipal level. Because of the importance of school quality to property value (see Brasington, 2000, among others), the 6th Grade Mathematics Pass Rate in each school district in 2000 was included (B. Mikelbank, personal communication, September 9, 2003; M. Nicholson, personal communication, November 25, 2003). This test is considered to be particularly difficult because of its content and form. Therefore it is particularly good at differentiating better school districts from worse ones. It is a short answer test, not the typical multiple choice test so the results depend less on chance and more on learned knowledge (M. Nicholson, personal communication, November 25, 2003). Lot and housing characteristics, such as rolling topography, lot size, square footage, number of full bathrooms, and age of the housing unit, were added to the analysis. Age of the housing unit was paid special consideration. Curve fitting exercises were undertaken based on the assumption that the age of the housing unit and property value are related in a nonlinear fashion (Goodman & Thibodeau, 1997). Tests with classified dummy variables (Age of Housing Unit 0 to 10 Years, Age of Housing Unit 11
45
to 20 Years, etc.) showed that in most cases the coefficients describing their relationship with property value (sale price) change from positive (for younger housing units) to negative (for older housing units). Therefore, the (unclassified) variable Age of the Housing Unit was squared (Cannaday & Sunderman, 1986; Clark & Herrin, 2000; Grether & Mieskowski, 1974; Malpezzi, Ozanne & Thibodeau, 1987). With respect to socioeconomic factors, the Proportion of Non-Hispanic White household heads and Average Household Income were chosen (T. Bier, personal communication, September 10, 2003). These variables were obtained from the 1980, 1990, and 2000 U.S. Censuses at the census tract level after adjusting for boundary changes over time. The Census tract level was used for the quantitative analysis because the PaceNet data set provided transactions accompanied by their respective census tract number. Based on information provided by the censuses, yearly rates and values were calculated by building weighted averages. For example, for rates (values) for 1981, the 1980 rates were weighted by .9 and the 1990 rates were weighted by .1, and then the two were averaged. The extensive discussion of racial composition and its impact on house prices in the literature suggests that this variable is likely to contribute to the factors that influence property values. However, it is unclear what race stands for—it might, for example, indicate perceptions about African Americans living in the same neighborhood or indicate socioeconomic phenomena not caught by other direct socioeconomic variables such as income. Nevertheless, the variable Proportion of non-Hispanic Whites in Census Tract is used for the quantitative analysis.
46
Average Household Income was used as a proxy for demand as well as an indicator of preference to live in neighborhoods characterized by high household incomes and low crime rates. The higher the average household income, the greater is the demand for properties with higher property values. More specifically, Average Household Income per Census Tract was taken, although the median should be preferred over the average because the latter is prone to be influenced by outliers. Because of the longitudinal nature of this dissertation and the fact that census tract boundaries change over time, as discussed below, the Average Household Income per Census Tract had to be used. In sum, the following variables were selected for the quantitative analysis: Dependent variable: •
Sale Amount in Dollars (natural log)—The sale amount serves as a proxy for the value of the house. The semi-log specification has advantages over the linear form because it helps to normalize the distribution of the error term, a desirable characteristic for ordinary least squares estimators (Kennedy, 2003).
Independent variables: Time component: •
Dummy Sale Season—The coefficient is expected to be neutral for all seasons except for winter quarter due to lower demand for homes during that season (see for example Goodman and Thibodeau (1997), among others).
47
•
Dummy Sale Year—It is hypothesized that the coefficient is positive because homes appreciate nominally over time (see for example Bogart and Cromwell (1997, 2000), among others).
Space component: •
Distance from Geographic Centroid of Census Tract to Downtown (miles)—Since neoclassical economic urban theory (Muth, 1969, among others) suggests that land prices should decrease outward from the center in order to compensate for costs of commuting, the sign is expected to be negative. On the one hand, downtowns still have some employment and amenities. On the other hand, there are several major job centers around the outerbelt in each metropolitan area. In addition, there is no evidence that structure prices necessarily decline with distance from the CBD, all other things being equal, only land prices. Some ambiguity is expected when it comes to the sign of this variable (see Kim & Morrow-Jones, 2005; see also Rachlis & Yezer, 1985).
•
Distance from Geographic Centroid of Census Tract to Nearest Interstate/State Route Access Point (miles)—Neoclassical economic urban theory (Muth, 1969, among others) suggests that commuting costs and land prices interact, so commuting costs should also have an impact on house prices. However, it might be the case that commuting time does not have a major impact on house prices because fuel prices have decreased and the efficiencies of many cars have increased in the long run. Since few studies have utilized this variable (see
48
Bruinsma, Rienstra & Rietveld, 1997 for related aspects), some ambiguity is expected. Tax district level: •
Effective Property Tax Rate in Tax District at Year of Sale—Consistent with the majority of works in the capitalization literature, such as Mieszkowski and Zodrow (1989)), the sign is expected to be negative—the higher the effective property tax rate is in a tax district, the lower the sales amount, as long as everything else is held constant.
Municipal level: •
Crime Rate in Municipality in 2000—Consistent with works in public safety (Clark & Cosgrove, 1990), murder per capita as a representative type of violent crime is expected to have a negative sign (the higher the murder per capita rate the lower the sales amount, everything else being equal). 12
School district level: •
Pass Rate on the State 6th Grade Mathematics Test in School District in 2000 13 — The coefficient is expected to be positive, as suggested by Jud and Watts (1981) (the higher the math test pass rate the higher the sales amount, all other things being equal). 14
12
Since longitudinal data could not be located endogeneity is suspected to be present. Ohio Department of Education. (2003). Local Report Card District Ratings, 2002 (2000-2001 School Year) [Data file]. Available from ODE Web site, http://www.ode.state.oh.us/reportcard/ratings.asp 14 Since longitudinal data could not be located endogeneity is suspected to be present. 13
49
Census tract level: •
Proportion of Non-Hispanic White Household Heads in Census Tract at Year of Sale 15 —Consistent with the Chicago School literature and Hoyt (1939), the coefficient is expected to be neutral or positive (the higher the proportion of nonHispanic Whites in a Census tract the higher the sales amount, all other things being equal) 16 .
•
Average Nominal Household Income in Census Tract at Year of Sale 17 — Consistent with Fortura and Kushner (1986), the coefficient is expected to be positive (the higher the median household income in a Census tract the higher the sales amount, all other things being equal) 18 .
Housing unit level: •
Dummy Rolling Topography—The coefficient is expected to be positive since homeowners prefer a rolling topography that adds to the aesthetics of the setting (Appraisal Institute, 2001).
•
Lot Size 19 —The coefficient is expected to be positive (the larger the lot size the higher the sales amount, if everything is held constant) (Appraisal Institute, 2001; Mills, 1972).
15
Variable multiplied by 100 for the sake of simplicity and convenience of interpretation. Variable Average Change in the Proportion of Non-Hispanic White Household Heads in Census Tract First Five Years after Year of Sale was created and integrated into the model. Whereas this variable did not cause any issues in most models, it caused the model to become full rank in the case of the Cincinnati model. Thus, this variable was not integrated into the model. 17 Variable divided by 1,000 for the sake of simplicity and convenience of interpretation. 18 Variable Average Change in the Average Nominal Household Income in Census Tract First Five Years after Year of Sale was created and integrated into the model. Unfortunately, including this variable caused the model to become full rank. Thus, this variable was not integrated into the model. 19 Variable divided by 1,000 for the sake of simplicity and convenience of interpretation. 16
50
•
Square Footage of Unit at Year of Sale 20 —The coefficient is expected to be positive (the higher the square footage of the unit the higher the sales amount, all other things being equal) (Mills, 1972).
•
Number of Full Bathrooms of Unit at Year of Sale—The coefficient is expected to be positive (the higher the number of full bathrooms the higher the sales amount, all other things being equal) (Mills, 1972).
•
Age of Unit in Years at Year of Sale (Squared)—The coefficient is expected to be negative for younger housing units (need for maintenance/investments) but positive for older housing units (vintage effect outweighing the need for maintenance/investments), all other things being equal (see Goodman & Thibodeau, 1997; Haughwout, 1997).
3.3 Methods This dissertation uses the hedonic approach, based on neoclassical economic models of urban structure. The hedonic approach treats the house as a bundle of characteristics that cannot be repackaged and that is sold for a single price (Epple, 1987; King, 1976; Lancaster, 1966; Rosen, 1974; Witte, Sumka & Erekson, 1979). According to Rosen, “Hedonic prices are defined as the implicit prices of attributes and are revealed to economic agents from observed prices of differentiated products and the specific amounts of characteristics associated with them” (Rosen, 1974, p. 34). Thus, property values can be disaggregated into the price paid for different components of the housing bundle with the help of regression models.
20
Variable divided by 1,000 for the sake of simplicity and convenience of interpretation.
51
The functional form chosen for the model was the semi-log form, which is often chosen when the dependent variable has a wide spread (I. Maric, personal communication, October 19, 2003) and when several independent variables have a concave function (Cannaday & Sunderman, 1986). Tests for multicollinearity and heteroskedasticity were performed. Multicollinearity (see Tables A.10, A.11, A.12 and A.13 in the Appendix) arises when two or more variables are highly correlated with each other, causing standard errors for the regression coefficient to be very high (Pindyck & Rubinfeld, 1998). Results of the test for multicollinearity show that no pair of variables has a correlation higher than 0.75, the threshold chosen for this analysis. 21 Heteroskedasticy leads to inefficiency and biased standard error estimates, making the usual standard errors, tests, and confidence intervals untrustworthy (Hamilton, 1992). To test for non-constant variance, Breusch-Pagan and White’s tests were conducted (see Table A.14 in the Appendix). Results show that there is heteroskedasticity. In this case, ordinary least-squares estimation places more weight on the observations with large error variances than on those with smaller error variances. The reason for this weighting is that the sum-of-squared residuals associated with large variance error terms are likely to be greater than the sum-of-squared residuals associated with low variance errors. This implicit weighting causes the ordinary least-squares parameter estimators to be inefficient (i.e., the variances of the estimated parameters are
21
Strong multicollinearity, such as a multicollinearity of 0.75, is common and still permits estimation, although it makes it less precise (Hamilton, 1992). The third research question asks about factors that influence property values of single family homes in mature suburbs. It is assumed that there are multiple factors that influence property values. Thus, multicollinearity is tolerated in this study.
52
not the minimum variances and are also biased, although the parameters are still consistent) (Pindyck and Rubinfeld, 1998). In order to address heteroskedasticity, a weighted least squares estimation was undertaken that places less weight on the observations with large error variances than on those with smaller error variances. Since the variance and the weight are not known, they must be estimated. Nielsen (2002) suggests several steps to calculate the weights. First, estimate an OLS regression of the dependent variable on the independent variables to obtain the residuals; second, estimate a standard deviation function based on visual evidence gained from a residual plot by regressing the residuals on the dependent variable; and third, calculate the weights by inversing a fitted value (estimate) of the variance based on a regression with the variance as dependent variable. This procedure yields efficient parameter estimators which satisfy all the assumptions of the classical linear regression model. Results will be discussed in Chapter 5. Many authors argue that there are housing submarkets (see Bourassa, Hoesli & Peng (2003) for a good overview; see also Adams, 1991; Goodman & Thibodeau, 1998) that could affect the results. In order to test whether the groupings used in this study define different submarkets, Chow tests were conducted. 22
F=
(ess _ c − (ess _ 1 + ess _ 2 + ess _ 3) / k (ess _ 1 + ess _ 2 + ess _ 3) /( N _ 1 + N _ 2 + N _ 3) − 2 * k )
(1)
Formula (2) below tests whether there are three different regional submarkets, such as Cuyahoga County versus Franklin County versus Franklin County.
