CHAPTER SIXTEEN GWO-WEI TORNG A SPATIAL PERSPECTIVE OF THE INTERSTATE HIGHWAY SYSTEM'S EFFECTS ON SUBURBANIZATION Introduction The role of transportation systems in fostering growth and affecting land use structures has been of great interest to academics, planners, investors, and politicians. Peter O. Muller formulated his explanation of urban form in his article, "Transportation and Urban Form: Stages in the Spatial Evolution of the American Metropolis", by identifying four transportation eras (Figure 1). Figure 1 clearly illustrates the concept of dependency of land use on transportation system changes. Indeed, innovations in transportation technologies in the past have greatly influenced people's travel attitudes, travel behaviors, and then extended the boundaries of socio-economic activities, such as workplaces, residential areas, and shopping. While examining Figure 1, it is important to distinguish the growth effects from the distribution effects. In Figure 1, the distribution effects can be interpreted as the changing shapes of urban forms, while the growth effects are the net change of total areas within the urban boundaries. Figure 1, Intraurban Transport Eras and Metropolitan Growth Patterns: (I) Walking-Horsecar Era, (II) Electric Streetcar Era, (III) Recreational Auto Era, (IV) Freeway Era. Source: Peter O. Muller, Transportation and Urban Form: Stages in the Spatial Evolution of the American Metropolis.
The effects of transportation infrastructure on shaping urbanized areas have drawn significant attention in the past. The relationship between transportation and land use is interactive. Empirical urban development research reveals that land use patterns influence transportation investment decisions, which may later affect further land use changes. In this research, only the effects of highway/expressway systems on urbanized area growth will be explored. The primary intention in building interstate highways was for "interstate" transportation purposes, e.g., national defense and interstate commerce. Nevertheless, several major metropolitan transportation studies, e.g., a study done by the Chicago Area Transportation Study (CATS) in 1955, suggested that one purpose in developing the transportation networks was to serve travel generated by projected land-use expansions (Teaford, 1986). Demand-driven transportation planning schemes were very much a part of the thinking. Indeed, more than 30 years' experience tells us that those limited-access highways have very significant impacts at the local level, in addition to their impact on land use at the regional or even broader geographical scale. Rapid urban sprawl in the past several decades is perceived to be directly or indirectly induced by the emergence of highway systems. Because interstate highways are relatively "lineal" featured (one dimensional) with respect to overall urban areas, their effects should be considered in terms of directions as well as densities. This research examines the land use impact of interstate highway investments from a geographical perspective. It begins with a brief literature review which summarizes the objectives, motivation, methodology, and results of previous research on similar issues. In the second section, major research questions are illustrated. Next, the methodologies applied in this research, e.g., grid-cell based raster spatial analysis, are introduced followed by the development of models and discussions of modeling results. Then, the transition of spatial effects is analyzed based on the derived modeling results. In the final section, conclusions drawn from research findings are summarized and some policy implications derived from this study are discussed. Literature Review Muller (1986) pointed out that Americans were not urban dwellers by design. Many people live for a dream of owning a private, single family home, near-to-nature, rural, with huge front and back yards. As described by Schaeffer and Sclar (1975), "people have found a way to isolate themselves;... a way to privacy among their peer group;... They see nothing except what they are determined to see." For a majority of urban dwellers, living within large cities was actually not a personal preference but a personal choice given no other viable choices. Starting from 1945, after a long period of economic depression and war, the U.S. showed a great potential for economic revitalization. The automobile was no longer a luxury good but a necessity for commuting, shopping, socializing, and running personal errands. Meanwhile, America's birth rate rose, many young couples left crowded central city neighborhoods and moved to spacious suburbs where their kids could play in their own front or back yards. Two-car families became commonplace. In many aspects, the suburbanization appeared to be an inevitable tendency (Teaford, 1986). The passage of the Interstate Highway Act of 1956 signaled the arrival of the suburbanization era. The booming highway systems acted as catalyst which turned the historic metropolitan cities inside out and reshaped the traditional urbanized areas . This suburbanization phenomenon can be well explained by the increase in people's mobility. People can afford to live outside the central cities by traveling at a much higher travel speed on the newly constructed highways while not increasing travel time. Figure 2 illustrates the concept of the dynamic relationships between population density and urban area expansion from a spatial perspective. The resulting increase in mobility indeed reduced the relative attractiveness of old and crowded central cities. Figure 2, Type of Urban Area Expansion: (I) Pure Growth (increased population are even distributed), (II) No Growth, Redistribution (no change in total population), and (III) Growth + Redistribution (net gain in population and population redistribution) Meanwhile, the gasoline cost in the U.S. was low enough to support the highway commute choice. Anthony Down (1992) pointed out that the inflation adjusted price of unleaded gasoline actually went down by 27.4 percent from 1975 to 1987. The gasoline prices in other industrial countries in 1991 were at least 50 percent higher than those in the United States. In some areas, like Italy and Japan, the gasoline prices were three times or higher compared to the prices in the U.S. The relatively lower gasoline cost reduces the auto operating cost and thus indirectly encourages the urban decentralization. Nevertheless, it is not appropriate to attribute the