SUBURBANIZATION AND SPRAWL:
A CASE STUDY AND POLICY IMPLICATION
MOHAMMAD H. POURFAYAZ
SUBURBANIZATION AND SPRAWL: A Case Study and Policy Implication
Mohammad H. Pourfayaz
Suburbanization and Sprawl:
A Case Study and Policy Implication
Mohammad H. Pourfayaz
has been approved for the School of Architecture and Planning
Professor Thomas A. Clark
A thesis submitted in partial fulfillment of the requirements for the degree of Master of Urban and Regional Planning.
The author wishes to express his gratitude to those who served on his committee, without whose understanding and cooperation this thesis may never have been compieted. I greatly appreciate their willingness to serve on this committee and their input on the various drafts. Professor Thomas Clark, of the University of Colorado at Denver, School of Architecture and Planning, encouraged the author throughout the preparation of this thesis. He was not only my thesis advisor, but he was supportive of me on many other levels. Professor Yuk Lee gave me valuable guidance and criticism during the course of this research. 1 also received a great amount of advice and help from Mr. Larry Mugler, the Director of Development Service of DRCOG. Despite his tight schedule with DRCOG, he was very generous with his time and advice. Finally, a special thank you to my good friend, Elk Friesell, who helped me in editing my thesis.
TABLE OF CONTENTS
Chapter I: Introduction 1
1 - A: Problem Statement 1
2 - B: Thesis Goals 8
3 - C: Thesis Organization 9
Chapter II: Methodology 12
1 - A: Thesis Hypotheses 12
2 - B: Methodology and Literature Sources 13
3 - C: Problems Involved 14
Chapter III: Description of the Case Study Area 15
1 - The Metropolitan Setting 15
2 - Two Suburban Perspective 17
A: Function View of Suburbs 17
B: Stages of Development 27
Chapter IV: Case Study "Denver Metropolitan Area" 34
1 - Variable Considerations 34
2 - Variables and Sources 35
3 - Correlation Matrix 38
4 - Regression Models 43
5 - Nonlinearity 53
A: Polnomial Equations 53
B: Logarithmic Transformations 58
C: Semi Log 59
6 - Population Changes 60
7 - Evaluation and Conclusion 61
Chapter V: Analysis of the Determinants of 64
1 - The Value of land 64
2 - Amenity 66
3 - Shopping Center 67
4 - Population and Employment Consideration 68
5 - The Relationship Between Population and
Employment Changes 69
6 - Employment and Population Policy 71
Chapter VI: Conclusion and Policy Recommendations 74
1 - A: Renewal 75
2 - B: Preservation of Agricultural Land 76
3 - C: Prevention of Unplanning Sprawl 77
4 - Landuse Transportation Interaction 78
5 - Conclusion 80
6 - Policy Recommendation for Denver 81
7 - Policy Recommendation for the Rest 84
A: Equations 86
B: Bibliography 89
LIST OF TABLES
1.1 Types of Costs Analyzed 4
3.1 The Ratio of: Total Employee/Total Population in Each City in 1980 19
3.2 The Ratio of: Total Employee/Total Population in Each City in 1985 21
4.1 Correlation Matrix for Period A 39
4.2 Correlaion Matrix for Period B 39
4.3 The Results of Stepwise Regression for Period B 49
4.4 The Results of Stepwise Regression Period A 49
5.1 The Ratio of Incremental Employment to Incremental Population 70
LIST OF FIGURES
3.1 Histogram and Frequency Polygon 30
4.1 Population Number of Functions 45
4.2 Population Number of Functional Units 45
5.1 Conceptua odel 73
6.1 Land Use Transportation Interaction 79
6.2 Cohort Survival Models for 1980 and 2000 82
Chapter I Introduction
A: Problem Statement
Although urban sprawl and suburbanization (the decentralization of activities and population from the city to the adjacent fringe areas) is not a new phenomenon for American metropolitan areas since the end of World War II, in the last decade the growth fringe has been most striking and a complicated problem in this society. While this phenomenon has been synonymous with residential dispersion and employment dispersion, the latter seems to play a more important role; in recent years office employment in many suburbs represents a growing, and in the view of many, irreversible trend. The geographic extent of recent suburban office growth can be seen in many U.S. metropolitan areas like Atlanta, Dallas, Denver, Los Angeles, etc. A good example of this case is at Denver's southeast 1-25 corridor, comprised of several high-tech office park developments, it has produced more office space than all of downtown Denver. The downtown's share of total regional employment is expected to decline from 40 percent in 1975 to 25 percent in the year 2000, despite major public efforts to curb this exodus.
Sprawl caused due to either residential dispersion or employment dispersion, leads to such difficulties as: consuming open space, destroying community, increasing inordinate social costs, less efficient resource use (land, water, and energy) and final serious internal congestion (trip patterns are
becoming more and more diffuse in SMSA's) which may jeopardize economic growth.
Federal, state, and local government decisions were most responsible for present situations. Establishment of the federal housing administration made home ownership available to a much larger segment of society, and the single family suburban home became the American dream. Also from the federal level came funding for highways, including the interstate highway system. These programs made low density suburbanization possible. States have tended to react to trends, rather than establishing them, and have also engaged vigorously in road building with little conception of consequences. Local governmental policy usually allows dispersed development to continue unhelped for fear of encroaching on developer's "rights" to real estate profits, despite the increased cost-to-tax base ratio that suburbanization creates. Policy makers are reluctant to take any measures restricting auto usage lest their popularity wane among their constituents.
At this point it is important to clarify the intent of this research.
1. It is important to define "urban sprawl" quite carefully. This term has often been equated to suburban growth. Normally "sprawl" leads to suburbanization.
2. While the term "urban sprawl" is occasionally focused on fine grained development patterns in rather small areas, documentation at this scale is difficult. Based upon these reasons I have substituted suburban data for data at the subjurisdictional areas.
3. It should be stressed that the decentralization (i.e., the redistribution of population and employment away from urban centers to outlying areas) most normally is associated with
planned dispersal. "Sprawl," on the other hand, is unplanned. Still, both decentralization and sprawl are fostered by similar factors.
4. Moreover, the core of this study does not attempt to evaluate sprawl. Rather it pursues the identification of factors which promote sprawls, decentralization and suburbanization, in order to present the best possible recommendations for limiting sprawl.
3. I seek to examine the phenomenon of decentralization and
sprawl by examining the reasons why different suburban jurisdictions in the same metropolitan region grow at different rates. Here, I focus on changes in both population and employment.
The significance of the problems associated with sprawl would obligate us to take a closer look at the cost of sprawl. Hence we refer to the governmental documents (Real Estate Research Corp. 1974, Vol. 2, pp. 5-8 and Vol. 3, pp. 1-6). In these two reports sprawl development has been compared with orderly development. Table 1.1 shows the types of cost that have been included.
Table 1.1 Types of Costs Analyzed
(capital and operating) Environmental Effects
residential (capital only) air pollution
open space/recreation water pollution, erosion
streets and roads vegetation and wildlife
utilities (sewer, water, visual effects
storm, drainage, gas water and energy consumption
public facilities and services: Personal Effects
police, fire, solid waste
collection, library, health use of discretionary time
care, churches, general psychic costs
government travel time
land traffic accident crime
Here are some brief points:
1. Higher residential densities require substantially less land for
housing, perhaps as little as one-tenth of the single-family residential land requirements for a given number of dwelling units.
2. Likewise land requirements for rights-of-way, recreation,
schools, and institutions decrease with higher densities but not as greatly as residential land.
3. Housing costs decrease as density increases because of
economies of scale and common elements as well as reduced unit sizes and less amenities.
4. Roadway and utility networks are reduced as development
patterns become more orderly and contiguous and as densities increase.
3. Costs of schools, library, recreation, health, administrative
facilities and services primarily vary with the population served. Since
the population characteristics depend on the pattern of development, their costs can differ markedly. Lower densities and scattered developments increase the likelihood that multiple locations for these operations are needed.
6. Lower densities entail longer travel distance by users to reach dispersed facilities.
Air quality, water quality, noise, soil erosion, and vegetation/wildlife generally are adversely affected by increasing densities. However, by clustering development in a rational manner, it is possible to conserve ecologically sensitive areas and minimize adverse effects at higher densities. While higher densities cause greater automobile concentrations, driving trips and distances are reduced. In general, single-family home ownership is more costly, even after tax adjustments, than renting apartments. Living in single-family homes also requires more in terms of time, both for household chores and auto travel. In higher density areas, travel time is reduced due to shorter trips, and household time is reduced for maintenance and therefore increased for leisure. Traffic accidents may be more frequent in higher density areas due to congestion, but this is perhaps offset by more auto travel in low density areas. In either case, well-designed and planned streets and highways reduce accident rates. As regards psychic costs, there are tradeoffs among housing types and densities as to privacy, comfort and security, possession and responsibility, and aesthetic values. In this regard, individual preferences, attitudes and tolerances will tailor choices among types and amenities.
Duane Windsor, in his study (Windsor, 1979) not only helps clarify the
meaning of sprawl, but also leads us into a discussion of the costs associated with this form of growth. Windsor concluded that for every cost and impact measure used in the costs of sprawl, the high density planned community is superior to low density sprawl. He said that sprawl patterns are characterized by leap frog development, an utter lack of coordination above the neighborhood level; while planned patterns are characterized by coordinated and compact development.
Later in his report, he stated that the cost of sprawl was intended to encourage higher densities in the urban fringe around central cities. The message intended for suburban communities is that the fiscal and environmental effects of residential development can be minimized by higher densities through clustering. The message intended for federal officials is that the costs and impacts of expected suburban growth can be reduced considerably by avoiding present sprawl patterns. Finally, he says that: there are two main reasons for the existence of low density sprawl.
1. At least in principle, the real estate market does precisely that through the interaction of buyers and sellers all trying to maximize their own net benefit. The typical development patterns employed in the costs of sprawl are at least partly the product of such market forces. It is likely that postwar patterns of low density sprawl on the urban fringe are the result of consumer demand. In other words, that low density residential development is income elastic (Dougharty, 1975 p. 7). Consumers choose to live at high densities only where land costs are very high. In central cities on the urban fringe, land costs are lower and thus permit low densities. The costs of sprawl artificially drives up land costs in high density prototypes to reflect this relationship. In fact, suburban land is
available at lower costs.
2. Abundant evidence indicates that suburban officials and voters are
strongly opposed to population growth and multifamily development, particularly in high-rise structures. The urban fringe around central cities appears to be strongly fortified with exclusionary land use controls and growth management programs.
In general the nature of the problem of the cost of sprawl can be investigated from different points of view and in the following we do this from two different dimensions.
There are those who argue that there is, in fact, a serious problem facing the United States in terms of the agricultural resource base because of sprawl; they use a variety of arguments to justify their position. The basic argument can be summarized as those relating to the following:^ 1) urban shadow effect, 2) quality of cropland converted, 3) location of prime cropland loss, and k) costs of farmland conversion.
