If you have noticed, this blog has been largely silent over the past few months. The reason was not just more frequent Twittering but in fact, real-world events like finishing up my doctoral dissertation. Those who know me well are aware that this has been a long time coming. The process of researching, writing, and defending a dissertation in any field is an intensely time-consuming and arduous task. It takes up you every waking moment whether you are working on it or not. Regardless of how much you enjoy what you are doing, focusing on a single task for couple of years fine-tuning to satisfy rigorous academic standards and personal wishes of your committee members who want you to give your best can be draining. Personally, I think a Ph.D. is more about testing your patience and endurance more than your intelligence. Below, I hope to share the research objectives and findings of my dissertation hopefully devoid of any jargon.
With undergraduate and graduate degrees in Architecture and Public Policy respectively, I chose to pursue my doctoral education in Urban and Regional Science which amalgamates urban planning principles with social science & economics theory. My dissertation focused on examining the housing price impacts on single-family residential properties before and after proximate Superfund remediation along with contemporaneous sociodemographic change. Superfunds are highly toxic and contaminated sites from erstwhile commercial and industrial use and are often abandoned due to health and liability concerns. These sites more often than not have a detrimental effect on the surrounding neighborhood and it is pertinent from the perspective of urban planning to understand the extent of impact of such properties. However in the literature, little evidence has been presented to evaluate the impact of remediation of such contaminated properties; partly since remediation is a recent phenomenon and economic and social changes post-remediation has longer lag time. Further, contemporaneous sociodemographic change has been rarely explored while examining economic impacts of contamination leaving a significant gap in the literature pertaining to neighborhood change following change in contamination status. The sociodemographic change can also be examined from the perspective of environmental and social justice to analyze if minorities and low-income people are disproportionately exposed to contamination and if they reap the benefits of subsequent remediation.
Hedonic models are used to isolate the effect of proximate environmental amenities or disamenties on housing price while controlling for associated structural and neighborhood factors that typical explain housing value. The environmental disamenities is measured by the distance to the nearest Superfund site and controlled by number of other Superfund sites in close proximity, type of contamination, size of contamination and even proportion of industrial and commercial properties to account for minor contamination. The dataset I used consisted of nearly 270,000 single-family residential properties complete with associated structural details like number of bedrooms, bathrooms, lot size, living area, and even length of resident tenure and tax rates. This dataset was complemented by information from USEPA and state environmental agencies on Superfund sites and other contamination and Census data like proportion of minorities, housing condition, unemployment and poverty rates, etc. for neighborhood factors. This allowed me to isolate the effect of the proximity of the Superfund site and compare the before and after remediation effects.
If that wasn’t enough, I incorporated the effect of spatial dependence in my analysis. Although, spatial dependence in housing markets (observations located closer influence each other more than observations located further away) is obvious, it has been rarely used in hedonic analysis; partly due to computational limitations. Recent advances in spatial statistics and development of GIS tools via softwares like ArcGIS, GeoDA, and extensions within Matlab make it easier to incorporate correction for spatial dependence. This helps in correcting for biased and inefficient estimators and even corrects for omitted variable bias which often plagues housing valuation. Dividing the region into housing submarkets using school quality also further refined the findings.
After running hazaar OLS and spatial models, the findings revealed a 2.2 percent increase in sales prices with every additional mile away from the contaminated Superfund and this effect diminished significantly after remediation indicating that although contamination has a negative effect, the subsequent cleanup had a net positive effect. Additionally, properties around unremedied Superfund sites sold for nearly 10.1 percent less than properties around remedied Superfund sites. The sociodemographic change analysis expectedly revealed that Superfund sites were disproportionately located in low-income minority neighborhoods causing an unfair burden. However, remediation did not lead to gentrification i.e. displacement of low-income population by upwardly mobile population with college-level education, as expected. It was also observed that Superfund sites in primarily premier housing submarkets were more likely to be remedied than those located in low-income submarkets again alluding to an environmental justice concern. This was disturbing since it was also found that remediation of Superfund sites located in low-income minority neighborhoods had a greater positive price effect than those located in premier submarkets; possibly because other amenities in premier submarkets such as better schools, coastline proximity, lower crime, etc. compensated for the presence of Superfund sites.
I was extremely pleased with my efforts and the results it produced and thankfully so was my committee since they did consider this as a significant contribution to the literature. Of course, it has to be developed further and refined for journal publication. Of course, I’ve not included all theoretical constructs, literature review, and model specification criteria that went into the analysis. For that, you’ve to read my 280-page ‘magnus opus’ which even my advisor couldn’t bear to look at again after her approval.
I successfully defended this dissertation last month and as of last week, the manuscript has been approved by the Thesis Office and Office of Graduate Studies at Texas A&M University. I’m set to graduate in August with a Doctor of Philosophy (Ph.D.) degree in Urban and Regional Sciences. And all I can say is, whew!