only cause of suburbanization to the transportation systems. Bradbury, et al. (1982) concluded that several possible causes of urban decentralization exist. The results suggested that national economic conditions, technological change, and federal policies were also responsible for the urban decline in the past decades. There are some studies concerning highway impacts on land use changes. In the early interstate highway era, the majority of research, such as Adkins (1959), and Mohring (1961), tended to justify the positive effects of interstate highways through economic reasons, such as increase of land value. However, these studies failed to address the difference between growth and distribution effects. That is, whether the land value increases near the highway were simply a distributional shift or a net gain within the region as a whole was an issue. Both researches used the highway length as a indicator rather than using the accessibility measures of the highway interchange. It was recognized later that the highway interchange is a more meaningful and appropriate research unit when exploring highway systems' impact on land use. Finally, both researches were conducted in the late 1950s and early 1960s which might be too early to capture "time lag" effects of the highway investment. Thus an objective evaluation of the effects of these newly constructed highways was virtually impossible. A more recent and comprehensive highway impact studies was conducted by Payne-Maxie Consultants for U.S. Department of Transportation (DOT) and U.S. Department of Housing and Urban Development (HUD) in 1980. This study was designed specifically to explore the impact of beltways (circumferential limited-access highways) instead of overall interstate highways. It considered influential factors, such as city ages and highway location. In addition, this study applied the concept of interchange density using it as a measurement of highway accessibility. Most importantly, an effort was made to distinguish the distributional impact from the growth impact by controlling regional growth related variables. In other words, the study attempted to control for the regional changes and then identify the "pure" distributional impact. The results unveiled the fact that neither the existence of beltway itself nor the interchange density has consistently significant impact on population growth in SMSA during the study period (1960-1977). Regarding the spatial analysis of land use change literature, a recent work, BART AT 20: A Preliminary Look at the Economic, Property Market, and Land Use Impact of the Bay Area Rapid Transit System, by John D. Landis (1994), studied the impact of Bay Area Rapid Transit stations on surrounding area land use conditions and evolution. This study analyzed 1965-1990 land use changes at nine suburban BART stations. Grid-cell based analytical methods were used and probability models were developed in the study. The results suggest that the distance to the existing BART stations have significant influences on land use change. The farther a grid-cell is away from BART stations, the less likely this grid-cell would change its land use. In addition, adjacent land use type also stood out as a highly significant and consistently important variable, especially during the 1975-90 period. The positive coefficient for this adjacent land use variable indicates that land use of each individual grid-cell is strongly affected by the land use pattern of its neighboring, or adjacent grid-cells (the existence of the spatial neighboring effect). Research Questions As mentioned before, the main purpose of this study is to demonstrate the impact of interstate highway systems on population decentralization since the 1960s. In order to achieve this goal, the following three major research questions are proposed. 1. Spatial correlation: Are the expansions of urbanized areas spatially correlated with the development of interstate highway systems? That is, do the directions of population decentralization match the directions of highway system development? Figure 3 illustrates the spatial correlation effects. The inner shaded circle stands for the existing urbanized area before the highway systems were constructed. The thick black line represents a radial highway system which goes right through center city area. It is assumed that the net expansion of urbanized area are equal for both types of growth (A and B). Should there be no spatial correlation effect, type A expansion is expected. Otherwise, the new urbanized area would look like type B. 2. Spatial neighboring effects: Do spatial neighboring effects play a significant role in the expansions of urbanized areas? That is, all else being equal, do lands which are closer to the existing urbanized land have a higher chance of changing their urbanization status? 3. Time lag effects: Do time lag effects of highway systems on population suburbanization exist? With other conditions controlled, would a non-urbanized land cell which is closer to a younger highway interchange have a higher probability of turning into an urbanized land cell than that one whose closest highway interchange is older? Figure 3, The concept of Spatial Correlation Effects 4. Beltway effects: Do beltway systems have any effect on encouraging in-fill type of urbanized area expansion (type A in figure 3) rather than radial expansion along the highway corridors like type B in figure 2? 5. Transitions of spatial effects on suburbanization: Whether two major spatial effects, spatial correlation and spatial neighboring effects, are experiencing transitions in terms of their impacts on suburbanization. Methodology Case Study Area Selection: Oklahoma City, Oklahoma In this study, Oklahoma City, Oklahoma, is selected for longitudinal analysis over three decades (1960-1990) of interstate highway evolution and urbanized area growth. The following information, including some digital map files, about the case study area are collected for research purposes. 1. Urbanized area boundaries of case study area for year 1960, 1970, 1980, and 1990 2. Street center line maps for the case study area (digital format) 3. Interstate highway system evolution from 1960 through 1990 4. Interchange locations and their years of completion 5. Locations of undevelopable lands, e.g., rivers (digital format) 6. Other arbitrarily defined boundaries, e.g., city boundary and county boundary (digital format) 7. Classification of highway types, i.e., beltway vs. radial systems.
