|
|||
|
|
|
|
|
|
U.S. Department of Agriculture’s Data. Developed land is defined by the U.S. Department of Agriculture’s Natural Resources Conservation Service (NRCS) in its 1997 National Resources Inventory (NRI), originally released in December 1999 and then re-released in January 2001 with revised and corrected figures.174 The NRI is primarily oriented toward private and non-federal lands, with an emphasis on the quality and quantity of the nation’s productive resource land base, that is to say, croplands, range, pasture, and forestlands. But it also quantifies the loss of other non-federally owned open spaces and natural habitat, as well. The NRCS identifies Developed Lands as those non-federal lands that have been removed permanently from the rural land base. The Developed Land category includes: (1) large tracts of urban and built-up land of 10
acres or more; U.S. Census Bureau’s Urbanized Areas. Our second measure of sprawl is obtained by drawing on decennial statistics for population, for urbanized land per average resident, and for total urbanized land from U.S. Bureau of Census data on Urbanized Areas of the United States.176 Most recent studies and reports that have attempted to quantify land consumption due to sprawl have used the same data. (The Census definitions and methodology for measuring each Urbanized Area are described in Appendix F.) An "Urbanized Area" (UA), as defined by the Census Bureau, is a continuously built-up or developed area with a population of at least 50,000.177 It consists of one or more "central places" as well as densely-settled surrounding areas which the Bureau terms "urban fringe." The central place(s) and urban fringe may be thought of as the urban core and suburbs of a given UA. The Census Bureau introduced the UA concept in the 1950 census as part of its efforts to differentiate the urban and rural portions of the nation’s population (See Appendix F). Per Capita Consumption Factors Alone Cannot Explain
Overall Sprawl Defining ‘Per Capita Sprawl.’ To illustrate, we will use Georgia, where the public has grown increasingly alarmed at the pace of sprawl. Per capita land consumption in Georgia was 0.419 acres in 1982. It grew to 0.529 acres in 1997. Thus, the Per Capita Sprawl over that period was 0.110 acres, or 26.2 percent. From this information, it is obvious that Georgians must address Per Capita Sprawl if they are to slow overall sprawl in the state. The concept of Per Capita Sprawl, or per capita land consumption growth, is immensely useful because it compresses into a single figure the results of dozens of factors listed earlier as probable causes of consumption increases. Per capita urban or developed land consumption is not limited to the size of a person’s house lot or to a person’s proportion of the land covered by an apartment complex. It also includes a portion of all the other land that has been converted from rural to urban use to provide for jobs, industry, commercial establishments, recreation and entertainment, shopping, parking, transportation, storage, government services, religious and cultural opportunities, waste handling/disposal, and education. In more rural settings, it would include rural housing and vacation homes and also built-up or heavily modified landscapes associated with agriculture and resource extraction like food processing facilities, mines, mills and smelters, quarries, port facilities, sawmills and lumberyards, hydroelectric dams, and so forth. Thus, the level of per capita land consumption is based both on direct individual decisions and behavior, and on collective decisions made through the government and the marketplace. The effect of all urban planning, zoning, development, and transportation decisions shows up in the per capita land consumption figure. The amount of developed land in predominantly rural settings is a function of similar decisions, as well as those made by federal and state land management agencies, agribusiness and resource industries. In the end, per capita land consumption under Natural Resources Conservation Service data is calculated by dividing the total developed land area of a given state by the total number of residents in that state. Under the Census Bureau data, the total amount of land in an Urbanized Area is divided by the population of that area. (See Appendix D for more on calculating per capita land consumption.) Per Capita Sprawl and Smart Growth. It is very difficult to measure precise effects of trying to change any one of the planning, consumption, and other behavioral factors as causes of land consumption growth. But we can determine the overall effect of all those