Exploring the Value of Public Investment in “Generative” Fiber Infrastructure


In an earlier post I described the BTOP Comprehensive Community Infrastructure (CCI) program as a “very good investment of public funds.” My reasons were twofold, the first one being that it expanded the availability of high-speed connectivity in underserved areas, including more than 42,000 miles of new and 24,000 miles of upgraded fiber infrastructure. The second was that research by ASR Analytics suggests that the CCI program accomplished this expansion in a way that addresses both forms of economic harm claimed by advocates on both sides of the special access regulation debate. As a result, I suggested “that the federal government consider expanding its CCI investment in geographic areas that the FCC’s special access data collection project indicates still face a lack of competitive options and an abundance of excess-profit-extracting prices in the special access market.”

In five related posts, I considered a number of issues and perspectives that inform this policy suggestion, including the following:

In this post I’m going to:

I’ll start with an excerpt from an earlier post:

According to Table 7 on pg. 15 of ASR’s final report, the total amount (including both federal grants and matching funds) budgeted for 109 CCI projects was $3.9 billion. The table also indicates that, at the time the study was done, these projects had connected 21,240 CAIs, at a budgeted cost of $184,141 per CAI. Assuming federal grants paid for 80% of this total cost, the average federal grant amount per CAI would be in the neighborhood of $147,300.

Table 13 on pg. 34 of the report shows the changes in subscription speeds and pricing experienced by the 86 CAI locations providing this information to ASR. The table shows very large increases in speed and, depending on the category of CAI, dramatic 94-96% average reductions in per-Mbps pricing. Table 14 on pg. 36 uses these reported changes in speed and price to extrapolate CAI cost savings from switching to CCI-provided fiber connections. Averaged across all CAI categories, the per-CAI annual savings amounted to $236,151.

This means that, in just one year, the average CAI saved 28% more in operating costs ($236,151) than the total capital cost ($184,141) required to connect it to a CCI fiber network, and 60% more than the federal government’s share of that investment ($147,300). Based only on these direct social costs and benefits, I’d consider this a good investment of public funds.

But these direct cost savings to CAIs were not the only impacts of the federally supported BTOP fiber deployment program that were considered by ASR. It also estimated economic benefits driven by increased broadband availability in areas newly reached by the BTOP fiber networks. Using matched pair county-level analysis, ASR found that CCI-impacted counties achieved broadband penetration two percentage points higher than control counties. Based on this, ASR derived estimates of economic benefits of the $3.9 billion in CCI network investments using a number of widely accepted economic impact models.  These impacts included:

These findings of the ASR Analytics study suggest that:

Building on the ASR Analytics evaluation study

As noted above, ASR’s BTOP evaluation study used matched pair analysis of CCI-impacted counties to compare their growth in broadband availability to that of counties that were comparable on key control variables. ASR used NTIA availability data for multiple time periods to measure and compare these changes in availability (for more details see Appendix D of the ASR final BTOP evaluation report).

As discussed above, ASR found that, on average, the increase in broadband availability for CCI-impacted counties was two percentage points higher than in control counties, using the then-current broadband speed threshold of 3Mbps downstream service. ASR then used this differential to estimate and extrapolate economic impact variables (e.g., GDP, job growth, income) using the broadband impact models referenced above.

In light of ASR’s well-documented research and its promising though preliminary findings, an effort to update and expand on the strong foundation it and NTIA have built strikes me as timely, especially with special access policy questions getting focused attention from the FCC. More specifically, what I’d propose is to:

1. Use updated FCC availability data to explore how the matched-county broadband availability differential has evolved over a longer period of time.

2. Examine this broadband availability differential using speed thresholds higher than the 3 Mbps downstream level used by ASR, including the FCC’s current threshold of 25 Mbps downstream and 3 Mbps upstream.

3. For counties for which data is available, add to the matched pair comparison an analysis of broadband adoption data derived from the Census Bureau’s American Community Survey (ACS). Beginning with 2013 data, this data is being released annually for geographies with populations greater than 65,000, and should be available for virtually all counties on a blended five-year basis starting in 2017.

4. Examine and compare actual county-level economic indicators (e.g., County Business Patterns and other datasets available from the Census Bureau and other sources) for the matched pair counties.  The goal here would be to explore the extent to which the economic impacts predicted by the models used by ASR actually occurred and/or whether there were other impacts suggested by these economic indicators.