Figure 4 provides an overview of the evolution process of urbanized area of Oklahoma City from 1960 through 1990. In this study, spatial models are developed to distinguish the directional effects of the interstate highway system from its density effects on urban decentralization. GIS Analysis By using a GIS software (ARC/INFO for UNIX running on IBM RS6000 platform), this research develops several grid-cell based spatial analysis models to explore the spatial evolution process of Oklahoma City urbanized area and its relationships with interstate highway constructions between every observation time period (every 10 years) or every other observation time period (every 20 years) . In the following multivariate logit models, change of urbanization status (dichotomous variable) will be used as the dependent variable throughout all models. Each cell is treated as a research unit and each non-urbanized cell's probability of changing its urbanization status in the next decade is represented as a nonlinear function of its distance to the closest existing urbanized cell, its distance to the closest highway interchange, year of completion of the closest highway interchange, and type (beltway/radial) of the closest highway interchange. Cell size selection The urbanized area boundary coverages for Oklahoma City, Oklahoma are determined from census data issued every 10 years by the Census Bureau. Due to lack of precision of the hard copy maps in the census books, the source of the urbanized area data, cell size is set to equal one half square mile per cell (about 3734 feet on each side). In other words, this study employs neighborhood-scaled observations in the models.
The definition of study areas Empirical evidence shows that the absolute magnitude of urbanized area growth is actually a function of existing urbanized area's size. Larger urbanized areas tend to have larger expansion. Thus, a study area should be defined in a way which is sensitive to the area's size. It is clear that traditional methods for defining study areas, such as using politically defined boundaries, are not appropriate. Besides, the numbers of changed cells and non-changed cells should not be heavily weighted to either side in order to make the forecast model more meaningful. Therefore, only those cells which are perceived by authors to have a reasonable chance of changing their urbanization status are included in the study area. In this case study city, Oklahoma City, Oklahoma, the study area is defined as a 3-mile buffer of 1960 urbanized areas for 1960-1970 model, 3.5-mile buffer of 1970 urbanized area for 1970-1980 model and 1960-1980 model, and 4-mile buffer of 1980 urbanized area for 1980-1990 model as well as 1970-1990 model . Within each study area boundary, areas covered by water body, e.g., lakes and rivers, are regarded as undevelopable areas and thus are taken out of the sample set.
Logit Model Development Among others, there are three commonly used probability models: Linear Probability Model (LPM), Logit Model, and Probit Model. The difficulty with LPM is its linear assumption which may result in observations with probability greater than 1 or less than 0, a meaningless result in a probability model. Both logit model and probit model belong to the nonlinear probability model family with values always bounded by zero and one. Usually, both models have similar performance in terms of their prediction power and accuracy rate. However, the logit model is relatively easier to be interpreted. In this study, the logit model with dichotomous dependent variable (change of urbanization status during the study period, 0 = no changed, 1 = changed) is used with each grid-cell in the defined study area acting as a unique observation. The basic format for dichotomous logit model is as follows: P(Y = 1 | X) = exp( sum(bx))/[1 + exp( sum(bx))] where X are observed variables and the remaining unknowns are the parameters b.
Spatial correlation phenomenon It is commonly believed that the massive interstate highway system construction since the mid 1950s' is one of the major causes for the rapid growth of urbanized lands. The Euclidean distance of each cell's centroid to its closest highway interchange is measured to represent the actual distance between the two. This independent variable (DIST_INT) is expected to have significant negative coefficients throughout all multivariate logit models.
Spatial neighboring phenomenon By definition, the urbanized area is a cluster of land cells with their population density reaching a certain threshold. It is assumed that the presence of an urbanized area cell can enhance the chance of urbanization of its neighboring cells. Authors assume that the evolution of urbanized areas was affected by interstate highway systems but also by the spatial neighboring effect. The rationale for expecting the neighboring effect is land developers' avoidance of costly extension of urban infrastructure. Using the census definition of urbanized area as places with population densities greater than 1000 persons per square mile implies the need for water and sewer line infrastructures in the newly developed lands. Land developers are inclined to develop land near existing systems to avoid costly mainline extensions. In order to distinguish the interstate highway spatial correlation effect from spatial neighboring effect, the latter is controlled by adding an independent variable, DIST_UA which measures the Euclidean distance from each observed cell to its closest urbanized area cell during the previous decade. Consequently, significant and negative coefficients for this predictor (DIST_UA) are expected throughout all logit models. That means the farther a cell is away from existing urbanized cells, the less likely this cell will turn into an urbanized cell.
Time lag effects There remains an unknown issue regarding whether a time lag effect exists in terms of encouraging non-urbanized areas growing along highway corridors. Urbanized area's growth directions may not reflect the influences of nearby highway interchanges until much later than the interchange's completion year. In order to explore this issue, this research includes the closest highway interchange years of completion as another independent variables (INT_60, INT_70, and INT_80). These time lag variables appear as dummy variable formats on cell by cell base.