22
Adapted by author from Gould (2005).
53
FThreeCounties =
(58,040 − 56,148) / 30 = 315.3782621 56,148 /(280,840 − 60)
(2)
Formulas (3), (4), and (5) test whether there are submarkets within each county, such as Cleveland, Cuyahoga County’s mature suburbs, and Cuyahoga County’s developing suburbs. (32,087 − 29,182.59087) / 30 = 261.6274045 29,182.59087 /(157,785 − 60)
(3)
FFranklinCounty =
(11,475 − 14,062.13565) / 30 = −389.2072729 14,062.13565 /(63,525 − 60)
(4)
FHamiltonCounty =
(12,586 − 11,158.87796) / 30 = 76.57271085 11,158.87796 /(18,022 − 60)
(5)
FCuyahogaCounty =
Fcritical (30,60) = 1.65 23 where
(6)
F = test statistic ess_c = error sum of squares from the pooled (constrained) regression ess_x = error sum of squares from the separate regression k = number of estimated parameters N_x = number of observations in a group
Results in equations (2) through (5) show that it is incorrect to assume equal coefficients in the cases of Cuyahoga and Hamilton Counties, although it is correct to assume equal coefficients in the case of Franklin County. Regressions were run based on the definition given above. In order to be able to compare the three counties, analyses for Franklin County were pursued despite the results gained in the Chow test. Future research will address this issue. 23
Pindyck & Rubinfeld (1998).
54
The nine submarkets selected for analysis are as follows: •
Cuyahoga County Central City [Cleveland]
•
Cuyahoga County Mature Suburbs
•
Cuyahoga County Developing Suburbs
•
Franklin County Central City [Columbus]
•
Franklin County Mature Suburbs
•
Franklin County Developing Suburbs
•
Hamilton County Central City [Cincinnati]
•
Hamilton County Mature Suburbs
•
Hamilton County Developing Suburbs
These submarkets are the basis of the discussion of the results in Chapter 5. Chapter 4, the next chapter, will discuss public policies that benefit homeowners in mature suburbs. 24
24
Spatial autocorrelation will be addressed in future research efforts. See Can (1990) and Can & Megbolugbe (1997).
55
CHAPTER 4
PUBLIC POLICIES
4.1 Current Policies The second research question asks what public policies benefit homeowners in mature suburbs. Puentes and Orfield (2002) point out that “[…] first suburbs are caught in a policy blindspot. Unlike central cities, they are not poor enough to qualify for many federal and state reinvestment programs and not large enough to receive federal and state funds directly” (Puentes & Orfield, 2002, p. 3; K. Montlack, personal communication, September 11, 2003; see also Persky & Kurban, 2001). However, the expert interviews and the literature revealed a range of existing and proposed policies that benefit mature suburbs and their residents, especially homeowners. Public policies for home owner investments are of special interest since housing units are fixed in space. Even when homeowners move out of the community the investment value stays (Joint Center for Housing Studies of Harvard University, 2003). The following paragraphs will differentiate between policies funded by the federal government, the State of Ohio, and select counties, municipalities, and non-profit organizations. None of these programs are aimed primarily at mature suburbs, but all could be used to benefit them.
56
4.1.1 Federal Policies One example of a federally funded program that currently benefits some mature suburbs is the Community Development Block Grant (CDBG) Program, which grew out of President Nixon’s New Federalism policy strategy and was initiated in 1974 (Conlon, 1981; Markusen, Saxenian & Weiss, 1981). CDBG programs are divided into Entitlement Communities and State Administered CDBG. These two programs provide grants on a formula basis to entitled cities and counties to provide decent housing and a suitable living environment, primarily to low and moderate income residents, among other goals. 25 Eligible grantees in the Entitlement Communities program are cities with populations of at least 50,000 and qualified urban counties with populations of at least 200,000 (M. McGinty, personal communication, September 8, 2003). Eligible grantees in the State Administered CDBG are non-entitlement areas that do not qualify for the Entitlement Communities program. As can be seen in Tables A.15, A.16, and A.17 in the Appendix, Cleveland Heights (population 50,769 26 ), Euclid (population 52,717), Lakewood (population 56,646) and Parma (85,655) are the only mature suburbs in this study that are Entitlement Communities. As an example, Parma provides for grants of up to 50 percent of housing repair costs up to a maximum of $2,000 subject to income limits (M. McGinty, personal communication, September 8, 2003).
25
U.S. Department of Housing and Urban Development. (n.d.). Index of Community Development Programs. Retrieved February 21, 2003, from http://www.hud.gov/offices/cpd/communitydevelopment/programs/index.cfm 26 According to the 2000 U.S. Census, the population of Cleveland Heights was 49,958. However, the city petitioned for a recount when it became obvious that the Census had missed several multi-story buildings. The adjusted number is 50,769. J. Hull, personal communication, April 19, 2006.
57
Another program with potential to help mature suburbs is the Federal HOME program, which provides up to $15,000 per household to residents in a community who are purchasing their first homes ($5,000 for a single family home, $10,000 for a condominium converted from a two-family structure, and $15,000 to purchase a twofamily home). Eligible applicants must have resided within the community for at least one year immediately prior to application and be either first-time home buyers, displaced homemakers, or people who have not owned homes for three years. The program is also subject to federal income limits and to a minimum down payment of three percent of the sale price from personal funds (M. McGinty, personal communication, September 8, 2003). The City of Parma and the City of Cleveland Heights promote this program in a very active fashion on their websites. 27
4.1.2 State, Regional, and County Policies State, regional, and county programs vary more than federal programs, as each regional planning organization or county works to develop ideas for their respective areas. Some current ideas could be used in other regions as well, so these regional entities should try to share their best practices. At the state level, the Community Housing Improvement Program (CHIP) provides grants to eligible communities that undertake infrastructure improvements and housing-related activities, such as the improvement and provision of affordable housing
27
City of Cleveland Heights, What’s New? (n.d.). Retrieved April 23, 2006 from http://www.clevelandheights.com/whatsnew.asp?id=219, City of Parma, City Hall, Community Development (n.d.). Retrieved April 23, 2006 from http://www.cityofparma-oh.gov/cityhall/commdev.htm
58
for low- and moderate-income persons. The grant ceiling is $500,000 and the funding sources are CDBG/HOME and the Ohio Housing Trust Fund. 28 One good example of a regional program is the weatherization and repair program of People Working Cooperatively (PWC), a non-profit organization active in Southwestern Ohio and Northern Kentucky, including the mature suburbs of Hamilton County. 29 PWC provides services for homeowners who are low-income, disabled, or senior citizens. PWC is funded by several governments (State of Ohio, Commonwealth of Kentucky, Boone County, Campbell County, Hamilton County, Kenton County), organizations (United Way; Kentucky Housing Corporation, etc.), and private donors. An example of a program at the county level is Cuyahoga County’s Housing Enhancement Loan Program (HELP), which is funded by Cuyahoga County. 30 HELP allows homeowners in eligible communities to borrow money to alter, repair, or improve a property to protect or improve its basic livability and enhance the property’s value (with the exception of luxury items). The interest rate is three percentage points below banks’ market interest rate for home improvement loans. The minimum loan amount is $1,500, and the maximum is $200,000 (for single family or two family dwellings; buildings with three or more units are subject to the lesser of $200,000 or $20,000 per unit with a maximum amortization of five years). Eligible individuals are owner-occupants, as well as investors who own single- or two-family dwellings with market values of less than $250,000 (there is no market value limit for buildings with three or more units). There is 28
Ohio Department of Development, Housing, Shelter and Supportive Services Programs, Community Housing Improvement Program (CHIP) (n.d.). Retrieved May 13, 2006 from http://www.odod.state.oh.us/cdd/ohcp/hssp.htm 29 People Working Cooperatively, Repair Affair (n.d.). Retrieved March 20, 2006 from http://www.pwchomerepairs.org/ 30 Banks charge the mortgagee an interest rate three percent below the current mortgage interest rate. In turn, Cuyahoga County’s Treasurer buys the mortgages from the banks for an interest rate three percent below the current certificate of deposit interest rate (B. Nimrick, personal communication, May 3, 2006).
59
no maximum income ceiling or restriction based on the age of the property (D. Beach, personal communication, September 9, 2003; M. McGinty, personal communication, September 8, 2003; J. Rokakis, personal communication September 8, 2003). 31 Communities eligible for HELP are those that have had a property appreciation rate less than Cuyahoga County’s median of 2.34 percent over the past 18 years. Table A.18 in the Appendix shows that 33 municipalities in Cuyahoga County are eligible for HELP loans, including Cleveland, 24 out of 29 mature suburbs (82.76 percent), and 8 out of 18 developing suburbs (44.44 percent). However, none of the nine communities that have spilled over to the neighboring counties are eligible. The high proportion of eligible mature suburbs confirms that they have appreciation issues. On the other hand, the relatively high proportion of developing suburbs shows that they already have problems that might warrant an increased focus over the coming years. As of February 8, 2005 the Cuyahoga County Treasurer’s Office had made possible over 5,200 loans totaling more than $63.3 million. A follow-up survey (3,098 questionnaires sent out and 295 completed for a response rate of 9.5 percent) asked recipients of HELP loans whether they would have made an investment were it not for the loan. About 55 percent of the respondents answered that they would not have made an investment and about 78 percent answered that they were more likely to remain in their homes longer than they had anticipated because of their improvements (B. Nimrick, personal communication, April 18, 2006). In a similar vein, Hamilton County administers a Home Improvement Program (HIP). The lower limit of this loan is $1,500 and the upper limit is $50,000 (R. Bittner,
31
Cuyahoga County, Treasurer, Home Improvement Loans, Housing Enhancement Loan Program (n.d.). Retrieved March 24, 2006 from http://www.cuyahogacounty.us/treasurer/homeimprove/HELP.htm
60
personal communication, September 18, 2003). Hamilton County offered participation in HIP to all communities in the county and all municipalities. All places except Arlington Heights, which was classified as a mature suburb, and Milford, which has spilled over into the adjacent county, agreed to participate in the program (see Table A.19 in the Appendix). This includes the City of the Village of Indian Hills, which has a median household income of $158,742 and a median value of $750,000 for all owner-occupied housing units. In comparison, Hamilton County’s median household income is $40,964 with a median owner-occupied housing unit value of $109,000 (2000 U.S. Census). Franklin County does not have an equivalent to the HELP or HIP programs due to the fortunate situation of most of its suburbs. Another example of a county program, administered and partly funded by Cuyahoga County’s Department of Development and matched by cities, is the Operation Home Improvement Exterior Maintenance Grant Program for homeowners who have been cited for code violations and who meet program qualifications and income guidelines. 32 Forty-eight urban communities in Cuyahoga County are eligible to apply for this program.
