What are the environmental consequences of suburban growth? Environmentalists advocate living in public transit friendly cities featuring higher population densities. In future posts, I’d like to discuss the benefits of sprawl but here I’ll focus on analyzing some of the key environmental costs caused by sprawl.
Here are some candidates:
A. More Driving leading to more smog, more SUVs and more CO2 production?
B. Longer Commutes?
C. More road paving?
D. More water consumption?
E. More conversion of farmland into suburbia?
Let’s start with Topic A and to earn some street credibility let me show you some mildly interesting facts. Using the 2001 National Household Transportation Survey, the unit of analysis is a household. For 13658 households who live in one of 49 metropolitan areas (these are all of the metro areas identified in the data set), I use data on the aggregate annual gallons of gasoline consumed by the household. The log of this variable is the dependent variable.
Controlling for the log of household income (linc), the household’s size (hhsize) and the head of household’s age (hhr_age and age squared), I want to show you new evidence on the relationship between urban population density and gallons of gasoline consumed by driving. Below, log(MSA Density) = log(average Population density in the Metropolitan area). This variable varies across metro areas but not within metro-areas. Log(Census Tract Density) = log(population density in the census tract where the household lives). The variablevaries across and within metropolitan areas. From reading Jim Hamilton’s blog I see that he posts univariate regressions so I know that folks are hungry for more regression output, so here it is!
Log(Gallons) = controls -.252*log(MSA Density) - .307*log(Census Tract Density) + .934*log(household Income)
You learn 3 things from this multivariate OLS regression. Gasoline’s income elasticity is close to 1. Controlling for household demographics, a 10% increase in a metropolitan area’s density reduces gasoline consumption by 2.5%. According to my Data, San Francisco’s “ldens” = 9.33 while Houston’s “ldens” = 8.37, so I predict that all else equal if a household moved from San Francisco to Houston and lived at the same population density at the census tract level (ldens2), its gasoline consumption would increase by: exp( -.25*(8.37-9.33))- 1 = 27.1% That’s a pretty serious sprawl effect!!
Controlling for a metro area’s overall density, a 10% increase in census tract population density reduces gasoline consumption by 3.1%. For those of you who care about robustness, the within metro area elasticity does not change if I include MSA fixed effects.
When households live in the suburbs, they drive more miles and they are more likely to own a SUV (see my book Green Cities: Urban Growth and the Environment posted to my Fletcher webpage). Both of these facts generate the gasoline result.
So, let’s return to our checklist;
More Driving (yes), more SUVs (yes) and more CO2 production (yes)?
More smog? No! To appreciate this point let’s go to where the action is ---
The Los Angeles Basin suffers from the highest levels of air pollution in the United States, with the pollution caused mainly by vehicle emissions. But Los Angeles has made dramatic progress on air pollution over the last 25 years. For ambient ozone, a leading indicator of smog, the average of the top 30 daily peak one-hour readings across the county’s 9 continuously operated monitoring stations declined 55% from 0.21 to 0.095 parts per million between 1980 and 2002. The number of days per year exceeding the federal one-hour ozone standard declined by an even larger amount—from about 150 days per year at the worst locations during the early 1980s, down to 20 to 30 days per year today. (Data source: California Ambient Air Quality Data CD, 1980-2002 (California Air Resources Board). Recent pollution gains are especially notable because Los Angeles County’s population grew by 29 percent between 1980 and 2000, while total automobile mileage grew by 70 percent (Census of Population and Housing 1980 and 2000; California Department of Transportation 2003). For air quality to improve as total vehicle mileage increases indicates that emissions per mile of driving must be declining sharply. Returning to the language of Scale, composition and technique --- for urban air pollution composition and technique effects have offset scale effects.
Areas such as Riverside and San Bernardino in California have experienced enormous growth while ozone levels have fallen sharply. Sprawl didn’t cause this pollution reduction but it also didn’t seriously exacerbate the initial problem. For more evidence on this see my paper --- Smog Reduction's Impact on California County Growth. Journal of Regional Science, August 2000.
An Open Research Question: As cities cleanup their smog problems, what is the marginal benefit from further smog reductions? Are there diminishing returns to clean air days? To push this point a little further --- in a heterogeneous population where some people (think asthmatics) are more at risk than others to suffer from pollution, if our goal is to protect these folks must we scrub the air such that there are no “Smog Days” each year? Or could a city be a “Green City” if it has 5 smoggy days a year but does a great job predicting when those days will occur and spreading the information using the media and the Internet to get the word out to susceptible people to stay inside or go away those days.
In future days, I will return to points B-E above.