factors together by looking at the simple statistic of the average amount of developed or urban land per resident in an entire state or any Urbanized Area. If that per capita land consumption figure goes up markedly, then we know that Smart Growth efforts related to the above factors either have not been undertaken or are failing to achieve their desired result of higher densities. If the per capita figure grows only slightly, or remains the same, and especially if it decreases, then planning, consumption, and behavioral factors are collectively moving in the direction desired by the anti-sprawl leaders. It is difficult to determine whether or not their efforts have any impact, but we do know in such cases that per capita land consumption patterns are being brought under control. Per Capita Sprawl Rate Far Less Than Overall
Sprawl Rate 
Our literature search found that most media stories, advocacy programs, governmental reports, and political statements about sprawl have focused almost entirely on development esthetics, reducing public costs, and the land-use and consumption factors that cause per capita land growth. This would suggest that Per Capita Sprawl explains most, if not all, of the Overall Sprawl in the nation’s Urbanized Areas and the increase in overall developed lands throughout the countryside. Our hypothesis questioned the validity of such a supposition that appears to deny that population growth explains a significant amount of sprawl. Per Capita Sprawl Is Only One Part of the Story. One way to determine if growth in per capita land consumption indeed explains most of sprawl is to compare the percentage growth of per capita land consumption with the percentage growth of all developed land, which is what Table 5 does for the 49 states surveyed in the 1997 National Resources Inventory. Using Georgia as an example, we see a 26 percent increase in per capita land consumption. But the overall development of rural land increased by 67 percent. If the factors causing growth in per capita land consumption were the overwhelming cause of Georgian sprawl, their percentage growth would be nearly as high as the 67 percent Overall Sprawl, or certainly well over half of 67 percent. Instead, Per Capita Sprawl was less than half. Thus, the simple exercise of comparing the two percentage growth rates shows the invalidity of the supposition that growth in per capita land consumption is the overwhelming cause of rural land development in Georgia. When we look at the side-by-side percentage comparisons for all states, we find: • Most states were like Georgia, with the percentage growth in developed land being considerably larger than the percentage growth in per capita consumption. It was twice as high in Wisconsin, three times as high in Kansas, four times as high in Texas, seven times as high in Colorado and 16 times as high in California. • In only six states were the percentages in the two columns close enough to suggest that nearly all the sprawl was related to growth in per capita land consumption (Iowa, Louisiana, North Dakota, Pennsylvania, West Virginia, and Wyoming). • In 24 states, per capita land consumption did not grow at even half the rate of Overall Sprawl. Obviously something more than just per capita sprawl is at work in producing the loss of undeveloped land in most states. (That is especially true in Arizona and Nevada where per capita land consumption did not grow at all. That phenomenon is discussed later in the report.) By comparison, we find much the same situation in the 100 largest Urbanized Areas: there, as well, very few of the Per Capita Sprawl percentages are even close to as high as the Overall Sprawl percentage. Focusing Only on Per Capita Sprawl is Too Narrow. Figure 2 shows that for the 100 largest metropolitan areas combined, Per Capita Sprawl growth was a significant 22.6 percent. But overall sprawl growth was more than twice as high at 51.5 percent. We can see in both the state and city figures that all the factors leading to growth in per capita land consumption simply have not produced enough sprawl to explain the overall increase in land development. Clearly per capita land consumption growth is a major factor — but not the overwhelming factor — in America’s urban sprawl and the overall spread of development in states’ rural areas. Though the statistics for a few states as well as some of the Urbanized Areas seem to justify a single-factor anti-sprawl approach, most of the states and cities fit another explanation, one in which both Per Capita Sprawl and the other major factor — population growth — must be tackled if Overall Sprawl is to be seriously slowed.