5. Where notably large variations are found among the matched pair differentials in broadband availability and/or penetration, and/or in actual economic impact variables, explore potential reasons for these differentials based on qualitative and/or quantitative analysis of CCI projects and CCI-impacted counties exhibiting these large variations. The goal here would be to extract additional insights and lessons learned regarding how CCI networks can best deliver social value, as well as the contextual factors impacting how effective different approaches are in achieving that value.

Factors to be considered in #5 might include the ownership and management models employed by CCI grant recipients, the specific approach they take to providing “open access” to their fiber networks, as well as other policies and strategies they employ in relation to wholesale and last mile providers, CAIs, local community development programs, and local economic, demographic and institutional factors.

For example, the specifics of how CCI grantees approached the BTOP program’s open access requirement have not been uniform, as explained on pages 28-29 of ASR’s final evaluation report:

To help expand service within unserved and underserved areas…[e]ach of the grantees in the evaluation study sample implemented at least one strategy, and in many cases a combination of strategies, to ensure open access to the BTOP-funded network by third-party service providers. For example, the research and education network and the healthcare network in Arkansas established a partnership to deploy new and upgraded fiber and colocation facilities. Merit Network in Michigan offered indefeasible right-of-use agreements to private third- party service providers. MassTech fostered competition by helping CAIs compare services and prices offered by third-party providers that use the BTOP-funded network.

Similarly, different CCI grantees adopted different usage and pricing policies to support positive impacts of their network investments.  For example, as described on pages 3-4 of its case study of Merit Network, a Michigan-based CCI grantee owned by member institutions of higher learning, ASR explained that:

The Merit network connects institutions of higher learning and facilitates collaboration by allowing them to freely connect to other institutions on the network, or access on-net services at speeds up to 1 Gbps. This allows institutions to collaborate on research, and to cut costs by sharing services, including hosting. Merit provides some content over this network as well, including Internet2. These services give faculty, staff, and students fast and reliable access to educational and research opportunities…The free on-net services provide incentive for CAIs to create wide area networks (WAN) using Merit fiber…[and] cost-savings [and greater efficiency] for any CAI organization with multiple locations.

Merit is an example of the Research and Education Network (REN) category of CCI grantee.  Owned by member universities, it exhibits a range of generative characteristics, including support for training, collaboration and feedback among its user community, including:

[Regular] opportunities for Members to learn from each other and share best practices in the networking arena. Forums include the Michigan Information Technology Executive (MITE) Forum, Merit Joint Technical Staff (MJTS), Networking Summit, Bring Your Own Device (BYOD) Summit, and the Merit Member Conference (MMC).

The Merit Advisory Council (MAC) has a direct voice to our Board of Directors and leadership through which feedback and recommendations are provided.

The Merit Services Innovation Group enables Members to provide suggestions and feedback regarding current and future services.

Merit facilitates collaboration between Members and regularly contributes staff and resources to educational and research activities.

Professional Learning events are tailored to the needs of our Members and are offered at reduced cost.

One starting point for considering the impacts of differential policies, structures, strategies and programs among CCI grantees would be a careful review of the dozen CCI case studies conducted by ASR. Another would be using the expanded matched pair county analysis described above to identify differences in the availability, penetration and economic impact variables across CCI projects.

In my view, a research project along these general lines would: 1) help maximize the ongoing value provided by existing CCI projects; 2) provide valuable guidance for consideration of future programs designed to build on the success of and lessons learned from the BTOP program and; 3) shed light on policy debates and options related to special access and perhaps other communication and infrastructure policy issues.

Tags: , , ,

Community Broadband As Generative Infrastructure


April 16th, 2016

In an earlier post I discussed Marjorie Kelly’s framework for distinguishing “generative” vs. “extractive” ownership models.  In this post, I’ll try to further clarify this distinction by considering some key characteristics of community-owned local access networks in relation to Kelly’s framework (in my next post I’ll shift my focus to middle mile and special access fiber).