Beltway effects As mentioned, beltway systems are believed to play important roles in forming urbanized area boundaries. If a non-urbanized cell's nearest highway interchange is a beltway interchange, it is expected to have higher probability of changing its urbanization status (positive effect). In the following logit models, dummy variable (INT_BELT) is used to explore the beltway effect on urbanization.
Spatial Effects Transitional Analysis Based on logit modeling results, several curve fitting techniques, such as logistic curve fitting, are used to depict the historical trend and predict the future trend of spatial effects on suburbanization. The probability of changing urbanization status during different time periods are then discussed. The aggregate trend of urbanization area growth (without directional concerns) is used to adjust for the predicted results. Modeling Results Logit Modeling In this study, a framework of the binomial logit model for predicting the probabilities of changing urbanization status among non-urbanized lands is specified as follows:
P(UA_CHANGE=1 / DIST_INT, DIST_UA, INT_60, INT70, INT_80, INT_BELT) = exp(b0+b1*DIST_INT+b2*DIST_UA+b3*INT_60+b4*INT_70+b5*INT_80+b6*INT_BELT) / [ 1 + exp(b0+b1*DIST_INT+b2*DIST_UA+b3*INT_60+b4*INT_70+b5*INT_80+b6*INT_BELT)] where UA_CHANGE: dichotomous dependent variable, 1 = from non-urbanized area to urbanized area during study period, 0 = otherwise DIST_INT: distance to the closest highway interchange. DIST_UA: distance to the closest urbanized cell The following three dummy variables are used if applicable. INT_60: dummy variable for testing time lag effect. INT_60 = 1 if the nearest interchange was completed before 1960, otherwise INT_60 = 0 INT_70: 1 if the nearest interchange was completed between 1961-1970, 0 otherwise INT_80: similarly, 1 if the nearest interchange was completed between 1971-1980, 0 otherwise INT_BELT: 1 if the nearest interchange is part of the beltway system, 0 otherwise Three logit models by each decade within the study period (1960-1990) are established and summarized in table 1.
1960-1970 Model Figure 5 shows the urbanization area expansion between 1960-1970, highway systems as of 1960, and the study area (3 mile buffer of1960 urbanized area boundary) defined for developing logit model. For this 1960-1970 model, both DIST_INT and DIST_UA unveil significant explanatory power on UA_CHANGE. DIST_UA plays a more important role in predicting the change of urbanized status than DIST_INT. The negative coefficients of DIST_INT and DIST_UA support previous hypothesis which suggest that both the distance to nearest interchange and the distance to the nearest urbanized cell have negative effects on the probability of changing urbanization status for non-urbanized cells. Nevertheless, this 1960-1970 model is not predicting any cell with urbanization status change even though with 90.25% overall fit percentage. My explanation is that the sample set is highly skewed with 657:71 no-change/change ratio and two distance variables can only explain a limited portion of the dependent variable's variance although both of them are statistically significant. The beltway effect does not appear to be significant. In fact, based on the status of the Oklahoma City highway systems as of 1960, as shown in figure 5, no beltway effect should have been expected.
Limitations of 10-year Models and the Concepts of 20-year Models Because of the inherent constraint of our arbitrarily defined study areas, e.g., 3-mile buffer, 4-mile buffer, etc., it would be unfair to conclude that the widely held expectation that highways encourage urban sprawl has proven to be incorrect. For example, figure 5 shows the change of urbanized area boundaries between 1960 and 1970. It is obvious that south portion of urbanized area growth is well beyond the 3-mile buffer of 1960 boundary. In order to avoid the potential of losing information, we developed another set of models which examine the urbanized status change in every 20 years. That is, two models, one for year 1960 to 1980, the other for year 1970 to 1990, are developed based on a very similar framework. For the 1960 - 1980 model, the distance to the nearest highway interchange is recorded for each cell based on highway networks completed by 1970. The distance to the nearest existing urbanized area cell is measured based on 1960 urbanized area boundary. Correspondingly, for the 1970-1990 model, the distance to the nearest highway interchange is based on highway networks by 1980 while the distance to the nearest urbanized cell is measured based on 1970 urbanized area boundary. Table 2 summarizes the results of both models.
1960-1980 Model Figure 8 shows the urbanized area expansion between 1960-1980, highway systems completed by 1970, and the defined 3.5-mile buffer (of 1970 urbanized area) study area. For 1960-1980 model, both distance variables coefficients show significant explanatory power. Unlikely previous 10 year period models, the distance to the nearest highway interchange variable (b = -0.1881) becomes a more influential factor in changing urbanization status than the distances to the nearest existing urbanized cell variable (b = -0.0752). Once again, the time lag variables, INT_60, is not a significant factor in this model. As expected, the beltway effect, INT_BELT (b = 0.2775), is not a significant predictor. The accuracy rate of prediction is 68.32%. Apparently, this model is much more reliable when predicting a cell as no change (0 to 0, 96.71% accuracy rate) than it is when predicting a cell as change (1 to 1, 11.68% accuracy rate). However, this is already an improvement in contrast to the 1960-1970 model and the 1970-1980 model where cells with changed urbanization status are predicted with 0% accuracy rate.
