Turning Housing Into Driving: Parking Requirements and Density in Los Angeles and New York

Abstract

This article examines the idea that residential minimum parking requirements are associated with lower housing and population densities and higher vehicle densities (residential vehicles per square mile). Cities frequently use minimum parking requirements to manage traffic, but parking requirements accommodate vehicles, suggesting they should lead to more driving and congestion rather than less. If parking requirements reduce congestion, they likely do so not by reducing the number of vehicles in an area but by reducing the densities of housing and people. We support this idea by comparing the Los Angeles and New York urbanized areas. We show that differences in housing, vehicle, and population densities across and within these urbanized areas are closely correlated with differences in the share of housing units that include parking, and that the share of housing units that include parking is in turn correlated with the stringency of parking requirements. Compared with Los Angeles, New York shifts less of the cost of driving into its housing market. We further show that within New York City, a 10% increase in minimum parking requirements is associated with a 5% increase in vehicles per square mile, a 4% increase in vehicles per person, and a 6% reduction in both population density and housing density. These relationships remain even after controlling for street layout and proximity to the subway.

Keywords:

• parking requirements
• regulatory barriers
• density
• transportation
• land use

Acknowledgments

Katie Matchett, Ben Cummins, and Adam Blair provided excellent research assistance. Dan Chatman, Namji Jung, George Homsy, Kate Lowe, Michelle Craven, Greg Pierce, and Paavo Monkonnen gave helpful comments, as did seminar participants at the ACSP and the New York City Department of City Planning. We thank Simon McDonnell, Josiah Madar, and Vicki Been for sharing parking requirement data and David King for subway data.

Notes

^1. Shoup (2005) summarizes this evidence and provides an overview of parking requirements.

^2. These relationships hold at the census tract level as well. Across the 18,080 tracts in the country's 25 largest urbanized areas in 2000, the correlation between population density and vehicles per person was 0.6, while the correlation between density and vehicles per square mile was 0.75.

^3. Fuel taxes are user charges, but evidence suggests that American fuel taxes are inefficiently low (Parry & Small, 2005).

^4. In 2012, the price of a Volkswagen Passat in Singapore (including taxes and fees) was $152,000, just under the median price of a U.S. metropolitan area home (Goodman, 2012).

^5. In both Los Angeles and San Francisco, the two urbanized areas with higher population densities than New York, vehicle density is 55% of population density.

^6. Seph Murtagh, Ithaca Planning and Economic Development Commission, June 13, 2012.

^7. For example, the City of San Francisco estimates that off-street parking in its downtown costs between $50,000 and $60,000 per space (Lee & Mayor of San Francisco, 2012).

^8. The year 2003 was the last time when New York, northern New Jersey, and Los Angeles were surveyed in the same year.

^9. Newark, New Jersey, which we count as a suburb, also contributes to this difference. If Newark is removed, 91% of the housing in New York's suburbs has bundled parking.

^10. The city introduced some parking requirements in 1950, but the 1961 zoning resolution is generally considered the advent of parking requirements in New York City (New York City Department of City Planning, 2009).

^11. This figure doubtless reflects some sampling error—surely a housing unit without included parking spaces was built somewhere in Los Angeles—but the difference between the two areas is nevertheless substantial.

^12. This building would also qualify for a waiver from parking requirements, as we discuss later.

^13. The year 1970 might be a more appropriate baseline because the city probably used 1970 census data when it decided the location of parking maximums. But using 1970 as the baseline does not change the results and, if anything, reinforces them.

^14. An example of a denominator-driven vehicle-to-income ratio is Greenwich, Connecticut, which is in the New York urbanized area. Greenwich has a higher PCPI ($74,000) than lower Manhattan does and triple Manhattan's vehicles per $10,000 of personal income (0.10). Unlike Manhattan, however, Greenwich has a 70% drive-alone commute share, and 95% of its households have vehicles.

^15. African Americans own fewer vehicles than other groups do, possibly as a result of discrimination in vehicle financing (Ayres, 2001; Giuliano, 2003). Buehler (2011) reviews the determinants of vehicle ownership and use.

^16. Because we use spatial data, our linear regressions may suffer from spatial autocorrelation because of omitted variables. However, we believe our independent variables capture enough spatial variance to minimize this risk.

^17. AHS New York 2003, pages 59 and 109.