The Impacts of State Growth Management Programs on Urban Sprawl in the 1990S

ABSTRACT:

The urban landscape in the United States has been characterized by sprawl for decades. Despite arguments in favor of sprawl or its component attributes, it is widely believed that sprawl leads to many contemporary United States urban and environmental problems. Since 1961, 15 states in the United States have adopted state growth management programs (SGMPs) with various goals, including curtailing sprawl. This article, among a few previous studies, examines the effectiveness of SGMPs on containing sprawl. We create sprawl indices for 294 metropolitan areas in 1990 and 2000, measuring two major dimensions of sprawl: density and land-use mixture. Our examination of SGMPs involves not only a dummy variable indicating whether a SGPM exists or not, but a score system measuring the degree of state involvement in local growth management and three variables measuring three major attributes of SGMPs. With the refined measurements for both sprawl and SGMPs, we examine the impacts of SGMPs on the sprawl measures in the 1990s of the selected metropolitan areas in the United States. Our statistical results show that SGMPs effectively promoted compact development in terms of population density and land-use mixture. However, the statistical results do not support the claim that SGMPs with a higher degree of state involvement in local growth management, on average, worked better at curtailing sprawl than those with a lower-degree involvement in the 1990s. This article suggests that state governments in the United States should more fully exercise their responsibilities to control urban sprawl rather than just leave this issue to local devices.

Notes

1 Their study was conducted prior to the implementation in 1990 of the state of Washington’s growth management legislation.

2 We calculate the variable from Zip Code Business-Patterns data rather than from Census Transportation Planning Package (CTPP). Another difference is that we measure the degree of population-serving enterprise mix rather than the degree of population-serving job mix.

3 Zip code business data for 1990 are not available from the Census Bureau. Therefore, data for 1994 are used.

4 According to the geographic boundary of MAs defined in 1990, we know counties in each metropolitan area in 1990 and 2000. From National Association of Counties (http://www.naco.org), we retrieved all cities in each county on December 12, 2004; and we know the zip codes in each city from zip code business data 1994 and 2000. Based on all of the information, we calculate this variable.

5 The correlation coefficient between gross population density in 1990 and population-serving enterprise mix in 1994 is about −0.21, and the correlation coefficient between density in 2000 and the enterprise mix in 2000 is about −0.22.

6 Since the list of the sprawl scores for the 294 MAs would take a lot of space in the article, we advise readers that these are available from the authors by request.

7 In November 2004, Oregon voters passed Ballot Measure 37, which took effect in December 2004. Measure 37 requires that state and local governments provide compensation if they enact or enforce land use regulations that restrict the use of property or reduce its fair market value.

8 We thank Tim Chapin for bringing this to our attention.

9 We thank Tim Chapin for bringing this to our attention.

10 We collect information about population and the average wage for counties in 2000 from Census of population and online database provided by Bureau of Economic Analysis. Average wage for each MA in 2000 is then calculated, according to the geographic boundaries of MAs defined in 1990.

11 We compare the number of districts for each MA in 2002 from the Census website. We acknowledge that this set of data in 2000 may be more helpful but we expect that the number would not change much from 2000 to 2002.

12 We thank Tim Chapin for suggesting this to us.

13 See http://www.sierraclub.org/sprawl/report99/openspace.asp#ratings.

14 Since agricultural revenue data for 1990 are not available, 1992 data are used.

15 We acknowledge that growth rates in house prices may be endogenous with growth rates in densities. The standard error for this variable may be biased. However, this variable is not a key variable that we are interested in; therefore, the endogeneity is tolerable.

16 We collect information about median house values for counties in 2000 from the Census Summary File 3. Median house value for each MA in 2000 is then calculated as the average of the median house values for counties, according to the geographic boundaries of MAs defined in 1990.

17 We thank an anonymous referee for bringing this to our attention.