Table 2, Logit Modeling Results: 20-Year Models
Dependent Variable UA_CHANGE Study Period 1960-1980 1970-1990
DIST_INT -0.1881( .0000) -0.1486( .0000) DIST_UA -0.0752( .0044) -0.4223( .0000) INT_60 0.1015( .5195) -0.4539( .0267) INT_70 -0.2396( .1333) INT_BELT 0.2775(.3468) 3.3636(.0011)
Constant 0.1672( .2704) 1.3990( .0000)
Observations 1051 1299 Change 351 551 Predicted 41 313 % Predicted 11.68% 56.81% No Change 700 748 Predicted 677 597 % Predicted 96.71% 79.81% Overall Fit 68.32% 70.05%
Note: Numbers in the parenthesis show the t statistics for the corresponding variables. Numbers with bold font are statistically significant at 0.05 level.
1970-1990 Model Figure 9 shows the urbanized area expansion between 1970-1990, highway systems completed by 1980, and the defined 4-mile buffer (of 1980 urbanized area) study area. In this 1970-1990 model, both distance variables appear significantly with negative coefficients as expected. The distance to the nearest urbanized area cell, DIST_UA (b = -0.4223) is a more important factor in terms of its impact on urbanization status change than DIST_INT (b = -0.1486). Table 3 compares the coefficients of two distance variables derived from both 1960-1980 model and 1970-1990 model. On one hand, the decrease of coefficient value of DIST_INT indicates that the impact of highway systems on suburbanization is declining. On the other hand, the increase of coefficient value of DIST_UA suggests that the spatial neighboring effects on suburbanization is growing. A further discussion regarding this phenomenon will be made in the concluding section. The coefficient of INT_BELT unveils the same information as previous 1980-1990 model that beltway effect has tremendous predict power (b = 3.3636, p = 0.0011) during the last 10 to 20 years or so. The negative and significant coefficient of INT_60 suggests that having a more than 20 years old highway interchange as the nearest interchange, a non-urbanized land cell would have less chance of changing its urbanization status than others.
Table 3, Comparison of Distance Variables' Coefficients between 20-year models
Variable 1960-1980 model 1970-1990 model % change
DIST_INT -0.1881 -0.1486 - 21%
DIST_UA -0.0752 -0.4223 + 462%
Stronger Predictor* DIST_INT DIST_UA
* Predictor's strength is decided by its absolute value of coefficient
Because logit model is a nonlinear probability model, the
Interpretation of Models' Coefficients
Spatial Effects Transitional Analysis
Based on previous modeling results, I conclude that the spatial
neighboring effect is playing a more important role than the spatial
correlation effect, which was a dominant factor during the massive
highway construction era. In order to explore this phenomena more
clearly, I use curve fitting techniques to depict this transition
graphically, as well as to predict the future urbanization probabilities
for non-urban lands with different physical properties, e.g., distance to
interchange. Twenty-year modeling results are used as the base of this
Figures 10 and 11 illustrate the probabilities of urbanization of
lands with various distances to the existing urbanized area and the
nearest interchange. As indicated, the impact of the two spatial effects
can be examined through the differences of slopes. A larger difference
between the slope of the axis and the slope of observed probabilities,
when controlling for the other axis (effect), reveals a greater impact of
that spatial effect. We can see in Figure 10 that the slope facing the Y
axis (spatial correlation effect) is steeper than that facing the X axis
(spatial neighboring effect). Once again, a stronger spatial correlation
effect for the 1960-1980 period is confirmed. In Figure 11, the opposite
result is observed. The slope facing the X axis is much steeper than
that facing the Y axis in the 1970-1990 period. Clearly, a spatial
effect transition took place between the two time periods. If there are
no other major technological and policy changes in the near future, the
spatial neighboring effect is expected to play a more important role in
population suburbanization in the next few years.
Figure 12 shows the probabilities of changing urbanization status
for lands 1) five miles from nearest interchange and one mile away from
existing urbanized area (UA1INT5, solid line in the Figure), or 2) one
mile from nearest interchange and five miles away from existing urbanized
area. Both types are controlled for the age of the nearest interchange
as "1970" and the type of the nearest interchange variables as
"beltway". Time period 0 reflects the probabilities during the 1960-1980
period while time period 1 stands for 1970-1990 period. In other words,
curve segments beyond time period 1 are the predicted probabilities. The
logistics curve fitting model is used due to its desirable bounded growth
property (probability * 0 and * 1). As expected, distance to the
existing urbanized area is becoming more and more decisive in terms of
influencing the urbanization probabilities. Any developable land which
is close to the existing urbanized lands, e.g., UA1INT5, has a tremendous
potential to become urbanized land in the near future.