4.1.3 Local Policies Localities do not always have to wait for state, regional, or county policies for assistance. Mature suburbs in particular may need to develop policies to help themselves. Many cities have programs aimed at improving properties. Some examples at the city level include the following:
32
City of South Euclid, Building Department, Property Maintenance Programs (n.d.). Retrieved March 24, 2006 from http://www.cityofsoutheuclid.com/propertymaintenance.htm
61
•
the Public Services Program of Parma, in which the city offers supplemental grass cutting and snow removal for senior or disabled residents (M. McGinty, personal communication, September 8, 2003); 33
•
the Sidewalk Repair Program of Deer Park, in which the city co-sponsors sidewalk repairs (D. O’Leary, personal communication, September 24, 2003);
•
the Pride of Whitehall, a volunteer program that assists low-income seniors in addressing code violations (L. Ochsendorf, personal communication, November 26, 2003);
•
Home Handyman of Euclid, a city program that provides help for eligible homeowners in projects such as minor code repairs, vinyl siding, smoke alarm or dead bolt lock installation;34
•
Senior Services, a volunteer program in Euclid that provides information on agencies that provide services such as home maintenance and repair services and personal care; 35 and
•
Project Help of the City of Parma, a program which prevents people from losing their homes because of a sudden loss of income; the city pays for home-related expenses such as utilities or the mortgage or rent for two months (M. McGinty, personal communication, September 8, 2003).
33
City of Parma, City Hall, Service Department (n.d.). Retrieved March 20, 2006 from http://www.cityofparma-oh.gov/cityhall/service.htm 34 City of Euclid, Resident Information, Neighborhood Programs (n.d.). Retrieved April 23, 2006 from http://www.ci.euclid.oh.us/residentinformation/neighborhood.cfm 35 City of Euclid, Resident Information, Neighborhood Programs (n.d.). Retrieved April 23, 2006 from http://www.ci.euclid.oh.us/residentinformation/neighborhood.cfm
62
As with regions and counties, it would be helpful for cities to exchange best practices, especially those that are mature suburbs.
4.2 Current Suburban Tools As discussed above, public policies provide resources that are complemented by local suburban tools in order to maintain and improve the housing stock of suburban communities. Several tools were mentioned during expert interviews. These fall into two groups: suburban tools that punish homeowners (“sticks”) and those that encourage them (“carrots”).
4.2.1 “Sticks” Many communities conduct code enforcement every few years for the exterior of their housing units (R. Bittner, personal communication, September 18, 2003; D. Brooks, personal communication, September 18, 2003; W. Creager, personal communication, September 17, 2003; S. Upton Farley, personal communication, September 16, 2003; D. Feinstein, personal communication, October 10, 2003; J. Kocevar, personal communication, September 9, 2003; D. Lorek, personal communication, November 3, 2003; M. McGinty, personal communication, September 8, 2003; L. Ochsendorf, personal communication, November 26, 2003; D. O’Leary, personal communication, September 24, 2003; D. Savage, personal communication, September 18, 2003; B. Siegel, personal communication, September 16, 2003; E. Tollerup, personal communication, September 11, 2003; J. Wright, personal communication, September 17, 2003). 36 A few
36
City of University Heights, Building (n.d.). Retrieved April 23, 2006 from http://www.universityheights.com/building.html
63
communities undertake architectural reviews (D. Lorek, personal communication, November 3, 2003; D. Savage, personal communication, September 18, 2003), while others require an annual occupancy permit for rental properties with exterior and interior inspections to address overcrowding and to ensure proper maintenance of properties. 37 Similarly, point of sale inspections are quite popular (W. Creager, personal communication, September 17, 2003; D. Lorek, personal communication, November 3, 2003; E. Tollerup, personal communication, September 11, 2003). 38 The City of North College Hill takes the point of sale inspection a step further: If the seller is unable to fix the points cited at the point of sale inspection before the sale, then he or she must pay 150 percent of the repair costs into an escrow account before the sale takes place. This repair strategy is especially popular with affluent senior citizens who are physically unable to renovate and/or are mentally unable to coordinate a renovation (R. Corrigan, personal communication, September 9, 2003). One community, North College Hill, is considering taking both code enforcement and point of sale inspections a step further by making a concentrated effort to motivate and guide owners in becoming more active in terms of renovations. At an instructional level, the city would like to offer home improvement classes at the local school, increasing the connection between residents and the facility. At a practical level, they would like to focus their renovation efforts on one block at a time, making a concentrated effort in a small area (“taking back the neighborhoods;” D. Brooks, personal communication, September 18, 2003).
37
City of South Euclid, Building Department, Property Maintenance Programs (n.d.). Retrieved March 24, 2006 from http://www.cityofsoutheuclid.com/propertymaintenance.htm 38 City of University Heights, Building Department (n.d.). Retrieved March 23, 2006 from http://www.universityheights.com/building.html
64
Programs at the individual housing unit level are complemented by programs at the neighborhood level, such as the City of Euclid’s Sidewalk Inspection in which every sidewalk in the community is inspected every five years. 39 When necessary, homeowners can either hire a contractor or let the city do repairs. The city will add the repair costs to property taxes for the next two years (J. Kocevar, personal communication, September 11, 2003).
4.2.2 “Carrots” The First Suburbs Consortium Housing Initiative has created several potential unit designs to address issues that are characteristic of suburban bungalows: small lots; small rooms, kitchens, and bathrooms; small number of rooms, kitchens, and bathrooms; insufficient number of electrical outlets; poor construction quality; aging issues; and monotonous architectural styles (T. Bier, personal communication, September 10, 2003; R. Bittner, personal communication, September 18, 2003; First Suburbs Consortium Housing Initiative, 2002; D. Lorek, personal communication, November 3, 2003; J. Magill, personal communication, November 6, 2003; M. McGinty, personal communication, September 8, 2003; R. Miller, personal communication, September 23, 2003; L. Ochsendorf, personal communication, November 26, 2003; J. Rokakis, personal communication, September 8, 2003; E. Tollerup, personal communication, September 11, 2003; Wirt et al., 1972). The City of Wyoming embraces alternative approaches that encourage homeowners to improve their homes. The City gives annual awards for property
39
City of South Euclid, Building Department, Property Maintenance Programs (n.d.). Retrieved March 24, 2006 from http://www.cityofsoutheuclid.com/propertymaintenance.htm
65
appearance and property improvement along with tree awards, recognizing those trees that make the community special. 40 In addition to these activities, the City encourages homeowners not only to apply to place their homes on the National Register, but also to enter the Prettiest Painted Places in America contest of the Paint Quality Institute. 41 Moreover, the community tries to pave its streets fairly often, which typically encourages homeowners to improve the appearance of their properties (D. Savage, personal communication, September 18, 2003). Similarly, the City of Bedford’s Blue Ribbon Program encourages homeowners to upgrade property values, to improve the appearance of their neighborhoods, to provide greater safety for neighborhood residents, and to increase community spirit. Upon application and approval by Bedford’s City Council, a neighborhood is allowed to use the phrase “A Bedford Blue Ribbon Neighborhood” in promotional materials. 42 The City of Bedford’s Beautification program has the purpose of making the city a clean and attractive place to live, maintaining a healthy environment, educating citizens about the ongoing efforts to keep the city beautiful, and generally beautifying the city. The City’s Beautification Commission gives out awards to businesses in the community for beautification activities as well. 43
40
City of Wyoming, News, Notices, & Events (n.d.). Retrieved March 24, 2006 from http://www.wyoming.oh.us/index.cfm?fuseaction=home.ViewPage&page_id=C4781EA3-6ED5-4EBC8EC2BB0CA808F5EF. 41 The Old House Web, Features, Design (n.d.). Retrieved April 15, 2006 from http://www.oldhouseweb.com/stories/Detailed/10322.shtml and The Rohm and Hass Paint Quality Institute (n.d.). Retrieved April 15, 2006 from http://www.paintquality.com/ 42 City of Bedford, City Council. (n.d.). Blue Ribbon Program. Retrieved April 23, 2006 from http://www.ci.bedford.tx.us/council/assets/Blue%20Ribbon%20Program%20Application.pdf 43 City of Bedford, City Council (n.d.). Retrieved April 23, 2006 from http://www.ci.bedford.tx.us/council/boards/beautification_awards.htm
66
4.2.3 Other Tools Falling between suburban tools that punish and encourage homeowners, some communities, such as Cleveland Heights and Euclid, prevent issues by offering training on home maintenance for first-time homebuyers in their communities (E. Tollerup, personal communication, September 11, 2003).44 Yet another community, North College Hill, was at one point in time thinking about requiring owners of rental properties to have an escrow account at the time of the purchase of the property in order to pay for future repairs that might arise in connection with code enforcement (Daniel Brooks, personal communication, September 18, 2003). However, this idea did not pass City Council (R. Weber, personal communication, March 27, 2006).
Puentes and Orfield (2002) point out that most first suburbs are not poor enough to qualify for many federal and state reinvestment programs and not large enough to receive federal and state funds directly. Although the expert interviews revealed a range of existing and proposed policies that benefit mature suburbs and their residents, and especially their homeowners, it remains to be seen whether communities and residents can take advantage of the policies and ideas. Also, if the communities are able to use some of the policies discussed, will they be enough to address issues in those suburban communities that battle the imbalance between community resources and needs for expenditure in the future?
44
City of Cleveland Heights, What’s New? (n.d.). Retrieved April 23, 2006 from http://www.clevelandheights.com/whatsnew.asp?id=325
67
CHAPTER 5
RESULTS
Chapter 1 was an introduction, Chapter 2 provided a literature review in connection with the variables, Chapter 3 focused on a definition of mature suburbs as well as data and methods used for the analysis, and Chapter 4 provided an overview of public policies. This chapter will concentrate on results based on the expert interviews and the quantitative analyses.
5.1 Spatial Patterns The housing markets of the three study counties were split into submarkets: Central City (U.S. Census definition), Mature Suburbs (author’s definition) and Developing Suburbs (everything else). These submarkets are shown in Figures 1, 2, and 3 below, one for each of the three counties. Townships were left out of the analysis. The Chow test conducted above confirms that there are submarkets, although they might have to be refined in future research efforts. As discussed above, classic models of urban change suggest a concentric ring structure (Burgess, 1925), sector structure (Hoyt, 1939), or multiple nuclei structure (Harris & Ullman, 1945). The spatial structures of mature suburbs in the three counties 68
provide examples of these classic models. Cuyahoga County’s mature suburbs have a concentric ring structure (interrupted by Lake Erie). Franklin County’s mature suburbs have a scattered structure that can be attributed to Columbus’s aggressive annexation efforts in the 1950s and 1960s (Bier & Howe, 1998; P. Feldmann, personal communication, November 3, 2003; Howe et al., 1998; T. Marsh, personal communication, October 28, 2003). Many places that would be mature suburbs in a city like Cleveland are actually within the municipal boundaries of Columbus. One such place is Clintonville, a former 1920s streetcar suburb that is now part of the city of Columbus (T. Bier, personal communication, September 10, 2003; T. Marsh, personal communication, October 28, 2003). Hamilton County’s mature suburbs have a structure that is partly scattered, and partly sector shaped. This can be attributed to the hilly topography in the western part of the city and the railroad line that coincides with the chain of mature suburbs in the eastern part of the city (S. Armstrong, personal communication, September 18, 2003; Howe et al., 1998; R. Miller, personal communication, September 23, 2003; W. Miller & J. Keller, personal communication, September 16, 2003; D. Savage, personal communication September 23, 2003). The spatial patterns of these three counties illustrate the importance of local characteristics, such as annexation history and topography, which affect mature suburban communities and their locations. These factors should be considered in any research on urban spatial patterns.