The other dimension of sprawl can be seen in the congestion problem (Suburban MobilityThe Coming Transportation Crisis, C. Kenneth Orshi. Transportation Quarterly, April 1985). Today, some of the worst traffic snarls occur far from the central urban core: on circumferential highways, in suburban centers, and on approaches to suburban office parks. Congestion has lost its directional bias: people commuting from one suburb to another or driving from their suburban homes to a shopping center are just as likely to run into heavy traffic as are commuters on their way downtown. Suburban traffic congestion seems to be rapidly spreading in space and in time. Highway corridors in which traffic flowed smoothly only a few short years ago now seem hopelessly clogged. Bumper-to-bumper, stop-and-go traffic that used to occur
^Richard H. Jackson, Land Use in America, Chap. 8.
only during the morning and evening rush hours now continues all day. In Houston, the rush hour is said to last 10 hours a day. In Dallas certain portions of the freeway system remain congested with bumper-to-bumper traffic throughout the day. Endemic congestion affects large portions of the Los Angeles and Orange Counties, Houston, Dallas, Denver, northern New Jersey, Long Island and Washington D.C. In many of these areas, public opinion surveys indicate that traffic congestion has superseded crime, housing, and pollution as the number one concern.
The concern should not be taken too lightly in an economic sense either, since the time and opportunity could be lost and wasted heavily, in the traffic congestion. Therefore, the policy makers have to combat with this problem very seriously, as they do with any other economic problems that hamper growth and prosperity.
B: Thesis Goals
Within the context of sprawl development dilemma of how to control and limit sprawl development with improvement of quality of life, this thesis focuses on two significant objectives.
1. The study is primarily intended to identify the forces which are behind the suburbanization and to find out the degree of relative importance of these forces on decentralization. This has been achieved through different regression models as well as studying research done by others. Obviously the Denver metropolitan area has been chosen as the case study and is based upon purposes of this research. All explanatory variables that have been chosen will help to explain sprawl as a form of differentiated growth. We will study the roles some of these
variables play in encouraging suburbanization. The factors emphasized can be characterized as those which have contributed as centrifugal forces in our growth. Of these forces, various economic factors seem dominant. The nature of suburban development is profoundly affected by the interrelationships amongst our economic forces. The employment variable seems to have strong relationship to the creation of sprawl. We will observe and study the past and present patterns of employment changes.
2. Study and observation, on the large scale, leads us to study in the specific, including case study, and other related research. We then present general as well as specific policy recommendations.
Overall, this thesis is seeking both primary and proximate causes, since there is no single cause. Therefore, a relative solution to the overall problem requires knowledge of the many causative factors. In the meantime the study attempts to examine the sources of policy influence on urban land development by which we are able to sense different factors that shape land use in any particular area. It also gives a sense of the approach and analysis for the causal chain which eventually produces the pattern of land use and social environment.
In short, it can be said that the goal of this thesis is to present expedient strategies for limiting sprawl and promoting orderly development.
C: Thesis Organization
The main body of the thesis is organized as follows:
Chapter Two: Methodology. Lists the thesis hypotheses, addresses the
methodologies, literature, and sources of data; and finally some problems involved in this research.
Chapter Three: Description of the Case Study Area and Identification of
the Constituent Suburbs. Chapter Three is trying to present a description of the case study area. It starts with the Denver metropolitan setting. In this section we review very briefly some general information about the Denver metropolitan area from the point of its evolution structure. The attempt of the next section is to identify 31 cities of study by the function that they perform. The last section focuses on the degree of physical development of each city based on two assumptions.
We have tried to show a visual representation of the features of different cities' relative population densities,
i.e., RPD's. These features would help us in studying the patterns of physical development.
Chapter Four: Case Study: Denver metropolitan area. This chapter is
the core of the thesis with consideration of key variables. The next section introduces each variable under study and its sources. Then the results of some regression models which were fitted to case study variables will be shown and analyzed. Here is presented a comparison between linear and nonlinear function forms. The result demonstrates the relative priority of the various variables in explaining the overall decentralization. By using the criteria two factors emerge as having inordinate importance: population change and employment change.
Chapter Five: Analysis of the Determinants of Suburbanization. Chapter Five is based on two points: 1) study of different aspects of key variables solely or jointly; 2) schematic representation of interaction among these variables. The first factor which will be discussed is the value of land and later, respectively, the issues of amenity, shopping centers and finally population and employment consideration. So, the main focus in this issue is population and employment growth. Finally a model is reported which explains most of the key variable relationships.
Chapter Six: Conclusion and Policy Recommendations. The last chapter consists of policy recommendations. These address three considerations: 1) renewal and rehabilitation of central cities and the suburbs which are in our advanced stage of development; 2) preservation of agricultural land; 3) prevention of unplanned sprawl development. Finally, in the end of this chapter policy recommendations for the Denver metropolitan area are presented as well as for the
rest of the variables.
Chapter II Methodology
A. Thesis Hypotheses
The thesis examines six hypotheses, addressing the process of urban sprawl in the Denver metropolitan area. The hypotheses are:
1. Sprawl activity is a function of rent or price of land. People or business operations go outside the central city because the value of land or rent is cheaper. Alonso's argument that the price of rent trades off against the greater transportation cost associated with commuting to the CBD is not applicable here because of the following reason:
During the last decade there has been a proportional shift from industrial jobs to service and knowledge base jobs, which are more decentralized, and can function independently of CBD. Therefore, the transportation cost to CBD has become a less serious concern. Also from residential consideration, there are enough facilities (shopping center, grocery store, movie, barber, etc.) outside the CBD to provide for all needs.
2. The residential activities or employee activities go outside the central city because of: development within the central city is generally restricted to small areas and perhaps parcels are not contiguous, while development outside the central city has less
constraint for space.
3. The physical distance of each city to the closest major highway could be another force for residential activities as well as employee.
4. Shopping centers provide many different goods and services that are conveniently and attractively available to consumers. Therefore, accessibility to shopping centers could be noted as one of the factors which contribute to the degree of attractiveness of suburbs.
5. People who are richer can afford to live outside the cities, therefore, those may cause decentralization and sprawl.
6. People are concerned about the amenity level of the place that they choose for their activities (outside the central city we are faced with less pollution, less crowded, less crime, more spaces, etc.). Therefore, this can be one of the reasons that they move outside of the city.
B. Methodologies and Literature Sources
Here methodologies, general conceptural literature and sources of data are addressed. Referred to often, is the general body of literature on suburbanization and sprawl. The source of data for each variable has been noted in Chapter IV (Case Study) after the description of the specific variables. Also a step by step explanation of the methodologies to develop a different regression model is discussed in this chapter. For an indication of the scope of overall literature see the annotated bibliography prepared by this author. (M. Pourfayaz, May 1988)
C. Problems Involved
The main part of the problem that this study was facing, was the gathering of disaggregate data for each suburb. The difficulty arose when the approach was made to specialized departments at the city administration level, although in most cases, the city's officials were cooperative. Nevertheless, the lack of detailed statistics was an obstacle to research at times. Plus, the discrepancies of information and various resources made room for personal judgements and adjustment. For example, such a case was the disaggregate household income for Denver metropolitan area reported by the Denver Regional Council of Governments (DRCOG) which was often different from the census of population and housing information supplied by the U.S. Department of Commerces, the Bureau of Census, for recent years.
DESCRIPTION OF THE CASE STUDY AREA
Description of the Case Study Area "Identification of the Constituent Suburbs"
The Metropolitan Setting: The first settlement of the region was at the confluence of Cherry Creek and the Platte River in 1858. It was to be called Denver, later to be known as the Queen City of the Plains. It started and remains as the center for retail and wholesale trade in the Rocky Mountain region.
The city and county of Denver, a single urban government, was created in 1904 with the consolidation of the city of Denver and six contiguous townsArgo, Berkely, Elyria, Globeville, Montclair, and Valverde. This new Denver incorporated an area of 58.75 square miles with a population of some 140,000. This jurisdiction, for all practical purposes, excluding education, served as a unitary metropolitan government for the area until World War II.
From World War II to 1975 the city and county of Denver changed from the metropolitan unit to one of a number of government units. The Denver metropolitan area is now (1987) populated by 1,876,500 people (based on DRCOG). This amount, based on DRCOG's estimation, would increase to 2,340,800 in the year 2000. The Denver metropolitan area contains six county areas of Denver, Adams, Arapahoe, Jefferson, Boulder and Douglas. This region is the major commercial center with a diversified economic base. New "bedroom communities" sprang up around the city. The increased population necessitated new development in areas peripheral to the urban core. The
overriding characteristics of the metropolitan area today are its low density sprawl and leap frog development pattern.
The Denver metropolitan area, with the city of Denver, contains ^1 municipalities. The author, for research purposes, confined the study to 31 cities. This was because of nonavailability of necessary information for other cities.
The cities which are under study can be seen in Map A and are as follows: Arvada, Aurora, Boulder, Brighton, Broomfield, Castle Rock, Cherry Hills, Commerce City, Edgewater, Englewood, Evergreen, Federal Heights, Glendale, Golden, Greenwood, Highlands Ranch, Lafayette, Littleton, Longmont, Louisville, Lakewood, Lyons, Morrison, Mountain View, Nederland, Northglenn, Parker, Sheridan, Thornton, Westminster, Wheat Ridge.
The concern of this chapter is the study of these 31 cities from two points of view: 1) to identify these suburbs by the function that they perform and 2) to find out the degree of physical development of each suburb.
Two Suburban Perspectives A: Functional View of Suburbs:
Since suburbia is not a homogeneous, undifferentiated lot, there are fundamental differences among suburban communities in their function. Planners of suburbs have noted that in general there are two major types of suburbs, which can be identified by the functions of how they perform. These two types of suburban communities are: the employing satellite, or suburbs of production, and the residential suburbs; in other words, the suburbs which are employee oriented and the suburbs which are residentially oriented. The suburbs or cities which are employee oriented, usually have residents of their own but their principal function is that of providing goods or services for non-
local consumption. A good example of this kind of suburb in the Denver metropolitan area is Greenwood Village by its high office activities. The cities or suburbs which are residentially oriented are communities of home owners and, in some cases, renters. Economic activities found in them are mainly retail trade and personal services for the residents; the total character is residential. A good example of this kind of city in the Denver metropolitan area is the city of Lafayette.
To distinguish empirically between residential and employment cities in the Denver metropolitan area we compare the pattern of aggregate employment in the city to the number of total residents who live in that city. In other words the E/R ratio is a quantitative indicator that can show the nature of function in the suburb or city. This ratio was used for 31 cities under study (for years 1980 and 1983). The results appear in Table 3.1 and Table 3.2. The ratios would be compared with regional average points for the metropolitan area in 1980 and 1985, respectively (regional average points in 1980 and 1985 were respectively .53 and .56)^ For example, in 1980 if one of the ratios is greater than .53 the related city is more employment oriented, otherwise residentially oriented.
These numbers were given by DRCOG.
Table 3.1 The Ratio of: Total Employee/Total Population in Each City in 1980
Cities Ratios Cities Ratios
1 Arvada .17 17- Lafayette .13
2 Aurora .29 18- Littleton .47
3 Boulder .65 19- Longmont .35
4 Brighton .32 20- Louisville 1.38
5 Broomfield .29 21- Lakewood .48
6 Castle Rock .39 22- Lyons .23
7 Cherry Hills .23 23- Morrison .42
8 Commerce City .67 24- Mountain View .34
9 Edgewater .27 25- Nederland .02
10- Englewood .77 26- Northglenn .20
11- Evergreen .47 27- Parker .02
12- Federal Heights .20 28- Sheridan .53
13- Glendale 2.6 29- Thornton .19
14- Golden .59 30- Westminster .30
15- Greenwood 2.4 31- Wheat Ridge .41
16- Highlands Ranch .003
As it is presented in the table among the wide range of ratios, we have two extremes: the city of Highlands Ranch with a ratio of .003 and the city of Glendale with a ratio of 2.6. It should be mentioned that the city of Highlands Ranch in 1980 was in the very first stage of development and during
that time the total number of employees were only a few who were working for the city (according to a telephone conversation with a city employee). Also a similar situation existed for the city of Parker with a ratio of .02. So overall those ratios show how much those two communities were residentially oriented in 1980. In fact they are represented as a limited residential area (small village). As for the city of Glendale with the highest ratio of 2.6, the city was already employee-oriented in 1980. As a whole, in 1980 eight cities could be considered as employee centered or employee oriented: Glendale, Greenwood, Louisville, Englewood, Commerce City, Boulder, Golden, Sheridan; the rest were residentially oriented.
