What Drives the Partnership Decision? Examining Structural Factors Influencing Public-Private Partnerships for Municipal Wireless Broadband

ABSTRACT

Municipalities across the United States are in various stages of wireless broadband implementation. Policymakers establish wireless broadband networks for a number of reasons, including promoting economic development, supporting internal operations, and providing affordable access to citizens. While some governments provide wireless broadband access as a public service, others partner with private firms. This article exploits the variation in approaches and examines the structural factors that give rise to public-private partnerships. The results show that partnerships for wireless broadband are more likely among municipalities with political-administrative autonomy and greater economic viability. For municipalities seeking to address inequities in broadband access, these results can have profound implications, as partnerships are often the most viable deployment option. In addition to examining determinants of wireless broadband partnerships, this study aids in understanding which municipalities are more likely to benefit from partnership approaches as this form of alternative service delivery increases.

ACKNOWLEDGEMENTS

I am grateful to Irfan Nooruddin for his insightful recommendations. I also thank the anonymous reviewers and symposium editors, Mildred Warner, Trevor Brown, and Germà Bel, for their valuable comments and support.

Notes

Notes: Adj. count R ² = 0.56; N = 189. Sample weighted at 1%.

^*** p < .01; ^** p < .05; ^* p < .10, two-tailed tests.

^a Percentages are sub-1% due to the number of events relative to the population. This is accounted for by weighting the sample in the rare events logit estimation (from which these values are derived).

Estimating broadband coverage in the U.S. is problematic. The FCC has long been under fire for erroneously reporting that 99% of Americans had broadband access, arriving at this figure by claiming that when at least one household in the zip code reported access to a high-speed Internet service provider, the zip code was coded as having access (Bosworth 2008). Recognizing the perils of biased reporting, Congress passed The Broadband Data Improvement Act of 2008 and charged the FCC with improving the quality of the data on broadband access.

The last mile represents the costly extension of fiber between the residence/business and the back-end infrastructure that supplies broadband service.

Wireless is provided primarily using Wi-Fi and WiMax solutions. Wi-Fi is a wireless local area network (LAN) with service range of 100 yards, whereas WiMax covers a broader service area (3 to 30 miles); however, WiMax can be less successful than Wi-Fi for indoor use. Municipal wireless networks are sometimes a combination of the two technologies (e.g., Houston, Texas).

Three dominant reasons are identified for achieving affordability in wireless technologies: (1) Wi-Fi operates in unlicensed spectrum (WiMax operates in either licensed or unlicensed spectrum), which means that there has been high participation in developing solutions in this area; (2) strong levels of interoperability through industry-led early efforts to standardize technology; and (3) low unit cost for equipment—spurred in large part through the mass integration of Wi-Fi chipsets in laptops and mobile devices (Bar and Galperin 2004).

There is a correlation between age and rural living; the Pew Internet and American Life Project (2006) cites that rural America is “older” than metropolitan counterparts (43% of rural Americans are over 50 whereas 37% of non-rural Americans are over 50).

Consider the town of Thomaston, Maine, with a population of 3,748. Thomaston entered a public-private partnership with a local startup firm, RedZone, to provide municipal wireless access. RedZone's mission is to bring wireless broadband solutions to towns with small populations that are underserved by affordable broadband (Graychase 2006).

Mullins and Pagano (2005) note that 46 states have passed legislation limiting revenue generation and expenditures.

The dataset includes demographic data collected from the U.S. Census Bureau 2000 Census, 2002 Census of Governments, and U.S. Department of Agriculture 2003 Rural-Urban Continuum identification. Crosschecking was conducted via municipal and industry Web sites.

A total of 152 cities met the criteria but were dropped due to missing data because they failed to complete the U.S. Census Bureau's 2002 Local Government Directory Survey. The response rate for said survey is 70.3%; the response rate among the cities providing municipal wireless technology is 77.1%.

Of the 111 provisioning cities, 78 are partnerships and 33 provide the service in-house. As municipal-provision is not the unit of analysis of this study, these 33 municipalities are not included in the study.

For examples, see Nielsen et al. 2011; Davis and Bermeo 2009; Wade and Reiter 2007; Maestas, Maisel, and Stone 2005; King and Zeng 2001a.

The random selection of municipalities was generated using U.S. Census Bureau 2002 Census of Governments data. I limited the dataset to general purpose governments coded by the U.S. Census Bureau as municipalities, towns, or townships. Next, I assigned a random value between 0 and 1 to each observation (municipality, town, or township) and then sorted numerically on the random value. The first 111 observations (i.e., 111 lowest values between 0 and 1) were selected and assigned “0” values to create the non-events in the dependent variable. Two observations randomly assigned in the 111 set were already included in the dataset as partnering providers (“1” value), therefore I included the 112th and 113th observations to complete the non-events selection of 111 observations.

The USDA uses a nine-point scale, but given the number of observations in the model, some of the categories required collapsing due to insufficient variance. In the statistical models, suburban area is the omitted comparison category.

Factor analysis reveals that population and the metropolitan/suburban/rural indicators are not measuring the same underlying factor; metropolitan/suburban/rural is measuring density of the surrounding area while population is measuring the number of residents in a city.

The U.S. Census Bureau's 2002 Local Government Directory Survey does not include contracting data on information technology functions.

Additional information

Notes on contributors

Amanda M. Girth

Amanda M. Girth ([email protected]) is an Assistant Professor at the John Glenn School of Public Affairs at The Ohio State University. Her research examines government contracts, focusing on performance and accountability in the implementation process. She also studies the development and management of public-private partnerships.