This finding would indicate that most Smart Growth efforts are too narrow to succeed in substantially halting sprawl. It is not that Smart Growth efforts are focused on the wrong factors, but that they are focused too narrowly. Obviously, there is another factor involved in sprawl. Without also addressing population growth, Smart Growth programs as currently envisioned, promoted, and implemented, are destined to fall far short of protecting agricultural land and natural habitats from the spread of asphalt, concrete, gravel, and steel. Per Capita Land Consumption Growth • 400 residents (0.200 acre) * (400 people) = 80 acres Let’s say we revisit this village a few years later and find that the fully developed area has expanded 50 percent to 120 acres. There can be only three types of explanation for this overall growth in developed land: 1. Per capita land growth alone: The 400 villagers have expanded their per capita land consumption by 50 percent from 0.200 acre to 0.300 acre. (0.300 acre) * (400 people) = 120 acres This could have happened by households dividing through divorce or children leaving home and starting new households, by people expanding the size of their houses and yards, by constructing additional public and commercial buildings, and by abandoning homes and stores within the old boundaries to move just outside those boundaries, perhaps adding a shopping mall and large parking lot on the town’s edge. In whatever way, the 400 villagers have expanded into the surrounding countryside without adding any extra population. Such a situation is precisely what most of the nation’s Smart Growth programs are designed to address. 2. Population growth alone: The per capita land consumption did not rise at all while 200 additional residents moved into the village, causing a 50 percent increase in population to 600. This is the situation that best fits the prescription of "population hawks" who believe most problems can be resolved simply by stopping population growth. (0.200 acre) * (600 people) = 120 acres 3. Combination of per capita land growth and population growth: There may have been some combination of both per capita land consumption growth and population growth. One example would be that per capita land use grew 20 percent to 0.240 acre and population grew by 25 percent to 500. This situation requires a two-pronged approach. (0.240 acre) * (500 people) = 120 acres Notice that although population grew by 25 percent and per capita consumption grew by 20 percent, the total land development grew by a percentage (50 percent) that is more than the sum of the percentage of both growth factors (25 percent and 20 percent). This is due to "second-order terms" and should not suggest that the two major factors account for less than 100 percent of the sprawl when working together. In each state and Urbanized Area, sprawl has occurred under one of those three scenarios. But, as we found in the comparisons in the previous section, most states and cities fall in the third scenario. Despite the considerable complexity of sprawl and of the development of rural land, nearly all the complexity can be boiled down to what end up being two rather simple factors in an equation: The amount of Overall Sprawl in an area is equal to the change in per capita land consumption multiplied by the change in population. Lining Up the Two Sprawl Factors Side By Side • • In 20 states, population grew faster than per capita land consumption. • In one state (Missouri), per capita land consumption and population grew at an equal rate (10 percent). • In four states, population shrank while per capita land consumption increased. • In two states, per capita land consumption declined while the population grew. • In no state did both per capita land consumption and population decline.
We also find that those states that had higher population growth tended to have less growth in per capita land consumption, i.e. less Per Capita Sprawl. For example, the 10 states with population growth of 25 percent or more averaged only a 4 percent rise in per capita land consumption — that compared to a 16 percent rise in per capita land consumption averaged for all 49 states. There could well be a correlation between higher population growth and lower Per Capita Sprawl, perhaps due to greater regulation and land use planning that become politically feasible with intense population pressure, and that have the net effect of pushing up densities in an effort to limit sprawl. It also could be that the construction and development industry are not able to keep pace with rapid population growth, which means that per capita land consumption may eventually rise when industry has had more time to respond. Combining the data of all 49 states, we find that growth in both population and per capita land consumption occurred at the same rate — 16 percent over the 15-year period of 1982-1997. We find a similar result when combining data for all 100 of the Census Bureau’s largest Urbanized Areas. Those results are summarized in Figure 3. Overall population growth in these cities from 1970-1990 was 23.6 percent and their Per Capita Sprawl or per capita land consumption growth was 22.6 percent. Thus, it is evident that the roles of the two growth factors are nearly identical in both the states from 1982-1997 and in the largest cities from 1970-1990.
Scatter Plot of Population Growth and Sprawl. One
of the most common and straightforward ways of examining the
relationship between two variables is to use what is called a scatter
plot. Figure 4 is a scatter plot of the increase in each state’s
population (1982-1997) on the x (horizontal) axis and percent sprawl
(1982-1997) on the y (vertical) axis. The scatter plot also shows a
straight line, which represents the "best fit" to the data points. That
is, the line smoothes out fluctuations in the data and shows the pattern
or relationship between population growth and sprawl more clearly. The
upward or positive slope of the line indicates that there is a positive
relationship between population increase and sprawl — states with more
population growth are also states where more land was developed. If the
contention of some observers that sprawl and population are unrelated
was correct, then the line should be flat or even negative. This is
clearly not the case. Of course, the scatter plot does not conclusively
prove that population growth causes sprawl, but it does strongly suggest
the two are closely
Scatter Plot of Per Capita Increase in Land Use and Sprawl. Population growth is not the only factor affecting sprawl. As already discussed, increases in land use per-person must also play a role in the expansion of developed land. Figure 5 shows a scatter plot with percent sprawl again on the y-axis, but instead of population growth on the x axis as in Figure 4, the increase in per capita land use is substituted. The positive slope of the line indicates that increases in land use per person are also positively correlated with sprawl. Taken together, Figures 4 and 5 indicate that both factors are important reasons for sprawl.