In comparison with 1985, the following ratio changes have occurred
Table 3.2 The Ratio of: Total Employee/Total Population in Each City in 1985
Cities Ratios Cities Ratios
1 Arvada .28 17- Lafayette .21
2 Aurora .45 18- Littleton .81
3 Boulder 1.19 19- Longmont .66
4 Brighton .51 20- Louisville 2.06
5 Broomfield .49 21- Lakewood .82
6 Castle Rock .57 22- Lyons .37
7 Cherry Hills .46 23- Morrison 1.17
8 Commerce City 1.35 24- Mountain View .58
9 Edgewater .51 25- Nederland .22
10- Englewood 1.42 26- Northglenn .34
11- Evergreen .25 27- Parker .45
12- Federal Heights .35 28- Sheridan .76
13- Glendale 4.93 29- Thornton .30
14- Golden .73 30- Westminster .45
15- Greenwood 3.53 31- Wheat Ridge .88
16- Highlands Ranch .24
In this case Glendale, with a ratio of 4.93, is still at the top from an employee oriented point of view. But the cities of Highlands Ranch and Parker gave their ranks to the city of Lafayette and Nederland with ratios respectively .21 and .22. As it appears in Table 3.2, in 1985, the number of cities which were employee oriented had increased from eight to fourteen. This indicates
that during 1980-85 the tendency was to strike a balance between the growth of major employment centers and residential areas developing in all parts of the metro area. Our point here could be supported by a brief extract of the Buyer's Guide, published by the Denver Chamber of Commerce, 1986-87. We present a summary of the report of the major corridor in the metropolitan area in the following. As can be seen in these pages, the employment base is shifting out along each corridor.
West 1-70 and C-470: This area includes most of Jefferson County and a number of suburban communities including Arvada, Lakewood, Golden and Wheat Ridge.
Major employers include:
Company Product Denver Employment
1. *Cobe Laboratories medical equipment 1,100
2. *Adolph Coors Company malt beverages, food preparation, containers ceramic parts 9,700
3. Denver Federal Center complex of offices for various federal agencies 8,000
4. Manville Corp. corporate headquarters for forest, roofing, pipe products 1,400
5. Martin Marietta/ Denver Aerospace aerospace and defense related research and production 13,900
Headquarters Major business parks include:
Business Park Location Acreage
1. Academy Park Hampden & Wadsworth 250
2. Ken-Caryl Business Ctr. Kipling & Chatfield 245
3. Denver West Office Park 1-70 Sc W. Colfax 600
4. Foothills Business Park Hwy. 93 & 6th Ave. 125
5. Genessee Business Ctr. 1-70 at Exit 254 120
South 1-25 and C-470: This corridor has the largest concentration of office space outside the central business district. It includes parts of Arapahoe and Douglas counties with the communities of Littleton, Greenwood Village, Highlands Ranch, Castle Rock and part of Denver. A large part of the growth in this corridor is taking place in unincorporated area.
Major employers include:
1. Diners Club credit card processing 1,300
2. Great Western Life life insurance, U.S.
Assurance Company Headquarters 1,000
3. Honeywell Control
and Test scientific apparatus
Instrument Division and instruments 1,070
4. Information Handling indexed technical
Services information 750
5. U.S. West communications and related
services, corporate headquarters 250
Major business parks include:
Business Park Location Acreage
1. Denver Technological Center 1-25 5c 1-225 776
2. Dove Valley Business Park Dry Creek Road 5c Peoria 1,000
3. Greenwood Plaza 1-25 5c Quebec 363
4. Highlands Ranch C-470 5c Broadway 930
5. Inverness Business Park 1-25 5c County Line Road 980
6* Meridian International Business Center 1-25 5c C-470 1,158
7. South Park Mineral Ave. 5c Broadway 267
East 1-70 and 1-225: This corridor has the heaviest concentration of warehouse/distribution facilities in the metro area, as well as a mixture of other commercial users. It includes parts of Adams, Arapahoe and Denver counties and the cities of Aurora, Commerce City and part of Denver.
Major employers include:
Company Product Denver Employment
1. AT&T communications equipment
and services 9,000
2. Associated Grocers grocery distributor
of Colorado 750
3. Samsonite Corp. luggage 1,400
4. Stapleton International Airport 15,000
United Airlines 5,000
Frontier Airlines 3,400
Continental Airlines 2,800
This area also has three major military installations which are large civilian employers:
Buckley Air Base 1,400 employees Fitzsimons Army Medical Center 2,850 employees Lowry Air Force Base and Technical Training Center 4,800 employees
Major business parks include:
Business Park Location Acreage
1. Aurora Business Center 1-70 & Tower Road 900
2. Aurora Centre Tech 6th & Chambers 360
3. Centre Point Alameda & Chambers 112
4. Denver Business Center 51st & Havana 256
5. Irondale Industrial Park 1-76 & 88th Ave. 230
6. Stapleton Industrial
Park 1-270 & Quebec 186
7. Upland I, II, III 1-70 & Chambers 326
1-70 & Tower Road 284
Boulder Turnpike and North 1-25: This corridor has the highest concentration of "advanced technology" companies in the metropolitan Denver area. In particular, Boulder county has the largest number of these firms in the state of Colorado. The corridor includes parts of Adams, Boulder and Jefferson Counties and the communities of Boulder, Broomfield, Longmont, Louisville, Northglenn, Thornton and Westminster.
Major employers include:
Company Product Denver Employment
1. Auto-trol Technology computer-aided
Corporation design systems 500
2. Ball Corporation aerospace research and
development, containers 2,800
3. IBM computer printers and
copying equipment 6,000
4. *NBI, Inc. word processors and office
automation systems 1,900
5. Rockwell Int'l. weapons components research
and production 6,500
6. Storage Technology
Corporation computer memory systems 4,200
7. Sundstrand Corp. aerospace components 1,200
Major business parks include:
Business Park Location Acreage
1. Colorado Tech. Ctr. 104th 5c Dillon Rd.,
2. Flatirons Industrial 55th 5c Central Ave.,
Park Boulder 250
3. Interlocken Bus. Ctr. U.S. 36 5c Colo. 121 570
4. Longs Peak Industrial
Park Longmont 105
5. Northglenn Industrial
Park Irma Drive 5c E. 104th Ave. 100
6. Park Centre Huron 5c 120th Ave. 200
7. Westech Broomfield 410
B: Stages of Development
Since there is not any precise criteria to measure the degree of physical development of each city, planners often are looking to find a way that could express the relative degree of physical development of cities. Sometimes this strive is based just on subjective judgement. However in our case, in order to make a fair judgement we have to make two assumptions, and based on those restrictions we study the pattern of physical development in the Denver metropolitan area.
1. We are assuming that Denver, as a city, has been a relatively developed area.
2. We consider heavily populated areas as developed cities,
so we are using the criteria of the population density as the main element.
Our purpose for doing this is based on our belief that: every kind of residential or non-residential activity occupies some space in buildings themselves. Therefore, a person either in his residential or work place would be occupying a certain area of building space and these buildings could be considered as a sign of physical development. Here it should be mentioned that in employee activities, we are not concerned with the industrial employees since their activities occupy some areas which do not often have building values (at least in aesthetic character). We are, rather, concerned more about employee activities which use office space, including other employee activities such as shopping centers.
Therefore, it was better if it was possible instead of: total population/total square feet of area, we could have used total population + total nonindustrial employees/total square feet of area.
Although in this case we could consider some building space for
residential as well as nonresidential purposes. But because of non-availabilities of appropriate means for assessment, reluctantly we have relied upon population density.
There has been a comparison made between population density of each city to that of Denver (relative population density = RPD). We have tried to show a visual representation of different cities' RPD's by putting different ratios on x-axis and the frequencies on y-axis. This way we can have some of the features of different ratios (histogram and frequency polygon). These features could be helpful for study of pattern of physical development and interpretation.
Note: it should be mentioned that we are aware that the following problems have negative impact on the validity of our methodology.
1. Some cities may be highly employee oriented and may be physically developed, but they are not residentially oriented. Consequently, they are not heavily populated. Apparently they belong in the categories of cities which are in the early stage of development.
2. The other problem that exists here is the average population density for each city. Sometimes there is a lot of vacant land within the cities with low or zero population, as compared to parts of cities with heavily populated areas. Therefore, this does not represent real distribution of population density.
3. Some cities might lose or gain more population because
of the economic situation, and this makes a more difficult condition for interpretation. t
4. Finally, annexation is the other problem which creates a
disturbance in interpretation.
Despite all of these problems, we have used the above procedure. The results appear in the following diagram.
Several points could be drawn from the foregoing study:
1. The frequency polygon in 1985 with regard to the frequency polygon in 1980 and 1975 has more moderate fluctuations. This means that overall in 1985, as compared with 1980 and 1975, we have a better distribution of population in the Denver metropolitan area. Also from a development point of view the cities are more similar to each other than in 1980 and 1975.
2. Despite the fact that three frequency distributions in
three periods are skewed to the right, as times proceed the mode goes toward the right. In other words, the cities which are in different stages of development have a tendency toward Denver (with RPD=1).
3. Those cities which have their ratio located between .9
and 1.1 are considered as being similar to the situation of Denver. In 1975 we had two cities in the Denver category: Arvada and Glendale; in 1980 there were five cities: Arvada, Boulder, Englewood, Longmont and Northglenn; in 1985 there were three cities: Arvada, Englewood, and Northglenn.
4. In general we divide the range of RPD into three categories:
A: from .05 to .2 B: from .2 to .7 C: .7
We consider the cities of category A to be in the early stages of development. A good example of this category in 1975 is the city of Lafayette with a ratio of .18 which moved to
category B in 1980 and remained in the same category in 1985 with a ratio respectively of A2 and .55. The cities of category B are those cities that have some degree of physical development. A good example of this category is the city of Aurora which in 1975 and 1980 had, respectively, ratios of .52 and .62 and was in category B, but in the year of 1985 with RPD Jk transferred to category C. Finally those cities which have a ratio greater than .7 are considered as relatively developed cities.
5. In 1975 six cities that represented highest frequency
(mode) were in the early stage of development (RPD between .1 and .2); they include: Greenwood Village, Lafayette, Nederland, Castle Rock, Cherry Hills and Evergreen. In 1980 the majority of the cities (RPD between A and .5) were Golden, Lafayette, Thornton, Westminster, and Sheridan. In 1985 the majority of cities (RPD between .5 and .6) were Greenwood Village, Littleton, Sheridan, Thornton and Westminster.
6. In all three periods the city of Parker, with the lowest
ratio, was at the extreme left side, and the city of Edgewater, with the highest ratio, was at the extreme right of diagram.
By studying the pattern of the data for different variables, the following points could be stated. (The following points were confirmed with correlation matrix that comes in the next chapter.)
A. There is a negative relationship between the distance of
each city to CBD and the degree of physical development. As we go farther from CBD, obviously the physical development
of other cities diminishes. The cities of Englewood and Parker could be sited as good examples.
