Back to Black … and Green? Location and policy interventions in contemporary neighborhood housing markets

Abstract

The post-war flight from US central cities led to widespread decay and devaluation in downtown housing markets. In a reversal of fortunes, distant housing prices soared while the dense urban core lagged. However, over the course of the 2000–2006 housing bubble, we find that the markets in often ignored mid-sized cities shifted back to the downtowns. This research examines the factors influencing neighborhood housing values, including location and public policy interventions. Our analysis period begins with 2000 and has two end points: one at the close of the national housing bubble in 2006 and another in 2008 during the housing market collapse. Based on OLS and spatial regression analyses of percent increases in neighborhood housing values for Louisville, Kentucky, we find that higher downtown property increases are due in large part to historic preservation districts, a university–community partnership, and a HOPE VI site. We confirm that our findings hold even through the 2007–2008 housing crisis. We ultimately theorize that higher downtown appreciation is due to three factors: green urbanism, planning/policy successes, and the surprising non-significance of the traditionally negative predictor race (nonwhite percentage).

Keywords:

• housing
• location
• neighborhood
• policy
• preservation

Acknowledgments

The authors thank the following individuals for their helpful comments or assistance in obtaining data: Tony Lindauer, Donna Hunt, Jay Mickle, Ron Crouch, Vernon Smith, Michael Price, Martye Scobee, Tim Clay, Frank Goetzke, Casey Dawkins, and the anonymous reviewers, as well as participants at the ACSP-AESOP Fourth Joint Congress in Chicago, the 2008 Fulbright Summer School at the Vienna University of Economics and Business Administration, and the Rethinking Transportation for a Sustainable Future conference in Louisville. Special thanks to Katrin Anacker the former managing editor of Housing Policy Debate for her detailed suggestions. Authors' names are in alphabetical order signifying that each made an equal contribution.

Notes

¹Louisville, Louisville Metro, and Jefferson County, KY, now refer to the same geographic area. Louisville Metro, however, is not synonymous with the larger multi-county Louisville Metropolitan Statistical Area, which covers several Kentucky and Indiana counties. This analysis is restricted to Jefferson County.

²The Louisville Metro government “double-counts” the populations of remaining, independent lower-class cities within Jefferson County and, when they are added to the population total for Louisville Metro, claims Louisville is the 16th largest city in the United States.

³The inner and outer beltways are the Watterson Expressway (I-264) and the Gene Snyder Expressway (I-265).

⁴For simplicity, we define the east-end as neighborhoods east of I-65 and the west-end as neighborhoods west of I-65. I-65 is the vertical highway included in Figures 1 –3.

⁵A list can be found at Louisville Metro's Historic Landmarks and Preservation Districts Commission website at: http://www.louisvilleky.gov/PlanningDesign/Historic+Landmarks+and+Preservation+Districts+Commission.htm.

⁶Percent nonwhite approximates percent black in the Louisville context since there are few Hispanics as of the 2000 Census, although this is beginning to change.

⁷We also tried using housing density (units per square mile) as a measure of average housing size but the variable was not significant.

⁸This variable is only an approximation of proximity to employment since adjacent tracts may also have high and proximate employment. Thus, our model likely underestimates the impact of this variable. A spatially lagged independent variable may be an improved measure. We use density instead of raw numbers of employees since census tracts on the periphery are much larger than downtown tracts. This variable excludes the highest employment tract, the central business district, because it is excluded from the analytic sample.

⁹Countywide crime rates for 2000–2002 are not available since the county was, at that time, served by many police agencies. The first post-consolidation numbers for 2003 are unreliable, so 2004 was the best year available to control for crime.

¹⁰Three districts are grouped at 2 and two are grouped at 3 due to nearly identical crime rates.

¹¹Typically neighborhoods are grouped together based on common crime levels and types of regular offenses.

¹²We also ran SLM and SEM versions of all other specifications (not shown). The results are very similar to those for specification 3.

¹³Note that Table 7 does not include the variable percent change in units. The JCPVA has numerous classes of residential properties. While every attempt was made to include the same classes of residential properties in 2008 as for 2000, it was not possible to 100% verify that the same exact properties were selected since we did not do the selection for 2000.

¹⁴As Tiebout (1956, 416) noted, “while the residents of a new government housing project are made better off, benefits also accrue to other residents of the community in the form of the external economies of slum clearance.”