Top- and Bottom-Ten Sprawling States. Another simple way to examine the relationship between population growth and sprawl is to examine the amount of population growth exhibited by those states that sprawled the most and the least. If population growth were not an important underlying cause of sprawl, one would expect that states which sprawled the most would have very similar rates of population growth as states that sprawled the least. In fact, the opposite is true. Increases in developed land are associated with population growth: • In the 10 states that had the largest percentage increase in developed land between 1982 and 1997, population grew on average by 19 percent. • In contrast, the 10 states that had the smallest percentage increase in developed land had population growth averaging only 4 percent. States that sprawled the most grew dramatically more in population than those with the least sprawl. By itself, of course, this does not prove that population growth is an important underlying cause of sprawl. But it certainly indicates that where there is a great deal of sprawl there is also a great deal of population growth.
The Impact of Population Growth and Density Changes
on State Sprawl Population Is Closely Associated with Sprawl. Figure 7 shows the relationship between population growth and sprawl by separating the states that had population growth according to their rate of growth. The figure shows the same general pattern as in the above discussion: • As we have seen, in the four states in which the population declined, developed land increased 20 percent on average. • Of 22 states that grew in population by between 0 and 10 percent, developed land increased 26 percent on average. • In the seven states with population growth of more than 10 percent and less than 20 percent, there was a 38 percent expansion in developed land on average. • In the 10 states with population growth of more than 20 percent and less than 30 percent, there was 41 percent expansion in developed land on average. • In the six states that grew by more than 30 percent, there was 46 percent expansion in developed land on average. Population growth appears closely associated with sprawl. In general, the more a state grows in population, the more land that is lost to development. Of course, these figures also show that even where there is population decline, there is still sprawl, indicating that increase in population is certainly not the only reason for sprawl.
Ranking the States by Absolute Population Growth. Ranking the 49 states based on total population change and not percentage change in population confirms yet again the striking positive relationship between population increase and expansion of developed land. If the states are ranked by absolute population growth from highest to lowest, then divided into three groups we get the following results: • The top third averaged population growth of 1.7 million from 1982 to1997 and lost on average 871,200 acres to development. • The middle third had population growth averaging 365,707 and lost on average 393,619 acres. • The bottom third had population growth averaging 56,000 and lost on average 229,000 acres. Arizona and Nevada: Success Stories? Another way to think about the relationship between population growth and the expansion of developed land is to look at states in which land use per person actually went down. That is, there was no Per Capita Sprawl. There were only two states where land use per person decreased between 1982 and 1997 — Arizona and Nevada. This is a goal of the Smart Growth movement. So in effect, these states should be a real success story. But in both states, the amount of sprawl was enormous. Arizona experienced a 40 percent increase in total developed land while Nevada’s increase was 37 percent. These two states can hardly be described as success stories if the goal is to preserve open space and protect undeveloped land. Both states had such high population growth that it negated the gains from a reduction in per capita consumption, with the result that an enormous amount of land was lost to development. Arizona’s per capita land consumption declined by 13 percent, but with massive population growth of 58 percent the state suffered the loss of 629 square miles of undeveloped land. Nevada reduced per capita land consumption by 26 percent but still lost 171 square miles to development. The lesson of Arizona and Nevada is that if there continues to be dramatic increases in population, controlling the density of new settlements by itself will not prevent very high rates of sprawl. Land Use and Population Both Matter. All of the findings in this section indicate that population growth is an important underlying cause of sprawl. Of course, our analysis makes clear that increases in per capita land consumption also play a role. Thus, the Smart Growth movement is correct to focus on factors that would reduce land use per person. Nonetheless population growth also seems to be central to understanding the loss of rural and undeveloped land. To focus exclusively on per capita land use as the movement has largely done, however, is almost certainly not going to prevent massive loss of rural and wild land. Population Growth’s Share of Sprawl Quantifying the Role of Population Growth. Since both sprawl factors together account for 100 percent of the sprawl in each state or city, the exercise in this section will merely convert the relationship between the two sprawl factors into two percentages that will add up to 100. As a result of this exercise, we will be able to say that