B. The median monthly rent (DRCOG, 1970-80 General Population, Housing and Socio-Economic Characteristic) and consequently the price of land of cities which are in the second stage of development is more expensive than for the cities which are in the early stage and third stage of development. This might be because the cities which are in the early stage of development cannot provide enough services for their residents and the cities which are in the third stage of development are crowded with high pollution. Therefore, the cities in the second stage of development are more attractive.
C. Based on median household income (DRCOG, 1970-80 General Population, Housing and Socio-Economic Characteristic) there is no relationship between the degree of physical development and median household income of each city, but in 1980 there is a negative relationship between the degree of physical development and median household income of each city. This might show the point that in 1980 by increasing population in Colorado, especially in the Denver metropolitan area people with higher income desired to live in
less crowded areas.
CASE STUDY "DENVER METROPOLITAN AREA"
Case Study "Denver Metropolitan Area"
The main purpose of this chapter is to prioritize the relative importance of the various factors which determine the location of employment and residential development in the Denver metropolitan area. These insights will be used in order to establish policy recommendations regarding suburban development.
The other concern of this chapter is to investigate and interpret the existence of high and moderate correlations^ among some explanatory variables.
It should be reminded that the primary intention of this study is to identify the forces which are behind suburbanization and centrifugal growth; "we are seeking primary and proximate causes of this growth." The variables which were considered in this study are based upon accepted research in conjunction with thesis objectives and are composed of various social, economic and geographic characterics of the region (Denver metropolitan area).
Despite the fact that the theory could justify more than ten independent variables in explaining employment growth and population growth,
^There is considered to be a moderate correlation between two variables when: .5 ^r <.8 and a high correiation when rS.8.
the study is restricted to ten explanatory variables for the following reasons:
(1) One of our main concerns was with the degree of freedom. With the number of observations (31 cities) being constant, if we had increased the number of independent variables, it would have led to a decrease of freedom and, consequently, hamper the statistical results. (2) Further, regression models with a limited number of independent variables are easier to analyze and understand. (3) The presence of many highly intercorrelated independent variables (i.e., the number of employees and labor force) may add little to the predictive power of the model while detracting from its descriptive abilities and increasing the problem of round off errors.^
Variables and Sources The variables which were utilized are:
1 - EMPC: as dependent and independent variables. These variables were measured by employment change for each city between the period of 1975 and 1980 (period A) and the period of 1980 and 1985 (period B). Sources: 1975-80 Employment Estimates by Industrial, DRCOG and 1980-83 Employment Estimates by Industry, DRCOG.
2 - POPC: population change as explanatory and dependent variables for each city in periods A and B. Sources: 1985
The degree of freedom associated with a calculated statistic is the number of available observations minus the number of constraints placed on the data by the calculation or in other words: d.f = # of observations -the number of parameters (each independent variable creates one parameter).
This discussion can be found in many statistic books.
^3ohn Neter, "Applied Linear Statistical Models," (Homewood, Illinois, 1974), p. 374.
(Municipality Population and Household Estimate, DRCOG; 1970-80 General Population, Housing and Socio-economic Characteristics, DRCOG.
3 - VLAND; total vacant lands which have potential use for development in the future (based on years 1975 and 1980). Measure: square feet. Source: personal interview with each city government. Date: during May 1987.
4 - RENT: one of the predominant variables that can be considered as an explanatory variable for our study, is an index that shows the average rent for each city. Since it was impossible to gather this index for the Denver metropolitan area, I was obligated to use the median monthly rent for each city (in 1975 and 1980), which has almost the same reflection in equation form. Measure: $ Source: 1970-80 General Population, Housing and Socio-economic Characteristics, DRCOG, January 1983.
5 - T.CBD: A table was provided by DRCOG which included a total of 1,334 different unit zones. For each unit zone the
time distance to CBD (in 1980) was calculated.^ These units were transferred to the 31 cities, and an average time to CBD from each city during peak hours on week days was calculated.
Based on DRCOG's assumption, the T.CBD for 1980 was multiplied by a ratio of 9/10. The result gave a time distance to CBD in 1975 (in 1975 there was less congestion as compared to 1980). Measure: minute.
6 - T.WORK: included the average minutes for all workers living in 16th St. and California St. are assumed as the center of CBD.
each city to travel to their work place (in 1980). Again, to compute T.WORK for 1975, this variable was multiplied by a ratio of 9/10. Measure: minute. Source: Foresight Data, Inc.
7 - DFREEW: This variable is the physical distance between the approximate center of each city to the nearest major highway (1975 and 1980). Measure: mile. Source: usage of maps.
8 - INCOME: Median Household Income (1975, 1980). Measure: $ Source: 1970-80 General Population, Housing and Socioeconomic Characteristics, DRCOG.
9 - RPD: This independent variable, which represents the relative
population density, has been calculated as follows: (a) The
population of Denver (in 1975 and 1980) was divided by its total land area (in 1975 and 1980). The result gives the population density for Denver, respectively in 1975 and 1980. (b) Using this same method the population density for each city in 1975 and 1980 was calculated, (c) The ratio of population density for each city was divided by the population density of Denver. This could be considered as a measurement of relative population density. It is obvious that if this ratio (RPD) is greater than one, the given city is more congested than Denver and vice versa if the ratio is smaller than one.
10 - SHOP: This variable was calculated by subtracting the total leasable square footage (in 1975 and 1980) from the square square footage currently available (in 1975 and 1980). The result gives the amount of space occupied by shopping centers for each city in 1975 and 1980. The cities that did not have any shopping cehters gained a zero value for the SHOP variable. Measure: square feet. Source:
Denver Business Journal, March 30, 1987.
11 - AMEN: The amenity variable shows the degree of convenience and desirability for each city and was measured by:
1. summing up the following expenditure for each city:
A judicial public safety, which includes:
Aj: law enforcement
B public works, which includes:
Bj: roads and highways
B2: solid waste service
C culture-recreation, which includes:
2. After calculating A+B+C for each city (in 1975 and 1980),
the result was divided by the total population of each study city (in 1975 and 1980) in order to get amenity per capita for 1975 and 1980. Measure: $/person. Source: 1975 and 1980 Local Government Financial Compendium, State of Colorado. Before we consider the multivariate relationship, we would rather pay our attention to partial relationships, hence focusing on the correlation matrix.
The correlation matrix for period A and B that have been shown in Tables 4.1 and 4.2 can be useful to some degree in giving us a picture of casual relationship. The importance of these tables could be more clear if we are
POPC VLAND RENT TCBD TWORK DFREEW AMEN RPD INCOME SHOP EMPC
POPC 1.00 .85 .21 -.02 -.05 -.10 -.01 -.13 .06 .26 .72
VLAND .85 1.00 .18 -.03 .04 -.02 .004 -.24 .03 .36 .62
RENT .21 .18 1.00 -.09 .002 .14 .11 -.15 .60 .22 .42
TCBD -.02 -.03 -.09 1.00 .76 .54 -.32 -.41 -.29 -.31 -.14
TWORK -.05 .04 .002 .76 1.00 .52 -.51 -.53 -.16 -.15 -.20
DFREEW -.10 -.02 .14 .54 .52 1.00 -.19 -.36 -.07 -.26 -.20
AMEN -.01 .004 .11 -.32 -.51 -.19 1.00 .16 -.005 .20 .22
RPD -.13 -.24 -.15 -.41 -.53 -.36 .16 1.00 -.18 .35 .22
INCOME .06 .03 .60 -.29 -.16 -.07 -.005 -.18 1.00 .07 .15
SHOP .26 .36 .22 -.31 -.15 -.26 .20 .35 .07 1.00 .55
EMPC .72 .62 .42 -.14 -.20 -.20 .22 .22 .15 .55 1.00
Table 4.1 Correlation Matrix for Period A
POPC VLAND RENT TCBD TWORK DFREEW AMEN RPD INCOME SHOP EMPC
POPC 1.00 .86 .06 -.06 -.02 -.10 .05 -.05 -.01 .64 .75
VLAND .86 1.00 .07 -.03 .03 .01 .000 -.22 -.07 .65 .59
RENT .06 .07 1.00 .04 .02 .21 .15 -.36 .59 -.003 .03
TCBD -.06 -.03 .04 1.00 .75 .53 -.32 -.42 -.12 -.28 -.04
TWORK -.02 .03 .02 .75 1.00 .52 -.51 -.59 .06 -.11 -.21
DFREEW -.10 .01 .21 .53 .52 1.00 -.19 -.43 .06 -.25 -.19
AMEN .05 .000 .15 -.32 -.51 -.19 1.00 .22 -.06 .14 .20
RPD -.05 -.22 -.36 -.42 -.59 -.43 .22 1.00 -.41 .17 .11
INCOME -.01 -.07 .59 -.12 .06 .06 -.06 -.41 1.00 -.05 -.10
SHOP .64 .65 -.003 -.28 -.11 -.25 .14 .17 -.05 1.00 .55
EMPC .75 .59 .03 -3- o 1 -.21 -.19 .20 .11 -.10 .55 1.00
Table 4.2 Correlation Matrix for Period B
aware of each element of these tables present the degree of association between two variables, while controlling the affect of other variables without manipulating the raw data.
Since Table 4.2 is for period B (1980-85), we pay closer attention to
this table for interpretive point of view. Then if there is any significant change between corresponding elements of Table 4.1 as compared to Table 4.2, we try to explain why. Finally we should say that by considering the formula
we are concerned about those r's that are significant also that have a value greater than .5 by using this formula.
1. POPC and VLAND = .86
The high correlation that exists between POPC and VLAND could be due to two factors: (1) there may be some part of collinearity, depending on sample error, since most of city organizations gave me data about the vacant land according to estimation and guesses. So the results might have been either too high or low rather than actual figures; (2) Even if the estimates given were exactly correct, we should expect a high correlation between POPC and VLAND since people for, residential purposes, are looking at places which have vacant land opportunities for different uses, especially for abiding
r = simple correlation
d.f = degree of freedom
mean. The amount of vacant land that I received was the total
amount of vacant land which had potential use for commercial, industrial and residential purposes. However, if the amount of vacant lands which was given was just for residential purposes, we should expect an even higher correlation.
2. POPC and SHOP = .64
The moderate correlation of POPC and SHOPPING center may be due to the fact that people want to choose their home near places that have full accessibility to shopping centers or the reverse. It could be said also that: shopping centers absorb many employees, and if these employees could be given choices, they would rather live closest to their work places. This is more clear when we look at the period A, because there are less shopping centers than period B according to our low r.
3. POPC and EMPC = .73
The relationship between EMPC and POPC is very clear. These two factors have a mutual relationship and can be cause and affect of each other. Very briefly we can say if people were given the choice, they would try to have the shortest distance between the place that they live and the place that they work.
4. VLAND and SHOP = .63
Between 1981 and 1983 several shopping centers were built in the Denver metropolitan area. This shows that in 1981 there was a rather high demand for vacant lands to fulfill shopping center building purposes. This was not true in 1976, and hence we do not see a high correlation for period A. *
EMPC and VLAND = .39
This correlation can be justified in such a way that by increasing in office activities and other employment activities, the vacant lands receives very important attention. In other words, employees need some space in which they designate for their work.
6. RENT and INCOME = .59
The interpretation of this correlation is obvious: always some part of the income goes for the rent. Even if we consider the RENT variable as representative variable instead of price of the land, we can say that everyone who has more income can afford to have more expensive property for residential or other purposes.
