Can we use administrative data to quantify informal housing additions at the parcel level? An analysis of Austin, USA

Abstract

Scholarship on informal housing is broadening to include more studies of cases in the Global North, with many of them taking advantage of ready data availability to produce quantitative estimates of the extent of the phenomenon even in diffuse contexts where studying informal housing has heretofore been difficult. We introduce a novel technique, Informal Housing Addition at the Parcel Scale (IHAPS). Relying on publicly available administrative tax assessment and building permit datasets and drawing inspiration from building science, IHAPS seeks to identify residential parcels within a city that underwent apparent increases in building area but that did not obtain a legally required building permit. Applying IHAPS to Austin, Texas, USA, we demonstrate the basic soundness of the technique in spite of its generation of false-positive identifications. We conservatively estimate that approximately one out of every 2400 single-family residential parcels in Austin experienced an informal (unpermitted) addition between 2008 and 2018. IHAPS has promise to reveal new insights into informal city-building processes in a Global North context, but also raises ethical and privacy concerns that researchers must consider.

Keywords:

• Informal housing
• Global North
• administrative data
• parcel data
• building science

Acknowledgement

The authors thank Dr. Nicole Gurran and the participants in the Informal Housing Research Colloquium at the University of Sydney in June 2019 for in-depth feedback that shaped this research.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Because of our concerns about subjecting Austin residents to building inspection and law enforcement scrutiny, discussed at length within the article, we do not make our data set freely available. We will consider individual requests provided that adequate protection measures are used.