69
Legend: Cleveland Cuyahoga County’s Mature Suburbs Cuyahoga County’s Developing Suburbs Cuyahoga County’s Invalid Places Cuyahoga County’s Townships
Figure 1: Mature Suburbs in Cuyahoga County Note: 30 out of 55 suburban municipalities
70
Legend: Columbus Franklin County’s Mature Suburbs Franklin County’s Developing Suburbs Franklin County’s Invalid Places Franklin County’s Townships
Figure 2: Mature Suburbs in Franklin County Note: 11 out of 24 suburban municipalities
71
Legend Cincinnati Hamilton County’s Mature Suburbs Hamilton County’s Developing Suburbs Hamilton County’s Invalid Places Hamilton County’s Townships
Figure 3: Mature Suburbs in Hamilton County Note: 12 out of 35 suburban municipalities
72
5.2 Themes from Expert Interviews As mentioned above, this study uses qualitative methods to facilitate quantitative methods. The following paragraphs discuss suburban issues associated with population, municipal financing, housing units, and school quality because these were the principal themes that the experts in the three counties consistently brought up during their interviews. Subsequently, these and other factors were used as variables in the quantitative analysis, the results of which will be discussed later.
5.2.1 Population With respect to population, some communities have experienced a loss (Howe et al., 1998; D. O’Leary, personal communication, September 24, 2003). A shrinking population can be partly attributed to outmigration. For example, children may leave parental homes or residents may move to other areas (P. Feldmann, personal communication, November 3, 2003). Shrinking household size has occurred nationwide, but actual loss of households to outmigration can signal a problem. Tables A.15, A.16, and A.17 in the Appendix illustrate the shrinking population in the suburbs of the three study communities from 1990 to 2000. The majority of mature suburbs experienced negative population growth: 84.62% in Cuyahoga County, 72.72% in Franklin County, and 61.54% in Hamilton County. However, this finding is not consistent for developing suburbs: Negative population growth was experienced by 52.94% of developing suburbs in Cuyahoga County, 28.57% in Franklin County, and 55.55% in Hamilton County. The positive population growth in the majority of
developing suburbs in Franklin County can be attributed to Columbus’s eager annexation activities in the past. A shrinking population can also partly be attributed to a rate of death that is higher than the rate of birth or to the unavailability of facilities to provide care for senior citizens who have aged in a community but now need more assistance (Fitzpatrick & Logan, 1985; Logan, 1984). Many suburbs have a high proportion of senior citizens who often face challenges keeping up their lifestyles (P. Feldmann, personal communication, November 3, 2003; Fitzpatrick & Logan, 1985; Logan, 1984). Those who cannot live in their own homes move to homes of relatives or friends, assisted living facilities, nursing homes, or hospices (Pynoos, 1998). These facilities might not be located in the same community when senior citizens need to leave their homes or those in a community might not have space available at the time senior citizens need to move (S. Armstrong, personal communication, September 18, 2003; D. Beach, personal communication, September 9, 2003; D. Brooks, personal communication, September 18, 2003; R. Corrigan, personal communication, September 9, 2003; P. Feldmann, personal communication, November 3, 2003; M. McGinty, personal communication, September 8, 2003; B. Mikelbank, personal communication, September 9, 2003; S. Neal, personal communication, September 17, 2003; L. Ochsendorf, personal communication, November 26, 2003; D. O’Leary, personal communication, September 24, 2003; B. Siegel, personal communication, September 16, 2003; E. Tollerup, personal communication, September 11, 2003; J. Wright, personal communication, September 17, 2003).
74
When senior residents leave their homes, some of the housing units may be bought by in-movers with lower socioeconomic profiles than current residents. This affects municipal finances (L. Ochsendorf, personal communication, November 26, 2003). Many present and in-moving residents demand more municipal services over time (D. Beach, personal communication, September 9, 2003; W. Creager, personal communication, September 17, 2003; D. Feinstein, personal communication, October 10, 2003; S. Neal, personal communication, September 17, 2003). However, some municipalities, especially the so-called graying suburbs (Fitzpatrick & Logan, 1985; Logan, 1994), have faced or are about to face a declining municipal income and tax base because of retirements, aging building stock, and business closings (Gurwitt, 1993). Nevertheless, many experts pointed out that current tax rates in mature suburbs are low in comparison with those in central cities and developing suburbs (D. Beach, personal communication, September 9, 2003; D. Brooks, personal communication, September 18, 2003; R. Corrigan, personal communication, September 9, 2003; W. Creager, personal communication, September 17, 2003; M. McGinty, personal communication, September 18, 2003; J. Kocevar, personal communication, September 11, 2003; R. Miller, personal communication, September 23, 2003; S. Neal, personal communication, September 17, 2003).
5.2.2 Resource/Expenditure Imbalance A second theme emphasized by the interviewees is the imbalance between community resources and needs for expenditure (Gurwitt, 1993). The most important source of income for municipalities is income tax, followed by property tax, sales tax, 75
and (formerly) the estate tax (see Bowman, 1981; Netzer, 1981). Residents generally pay income tax to the community in which they work, not where they live. Although they use municipal services such as police, fire, and trash removal in their communities of residence, they pay income tax to their workplace communities. Property tax is the second most important source of income, but if property values are not increasing, or if residents refuse to tax themselves at a higher rate, property tax revenues will not increase over time. Although sales tax is the third most important source of income, many suburbs do not have many retail businesses. Given the difficulties discussed below, increases in sales tax revenue seem to be unlikely in the future. Estate tax, which used to be another important source of income, was abolished by the State of Ohio in 2003 (R. Bittner, personal communication, September 18, 2003; M. Burns, personal communication, September 16, 2003; R. Corrigan, personal communication, September 9, 2003; D. Lorek, personal communication, November 3, 2003). Many municipalities have a very high proportion of residential land use, so they do not have many employers that contribute to their tax base (V. Barney, personal communication, November 3, 2003; Gurwitt, 1993; D. Lorek, personal communication, November 3, 2003). If suburban communities have downtown areas—and some of them do not—they often experience the loss of privately owned small businesses and have difficulty attracting new businesses, especially when there is not much land available for business use in the first place (R. Bittner, personal communication, September 18, 2003; R. Corrigan, personal communication, September 9, 2003; J. Kocevar, personal communication, September 11, 2003; D. O’Leary, personal communication, September 24, 2003; B. Siegel, personal communication, September 16, 2003). 76
Several communities have a few large employers, or perhaps only one. This can be considered beneficial, but it is also risky because the large employers might leave the city. Examples of cities with a few or one large employer include Ben Venue Laboratories in the City of Bedford (R. Corrigan, personal communication, September 9, 2003), General Motors in the City of Parma (M. McGinty, personal communication, September 8, 2003), Proctor and Gamble in the City of St. Bernard (B. Siegel, personal communication, September 16, 2003), and the Defense Supply Center Columbus in Whitehall (L. Ochsendorf, personal communication, November 26, 2003). At the same time, many mature suburbs battle aging infrastructure that needs maintenance and repairs (D. Beach, personal communication, September 9, 2003; R. Corrigan, personal communication, September 9, 2003; J. Kocevar, personal communication, September 11, 2003; K. Montlack, personal communication, September 11, 2003; D. O’Leary, personal communication, September 24, 2003). A few communities provide infrastructure services that are independently run and are not integrated in a regional system. These communities cannot take advantage of economies of scale. For example, the City of Bedford’s water department and wastewater plant will run into serious maintenance and repair issues at some point because the city (population 14,214 in 2000) will have trouble affording the needed repairs (R. Corrigan, personal communication, September 9, 2003).
5.2.3 Housing The third theme relates to housing. Many mature suburbs house a high proportion of senior citizens (as illustrated by Tables A.20, A.21, and A.22 in the Appendix), 77
although only Cuyahoga County’s mature and developing suburbs and Franklin County’s developing suburbs had a growth in the proportion of senior citizens from 1990 and 2000. Some senior citizens and in-moving home buyers who have lower incomes might face challenges maintaining and repairing their homes. Financial and technical municipal help would be useful both for these households and for maintaining the community, but it would cost a municipality money it might not have (M. McGinty, personal communication, September 8, 2003; J. Kocevar, personal communication, September 11, 2003; S. Neal, personal communication, September 17, 2003). Communities might have issues with landlords who do not reside in the city in which they own rental properties and, therefore, neglect their properties (R. Bittner, personal communication, September 18, 2003; D. Brooks, personal communication, September 18, 2003; D. O’Leary, personal communication, September 24, 2003). Some communities have encountered problems related to vacancies as well as decay (B. Siegel, personal communication, September 16, 2003; E. Tollerup, personal communication, September 11, 2003). Vacancies often happen in connection with a sale or inheritance. In the latter case, heirs may have difficulty deciding what should happen to a property, or they may be waiting for property values to increase. Vacancies and lack of maintenance can create problems (L. Ochsendorf, personal communication, November 26, 2003). In the extreme case, communities can become characterized by abandoned housing, as is the case in East Cleveland (T. Bier, personal communication September 10, 2003). Mature suburbs are often characterized by housing units built on small lots. They often have small rooms, few rooms, and only one or one-and-a-half baths (M. McGinty, personal communication, September 8, 2003; R. Miller, personal communication, 78
September 23, 2003; J. Rokakis, personal communication, September 8, 2003). The study of First Suburbs Consortium Housing Initiative (2002) showed that small lots; small rooms, kitchens, and bathrooms; a small number of rooms, kitchens, and bathrooms; and an insufficient number of electrical outlets are issues in their analyzed suburbs. Some housing units that were built in the 1960s and 1970s have not aged well due to shoddy initial construction (T. Bier, personal communication, September 10, 2003; R. Bittner, personal communication, September 18, 2003; D. Lorek, personal communication, November 3, 2003; J. Magill, personal communication, November 6, 2003; L. Ochsendorf, personal communication, November 26, 2003). Many properties do not have interesting architectural styles to make up for age or small size (T. Bier, personal communication, September 10, 2003; E. Tollerup, personal communication, September 11, 2003).
5.2.4 School Quality The final theme among the experts interviewed is school quality. In several communities schools are in academic emergency, a situation triggered by in-movers with a high need for social services as well as many transitional students, many of them international, who face severe language barriers (R. Bittner, personal communication, September 18, 2003; L. Ochsendorf, personal communication, November 26, 2003). The additional needs of these students create additional tax burdens at the same time that tax revenues are declining or flat. These four themes informed both the choice of variables for the quantitative analysis and the interpretation of the quantitative results.
79
5.3 Descriptive Statistics 5.3.1 Metropolitan Area Characteristics Property values are affected by factors associated with the unit, the neighborhood, and the municipality. They are also affected by conditions at the metropolitan level and, in particular, population growth. These metropolitan-wide factors will be discussed below. All three metropolitan areas had positive population growth between 1990 and 2000. The Cleveland-Akron, OH Consolidated Metropolitan Statistical Area (CMSA) had a growth of 6.74 percent (1990 population: 2,759,823; 2000 population: 2,945,831). 45 The Columbus Metropolitan Statistical Area (MSA) had a growth of 11.81 percent (1990 population: 1,377,419; 2000 population: 1,540,157). 46 The CincinnatiHamilton, OH-KY-IN Consolidated Metropolitan Statistical Area (CMSA) had a growth of 13.48 percent (1990 population: 1,744,124; 2000 population: 1,979,202). 47
5.3.2 Comparisons among Central Counties Although the three central counties are all in Ohio, their housing markets and residents have different characteristics (see Table 2 below). The statistics for properties sold between 1985 and 2000 indicate that the differences in spatial patterns of mature suburbs are accompanied by complex systems of other differences.