7. T.CBD and T.WORK = .75
This could be due to: (1) those who used the route to go to a work place which is located in CBD; (2) those who use some part of the route to get to work as well as to reach the CBD.
8. DFREEW and T.WORK = .52
9. DFREEW and T.CBD = .53
The second reason that we used for justifying collinearity between T.CBD and T.WORK, can be also applied for (8) and (9).
10. AMEN and T.WORK =-.51
First of all, they have opposite direction. This means that if time to work would increase, amenity for related city would decrease since people have to spend more time. Therefore the opposite direction of these two factors is evident. On the other hand, the way we measured the amenity, is that a large
portion of expenditure goes to highway section. It is clear that this amount for highway can have an important impact on changing the time to work. For instance: with good facilities, maintenance, and enough land on a highway, we would have a better level of service and, consequently, would spend less time in traveling to work or home.
11. RPD and T.WORK = -.59
RPD, as we mentioned, is relative congestion of each city to Denver. As this ratio decreases, it means the city under study, as compared with Denver, has less population density. On the other hand, conceptually we know that usually people have more desire to live in the places close to highways and freeways in each city in order to make the shortest possible daily trips. This means that the places close to highways are more congested than the places far away from the highways. Consequently, the places that represent less congestion have lower RPD and people have to travel a longer period of time to reach to highways, to go to their work places (except for those who work close to their residential area).
12. EMPC and SHOP = .55
This correlation could indicate that a significant part of total employees were working in shopping centers in period B, due to increase in number of shopping centers at Denver metropolitan area.
To develop an understanding of the empirical relationship we
employed various regression models. In order to find the best behavioral equation which would describe the response of planning agents to employment change and population change, different functional forms (linear, polynomial, semilog and logarithm) were examined. Some of these studies, which seem to have more interesting results, will be shown in this chapter or Appendix A. It should be mentioned that different functional forms (equation models) were tested in two different intervals [ (1 (1975-1980) 2 (1980-1985)1 This has been done in order to find the relationship that exists between different variables and dependent variables in one interval of time. Can it be supported by previous intervals of time, or have there been any changes in previous intervals of time? In the event of changes, how can the changes be justified?
As has been previously stated, the sprawl development phenomenon is for two purposes: (1) residential activities and (2) employment activities. These two factors have a strong relationship to each other, and it is difficult to say which one is cause and which one affect; both can serve as cause and affect. Hence, population change, which was used as an explanatory variable, has also been treated as a dependent variable which follows employment change. In this way there can be a better appreciation of cause and affect.
The first functional form which was considered for hypothesizing the ten independent variables as influences upon employment change, had a linear relationship, with this general form:
y = a + bixi + + biox10
The question raised here is whether this equation has an intercept or goes through the origin. In order to answer this question, it would be useful to
consider the theory of central place functions'7 very briefly.
Figure 4.1 Number of Functions
According to this theory: (1) there is always a positive relationship between the size of population and the number of functions (Figure A); (2) a nonlinear relationship, such as the one shown in Figure B, might be expected between the number of functional units and population size. This suggests that as population size increases, the number of functional units increases fairly rapidly up to a point but at a slower rate thereafter. This would be consistent with the notion that in the larger centers economies of size become feasible thereby allowing larger operating units to be substituted for several smaller ones. Studies have
^Leslie J. King, "Central Place Theory," McMaster University. Beverly Hills: Sage Publications, pp. 20-25.
confirmed that this relationship between size and the number of functional units is also a strong, positive one. This relationship has provided a basis for estimating the threshold level for a function. The threshold level is defined as the minimum level of support, as measured by numbers of population, required to support a function in a particular community.
As can be seen in both Figures A and B, there is an intersection between population axis and curves so the curves cannot go through the origin. A similar logic can be applied when, instead of population size, the number of employees is considered. In this respect, (1) there is a positive relationship between the number of employees and the number of functions; (2) there is a positive relationship between the number of function units; (3) there is always a minimum level of employees (threshold) needed to support a function in a particular community.
When dealing with population changes or employment changes, (instead of population and employment) the following reasons for having intercept could be presented.
a. The central place functions can be applied for population and employment changes. However, in population and employment cases the intercept is only positive, but for population changes and employment changes, there could be a negative intercept as well.
b. Population growth or employment growth are dynamic processes, so they can never remain constant. For example, population changes are due to two factors: (1) changes in natural increases, (2) migration. This implies that population and employment growth cannot go through the origin.
c. When population change or employment change is a dependent
variable, there are some unknown or unpredictable factors in the equation that can be explained by error terms. These error terms can be justified as intercept or some part of intercept,
d. Based on intuitive logic, it could be said that there is no place in this world that could have a "zero degree" of activity when there is land available for potential use of various activities or for residential purposes, even if all explanatory variables in our equation are zero, (especially important variables like RENT, SHOP, etc.) Based on this theory, the point can be supported that our functional forms must have intercept. Therefore, every functional form in this study concerning employment change or population change would be based on an intercept formula.
The first technique for analysis was multiple regression. The
statistical results for both periods were poor, as shown below.
For Period B (1980-85):
EMPC = 10543.79 + .41 POPC .58 T.WORK + .45 T.CBD + .34 SHOP .44 VLAND (1.48) (3.4) (-2.2) (2.13) (1.69) (-1.36)
-.21 RPD .09 INCOME .04 DFREEW .01 AMEN .003 RENT Eq. (1)
(-.92) (-.52) (-.23) (-.08)
MR = .84 R~2 = .57 F10 = 4.92
R2 = .71 SE = 2898.67 20
For Period A (1975-80):
EMPC = 9340.37 + .56 POPC + .02 T.WORK + .20 T.CBD + .18 SHOP + .11 VLAND (2.21) (2.72) (.13) (1.11) (1.22) (.24)
+.35 RPD + .05 INCOME .09 DFREEW + .16 AMEN + .28 RENT Eq. (2)
(2.18) (.33) (.74) (1.25) (1.96)
MR = .89 F~2 = .69 F10 = 7.7
R2 = .79 SE = 1988.91 20
The numbers in parentheses, beneath the parameter estimate, are t- statistic.
The main reason for unacceptable results is due to four possibilities: (1) poor data, (2) strong collinearity among some explanatory variables, (3) the lack of linear relationship between dependent variables and independent variables, (4) it could be a combination of three problems or some of those. For possibility number 1 we do not have any control, and neither are we able to increase the number of observations (31 cities). So, we should be more concerned about three other possibilities in this study.
Based on Tables 4.1 and 4.2 (correlation tables for periods B and A), there is serious muiti-collinearity in equations (1) and (2). The other symptoms of this multi-collinearity can be seen in low + values and high standard errors. These characters can lead to serious problems and ambiguities for our analysis. Let's see what happens to our estimation. Consider equation (1). We know that a variable coefficient, for instance T.CBD's coefficient, can be interpreted as the affect of one unit change in T.CBD on the EMPC variable (if nine other explanatory variables remain constant). But here the variation in T.CBD is consistently related to T.WORK and DFREEW. Consequently the resulting variation in EMPC cannot be accurately attributed to the T.CBD source. In other words the associated variables of employment change are too unreliable to be used for analysis.
One of the ways that we can mitigate the multi-collinearity problem is to drop one or several independent variables from our equation. But the question that still remains here is, which set of independent variables should be kept and which set should be excluded? In order to solve this question we employed the stepwise regression technique. This technique is different from the multiple regression or technique in that the first independent variable to be entered accounts for the largest proportion of the total variance in the dependent variable and so on. The criterion for adding or deleting an
independent variable can be stated equivalently in terms of the sum of squares reduction, coefficient of partial correlation or F statistics.
A low F-value of inclusion was purposely selected for the regression to provide a picture of the relative order of all variables considered. But the statistical criteria did not allow for choosing more than four independent variables for both periods. The statistical results can be seen in Tables 4.3 and 4.4.
Table 4.3 For Period B
Variable B B SE B (t)
1 -POPC .62 .28 .07 4.21
2 T.WORK -.50 -249.56 85.71 -2.91
3 T.CBD .44 100.89 41.41 2.44
4 SHOP .22 .01 .01 1.43
6077.13 2263.89 2.68
R2 = .67 R'2 = .62
MR = .82 SE = 2679.13
F^ = 13.75 26
Table 4.4 For Period A
Variable B B SE B (t)
1 -POPC .64 .28 .04 6.16
2 RENT .28 26.70 9.92 2.69
3 RPD .28 1901.34 752.41 2.53
4 SHOP .22 .02 .01 1.94
-6171.17 2290.07 -2.69
R2 = .76 MR = .87
R-2 = .72
SE = 1896.04
F^ = 20.19
Despite the statistical results acquired with the stepwise technique, it presents a better outcome for both periods as compared to the multiple linear regressions technique. But the conclusions were not as expected, especially for period B.
For period B we are still faced with a multi-collinearity problem. There is also high standard error. In the case of A, even though the statistical results, as compared with B, are better and we are not faced with a multicollinearity problem, the standard error is relatively high. Therefore, the existence of linear relationship is questionable. It is wise to take a close look at the other alternative "nonlinearity relationship between dependent variable and independent variables."
It is better, for the moment, to stop here and assume that some linear relationship exists and, based on this assumption, make some interpretation of the results. Even though it could have been proven certainly that linearity is rejectable, the results that have been received from linearity could represent some reflection of reality. The results are invalid for prediction purposes but could be useful to some degree for explanatory purposes (overall).
By studying and comparing the results of two different periods the following points are concluded:
A. POPC and SHOP in both periods are considered as the explanatory variables which could justify some part of employment change. However, because of low t-value for shopping center, particularly in period B, you cannot rely as much on that as an explanatory variable. It is possible that low
t-value is due to multi-collinearity in period B.
B. Having T.WORK and T.CBD as explanatory variables in period B is trying to demonstrate the point of how much worse the traffic congestion is in period B than in period A. Because of congestion and traffic jams employees in period B are more concerned about the time that they should spend for business and work purposes. Therefore, these two mentioned variables attract more attention in our results.
C. One of the main reasons for formless urban sprawl from an employment point of view: it is cheaper for businesses to operate in those areas than T.CBD. While our results in period B the proxy variable RENT which was used instead of price of land, does not imply any role in growth for jobs. For example: many buildings in the downtown area, up to six to seven years ago, were occupied for different office activities; today they are empty. This shows that since 1980, overall, there has been no demand for office activities. Consequently, the price for land or rent cannot play an important role in the case of the Denver city and metropolitan area.
D. Even in period A, the RENT variable can be considered as an explanatory variable due to its positive correlation with EMPC; however, this does not have logical direction. Employment activities move outside CBD because of price of land or rent is cheaper than inside CBD. Therefore, there is an opposite direction. Although this is true for most cities like Los Angeles, New York, etc., for the particular case of Denver this logic does not hold true. For the Denver metropolitan area we
can say that rent or price of land follows the employment changes. In other words rent or price of land depends on demand for space for employment activities. The question may be raised that for period A as employment changes were between 1975-80, while the rent variable was considered just for year 1975. If rent was supposed to follow the employment changes, it needed time to justify itself, since this happened in the period of 1975-80. The answer to this question is: when a plan (growth in employment activities) for development is revealed on a specific area for future, the demand of the given land would show promising and would rise, so the rent or price of land in 1975 begin to upward.
"Some of the following could be helpful for justifying C and D
1. In period B because of the national recession and bad economic situation in Colorado in particular, logically we could not have the normal pattern of growth for jobs. For example: many buildings in the downtown area up to six to seven years ago were occupied for different office activities, while today they are empty. This shows that since 1980, overall, there was no demand for office activities. Consequently, price of land or rent cannot have an important role in the case of Denver city and metropolitan area.