around 44 percent of the sprawl in Georgia is related to increases in per capita land consumption and that around 56 percent of the sprawl is related to population growth. This is merely a mathematical way of expressing the relationship we already could observe when looking at the 26.2 percent growth in per capita land consumption and the 32.5 percent growth in population. We will use two statistical methods to calculate the relationship. Both are based on a simple equation. As already mentioned, and as explained more fully in Appendix D, the amount of land taken up by a city, town, metropolitan area, developed area, or urbanized area is the simple product of the number of residents times the amount of land used (or consumed) per resident, as shown in the following equation: A = (P) * (a) Where: A = Area of urbanized/developed land in acres or square miles P = Population of the urban/suburban area or state a = developed or urbanized land per person (i.e. the inverse of density, which is number of people per unit area of land) Sprawl then is the increase of ‘A’ over time. The Simple Ratio Approach. A simple way to calculate the ratio of any two figures to each other is to add them together to obtain a sum, which can then be divided into each figure to yield two percentages. The two percentages thus obtained will add up to 100 percent. We call this the "simple ratio" method. For example, in the case of Georgia, we add the per capita consumption growth percentage of 26.2 to the population growth percentage of 32.5, yielding a sum of 58.7 percent. When we divide 58.7 into each growth figure we find that: • The 26.2 percent growth in per capita consumption is 44 percent of the power of the two growth figures combined (26.2 / 58.7 = 44 percent). • The 32.5 percent growth in population is 56 percent of the power of the two growth figures combined (32.5 / 58.7 = 56 percent). Based on that, we can say that 44 percent of Georgia’s sprawl is explained by, or related to, per capita consumption growth, and that 56 percent of Georgia’s sprawl is explained by, or related to, population growth. Simple Ratio Analysis of States. Applying the "simple ratio" to states in Table 7 we find a great deal of variation from state to state in the source of sprawl: In six states (Iowa, Louisiana, North Dakota, Pennsylvania, West Virginia, Wyoming) 90 percent or more of the sprawl is related to declining population density (that is, rising per capita land consumption). In five other states (Arizona, California, Hawaii, Nevada, Washington) at the opposite end of the spectrum, 90 percent or more of the sprawl is explained by population growth. The first group of six states has experienced relatively low population growth (or even "negative growth") while the latter group of five states has undergone explosive growth in the number of residents. The six high per capita sprawl / low-population growth-states converted 3,305 square miles of rural to developed land in the 15 years from 1982 to 1997. The five low Per Capita Sprawl / high-population-growth states lost 3,733 square miles of rural land over the same time period. We find that in the average state (the mean of the 49 state percentages), per capita sprawl explained 55 percent of the new land development, while population growth explained 45 percent. When calculating all the states’ population growth and land consumption growth together (the weighted average), we find that 50.3 percent of new land development is related to per capita sprawl and 49.7 percent to population growth. (Both are rounded to 50 percent in Table 7.) From this, one can easily say that roughly half of the increase in developed land in the 49 states from 1982 to 1997 was related to increase in land consumption per state resident, and half to the increase in the number of state residents.
It should be pointed out that while the percentage figures themselves are exact, this does not mean that actual sprawl corresponds precisely to these percentages. For example, it is unlikely that all sprawl (100 percent) is due only to population growth in Arizona and Nevada. Certainly there are places within these two states where sprawling, low-density subdivisions have indeed eliminated desert and cropland. Our analysis of Census and developed land data strongly suggests that in such cases, the overwhelming preponderance of sprawl is due to population growth. Conversely, even in those states that lost population overall from 1982-1997 (Iowa, Louisiana, North Dakota, West Virginia, Wyoming), and which thus show 100 percent of sprawl related to "per capita sprawl," there are certainly places in which population growth has played some role. But in these states as a whole, it’s a negligible one. The ‘Holdren Method.’ Apportioning shares of sprawl or the rate of sprawl between rising per capita land consumption (declining population density) and population growth can also be accomplished by means of applying a more mathematically rigorous method first described and partially developed by Harvard physicist John Holdren, internationally honored in 2000 for his achievements in environmental science.179 This method can be applied to virtually any type of resource use. Perhaps its best-known application has been in understanding how total U.S. energy use has risen in recent decades. The method enables analysts to apportion shares of the total rate of increase of energy consumption in a state, country, or the world as a whole to (1) the change in per capita energy use and to (2) the change in population. The Holdren method can also help us understand how much