45
In comparison with the 1990 census, the U.S. Census Bureau included Ashtabula, Lorain, and Portage Counties in the definition of a metropolitan area. 46 The Columbus MSA grew 11.81 percent despite the fact that the U.S. Census Bureau did not include Union County in the metropolitan definition in 2000, although it did so in 1990. 47 Comparing 1990 with 2000, the U.S. Census Bureau included Brown, OH, Gallatin, KY, Grant, KY, Ohio, IN, and Pendleton, KY counties in the definition of a metropolitan area.
80
Variable (Mean)
NB: Sale Price at Year of Sale Winter Season (Dummy) Spring Season (Dummy) Fall Season (Dummy) Sale Year 1985 (Dummy) Sale Year 1986 (Dummy) Sale Year 1987 (Dummy) Sale Year 1988 (Dummy) Sale Year 1989 (Dummy) Sale Year 1990 (Dummy) Sale Year 1991 (Dummy) Sale Year 1992 (Dummy) Sale Year 1993 (Dummy) Sale Year 1994 (Dummy) Sale Year 1995 (Dummy) Sale Year 1996 (Dummy) Sale Year 1997 (Dummy) Sale Year 1998 (Dummy) Sale Year 1999 (Dummy)
Cuyahoga, Franklin, Hamilton Counties $109,858
Cuyahoga County
Franklin County
Hamilton County
$108,560
$103,485
$118,049
17.41%
16.92%
17.50%
18.16%
28.21%
28.33%
28.06%
28.13%
24.90%
24.92%
25.14%
24.63%
3.55%
3.89%
3.07%
3.44%
4.53%
4.58%
4.20%
4.77%
4.62%
4.70%
4.25%
4.83%
4.44%
4.85%
3.86%
4.31%
4.48%
4.71%
4.03%
4.51%
4.66%
4.92%
4.29%
4.58%
4.66%
4.82%
4.28%
4.75%
5.37%
5.49%
5.07%
5.45%
5.87%
5.64%
6.18%
5.97%
6.43%
6.40%
6.56%
6.37%
6.66%
6.61%
7.00%
6.43%
7.62%
7.42%
8.10%
7.49%
8.13%
7.96%
8.51%
8.06%
9.35%
9.19%
10.00%
9.01%
9.90%
9.53%
10.29%
10.14%
Table 2: Variable Means for Cuyahoga, Franklin, and Hamilton Counties for Properties Sold between 1985 and 2000 (Population of Sales)
81
Table 2 (continued)
Distance to Downtown (miles) (2000) Distance to Nearest Interstate or State Route Access Point (miles) (2000) Property Tax Rate at Year of Sale Number of Murder per 1,000 Population in Municipality (2000) Math Test 6th Grade Pass Rate in School District (2000) Proportion of NonHispanic Whites in Census Tract at Year of Sale Average Household Income in Census Tract at Year of Sale Rolling Topography (Dummy Variable) Lot Size at Year of Sale Square Footage of Housing Unit at Year of Sale Number of Bathrooms of Unit at Year of Sale Age of Housing Unit at Year of Sale Number of Observations Proportion of Total Number of Observations
8.33
9.08
6.61
8.69
1.27
1.32
0.98
1.40
55.80%
58.40%
53.21%
53.88%
5.47
5.13
7.05
4.26
56.52%
57.20%
54.30%
57.45%
81.87%
81.14%
81.67%
83.30%
$50,349
$49,640
$47,676
$53,943
6.01%
1.24%
0.41%
19.27%
12,576
14,860
8,781
11,940
1,638
1,629
1,566
1,718
1.410
1.315
1.447
1.534
42.45
46.12
34.69
43.32
357,754
169,595
91,225
96,934
100%
47.41%
25.50%
27.10%
Some of the variables given in Table 2 above are noteworthy, such as the Distance to Downtown, the Distance to the Nearest Interstate or State Route Access Point, and the Property Tax Rate. The distance to downtown reflects the size of a county, but it might also reflect the locations of housing turnovers. Cuyahoga County measures 82
about 458 square miles, 48 Franklin County measures about 540 square miles, 49 and Hamilton County measures about 407 square miles. 50 From this data, one would assume that the average distance from a property to downtown is highest in Franklin County and lowest in Hamilton County. However, Cuyahoga County is located at Lake Erie and Cleveland’s downtown is near the shore, and Cincinnati’s downtown is located in the southern part of Hamilton County. Therefore, distances from properties to downtown in Cleveland (9.08 miles) and in Cincinnati (8.69 miles) are longer than in Franklin County (6.61 miles), where Columbus is located in the county’s center. The distance from a property to the nearest Interstate or State Route access point might be an indicator of the location of a property in each county. One can assume that the density of access points is higher near a downtown, but the relative location of a downtown influences road networks. Thus, Cleveland has a higher density of access points around downtown, truncated on the north by Lake Erie. Columbus has access points that are well spread over its entire area due to its spatial structure and the fact that downtown can be accessed from all directions. Cincinnati’s road structure is affected by the Ohio River. These interpretations might explain the relatively high distances to access points in Cuyahoga County (1.32 miles) and Hamilton County (1.40 miles) and the low distance to them in Franklin County (0.98 miles). Property tax income is used for school funding and other public purposes. It can also be considered a proxy for municipal services. A high property tax rate might reflect
48
City-Data.com. (2006). Cuyahoga County, Ohio (OH). [Data file]. Available from City-Data.com Web site, http://www.city-data.com/county/Cuyahoga_County-OH.html 49 City-Data.com. (2006). Franklin County, Ohio (OH). [Data file]. Available from City-Data.com Web site, http://www.city-data.com/county/Franklin_County-OH.html 50 City-Data.com. (2006). Hamilton County, Ohio (OH). [Data file]. Available from City-Data.com Web site http://www.city-data.com/county/Hamilton_County-OH.html
83
the need for school and other public services and/or good public services. Comparing among the three counties, Cuyahoga County has the highest property tax rate (58.40), but the average tax rates in Franklin (53.21) and in Hamilton Counties (53.88) are lower. Another difference worth noting is the range of average household incomes within the three counties. Hamilton County has the highest average household income ($53,943), Franklin County has the lowest ($47,676), and Cuyahoga County falls in between. The difference might be explained by the relatively low average age of household heads in Franklin County, since the average household income is expected to be low when the head of household is young. Hamilton County’s population median age is 35.5 years 51 , Cuyahoga County’s is 37.3 years, 52 and Franklin County’s is 32.5 years. 53
5.3.3 Mature Suburbs Compared to Central Cities and Developing Suburbs Consistent with ecological theory (and as seen in Figure 1 above), Cuyahoga County’s mature suburbs fall between the central city and developing suburbs in many characteristics studied (as shown in Table 3 below). In comparison with both Cleveland and Cuyahoga County’s developing suburbs, the properties in Cuyahoga County’s mature suburbs are distant from the nearest highway access points (1.52 miles compared with 0.80 for the city and 1.44 for the developing suburbs). The property tax rate of Cuyahoga County’s mature suburbs is highest among the three groups (60.87 percent versus 58.52
51
Ohio Office of Strategic Research (n.d.). Ohio County Profiles: Hamilton County. Available from Ohio Department of Development website, http://www.odod.state.oh.us/research/files/S0/Hamilton.pdf 52 Ohio Office of Strategic Research (n.d.). Ohio County Profiles: Cuyahoga County. Available from Ohio Department of Development website, http://www.odod.state.oh.us/research/files/S0/Cuyahoga.pdf 53 Ohio Office of Strategic Research (n.d.). Ohio County Profiles: Franklin County. Available from Ohio Department of Development website, http://www.odod.state.oh.us/research/files/S0/Franklin.pdf
84
for Cleveland and 53.19 for developing suburbs). This reflects the excellent municipal services that may be provided in the mature suburbs, such as superior and fast police, fire, and medical services (T. Bier, personal communication, September 10, 2003; M. McGinty, personal communication, September 8, 2003) and good trash removal (M. McGinty, personal communication, September 8, 2003) despite the relative lack of employers to carry the tax burden. These suburbs tax their residents at a relatively high rate and the question remains whether suburban communities will be able to keep their superior services in the long run, given the bleak future outlook with respect to municipal incomes.