To get started, I’ll reiterate one of my key working premises, that when an extractive ownership model is combined with a lack of competitive pressure or corrective regulation (as is currently the situation in much of the nation’s local access and special access markets), service providers can achieve high levels of financial extraction.  This can lead to broad and substantial economic harm, especially in the case of core infrastructure like the Internet and telecommunications in general.  That’s because these infrastructure resources tend to be “spillover rich” when managed in a non-discriminatory way, but less so when constrained by dominant ISPs’ internal monetization priorities, as I discussed in the latter section of an earlier post, and which Brett Frischmann addressed in far more depth in his book, Infrastructure: The Social Value of Shared Resources.

As discussed in my last post, I view the longstanding focus on shareholder returns at the expense of network upgrades by AT&T, the nation’s largest ILEC, as one example of this dynamic.  Another relates to the nation’s leading cable operators.  On one hand these companies have been able—thanks to their much higher speeds relative to DSL—to capture nearly all net broadband customer growth in recent years (see the table here for 2015 data).  But, at the same time, these same companies have consistently been ranked at or near the bottom among all U.S. industries in customer satisfaction surveys.  Simple economic logic tells us that, if there was minimally healthy competition in the market for these higher-speed broadband connections, this combination of strongly positive market share gains with strongly negative customer satisfaction would be a very unlikely outcome.  As the title of Susan Crawford’s 2013 book points out, these growing ranks of customers signing up for and retaining cable modem service are, in a very practical sense, a Captive Audience.

Later on in this post I’ll suggest some research questions I believe are worth pursuing related to the operation and impacts of community-owned networks and the relevance of Kelly’s ownership framework to the broadband access market. But before I do, I want to consider how some of Kelly’s “generative” characteristics apply in theory and practice to community broadband.

In applying Kelly’s framework to local access ownership models, it makes sense to start with Purpose, which Kelly describes as the most fundamental design element.

Having studied a number of community-owned broadband networks, I’d say that all or most were undertaken with a purpose along the following lines: to provide local households, businesses and public service organizations (e.g., schools, healthcare providers, public safety, etc.) with affordable, reliable, symmetrical, high-capacity broadband connectivity and related services; to support their ability to thrive in an increasingly competitive and knowledge-based global economy and; to provide decent paying jobs to local citizens who, in turn, provide high quality customer service to the network’s customers.

While privately owned networks, including those owned by publicly-traded corporations, might claim to have the same or a very similar purpose, years of listening to earnings calls of these publicly-traded companies tells me that such goals are, at best, secondary priorities to the overriding goal of maximizing shareholder value, and ones that will be jettisoned if they conflict with the latter goal. And if shareholder value-maximizing decisions are not made with sufficient speed, vigor and clarity of (extractive) purpose, a publicly-traded firm’s management is likely to face intense pressure from investors, particularly those focused on a relatively short time horizon for measuring shareholder value.

It’s also clear to me that compared with most publicly-owned cablecos and telcos, community-owned networks tend to have locally Rooted Membership (vs. the Absentee Ownership characterized by publicly-traded stocks); Mission-Controlled Governance focused on achieving benefits for the local community (vs. Governance by Markets, stock price and related measures of financial profitability); and tend to function as part of local and national Ethical Networks focused on supporting sustainable community development (vs. Commodity Networks geared toward maximizing financial extraction).  And, as discussed further below, while the capital intensive nature of last mile access networks makes reliance on Stakeholder Finance challenging, there are models developing on this front as well.

A mix of generative characteristics and results

The general family of community-owned networks includes both municipal ownership (often through an existing municipal power utility) and end-user cooperatives (often through an existing rural electric co-op). While there are legal, organizational and financial differences between these ownership models, my preliminary research suggests these aren’t large enough to significantly alter the fundamental ways in which they differ from ownership by publicly-traded cable and telephone companies.

That being said, different combinations of generative characteristics may yield somewhat different sets of strengths and weaknesses, with these sometimes impacted by other factors, including relevant state laws and local regulations, existing institutional relationships and expertise, local market dynamics, and the mix of stakeholders supporting the project.

The mix of generative characteristics and situational factors can also impact how a community network evolves over time, and whether its particular model is sustainable within the environment in which it has taken root.