Figure 12, Transitional Trend of Probability of
Note: UA1INT5 stands for lands with one mile away from existing urbanized
area and five miles from the nearest interchange. UA5INT1 stands for
land with five miles away from existing urbanized area and one mile from
the nearest interchange.
A) According to urban development theories and past experiences, we know that American people tend to live and work in low-density areas. From the lessons we learned from opening interstate highway systems, we conclude that many people are willing to trade travel time savings for a suburban living environment. In other words, people are willing to move as far away from high density areas, especially the central city, as they possibly can, if the commuting time remains unchanged. The combination of the above two facts suggests that travel time, not distance, is the key in deciding the directions of suburbanization. During the massive interstate highway construction era (1960s' and 1970s') when people were able to travel on the highway at a relatively high speed, living in a low density suburban neighborhood became affordable as long as the neighborhood was close enough to a highway interchange. Not surprisingly, this research found that the shapes of urbanized area expansion were heavily influenced by the directions of interstate highway systems during this period of time. However, since the 1980s', when many metropolitan areas started suffering traffic congestion problems, the daily commute between suburbs and central cities became a nightmare for many commuters. Interstate highway systems were mostly overloaded, with the average vehicle travel speed down to about 30mph or lower during peak hours. Apparently, the travel speed and travel time advantage for interstate highway systems no longer exists. This may explain why the highway effects on shaping urbanized areas are fading.
B) The other possible explanation for this phenomena is suburbanization of employment since the 1980s. As mentioned, during the massive highway construction era, residential suburbanization was the major tendency. During the 1980s', the local economic climates, e.g., labor force supply and transportation accessibility, and the emergence of telecommunication technologies made the suburban environment attractive to many employers. American businesses began to show a preference for low-density workplaces. Nodal development around the traditional center cities is a common land use development pattern since the 1980s. (Levine, 1992) The development since then tends to infill those vacant lands between center cities and new town centers rather than moving further out.
2. The beltway effect has become a dominant factor in predicting urban area boundaries since the 1980s. With all other predictors in the model controlled, any non-urban land cell with a beltway interchange as its nearest interchange would increase the odds (P/1-P) of changing its urbanization status by a factor of 28 (= e 3.36, based on 1970-1990 model) or more. This conclusion is consistent with the conclusion derived before the urban sprawl during the 1980s had a tendency of in-filling the surrounding non-urban land, rather than stretching along the radial highway corridors.
3. Even though both spatial correlation effects and spatial neighboring effects appear to have significant prediction power, there still exist some other crucial determinants which are influential factors in predicting urbanized area growth, but are not included in the models. Throughout all developed models, the accuracy rates of predicting changed cells (dependent variable = 1) are consistently much lower than the accuracy rates of predicting non-changed cells. All five models tend to over predict non-changed probabilities. The skewed sample distribution is the fundamental cause of this tendency. The combined prediction power of the two distance variables appears to be insufficient to detect those underrepresented cells correctly under this circumstance.
4. The 10-year time lag effects are consistently insignificant through all logit models. The signs and magnitudes of the 10-year time lag variable also vary from model to model. However, the 20-year time lag variable shows negative and significant coefficients in both the 1980-1990 and the 1970-1990 models. It seems that having an older highway interchange as the nearest interchange makes a non-urban land cell less likely to urbanize. Unfortunately, from a modeling point of view, there exists less than 40 years of history about the interstate highway system. And my observation time interval is limited to every ten years due to the limited availability of census data. The inherent limitation also keeps us from exploring any time lag effect of less than 10 years. The results must be viewed with caution.
Policy Implications Intelligent Transportation Systems (ITS) is considered by some as a major contribution to transportation technology as we move toward year 2000. The deployment of ITS technologies is expected to influence people's travel behavior in the near future. Among various ITS goals, saving travel time, both individual time and commercial time, is one of the most important components. Many time-saving related technologies, such as adaptive traffic signal control systems and dynamic route guidance systems, are either commercially available or under field test status. From the lessons of interstate highway development in the past forty years, we have already learned that people show their willingness to trade travel time savings for greater travel distance to meet their desires for low density neighborhoods and low density workplaces. Urban sprawl in the past decades has indeed caused enormous social and environmental costs. We have made mistakes and have paid for them. We want to make sure that ITS can be deployed in such a way that people are encouraged to preserve the benefits of travel time saving rather than substitute them with any other potentially detrimental decisions, such as moving further away. While we are lucky enough to see that the driving force for suburbanization is getting weaker, we have to prevent other causes from emerging. The potentials of some ITS technologies, such as Automated Traffic Management Systems (ATMS) and Automated Traveler Information Systems (ATIS) for encouraging urban sprawl should be properly addressed. And the benefit cost analysis of ITS deployment should also be conducted with extreme caution so as not to overestimate ITS benefits, e.g., travel time saving, with the assumption of current travel patterns remaining unchanged. From experiences, our society has learned that a net increase in people's mobility tends to enhance the magnitude of urban sprawl, i.e., the amount of urbanized area expansion. An unbalanced distribution of social resources, e.g., income and mobility, will contribute to the sprawl of urban forms. The combination of rapid and unbalanced urban area expansion causes enormous problems, such as traffic congestion and some derived environmental problems, such as air pollution. Many transportation-land use studies in the past explored many different solutions, such as job-house balancing strategy, to cure the urban sprawl problems. According to Levine (1995), one extreme of thought argues that we should let the market mechanism work by itself. In contrast, the other extreme of thought, in contrast, proposes that a stricter land use control procedure should be enforced. Both ideas have proven to be inefficient solutions. If the market could work by itself, we would not have had this urban sprawl problem in the first place. There exist negative externalities associated with the residential location decisions which create the improperly distribution of social resources. We saw richer people moving to the suburbs, with the social costs of their relocating behaviors burdening the whole society. This distorted market then creates some inefficient results. On the other hand, local land use control seemed to serve richer people better than it did to the poor. In many cases, we saw clearly that richer people moved to the suburbs and then used land use control as an excuse to keep poor people from following. Even the job-house balancing policy does not appear to be an effective solution to this sprawl problem for several reasons, such as the fact that many household have two or more workers going in various directions. Considering the fact that the highway effect on suburbanization is diminishing and ITS technologies are emerging, I am proposing the following four policy suggestions to cope with the current transportation problems and related issues.