85
Variable (Mean)
NB: Sale Price at Year of Sale Winter Season (Dummy) Spring Season (Dummy) Fall Season (Dummy) Sale Year 1985 (Dummy) Sale Year 1986 (Dummy) Sale Year 1987 (Dummy) Sale Year 1988 (Dummy) Sale Year 1989 (Dummy) Sale Year 1990 (Dummy) Sale Year 1991 (Dummy) Sale Year 1992 (Dummy) Sale Year 1993 (Dummy) Sale Year 1994 (Dummy) Sale Year 1995 (Dummy) Sale Year 1996 (Dummy) Sale Year 1997 (Dummy) Sale Year 1998 (Dummy) Sale Year 1999 (Dummy)
Cuyahoga County
$108,560
Cuyahoga County: Central City [Cleveland] $54,833
Cuyahoga County: Mature Suburbs (grouped) $113,011
Cuyahoga County: Developing Suburbs (grouped) $152,605
16.92%
18.82%
16.16%
16.54%
28.33%
26.86%
29.04%
28.40%
24.92%
26.51%
24.52%
24.21%
3.89%
3.42%
3.96%
4.17%
4.58%
3.84%
4.67%
5.14%
4.70%
4.14%
4.75%
5.16%
4.85%
4.82%
4.73%
5.16%
4.71%
4.80%
4.55%
4.96%
4.92%
5.12%
4.71%
5.14%
4.82%
4.72%
4.65%
5.24%
5.49%
4.99%
5.33%
6.30%
5.64%
4.93%
5.71%
6.19%
6.40%
6.36%
6.35%
6.55%
6.61%
7.03%
6.53%
6.33%
7.42%
7.97%
7.25%
7.20%
7.96%
8.44%
7.87%
7.60%
9.19%
9.45%
9.40%
8.55%
9.53%
10.04%
9.86%
8.31%
Table 3: Descriptive Statistics of Cuyahoga County, Cleveland, and Cuyahoga County’s Mature and Developing Suburbs
86
Table 3 (continued) Distance to Downtown (miles) (2000) Distance to Nearest Interstate or State Route Access Point (miles) (2000) Property Tax Rate at Year of Sale Number of Murder per 1,000 Population in Municipality (2000) Math Test 6th Grade Pass Rate in School District (2000) Proportion of Non-Hispanic Whites in Census Tract at Year of Sale Average Household Income in Census Tract at Year of Sale Rolling Topography (Dummy Variable) Lot Size at Year of Sale Square Footage of Housing Unit at Year of Sale Number of Bathrooms of Unit at Year of Sale Age of Housing Unit at Year of Sale Number of Observations Proportion of Total Number of Observations
9.08
5.48
8.85
12.96
1.32
0.80
1.52
1.44
58.40%
58.52%
60.87%
53.19%
5.13
14.80
2.04
1.78
57.20%
24.43%
62.59%
78.00%
81.14%
61.96%
85.99%
90.05%
$49,640
$31,704
$51,662
$63,419
1.24%
0.32%
1.13%
2.39%
14,860
5,463
11,444
30,073
1,629
1,274
1,605
2,028
1.315
1.0806
1.2838
1.6101
46.12
66.24
45.86
26.96
169,595
40,046
81,260
40,074
47.41%
11.19%
22.71%
11.20%
87
Variable (Mean)
Sale Price at Year of Sale Winter Season (Dummy) Spring Season (Dummy) Fall Season (Dummy) Sale Year 1985 (Dummy) Sale Year 1986 (Dummy) Sale Year 1987 (Dummy) Sale Year 1988 (Dummy) Sale Year 1989 (Dummy) Sale Year 1990 (Dummy) Sale Year 1991 (Dummy) Sale Year 1992 (Dummy) Sale Year 1993 (Dummy) Sale Year 1994 (Dummy) Sale Year 1995 (Dummy) Sale Year 1996 (Dummy) Sale Year 1997 (Dummy) Sale Year 1998 (Dummy) Sale Year 1999 (Dummy)
Franklin County
$103,485
Franklin County: Central City [Columbus] $85,672
Franklin County: Mature Suburbs (grouped) $158,113
Franklin County: Developing Suburbs (grouped) $128,842
17.50%
18.10%
15.73%
17.21%
28.06%
27.74%
29.76%
28.06%
25.14%
25.61%
23.62%
24.75%
3.07%
3.01%
3.50%
2.94%
4.20%
3.93%
5.04%
4.24%
4.25%
4.21%
4.33%
4.19%
3.86%
3.92%
3.64%
3.70%
4.03%
4.09%
4.10%
3.59%
4.29%
4.36%
4.41%
3.78%
4.28%
4.16%
4.69%
4.42%
5.07%
4.82%
5.29%
5.72%
6.18%
6.01%
6.58%
6.18%
6.56%
6.59%
5.81%
7.13%
7.00%
7.15%
6.03%
7.34%
8.10%
8.14%
7.18%
8.60%
8.51%
8.78%
8.23%
8.05%
10.00%
9.93%
10.04%
10.32%
10.29%
10.36%
10.59%
10.22%
Table 4: Descriptive Statistics of Franklin County, Columbus, and Franklin County’s Mature and Developing Suburbs
88
Table 4 (continued) Distance to Downtown (miles) (2000) Distance to Nearest Interstate or State Route Access Point (miles) (2000) Property Tax Rate at Year of Sale Number of Murder per 1,000 Population in Municipality (2000) Math Test 6th Grade Pass Rate in School District (2000) Proportion of Non-Hispanic Whites in Census Tract at Year of Sale Average Household Income in Census Tract at Year of Sale Rolling Topography (Dummy Variable) Lot Size at Year of Sale Square Footage of Housing Unit at Year of Sale Number of Bathrooms of Unit at Year of Sale Age of Housing Unit at Year of Sale Number of Observations Proportion of Total Number of Observations
6.61
5.70
5.88
9.97
0.98
0.83
0.94
1.37
53.21%
49.36%
58.83%
59.31%
7.05
9.40
3.24
0.23
54.30%
43.15%
80.95%
70.47%
81.67%
74.40%
93.72%
93.80%
$47,676
$40,813
$68,295
$54,468
0.41%
0.00%
0.04%
0.91%
8,781
7,417
11,064
10,375
1,566
1,437
1,889
1,843
1.447
1.3397
1.6594
1.7192
34.69
37.37
41.39
19.38
91,225
54,780
13,764
14,888
25.50%
15.31%
3.85%
4.16%
89
Table 4 illustrates that Franklin County’s mature suburbs rank highest in the county for several characteristics, such as Sale Price, Pass Rate for the Mathematics Test, Square Footage, Age of the Housing Unit, and Average Household Income. These phenomena reflect Columbus’s relatively slow spatial development in the early part of the 20th century and the city’s aggressive annexation history in the 1950s and 1960s. In other characteristics, Franklin County’s mature suburbs fall between Columbus and its developing suburbs.
90
Variable (Mean)
Sale Price at Year of Sale Winter Season (Dummy) Spring Season (Dummy) Fall Season (Dummy) Sale Year 1985 (Dummy) Sale Year 1986 (Dummy) Sale Year 1987 (Dummy) Sale Year 1988 (Dummy) Sale Year 1989 (Dummy) Sale Year 1990 (Dummy) Sale Year 1991 (Dummy) Sale Year 1992 (Dummy) Sale Year 1993 (Dummy) Sale Year 1994 (Dummy) Sale Year 1995 (Dummy) Sale Year 1996 (Dummy) Sale Year 1997 (Dummy) Sale Year 1998 (Dummy) Sale Year 1999 (Dummy)
Hamilton County
$118,049
Hamilton County: Central City [Cincinnati] $98,570
Hamilton County: Mature Suburbs (grouped) $112,773
Hamilton County: Developing Suburbs (grouped) $142,136
18.16%
18.82%
17.90%
17.58%
28.13%
27.43%
27.78%
28.31%
24.63%
25.75%
25.62%
24.48%
3.44%
2.97%
3.93%
3.81%
4.77%
4.08%
4.53%
4.82%
4.83%
4.01%
4.77%
4.72%
4.31%
3.65%
4.67%
4.36%
4.51%
3.84%
4.75%
4.56%
4.58%
4.18%
4.25%
4.47%
4.75%
4.30%
4.37%
5.03%
5.45%
5.09%
4.99%
5.41%
5.97%
6.08%
6.25%
5.81%
6.37%
6.11%
6.13%
6.28%
6.43%
6.46%
6.46%
6.47%
7.49%
7.63%
7.55%
7.53%
8.06%
8.43%
7.88%
8.01%
9.01%
9.67%
9.32%
8.85%
10.14%
11.54%
10.37%
10.10%
Table 5: Descriptive Statistics of Hamilton County, Cincinnati, and Hamilton County’s Mature and Developing Suburbs
91
Table 5 (continued) Distance to Downtown (miles) (2000) Distance to Nearest Interstate or State Route Access Point (miles) (2000) Property Tax Rate at Year of Sale Number of Murder per 1,000 Population in Municipality (2000) Math Test 6th Grade Pass Rate in School District (2000) Proportion of Non-Hispanic Whites in Census Tract at Year of Sale Average Household Income in Census Tract at Year of Sale Rolling Topography (Dummy Variable) Lot Size at Year of Sale Square Footage of Housing Unit at Year of Sale Number of Bathrooms of Unit at Year of Sale Age of Housing Unit at Year of Sale Number of Observations Proportion of Total Number of Observations
8.69
5.34
7.68
11.81
1.40
1.26
0.85
1.22
53.88%
57.74%
51.96%
52.61%
4.26
6.60
2.24
2.91
57.45%
28.21%
57.90%
69.71%
83.30%
68.90%
86.83%
83.37%
$53,943
$40,879
$52,437
$60,913
19.27%
17.72%
12.18%
16.19%
11,940
8,391
10,328
12,486
1,718
1,622
1,713
1,766
1.534
1.3472
1.4757
1.5989
43.32
66.08
56.25
36.97
96,934
26,485
9,725
19,869
27.10%
7.40%
2.72%
5.55%
92
Hamilton County’s mature suburban properties are nearer to the nearest access point than properties in Cincinnati are, as can be seen in Table 5 above. Other factors pertaining to Hamilton County’s mature suburbs are consistent with the ecological models. In sum, the spatial structures of Cuyahoga and Hamilton Counties are more consistent with the ecological theory because they developed about the same time as Chicago. The theory is descriptive of a historical era of urban development.
5.3.4 Mature Suburbs in the Three Central Counties This section will focus on comparing mature suburbs in Cuyahoga, Franklin, and Hamilton Counties (see Tables 3, 4, and 5 above and Table 6 below). The characteristics of Franklin County’s mature suburbs are influenced by Columbus’s aggressive annexation history in terms of their Sale Price ($158,113 versus $113,001 in Cuyahoga County and $112,773 in Hamilton County), their Math Pass Rate (80.95% versus 62.59% in Cuyahoga County and 57.90% in Hamilton County), Square Footage (1,889 square feet versus 1,713 square feet in Hamilton County and 1,605 square feet in Cuyahoga County), and Average Household Income ($68,295 versus $52,437 in Hamilton County and $51,662 in Cuyahoga County). If Columbus had not annexed land, current central city neighborhoods in Columbus would be in the suburbs. Columbus’s annexation history is the reason why mature suburbs in Franklin County are much better off, on average. Among the three subgroups, the property tax rate is highest in the mature suburbs of Cuyahoga County (60.87), followed by those in Franklin County (58.83) and Hamilton County (51.96). As mentioned above, a high rate might be explained by superior 93
municipal services. 54 Indeed, several experts in Hamilton County’s mature suburbs mentioned excellent and fast police, fire, and medical services (T. Bier, personal communication, September 10, 2003; W. Creager, personal communication, September 17, 2003; B. Siegel, personal communication, September 16, 2003) and good trash removal that might even include free trash bags delivered to residents’ houses and a flexible pickup schedule (B. Siegel, personal communication, September 16, 2003). The City of St. Bernard, among other cities, offers heavily subsidized transportation within their community—a local bus ride for 10 cents, for example, or transportation to medical facilities within the I-275 loop, house calls by the municipality’s health department, and a lifeline (B. Siegel, personal communication, September 16, 2003). One would expect that the superior services in Hamilton County’s communities would be reflected in a high property tax rate, but instead Hamilton’s mature suburbs have the lowest average tax rate of the three counties.
54
Currently, only crime and the pass rate of the 6th grade math test were incorporated as a proxy of municipal services. Better proxies of municipal services could be the square footage of parks, the number of books in a library, or services that are offered for senior citizens.