For example, in the case of municipally-owned networks, the departure of strong founding project leadership has sometimes led to a migration of decision-making to political leaders lacking appropriate industry expertise, particularly if the network was deployed in a community that lacked an existing public utility that was well-managed and enjoyed the loyalty and respect of its local customer base. In my view this type of management transition led to problems for the community-owned network in Burlington, VT, and highlights a potential point of design weakness for municipally-owned networks lacking a well-established and sufficiently independent, professional, non-political utility management unit.

In other cases, management of a municipally-owned network has experienced disruptive discontinuities when local political leadership has changed. This seems to be most likely if such change occurs when the network and/or city is facing financial or other challenges, and/or when candidates or newly elected officials view the shortcomings of their predecessors’ network project as a useful rallying point for mobilizing political support.  This also supports the notion that, once launched, the ongoing management of a community network needs to be sufficiently sheltered from day-to-day and election-to-election political pressures.  My research suggests that this is most readily achieved in communities that already have some form of municipal utility infrastructure (or set up a strong-enough one when a network project is launched).

Similarly, the interaction of political dynamics among multiple jurisdictions has sometimes complicated and/or delayed decision-making, adding to the challenges of addressing unexpected financial or operational problems. My sense is that this has been a factor for the multi-city UTOPIA fiber network in Utah, one that was made even more difficult by state restrictions on the network’s ability to provide retail services.

This latter point highlights the significance of state regulation as a factor that can influence which generative characteristics are likely to be most effective (or even legal) in achieving a project’s internal and external goals.  In the case of UTOPIA, the retail-service prohibition presented  serious and not-well-understood challenges and risks related to marketing, finances, technology and overall management of a wholesale-only local access enterprise.  As those risks became more clear with time, the task of addressing them was made even more challenging by the project’s multi-city management and financial structure.  As a result, this pioneering and arguably over-ambitious project, launched a dozen years ago, has become the perennial poster boy used by critics to make their case against community broadband.

In my view, the risks related to local political dynamics suggest that communities without existing public utilities take these risks very seriously.  While establishing a strong and sufficiently independent municipal utility structure is one option, another would be to adopt a form of cooperative structure (e.g., similar to the rural telephone and electric cooperatives common in very rural areas).  This would help ensure that the network is responsive to end-users rather than to local politicians with many and often conflicting priorities and, in some cases, focused too much on the next election and lacking appreciation for the management requirements of the network. That being said, cooperative managers are not immune from losing touch with their members needs, nor can these members always be relied upon to wisely exercise the rights and responsibilities of their membership.

These challenges highlight the value of initial and sustainable stakeholder “buy-in” for a community network to succeed, a factor that relates to Kelly’s concepts of Rooted Membership and Mission-Controlled Governance.

The value of building and maintaining stakeholder buy-in seems especially important during a network’s early years, when the bulk of construction is underway. This is because costs are especially high during this startup phase, while revenues are just beginning to ramp up. Case study research suggests that it’s during this startup period that incumbents have the most leverage to mobilize their considerable resources to weaken both the economic viability and credibility of community networks.

Given its role as core communication infrastructure,  a community-owned broadband network is a resource likely to impact virtually every organization in a community, including local government, public safety, education, healthcare, non-profits, and businesses both large and small.  This suggests that important elements of generative structure (e.g., Rooted Membership and Mission-Controlled Governance) will be closely tied to how these various stakeholders (as well as residential users) are involved in setting goals and priorities and the decision-making processes related to network management, build-out plans, resource allocation, service development, pricing, etc. This governance issue also relates to the Ethical Network element of generative ownership, as reflected in the relationships among community leaders, their local constituencies and their counterparts in other communities that have invested in a community-owned network or are considering such an investment.

Stakeholder finance: challenging but potentially fortifying

Deploying a community broadband network that relies on Stakeholder Finance strikes me as more challenging than Kelly’s other generative design elements, in large part because the high upfront cost and capital-intensive nature of communication networks has typically required access to public debt markets.

One of the clearest and apparently successful examples of local stakeholder financing is Vermont’s ECFiber, now officially known as the East Central Vermont Telecommunications District.

As the following excerpt from ECFiber’s website indicates, its decision to initially rely on local stakeholder financing was made out of necessity:

On Town Meeting Day 2008, 24 towns voted to join ECFiber…In August, 2008 23 towns signed the Inter-local Contract and by early September, the initiative’s underwriter, Oppenheimer & Co., had pledges of $70 million. One week later the international financial markets collapsed taking ECFiber’s initial funding effort with it.