of the sprawl rate is related to declining population density, or rising per capita land use, and how much should be attributed to population growth. As in the case of national energy consumption, the question here is how much of the increased total consumption of rural land (Overall Sprawl) is related to per capita change in land consumption (per capita sprawl) and how much is related to increase in the number of land consumers (population growth). See Appendix E for further description. For all the complexity of this method and its use of logarithms, it produces only slightly different results in Table 8 than the ones in Table 7 from the more transparent calculation explained above (the "simple ratio" method). For now, we will simply provide the equation from the Holdren method that we use to determine the percentages brought about by rising population and falling density: Population’s share of growth rate = Annualized average population growth rate / Annualized average land development growth rate The term "annualized" means that the natural logarithm (ln) is applied to the rate of increase in each of the factors. This avoids the distorting effect of what is called in mathematical parlance a "second-order term." Another way of expressing this mathematical relationship is the following: ln (final population / initial population) + ln (final per capita land area / initial per capita land area) = ln (final total land area / initial total land area) Once these numbers are obtained, to find the percentage of the growth rate due to either population or density, either factor can simply be divided by the land area factor (and multiplied by 100 to convert to percent). For a more complete explanation, please refer to Appendix E. Holdren notes that this kind of numerical analysis cannot reveal the whole, complex story of population’s role in rising energy consumption, because of "nonlinearities." That is, in the numerical equation, population and per capita consumption are treated strictly as if they were separate, independent variables, when in fact, in the real world, they may interact with each other. In other words, they are likely to be interdependent in many cases. For instance, per capita energy use may depend on population size or growth rate, and so forth. But teasing these complexities apart in an effort to quantify them is all but impossible. Analogously, in a number of sprawl cases there is probably some degree of interaction between the amount of total urbanized or developed land, per capita land consumption, and population size or growth rates. Overall sprawl is not entirely a dependent variable, and per capita land consumption growth and population growth are not entirely independent variables. For instance, high rates of sprawl under the pressure of rapid population growth probably generate political pressure to implement more stringent land use controls and zoning, which may lead to higher residential densities than would occur otherwise. In general though, this numerical analysis allows us to quantify, with fairly high confidence in the broad accuracy of the results, the relative strength of the two factors — per capita land consumption and population — in forcing urban sprawl and land development. Holdren Method Analysis of States. Table 8 reports the results of the Holdren method using the state data. In comparison to the simple ratio method found in Table 7, the results in Table 8 obtain very similar results when looking at the nation as a whole. The state average (mean) for the two methods produces identical results: 55 percent of overall sprawl related to per capita consumption and 45 percent related to population growth. The weighted averages (obtained by aggregating the figures for growth in population and developed land for all 49 states) are very close for both methods. In the simple ratio method, the weighted average shows 51 percent of total sprawl related to per capita consumption, and 49 percent to population growth. In the Holdren method, these percentages are essentially reversed — 48 percent for per capita sprawl and 52 percent for population. Overall, analysis of the NRI data on increase in developed land for 48 contiguous states plus Hawaii from 1982-1997 strongly suggests that per capita sprawl (rising land consumption per person) explains roughly half of sprawl, and population growth the other half.
Regression Analysis of State Data. While the "simple ratio" and "Holdren" formulas above are very useful for apportioning the total amount of sprawl attributable to changes in per capita land use or population growth, they do not provide an estimate of how much land is actually lost to development, holding other factors constant. For example, the developed area of California and Vermont expanded by roughly the same percentage (32 percent and 31 percent, respectively) between 1982 and 1997. However, because California had a much larger developed area to begin with, its 32 percent expansion was 2,060 square miles while the 31 percent increase for Vermont was only 117 square miles. In order to address this question, we performed a regression with the state data. While regression is a very different approach than that utilized in the this study so far, the results that follow buttress the above analysis by showing the importance of population growth as a cause of sprawl, even after controlling for other factors. The Ordinary Least Squares (OLS) regression model takes the following form: Acres lost to development 1982-97 in 1000si
= where
 is change in per capita land use 1982-97 and
 is an error
term.