1. Internalize suburbanization externalities People who move to suburban areas and generate negative externalities should be paying for the "true social cost" of their choice. The higher cost would have the effect of discouraging urban sprawl. 2. Re-prioritize ITS deployment procedures Many social policies, such as taxation, aim at income redistribution. In many respects, mobility can be treated as another type of property owned by individuals. Most ITS development efforts in the U.S. seem to focus on the private and commercial transportation services. It is important that public transportation applications of ITS are also promoted in such a way that public transit users are not discriminated against by this technology. This is very crucial in the sense that unbalanced distribution of mobility is also a major cause of urban sprawl. The potentials of ITS for private transportation are expected to enhance the users' mobility significantly. The deployment process of ITS user services should be scheduled in such a way that public transit users (usually having less mobility) can receive the benefits of ITS at the same time as private mode users, if not earlier. 3. Equip people with resources and provide them with more options The major reason why job-house balancing did not work well is that no one knows how to meet people's needs better than people themselves. Instead of offering people the "assumed" best choice, people should be given the power to make their own choices. In this case, people should be allowed to live any place they want (of course, with certain restrictions), as long as they are willing to pay the "true cost". Government's role is to provide people with similar resources, e.g., income, education, etc. so that they will have adequate knowledge and information to maximize their personal utilities. 4. Let market run by itself when the situation "matures" In most cases, government interventions became the biggest cause of inefficiency and society loss. I agree that to a certain extent government intervention is necessary. But the goal of government intervention should aim at maintaining a healthy environment and social justice (i.e., prevent the more powerful from ripping off weaker people), instead of providing people with "end products" that government believes are best. In the ITS-transportation case, the government should make sure of the existence of equal opportunities in terms of selecting house and workplace locations. Then, let the market mechanism generate the optimal solution. It has been 38 years since the passage of Federal-Aid Highway Act of 1956. The passage of Intermodal Surface Transportation Efficiency Act (ISTEA) of 1991 authorized $21 billion out of its $151 billion total budget over six years for national highway system construction and maintenance. As a response to the ISTEA, the Secretary of Transportation, Federico Pena, brought the proposed 158,000-mile National Highway System (NHS) before Congress in December 1993, giving birth to the National Highway System Designation Act of 1994. In other words, tremendous attention to and action on the interstate highway system is foreseeable. Being urban planners, we have observed the profound impact of highway systems on urban forms. We have recognized the problems with certain types of urban sprawl. Hopefully, this research will contribute to clarifying factors influencing social benefits of highway investments. Future Study Suggestions Expand the Scope of this Study This research can be expanded on three different dimensions: issue, space, and time. On the first dimension - issue, I would like to explore the transitional dynamics between interstate highway systems and other social phenomenon, such as population, migration, and environment dynamics. On the second dimension - time, I intend to include a longer period of time, e.g., starting from the pre-auto era to the present. Finally, on the third dimension - space, I wish to extend the study to examine more cities in the U.S. or even the rest of the world, in order to categorize transitional patterns. This study should be treated as a pilot study, to be followed by a series of further in-depth explorations. The final goal of this study is to develop a comprehensive methodology which is replicable to other urbanized areas in/out of the U.S. Accommodate Multi-Level Model Structure Apparently, the proposed model would encounter the dilemmas of a hierarchical data structure. The Level-1 units are grid-cells, which are nested within the Level-2 units of highway completion year. As indicated by A. S. Bryk and S. W. Raudenbush (1992), the most common concerns regarding hierarchical data analysis include aggregation bias, misestimated precision, and the "unit of analysis" problem. Nevertheless, this study would still use a conventional dichotomous logit model to predict the probability of urbanized status change due to the following two reasons:
1. First of all, unlike a hierarchical linear model, which is substantively developed, there is not yet any hierarchical nonlinear model which is universally accepted.