2. Based on the data that we received from DRCOG^, the average median rent for suburbs, as compared with the city of
1970-80 General Population, Housing and Socio-economic Characteristics.
Denver, shows that in 1976 the median rent for Denver was $185. while the median rent for suburbs was $219. Therefore, we are led to believe that the price of land in the Denver area is cheaper than its suburbs, which is completely different than other U.S. major cities.
Of course, we should not ignore the point that most of the cities that we have studied, like Lakewood, Glendale, etc., were not at the first stage of development. Some of them already have passed through different stages of development. Therefore, as we have seen, the prices of land in those cities is even are more expensive than in Denver city. Maybe it would have been better, if we had chosen some areas that were in the first stage of development to be compared with Denver city. However, since gathering data for these areas would be beyond the time span of this research, we had no choice but to restrict ourselves to current situations.
So far what we have covered was based on some degree of linearity between dependent and explanatory variables. Let us see the pattern of dependent variable with each independent variable in both periods and find out whether the overall pattern of the scatter plot shows any particular relationship or not.
As we can see in pages 5^, 55 and 56, the patterns indicate an irregular nonlinearity and, in the case of nonlinearity, even the points do not show any distinct pattern. Nevertheless, we have studied several different nonlinear alternatives by using the stepwise regression technique in the hopes of being able to present a more acceptable result.
A. The first case that we considered was a polynomial equation of
W,:tLU IUJTCLO UJEO.U
EMPC AND POPC
-( 16000+ I I I + 1 I I I
I 3000+ 1 1 I +
I 1 1 I
I 1 I
I 1 1 2 I
I 1 4 1 1 I
0 + A31 +
31 cases plotted.
EMPC AND POPC
!500 + I I I
1 1 1 1 1131 1 0+ C 51 I 1 I
+ +---H-----+----+ -
31 cases plotted
1 I I I + I I I I + I I
EMPC WITH SHOP
I 1 I
I 1 1 I
3000 + +
I 1 1 I
I 1 I
I 1 1 1 1 I
122 1 1 1 I
0+0 2 1 +
+ -I +
0 70000 1400C0
EMPC AND AMEN
I 1 I
I 1 1 I
3000 + +
I 1 1 I
I 1 I
I 1 1 1 1 I
I 24 1 I
175 350 525 700
EMPC WITH SHOP
I 1 I
E 12500+ 1 f-
M I I
P I 1 1 I
r I 1 1 I
12 1 1 1 1 I
0 + 942 1 1 1 +
I 1 I
60000 1 300 00
0 120000 240000
EMPC AND AMEN
I 1 I
r 12500+ 1 4-
M I i
P I 1 1 i
C I I 1 I
13 11 1 I
0+5552 1 b
I t i
EMPC WITH TCDD
1 1 I
I 1 1 I
I 1 1 I
I 1 I
I 1 1 1 1 I
I 111121 I
0 + 1 2 112 1 2 1 1 2 +
17 5 35 52 5 70
EMPC WITH TCBD
1 l I
500 + l +
I i l 1
I i l I
I linn I
0 + 121313 12 1 1 2 +
I 1 I
+ + + 20 60
EMPC AND RENT
I 1 I
E I I
M 7 1 1 1 1
P 8000+ 4
C I 1 1 I
I 1 I
I 1 1 1 1 I
I 1 11 1 21 I
0+321 33 11 +
1 BO 270
EMPC AND VLAND
i 1 I
E 12500+ 1 +
M 1 I
P I 1 1 I
r I 2 I
121 1 1 1 I
0+942 1 1 1 4
I 1 I
0 7000 14000
-4- + - EMPC AND RENT
E i i 12500+ 1
r, p T 7 1 1
C 7 i 1 1
I Q-f- 11211 232135 1 1
i i 4 1
160 240 320 400
EMPC AND VLAND
I 1 I
E 1 1
M I 1 1 I
F 8000+ +
C I 1 1 I
I 1 I
1 1 C- 1 I
I 12 1 1 1 1 I
0 + 831 1 1 +
0 7000 14000
EMPC AND INCOME
+ + - 1 l i + l l 1 i + 1 1 1 1 + l
I 1 I
E 1 I
M I 1 1 I
P 8000+ +
r 1 1 1 I
I 1 I
I 121 I
I 1 222 I
0 + 3 261 1 1 +
+ + -
0 3500 7000
EMPC AND RPD
I 1 I
E I I
M I 1 1 I
P 8000+ +
r I 1 f 1 I
I 1 I
7 i 1 1 1 1 I
I 1 1 2 1 1 1 I
0+1A11 3 2 1 1 +
+ + -
. 4 5 1 35
0 . A ) 1. G
EMPC AND DFREEW -*---+----+----+'
I 1 I 1
11; i a;
i i i
0+0 23 1 2
2. 25 6 75
A. 5 9
EMPC AND INCOME 1
I 1 1 I 1 1 112 21 0+ 633211 I 1 I
1500 A 500
+ I I 1 + I I I 1
1 + 1 1
EMPC AND RPD
I 1 I
E 12500+ 1 +
M 1 1
P I 1 1 I
C I 1 1 I
I 1 12 11 I
0+12 313 2 13 1 1 +
1 1 I
+ H - + +
375 1. 125
0 .75 c. U
EMPC AND DFREEW
E 12500+1 +
M I I
F I 2 I
I 1 1 1231 0+6 25 I 1 I +
+ 6. 75 A. 5
1 + I I
EMPC AND TWORK
I 1 I
E I I
M 1 1 1 I
P 8000+ +
C I 1 1 I
I 1 I
I 21 1 I
I 1 1212 I
0 + 1 11 14 31 2 +
EMPC AND TWORK
I i I
E 12500+ i +
M I I
P I i i I
C I -> C- I
I 1 2 3 I
0 + 1 11153 121 2 +
I 1 I
12. 5 n c cl u 37. 5 50 62 5
degree 2, which we intended to hypothesize:
1) The interaction between T.CBD and T.WORK. Since they are strongly related to each other, it was worth it to see their joint effect even though the results were not acceptable.
2) The interaction of rent and INCOME* ^he rat^
INCOME can a measure ability to pay rent with respect to income. The results can be seen in Appendix A by equations (1) and (2). The results show that in period B the interaction affect just for rent/income was to some t-level significant. But overall it did not have any superiority to the linear case, so it was not considered. For period A, the results were exactly the same as the results received from linear case.
B. To further our analysis, logarithmic forms were tested. This
time the results were statistically significant, but we could not completely rely on results for interpretation. The difficulty is due to our dependent and four independent variables. Since in some observations, these variables had zero or negative number. On the other hand, the argument for the natural log function cannot be less than or equal to zero. So the computer has to define these cases as missing value and ignore them (equation A^ and A^ in Appendix A). So in this case we lose
Anthony Chandor, The Penguin Dictionary of Computers, Second Edition, p. 36.
some degrees of freedom and this seriously hampers interpretation. For example, in equation our observations from 31 cities decreased to just 17 cities. The other
alternative is that we make some adjustment for those variables which have some negative or zero numbers. For example, the city of Englewood (based on DRCOG's data*) had the biggest loss in population during 1980-85, among the cities of Denver metropolitan area, namely a loss of 2,052. In order to have the smallest population change (1) in the city of Englewood, we should add 2,053 to each city population (-2,052 + 2,053 = 1). It is clear that if we do this kind of adjustment to five different variables in our study, we have been manipulating the variable values to achieve intelligible results Despite the criticism that has been mentioned for logarithmic forms, in the process of decision making about the relative importance of explanatory variables, great concern has been given to equation 5 and 6. In the following we study equation 5 and 6 very briefly. Here we assumed that: if from a theoretical point of view, a variable "x" can explain variable "y," then there is a reasonable justification that logarithm (x) can explain logarithm (y) and vice versa [ if L(x) can explain L(y), then x can explain y ] .
1. In both periods employment changes have been explained by the explanatory variables VLAND, RPD, RENT and POPC.
*1985 Municipality Population and Household Estimate.
2. These periods are differentiated for this variable "SHOP" as follows:
a. In period A, as compared with period B, people had less access to private cars. Therefore, the accessibility to different places, such as shopping centers, is more restricted. Thus, people were more concerned with this variable in period A as compared to period B.
b. After 1980, we see the development of new shopping centers and the expansion of some existing shopping centers. Thus, in period B the new abundance of shopping centers, combined with their readily accessible locations makes the variable "SHOP" less significant than in the previous period.
c. In period B the variable amenity (consisting of such services as recreation, parks, libraries, law enforcement, road systems, fire protection, and water disposal) has a greater linkage to employment changes. New employment centers developed in areas sensitive to these amenities, thus luring new employers and their employees. Competition amongst municipalities often was based, it appears, on their ability to provide such amenities.
3. The functional form which was considered here was
This can have two forms:
1) Ly = oc + bjXj + b]oxlO
2) y = L( oc + t)|X| + + bjQXjq)
We tested both forms and the results are shown in equation A^, Ag, and Ajq in Appendix A (we are trying to show the cases where the computer assumed zero or negative numbers as missing values, manipulated cases were not put in to use here). Also, we did not accept this functional form,
1) for the same reason as the case of logarithm,
2) due to a low or relatively low R
Whatever we have done for EMPC, we have tried also for POPC, as dependent variable in order to have a better scrutinizing of cause and affect. Because of we do not want to make this chapter longer, two equations that we conceived for linearity cases with stepwise technique, are shown in Appendix A (eq. Ajj and eq. A^)*
In all cases regardless of whether EMPC was dependent variable or POPC, period A (1976-80) shows better results. This may be due to: (1) a factor that was not important in period A, but played a vital role in period B in explaining dependent variables and since we ignored the variable, that reflected the consequences in period B; (2) as we mentioned earlier, there was a better economics situation in period A than period B, and consequently period A had more regulated employment and residential activities than period B; (3) despite the fact that data in both periods were from the same sources, it seems the data for period A were more precise. More data for variables in period A are completely accessible and kept on file in contrast to period B. For example, the data for number of employees which we received from DRCOG was just for
1980-83. So for the interval of 1980-85 I had to multiply this figure by .
Also other estimation and manipulating at this period have been used.
EVALUATION AND CONCLUSION
The nature of the criteria that we have used in this research (correlation matrix, linearity and nonlinearity) are such that, unfortunately, we cannot rely upon each of these criteria completely for making a decision because of:
1. Correlation matrix is a criteria that usually cannot show the whole picture of causal relationship precisely, especially in our case that more or less we are faced with the same problem of multicolinearity as we are faced in linearity.
2. The problems that we faced with linearity (especially collinearity) has been discussed enough earlier.
3. For our third criteria, as we have seen already, in the several nonlinearity cases that were tested, only logarithmic forms had some degree of validity. Of course every phenomenon of this world, including our case, can be fitted with an exact polynomial equation. But the question is, with our limitation of time and energy is it worth it to follow this equation any further? For better results we could have taken into consideration cases with different degrees of polynomial. Obviously it is easy to imagine how much frustration and time consumption would result. So by accepting this fact that this is "EXPRESSION OF WHAT EXISTS" then we should convince ourselves to look at the criteria positively and make compromises among them for final judgement about the relative importance of each variable. By focusing equal attention to each of these three criteria the
policy for consideration is:
a. Considering the correlation matrix, which shows the degree of association between each pair of variables and especially paying more attention to correlation between dependent and each independent variable.
b. Considering linear relationship between dependent and independent variables can be, to some degree, useful (at least from decision making and explanatory points of view, if not for prediction purposes).
c. Furthermore, among nonlinearity cases to pay special concern to logarithmic forms.