Results of Regression Using State Data. Table 9 reports the results of the regression. All of the variables are significant at the 0.01 level. The high R-squared indicates the strength of the statistical model. Turning to our variable of interest, population growth, we find that it behaves as expected. The results show that each 10,000-person increase in state population between 1982 and 1997, resulted in 1,600 acres of previously undeveloped land being developed. This is the case even after controlling for other factors such as the initial size of the state’s population or the total land area that was developed at the start of the time period in 1982. Also as expected, Table 9 shows that changes in per capita land use accounted for a good deal of the increase in developed land. For each 1,000-acre increase in the amount of developed land used per 10,000 people, 256 acres were developed within a state over the time period of the study. Overall the regression results lend strong support to our earlier findings indicating that population growth is an important underlying reason for sprawl.
Population Growth & Sprawl – Analysis of City Data Population Growth in the Worst and Least Sprawling
Cities. Table 10 reports the average population growth for cities
with the largest and smallest percentage increases in urbanized area
(sprawl). Tables with all data for the 100 largest Urbanized Areas can
be viewed on-line at
http://www.sprawlcity.org/hbis/index.html. As in the case of states,
this exercise shows that increases in urbanized area are associated with
population growth. Probably the starkest contrast can be found by
comparing the top-10 worst sprawlers with the 10 cities that sprawled
the least. In the 10 worst cities, population grew by 103 percent on
average between 1970 and 1990. In contrast, in the 10 cities that
sprawled the least, population grew by only 7 percent on The same general pattern exists if we look at the top and bottom five sprawlers as well and the top- and bottom-20 sprawlers. While other factors surely have also played a role, it is clear that those cities that expanded the most in size had dramatically higher levels of population growth than those cities that expanded the least. On its face, this indicates that there is likely to be a causal relationship between population growth and sprawl.
The Impact of Population Growth on City Sprawl. Table 11 examines the relationship between population and sprawl from the opposite direction as Table 10. Of the 100 Urbanized Areas examined in this study, 11 declined in population, while 89 experienced population growth between 1970 and 1990. • In the 11 cities that declined in population, urbanized area increased in size by 26 percent on average. This means that the density of these cities (land per person) fell, and the built-up area expanded. Thus, even without population growth there is still significant sprawl. • However, the expansion in urbanized area in cities that declined in population was modest in comparison to the 89 cities where population increased. The built-up area of cities that grew in population expanded in size 75 percent on average — or almost triple that of the 11 cities that declined in population. The relationship between sprawl and population growth holds for cities with differing levels of population growth; with more population growth comes more sprawl. In the 16 cities that grew in population by 10 percent or less between 1970 and 1990 (but whose population did not decline), urbanized area expanded 38 percent — more than in cities that declined in population but considerably less than in the cities where population increased more dramatically. Cities that grew in population by between 10 and 30 percent sprawled 54 percent on average. Cities that grew between 31 and 50 percent sprawled 72 percent on average. Cities that grew in population by more than 50 percent sprawled on average 112 percent. These findings confirm the common sense, but often unacknowledged proposition, that there is a strong positive relationship between sprawl and population growth. This relationship is depicted graphically in Figure 8.
Denser Settlement Did Not by Itself Prevent Sprawl. A central goal of Smart Growth is to slow or even stop the conversion of rural and less developed areas to urbanized land by preventing a decline in density. Thus places where density increased should be the success stories. Between 1970 and 1990, there were 18 cities where the density of their urbanized areas either remained the same or increased (another way of saying that the per capita land consumption declined). However, these cities experienced very significant sprawl. As the bottom of Table 11 shows, the urbanized area of these 18 cities, which reduced per capita consumption by increasing density, expanded 52 percent on average into the surrounding rural area. The reason these 18 cities sprawled so much despite an increase in density is that their population grew even faster than did their density. The population of these 18 cities increased 86 percent on average, while density increased only 17 percent. As a result, these 18 cities encroached on rural and undeveloped areas at a 52 percent rate, less than the average city which sprawled 70 percent, but considerably more than the 26 percent increase in cities that experienced population decline.Applying the Holdren Method to the Urbanized Areas. As we have seen, the Holdren formula can be used to apportion the amount of sprawl related to increases in land use per person and increases