ECFiber then submitted several funding proposals under the American Recovery and Reinvestment ACT (stimulus program), but with no operating history at that point, we were edged out by competing proposals from other local companies.

But undaunted, ECFiber returned to the Vermont roots of self-reliance and initiated our current program of grass-roots funding. With the advice of local counsel, ECFiber developed a program of issuing promissory notes in a private placement offering. The notes are offered in $2500 units. The first round of financing, in January,2011, raised $912,000, which enabled us to build our first 20+ mile loop… Additional rounds of financing have brought total investment to nearly $5 million…It is ECFiber’s intention, at some suitable point, to return to the capital markets to seek sufficient funding to build out the entire network in all member towns.

According to a March 11, 2016 press release, ECFiber has reached the point where it is ready to augment its initial base of local stakeholder financing via institutional capital markets:

ECFiber…plans to activate 110 miles of network in 2016 and build an additional 250 miles in 2017. “Working with bond underwriters, we believe ECFiber has reached the point in its financial development that allows us to access institutional capital markets for the first time in 2016,” says Irv Thomae, District Chairman.

While the financing model pioneered by ECFiber may lead to slower network buildout, it may strengthen the Stakeholder Financing element of its generative ownership design, since most of its initial funding came from community members with a three-pronged interest in its success—as customers, as local community members, and as direct financial investors.

This underscores a broader and important point: community broadband planners are likely to increase their chances for success if, from the beginning, they keep in mind all elements of Kelly’s generative ownership structure, including how they interact with each other to support the project’s Living Purpose.

To a large extent, all of Kelly’s ownership characteristics relate to the effective and sustainable harnessing of stakeholder support and participation.  As both successful and unsuccessful community broadband projects have demonstrated, strong and sustained support from community stakeholders provides a solid—and perhaps the most essential— foundation upon which to build a community network.

As noted above, this foundation has proven to be especially important during a project’s startup phase, when learning curves and financial pressures abound, and when community networks are likely to be most vulnerable to well-financed political, legal and predatory pricing attacks by incumbent service providers. The more fully and firmly that stakeholder support is embedded in a community network’s design, the more likely it is to weather these startup storms and any squalls that might follow in later years. And the more likely it will be to remain focused on prioritizing the social benefits and community development goals that give it the Living Purpose that distinguishes it from the financial priorities of extractive ownership models.

More research can help

The existing body of research focused on community broadband networks tends to be heavily polarized and somewhat anecdotal, with proponents focusing on success stories and opponents on the sector’s most notable failures, even if, as with UTOPIA, they were launched many years ago and were subject to a unique mix of situations and constraints that virtually guaranteed they would illustrate painful lessons for others to learn from. To a large extent the tone and content of existing research reflects the often intense political battles at the state, local and national levels regarding restrictions on community broadband projects.  As with most politically charged policy-related research, the result is a strong tendency toward cherry-picking of projects to study and data to analyze.

My own view is that case study-oriented research by myself and others provides clear evidence that community-owned broadband networks can and often do succeed in terms of both their internal economics and in bringing to their communities lower prices, faster speeds, better customer service and more robust support for the potentially large but difficult-to-internally-monetize social goods discussed in Frischmann’s book.

That being said, I also believe that state and local policymakers, local decision-makers and communication scholars could benefit from additional and less agenda-driven research in this area, perhaps conducted by a team of researchers representing a range of perspectives and expertise, and well sheltered from bias based on the source of their funding.  Among the questions I view as worthy of such research are the following:

Given the intensity of debate surrounding state laws restricting the ability of communities to finance and control their local broadband networks; the FCC’s efforts to preempt such state restrictions and; the expansion of both publicly-owned and privately-owned (e.g., Google Fiber) competitive network models, I believe research focused on these questions can help local leaders, state and federal policymakers, and private sector players make better-informed decisions about how best to leverage the power of high-speed Internet access to benefit our nation’s citizens, businesses and public institutions.

While the focus of this post has been the relevance of Kelly’s ownership framework to the local broadband access market, I believe it is also relevant to policy issues and research questions related to the special access market, which is the focus of the final blog post in this series.


Tags: , , , ,