94
Variable (Mean)
Sale Price at Year of Sale Winter Season (Dummy) Spring Season (Dummy) Fall Season (Dummy) Sale Year 1985 (Dummy) Sale Year 1986 (Dummy) Sale Year 1987 (Dummy) Sale Year 1988 (Dummy) Sale Year 1989 (Dummy) Sale Year 1990 (Dummy) Sale Year 1991 (Dummy) Sale Year 1992 (Dummy) Sale Year 1993 (Dummy) Sale Year 1994 (Dummy) Sale Year 1995 (Dummy) Sale Year 1996 (Dummy) Sale Year 1997 (Dummy) Sale Year 1998 (Dummy) Sale Year 1999 (Dummy)
Cuyahoga, Franklin, Hamilton Counties: Mature Suburbs $118,252
Cuyahoga County: Mature Suburbs (grouped) $113,011
Franklin County: Mature Suburbs (grouped) $158,113
Hamilton County: Developing Suburbs (grouped) $142,136
16.27%
16.16%
15.73%
17.58%
29.02%
29.04%
29.76%
28.31%
24.50%
24.52%
23.62%
24.48%
3.90%
3.96%
3.50%
3.81%
4.70%
4.67%
5.04%
4.82%
4.70%
4.75%
4.33%
4.72%
4.58%
4.73%
3.64%
4.36%
4.51%
4.55%
4.10%
4.56%
4.63%
4.71%
4.41%
4.47%
4.63%
4.65%
4.69%
5.03%
5.29%
5.33%
5.29%
5.41%
5.88%
5.71%
6.58%
5.81%
6.26%
6.35%
5.81%
6.28%
6.46%
6.53%
6.03%
6.47%
7.26%
7.25%
7.18%
7.53%
7.92%
7.87%
8.23%
8.01%
9.48%
9.40%
10.04%
8.85%
10.01%
9.86%
10.59%
10.10%
Table 6: Descriptive Statistics of Mature Suburbs in Cuyahoga, Franklin, and Hamilton Counties (combined and separate)
95
Table 6 (continued) Distance to Downtown (miles) (2000) Distance to Nearest Interstate or State Route Access Point (miles) (2000) Property Tax Rate at Year of Sale Number of Murder per 1,000 Population in Municipality (2000) Math Test 6th Grade Pass Rate in School District (2000) Proportion of Non-Hispanic Whites in Census Tract at Year of Sale Average Household Income in Census Tract at Year of Sale Rolling Topography (Dummy Variable) Lot Size at Year of Sale Square Footage of Housing Unit at Year of Sale Number of Bathrooms of Unit at Year of Sale Age of Housing Unit at Year of Sale Number of Observations Proportion of Total Number of Observations
8.35%
8.85
5.88
11.81
1.43
1.52
0.94
1.22
59.77%
60.87%
58.83%
52.61%
2.22
2.04
3.24
2.91
64.57%
62.59%
80.95%
69.71%
87.08%
85.99%
93.72%
83.37%
$53,847
$51,662
$68,295
$60,913
0.02%
1.13%
0.04%
16.19%
11,289
11,444
11,064
12,486
1,653
1,605
1,889
1,766
1.3509
1.2838
1.6594
1.5989
46.27
45.86
41.39
36.97
104,740
81,260
13,764
19,869
29.28%
22.71%
3.85%
5.55%
96
5.4 House Price Behavior/Appreciation The third research question asks how property values of single family homes have behaved in mature suburbs compared with those of single family homes in central cities and developing suburbs. Tables 7 through 11 provide answers for the years 1986 to 1999 based on regression coefficients of the dummy sale year variables. These tables list the appreciation rates for the uncontrolled case (a regression with dummy sale years only as independent variables) and the controlled case (a regression with all independent variables as discussed in Chapter 3) although only the controlled cases will be discussed below. In order to avoid perfect multicollinearity, dummy sale year 2000 was removed from the analysis, so results for appreciation were only calculated up to 1999. Table 7 provides information on the sale price appreciation for the three counties combined as well as separated. Table 8 gives information on the sale price appreciation for Cuyahoga County, Cleveland, and Cuyahoga County’s Mature and Developing Suburbs. Table 9 gives information on the sale price appreciation for Franklin County, Columbus, and Franklin County’s Mature and Developing Suburbs. Table 10 gives information on the sale price appreciation for Hamilton County, Cincinnati, and Hamilton County’s Mature and Developing Suburbs. Finally, Table 11 provides information on the sale price appreciation for all mature suburbs and mature suburbs (grouped) in Cuyahoga, Franklin, and Hamilton Counties. Figure 4 and Table 7 below illustrate the volatility of appreciation in the three counties, especially in Franklin County, with positive peaks in 1986 (12.59 percent) and 1987 (14.52 percent) and a negative peak in 1988 (-12.20 percent). Overall, there seems
97
to have been more volatility in the late 1980s and early 1990s than in the mid- to late1990s, as well as a slight overall downward trend.
Sale Price Appreciation 1986 - 1999 Cuyahoga, Franklin, and Hamilton Counties (combined and separate)
CuyFraHamCounties Cuyahoga Franklin Hamilton
15.00
Appreciation
10.00 5.00 0.00 -5.00 -10.00
99 19
98 19
97 19
96 19
95
94 19
19
93 19
92 19
91 19
89
90 19
19
88 19
87 19
19
86
-15.00
Years
Figure 4: Sale Price Appreciation 1986 – 1999: Cuyahoga, Franklin, and Hamilton Counties (combined and separate) 55
55
Figures 4 through 8 illustrate the controlled case only.
98
Uncontrolled Rate Controlled Rate 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 1986-1999
Cuyahoga, Cuyahoga Franklin Hamilton County County County Franklin, Hamilton Counties 8.41% 7.22% 12.54% -7.33% 9.14% 2.47% 12.59% 5.82% 6.02% 3.24% 5.16% 11.27% -1.59% 8.87% 14.52% 12.75% -1.86% -3.89% -0.52% 2.05% 5.86% -0.18% -12.20% -6.08% 8.98% 13.07% 4.61% 4.28% 7.61% 13.60% 3.67% 10.98% 5.15% 6.44% 2.16% 5.75% 4.86% 5.73% 4.06% 5.49% 5.37% 5.14% 7.05% 4.09% 4.46% 8.67% -0.31% 3.64% 11.64% 17.18% 6.42% 6.05% 2.23% 0.59% 8.55% 5.57% -0.00% -0.55% 3.24% -1.51% 3.27% 6.05% -0.09% 0.52% 0.37% -5.57% 5.35% 4.90% 10.66% -2.76% 5.04% 7.07% 2.98% 2.36% 2.43% 4.85% -4.73% 2.59% 4.29% -8.51% 4.80% -4.67% 6.71% 3.02% 9.42% 8.39% 5.31% 12.76% 4.02% 4.67% 3.45% 3.43% 3.66% -0.46% 5.38% 4.70% 3.58% 2.76% 6.72% 1.90% 0.89% 6.91% 2.31% 6.81% 3.23% 3.31% 2.57% 3.54% -4.89% 7.15% 0.19% -0.43% 4.48% 3.62% 4.85% 3.31% 3.63% 4.83% 3.81% 4.36%
Table 7: Sale Price Appreciation 1986 – 1999: Cuyahoga, Franklin, and Hamilton Counties (combined and separate) Figure 5 and Table 8 below illustrate the volatility that is especially the case in Cleveland, with positive peaks in 1993 (18.96 percent) and 1995 (14.08 percent) and a negative peak in 1994 (-16.69 percent). As mentioned above in connection with Table 7, 99
there seems to have been more volatility in the late 1980s and early 1990s than in the mid- to late-1990s.
Sale Price Appreciation 1986 - 1999 Cuyahoga County, Cleveland, and Cuyahoga County's Mature and Developing Suburbs
Appreciation
Cuyahoga 20.00
Cleveland
15.00
Cuy Mature
10.00
Cuy Developing
5.00 0.00 -5.00 -10.00 -15.00
99 19
98 19
96
97 19
19
95 19
19
94
93
92
19
19
91 19
90 19
89 19
87
88 19
19
19
86
-20.00
Years
Figure 5: Sale Price Appreciation 1986 – 1999: Cuyahoga County, Cleveland, and Cuyahoga County’s Mature and Developing Suburbs
100
Uncontrolled Cuyahoga Cleveland County Rate Controlled Rate 7.22% 10.21% 1986 2.47% 1.63% 3.24% -4.82% 1987 8.87% -9.45% -3.89% 2.50% 1988 -0.18% 9.68% 13.07% 12.64% 1989 13.60% -4.51% 6.44% 6.91% 1990 5.73% 6.91% 5.14% 0.25% 1991 8.67% 0.56% 17.18% 10.74% 1992 0.59% -0.90% -0.55% 15.46% 1993 6.05% 18.96% -5.57% -10.56% 1994 -2.76% -16.69% 2.36% 4.52% 1995 2.59% 14.08% -4.67% 6.68% 1996 8.39% -1.82% 4.67% 7.14% 1997 -0.46% 3.31% 2.76% 5.76% 1998 6.91% -1.15% 3.31% 5.46% 1999 7.15% 2.46% 1986-2000 3.62% 5.21% 4.83% 1.65%
Cuyahoga County: Mature Suburbs 5.12% -2.28% 4.27% 4.86% 4.02% 7.90% 6.88% 1.56% 4.47% 8.86% 6.85% 5.30% 5.99% 0.82% 4.66% 6.51% -0.02% -14.50% 3.26% 18.36% 2.85% 2.54% 4.64% 6.58% 2.37% 3.44% 4.01% 4.48% 4.24% 3.89%
Cuyahoga County: Developing Suburbs 5.61% 6.56% 10.92% 5.01% 8.20% 4.76% 7.36% 6.09% 8.41% 5.32% 3.40% 2.45% 7.44% 4.98% 2.01% 4.42% -0.86% 1.35% 5.62% 3.92% -5.95% 5.57% 2.32% 2.65% -1.24% 1.63% 3.00% 5.27% 4.02% 4.28%
Table 8: Sale Price Appreciation 1986 – 1999: Cuyahoga County, Cleveland, and Cuyahoga County’s Mature and Developing Suburbs Figure 6 and Table 9 below illustrate the volatility in Franklin County, although the majority of appreciation rates is positive. This volatility decreases in the second half of the 1990s when appreciation rates were still mostly positive. 101
Sale Price Appreciation 1986 - 1999 Franklin County, Columbus, and Franklin County's Mature and Developing Suburbs Franklin Columbus 14.00
Fra Mature
12.00
Fra Developing
Appreciation
10.00 8.00 6.00 4.00 2.00 0.00
98
99 19
19
97 19
96 19
95 19
94 19
93 19
91
92 19
19
90 19
89 19
88 19
87 19
19
86
-2.00
Years
Figure 6: Sale Price Appreciation 1986 – 1999: Franklin County, Columbus, and Franklin County’s Mature and Developing Suburbs
102
Uncontrolled Franklin Columbus County Rate Controlled Rate 12.54% 11.35% 1986 12.59% 7.20% 5.16% 7.47% 1987 14.52% 0.86% -0.52% -0.21% 1988 -12.20% 11.08% 4.61% 4.09% 1989 3.67% 1.79% 2.16% 2.01% 1990 4.06% 2.85% 7.05% 5.31% 1991 -0.31% 3.89% 6.42% 5.47% 1992 8.55% -3.89% 3.24% 6.53% 1993 -0.09% 16.51% 5.35% 5.30% 1994 5.04% 1.23% 2.43% 4.38% 1995 4.29% 0.78% 6.71% 5.15% 1996 5.31% 9.05% 3.45% 4.15% 1997 5.38% 0.97% 6.72% 8.15% 1998 2.31% 8.96% 2.57% 0.11% 1999 0.19% -1.17% 1986-2000 4.85% 4.95% 3.81% 4.29%
Franklin County’s Mature Suburbs 7.10% 6.47% 8.90% 7.48% 4.34% 2.35% 3.89% 1.70% -0.63% 2.26% 9.34% 11.50% 2.36% 3.79% 1.58% 0.24% 6.86% 5.88% 2.01% 8.70% 5.49% -1.16% 3.03% 4.36% 1.74% 0.13% 6.32% 16.83% 4.45% 5.04%
Franklin County’s Developing Suburbs 12.19% 3.27% 8.40% 6.75% -0.00% -0.01% 2.21% 5.21% 9.44% 14.27% -1.03% -12.56% 10.10% 4.77% 0.00% 3.57% 9.10% 2.71% 3.82% 2.17% 5.53% 3.98% 5.54% 3.74% -0.00% 1.43% 3.74% 5.69% 4.93% 3.21%
Table 9: Sale Price Appreciation 1986 – 1999: Franklin County, Columbus, and Franklin County’s Mature and Developing Suburbs
103
Figure 7 and Table 10 below illustrate the sale price appreciation in Hamilton County and its submarkets with the exception of the model for Cincinnati, which is not full rank. All submarkets are very volatile, as can be seen in Figure 7.