Although for this research we are not able to precisely identify which variables can be used to explain decentralization and sprawl growth, by considering the above policy, both "EMPC" and "POPC" did indeed interact in a circular fashion. Other variables which help to explain the phenomenon can be categorized in three groups. These three groups are determined by the following process. In a table, reasonable specific values were assigned to each of the nine independent variables according to their rank for one criterium for period A. Then, for the other two criteria the variables were similarly assigned values according to their rank. The assigned values of each variable for each criteria were totaled, resulting in a score for each variable. The process was repeated for period B. The scores of each variable from period A were added to the scores from period B to arrive at final scores. Then on the basis of the final scores, we categorized them in three groups:
A. The variables which had an important role:
1. VACANT LAND
B. The variables which had a relatively important role: it. SHOP
C. The variables which did not show a significant role:
Therefore, it is recommended for better policy making in the future that the main emphasis should be focused upon variables in priority group A, as well as B.
ANALYSIS OF THE DETERMINANTS OF SUBURBANIZATION
Analysis of the Determinants of Suburbanization
The variables that we studied in the last chapter are somehow related to one another. In other words, they can be put in a framework to show diagrammatically the logical relationship among them. Thus, at this stage we consider two points:
1. to study different aspects of key variables solely or jointly,
2. a pictorial representation of variables which can be considered as a conceptual model.
Therefore, the thrust of this chapter, would be around these two
The Value of Land
The price of land or rent is one of those forces that plays a tremendous role in explaining sprawl development. The validity of this point has been proven by our case study also. The level of significance of this variable here will necessitate a deeper study of this regard.
Since the supply of land is fixed, land value is determined by demand for space. In general the value of land can be explained by a very simple
theoretical model. ^
Vi = f (Hi, Ti, Ai, Pi)
where: i = is index and indicate industrial, commercial, or residential
V = the value of a particular urban area site H = certain historical factors T = topography A = amenities
P = accessibility to economic activities Each of these factors has a different degree of importance in describing the value of land.
Hi In most areas that have started sprawl development recently
the historical factor is zero and therefore the key variable reduces to three factors.
_P Site's accessibility is a joint function of its proximity to
different activities and the transportation system connecting it with these possible destinations.
A_ The amenities characteristic could be determined by
social and atmospheric conditions associated with the site.
_T Topography means the natural physical characteristics of a
particular site. Foremost among these features are slope, elevation, and soil condition. Topography has two separate impacts on measured land values.
A site's terrain has an affect on its amenity index. For example, a site situated high up on a steep hill might be above the smog level and might
^Bourne, Larry S. 1971. Internal Structure of the City, Eugene F. Brigham, The Determinants of Residential Land Values. New York, Oxford University Press, pp. 160-162.
have a commodity view of the city; both factors would tend to increase its amenity level. (In our case study "Denver metropolitan area" mountain view is very important from an amenity point of view. On the other hand, this steep hilly site might be excessively windy and create negative amenities for people who dislike wind or who have small children). Thus topography's affect on a site's amenity ranking would vary from family to family depending on their composition and preferences. Besides its amenity impact, topography also has an influence on land values through its affect on development cost.
Amenity, in general, refers to the pleasantness of the urban environment as a place in which to live, work, and spend one's leisure time. This variable does play a very important role in explaining sprawl development phenomenon. Despite the fact that in our case study amenity apparently did not show a significant role, this variable even in the Denver metropolitan case, could be considered very important for two reasons.
1. As an element of the public interest, amenity is closely
associated with convenience. Indeed amenity and convenience are frequently linked together and since convenience could be seen in terms of time/distance we can say that part of amenity level could be interpreted in time to work and time to CBD. We have seen that these two variables, especially time to work showed very significant role or in other words we can claim that despite, it was tried to limit the degree of multicollinearity in this case study but because of high interaction among some variables, still we can see the reflection of amenity in other
2. The amenity level for each city is a qualification factor that
should be determined subjectively by different individual's values, but since this was not possible for us, we tried to convert that in a quantitative framework in order to have a tangible value.
Shopping centers could provide "crystallization points for suburbia's community life" by affording opportunities for social life and recreation in a protected pedestrian environment, by incorporating civic and educational facilities. Therefore urban theorists began to conceive of the shopping center as a vehicle for social and civic reform distressed by the formless outward spread of the city, they viewed the planned development of shopping centers as an antidote.
Although shopping centers varied greatly in the number and types of amenities and conveniences they afforded, they quickly attracted attention for their potential contributions to the civic and commercial vitality of the suburbs. So, some planners believe that no construction is more dynamic than the shopping center or as likely to influence a reform of the usual urban and suburban hodge podge ... a new generation of development store men and women are showing the same high responsibility that their grandfathers showed when they helped to create the great downtowns which we know today in hundreds of cities.
In general, the attractiveness of a shopping center refers to the location of the center, its size and overall character of its business provisions. It is difficult to define more specifically the attractiveness of shopping centers, since attractiveness depends on additional considerations to do with the
amenities found and the imagery involved and these are not easily translated into quantitative terms. The main weakness common to all indices of shopping center attractiveness lies in their inability to properly distinguish between the portions of trade which accrues from the local population against that which is drawn from people in surrounding areas.
Overall, for establishing a shopping center a careful and all around study from supply side (locational characteristic of shopping center) as well as demand side (consumer behavioral patterns) should be done.
Population and Employment Consideration
Due to interdependency between population growth and employment growth, it is difficult to determine which can gain more weight for explaining decentralization or urban sprawl (except since last decade that urban village phenomenon happened for employee activities and especially office activities). Therefore, these two variables can be studied simultaneously, just for a better identification of one can present a better control of the other.
The Spatial Distribution of Employment: area employment changes result
from three basic forces:
1. Given labor supply and demand conditions, if labor markets are out of equilibrium, then wage changes that result from equilibrating market forces may result in employment changes.
2. Given labor supply curves that are not perfectly inelastic, changes in labor demand can cause changes in employment. Changes in labor demand can be caused by a number of factors, such as change in the national economy, changes in area income,
Michael 3. Greenwood, "Population Redistribution and Employment Policy."
Manuscript, pp. 115-117.
population changes brought on by natural increase or net migration, changes in population composition, and changes in consumer preferences.
3. Given labor demand curbs that are not perfectly inelastic,
changes in labor supply can cause change in employment. Changes in labor supply can also result from a number of factors, including changes in the working-aged population brought on by natural change or by net migration and changes in labor-force participation rates that are independent of wage levels.
Although the three forces generally operate simultaneously in local economics that are growing or declining, the latter two are likely to be of primary importance. Each of these forces, however, unfolds in the context of the national economy, and therefore, regional and suburban changes in employment and population should not be viewed apart from their national setting. At the national setting we should be concerned about three factors: 1) the national rate of employment growth; 2) the age of labor force; 3) fertility patterns, family composition, and marriage rates. These three factors are, of course, not independent of one another or of other forces operating in the economy and in society in general. Furthermore for analysis purposes the relevance of each factor to changes in the spatial distribution of population must be considered seriously.
The Relationship Between Population and Employment Changes
Although population changes reflect corresponding employment changes, the extent to which population changes lead, lag, or occur simultaneously with employment changes is unknown. Quantitative analysis of
Denver metropolitan area's data suggests an unstable relationship between the variables POPC and EMPC over time as well as a relationship that differs considerably between suburbs, as it appears in Table 5.1 It should be mentioned that we selected those cities in the Denver metropolitan area for this study which showed significant population changes (greater than 500).
Table 5.1 The Ratio of Incremental Employment to Incremental Population
1 - Arvada .35 -.27
2 - Aurora .36 .35
3 - Boulder 2.53 2.97
4 - Broomfield .33 .49
5 - Englewood -.45 -.74
6 - Federal Heights .39 .47
7 - Golden 1.56 -.98
8 - Lafayette .02 .52
9 - Lakewood 1.00 1.01
10- Littleton 5.4 1.47
11- Longmont .34 1.74
12- Louisville 2.17 1.71
13- Northglenn -.22 -.21
14- Thornton .13 .17
15- Westminster .51 .19
16- Castle Rock .45 .43
This table indicates the ratio of incremental employment to incremental population, or what might be interpreted as the extra employment associated with one extra person. The most remarkable characteristic of the data is the wide range of values for each selected city in Denver metropolitan area. For example, the City of Golden varies from a relatively high of 1.56 extra jobs per extra person (1975-80) to a low of -.98 extra jobs per extra person (1980-85). These data suggest that extra population, alone, does not account for additional employment.
Among the other factors that contribute to city employment growth or that yield differentially high relationship between extra population and extra employment are the demographic and socio-economic composition of the population, the composition of employment, the labor-intensity of production process, incremental city income, incremental regional income, technological factor and national economy.
Employment and Population Policy
In the intraurban (metropolitan) contents, the distinction between population policy and employment policy are difficult. The difficulty is due to the interdependency between population growth and employment growth. In many instances the best way to influence employment growth is to influence population growth. A number of studies suggest that the primary cause of spatial employment redistribution had been spatial redistribution of market demand resulting from population redistribution. Reasons for this may be that higher income, better educated, and younger persons tend to have greater migration rates and that the local employment effects of migration tend to be especially sensitive to persons with these characteristics. On the other hand employment redistribution also causes population redistribution. Very briefly,
this is a mutual relationship.
Overall, here our concerns should be devoted to population and employment growth.
1. By using different policy would enable us to discourage and encourage migration as well as birth control.
2. To have some degree of controls on distribution of different economics activities in the regional and city levels.
Achievment of these two goals is parallel for controlling the sprawl development objectives or in other words the analogous is fully compatible with the concept of decentralization control. Therefore, it seems that each one of those can be supplementary to the other. Hence we focus on the explanatory variables which we have emphasized greatly in the last chapter.
The variables that we consider with regard to what we discussed earlier could be put in the following model (Figure 5.1). Further we should say that in our case study the place of the variables (T.CBD, T.WORK, and DFREEW) could be found in the infrastructure section.
A^: buildingshomes, school, shop, offices and factories
B^: transport facilitiesroads, railway lines, airports, ports, etc.
As we have seen, these three explanatory variables are a reflection of road situation. Therefore they should be studied in the infrastructure section.
CONCLUSION AND POLICY RECOMMENDATIONS
Conclusion and Policy Recommendations
In order to enhance the quality of life of those who rely on the land resources and while keeping the tranquility of highly prized natural environments, a set of devices or objectives should be designed in the direction of achieving this goal. One of the objectives that can serve this goal is to have some degree of control on urban sprawl. This can be achieved through the following considerations: (1) renewal and rehabitation of central cities and the suburbs which are in the third stage of development (i.e., Arvada in our case study); (2) preservation of agricultural land; (3) prevention of unplanned sprawl development.
Succeeding in the above suggestions could provide us with a fair balance between the growth of cities and suburbs. The attempt of this last chapter is to present a set of policies and recommendations that can be helpful for controlling the sprawl development. It should be noted that the author is not concerned with the different aspects of each policy, such as, the feasibility of the particular policies from economical, political, and other standpoints; this responsibility is left to the appropriate authorities.
Many of the policies within each consideration could also be employed in another. In other words, there are some overlap policies. In the following, we pursue each of the considerations by studying them in depth.