in population. Using this approach reveals results very similar to those of the analysis done on the states, with roughly half (52 percent) of sprawl related to population growth and 48 percent related to increases in per capita land use. Taking the measure of all cities and states analyzed, population growth accounts for slightly more than half of sprawl in the nation’s cities, and somewhat less than half in the nation’s states. In sum, it would appear that population growth accounts for about half of sprawl. The other half of sprawl reflects an increase in land per resident, which itself is the outcome of at least two-dozen factors. Thus, it is reasonable to conclude that of the dozens of factors contributing to sprawl, population growth is the single most important. It is also reasonable to conclude that population growth has approximately the same influence on sprawl as all the other factors combined. There Is Great Variation in the Reasons for Sprawl from Place to Place. As Table 11 indicates, there is tremendous variation in the percent of sprawl due to population growth vs. declining density in the nation’s 100 largest urbanized areas between 1970 and 1990. In 18 of the cities, population growth accounted for 100 percent of sprawl. In 11 cities, population growth accounted for 0 percent of sprawl. In the remaining 71 cities, some combination of population growth and declining density was responsible for sprawl. Thus focusing on just one factor such as density or population growth cannot solve the problem. Clearly a multifaceted approach that addresses all of the causes of sprawl is necessary. Continue to Policy Implications
Center for Immigration Studies Home Page Endnotes 174 USDA Natural Resources Conservation Service. 2000. Summary Report – 1997 National Resources Inventory (revised December 2000). 84 pp. Prepared in collaboration with the Iowa State University Statistical Laboratory. 175 The NRI is based on scientifically-selected sample sites in 49 states. The sample constitutes a two-stage, stratified area sample of the entire country, with 800,000 sample points representing each county of every state except Alaska. Photo-interpretation and remote sensing were utilized extensively. Since 1982, the NRCS has conducted an NRI every five years under the same procedures and protocols, so that nationally-consistent databases are available for 1982, 1987, 1992, and 1997. (Although the next survey was conducted in 2002, the results will not be available for at least another year or two.) Analysts can be confident that variations in land figures between different inventories reflect differences "on the ground" rather than in sampling methods. Nevertheless, as in all statistical samples, there are margins of error. 176 The U.S. Census Bureau data sources used in this study are: 1990 Census of Population and Housing, Summary Population and Housing Characteristics — United States, Table 8 – Land Area and Population Density; 1980 Census of Population, Number of Inhabitants, United States Summary, Table 34 – Population, Land Area, and Population Density of Urbanized Areas: 1980; 1970 Census of Population, Volume 1 Characteristics of the Population, Part 1, United States Summary (issued June 1973), Table 20 – Population and Land Area of Urbanized Areas: 1970 and 1960. All of these are available from the Statistical Information Office (Population Division) of the U.S. Department of Commerce’s Bureau of the Census in Maryland (301-457-2422). 177 U.S. Census Bureau. No Date. Chapter 12. "The Urban and Rural Classifications." 178 West Virginia is a special case for two reasons: First, the topography of the "mountain state" is such that a great deal of land often has to be leveled for even modest development or resource extraction projects like "mountain-top removal" surface coal mining. Second, West Virginia has seen an increase in vacation home construction and tourism-related development as a consequence of its natural beauty, low land prices, and proximity to the rapidly growing and affluent Washington, D.C. metropolitan area and to a lesser extent the Pittsburgh metropolitan area. The building of vacation or second homes and tourist facilities is unusual because land is cleared for housing and the infrastructure necessary to support it, but there is little or no corresponding increase in population because the Census continues to count persons as living at their primary place of residence out-of-state. 179 John P.
Holdren. 1991. "Population and the Energy Problem." Population and
Environment, Vol. 12, No. 3, Spring 1991. Holdren is Teresa and John
Heinz Professor of Environmental Policy and Director of the Program on
Science, Technology, and Public Policy at Harvard University’s Kennedy
School of Government, as well as Professor of Environmental Science and
Public Policy in the Department of Earth and Planetary Sciences at
Harvard University. Trained in aeronautics/astronautics and plasma
physics at MIT and Stanford, he previously co-founded and co-led for 23
years the campus-wide interdisciplinary graduate degree program in
energy and resources at the University of California, Berkeley. In 2000,
he was awarded the Tyler Prize for Environmental Achievement at the
University of Southern California, which administers the award. The
Tyler Prize is the premier international award honoring achievements in
environmental science, energy, and medical discoveries of worldwide
importance. |
is state population in 1982;
is change in state population 1982-97;
is acres of developed land in 1982;