Sale Price Appreciation 1986 - 1999 Hamilton County, Cincinnati, and Hamilton County's Mature and Developing Suburbs Hamilton Cincinnati 20.00
Ham Mature Ham Developing
Appreciation
15.00 10.00 5.00 0.00 -5.00
99 19
98 19
97 19
95
96 19
19
19
94
93 19
92 19
91 19
90 19
89 19
87
88 19
19
19
86
-10.00
Years
Figure 7: Sale Price Appreciation 1986 – 1999: Hamilton County, Cincinnati, and Hamilton County’s Mature and Developing Suburbs Note: Cincinnati model not full rank
104
Uncontrolled Hamilton Cincinnati County Rate Controlled Rate -7.33% NFR 1986 5.82% 11.27% NFR 1987 12.75% 2.05% NFR 1988 -6.08% 4.28% NFR 1989 10.98% 5.75% NFR 1990 5.49% 4.09% NFR 1991 3.64% 6.05% NFR 1992 5.57% -1.51% NFR 1993 0.52% 4.90% NFR 1994 7.07% 4.85% NFR 1995 -8.51% 3.02% NFR 1996 12.76% 3.43% NFR 1997 4.70% 1.90% NFR 1998 6.81% 3.54% NFR 1999 -0.43% NFR 1986-2000 3.31% 4.36%
Hamilton County’s Mature Suburbs 4.62% -7.96% 21.83% 15.93% -8.37% -4.62% 8.94% 18.30% 7.03% 8.05% 5.13% -5.65% 2.67% 4.81% 3.03% -1.99% 9.01% 14.27% 2.52% -4.11% -1.78% 0.01% 6.42% 2.50% -0.68% 0.95% 7.95% 8.51% 4.88% 3.50%
Hamilton County’s Developing Suburbs 9.20% 13.32% -2.53% -2.08% 4.63% -5.80% -1.57% 18.79% 10.06% 7.15% 0.64% -1.94% 10.68% 9.53% -3.62% 2.41% 8.86% 6.11% 4.38% 4.19% 2.38% 3.19% 4.97% 3.16% 0.27% 4.77% 3.76% 4.60% 3.72% 4.81%
Table 10: Sale Price Appreciation 1986 – 1999: Hamilton County, Cincinnati, and Hamilton County’s Mature and Developing Suburbs Note: NFR means model not full rank
105
Finally, Figure 8 and Table 11 for mature suburbs show that, with the exception of Hamilton County’s mature suburbs in the late 1980s, the appreciation in mature suburbs is much less volatile than in other comparisons discussed above. Over the 15year period of the study the appreciation rate centers around five percent, although in the late 1990s the appreciation rate seems to be lower than five percent. This might indicate a trend of increasing appreciation rates at a decreasing rate. Tables A.23 through A.27 in the Appendix show the appreciation in all submarkets based on two-year moving averages. Results indicate that creating moving averages smooths out results. Sale Price Appreciation 1986 - 1999 Mature Suburbs in Cuyahoga, Franklin and Hamilton Counties (combined and separate) Mat Suburbs Cuy Mature
25.00
Fra Mature
20.00
Ham Mature
Appreciation
15.00 10.00 5.00 0.00 -5.00
99 19
97
98 19
19
96 19
94
95 19
19
93 19
92 19
91 19
90 19
89 19
87
88 19
19
19
86
-10.00
Years
Figure 8: Sale Price Appreciation 1986 – 1999: Mature Suburbs in Cuyahoga, Franklin, and Hamilton Counties (combined and separate)
106
Uncontrolled Rate Controlled Rate
1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 1986-2000
Cuyahoga, Franklin, Hamilton Counties: Mature Suburbs 6.03% 9.14% 5.83% -1.59% 2.32% 5.86% 7.20% 7.61% 4.34% 4.86% 7.22% 4.46% 5.18% 2.23% 4.30% 3.27% 1.22% 10.66% 3.04% -4.73% 2.92% 9.42% 4.70% 3.66% 2.18% 0.89% 4.55% -4.89% 4.36% 3.63%
Cuyahoga County: Mature Suburbs (grouped)
Franklin County: Mature Suburbs (grouped)
5.12% -2.28% 4.27% 4.86% 4.02% 7.90% 6.88% 1.56% 4.47% 8.86% 6.85% 5.30% 5.99% 0.82% 4.66% 6.51% -0.02% -14.50% 3.26% 18.36% 2.85% 2.54% 4.64% 6.58% 2.37% 3.44% 4.01% 4.48% 4.24% 3.89%
7.10% 6.47% 8.90% 7.48% 4.34% 2.35% 3.89% 1.70% -0.63% 2.26% 9.34% 11.50% 2.36% 3.79% 1.58% 0.24% 6.86% 5.88% 2.01% 8.70% 5.49% -1.16% 3.03% 4.36% 1.74% 0.13% 6.32% 16.83% 4.45% 5.04%
Hamilton County: Mature Suburbs (grouped)
4.62% -7.96% 21.83% 15.93% -8.37% -4.62% 8.94% 18.30% 7.03% 8.05% 5.13% -5.65% 2.67% 4.81% 3.03% -1.99% 9.01% 14.27% 2.52% -4.11% -1.78% 0.01% 6.42% 2.50% -0.68% 0.95% 7.95% 8.51% 4.88% 3.50%
Table 11: Sale Price Appreciation 1986 – 1999: Mature Suburbs in Cuyahoga, Franklin, and Hamilton Counties (combined and separate)
107
With respect to the average controlled appreciation rate from 1986 to 2000, mature suburbs in Cuyahoga County have a rate (3.89 percent) that is higher than Cleveland but lower than Cuyahoga County’s developing suburbs. Mature suburbs in Franklin County are characterized by an appreciation (5.04 percent) that is the highest of all submarkets in that county. Hamilton County’s mature suburbs have an appreciation rate (3.50 percent) that is lower than that of its developing suburbs. Whereas Franklin County’s mature suburbs seem to be healthy in terms of their housing stock due to Columbus’s eager annexation efforts, the mature suburbs of the other two counties have rates that might be indicative of a gradual suburban decline in terms of property values. However, it is currently too soon to conclude with certainty that there is suburban decline in terms of property values. Appreciation rates should be examined in the future to see whether this trend continues.
5.5 Regression Models The fourth research question asks what specific factors influence property values of single family homes in mature suburbs compared to those of single family homes in central cities and developing suburbs. A regression model was estimated for all nine submarkets and the results are shown in Tables 12, 13, and 14 below. The Weighted Least Squares (WLS) model corrects for heteroskedasticity. Only those coefficients that were significant will be discussed below.
108
WLS Coefficient (t value)
Intercept Winter Season (Dummy) Spring Season (Dummy) Fall Season (Dummy) Sale Year 1985 (Dummy) Sale Year 1986 (Dummy) Sale Year 1987 (Dummy) Sale Year 1988 (Dummy) Sale Year 1989 (Dummy) Sale Year 1990 (Dummy) Sale Year 1991 (Dummy) Sale Year 1992 (Dummy) Sale Year 1993 (Dummy) Sale Year 1994 (Dummy) Sale Year 1995 (Dummy) Sale Year 1996 (Dummy) Sale Year 1997 (Dummy) Sale Year 1998 (Dummy) Sale Year 1999 (Dummy)
Cuyahoga, Franklin, Hamilton Counties 10.22038 (1,192.67) -0.13803 (-60.92) -0.00866 (-4.88) -0.00963 (-5.13) -0.65301 (-108.37) -0.56163 (-119.54) -0.57749 (-131.65) -0.51892 (-81.89) -0.44283 (-89.33) -0.39427 (-90.71) -0.34971 (-84.66) -0.32745 (-75.91) -0.29476 (-96.71) -0.18814 (-57.16) -0.23548 (-67.47) -0.14130 (-47.12) -0.10469 (-38.06) -0.09579 (-32.00) -0.14466 (-51.57)
Cuyahoga County
10.58274 (921.43) -0.06003 (-20.88) -0.05223 (-24.27) -0.04607 (-18.94) -0.68065 (-118.25) -0.65594 (-132.65) -0.56726 (-99.70) -0.56902 (-98.10) -0.43300 (-86.25) -0.37575 (-71.41) -0.28905 (-59.91) -0.28317 (-62.13) -0.22272 (-53.24) -0.25028 (-56.92) -0.22438 (-55.13) -0.14049 (-32.92) -0.14514 (-36.19) -0.07609 (-20.19) -0.00461 (-1.26)
Franklin County
9.91251 (557.95) -0.03720 (-9.15) 0.02062 (7.16) -0.01750 (-5.10) -0.66829 (-53.78) -0.54242 (-84.65) -0.39719 (-80.98) -0.51923 (-74.31) -0.48249 (-66.64) -0.44194 (-65.77) -0.44509 (-47.70) -0.35963 (-55.38) -0.36054 (-58.75) -0.31016 (-46.94) -0.26726 (-40.92) -0.21414 (-38.44) -0.16039 (-36.08) -0.13729 (-24.95) -0.13538 (-28.39)
Hamilton County
10.41452 (587.72) -0.04287 (-8.57) NS -0.04078 (-10.57) -0.71003 (-40.47) -0.65181 (-39.12) -0.52436 (-70.72) -0.58515 (-30.94) -0.47536 (-32.44) -0.42051 (-39.07) -0.38412 (-40.82) -0.32842 (-45.23) -0.32323 (-44.64) -0.25250 (-33.40) -0.33755 (-53.58) -0.20991 (-34.55) -0.16292 (-26.24) -0.09487 (-14.74) -0.09916 (-16.66)
Table 12: Log-Linear Weighted Least Squares Regression: Cuyahoga, Franklin, and Hamilton Counties (combined and separate)
109
Table 12 continued
Distance to Downtown (miles) (2000) Distance to Nearest Interstate or State Route Access Point (miles) (2000) Property Tax Rate at Year of Sale Number of Murder per 1,000 Population in Municipality (2000) Math Test 6th Grade Pass Rate in School District (2000) Proportion of NonHispanic Whites in Census Tract at Year of Sale Average Household Income in Census Tract at Year of Sale Rolling Topography (Dummy Variable) Lot Size at Year of Sale Square Footage of Housing Unit at Year of Sale Number of Bathrooms of Unit at Year of Sale Age of Housing Unit at Year of Sale R-Square Adjusted R-Square F Value Pr > F
0.00668 (23.92)
-0.00661 (-16.07)
0.00505 (9.12)
-0.00908 (-14.97)
NS
0.01123 (10.95)
0.01078 (4.88)
-0.04589 (-29.92)
0.00210 (20.38) NS
0.00110 (9.40) -0.00973 (-34.41)
0.00192 (7.08) 0.00708 (18.88)
0.00199 (9.07) NS
0.00304 (67.61)
0.00274 (45.97)
0.00070817 (8.42)
0.00153 (15.54)
0.00367 (67.45)
0.00371 (54.97)
0.00577 (56.83)
0.00404 (43.27)
0.00454 (96.90)
0.00336 (63.49)
0.00413 (47.14)
0.00430 (42.14)
-0.01665 (-4.11)
-0.03946 (-4.40)
NS
0.02136 (5.19)
-0.00009037 (-4.64) 0.41498 (200.05)
NS 0.37294 (147.98)
0.00994 (30.78) 0.38304 (98.71)
0.00337 (17.47) 0.45837 (123.41)
NS
0.02441 (6.23)
0.13149 (47.88)
0.01111 (2.87)
-0.00004724 (-79.52) 0.5211 0.5210 9,572.71