If we assume that growth of urban life in the future is inevitable, we should try, through effective devices and solutions, to change the direction of growth to some degree inward to the cities. This means revival and rehabilitation of many parts of the city as a means of utilizing existing resources. With this in mind hard decisions must be made concerning which land could or should be reused and how it should be used for the rebuilding of deteriorated areas. The policies than can be considered are: (1) increasing the rate on tax for idle lands and abandoned buildings in cities;* (2) more restrictions on land use control outside the cities as compared with inside. This means the lesser controlled area within the cities would become more attractive; (3) reformation of the Federal Tax Code: the nature of income tax must be somehow adjusted and, at the very least, the tax code should be neutral in regard to urban and suburban real estate development; (4) for residential and nonresidential activities inside the cities, especially for rehabilitation, the government should offer low-interest loans, interest subsidies, or guarantees to qualified firms located in urban areas where private lenders cannot meet business credit requirements; (5) tax base sharing in the metropolitan area. In general, sales tax and property tax in cities are higher than their suburbs. In order to adjust this difference, which is one of the reasons people to move to the suburbs, a portion (or all) of each jurisdiction's property tax base should be assigned to a pool that is then shared by every jurisdiction in the metropolitan area. The formula for sharing can vary, but essentially it is based on population and is adjusted to include property valuation compared to the metroplitan average per person. Under the adjustment, if a city is below the metropolitan
*When we say cities we mean central cities and those cities that are in third stage of development.
average, its government receives a larger share; if a city has property valuation above the metropolitan average, it receives a smaller share tax base sharing. This way each city would be free to set its own tax rate and hence its own level of expenditure, and this policy can decrease the need for relocation; (6) One of the reasons that suburbs are more attractive is because of less pollution, less noise and less crime as compared with cities. So as much as possible, different policies should make a better environment in cities in such a way that can compete with suburbs; (7) to aesthetically improve inside the cities, demolish some abandoned and uninhabitable structures and replace them with parks.
B. PRESERVATION OF AGRICULTURAL LAND
An important cost associated with loss of prime agricultural land is in the suburbanization process itself. Therefore, without regulations to protect those areas adjacent to cities that are more suitable for development, the waste and cost associated with sprawl will continue.
The general concern for agricultural land preservation can be
resolved through a variety of policies for its protection which are:
1. zoning ordinance resulting in exclusive agriculture zones
2. compensable regulations
3. development right easement
4. transferrable rights
5. land banking
6. preferential taxation
7. restrictions on state activities in agricultural districts
8. federal regulation and agencies
For more information refer to Richard H. Jackson, Land Use in America, New York, John Wiley & Sons, 1981, Ch. 8.
C. PREVENTION OF UNPLANNED SPRAWL DEVELOPMENT
Most of the recent research on urban sprawl has found that sprawl led to significantly higher price in the areas of land development costs, transportation costs, and social costs than did more carefully planned development with higher concentration of population. Therefore, we should try to plan for community development in advance, to prevent costly social disorder which is immensely important to every community environment. In short, planned development by definition creates a better environment and better residential services that enhance quality of life for achieving this objective. The following strategies can be used.
1. Land Banking. Communities purchase lands for urban facilities
such as fire and police stations in advance of their need. Banking can be used also to assemble the land needed for a park, or for a new urban community. In fact land banking has several advantages. First: promotion and encouragement of orderly growth and
development which is in harmony with the development plan for community. Second: minimizes the adverse affects of sprawl and unplanned growth. Third: reduces the price of land by controlling the land market and eliminating speculation. Fourth: it can preserve the prime agricultural land.
2. Much of the pressure for development in urban-fringe comes
as a result of public investment. Investment in infrastructure like highways, schools, etc., produces greater potential for sprawl. Therefore, for preventing unplanned development investment should be limited to only those types of programs which would not encourage sprawl. Very briefly, care should be taken with public
3. Growth strategies must be targeted to higher density
development in suburbs.
4. Focusing on transportation factor as catalyst of sprawl:
regardless of the fact that this factor was the main motivation for sprawl development, we should accept that the introduction of advanced transportation technology has allowed the urban area to expand and decentralize. This brief statement can show clearly the degree of importance that transportation has had in emergence of sprawl development. The significance of this factor is not neglected from our case study, and some how transportation factor show its decision reflection in our equations, by appearing in time to work variable in group A (the most important variables) and time to CBD in group B (the relative important policy variables). On the other hand one of the fundamental relationship in the study of transportation is the linkage between land use and transportation. Therefore, any kind of suggestion and recommendation about sprawl development that deals with transportation, should not be ignored from land use transportation interaction perspective.
LAND USE TRANSPORTATION INTERACTION3 We know that trip-making patterns, volumes, and model distributions are largely a function of the spatial distribution of land use. Likewise, the pattern of land use is influenced by the level of accessibility provided by the transportation system from one activity area to another. This land use-transportation relationship is presented by the diagram in Figure 6.1.
Michael D. Meyer and Eric D. Miller. 1984. Urban Transportation Planning, McGraw-Hill Book Company, p. 63.
Figure 6.1 Land Use Transportation Interaction
The development of land for a particular use results in the generation of new trips originating from that area or new trips attracted to that area, or both. The development of land in an urban area thus creates new travel demands and, consequently, a need for transportation facilities, whether in the form of new infrastructure or more efficient operation of existing facilities. Such improvements to the transportation system make the land more accessible to existing activity centers, thereby making it more desirable and affecting its monetary value. Increased accessibility and improved land values
in turn influence the locational decision of individuals and firms, once again spurring new land development and starting this cycle again, until an equilibrium is reached or until some other external factor intervenes. The time that it takes for the cycle to complete itself, in the short run, is the predominant influence of land use on the performance of the transportation system. For example, the impact of a new suburban shopping mall on the surrounding street system is likely to be severe, to the point of requiring major street and/or transit improvement. In the long run, the provision of
transportation infrastructure and the introduction of new technologies will influence urban form because of the improved accessibility that results. This impact obviously can be seen in the sprawl development phenomenon. One of the outstanding characteristics of this figure is that the public sector can intervene in the land development activities and, as a result, influence the process by different policies which encourage and discourage the developers such as: development fees, which developer is asked to share in the cost of area-wide improvements, making requirements more flexible or harder for developer, etc.
Briefly, the policies we have selected should, with public regulations, make the inside of cities more attractive through:
a. more flexible regulations inside of cities
b. more financial aid for building or rebuilding inside of the cities
c. tax rates becoming more uniform throughout the area
d. limiting public investment on outside of cities
POLICY RECOMMENDATION FOR DENVER METROPOLITAN AREA
The policies which have been recommended earlier are general and, more or less, could be applied in each part of the United States. In the following, we will be focusing on the policies which can work for the Denver metropolitan area.
Among eleven variables that we selected for this research, two (POPC and EMPC) showed a vital role in the development of urban sprawl. The two other variables, which demonstrated significant statistical results, thus influencing urban sprawl, are VLAND and RENT. Of course this does not mean that the rest of the variables are not important, but in our case study it would be more appropriate if we focus on these four variables. The particular geographic area will afffect the degree of relative importance of each variable. However, the variables EMPC and POPC have the greatest cause and effect relationship nationwide.
In the last chapter a section was devoted to employment and population policy. In the end of that chapter, a model (Figure 5.1) was presented. This model shows the interaction of key variables, including employment development and residential development. However, it would be very helpful and effective if any recommendation for POPC and EMPC would take account the above mentioned discussion, and particularly the model. Here is a brief recommendation for four of the most important variables in the Denver metropolitan area:
1. POPC: Limit population size (population cap), use a combination of the different tools which area available such as zoning restriction and the building permits issued each year.
2. EMPC: This is an exogenous factor and it depends on economic
We can have some degree of indirect control on EMPC by assuming that EMPC as dependent variable, not independent variable. This means that we should take a proactive role, to the extent possible, on factors which have an influence over EMPC. Overall control on POPC, whether viewed as dependent or independent variables is more manageable.
The data that developed as part of a DRCOG report titled Employment/Population Policy Forecast for the Denver Region 1980-2010 show the region's median age is climbing. This can be seen by comparing the two cohort survival models for 1980 and 2000 which are presented in Figure 6.2.
Distribution of Population by Age and Sex
75* - Females Males
70-74 - i rf
65-69 - i 1
60-64 r 1
55-59 - i 1
50-54 - ' 1 1
45-49- - ~l
Â§>-- _L er V'- |
35 39- | -vv: v* : i j
.Â§>3034- X .-* :
25-29- * 1
10-14- -i rgju r JV>, ?3 S/'V' |
5-9- > |
0-4- >Xv '. |
1 1 1 6 5 4 1 3 1 1 2 1 III III 1 2 3 4 5 6
Percent ol Population 2000
75* - Females
70-74 - i 1
65-69 - 1
60-64 - I
55-59 - | - , 1
45-49 ~ r 1
Â§ 40-44 - r ; - : -
CD 35-39 - | . 1
Â§>3034 - r-tSa-. .- '- : - - 1
25-29 - | ^ursti .
20-24 - -V-- - > s-~ - 1
15-19 - 1 V, ^ - ^ 1
10-14 - 1 - - 1
5-9 - 1 - 1
i i i i i i i i i i i i r
6 5 4 3 2 10 12 3 4 5 6
_____________________Percent ot Population
As it appears the majority of people in the year 2000 would be the age of 35-49. This group of people does not tend to be quite as mobile geographically. They tend to remain settled in their familiar environment. Therefore, there will be no major increase for new residential demand in the year 2000. Such demand which exists in 2000 can be satisfied by the amount of resident vacancy at this time. On the other hand, the DRCOG forecast showed that employment growth in the metro area over the next three decades is expected to occur at a rate of less than 3 percent annually, while at the present time the rate of office vacancy is 29.9%. The result is that this will not create a serious demand for new work space. In brief, POPC and EMPC as independent variables in the year 2000 cannot be considered significant as to need for new space and, therefore, will not encourage new sprawl. However, these variables as dependent variable may encourage sprawl in the year 2000. For example, many go outside Denver because the price of land or rent is cheaper in suburban neighborhoods than in Denver.
3. VLAND: This variable can be protected by different policies.
For example, government can buy land, can buy transfer development rights, zoning regulations, and develop taxing policy.
In our case study (Denver metropolitan area) we should make our zoning policies in Denver more creative and flexible, while at the same time making them more competitive with neighboring communities. This is to be done to offset the abundance of horizontal vacant land in the suburbs.
4. RENT or price of land: we can have influence by 1) zoning, 2) the form of tax policy.
The same creative policies which we advocated in the VLAND variable, have an impact on rents in Denver. With such adjustments, we will be able to maximize the beneficial uses of the land available to us in Denver,
while at the same time, this will make more attractive and competitive the price of land within Denver. Therefore, the demand outside of the city would decrease.
POLICY RECOMMENDATION FOR THE REST OF VARIABLES
5. AMEN: Public investment in infrastructure and also increased efforts in dealing with pollution, crime, and noise.
6. T.WORK, T.CBD, DFREEW: Investment in our infrastructure should be a shared cooperative effort among the various metropolitan jurisdictions which benefit either by their creation or the required maintenance involved. The funds created for these purposes should not be diverted for any other use. These funds could even by created by special user-fees or tollway policies.
7. SHOP: Development policies can encourage or discourage retail trade. Sales tax policies can have a similar effect.
8. RPD: By density control through existing government tools or new regulations.
9. INCOME: This is an exogenous factor and depends on the economic conditions at local, regional, and national levels.