2. Empirical evidences, e.g., school effects on teacher efficacy study by Bryk and Driscoll (1988), show that there is no significant difference between results from hierarchical linear models and these from conventional single level models in terms of the magnitudes of coefficients.
Add Zoning/Land Use Information to Exclude those Undevelopable Lands In this study, only water body is excluded from the sample set. In the real world, many types of lands are not developable due to limitations, such as unsuitable soil type, and environmentally sensitive and preserved areas. The forecast accuracy rate would be improved, if, before establishing models, future studies excluded these non-qualified areas from the sample set. Use Discrete Network Distance to Measure Real Travel Time rather than Making Continuous Space Assumption This study uses a simplified distance measurement method, Euclidean distance, to acquire distance information. An effort to apply the discrete network concept to measure actual travel time hopefully would improve the models' prediction capability. Conduct Residual Autocorrelation Analysis for each Prediction Model This study relies on the "eyeballing" method to examine modeling residuals' spatial autocorrelation phenomena. In fact, there are several spatial autocorrelation models available to explore the intensity of autocorrelation in a quantitative manner, e.g., Moran's I statistics and joint-count statistics (Upton and Fingleton, 1985). It is desirable to apply these available analytical tools to justify the models' assumptions and to improve their Definition of Beltways The most commonly used method to define and label a beltway is to base it on the highway numbers, such as found in Payne-Maxie Consultants, 1980. For example, I-96 is a main line while I-696 is a beltway. The same is true for I-75 vs. I-275. In this study, I also used this intuitive way to define beltway for the case study area. Nevertheless, I recognize that a set of objective and scientific criteria are necessary for academic research purposes, especially for a geographic analysis like this. I am currently developing a method of defining beltway from a geometric perspective. Hopefully, the follow-up study can apply this new method to objectively defined beltways.
1. John H. Aldrich and Forrest D. Nelson, Linear Probability, Logit, and Probit Models, SAGE publications, Newbury Park, California, 1984.
2. Katharine Bradbury, Anthony Downs, and Kenneth Small, Urban Decline and The Future of American Cities, The Brookings Institution, Washington D.C., 1982.
3. Anthony S. Bryk, Stephen W. Raudenbush, Hierarchical Linear Models: Application and Data Analysis Methods, Sage Publications, 1992.
4. Keith C. Clarke, Analytical and Computer Cartography, Prentice-Hall, Inc., New Jersey, 1990.
5. Judy S. Davis, Arthur C. Nelson, and Kenneth J. Dueker, "The New Burbs,: The Exurbs and Their Implications for Planning Policy", JAPA, Winter 1994, pp. 45-60.
6. Alfred Demaris, Logit Modeling: Practical Applications, SAGE publications, Newbury Park, California, 1992.
7. Anthony Downs, Opening Up the Suburbs, Yale University Press, New Heaven and London, 1974.
8. Anthony Downs, Stuck in Traffic, The Brookings Institution, Washington D.C., 1992.
9. Eran Feitelson, "The Spatial Effects of Land Use Regulations: A Missing Link in Growth Control Evaluation", JAPA, Autumn 1993, pp. 461-472.
10. Susan Hanson, The Geography of Urban Transportation, The Guilford Press, New York, 1986.
11. Jonathan C. Levine, "Decentralization of Jobs and the Emerging Suburban Commute", Research Report, College of Architecture and Urban Planning, The University of Michigan, 1992.
12. Mitchelle E. Macdonald, "Planned National Highway System to Upgrade 158,000 Miles of Roads", Traffic Management, February 1994, pp. 17-18.
13. Payne-Maxie Consultants, The Land Use and Urban Development Impacts of Beltways, Final Report No. DOT-OS-90079, U.S. Department of Transportation and Department of Housing and Urban Development, Washington, D.C., 1980.
14. Eudora S. Pendergrast, Suburbanizing the Central City : An Analysis of the Shifting in Transportation Policies Governing the Development of Metropolitan Toronto, 1959-1978, Department of Urban and Regional Planning, University of Toronto, 1981.
15. F. K. Plous, Jr., "Refreshing ISTEA", Planning American Planning Association, February 1993, pp. 11-16.
16. Michael Southworth and Peter Owens, "The Evolving Metropolis: Studies of Community Neighborhood, and Street Form at the Urban Edge", JAPA, Summer 1993, pp. 288-305.
17. Jon C. Teaford, The Twentieth-Century American City: Problem, Promise, and Reality, The Johns Hopkins University Press, 1986.
18. Graham J. G. Upton and Bernard Fingleton, Spatial Data Analysis by Example, John Wiley & Sons, 1985.
19. Sammis B. White, Lisa S. Binkely, and Jeffrey D. Osterman, "The Sources of Suburban Employment Growth", JAPA, Spring 1993, pp. 193-204.